Copyright © 1993-2001, Hugh Jack

page 1

Dynamic System Modeling

and Control

by
Hugh Jack

(Draft Version 2.6, December 20, 2004)

© Copyright 1993-2004 Hugh Jack

 page 1

PREFACE
How to use the book.
• read the chapters and do drill problems as you read
• examine the case studies - these pull together concepts from previous chapters
• problems at the ends of chapters are provided for further practice

Tools that should be used include,

• graphing calculator that can solve differential equations, such as a TI-85
• computer algebra software that can solve differential equations, such as Scilab

Supplemental materials at the end of this book include,

• a writing guide
• a summary of math topics important for engineers
• a table of generally useful engineering units
• properties of common materials

Acknowledgement to,
Dr. Hal Larson for reviewing the calculus and numerical methods chapters
Dr. Wendy Reffeor for reviewing the translation chapter

Student background
a basic circuits course
a basic statics and mechanics of materials course
math up to differential equations
a general knowledge of physics
computer programming, preferably in ’C’

Special notes
- despite all common wisdom, inertia is presented as a force, this makes it easier
for students attempting to learn, and keep sign conventions correct

TO BE DONE
small
italicize variables and important terms
fix equation numbering (auto-numbering?)
fix subscripts and superscripts
fix problem forms to include therefores, mark FBDs, etc.
check C programs for ANSI compliance
big
verify the phase angle relationships cos vs/ sin.
 page 2

chapter rotation
replace the rotational case with IC motor
chapter non-linear systems
develop chapter
chapter magnetic
consider adding/writing this chapter
chapter fluids
consider adding/writing this chapter
chapter thermal
consider adding/writing this chapter
chapter c programming
review section
add problems

Buzzword topics:
- Distributed systems
- Intelligent manufacturing systems
- Adaptive control
- Architectures for signal processing and control algorithms
- Discrete event systems
- Hybrid systems
- Predictive control
- Robust control
 page 1

1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1
1.1 BASIC TERMINOLOGY 1.1
1.2 EXAMPLE SYSTEM 1.3
1.3 SUMMARY 1.3
1.4 PRACTICE PROBLEMS 1.3

2. TRANSLATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1
2.1 INTRODUCTION 2.1
2.2 MODELING 2.3
2.2.1 Free Body Diagrams 2.4
2.2.2 Mass and Inertia 2.4
2.2.3 Gravity and Other Fields 2.8
2.2.4 Springs 2.10
2.2.5 Damping and Drag 2.18
2.2.6 Cables And Pulleys 2.21
2.2.7 Friction 2.23
2.2.8 Contact Points And Joints 2.25
2.3 SYSTEM EXAMPLES 2.25
2.4 OTHER TOPICS 2.35
2.5 SUMMARY 2.36
2.6 PRACTICE PROBLEMS 2.36
2.7 PRACTICE PROBLEM SOLUTIONS 2.41
2.8 ASSIGNMENT PROBLEMS 2.45

3. ANALYSIS OF DIFFERENTIAL EQUATIONS . . . . . . . . . . . . 3.1

3.1 INTRODUCTION 3.1
3.2 EXPLICIT SOLUTIONS 3.2
3.3 RESPONSES 3.16
3.3.1 First-order 3.17
3.3.2 Second-order 3.23
3.3.3 Other Responses 3.28
3.4 RESPONSE ANALYSIS 3.31
3.5 NON-LINEAR SYSTEMS 3.33
3.5.1 Non-Linear Differential Equations 3.34
3.5.2 Non-Linear Equation Terms 3.38
3.5.3 Changing Systems 3.41
3.6 CASE STUDY 3.47
3.7 SUMMARY 3.51
3.8 PRACTICE PROBLEMS 3.51
3.9 PRACTICE PROBLEM SOLUTIONS 3.56
3.10 ASSIGNMENT PROBLEMS 3.61

4. NUMERICAL ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1

 page 2

4.1 INTRODUCTION 4.1

4.2 THE GENERAL METHOD 4.1
4.2.1 State Variable Form 4.2
4.3 NUMERICAL INTEGRATION 4.10
4.3.1 Numerical Integration With Tools 4.10
4.3.2 Numerical Integration 4.15
4.3.3 Taylor Series 4.21
4.3.4 Runge-Kutta Integration 4.23
4.4 SYSTEM RESPONSE 4.29
4.4.1 Steady-State Response 4.30
4.5 DIFFERENTIATION AND INTEGRATION OF EXPERIMENTAL
DATA 4.31
4.6 ADVANCED TOPICS 4.33
4.6.1 Switching Functions 4.33
4.6.2 Interpolating Tabular Data 4.36
4.6.3 Modeling Functions with Splines 4.37
4.6.4 Non-Linear Elements 4.39
4.7 CASE STUDY 4.39
4.8 SUMMARY 4.46
4.9 PRACTICE PROBLEMS 4.47
4.10 PRACTICE PROBLEM SOLUTIONS 4.50
4.11 ASSIGNMENT PROBLEMS 4.60

5. ROTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1
5.1 INTRODUCTION 5.1
5.2 MODELING 5.2
5.2.1 Inertia 5.3
5.2.2 Springs 5.7
5.2.3 Damping 5.12
5.2.4 Levers 5.14
5.2.5 Gears and Belts 5.15
5.2.6 Friction 5.19
5.2.7 Permanent Magnet Electric Motors 5.22
5.3 OTHER TOPICS 5.23
5.4 DESIGN CASE 5.23
5.5 SUMMARY 5.28
5.6 PRACTICE PROBLEMS 5.28
5.7 PRACTICE PROBLEM SOLUTIONS 5.35
5.8 ASSIGNMENT PROBLEMS 5.44

6. INPUT-OUTPUT EQUATIONS . . . . . . . . . . . . . . . . . . . . . . . . . 6.1

6.1 INTRODUCTION 6.1
6.2 THE DIFFERENTIAL OPERATOR 6.1
6.3 INPUT-OUTPUT EQUATIONS 6.4
 page 3

6.3.1 Converting Input-Output Equations to State Equations 6.6

6.3.2 Integrating Input-Output Equations 6.9
6.4 DESIGN CASE 6.11
6.5 SUMMARY 6.20
6.6 PRACTICE PROBLEMS 6.20
6.7 PRACTICE PROBLEM SOLUTIONS 6.22
6.8 ASSGINMENT PROBLEMS 6.26
6.9 REFERENCES 6.27

7. ELECTRICAL SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1

7.1 INTRODUCTION 7.1
7.2 MODELING 7.1
7.2.1 Resistors 7.2
7.2.2 Voltage and Current Sources 7.4
7.2.3 Capacitors 7.8
7.2.4 Inductors 7.10
7.2.5 Op-Amps 7.11
7.3 IMPEDANCE 7.16
7.4 EXAMPLE SYSTEMS 7.18
7.5 ELECTROMECHANICAL SYSTEMS - MOTORS 7.26
7.5.1 Permanent Magnet DC Motors 7.26
7.5.2 Induction Motors 7.28
7.5.3 Brushless Servo Motors 7.29
7.6 FILTERS 7.32
7.7 OTHER TOPICS 7.33
7.8 SUMMARY 7.33
7.9 PRACTICE PROBLEMS 7.34
7.10 PRACTICE PROBLEM SOLUTIONS 7.38
7.11 ASSIGNMENT PROBLEMS 7.43

8. FEEDBACK CONTROL SYSTEMS . . . . . . . . . . . . . . . . . . . . . . 8.1

8.1 INTRODUCTION 8.1
8.2 TRANSFER FUNCTIONS 8.1
8.3 CONTROL SYSTEMS 8.3
8.3.1 PID Control Systems 8.5
8.3.2 Manipulating Block Diagrams 8.7
8.3.3 A Motor Control System Example 8.12
8.3.4 System Error 8.17
8.3.5 Controller Transfer Functions 8.21
8.3.6 Feedforward Controllers 8.21
8.3.7 State Equation Based Systems 8.22
8.3.8 Cascade Controllers 8.24
8.4 SUMMARY 8.24
8.5 PRACTICE PROBLEMS 8.24
 page 4

8.6 PRACTICE PROBLEM SOLUTIONS 8.31

8.7 ASSIGNMENT PROBLEMS 8.37

9. PHASOR ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1

9.1 INTRODUCTION 9.1
9.2 PHASORS FOR STEADY-STATE ANALYSIS 9.1
9.3 VIBRATIONS 9.8
9.4 SUMMARY 9.10
9.5 PRACTICE PROBLEMS 9.10
9.6 PRACTICE PROBLEM SOLUTIONS 9.12
9.7 ASSIGNMENT PROBLEMS 9.14

10. BODE PLOTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.1

10.1 INTRODUCTION 10.1
10.2 BODE PLOTS 10.5
10.3 SIGNAL SPECTRUMS 10.21
10.4 SUMMARY 10.22
10.5 PRACTICE PROBLEMS 10.22
10.6 PRACTICE PROBLEM SOLUTIONS 10.25
10.7 ASSIGNMENT PROBLEMS 10.35
10.8 LOG SCALE GRAPH PAPER 10.36

11. ROOT LOCUS ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1

11.1 INTRODUCTION 11.1
11.2 ROOT-LOCUS ANALYSIS 11.1
11.3 SUMMARY 11.10
11.4 PRACTICE PROBLEMS 11.11
11.5 PRACTICE PROBLEM SOLUTIONS 11.14
11.6 ASSIGNMENT PROBLEMS 11.25

12. NONLINEAR SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.1

12.1 INTRODUCTION 12.1
12.2 SOURCES OF NONLINEARITY 12.1
12.2.1 Non-Linear Relationships 12.1
12.3 NON-LINEAR ELEMENTS 12.2
12.3.1 Time Variant 12.3
12.3.2 Switching 12.3
12.3.3 Deadband 12.4
12.3.4 Saturation and Clipping 12.7
12.3.5 Hysteresis and Slip 12.7
12.3.6 Delays and Lags 12.8
12.4 SUMMARY 12.9
12.5 PRACTICE PROBLEMS 12.9
 page 5

12.6 PRACTICE PROBLEM SOLUTIONS 12.9

12.7 ASIGNMENT PROBLEMS 12.9

13. ANALOG INPUTS AND OUTPUTS . . . . . . . . . . . . . . . . . . . . 13.1

13.1 INTRODUCTION 13.1
13.2 ANALOG INPUTS 13.3
13.3 ANALOG OUTPUTS 13.10
13.4 NOISE REDUCTION 13.12
13.4.1 Shielding 13.12
13.4.2 Grounding 13.14
13.5 CASE STUDY 13.15
13.6 SUMMARY 13.15
13.7 PRACTICE PROBLEMS 13.15
13.8 PRACTICE PROBLEM SOLUTIONS 13.15
13.9 ASSIGNMENT PROBLEMS 13.16

14. CONTINUOUS SENSORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.1

14.1 INTRODUCTION 14.1
14.2 INDUSTRIAL SENSORS 14.2
14.2.1 Angular Displacement 14.3
Potentiometers 14.3
14.2.2 Encoders 14.4
Tachometers 14.8
14.2.3 Linear Position 14.8
Potentiometers 14.8
Linear Variable Differential Transformers (LVDT)14.9
Moire Fringes 14.11
Accelerometers 14.12
14.2.4 Forces and Moments 14.15
Strain Gages 14.15
Piezoelectric 14.18
14.2.5 Liquids and Gases 14.20
Pressure 14.21
Venturi Valves 14.22
Coriolis Flow Meter 14.23
Magnetic Flow Meter 14.24
Ultrasonic Flow Meter 14.24
Vortex Flow Meter 14.24
Positive Displacement Meters 14.25
Pitot Tubes 14.25
14.2.6 Temperature 14.25
Resistive Temperature Detectors (RTDs) 14.26
Thermocouples 14.26
Thermistors 14.28
Other Sensors 14.30
 page 6

14.2.7 Light 14.30

Light Dependant Resistors (LDR) 14.30
14.2.8 Chemical 14.31
pH 14.31
Conductivity 14.31
14.2.9 Others 14.32
14.3 INPUT ISSUES 14.32
14.4 SENSOR GLOSSARY 14.37
14.5 SUMMARY 14.38
14.6 REFERENCES 14.39
14.7 PRACTICE PROBLEMS 14.39
14.8 PRACTICE PROBLEM SOLUTIONS 14.40
14.9 ASSIGNMENT PROBLEMS 14.42

15. CONTINUOUS ACTUATORS . . . . . . . . . . . . . . . . . . . . . . . . . 15.1

15.1 INTRODUCTION 15.1
15.2 ELECTRIC MOTORS 15.1
15.2.1 Basic Brushed DC Motors 15.3
15.2.2 AC Motors 15.7
15.2.3 Brushless DC Motors 15.15
15.2.4 Stepper Motors 15.17
15.2.5 Wound Field Motors 15.19
15.3 HYDRAULICS 15.23
15.4 OTHER SYSTEMS 15.24
15.5 SUMMARY 15.25
15.6 PRACTICE PROBLEMS 15.25
15.7 PRACTICE PROBLEM SOLUTIONS 15.26
15.8 ASSIGNMENT PROBLEMS 15.26

16. MOTION CONTROL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.1

16.1 INTRODUCTION 16.1
16.2 MOTION PROFILES 16.2
16.2.1 Velocity Profiles 16.2
16.2.2 Position Profiles 16.11
16.3 MULTI AXIS MOTION 16.14
16.3.1 Slew Motion 16.15
Interpolated Motion 16.16
16.3.2 Motion Scheduling 16.17
16.4 PATH PLANNING 16.19
16.5 CASE STUDIES 16.21
16.6 SUMMARY 16.23
16.7 PRACTICE PROBLEMS 16.23
16.8 PRACTICE PROBLEM SOLUTIONS 16.24
16.9 ASSIGNMENT PROBLEMS 16.25
 page 7

17. LAPLACE TRANSFORMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.1

17.1 INTRODUCTION 17.1
17.2 APPLYING LAPLACE TRANSFORMS 17.3
17.2.1 A Few Transform Tables 17.4
17.3 MODELING TRANSFER FUNCTIONS IN THE s-DOMAIN 17.9
17.4 FINDING OUTPUT EQUATIONS 17.11
17.5 INVERSE TRANSFORMS AND PARTIAL FRACTIONS 17.14
17.6 EXAMPLES 17.21
17.6.1 Mass-Spring-Damper Vibration 17.21
17.6.2 Circuits 17.23
17.7 ADVANCED TOPICS 17.25
17.7.1 Input Functions 17.25
17.7.2 Initial and Final Value Theorems 17.26
17.8 A MAP OF TECHNIQUES FOR LAPLACE ANALYSIS 17.27
17.9 SUMMARY 17.28
17.10 PRACTICE PROBLEMS 17.28
17.11 PRACTICE PROBLEM SOLUTIONS 17.31
17.12 ASSIGNMENT PROBLEMS 17.35
17.13 REFERENCES 17.37

18. CONTROL SYSTEM ANALYSIS . . . . . . . . . . . . . . . . . . . . . . 18.1

18.1 INTRODUCTION 18.1
18.2 CONTROL SYSTEMS 18.1
18.2.1 PID Control Systems 18.3
18.2.2 Analysis of PID Controlled Systems With Laplace Transforms
18.5
18.2.3 Finding The System Response To An Input 18.8
18.2.4 Controller Transfer Functions 18.13
18.3 ROOT-LOCUS PLOTS 18.13
18.3.1 Approximate Plotting Techniques 18.17
18.4 DESIGN OF CONTINUOUS CONTROLLERS 18.21
18.5 SUMMARY 18.21
18.6 PRACTICE PROBLEMS 18.22
18.7 PRACTICE PROBLEM SOLUTIONS 18.27
18.8 ASSIGNMENT PROBLEMS 18.27

19. CONVOLUTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.1

19.1 INTRODUCTION 19.1
19.2 UNIT IMPULSE FUNCTIONS 19.1
19.3 IMPULSE RESPONSE 19.3
19.4 CONVOLUTION 19.5
19.5 NUMERICAL CONVOLUTION 19.6
19.6 LAPLACE IMPULSE FUNCTIONS 19.9
 page 8

19.7 SUMMARY 19.10

19.8 PRACTICE PROBLEMS 19.10
19.9 PRACTICE PROBLEM SOLUTIONS 19.10
19.10 ASSIGNMENT PROBLEMS 19.10

20. STATE SPACE ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.1

20.1 INTRODUCTION 20.1
20.2 OBSERVABILITY 20.13
20.3 CONTROLLABILITY 20.15
20.4 OBSERVERS 20.18
20.5 SUMMARY 20.18
20.6 PRACTICE PROBLEMS 20.19
20.7 PRACTICE PROBLEM SOLUTIONS 20.19
20.8 ASSIGNMENT PROBLEMS 20.19
20.9 BIBLIOGRAPHY 20.19

21. STATE SPACE CONTROLLERS . . . . . . . . . . . . . . . . . . . . . . . 21.1

21.1 INTRODUCTION 21.1
21.2 FULL STATE FEEDBACK 21.2
21.3 OBSERVERS 21.5
21.4 SUPPLEMENTAL OBSERVERS 21.11
21.5 REGULATED CONTROL WITH OBSERVERS 21.11
21.6 LQR 21.22
21.7 LINEAR QUADRATIC GAUSSIAN (LQG) COMPENSATORS 21.24
21.8 VERIFYING CONTROL SYSTEM STABILITY 21.24
21.8.1 Stability 21.25
21.8.2 Bounded Gain 21.26
21.9 ADAPTIVE CONTROLLERS 21.28
21.10 OTHER METHODS 21.31
21.10.1 Kalman Filtering 21.32
21.11 SUMMARY 21.32
21.12 PRACTICE PROBLEMS 21.33
21.13 PRACTICE PROBLEM SOLUTIONS 21.33
21.14 ASSIGNMENT PROBLEMS 21.33

22. SYSTEM IDENTIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1

22.1 INTRODUCTION 22.1
22.2 SUMMARY 22.10
22.3 PRACTICE PROBLEMS 22.10
22.4 PRACTICE PROBLEM SOLUTIONS 22.10
22.5 ASSIGNMENT PROBLEMS 22.10

23. ELECTROMECHANICAL SYSTEMS . . . . . . . . . . . . . . . . . . . 23.1

 page 9

23.1 INTRODUCTION 23.1

23.2 MATHEMATICAL PROPERTIES 23.1
23.2.1 Induction 23.1
23.3 EXAMPLE SYSTEMS 23.9
23.4 SUMMARY 23.16
23.5 PRACTICE PROBLEMS 23.16
23.6 PRACTICE PROBLEM SOLUTIONS 23.16
23.7 ASSIGNMENT PROBLEMS 23.16

24. FLUID SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.1

24.1 SUMMARY 24.1
24.2 MATHEMATICAL PROPERTIES 24.1
24.2.1 Resistance 24.2
24.2.2 Capacitance 24.4
24.2.3 Power Sources 24.6
24.3 EXAMPLE SYSTEMS 24.8
24.4 SUMMARY 24.10
24.5 PRACTICE PROBLEMS 24.10
24.6 PRACTICE PROBLEMS SOLUTIONS 24.10
24.7 ASSIGNMENT PROBLEMS 24.10

25. THERMAL SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.1

25.1 INTRODUCTION 25.1
25.2 MATHEMATICAL PROPERTIES 25.1
25.2.1 Resistance 25.1
25.2.2 Capacitance 25.3
25.2.3 Sources 25.4
25.3 EXAMPLE SYSTEMS 25.4
25.4 SUMMARY 25.7
25.5 PRACTICE PROBLEMS 25.7
25.6 PRACTICE PROBLEM SOLUTIONS 25.7
25.7 ASSIGNMENT PROBLEMS 25.7

26. OPTIMIZATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26.1

26.1 INTRODUCTION 26.1
26.2 OBJECTIVES AND CONSTRAINTS 26.2
26.3 SEARCHING FOR THE OPTIMUM 26.6
26.4 OPTIMIZATION ALGORITHMS 26.9
26.4.1 Random Walk 26.9
26.4.2 Gradient Decent 26.10
26.4.3 Simplex 26.10
26.5 SUMMARY 26.10
26.6 PRACTICE PROBLEMS 26.10
26.7 PRACTICE PROBLEM SOLUTIONS 26.10
 page 10

26.8 ASSIGNMENT PROBLEMS 26.10

27. FINITE ELEMENT ANALYSIS (FEA) . . . . . . . . . . . . . . . . . . . 27.1

27.1 INTRODUCTION 27.1
27.2 FINITE ELEMENT MODELS 27.2
27.3 FINITE ELEMENT MODELS 27.4
27.4 SUMMARY 27.12
27.5 PRACTICE PROBLEMS 27.13
27.6 PRACTICE PROBLEM SOLUTIONS 27.13
27.7 ASSIGNMENT PROBLEMS 27.13
27.8 BIBLIOGRAPHY 27.13

28. FUZZY LOGIC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28.1

28.1 INTRODUCTION 28.1
28.2 COMMERCIAL CONTROLLERS 28.7
28.3 REFERENCES 28.7
28.4 SUMMARY 28.7
28.5 PRACTICE PROBLEMS 28.8
28.6 PRACTICE PROBLEM SOLUTIONS 28.8
28.7 ASSIGNMENT PROBLEMS 28.8

29. NEURAL NETWORKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29.1

29.1 SUMMARY 29.9
29.2 PRACTICE PROBLEMS 29.10
29.3 PRACTICE PROBLEM SOLUTIONS 29.10
29.4 ASSIGNMENT PROBLEMS 29.10
29.5 REFERENCES 29.10

30. EMBEDDED CONTROL SYSTEM . . . . . . . . . . . . . . . . . . . . . 30.1

30.1 INTRODUCTION 30.1
30.2 CASE STUDY 30.3
30.3 SUMMARY 30.3
30.4 PRACTICE PROBLEMS 30.3
30.5 PRACTICE PROBLEM SOLUTIONS 30.3
30.6 ASSIGNMENT PROBLEMS 30.3

31. WRITING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.1

31.1 FORGET WHAT YOU WERE TAUGHT BEFORE 31.1
31.2 WHY WRITE REPORTS? 31.2
31.3 THE TECHNICAL DEPTH OF THE REPORT 31.3
31.4 TYPES OF REPORTS 31.3
31.5 LABORATORY REPORTS 31.3
An Example First Draft of a Report 31.5
 page 11

An Example Final Draft of a Report 31.11

31.6 RESEARCH 31.11
31.7 DRAFT REPORTS 31.11
31.8 PROJECT REPORT 31.12
31.9 OTHER REPORT TYPES 31.13
31.9.1 Executive 31.13
31.9.2 Consulting 31.13
31.9.3 Memo(randum) 31.13
31.9.4 Interim 31.14
31.9.5 Poster 31.14
31.9.6 Progress Report 31.14
31.9.7 Oral 31.15
31.9.8 Patent 31.15
31.10 LAB BOOKS 31.16
31.11 REPORT ELEMENTS 31.16
31.11.1 Figures 31.17
31.11.2 Graphs 31.18
31.11.3 Tables 31.19
31.11.4 Equations 31.19
31.11.5 Experimental Data 31.20
31.11.6 Result Summary 31.21
31.11.7 References 31.21
31.11.8 Acknowledgments 31.21
31.11.9 Abstracts 31.22
31.11.10 Appendices 31.22
31.11.11 Page Numbering 31.23
31.11.12 Numbers and Units 31.23
31.11.13 Engineering Drawings 31.23
31.11.14 Discussions 31.24
31.11.15 Conclusions 31.24
31.11.16 Recomendations 31.24
31.11.17 Appendices 31.25
31.11.18 Units 31.25
31.12 GENERAL WRITING ISSUES 31.25
31.13 WRITERS BLOCK 31.26
31.14 TECHNICAL ENGLISH 31.26
31.15 EVALUATION FORMS 31.29
31.16 PATENTS 31.31

32. PROJECTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1

32.1 32.1
32.2 OVERVIEW 32.1
32.2.1 The Objectives and Constraints 32.2
32.3 MANAGEMENT 32.3
32.3.1 Timeline - Tentative 32.3
 page 12

32.3.2 Teams 32.4

32.4 DELIVERABLES 32.5
32.4.1 Conceptual Design 32.5
32.4.2 EGR 345/101 Contract 32.5
32.4.3 Progress Reports 32.6
32.4.4 Design Proposal 32.6
32.4.5 The Final Report 32.7
32.5 REPORT ELEMENTS 32.8
32.5.1 Gantt Charts 32.8
32.5.2 Drawings 32.9
32.5.3 Budgets and Bills of Material 32.9
32.5.4 Calculations 32.10
32.6 APPENDICES 32.10
32.6.1 Appendix A - Sample System 32.10
32.6.2 Appendix B - EGR 345/101 Contract 32.18
32.6.3 Appendix C - Forms 32.19

33. ENGINEERING PROBLEM SOLVING . . . . . . . . . . . . . . . . . . 33.1

33.1 BASIC RULES OF STYLE 33.1
33.2 EXPECTED ELEMENTS 33.1
33.3 SEPCIAL ELEMENTS 33.2
33.3.1 Graphs 33.2
33.3.2 EGR 345 Specific 33.2
33.4 SCILAB 33.2
33.5 TERMINOLOGY 33.3

34. MATHEMATICAL TOOLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34.1

34.1 INTRODUCTION 34.1
34.1.1 Constants and Other Stuff 34.2
34.1.2 Basic Operations 34.3
Factorial 34.4
34.1.3 Exponents and Logarithms 34.4
34.1.4 Polynomial Expansions 34.5
34.1.5 Practice Problems 34.6
34.2 FUNCTIONS 34.9
34.2.1 Discrete and Continuous Probability Distributions 34.9
34.2.2 Basic Polynomials 34.9
34.2.3 Partial Fractions 34.11
34.2.4 Summation and Series 34.14
34.2.5 Practice Problems 34.16
34.3 SPATIAL RELATIONSHIPS 34.17
34.3.1 Trigonometry 34.17
34.3.2 Hyperbolic Functions 34.22
Practice Problems 34.23
34.3.3 Geometry 34.24
 page 13

34.3.4 Planes, Lines, etc. 34.41

34.3.5 Practice Problems 34.43
34.4 COORDINATE SYSTEMS 34.45
34.4.1 Complex Numbers 34.45
34.4.2 Cylindrical Coordinates 34.48
34.4.3 Spherical Coordinates 34.49
34.4.4 Practice Problems 34.50
34.5 MATRICES AND VECTORS 34.51
34.5.1 Vectors 34.51
34.5.2 Dot (Scalar) Product 34.52
34.5.3 Cross Product 34.57
34.5.4 Triple Product 34.59
34.5.5 Matrices 34.59
34.5.6 Solving Linear Equations with Matrices 34.64
34.5.7 Practice Problems 34.65
34.6 CALCULUS 34.70
34.6.1 Single Variable Functions 34.70
Differentiation 34.70
Integration 34.73
34.6.2 Vector Calculus 34.77
34.6.3 Differential Equations 34.79
First-order Differential Equations 34.80
Guessing 34.81
Separable Equations 34.81
Homogeneous Equations and Substitution 34.82
Second-order Differential Equations 34.83
Linear Homogeneous 34.83
Nonhomogeneous Linear Equations 34.84
Higher Order Differential Equations 34.86
Partial Differential Equations 34.86
34.6.4 Other Calculus Stuff 34.87
34.6.5 Practice Problems 34.87
34.7 NUMERICAL METHODS 34.93
34.7.1 Approximation of Integrals and Derivatives from Sampled Data
34.93
34.7.2 Euler First-order Integration 34.94
34.7.3 Taylor Series Integration 34.94
34.7.4 Runge-Kutta Integration 34.95
34.7.5 Newton-Raphson to Find Roots 34.95
34.8 LAPLACE TRANSFORMS 34.96
34.8.1 Laplace Transform Tables 34.96
34.9 z-TRANSFORMS 34.99
34.10 FOURIER SERIES 34.102
34.11 TOPICS NOT COVERED (YET) 34.102
34.12 REFERENCES/BIBLIOGRAPHY 34.103
 page 14

35. A BASIC INTRODUCTION TO ‘C’ . . . . . . . . . . . . . . . . . . . . . 35.1

35.1 WHY USE ‘C’? 35.1
35.2 BACKGROUND 35.2
35.3 PROGRAM PARTS 35.2
35.4 HOW A ‘C’ COMPILER WORKS 35.11
35.5 STRUCTURED ‘C’ CODE 35.13
35.6 ARCHITECTURE OF ‘C’ PROGRAMS (TOP-DOWN) 35.14
35.6.1 How? 35.14
35.6.2 Why? 35.15
35.7 CREATING TOP DOWN PROGRAMS 35.16
35.8 HOW THE BEAMCAD PROGRAM WAS DESIGNED 35.17
35.8.1 Objectives: 35.18
35.8.2 Problem Definition: 35.18
35.8.3 User Interface: 35.18
Screen Layout (also see figure): 35.18
Input: 35.19
Output: 35.20
Help: 35.20
Error Checking: 35.20
Miscellaneous: 35.21
35.8.4 Flow Program: 35.22
35.8.5 Expand Program: 35.22
35.8.6 Testing and Debugging: 35.24
35.8.7 Documentation 35.25
Users Manual: 35.25
Programmers Manual: 35.26
35.8.8 Listing of BeamCAD Program. 35.26
35.9 PRACTICE PROBLEMS 35.26

36. UNITS AND CONVERSIONS . . . . . . . . . . . . . . . . . . . . . . . . . 36.1

36.1 HOW TO USE UNITS 36.1
36.2 HOW TO USE SI UNITS 36.2
36.3 THE TABLE 36.2
36.4 ASCII, HEX, BINARY CONVERSION 36.6
36.5 G-CODES 36.8

37. ATOMIC MATERIAL DATA . . . . . . . . . . . . . . . . . . . . . . . . . . 37.1

37. MECHANICAL MATERIAL PROPERTIES . . . . . . . . . . . . . . 37.1
37.1 FORMULA SHEET 37.4

38. BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.1

38.1 TEXTBOOKS 38.1
38.1.1 Slotine and Li 38.1
 page 15

38.1.2 VandeVegte 38.1

39. TOPICS IN DEVELOPMENT . . . . . . . . . . . . . . . . . . . . . . . . . . 39.1

39.1 UPDATED DC MOTOR MODEL 39.1
39.2 ANOTHER DC MOTOR MODEL 39.4
39.3 BLOCK DIAGRAMS AND UNITS 39.8
39.4 SIGNAL FLOW GRAPHS 39.9
39.5 ZERO ORDER HOLD 39.9
39.6 TORSIONAL DAMPERS 39.9
39.7 MISC 39.10
39.8 Nyquist Plot 39.10
39.9 NICHOLS CHART 39.12
39.10 BESSEL POLYNOMIALS 39.14
39.11 ITAE 39.15
39.12 ROOT LOCUS 39.16
39.13 LYAPUNOV’S LINEARIZATION METHOD 39.16
39.14 XXXXX 39.17
39.15 XXXXX 39.18
39.16 XXXXX 39.18
39.17 XXXXX 39.18
39.18 XXXXX 39.18
39.19 XXXXX 39.18
39.20 XXXXX 39.18
39.21 SUMMARY 39.18
39.22 PRACTICE PROBLEMS 39.18
39.23 PRACTICE PROBLEM SOLUTIONS 39.19
39.24 ASSGINMENT PROBLEMS 39.19
39.25 REFERENCES 39.19
39.26 BIBLIOGRAPHY 39.19
 introduction - 1.1

1. INTRODUCTION

Topics:

Objectives:

1.1 BASIC TERMINOLOGY

• Modeling

• Explicit solutions

• Numerical solutions

• Empirical data

• Simulation

• Lumped parameter (masses and springs)

• Distributed parameters (stress in a solid)

• Continuous vs. Discrete

• Linear vs. Non-linear

 introduction - 1.2

• linearity and superposition

• reversibility

• through and across variables

• Analog vs. Digital

• process vs. controllers

• Basic system categories below,

Static

Dynamic Stochastic
Deterministic Distributed
Lumped Non-linear
Linear Continuous
Discrete

Figure 1.1 Model Classifications

 introduction - 1.3

• control system types: servo vs. regulating/process control

• open loop vs. closed loop

• disturbances

• component variations

• system error

• analysis vs. design

• mechatronics

• embedded systems

• real-time systems

1.2 EXAMPLE SYSTEM

• Servo control systems

• Robot

1.3 SUMMARY

1.4 PRACTICE PROBLEMS

1.
 translation - 2.1

2. TRANSLATION

Topics:
• Basic laws of motion
• Gravity, inertia, springs, dampers, cables and pulleys, drag, friction, FBDs
• System analysis techniques
• Design case

Objectives:
• To be able to develop differential equations that describe translating systems.

2.1 INTRODUCTION

If the velocity and acceleration of a body are both zero then the body will be static.
If the applied forces are balanced, and cancel each other out, the body will not accelerate.
If the forces are unbalanced then the body will accelerate. If all of the forces act through
the center of mass then the body will only translate. Forces that do not act through the cen-
ter of mass will also cause rotation to occur. This chapter will focus only on translational
systems.

The equations of motion for translating bodies are shown in Figure 2.1. These state
simply that velocity is the first derivative of position, and velocity is the first derivative of
acceleration. Conversely the acceleration can be integrated to find velocity, and the veloc-
ity can be integrated to find position. Therefore, if we know the acceleration of a body, we
can determine the velocity and position. Finally, when a force is applied to a mass, an
acceleration can be found by dividing the net force by the mass.
 translation - 2.2

x,v,a equations of motion

v ( t ) =  ----- x ( t )
d (1)
 dt
F d 2
a ( t ) =  ----- x ( t ) =  ----- v ( t )
d
(2)
dt dt
OR
x ( t ) = ∫ v ( t ) dt = ∫ ∫ a ( t ) dt (3)

v(t) = ∫ a ( t ) dt (4)

a(t) = F ( t )-
--------- (5)
M
where,
x, v, a = position, velocity and acceleration
M = mass of the body
F = an applied force

Figure 2.1 Velocity and acceleration of a translating mass

An example application of these fundamental laws is shown in Figure 2.2. The ini-
tial conditions of the system are supplied (and are normally required to solve this type of
problem). These are then used to find the state of the system after a period of time. The
solution begins by integrating the acceleration, and using the initial velocity value for the
integration constant. So at t=0 the velocity will be equal to the initial velocity. This is then
integrated once more to provide the position of the object. As before, the initial position is
used for the integration constant. This equation is then used to calculate the position after
a period of time. Notice that the units are used throughout the calculations, this is a good
practice for any engineer.
 translation - 2.3

Given an initial (t=0) state of x=5m, v=2m/s, a=3ms-2, find the system state 5 seconds
later assuming constant acceleration.
The initial conditions for the system at time t=0 are,
x 0 = 5m
–1 Note: units are very important and should nor-
v 0 = 2ms mally be carried through all calculations.
–2
a 0 = 3ms

The constant acceleration can be integrated to find the velocity as a function of time.
Note:
v ( t ) = ∫ a 0 dt = a 0 t + C = a 0 t + v 0 (6)
v ( t ) = a0 t + C
v0 = a0 ( 0 ) + C
v0 = C
Next, the velocity can be integrated to find the position as a function of time.
a 2
x ( t ) = ∫ v ( t ) dt = ∫ ( a 0 t + v 0 ) dt = -----0- t + v 0 t + x 0 (7)
2
This can then be used to calculate the position of the mass after 5 seconds.
a 2
x ( 5 ) = -----0- t + v 0 t + x 0
2
–2 2 –1
3ms
= ---------------- ( 5s ) + 2ms ( 5s ) + 5m
2
= 37.5m + 10m + 5m = 52.5m

Figure 2.2 Sample calculation for a translating mass, with initial conditions

2.2 MODELING

When modeling translational systems it is common to break the system into parts.
These parts are then described with Free Body Diagrams (FBDs). Common components
that must be considered when constructing FBDs are listed below, and are discussed in
following sections.

• gravity and other fields - apply non-contact forces

• inertia - opposes acceleration and deceleration
• springs - resist deflection
• dampers and drag - resist motion
• friction - opposes relative motion between bodies in contact
 translation - 2.4

• cables and pulleys - redirect forces

• contact points/joints - transmit forces through up to 3 degrees of freedom

2.2.1 Free Body Diagrams

Free Body Diagrams (FBDs) allow us to reduce a complex mechanical system into
smaller, more manageable pieces. The forces applied to the FBD can then be summed to
provide an equation for the piece. These equations can then be used later to do an analysis
of system behavior. These are required elements for any engineering problem involving
rigid bodies.

An example of FBD construction is shown in Figure 2.3. In this case there is a

mass sitting atop a spring. An FBD can be drawn for the mass. In total there are two obvi-
ous forces applied to the mass, gravity pulling the mass downward, and a spring pushing
the mass upwards. The FBD for the spring has two forces applied at either end. Notice that
the spring force, FR1, acting on the mass, and on the spring have an equal magnitude, but
opposite direction.

FBD Mass: FBD Spring:

M = 10 kg

K = 20 N/m
FR1

FR1
Mg
FR2

Figure 2.3 Free body diagram example

2.2.2 Mass and Inertia

In a static system the sum of forces is zero and nothing is in motion. In a dynamic
system the sum of forces is not zero and the masses accelerate. The resulting imbalance in
forces acts on the mass causing it to accelerate. For the purposes of calculation we create a
virtual reaction force, called the inertial force. This force is also known as D’Alembert’s
(pronounced as daa-lamb-bears) force. It can be included in calculations in one of two
ways. The first is to add the inertial force to the FBD and then add it into the sum of
 translation - 2.5

forces, which will equal zero. The second method is known as D’Alembert’s equation
where all of the forces are summed and set equal to the inertial force, as shown in Figure
2.4. The acceleration is proportional to the inertial force and inversely proportional to the
mass.

∑ F = Ma (Newton’s) (11)

∑ F – Ma = 0 (D’Alembert’s) (12)

Figure 2.4 D’Alembert’s and Newton’s equations

An application of Newton’s equation to FBDs can be seen in Figure 2.5. In the first
case an inertial force is added to the FBD. This force should be in an opposite direction
(left here) to the positive direction of the mass (right). When the sum of forces equation is
used then the force is added in as a normal force component. In the second case Newton’s
equation is used so the force is left off the FBD, but added to the final equation. In this
case the sign of the inertial force is positive if the assumed positive direction of the mass
matches the positive direction for the summation.
 translation - 2.6

D’Alemberts’s form: x

d 2 d- 2
Ma = M  ----- x ∑ x
+F = F – M  ----
dt F  dt x = 0
M or
d- 2
 ----
+
∑ x
F = – F + M  dt x = 0

Note: If using an inertial force then the direction of the force should be opposite to
the positive motion direction for the mass.

Newton’s form: x

d- 2
 ----
F + ∑ x
F = F = M  dt x
M
or
d- 2
 ----
+ ∑ x
F = – F = – M  dt x

Note: If using Newton’s form the sign of the inertial force should be positive if the
positive direction for the summation and the mass are the same, otherwise if they
are opposite then the sign should be negative.

Figure 2.5 Free body diagram and inertial forces

An example of the application of Newton’s equation is shown in Figure 2.6. In this

example there are two unbalanced forces applied to a mass. These forces are summed and
set equal to the inertial force. Solving the resulting equation results in acceleration values
in the ’x’ and ’y’ directions. In this example the forces and calculations are done in vector
form for convenience and brevity.
 translation - 2.7

5
F 1 = –4 N –7
F2 = –3 N
0
0
If both forces shown act through
the center of mass, what is the
acceleration of the ball? M=10kg

∑F = F 1 + F 2 = Ma

5 –7
–4 N + – 3 N = ( 10Kg )a
0 0

–2 – 0.2 – 0.2
 1   Kgm
∴a =  ------------- – 7 N = – 0.7  ----------- 1 m
- ------- = – 0.7 ----
10Kg 2  Kg 2
s s
0 0 0

Figure 2.6 Sample acceleration calculation

// A program to sum forces and calculate the acceleration

// define the given forces and mass

F1 = [5, -4, 0];
F2 = [-7, -3, 0];
M = 10;

function foo=Sum() // The sum of the applied forces

foo = F1 + F2;
endfunction

A = Sum() / M;
printf("The acceleration is ( %f, %f, %f) m/s^2 \n", A(1), A(2), A(3));
printf("The magnitude is |A| = %f m/s^2 \n", norm(A));

Figure 2.7 A Scilab calculation example

 translation - 2.8

2.2.3 Gravity and Other Fields

Gravity is a weak force of attraction between masses. In our situation we are in the
proximity of a large mass (the earth) which produces a large force of attraction. When ana-
lyzing forces acting on rigid bodies we add this force to our FBDs. The magnitude of the
force is proportional to the mass of the object, with a direction toward the center of the
earth (down).

The relationship between mass and force is clear in the metric system with mass
having the units Kilograms (kg), and force the units Newtons (N). The relationship
between these is the gravitational constant 9.81N/kg, which will vary slightly over the sur-
face of the earth. The Imperial units for force and mass are both the pound (lb.) which
often causes confusion. To reduce this confusion the unit for force is normally modified to
be, lbf.

An example calculation including gravitational acceleration is shown in Figure

2.8. The 5kg mass is pulled by two forces, gravity and the arbitrary force ’F’. These forces
are described in vector form, with the positive ’z’ axis pointing upwards. To find the equa-
tions of motion the forces are summed. To eliminate the second derivative on the inertia
term the equation is integrated twice. The result is a set of three vector equations that
describe the x, y and z components of the motion. Notice that the units have been carried
through these calculations.
 translation - 2.9

Assume we have a mass that is acted upon by gravity and F M = 5Kg

a second constant force vector. To find the position of
the mass as a function of time we first define the grav- Fg g
ity vector and position components for the system.

fx
0 0 x(t)
N- = m-
g = 0 Kg------ 0 ---- X( t) = y(t) F = f F g = Mg
2 y
– 9.81 – 9.81 s z( t) fz
Next, sum the forces and set them equal to inertial resistance.
d- 2
 ----
∑ F = Mg + F = M  dt X ( t )
fx
0 x(t)
d 2
+ f = 5Kg  ----- y ( t )
m-
5Kg 0 ----
2 y dt
– 9.81 s fz z(t)
fx
0 x( t)
m –1  d- 2
----
0 ----2- + 0.2Kg fy =   y ( t )
dt
– 9.81 s fz z(t)
Integrate twice to find the position components. Note: When an engineer
vx solves a problem they
fx 0 will always be looking
0 x(t)
m –1  d 
f y t + v y =  ----- y ( t ) at the equations and
0 ----2- + 0.2Kg 0 dt unknowns. In this case
– 9.81 s fz vz z ( t ) there are three equa-
0 tions, and there are 9
vx
fx 0 x0 constants/givens fx, fy,
0 x ( t ) fz, vx0, vy0, xz0, x0, y0
1--- m- + 0.2Kg – 1 2
0 ---- fy t + vy t + y0 = y ( t ) and z0. There are 4
2 2 0
– 9.81 s z(t) variables/unknowns x,
fz vz z0
2 y, z and t. Therefore
0.1f x t + v x t + x 0 0 with 3 equations and 4
0 unknowns only one
x(t)
2 value (4-3) is required
y(t) = 0.1f y t + v y t + y 0 m
0 to find all of the
z(t ) unknown values.
 –------------
9.81- + 0.1f  t 2 + v t + z
 2 z z0 0

Figure 2.8 Gravity vector calculations

 translation - 2.10

Like gravity, magnetic and electrostatic fields can also apply forces to objects.
Magnetic forces are commonly found in motors and other electrical actuators. Electro-
static forces are less common, but may need to be considered for highly charged systems.

Given,
FBD:
3 0 F1
F1 = 4 N N-
g = – 9.81 ------ M = 2Kg
Kg
0 0 M

Find the acceleration. g

ans.

1.5
a = – 7.81 m
----
s
0

Figure 2.9 Drill problem: find the acceleration of the FBD

2.2.4 Springs

Springs are typically constructed with elastic materials such as metals and plastics,
that will provide an opposing force when deformed. The properties of the spring are deter-
mined by the Young’s modulus (E) of the material and the geometry of the spring. A prim-
itive spring is shown in Figure 2.10. In this case the spring is a solid member. The
relationship between force and displacement is determined by the basic mechanics of
materials relationship. In practice springs are more complex, but the parameters (E, A and
L) are combined into a more convenient form. This form is known as Hooke’s Law.
 translation - 2.11

F
δ =  ------- F
L
EA
L
F =  ------- δ
EA
 L δ

F = Kδ (Hooke’s law)

Figure 2.10 A solid member as a spring

Hooke’s law does have some limitations that engineers must consider. The basic
equation is linear, but as a spring is deformed the material approaches plastic deformation,
and the modulus of elasticity will change. In addition the geometry of the object changes,
also changing the effective stiffness. Springs are normally assumed to be massless. This
allows the inertial effects to be ignored, such as a force propagation delay. In applications
with fast rates of change the spring mass may become significant, and they will no longer
act as an ideal device.

The cases for tension and compression are shown in Figure 2.11. In the case of
compression the spring length has been made shorter than its’ normal length. This requires
that a compression force be applied. For tension, both the displacement from neutral and
the required force reverse direction. It is advisable when solving problems to assume a
spring is either in tension or compression, and then select the displacement and force
directions accordingly.
 translation - 2.12

∆ x = deformed length ASIDE: a spring has a natural or undeformed length. When

at this length it is neither in tension or compression

Fc ∆x
compression as positive:
Ks Fc = Ks ∆ x

Fc ∆x

Ks F c = –K s ∆ x

Fc ∆x

Ks F c = –K s ∆ x

Ks
Ft
tension as positive:
Ft = Ks ∆ x
∆x
Ks
Ft
Ft = –Ks ∆ x
∆x
Ks
Ft
Ft = –Ks ∆ x
∆x
NOTE: the symbols for springs, resistors and inductors are quite often the same or simi-
lar. You will need to remember this when dealing with complex systems - and espe-
cially in this book where we deal with both types of components.

Figure 2.11 Sign conventions for spring forces and displacements

Previous examples have shown springs with displacements at one end only. In Fig-
ure 2.12, springs are shown that have movement at both ends. In these cases the sign of the
force applied to the spring is selected with reference to the assumed compression or ten-
sion. The primary difference is that care is required to correctly construct the expressions
for the tension or compression forces. In all cases the forces on the springs must be
assumed and drawn as either tensile or compressive. In the first example the displacement
and forces are tensile. The displacement at the left is tensile, so it will be positive, but on
the right hand side the displacement is compressive so it is negative. In the second exam-
 translation - 2.13

ple the force and both displacements are shown as tensile, so the terms are both positive.
In the third example the force is shown as compressive, while the displacements are both
shown as tensile, so both terms are negative.

∆x1 ∆ x2
F Ks F
F = Ks ( ∆ x1 – ∆ x 2 )

∆x1 ∆ x2
F Ks F
F = Ks ( ∆ x 1 + ∆ x 2 )

∆x1 ∆ x2
F Ks F
F = Ks ( – ∆ x1 – ∆ x2 )

Aside: it is useful to assume that the spring is

either in tension or compression, and then
make all decisions based on that assumption.

Figure 2.12 Examples of forces when both sides of a spring can move

The example in Figure 2.13 shows two masses separated by a spring. In the first
example the spring is assumed to be in tension. When x1 becomes positive it will put the
spring in compression, so it is made negative, however a positive x2 will put the spring in
tension, so it remains positive. In the second case the spring is assumed to be in compres-
sion and this time x2 will put it in tension so it is made negative.
 translation - 2.14

Consider the two masses below separated by a spring.

Ks
M1 M2

x1 x2

The system can be reduced to free body diagrams assuming the spring is tension.

M1 Ks ( – x1 + x2 ) M2

x1 x2

Note: In this example the spring is assumed to be in tension and the signs of the
magnitude are made negative for terms that result in compression for posi-
tive changes.

The system can be reduced to free body diagrams assuming the spring is in compression.

M1 Ks ( x1 – x2 ) M2

x1 x2

Figure 2.13 Drawing FBDs with interconnecting springs

Sometimes the true length of a spring is important, and the deformation alone is
insufficient. In these cases the deformation can be defined as a deformed and undeformed
length, as shown in Figure 2.14.
 translation - 2.15

∆ x = l1 – l 0
where,
∆ x = deformation
l 0 = the length when undeformed
l 1 = the length when deformed

Figure 2.14 Using the actual spring length

In addition to providing forces, springs may be used as energy storage devices.

Figure 2.15 shows the equation for energy stored in a spring.

2
K ( ∆x )
E P = ------------------
2

Figure 2.15 Energy stored in a spring

 translation - 2.16

Given,
N- Ks Ks
K s = 10 ---
m
∆ x = 0.1m
F
∆x
Find F assuming the springs are
normally 4m long when unloaded 10m

Aside: it can help to draw a

FBD of the pin.

ans. F=2N

Figure 2.16 Drill problem: Deformation of a two spring system

 translation - 2.17

Draw the FBDs and the equations of motion for the masses
Ks
F
M1 M2

x1 x2

ans.
··
x1 ( M1 ) + x1 ( Ks ) + x2 ( –Ks ) = F
··
x 1 ( K s ) + x2 ( –M2 ) + x2 ( – K s ) = F

Figure 2.17 Drill problem: Draw the FBDs for the masses
 translation - 2.18

2.2.5 Damping and Drag

A damper is a component that resists motion. The resistive force is relative to the
rate of displacement. As mentioned before, springs store energy in a system but dampers
dissipate energy. Dampers and springs are often used to compliment each other in designs.

Damping can occur naturally in a system, or can be added by design. The physical
damper pictured in Figure 2.18 uses a cylinder that contains a fluid. There is a moving rod
and piston that can slide within the cylinder. As the piston moves, fluid is forced through a
small orifice. When moved slowly the fluid moves easily, but when moved quickly the
pressure required to force the fluid through the orifice rises. This rise in pressure results in
a higher force of resistance. In ideal terms any motion would result in an opposing force.
In reality there is also a break-away force that needs to be applied before motion begins.
Other manufacturing variations could also lead to other small differences between damp-
ers. Normally these cause negligible effects.

orifice

motion
fluid
piston fluid

Figure 2.18 A physical damper

The basic equation for an ideal damper in compression is shown in Figure 2.19. In
this case the force and displacement are both compressive. The force is calculated by mul-
tiplying the damping coefficient by the velocity, or first derivative of position. Aside from
the use of the first derivative of position, the analysis of dampers in systems is similar to
that of springs.
 translation - 2.19

x
Kd
F = K d  ----- x
d (15)
F dt

Aside: The symbol shown is typically used for dampers. It is based on an old damper
design called a dashpot. It was constructed using a small piston inside a larger
pot filled with oil.

Figure 2.19 An ideal damper

Damping can also occur when there is relative motion between two objects. If the
objects are lubricated with a viscous fluid (e.g., oil) then there will be a damping effect. In
the example in Figure 2.20 two objects are shown with viscous friction (damping)
between them. When the system is broken into free body diagrams the forces are shown to
be a function of the relative velocities between the blocks.

x1
· ·
Fd = Kd ( x1 – x2 )

Kd
· ·
Fd = Kd ( x1 – x2 )

x2

Aside: Fluids, such as oils, have a significant viscosity. When these materials are put
in shear they resist the motion. The higher the shear rate, the greater the resis-
tance to flow. Normally these forces are small, except at high velocities.

Figure 2.20 Viscous damping between two bodies with relative motion

A damping force is proportional to the first derivative of position (velocity). Aero-

dynamic drag is proportional to the velocity squared. The equation for drag is shown in
 translation - 2.20

Figure 2.21 in vector and scalar forms. The drag force increases as the square of velocity.
Normally, the magnitude of the drag force coefficient ’D’ is approximated theoretically
and/or measured experimentally. The drag coefficient is a function of material type, sur-
face properties, object size and object geometry.

F F = –D v v

Figure 2.21 Aerodynamic drag

The force is acting on the cylinder, resulting in the velocities given below. What is the
applied force?

d- x = 0.1 m
---- ---- d- x = – 0.3 m
---- ----
dt 1 s dt 2 s

F F

k d = 0.1 Ns
------
m

ans. F = – 0.02N

Figure 2.22 Drill problem: Find the required forces on the damper
 translation - 2.21

2.2.6 Cables And Pulleys

Cables are useful when transmitting tensile forces or displacements. The centerline
of the cable becomes the centerline for the force. And, if the force becomes compressive,
the cable becomes limp, and will not transmit force. A cable by itself can be represented as
a force vector. When used in combination with pulleys, a cable can redirect a force vector
or multiply a force.

Typically we assume that a pulley is massless and frictionless (in the rotation chap-
ter we will assume they are not). If this is the case then the tension in the cable on both
sides of the pulley are equal, as shown in Figure 2.23.

T1

for a massless frictionless pulley

T1 = T2

T2

Figure 2.23 Tension in a cable over a massless frictionless pulley

If we have a pulley that is fixed and cannot rotate, the cable must slide over the
surface of the pulley. In this case we can use the coefficient of friction to determine the rel-
ative ratio of forces between the sides of the pulley, as shown in Figure 2.24.

T1

T2 µ ( ∆θ )
----- = e k
T1
∆θ
T2

Figure 2.24 Friction of a belt over a fixed drum

 translation - 2.22

Given,
µ s = 0.35 µ k = 0.2

M = 1Kg

Find F to start the mass moving up and down, and then

the force required to maintain a low velocity motion.
M

ans. π
0.2  ---
2
F up = 9.81e N
9.81N
F down = ---------------
π
0.2  ---
 2
e

Figure 2.25 Drill problem: Friction forces for belts on drums

Although the discussion in this section has focused on cables and pulleys, the the-
ory also applies to belts over drums.
 translation - 2.23

2.2.7 Friction

Viscous friction was discussed before, where a lubricant would provide a damping
effect between two moving objects. In cases where there is no lubricant, and the touching
surfaces are dry, dry coulomb friction may result. In this case the surfaces will stick in
place until a maximum force is overcome. After that the object will begin to slide and a
constant friction force will result.

Figure 2.26 shows the classic model for (dry Coulomb) friction. The force on the
horizontal axis is the force applied to the friction surfaces while the vertical axis is the
resulting friction force. Beneath the slip force the object will stay in place. When the slip
force is exceeded the object will begin to move, and the resulting kinetic friction force will
be relatively constant. (Note: If the object begins to travel much faster then the kinetic
friction force will decrease.) It is common to forget that friction forces are bidirectional,
but it always opposes the applied force or motion. The friction force is a function of the
coefficient of friction and the normal force across the contact surfaces. The coefficient of
friction is a function of the materials, surface texture and surface shape.

F result Block begins to slip and the

applied force exceeds the
Fs resultant and acceleration begins.

Fk

Fapplied
Fs

F
Fg Fk = µk N
F k, F s
Fs ≤ µs N
N

Note: When solving problems with friction remember that the friction force will always
equal the applied force (not the maximum force) until slip occurs. After that the fric-
tion is approximately constant. In addition, the friction forces direction opposes
applied forces, and motion.

Figure 2.26 Dry friction

 translation - 2.24

Many systems use kinetic friction to dissipate energy from a system as heat, sound
and vibration.

Find the acceleration of the block for

both angles indicated. µ s = 0.3
10 kg
µ k = 0.2 θ1 = 5 °
θ
θ 2 = 35 °

ans.

Figure 2.27 Drill problem: find the accelerations

 translation - 2.25

2.2.8 Contact Points And Joints

A system is built by connecting components together. These connections can be

rigid or moving. In solid connections all forces and moments are transmitted and the two
pieces act as a single rigid body. In moving connections there is at least one degree of free-
dom. If we limit this to translation only, there are up to three degrees of freedom, x, y and
z. In any direction there is a degree of freedom, a force or moment cannot be transmitted.

When constructing FBDs for a system we must break all of the components into
individual rigid bodies. Where the mechanism has been broken the contact forces must be
added to both of the separated pieces. Consider the example in Figure 2.28. At joint A the
forces are written as two components in the x and y directions. For joint B the force com-
ponents with equal magnitudes but opposite directions are added to both FBDs.

F Ay F By

A F Ax
F Bx

M1

B
M1 g M2 g

M2
F Bx
F By Note: Don’t forget that forces on con-
nected FBDs should have equal
magnitudes, but opposite direc-
tions.

Figure 2.28 FBDs for systems with connected members

2.3 SYSTEM EXAMPLES

An orderly approach to system analysis can simplify the process of analyzing large
systems. The list of steps below is based on general observations of good problem solving
techniques.
 translation - 2.26

1. Assign letters/numbers to designate force components (if not already done) -

this will allow you to refer to components in your calculations.
2. Define positions and directions for any moving masses. This should include the
selection of reference points.
3. Draw free body diagrams for each component, and add forces (inertia is
optional).
4. Write equations for each component by summing forces.
5.(next chapter) Combine the equations by eliminating unwanted variables.
6.(next chapter) Develop a final equation that relates input (forcing functions) to
outputs (results).

Note: When deriving differential equa- coefficient units

tions, the final value can be checked for
Kgm-
errors using unit analysis. This method F N = -----------
2
involves replacing variables with their s
unit equivalents. All the units should N-
match. Ks ---
m
·· ·
e.g., x 2 ( M 2 ) + x 2 ( B ) + x 2 ( K s2 ) + x 1 ( – K s2 ) = F Kd
Ns
------
m
∴----2 ( Kg ) + ----  ------ + m  ---- + m  – ---- = N
m m Ns N N
s s m m m
M Kg
∴N + ( N ) + ( N ) + ( – N ) = N

The units match, so there are no obvious problems.

Consider the cart in Figure 2.29. On the left is a force, it is opposed by a spring and
damper on the right. The basic problem definition already contains all of the needed defi-
nitions, so no others are required. The FBD for the mass shows the applied force and the
reaction forces from the spring and damper. When the forces are summed the inertia is on
the right side of the equation in Newton’s form. This equation is then rearranged to a sec-
ond-order non-homogeneous differential equation.
 translation - 2.27

x
Given the system diagram;
Kd
F
M1 Ks

The FBD for the cart is

F K d x·
M1
Ksx
M 1 x··
The forces for the cart are in a single direction and can be summed as,
+
∑ Fx = – F – Kd x· – Ks x = M1 x··
This equation can be rearranged to a second-order non-homogeneous diff. eqn.
Kd K
- x· + -------s- x = -------
F-
x·· + -------
M1 M1 M1

Aside: later on we will solve the differential equations, or use other methods to
determine how the system will behave. It is useful to have all of the ’output’
variables for the system on the left hand side, and everything else on the other.

Figure 2.29 A simple translational system example

 translation - 2.28

Develop the equation relating the input force to the motion (in terms of x) of the lefthand
cart for the problem below.
x1 x2

K d1 K d2
F
M1 K s1 M2 K s2

Figure 2.30 Drill problem: Find the differential equations

A simplified model of an elevator (M1) and a passenger (M2) are shown in Figure
2.31. In this example many of the required variables need to be defined. These are added
to the FBDs. Care is also taken to ensure that all forces between bodies are equal in mag-
nitude, but opposite in direction. The wall forces are ignored because they are statically
indeterminate, and x-axis force components are irrelevant to the forces in the y-axis.
 translation - 2.29

M1

M2
Kd
K S2

K S1

Assign required quantities and draw the FBDs

M2 g

· · y2
FD = Kd ( y2 – y1 ) M1 g

··
M1 y1 F S2 = K s2 ( y 1 – y 2 )
FD ··
M2 y2
F S2

y1 F S1 = – K s1 y 1

Figure 2.31 A multi-body translating system (an elevator with a passenger)

The forces on the FBDs are summed and the equations are expanded in Figure
2.32.
 translation - 2.30

Now, sum the forces in vector form, and substitute relationships,

∑ FM 1
= F S1 + F S2 + M 1 g + F D = M 1 a 1

∑ FM 2
= – F S2 + M 2 g – F D = M 2 a 2

At this point the equations are expanded

d
K S1 ( – y 1 ) + K S2 ( y 2 – y 1 ) + M 1 ( – 9.81 ) + K d ----- ( y 2 – y 1 ) = M 1 a y
dt
d
– K S2 ( y 2 – y 1 ) + M 2 ( – 9.81 ) – K d ----- ( y 2 – y 1 ) = M 2 a y
dt

Figure 2.32 Equations for the elevator

 translation - 2.31

Write the differential equations for the system below.

Assume it is in motion
A

M2
µ k1
K S1
M1 µ k2

ans.

Figure 2.33 Drill problem: A more complex translational systems

Consider the springs shown in Figure 2.34. When two springs are combined in this
manner they can be replaced with a single equivalent spring. In the parallel spring combi-
nation the overall stiffness of the spring would increase. In the series spring combination
the overall stiffness would decrease.
 translation - 2.32

Parallel Series

K S1 K S2
K S1

M K S2

Figure 2.34 Springs in parallel and series (kinematically)

Figure 2.35 shows the calculations required to find a spring coefficient equivalent
to the two springs in series. The first step is to draw a FBD for the mass at the bottom, and
for a point between the two springs, P. The forces for both of these are then summed. The
next process is to combine the two equations to eliminate the height variable created for
point P. After this, the equation is rearranged into Hooke’s law, and the equivalent spring
coefficient is found.
 translation - 2.33

First, draw FBDs for P and M and sum the forces assuming the system is static.

K S1 y 2
K S1
+ ∑ F y = K S1 y 2 – K S2 ( y 1 – y 2 ) = 0 (1)
P
K S2 ( y 1 – y 2 ) P
y2
K S2 ( y 1 – y 2 ) K S2

M
+ ∑ Fy = K S2 ( y 1 – y 2 ) – F g = 0 (2)
Fg
M
y1
Next, rearrange the equations to eliminate y2 and simplify.

(1) becomes K S2 ( y 1 – y 2 ) – Fg = 0
Fg
y 1 – y 2 = ----------
K S2
Fg
y 2 = y 1 – ---------- (3)
K S2

(2) becomes K S1 y 2 – K S2 ( y 1 – y 2 ) = 0

y 2 ( K S1 + K S2 ) = y 1 K S2 (4)

 Fg 
sub (3) into (4)  y – --------- - ( K + K ) = y K
 1 K S2 S1 S2 1 S2
Fg K S2
y 1 – ---------- = y 1 ----------------------------
K S2 K S1 + K S2
 K S2 
F g = y 1  1 – ---------------------------- K S2
 K S1 + K S2
 K S1 + K S2 – K S2
F g = y 1  -------------------------------------------- K S2
 K S1 + K S2 
 K S1 K S2 
F g = y 1  ----------------------------
 K S1 + K S2
Finally, consider the basic spring equation to find the equivalent spring coefficient.
K S1 K S2
K equiv· = ----------------------------
K S1 + K S2

Figure 2.35 Calculation of an equivalent spring coefficient for springs in series

 translation - 2.34

K S1 K S2

ans.

Figure 2.36 Drill problem: Find an equivalent spring for the springs in parallel

Consider the drill problem. When an object has no mass, the force applied to one
side of the spring will also be applied to the other. The only factor that changes is displace-
ment.
 translation - 2.35

Show that a force applied to one side of a massless spring is the reaction force at the other
side.

ans.

Figure 2.37 Drill problem: Prove that the force on both sides is equal

2.4 OTHER TOPICS

Designing a system in terms of energy content can allow insights not easily
obtained by the methods already discussed. Consider the equations in Figure 2.38. These
equations show that the total energy in the system is the sum of kinetic and potential
energy. Kinetic energy is half the product of mass times velocity squared. Potential energy
in translating systems is a force magnitude multiplied by a distance (that force was applied
over). In addition, the power, or energy transfer rate is the force applied multiplied by the
velocity.
 translation - 2.36

E = EP + EK (7)
2
Mv (8)
E K = -----------
2
E P = Fd = Mgd (9)

d- E
P = Fv = ---- (10)
dt

Figure 2.38 Energy and power equations for translating masses

2.5 SUMMARY

• FBDs are useful for reducing complex systems to simpler parts.

• Equations for translation and rotation can be written for FBDs.
• The equations can be integrated for dynamic cases, or solved algebraically for
static cases.

2.6 PRACTICE PROBLEMS

1. If a spring has a deflection of 6 cm when exposed to a static load of 200N, what is the spring
constant?
 translation - 2.37

2. Derive the effective damping coefficients for the pairs below from basic principles,
a) Kd1

Kd2

b) Kd1 Kd2

3. Write a differential equation for the mass pictured below.

M F

4. Write the differential equations for the translating system below.

Ks
F
M1 M2
 translation - 2.38

5. Write the differential equations for the system below.

x1 x2

K d1

K s1 M1 K s2 M2 F
B

6. Write the differential equations for the system given below.

Kd Ks
F
M1 M2

B1 B2

7. Write the differential equations for the system below.

x1 x2

K d1

K s1 M1 K s2 M2 F

8. Write the differential equations for the system below.

x1 x2

K d1 K d2 K d3

K s1 M1 K s2 M2 F
K s3
 translation - 2.39

9. Write the differential equations for the system below.

K s2 x2

M2 F
K d1
B
x1
K s1 M1

10. Write the differential equations for the system below.

K s2 x2

M2 F
K d1 µ s, µ k

K s1 M1 x1

11. Write the differential equations for the system below.

F

M1 x1

K s1
K d1

M2 x2

K s2
K d2
 translation - 2.40

12. Write the differential equations for the system below. Assume that the pulley is massless and
frictionless and that the system begins undeflected.

x
K s1 R
M

K s2

13. Write the differential equations for the system below. In this system the upper mass, M1, is
between a spring and a cable and there is viscous damping between the mass and the floor. The
suspended mass, M2, is between the cable and a damper. The cable runs over a massless, fric-
tionless pulley.

x1
Ks R
M1

M2 x2

Kd
 translation - 2.41

14. Write the differential equations for the system below.

x1
K s1
M1
µ s, µ k

M2 x2

K s2

2.7 PRACTICE PROBLEM SOLUTIONS

1. Ks = 33.3 N/cm = 3333 N/m

2.
a) K eq = K d1 + K d2 b) K d1 K d2
K eq = ------------------------
K d1 + K d2

3.

x + x  ----- = – -----
·· · B F
M M

4.
Ks –Ks
x 1 + x 1  ------- + x 2  --------- = -------
·· F
M1 M1 M1
Ks –Ks
x 2 + x 2  ------- + x 1  --------- = 0
··
 M 2  M2 
 translation - 2.42

5.
· K d1 K s1 + K s2 – K s2
x 1 + x 1  --------- + x 1  ----------------------- + x 2  ----------- = 0
··
M1 M1 M1

K s2 – K s2
x 2 + x 2  ------- + x 2  -------- + x 1  ----------- = -------
·· · B F
M2 M2 M2 M2

6.

· Kd + B1 Ks –K s
x 1 + x 1  ------------------- + x 1  ------- + x 2  --------- = 0
··
 M1   M 1  M1 
· B2 Ks –K s
x 2 + x 2  ------- + x 2  ------- + x 1  --------- = -------
·· –F
M2 M2 M2 M2

7.

· K d1 K s1 – K s2
x 1 + x 1  --------- + x 1  -------- + x 2  ----------- = 0
··
 M1   M1   M1 
Ks
x 1 + x 1  ------- + x 2 + x 2  ------- + x 2  ------- = -------
·· · B ·· · B F
M2 M2 M2 M2
 translation - 2.43

8.

FBDs: ··
M 1 x1
· · · ·
– K d1 x 1 K d2 ( x 1 – x 2 ) K d3 x 2
M1 M2
– K s1 x 1 K s2 ( x 1 – x 2 ) K s3 x 2
··
F M2 x2
For M1: · · · ··
+ ∑ F = – Kd1 x1 – Ks1 x1 – Kd2 ( x1 – x2 ) – Ks2 ( x1 – x 2 ) = M1 x1
·· · ·
x 1 ( M 1 ) + x 1 ( K d1 + K d2 ) + x 1 ( K s1 + K s2 ) + x 2 ( – K d2 ) + x 2 ( – K s2 ) = 0
· K d1 + K d2 K s1 + K s2 · – K d2 – K s2
x 1 + x 1  ------------------------ + x 1  ----------------------- + x 2  ------------ + x 2  ----------- = 0
··
 M1   M1   M1   M1 
For M2: · · · ·
+ ∑F = K d2 ( x 1 – x 2 ) + K s2 ( x 1 – x 2 ) + F – K d3 x 2 – K s3 x 2 = M 2 x 2
·· · ·
x 2 ( M 2 ) + x 2 ( K d2 + K d3 ) + x 2 ( K s2 + K s3 ) + x 1 ( – K d2 ) + x 1 ( – K s2 ) = F
· K d2 + K d3 K s2 + K s3 · – K d2 – K s2
x 2 + x 2  ------------------------ + x 2  ----------------------- + x 1  ------------ + x 1  ----------- = F
··
M2 M2 M2 M2

9.
·· · ·
x 1 ( M 1 ) + x 1 ( K d1 + B ) + x 1 ( K s1 ) + x 2 ( – B ) = 0
· K d1 + B- K s1 ·  – B 
x 1 + x 1  ------------------ + x 1  -------
·· - + x 2 ------- = 0
 M1   M1   M 1
·· · ·
x 2 ( M 2 ) + x 2 ( B ) + x 2 ( K s2 ) + x 1 ( – B ) = F
K s2
x 2 + x 2  ------- + x 2  -------- + x 1  ------- = -------
·· · B · –B F
 M 2  M2   M 2 M2
 translation - 2.44

10.
·· ·
x 1 ( M 1 ) + x 1 ( K d1 ) + x 1 ( K s1 ) = FF
· K d1 K s1 FF
x 1 + x 1  --------- + x 1  -------- = -------
··
 M1   M1  M1
··
x 2 ( M 2 ) + x 2 ( K s2 ) = F – F F
K s2 F – FF
x 2 + x 2  -------- = ----------------
··
M2 M2
· ·
where, FF ≤ µs M2 g if x1 = x2
· ·
 x1 – x2  · ·
F F = µ k M 2 g  ------------------
· · - if x1 ≠ x2
 x1 – x2 

11.
(assuming no gravity
·· · ·
x 1 ( M 1 ) + x 1 ( K d1 ) + x 1 ( K s1 ) + x 2 ( – K d1 ) + x 2 ( – K s1 ) = F
· K d1 K s1 · – K d1 – K s1
x 1 + x 1  --------- + x 1  -------- + x 2  ------------ + x 2  ----------- = -------
·· F
 M1   M1   M1   M1  M1
·· · ·
x 2 ( M 2 ) + x 2 ( K d1 + K d2 ) + x 2 ( K s1 + K s2 ) + x 1 ( – K d1 ) + x 1 ( – K s1 ) = 0
· K d1 + K d2 K s1 + K s2 · – K d1 – K s1
x 2 + x 2  ------------------------ + x 2  ----------------------- + x 1  ------------ + x 1  ----------- = 0
··
M2 M2 M2 M2

12.
if ( x ≥ 0 ) T = 0

if ( x < 0 ) T = –K s2 x

K s1
x + x  ------- + x  -------- = -------
·· · B T
M1 M1 M1
 translation - 2.45

13.

FBDs: Ks x 1 T T

· M1 ··
Bx 1 ·· M2 M2 x2
M1 x1

· ·· ·
For M1: +
∑ F = – Ks x1 – Bx 1 + T = M1 x1 M2 g Kd x2
·· ·
x1 ( M1 ) + x1 ( B ) + x1 ( Ks ) = T
Ks
x 1 + x 1  ------- + x 1  ------- = -------
·· · B T
 M 1  M 1 M1
For M2:
· ··
+ ∑F = T + Kd x2 – M2 g = –M2 x2
·· ·
x2 ( – M2 ) + x2 ( –K d ) = T – M 2 g
· Kd T – M2 g
x 2 + x 2  ------- = --------------------
··
 M 2 –M2

For T: if T <= 0 then T = 0N

if T > 0 x1 = x2

14.
if ( ( x 2 – x 1 ) < 0 ) T = 0

·
if ( x 1 = 0 ) F F ≤ M 1 gµ s

·
· x1
if ( x 1 ≠ 0 ) F F = M 1 gµ k -------
·-
x1
K s1 T – FF
x 1 + x 1  -------- = ---------------
··
M1 M1
K s2
x 2 + x 2  -------- = ------- – g
·· –T
 M2  M2

2.8 ASSIGNMENT PROBLEMS

 differential equations - 3.1

3. ANALYSIS OF DIFFERENTIAL EQUATIONS

Topics:
• First and second-order homogeneous differential equations
• Non-homogeneous differential equations
• First and second-order responses
• Non-linear system elements
• Design case

Objectives:
• To develop explicit equations that describe a system response.
• To recognize first and second-order equation forms.

3.1 INTRODUCTION

In the previous chapter we derived differential equations of motion for translating

systems. These equations can be used to analyze the behavior of the system and make
design decisions. The most basic method is to select a standard input type (a forcing func-
tion) and initial conditions, and then solve the differential equation. It is also possible to
estimate the system response without solving the differential equation as will be discussed
later.

Figure 3.1 shows an abstract description of a system. The basic concept is that the
system changes the inputs to outputs. Say, for example, that the system to be analyzed is
an elevator. Inputs to the system would be the mass of human riders and desired elevator
height. The output response of the system would be the actual height of the elevator. For
analysis, the system model could be developed using differential equations for the motor,
elastic lift cable, mass of the car, etc. A basic test would involve assuming that the elevator
starts at the ground floor and must travel to the top floor. Using assumed initial values and
input functions the differential equation could be solved to get an explicit equation for ele-
vator height. This output response can then be used as a guide to modify design choices
(parameters). In practice, many of the assumptions and tests are mandated by law or by
groups such as Underwriters Laboratories (UL), Canadian Standards Association (CSA)
and the European Commission (CE).
 differential equations - 3.2

inputs system outputs

Note: By convention inputs
are on the left, and outputs
forcing response are on the right.
function differential function
equations

Figure 3.1 A system with and input and output response

There are several standard input types used to test a system. These are listed below
in order of relative popularity with brief explanations.

• step - a sudden change of input, such as very rapidly changing a desired speed
from 0Hz to 50Hz.
• ramp - a continuously increasing input, such as a motor speed that increases con-
stantly at 10Hz per minute.
• sinusoidal - a cyclic input that varies continuously, such as wave height that is
continually oscillating at 1Hz.
• parabolic - an exponentially increasing input, such as a motor speed that is 2Hz at
1 second, 4rad/s at 2 seconds, 8rad/s at 3 seconds, etc.

After the system has been modeled, an input type has been chosen, and the initial
conditions have been selected, the system can be analyzed to determine its behavior. The
most fundamental technique is to integrate the differential equation(s) for the system.

3.2 EXPLICIT SOLUTIONS

Solving a differential equation results in an explicit solution. This equation pro-

vides the general response as a function of time, but it can also be used to find frequencies
and other characteristics of interest. This section will review techniques used to integrate
first and second-order homogenous differential equations. These equations correspond to
systems without inputs, also called unforced systems. Non-homogeneous differential
equations will also be reviewed.

The basic types of differential equations are shown in Figure 3.2. Each of these
equations is linear. On the left hand side is the integration variable ’x’. If the right hand
side is zero, then the equation is homogeneous. Each of these equations is linear because
each of the terms on the left hand side is simply multiplied by a linear coefficient.
 differential equations - 3.3

·
Ax + Bx = 0 first-order homogeneous
·
Ax + Bx = Cf ( t ) first-order non-homogeneous
·· ·
Ax + Bx + Cx = 0 second-order homogeneous
·· ·
Ax + Bx + Cx = Df ( t ) second-order non-homogeneous

Figure 3.2 Standard equation forms

A general solution for a first-order homogeneous differential equation is given in

Figure 3.3. The solution begins with the solution of the homogeneous equation where a
general form is ’guessed’. Substitution leads to finding the value of the coefficient ’Y’.
Following this, the initial conditions for the equation are used to find the value of the coef-
ficient ’X’. Notice that the final equation will begin at the initial displacement, but
approach zero as time goes to infinity. The e-to-the-x behavior is characteristic for a first-
order response.
 differential equations - 3.4

Given the general form of a first-order homogeneous equation,

Ax· + Bx = 0 and x ( 0 ) = x0

Guess a solution form and solve.

– Yt – Yt initial condition
x = Xe x· = – YX e
– Yt – Yt
A ( – YX e ) + B ( Xe ) = 0 Note: The general form below
is useful for finding almost
A ( –Y ) + B = 0 all homogeneous equations
Y = B --- – Yt
A x h ( t ) = Xe
Therefore the general form is,
–B
---t
A
x h = Xe

Next, use the initial conditions to find the remaining unknowns.

–B---t
A
x h = Xe
–B---0
A
x 0 = Xe

x0 = X

Therefore the final equation is,

B
– --- t
A
x ( t ) = x0 e

Figure 3.3 General solution of a first-order homogeneous equation

 differential equations - 3.5

Solve the following differential equation given the initial condition.

x· + 2x = 0 x(0) = 3

– 2t
ans. x ( t ) = 3e

Figure 3.4 Drill Problem: First order homogeneous differential equation

The general solution to a second-order homogeneous equation is shown in Figure

3.5. The solution begins with a guess of the homogeneous solution, and the solution of a
quadratic equation. There are three possible cases that result from the solution of the qua-
dratic equation: different but real roots; two identical real roots; or two complex roots. The
three cases result in three different forms of solutions, as shown. The complex result is the
most notable because it results in sinusoidal oscillations. It is not shown, but after the
homogeneous solution has been found, the initial conditions need to be used to find the
remaining coefficient values.
 differential equations - 3.6

Given,
Ax·· + Bx· + Cx = 0 x ( 0 ) = x0 and x· ( 0 ) = v 0

Guess a general equation form and substitute it into the differential equation,
Yt Yt 2 Yt
x h = Xe x· h = YXe x··h = Y Xe
2 Yt Yt Yt
A ( Y Xe ) + B ( YXe ) + C ( Xe ) = 0
2
A(Y ) + B( Y) + C = 0
2 2
– B ± ( B ) – 4 ( AC ) – B ± B – 4AC
Y = ---------------------------------------------------- = ------------------------------------------
2A 2A

Note: There are three possible outcomes of finding the roots of the equa-
tions: two different real roots, two identical real roots, or two complex
roots. Therefore there are three fundamentally different results.

If the values for Y are both real, but different, the general form is,
R1 t R2 t
Y = R 1, R 2 xh = X1 e + X2 e
Note: The initial conditions are then used to find the values for X1 and X2.

If the values for Y are both real, and identical, the general form is,
R1t R1 t
Y = R 1, R 1 xh = X1 e + X 2 te
The initial conditions are then used to find the values for X1 and X2.

If the values for Y are complex, the general form is,

Y = σ ± ωj σt
x h = X 3 e cos ( ω t + X 4 )
The initial conditions are then used to find the values of X3 and X4.

Figure 3.5 Solution of a second-order homogeneous equation

As mentioned above, a complex solution when solving the homogeneous equation

results in a sinusoidal oscillation, as proven in Figure 3.6. The most notable part of the
solution is that there is both a frequency of oscillation and a phase shift. This form is very
useful for analyzing the frequency response of a system, as will be seen in a later chapter.
 differential equations - 3.7

Consider the situation where the results of a homogeneous solution are the complex
conjugate pair.
Y = R ± Cj
This gives the general result, as shown below:
( R + C j )t ( R – Cj )t
x = X1 e + X2e
Rt Cjt Rt – Cjt
x = X1 e e + X2 e e

x = e  X1 e
Rt Cjt – Cjt
+ X2 e
 
Rt
x = e ( X 1 ( cos ( Ct ) + j sin ( Ct ) ) + X 2 ( cos ( – C t ) + j sin ( – C t ) ) )
Rt
x = e ( X 1 ( cos ( Ct ) + j sin ( Ct ) ) + X 2 ( cos ( ( Ct ) – j sin ( Ct ) ) ) )
Rt
x = e ( ( X1 + X 2 ) cos ( Ct ) + j ( X 1 – X 2 ) sin ( Ct ) )
Rt
x = e ( ( X1 + X 2 ) cos ( Ct ) + j ( X 1 – X 2 ) sin ( Ct ) )
2 2 2
Rt ( X 1 + X 2 ) + j ( X 1 – X 2 )
x = e ------------------------------------------------------------------------- ( ( X 1 + X 2 ) cos ( Ct ) + j ( X 1 – X2 ) sin ( Ct ) )
2 2 2
( X1 + X2 ) + j ( X1 – X2 )

X 1 + 2X 1 X 2 + X 2 –  X 1 + – 2 X 1 X 2 + X 2
2 2 2 2
Rt
x = e ------------------------------------------------------------------------------------------------------------- ( ( X 1 + X 2 ) cos ( Ct ) + j ( X 1 – X 2 ) sin ( C
X 1 + 2X 1 X 2 + X 2 –  X 1 + – 2 X 1 X 2 + X 2
2 2 2 2

Rt 4X 1 X 2
x = e --------------------- ( ( X 1 + X 2 ) cos ( Ct ) + j ( X 1 – X 2 ) sin ( Ct ) )
4X 1 X 2
Rt  ( X1 + X2 ) ( X1 – X2 ) 
x = e - cos ( Ct ) + j -----------------------
4X 1 X2  ------------------------ - sin ( Ct )
 4X X 4X 1 X 2 
1 2
Rt   ( X1 – X2 )  
x = e 4X 1 X 2 cos  Ct + atan  ------------------------- 
  ( X1 + X 2 ) 
Rt
x = e X3 cos ( Ct + X 4 ) where, X = 4X 1 X 2
3
 ( X1 – X2 ) 
X 4 = atan  -------------------------
frequency phase shift  ( X 1 + X 2 )

Figure 3.6 Phase shift solution for a second-order homogeneous differential equation
 differential equations - 3.8

Note: Occasionally a problem solution might consist of both a sine and cosine term with
the same frequency. These should normally be combined to a single term with a phase
shift as shown below.

Recall the double angle formula,

sin ( ωt + θ ) = sin ωt cos θ + sin θ cos ωt
This can be written in a more common form,

A ( sin ωt cos θ + sin θ cos ωt ) = A sin ( ωt + θ )

A cos θ sin ωt + A sin θ cos ωt = A sin ( ωt + θ )
B sin ωt + C cos ωt = A sin ( ωt + θ ) where,
B = A cos θ
B C C = A sin θ
A = ------------ = -----------
cos θ sin θ
sin θ-
θ = atan  ----
C C
----------- = ----
cos θ B  B

2 2
A = B +C

Consider the example,

3 + 4 sin  5t + atan  ---  = 5 sin ( 5t + 0.927 )

2 2 4
3 sin 5t + 4 cos 5t =
3

Figure 3.7 Phase shift solution form

 differential equations - 3.9

Solve the following differential equation given the initial condition.

·
x·· + 2x + x = 0 x( 0) = 1 x· ( 0 ) = 2

–t –t
ans. x ( t ) = e + 3te

Figure 3.8 Drill Problem: Second order homogeneous differential equation

The methods for solving non-homogeneous differential equations builds upon the
methods used for the solution of homogeneous equations. This process adds a step to find
the particular solution of the equation. An example of the solution of a first-order non-
homogeneous equation is shown in Figure 3.9. To find the homogeneous solution the non-
homogeneous part of the equation is set to zero. To find the particular solution the final
 differential equations - 3.10

form must be guessed. This is then substituted into the equation, and the values of the
coefficients are found. Finally the homogeneous and particular solutions are added to get
the final equation. The overall response of the system can be obtained by adding the
homogeneous and particular parts. This is acceptable because the equations are linear, and
the principle of superposition applies. The homogeneous equation deals with the response
to initial conditions, and the particular solution deals with the response to forced inputs.

Generally,
Ax· + Bx = Cf ( t ) x ( 0 ) = x0
First, find the homogeneous solution as before, in Figure 3.3.
–B
---t
A
xh = x0 e
Next, guess the particular solution by looking at the form of ’f(t)’. This step is highly
subjective, and if an incorrect guess is made, it will be unsolvable. When this hap-
pens, just make another guess and repeat the process. An example is given below. In
the case below the guess should be similar to the exponential forcing function.
For example, if we are given
4t
6x· + 2x = 5e
A reasonable guess for the particular solution is,
4t 4t
xp = C1 e \x· p = 4C 1 e

Substitute these into the differential equation and solve for A.

6  4C 1 e  + 2  C 1 e  = 5e
4t 4t 4t

5-
24C 1 + 2C 1 = 5 ∴C 1 = -----
26
Combine the particular and homogeneous solutions.
– 6--- t
5- e 4t + x e 2
x = x p + x h = ----- 0
26

Figure 3.9 Solution of a first-order non-homogeneous equation

The method for finding a particular solution for a second-order non-homogeneous

differential equation is shown in Figure 3.10. In this example the forcing function is sinu-
soidal, so the particular result should also be sinusoidal. The final result is converted into a
phase shift form.
 differential equations - 3.11

Generally,
Ax·· + Bx· + Cx = Df ( t ) x( 0 ) = x0 and x· ( 0 ) = v 0
1. Find the homogeneous solution as before.
σt
x h = X 3 e cos ( ω t + X4 )
or σt σt
x h = X 1 e + X 2 te
or σ1 t σ2 t
xh = X1 e + X2 e
2. Guess the particular solution by looking at the form of ’f(t)’. This step is highly sub-
jective, and if an incorrect guess is made it will be unsolvable. When this happens,
just make another guess and repeat the process. For the purpose of illustration an
example is given below. In the case below it should be similar to the sine function.
For example, if we are given
2x·· + 6x· + 2x = 2 sin ( 3t + 4 )
A reasonable guess is,
x p = A sin ( 3t ) + B cos ( 3t )
x· p = 3A cos ( 3t ) – 3B sin ( 3t )
x··p = – 9A sin ( 3t ) – 9B cos ( 3t )

Substitute these into the differential equation ans solve for A and B.
2 ( – 9 A sin ( 3t ) – 9B cos ( 3t ) ) + 6 ( 3A cos ( 3t ) – 3B sin ( 3t ) ) +
2 ( A sin ( 3t ) + B cos ( 3t ) ) = 2 sin ( 3t + 4 )
( – 18 A – 18B + 2A ) sin ( 3t ) + ( – 18 B + 18A + 2B ) cos ( 3t ) = 2 sin ( 3t + 4 )
( – 16A – 18B ) sin ( 3t ) + ( 18A – 16B ) cos ( 3t ) = 2 ( sin 3t cos 4 + cos 3t sin 4 )
( – 16A – 18B ) sin ( 3t ) + ( 18A – 16B ) cos ( 3t ) = ( 2 cos 4 ) sin ( 3t ) + ( 2 sin 4 ) cos ( 3t )
– 16A – 18B = 2 cos 4 18A – 16B = 2 sin 4
–1
– 16 – 18 A = 2 cos 4 A = – 16 –18 – 1.307 = – 0.0109
18 – 16 B 2 sin 4 B 18 –16 – 1.514 0.0823
Next, rearrange the equation to phase shift form.
x p = – 0.0109 sin ( 3t ) + 0.0823 cos ( 3t )
π
– 0.0109 + 0.0823 sin  3t + atan  ------------------- + ---
2 2 0.0823
xp =
–0.0109 2
3. Use the initial conditions to determine the coefficients in the homogeneous solution.

Figure 3.10 Solution of a second-order non-homogeneous equation

 differential equations - 3.12

When guessing particular solutions, the forms in Figure 3.11 can be helpful.

Forcing Function Guess

A C
Ax + B Cx + D
Ax Ax or Ax
e Ce Cxe
B sin ( Ax ) or B cos ( Ax ) C sin ( Ax ) + D cos ( Ax )
or Cx sin ( Ax ) + xD cos ( Ax )

Figure 3.11 General forms for particular solutions

 differential equations - 3.13

Solve the following differential equation given the initial condition.

x·· + 2x· + x = 1 x( 0) = 0 x· ( 0 ) = 0

–t –t
ans. x ( t ) = – e – te + 1

Figure 3.12 Drill Problem: Second order non-homogeneous differential equation

An example of a second-order system is shown in Figure 3.13. As expected, it

begins with a FBD and summation of forces. This is followed with the general solution of
 differential equations - 3.14

the homogeneous equation. Real roots are assumed thus allowing the problem solution to
continue in Figure 3.14.

Assume the system illustrated to the right starts

from rest at a height ’h’. At time ’t=0’ the sys-
Fg
tem is released and allowed to move. y
Ks Kd

Mg

My··
+ ∑ Fy = – Mg + K s y + K d y· = – M y··
Ksy K d y·
My·· + K d y· + K s y = Mg
Find the homogeneous solution.
At At 2 At
yh = e y· h = Ae y··h = A e
My·· + K d y· + K s y = 0
2 At
M  A e  + K d ( Ae ) + K s ( e ) = 0
At At
2
MA + K d A + K s = 0
2
– K d ± K d – 4MK s
A = --------------------------------------------------
2M
Let us assume that the values of M, Kd and Ks lead to the case of two different
positive roots. This would occur if the damper value was much larger than
the spring and mass values. Thus,
A = R 1, R 2
R1 t R2 t
y h = C1 e + C2 e

Figure 3.13 Second-order system example

The solution continues by assuming a particular solution and calculating values for
the coefficients using the initial conditions in Figure 3.14. The final result is a second-
order system that is overdamped, with no oscillation.
 differential equations - 3.15

Next, find the particular solution.

· ··
yp = C yh = 0 yh = 0
M ( 0 ) + K d ( 0 ) + K s ( C ) = Mg
Mg
C = --------
Ks
Now, add the homogeneous and particular solutions and solve for the unknowns
using the initial conditions.
Mg R t R t
y ( t ) = y p + y h = -------- + C 1 e 1 + C 2 e 2
Ks
y(0) = h y' ( 0 ) = 0
Mg 0 0
h = -------- + C 1 e + C 2 e
Ks
Mg
C 1 + C 2 = h – --------
Ks
R1t R2 t
y' ( t ) = R 1 C 1 e + R2 C2 e
0 0
0 = R1 C1 e + R2 C2 e
–R2
0 = R1 C 1 + R2 C2 C 1 = --------- C 2
R1
R2 Mg
– ------ C 2 + C 2 = h – --------
R1 Ks
K s h – Mg –R1 – R2 K s h – Mg –R1
C 2 =  -----------------------  ------------------ C 1 = ---------  -----------------------  ------------------
 K s   R 1 – R 2 R1 Ks R1 – R2
Now, combine the solutions and solve for the unknowns using the initial conditions.
K s h – Mg –R1 R t – R 2 K s h – Mg – R1
y ( t ) = -------- +  -----------------------  ------------------ e 1 + ---------  -----------------------  ------------------ e 2
Mg R t
Ks  Ks   R1 – R2  R1  Ks   R1 – R 2 
K s h – Mg –R1 K s h – Mg R2
y ( t ) = -------- +  -----------------------  ------------------ e +  -----------------------  ------------------ e
Mg R1 t R2 t
Ks  Ks   R1 – R2   Ks   R1 – R 2 

Figure 3.14 Second-order system example (continued)

 differential equations - 3.16

Given,
A sin ( ω t + θ ) (the desired final form)
A ( cos ω t sin θ + sin ω t cos θ )
( A sin θ ) cos ω t + ( A cos θ ) sin ω t
B cos ω t + C sin ω t (the form we start with)
where, B = A sin θ
C = A cos θ

To find theta,
A sin θ- = tan θ
B- = ---------------
---
C A cos θ

θ = atan  ----
B
C
To find A, (method #1)
B C-
A = ----------
- = -----------
sin θ cos θ
To find A, (method #2)
2 2
A = B +C

For example,
3 cos 5t + 4 sin 5t

3 + 4 sin  5t + atan 4---

2 2
3
5 sin ( 5t + 0.9273 )

Figure 3.15 Proof for conversion to phase form

3.3 RESPONSES

Solving differential equations tends to yield one of two basic equation forms. The
e-to-the-negative-t forms are the first-order responses and slowly decay over time. They
never naturally oscillate, and only oscillate if forced to do so. The second-order forms may
 differential equations - 3.17

include natural oscillation.

3.3.1 First-order

A first-order system is described with a first-order differential equation. The

response function for these systems is natural decay or growth as shown in Figure 3.16.
The time constant for the system can be found directly from the differential equation. It is
a measure of how quickly the system responds to a change. When an input to a system has
changed, the system output will be approximately 63% of the way to its final value when
the elapsed time equals the time constant. The initial and final values of the function can
be determined algebraically to find the first-order response with little effort.

If we have experimental results for a system, we can calculate the time constant,
initial and final values. The time constant can be found two ways, one by extending the
slope of the first part of the curve until it intersects the final value line. That time at the
intersection is the time constant. The other method is to look for the time when the output
value has shifted 63.2% of the way from the initial to final values for the system. Assum-
ing the change started at t=0, the time at this point corresponds to the time constant.
 differential equations - 3.18

y
t
y0 –-
τ y· + --1- y = f ( t )
y ( t ) = y 1 + ( y 0 – y 1 )e τ

y1 time constant

Note: The time will be equal to

τ t
the time constant when the
OR value is 63.2% of the way to
y the final value, as shown
below.
y1
– -t
τ
y ( τ ) = y 1 + ( y 0 – y 1 )e
–1
y ( τ ) = y 1 + ( y 0 – y 1 )e
y0
y ( τ ) = y 1 + ( y 0 – y 1 )0.368
τ t y ( τ ) = 1 + ( 0 – 1 )0.368
y ( τ ) = 0.632

Figure 3.16 Typical first-order responses

The example in Figure 3.17 calculates the coefficients for a first-order differential
equation given a graphical output response to an input. The differential equation is for a
permanent magnet DC motor, and will be examined in a later chapter. If we consider the
steady state when the speed is steady at 1400RPM, the first derivative will be zero. This
simplifies the equation and allows us to calculate a value for the parameter K in the differ-
ential equation. The time constant can be found by drawing a line asymptotic to the start of
the motor curve, and finding the point where it intercepts the steady-state value. This gives
an approximate time constant of 0.8 s. This can then be used to calculate the remaining
coefficient. Some additional numerical calculation leads to the final differential equation
as shown.
 differential equations - 3.19

For the motor, use the differential equation and the speed curve when Vs=10V is applied:

1400 RPM  ----

d-  K 2  -----
K-
 dt ω +  JR  ω =  JR V s
-------

1s 2s 3s
For steady-state
 ----
d- –1
 dt ω = 0 ω = 1400RPM = 146.6rads
 K 2
0 +  ------- 146.6 =  ------ 10
K
 JR  JR
K = 0.0682

1400 RPM
τ ª 0.8s

1s
 K 2 1
 ------- = ----------
 JR  0.8s
0.0682  ------ = ---------
K 1-
 JR 0.8s
K- = 18.328
-----
JR
ω· + ------1-
ω = 18.328Vs
0.8

Figure 3.17 Finding an equation using experimental data

 differential equations - 3.20

Find the differential equation when a step input of Vs=12V is applied:

1800 RPM 2
 ----
d-  -----
K   -----
K-
 dt ω +  JR ω =  JR V s
-

0.2s 0.4s 0.6s

 ----
d- 1 1800
ω + ---------- ω = --------------------- V s
ans.  dt 0.15 12 ( 0.15 )

Figure 3.18 Drill problem: Find the constants for the equation

A simple mechanical example is given in Figure 3.19. The modeling starts with a
FBD and a sum of forces. After this, the homogenous solution is found by setting the non-
homogeneous part to zero and solving. Next, the particular solution is found, and the two
solutions are combined. The initial conditions are used to find the remaining unknown
coefficients.
 differential equations - 3.21

Find the response to the applied force if the force is applied at F

t=0s. Assume the system is initially deflected a height of h.

F
y
Ks Kd
 d- y = 0
+ ∑ F y = – F + K s y + K d  ----
dt
K d  ----- y
d
Ks y  dt K d y· + K s y = F
Find the homogeneous solution.
Bt Bt
y h = Ae y· h = ABe
Bt Bt
K d ( ABe ) + K s ( Ae ) = 0
Kd B + K s = 0
–K
B = ---------s
Kd
Next, find the particular solution.
yp = C y· p = 0

Kd ( 0 ) + K s ( C ) = F F-
∴C = -----
Ks
Combine the solutions, and find the remaining unknown.
–Ks
---------t
Kd
F-
y ( t ) = y p + y h = Ae + -----
Ks
y(0) = h
0 F F
h = Ae + ----- - ∴A = h – ------
Ks K s
The final solution is,
–Ks
--------- t
Kd
y ( t ) =  h – ----- e
F F
+ -----
Ks Ks

Figure 3.19 First-order system analysis example

 differential equations - 3.22

Use the general form given below to solve the problem in Figure 3.19 without solving
the differential equation. Assume the system starts at y=-20.
t
– --
· τ
y + 10y = 5 y ( t ) = y 1 + ( y 0 – y 1 )e

ans. – 10t
y ( t ) = 0.5 – 19.5e

Figure 3.20 Drill problem: Developing the final equation using the first-order model
form

A first-order system tends to be passive, meaning it doesn’t deliver energy or

power. A first-order system will not oscillate unless the input forcing function is also
oscillating. The output response lags the input and the delay is determined by the system’s
time constant.
 differential equations - 3.23

3.3.2 Second-order

A second-order system response typically contains two first-order responses, or a

first-order response and a sinusoidal component. A typical sinusoidal second-order
response is shown in Figure 3.21. Notice that the coefficients of the differential equation
include a damping coefficient and a natural frequency. These can be used to develop the
final response, given the initial conditions and forcing function. Notice that the damped
frequency of oscillation is the actual frequency of oscillation. The damped frequency will
be lower than the natural frequency when the damping coefficient is between 0 and 1. If
the damping coefficient is greater than one the damped frequency becomes negative, and
the system will not oscillate because it is overdamped.

A second-order system, and a typical response to a stepped input.

2 2
y·· + 2 ζω n y· + ω n y = f ( t ) σ = ζω n ωd = ωn 1 – ζ

–σ t
y ( t ) = y ss + ( y 0 – y ss )e cos ( ω d t )

y ss

y0

ωn Natural frequency of system - Approximate frequency of control

system oscillations.
ξ Damping factor of system - If < 1 underdamped, and system will
oscillate. If =1 critically damped. If > 1 overdamped, and never
any oscillation (more like a first-order system). As damping factor
approaches 0, the first peak becomes infinite in height.
ωd The actual frequency of oscillation - It is below the natural fre-
quency because of the damping.

Figure 3.21 The general form for a second-order system

When only the damping coefficient is increased, the frequency of oscillation, and
overall response time will slow, as seen in Figure 3.22. When the damping coefficient is 0
the system will oscillate indefinitely. Critical damping occurs when the damping coeffi-
 differential equations - 3.24

cient is 1. At this point both roots of the differential equation are equal. The system will
not oscillate if the damping coefficient is greater than or equal to 1.

ξ = 0

ξ = 0.5

(underdamped)

1- = 0.707
ξ = ------
2

ξ = 1 (critical)

(overdamped)

ξ»1

Figure 3.22 The effect of the damping coefficient

When observing second-order systems it is more common to use more direct mea-
surements of the response. Some of these measures are shown in Figure 3.23. The rise
 differential equations - 3.25

time is the time it takes to go from 10% to 90% of the total displacement, and is compara-
ble to a first order time constant. The settling time indicates how long it takes for the sys-
tem to pass within a tolerance band around the final value. The permissible zone shown is
2%, but if it were larger the system would have a shorter settling time. The period of oscil-
lation can be measured directly as the time between peaks of the oscillation, the inverse is
the damping frequency. (Note: don’t forget to convert to radians.) The damped frequency
can also be found using the time to the first peak, as half the period. The overshoot is the
height of the first peak. Using the time to the first peak, and the overshoot the damping
coefficient can be found.

Note: This figure is not to scale to make details

tp near the steady-state value easier to see.
f d = --1- T
T
e ss

0.02 ∆ x b

0.02 ∆ x

0.9 ∆ x tr where,
ts
t r = rise time (from 10% to 90%)
∆x t s = settling time (to within 2-5% typ.)
0.5 ∆ x ∆ x = total displacement
T, f d = period and frequency - damped
b = overshoot
0.1 ∆ x t p = time to first peak
e ss = steady state error
–σ t π
x = ∆ xe cos ( ω d t ) ω d ≈ ---- (1)
tp

σ = ξω n (2)
–σ tp
b- = e
----- (3)
∆x
1
ξ = ----------------------------- (4)
π - 2
 -------
 t p σ + 1

Figure 3.23 Characterizing a second-order response (not to scale)

 differential equations - 3.26

Note: We can calculate these relationships using the complex homogenous form, and the
generic second order equation form.

2 2
A + 2ξω n A + ω n = 0
2 2 2
– 2ξω n ± 4ξ ω n – 4ω n
A = ----------------------------------------------------------- = σ ± jω d
2
– 2ξωn σ
---------------- = σ = – ξω n ω n = ------ (1)
2 –ξ
2 2 2
4ξ ω n – 4ω n
----------------------------------- = jω d
2
2 2 2 2
4ξ ω n – 4ω n = 4 ( – 1 )ω d

2
ωn – ξ ωn = ωd
2 2 2
ωn 1 – ξ = ωd
2 (2)

2 2
------ – ξ 2 σ
σ
------ = ω 2d
2 2
ξ ξ
1
1- ωd
2 ξ = --------------------
2 (3)
----
2
= ------ + 1
2 ωd
ξ σ ------ + 1
2
σ
The time to the first peak can be used to find the approximate decay constant
– σt
x ( t ) = C1e cos ( ω d t + C 2 )
π
ω d = ---- (4)
tp
– σt p
b ≈ ∆xe (1)

ln  ------
b (5)
 ∆x
σ = – ------------------
tp

Figure 3.24 Second order relationships between damped and natural frequency
 differential equations - 3.27

2.2

y(t)
2.0

t
0
4.0 5.0
Write a function of time for the graph. (Note: measure, using a ruler, to get val-
ues.) Find the natural frequency and damping coefficient to develop the dif-
ferential equation. Using the dashed lines determine the settling time.

ans.
t<4 y(t) = 0

t≥4 y(t) =

Figure 3.25 Drill problem: Find the equation given the response curve
 differential equations - 3.28

3.3.3 Other Responses

First-order systems have e-to-the-t type responses. Second-order systems add

another e-to-the-t response or a sinusoidal excitation. As we move to higher order linear
systems we typically add more e-to-the-t terms, and/or more sinusoidal terms. A possible
higher order system response is seen in Figure 3.26. The underlying function is a first-
order response that drops at the beginning, but levels out. There are two sinusoidal func-
tions superimposed, one with about one period showing, the other with a much higher fre-
quency.

Figure 3.26 An example of a higher order system response

The basic techniques used for solving first and second-order differential equations
can be applied to higher order differential equations, although the solutions will start to
become complicated for systems with much higher orders. The example in Figure 3.27
shows a fourth order differential equation. In this case the resulting homogeneous solution
yields four roots. The result in this case are two real roots, and a complex pair. The two
real roots result in e-to-the-t terms, while the complex pair results in a damped sinusoid.
The particular solution is relatively simple to find in this example because the non-homo-
geneous term is a constant.
 differential equations - 3.29

Given the homogeneous differential equation

d- 4
 ---- d 3 d 2
x + 13  ----- x + 34  ----- x + 42  ----- x + 20x = 5
d
 dt  dt  dt  dt
Guess a solution for the homogeneous equation,
At
xh = e
d- x = Ae At  ---- d- 2 2 At  ---- d- 3 3 At d- 4
 ---- 4 At
---- x = A e x = A e x = A e
dt h  dt h  dt h  dt h

Substitute the values into the differential equation and find a value for the unknown.
4 At 3 At 2 At At At
A e + 13A e + 34A e + 42Ae + 20e = 0
4 3 2
A + 13A + 34A + 42A + 20 = 0
A = – 1, – 10, – 1 – j, – 1 + j
–t – 10t –t
xh = C1 e + C2 e + C 3 e cos ( t + C 4 )

Guess a particular solution, and the solve for the coefficient.

xp = A d- x = 0 d- 2
 ---- d- 3
 ---- d- 4
 ----
---- x = 0
dt p  dt p  dt x p = 0  dt x p = 0

0 + 13 ( 0 ) + 34 ( 0 ) + 42 ( 0 ) + 20A = 5 A = 0.25

Figure 3.27 Solution of a higher order differential equation

The example is continued in Figure 3.28 and Figure 3.29 where the initial condi-
tions are used to find values for the coefficients in the homogeneous solution.
 differential equations - 3.30

Solve for the unknowns, assuming the system starts at rest and undeflected.
–t –10t –t
x ( t ) = C1 e + C 2 e + C 3 e cos ( t + C 4 ) + 0.25
0 = C 1 + C 2 + C 3 cos ( C 4 ) + 0.25 (1)
C 3 cos ( C 4 ) = – C 1 – C 2 – 0.25 (2)

d- x ( t ) = –C e – t – 10C e – 10t – C e – t cos ( t + C ) – C e – t sin ( t + C )

---- 1 2 3 4 3 4
dt h
0 = – C 1 – 10C 2 – C 3 cos ( C 4 ) – C 3 sin ( C 4 ) (3)

Equations (1) and (3) can be added to get the simplified equation below.
0 = – 9C 2 – C 3 sin ( C 4 ) + 0.25
C 3 sin ( C 4 ) = – 9C 2 + 0.25 (4)

d- 2
 ---- –t – 10t –t –t
x ( t ) = C e + 100C e + C e cos ( t + C ) + C 3 sin ( t + C 4 ) +
e
 dt h 1 2 3 4
–t –t
C 3 e sin ( t + C 4 ) – C 3 e cos ( t + C 4 )
0 = C1 + 100C 2 + C 3 cos ( C 4 ) + C 3 sin ( C 4 ) + C 3 sin ( C 4 ) – C 3 cos ( C 4 )
0 = C 1 + 100C 2 + 2C 3 sin ( C 4 ) (5)
Equations (4) and (5) can be combined.

0 = C 1 + 100C 2 + 2 ( – 9C 2 + 0.25 )
0 = – 17C 1 + 100C 2 + 0.5 (6)

d- 3
 ---- –t – 10t –t –t
x ( t ) = – C 1 e + ( – 1000 )C 2 e – 2C 3 e sin ( t + C 4 ) + 2C 3 e cos ( t + C 4 )
 dt h
0 = – C 1 + ( – 1000 )C 2 – 2C 3 sin ( C 4 ) + 2C 3 cos ( C 4 ) (7)

Figure 3.28 Solution of a higher order differential equation

 differential equations - 3.31

Equations (2 and (4) are substituted into equation (7).

0 = –C 1 + ( – 1000 )C 2 – 2 ( – 9C 2 + 0.25 ) + 2 ( – C 1 – C 2 – 0.25 )
0 = – 3C 1 + ( – 984 )C 2 – 1

C 1 =  – --------- C 2 – ---
984 1 (8)
3 3
Equations (6) and (8) can be combined.

0 = – 17   – --------- C 2 – --- + 100C 2 + 0.5

984 1
3 3
0 = 5676C2 + 6.1666667 C 2 = – 0.00109

C 1 =  – --------- ( – 0.00109 ) – ---

984 1
C 1 = 0.0242
 3  3
Equations (2) and (4) can be combined.
C 3 sin ( C 4 ) – 9C 2 + 0.25
---------------------------
- = ---------------------------------------
-
C 3 cos ( C 4 ) – C 1 – C 2 – 0.25
– 9 ( – 0.00109 ) + 0.25 C 4 = – 0.760
tan ( C 4 ) = ---------------------------------------------------------------------------
– ( 0.0242 ) – ( – 0.00109 ) – 0.25
Equation (4) can be used the find the remaining unknown.
C 3 sin ( – 0.760 ) = – 9 ( –0.00109 ) + 0.25 C 3 = – 0.377

The final response function is,

–t – 10t –t
x ( t ) = 0.0242e + ( – 0.00109 )e + ( – 0.377 )e cos ( t – 0.760 ) + 0.25

Figure 3.29 Solution of a higher order differential equation (cont’d)

In some cases we will have systems with multiple differential equations, or non-
linear terms. In these cases explicit analysis of the equations may not be feasible. In these
cases we may use other techniques, such as numerical integration, which will be covered
in later chapters.

3.4 RESPONSE ANALYSIS

Up to this point we have mostly discussed the process of calculating the system
response. As an engineer, obtaining the response is important, but evaluating the results is
more important. The most critical design consideration is system stability. In most cases a
system should be inherently stable in all situations, such as a car "cruise control". In other
 differential equations - 3.32

cases an unstable system may be the objective, such as an explosive device. Simple meth-
ods for determining the stability of a system are listed below:

1. If a step input causes the system to go to infinity, it will be inherently unstable.

2. A ramp input might cause the system to go to infinity; if this is the case, the sys-
tem might not respond well to constant change.
3. If the response to a sinusoidal input grows with each cycle, the system is proba-
bly resonating, and will become unstable.

Beyond establishing the stability of a system, we must also consider general per-
formance. This includes the time constant for a first-order system, or damping coefficient
and natural frequency for a second-order system. For example, assume we have designed
an elevator that is a second-order system. If it is under damped the elevator will oscillate,
possibly leading to motion sickness, or worse. If the elevator is over damped it will take
longer to get to floors. If it is critically damped it will reach the floors quickly, without
overshoot.

Engineers distinguish between initial setting effects (transient) and long term
effects (steady-state). The transient effects are closely related to the homogeneous solution
to the differential equations and the initial conditions. The steady-state effects occur after
some period of time when the system is acting in a repeatable or non-changing form. Fig-
ure 3.30 shows a system response. The transient effects at the beginning include a quick
rise time and an overshoot. The steady-state response settles down to a constant amplitude
sine wave.
 differential equations - 3.33

Steady-state

Transient

Note: the transient response is predicted with the homogeneous solution. The
steady state response in mainly predicted with the particular solution,
although in some cases the homogeneous solution might have steady state
effects, such as a non-decaying oscillation.

Figure 3.30 A system response with transient and steady-state effects

3.5 NON-LINEAR SYSTEMS

Non-linear systems cannot be described with a linear differential equation. A basic

linear differential equation has coefficients that are constant, and the derivatives are all
first order. Examples of non-linear differential equations are shown in Figure 3.31.

· 2
x+x = 5
Note: the sources of non-
·2 linearity are circled.
x +x = 5

·
x + log ( x ) = 5

·
x + ( 5t )x = 5

Figure 3.31 Examples of non-linear differential equations

 differential equations - 3.34

Examples of system conditions that lead to non-linear solutions are,

• aerodynamic drag
• forces that are a squared function of distance
• devices with non-linear responses

Explicitly solving non-linear differential equations can be difficult, and will typi-
cally involve complex solutions for simple problems.

3.5.1 Non-Linear Differential Equations

A non-linear differential equation is presented in Figure 3.32. It involves a person

ejected from an aircraft with a drag force coefficient of 0.8. (Note: This coefficient is cal-
culated using the drag coefficient and other properties such as the speed of sound and
cross sectional area.) The FBD shows the sum of forces, and the resulting differential
equation. The velocity squared term makes the equation non-linear, and so it cannot be
analyzed with the previous methods. In this case the terminal velocity is calculated by set-
ting the acceleration to zero. This results in a maximum speed of 126 kph.
 differential equations - 3.35

Consider the differential equation for a 100kg human ejected from an airplane. The
aerodynamic drag will introduce a squared variable, therefore making the equation
non-linear.
2 2
0.8 Ns
--------- ( y' ) · ) 2 – Mg = – My··
m
2 ∑ F y = 0.8 ( y
2 2
100kgy·· + 0.8 Ns N-
--------- ( y· ) = 100kg9.81 -----
2 kg
m
2 2
100kgy·· + 0.8 Ns
--------- ( y· ) = 981N
y 2
Mg m
2
s - ( y· ) 2 = 981kg ----
m- ------ m-
100kgy·· + 0.8kg ----
2 2 2
–1 2 s m –2 s
100y·· + 0.8m ( y· ) = 981ms
–3 –1 2 –2
y·· + 8 × 10 m ( y· ) = 9.81ms
The terminal velocity can be found be setting the acceleration to zero.

–3 –1 2 –2
( 0 ) + 8 × 10 m ( y· ) = 9.81ms
–2
9.81ms - = -------------------9.81 - m 2 s – 2 = 35.0 m
y· = ------------------------------ ---- = 126 km
-------
–3 –1 –3 s h
8 × 10 m 8 × 10

Figure 3.32 Development of a non-linear differential equation

The equation can also be solved using explicit integration, as shown in Figure
3.33. In this case the equation is separated and rearranged to isolate the ’v’ terms on the
left, and time on the right. The term is then integrated in Figure 3.34 and Figure 3.35. The
final form of the equation is non-trivial, but contains e-to-t terms, as we would expect.
 differential equations - 3.36

An explicit solution can begin by replacing the position variable with a velocity variable
and rewriting the equation as a separable differential equation.
·· –1 · 2 –2
100y + 0.8m ( y ) = 981ms
· –1 2 –2
100v + 0.8m v = 981ms
dv –1 2 –2
100 ------ + 0.8m v = 981ms
dt
dv –2 –1 2
100 ------ = 981ms – 0.8m v
dt
100
-------------------------------------------------
–2 –1 2
dv = dt
981ms – 0.8m v
100
--------------------
–1
–0.8m
∫ 981 –2 2- dv = ∫ dt
--------------------------------------------
-------------------- ms + v
–1
– 0.8m
– 125m
∫ --------------------------------------------
2
v – 1226.25m s
2 –2
- dv = t + C1

– 125m
∫ -----------------------------------------------------------------
m  m
- dv = t + C1
v + 35.02 ---- v – 35.02 ----
 s s

Figure 3.33 Developing an integral

 differential equations - 3.37

This can be reduced with a partial fraction expansion.

A B
∫ --------------------------------
 v + 35.02 m   m
- dv = t + C 1
- + -------------------------------

----  v – 35.02 ----
 s s
Av – A  35.02 ---- + Bv + B  35.02 m ---- = – 125m
m
s s
m
v ( A + B ) + 35.02 ---- ( – A + B ) = – 125m
s
A+B = 0 A = –B

35.02 m---- ( – A + B ) = – 125m

s
125 - s
( – ( –B ) + B ) = – ------------ B = – 1.785s
35.02
A = 1.785s

1.785s – 1.785s - dv = t + C
∫ --------------------------------- + -------------------------------
 v + 35.02 m   v – 35.02 m 
1
---- ----
 s  s
The integral can then be solved using an identity from the integral table. In this case
the integration constants can be left off because they are redundant with the one on
the right hand side.

1.785s ln v + 35.02 m ---- – 1.785s ln v – 35.02 m

---- = t + C 1
s s
m
v + 35.02 ----
s- = t + C –1 ln a + bx- + C
1.785s ln ---------------------------
m
v – 35.02 ----
1 ∫ ( a + bx ) dx = -----------------------
b
s
t -+C
v + 35.02 m ---- --------------- 1
s- = e 1.785s
---------------------------
v – 35.02 m ----
s
t -
m ---------------
v + 35.02 ---- C 1.785s
s- = e 1 e
---------------------------
v – 35.02 m ----
s

Figure 3.34 Solution of the integral

 differential equations - 3.38

m t -
v + 35.02 ---- ---------------
1.785s
s
---------------------------- = C 2 e
m
v – 35.02 ----
s
An initial velocity of zero can be assumed to find the value of the integration constant

m 0 -
0 + 35.02 ---- ---------------
1.785s 1 = C2
s
---------------------------- = C 2 e
m
0 – 35.02 ----
s
This can then be simplified, and the absolute value sign eliminated.

m t -
v + 35.02 ---- ---------------
s 1.785s
---------------------------- = ± e
m
v – 35.02 ----
s
t -
--------------- t -
---------------
m 1.785s − m 1.785s
v + 35.02 ---- = ± ve + 35.02 ---- e
s s
t t
 ----------------
m
----------------
m
v1 − −
1.785s 1.785s
+ e  = + 35.02 ---
- e – 35.02 ----
  s s
 ----------------
t

−  m −
0 = 35.02 ----  +
1.785s 1 – 1
+ =  ------------- = ---
m e – 1 1–1 0
v = 35.02 ----  ---------------------------- - ----------------
s s 1+1  −  1 + 1 2
t -
--------------- 
 1−+e
1.785s

 ---------------
t -

m e 1.785s
– 1 
v = 35.02 ----  ----------------------- -
s ----------------
t
1 + e 1.785s


Figure 3.35 Solution of the integral and application of the initial conditions

As evident from the example, non-linear equations are involved and don’t utilize
routine methods. Typically the numerical methods discussed in the next chapter are pre-
ferred.

3.5.2 Non-Linear Equation Terms

If our models include a device that is non-linear and we want to use a linear tech-
 differential equations - 3.39

nique to solve the equation, we will need to linearize the model before we can proceed. A
non-linear system can be approximated with a linear equation using the following method.

1. Pick an operating point or range for the component.

2. Find a constant value that relates a change in the input to a change in the output.
3. Develop a linear equation.
4. Use the linear equation for the analysis.

A linearized differential equation can be approximately solved using known tech-

niques as long as the system doesn’t travel too far from the linearized point. The example
in Figure 3.36 shows the linearization of a non-linear equation about a given operating
point. This equation will be approximately correct as long as the first derivative doesn’t
move too far from 100. When this value does, the new velocity can be calculated.
 differential equations - 3.40

Assume we have the non-linear differential equation below. It can be solved by lin-
earizing the value about the operating point

Given,
2
y· + 4y = 200 y ( 0 ) = 10

We can make the equation linear by replacing the velocity squared term with the
velocity times the actual velocity. As long as the system doesn’t vary too much
from the given velocity the model should be reasonably accurate.

y· = ± 200 – 4y

y· ( 0 ) = ± 200 – 4 ( 10 ) = ± 12.65

12.65y· + 4y = 20
This system may now be solved as a linear differential equation. If the velocity
(first derivative of y) changes significantly, then the differential equation should
be changed to reflect this.
Homogeneous:
12.65y· + 4y = 0
12.65A + 4 = 0 A = – 0.316
– 0.316t
y h = Ce
Particular:
yp = A
12.65 ( 0 ) + 4A = 200 A = 50
Initial conditions:
– 0.316t
y ( t ) = Ce + 50
0
10 = Ce + 50 C = – 40
– 0.316t
y ( t ) = – 40e + 50

Figure 3.36 Linearizing a differential equation

 differential equations - 3.41

If the velocity (first derivative of y) changes significantly, then the differential

equation should be changed to reflect this. For example we could decide to
recalculate the equation value after 0.1s.

– 0.316 ( 0.1 )
y ( 0.1 ) = – 40e + 50 = 11.24
d- y ( 0.1 ) = – 40 ( – 0.316 )e – 0.316 ( 0.1 ) = 12.25
---- Note: a small change
dt
12.25y' + 4y = 20
Now recalculate the solution to the differential equation.

Homogeneous:
12.25y· + 4y = 0
12.25A + 4 = 0 A = – 0.327
– 0.327t
y h = Ce
Particular:
yp = A
12.25 ( 0 ) + 4A = 200 A = 50
Initial conditions:
– 0.327t
y ( t ) = Ce + 50
0.1
11.24 = Ce + 50 C = – 35.070575
– 0.316t
y ( t ) = – 35.07e + 50
Notice that the values have shifted slightly, and as the analysis progresses the
equations will adjust slowly. Higher accuracy can be obtained using smaller
steps in time.

Figure 3.37 Linearizing a differential equation

3.5.3 Changing Systems

In practical systems, the forces at work are continually changing. For example a
system often experiences a static friction force when motion is starting, but once motion
starts it is replaced with a smaller kinetic friction. Another example is tension in a cable.
When in tension a cable acts as a spring. But, when in compression the force goes to zero.

Consider the example in Figure 3.38. A mass is pulled by a springy cable. The
 differential equations - 3.42

right hand side of the cable is being pulled at a constant rate, while the block is free to
move, only restricted by friction forces and inertia. At the beginning all components are at
rest and undeflected.

x2
x 1 = 0.1 m
---- t
N- s
K s = 1000 ---
m
µ k = 0.1
M = 100kg
µ s = 0.3

An FBD and equation can be developed for the system. The friction force will be left as
a variable at this point.

For the cable/spring in tension x 1 – x 2 Š 0

FF
Ks ( x1 – x2 ) ··
M = 100kg ∑ F x = – F F + K s ( x 1 – x 2 ) = Mx 2
··
Mx 2 N-  0.1 m
---- t – x 2 = 100kgx2
··
– F F + 1000 ---
m  s 
·· N- x = 1000 --- N- 0.1 m
100kgx 2 + 1000 --- 2 ---- t – F F
m m s
·· F F-
N -x = 1 ------- N -t – --------------
x 2 + 10 ---------
kgm 2 kgs 100kg
·· –2 FF
x 2 + 10s x 2 = 1 ---- m- t – --------------
-
3 100kg
s
For the cable/spring in compression x1 – x2 < 0
FF
··
M = 100kg ∑ Fx = –FF = Mx2
··
Mx 2 ··
– F F = 100kgx 2
··
100kgx 2 = – F F
·· FF
x 2 = – ---------------
100kg

Figure 3.38 A differential equation for a mass pulled by a springy cable

 differential equations - 3.43

x2
x 1 = 0.1 m
---- t
N- s
K s = 1000 ---
m
µ k = 0.1
M = 100kg
µ s = 0.3

An FBD and equation can be developed for the system. The friction force will be left as
a variable at this point.
N- = 981N
N = 100kg9.81 -----
kg
d- x £ 0 m
static friction ---- ----
dt 2 s
0N £ F F < µ s N < 294.3N

d- x > 0 m
kinetic friction ---- ----
dt 2 s
F F = µ k N = 98.1N

Figure 3.39 Friction forces for the mass

The analysis of the system begins by assuming the system starts at rest and undeflected.
In this case the cable/spring will be undeflected with no force, and the mass will be
experiencing static friction. Therefore the block will stay in place until the cable
stretches enough to overcome the static friction.
··
x2 = 0 x2 = 0 F F = 294.3N
·· –2 FF
m
x 2 + 10s x 2 = 1 ----- t – ---------------
3 100kg
s
–2 m-t – 294.3N
0 + 10s 0 = 1 ---- -----------------
3 100kg
s
m- t = 294.3kgm
1 ---- ------------------------
3 2
s 100kgs
t = 2.943s
Therefore the system is static from 0 to 2.943s

Figure 3.40 Analysis of the object before motion begins

 differential equations - 3.44

After motion begins the object will only experience kinetic friction, and continue to
accelerate until the cable/spring becomes loose in compression. This stage of motion
requires the solution of a differential equation.

·· –2 98.1N-
m- t – --------------
x 2 + 10s x 2 = 1 ----
3 100kg
s
For the homogeneous,
·· –2
x 2 + 10s x 2 = 0
–2 –1
A + 10s = 0 A = ± 3.16js
x h = C 1 sin ( 3.16t + C 2 )
For the particular,
· ··
x p = At + B xp = A xp = 0

–2 98.1N-
m- t – --------------
0 + 10s( At + B ) = 1 ----
3 100kg
s
–2 m-
10s A = 1 ---- A = 0.1 m
----
3 s
s
–2 98.1N
10s B = – --------------- B = – 0.0981m
100kg

Figure 3.41 Analysis of the object after motion begins

 differential equations - 3.45

For the initial conditions,

x ( 2.943s ) = 0m d- x ( 2.943s ) = 0 m
---- ----
dt s

x ( t ) = C 1 sin ( 3.16t + C 2 ) + 0.1 m

---- t – 0.0981m
s

0 = C 1 sin ( 3.16 ( 2.943s ) + C 2 ) + 0.1 m

---- ( 2.943s ) – 0.0981m
s
C 1 sin ( 9.29988 + C2 ) = – 0.1962
d- x ( t ) = 3.16C cos ( 3.16t + C ) + 0.1 m
---- ----
dt 1 2 s
0 = 3.16C 1 cos ( 3.16 ( 2.943 ) + C 2 ) + 0.1 m
----
s
C 1 cos ( 9.29988 + C 2 ) = – 0.0316
C 1 sin ( 9.29988 + C 2 )
- = –------------------
0.1962-
------------------------------------------------------
C 1 cos ( 9.29988 + C 2 ) – 0.0316

tan ( 9.29988 + C 2 ) = 6.209 C 2 = ( – 7.889 + π n )rad nŒI

– 0.1962
C 1 = --------------------------------------------------- = –0.199m
sin ( 9.29988 – 7.889 )
x ( t ) = – 0.199 m sin ( 3.16t – 7.889rad ) + 0.1 m ---- t – 0.0981m
s
----- x ( t ) = – 0.199 ( 3.16 )m cos ( 3.16t – 7.889rad ) + 0.1 m
d ----
dt s

Figure 3.42 Analysis of the object after motion begins

 differential equations - 3.46

The equation of motion changes after the cable becomes slack. This point in time can
be determined when the displacement of the block equals the displacement of the
cable/spring end.

m m
0.1 ---- t = – 0.199 m sin ( 3.16t – 7.889rad ) + 0.1 ---- t – 0.0981m
s s
– 0.199 m sin ( 3.16t – 7.889rad ) = 0.0981m
3.16t – 7.889 + π n = – 0.51549413 t = 3.328s
d- m
x ( 3.328 ) = 0.137m ---- x ( 2.333 ) = 0.648 ----
dt s
After this the differential equation without the cable/spring is used.

98.1N m
x 2'' = – --------------- = – 0.981 ---2-
100kg s
x 2 =  – 0.981 ----2 t + C 1
· m
s
0.648 ---- =  – 0.981 ----2 ( 3.328s ) + C 1
m m
s s
m
C 1 = 3.913 ----
s
 0.981 m  2 m
x 2 = – ------------- ----2 t + 3.913 ---- t + C 2
 2 s  s
0.137m =  – ------------- ---2- ( 3.328s ) + 3.913 ---- ( 3.328s ) + C 2
0.981 m 2 m
 2  s
s
C 2 = – 7.453m

x 2 ( t ) =  – ------------- ----2 t + 3.913 ---- t – 7.453m

0.981 m 2 m
 2 s  s
This motion continues until the block stops moving.

0 =  –0.981 ----2 t + 3.913 ----

m m
  s
s
t = 3.989s

The solution can continue, considering when to switch the analysis conditions.

Figure 3.43 Determining when the cable become slack

 differential equations - 3.47

3.6 CASE STUDY

A typical vibration control system design is described in Figure 3.44.

The model to the left describes a piece of

F M reciprocating industrial equipment. The
mass of the equipment is 10000kg. The
equipment operates such that a force of
1000N with a frequency of 2Hz is exerted
y on the mass. We have been asked to design
a vibration isolation mounting system. The
criteria we are given is that the mounts
should be 30cm high when unloaded, and
25cm when loaded with the mass. In addi-
tion, the oscillations while the machine is
running cannot be more than 2cm total. In
total there will be four mounts mounted
around the machine. Each isolator will be
composed of a spring and a damper.

Figure 3.44 A vibration control system

There are a number of elements to the design and analysis of this system, but as
usual the best place to begin is by developing a free body diagram, and a differential equa-
tion. This is done in Figure 3.45.
 differential equations - 3.48

F
+ ∑ Fy = F – 4K s y – 4K d y· + Mg = My··
M y
My'· + 4K d y· + 4K s y = F + Mg
4K d y· 4K s y F- + g
··y + ------------
- + ------------ = ---- 4K s y 4K d y· Mg
M M M
4K d y· 4K s y 1000N - sin ( 2 ( 2 π )t ) + 9.81ms – 2
y·· + ---------------------
- + ---------------------
- = ---------------------
10000Kg 10000Kg 10000Kg
–1 –1 –2 –2
y·· + 0.0004Kg K d y· + 0.0004Kg K s y = 0.1ms sin ( 4 π t ) + 9.81ms

Figure 3.45 FBD and derivation of equation

Using the differential equation, the spring values can be found by assuming the
machine is at rest. This is done in Figure 3.46.

When the system is at rest the equation is simplified; the acceleration and velocity
terms both become zero. In addition, we will assume that the cyclic force is not
applied for the unloaded/loaded case. This simplifies the differential equation by
eliminating several terms.
–1 –2
0.0004Kg K s y = 9.81ms
Now we can consider that when unloaded the spring is 0.30m long, and after loading
the spring is 0.25m long. This will result in a downward compression of 0.05m, in
the positive y direction.
–1 –2
0.0004Kg K s ( 0.05m ) = 9.81ms
9.81 –2 –1
K s = -------------------------------- Kgms m
0.0004 ( 0.05 )
–1
∴K s = 491KNm

Figure 3.46 Calculation of the spring coefficient

The remaining unknown is the damping coefficient. At this point we have deter-
mined the range of motion of the mass. This can be done by developing the particular
 differential equations - 3.49

solution of the differential equation, as it will contain the steady-state oscillations caused
by the forces as shown in Figure 3.47.

·· –1 · –1 –1 –2 –2
y + 0.0004Kg K d y + 0.0004Kg ( 491KNm )y = 0.1ms sin ( 4 π t ) + 9.81ms
·· –1 · –2 –2 –2
y + 0.0004Kg K d y + 196s y = 0.1ms sin ( 4 π t ) + 9.81ms
The particular solution can now be found by guessing a value, and solving for the
coefficients. (Note: The units in the expression are uniform (i.e., the same in each
term) and will be omitted for brevity.)
y = A sin ( 4 π t ) + B cos ( 4 π t ) + C
y' = 4 π A cos ( 4 π t ) – 4 π B sin ( 4 π t )
2 2
y'' = – 16 π A sin ( 4 π t ) – 16 π B cos ( 4 π t )
2 2
∴( –16 π A sin ( 4 π t ) – 16 π B cos ( 4 π t ) ) + 0.0004K d ( 4 π A cos ( 4 π t ) – 4 π B sin ( 4 π t ) )
+ 196 ( A sin ( 4 π t ) + B cos ( 4 π t ) + C ) = 0.1 sin ( 4 π t ) + 9.81
2
– 16 π B + 0.0004K d 4 π A + 196A = 0
–6
B = A ( 31.8 × 10 K d + 1.24 )
2
– 16 π A + 0.0004K d ( – 4 π B ) + 196A = 0.1
2 –3
A ( –16 π + 196 ) + B ( – 5.0 × 10 K d ) = 0.1
2 –6 –3
A ( – 16 π + 196 ) + A ( 31.8 × 10 K d + 1.24 ) ( – 5.0 × 10 K d ) = 0.1
0.1
A = --------------------------------------------------------------------------------------------------------------------------------------
2 –6 –3
-
– 16 π + 196 + ( 31.8 × 10 K d + 1.24 ) ( – 5.0 × 10 K d )
0.1
A = --------------------------------------------------------------------------------------------------------
2 –9 –3
-
K d ( – 159 × 10 ) + K d ( – 6.2 × 10 ) + 38.1
–6
3.18 × 10 K d – 0.124
B = --------------------------------------------------------------------------------------------------------
2 –9 –3
-
K d ( – 159 × 10 ) + K d ( –6.2 × 10 ) + 38.1
–2
C = 9.81ms

Figure 3.47 Particular solution of the differential equation

The particular solution can be used to find a damping coefficient that will give an
overall oscillation of 0.02m, as shown in Figure 3.48. In this case Mathcad was used to
find the solution, although it could have also been found by factoring out the algebra, and
finding the roots of the resulting polynomial.
 differential equations - 3.50

In the previous particular solution the values were split into cosine and sine compo-
nents. The magnitude of oscillation can be calculated with the Pythagorean formula.
2 2
magnitude = A +B
2
( 0.1 ) +  ( 3.18 ⋅ 10 )K d – 0.124
2 –6
magnitude = ---------------------------------------------------------------------------------------------------------------
2 –9 –3
K d ( – 159 ⋅ 10 ) + K d ( – ( 6.2 ⋅ 10 ) ) + 38.1
The design requirements call for a maximum oscillation of 0.02m, or a magnitude of
0.01m.
2
( 0.1 ) +  ( 3.18 ⋅ 10 )K d – 0.124
2 –6
 
0.01 = ---------------------------------------------------------------------------------------------------------------
2 –9 –3
K d ( – 159 ⋅ 10 ) + K d ( – ( 6.2 ⋅ 10 ) ) + 38.1
A given-find block was used in Mathcad to obtain a damper value of,

K d = 3411N ---s- Aside: the Mathcad solution

m

Figure 3.48 Determining the damping coefficient

The values of the spring and damping coefficients can be used to select actual
components. Some companies will design and build their own components. Components
can also be acquired by searching catalogs, or requesting custom designs from other com-
panies.
 differential equations - 3.51

3.7 SUMMARY

• First and second-order differential equations were analyzed explicitly.

• First and second-order responses were examined.
• The topic of analysis was discussed.
• A case study looked at a second-order system.
• Non-linear systems can be analyzed by making them linear.

3.8 PRACTICE PROBLEMS

1. The mass, M, illustrated below starts at rest. It can slide across a surface, but the motion is
opposed by viscous friction (damping) with the coefficient B. Initially the system starts at rest,
when a constant force, F, is applied. Write the differential equation for the mass, and solve the
differential equation. Leave the results in variable form.

M F

2. The following differential equation was derived for a mass suspended with a spring. At time 0s
the system is released and allowed to drop. It then oscillates. Solve the differential equation to
find the motion as a function of time.

+ ··
N ∑ Fy = K s y – Mg = – My
K s = 100 ----
m  100 ---
N-
y – ( 1Kg )  9.81 ------- = ( – 1Kg )y
N ··
 m  Kg
y  1 Nm
-------- y·· +  100 ---- y = 9.81N
N
M = 1Kg
 2  m 
s
( 1Kg )y +  100 -----------
·· Kgm- Kgm-
y = 9.81 -----------
 2  2
FBD: Ks y ms s
·· –2 –2
M y + ( 100s )y = 9.81ms
· –1
y = 0m y 0 = 0ms
Mg 0

3. Solve the following differential equation with the three given cases. All of the systems have a
 differential equations - 3.52

step input ’y’ and start undeflected and at rest.

·· · 2 ·
x + 2ξωn x + ω n x = y initial conditions x = 0
x = 0
y = 1
case 1: ξ = 0.5 ω n = 10

case 2: ξ = 1 ω n = 10

case 3: ξ = 2 ω n = 10

4. Solve the following differential equation with the given initial conditions and draw a sketch of
the first 5 seconds. The input is a step function that turns on at t=0.
·· ·
0.5V o + 0.6V o + 2.1V o = 3V i + 2 initial conditions V i = 5
Vo = 0
·
Vo = 0

5. Solve the following differential equation with the given initial conditions and draw a sketch of
the first 5 seconds. The input is a step function that turns on at t=0.
·· ·
0.5V o + 0.6V o + 2.1V o = 3V i + 2 initial conditions V i = 5
Vo = 0
·
Vo = 1

6. a) Write the differential equations for the system below. Solve the equations for x assuming that
 differential equations - 3.53

the system is at rest and undeflected before t=0. Also assume that gravity is present.

Kd1 = 1 Ns/m
Ks1 = 1 N/m Kd = 1 Ns/m

M = 1 kg
x1
Kd2 = 1Ns/m
Ks2 = 1 N/m

F=1N x

Ks = 1 N/m
x2
F=1N

b) State whether each system is first or second-order. If the system if first-order

find the time constant. If it is second-order find the natural frequency and
damping ratio.

7. Solve the following differential equation with the three given cases. All of the systems have a
sinusoidal input ’y’ and start undeflected and at rest.
·· · 2 ·
x + 2ξωn x + ω n x = y initial conditions x = 0
x = 0
y = sin ( t )
case 1: ξ = 0.5 ω n = 10

case 2: ξ = 1 ω n = 10

case 3: ξ = 2 ω n = 10

8. A spring damper system supports a mass of 34N. If it has a spring constant of 20.6N/cm, what
is the systems natural frequency?

9. Using a standard lumped parameter model the weight is 36N, stiffness is 2.06*103 N/m and
damping is 100Ns/m. What are the natural frequency (Hz) and damping ratio?
 differential equations - 3.54

10. What is the differential equation for a second-order system that responds to a step input with
an overshoot of 20%, with a delay of 0.4 seconds to the first peak?

11. A system is to be approximated with a mass-spring-damper model using the following param-
eters: weight 28N, viscous damping 6Ns/m, and stiffness 36N/m. Calculate the undamped nat-
ural frequency (Hz) of the system, the damping ratio and describe the type of response you
would expect if the mass were displaced and released. What additional damping would be
required to make the system critically damped?
·· ·
Mx + K d x + K s x = F

12. Solve the differential equation below using homogeneous and particular solutions. Assume
the system starts undeflected and at rest.
··· ·· ·
θ + 40θ + 20θ + 2θ = 4

13. What would the displacement amplitude after 100ms for a system having a natural frequency
of 13 rads/sec and a damping ratio of 0.20. Assume an initial displacement of 50mm, and a
steady state displacement of 0mm. (Hint: Find the response as a function of time.)

14. Determine the first order differential equation given the graphical response shown below.
Assume the input is a step function.
x
4

t(s)
0 1 2 3 4

15. Explain with graphs how to develop first and second-order equations using experimental data.

16. The second order response below was obtained experimentally. Determine the parameters of
 differential equations - 3.55

the differential equation that resulted in the response assuming the input was a step function.
0.5s
1s

10

0 t(s)
17. Develop equations (function of time) for the first and second order responses shown below.
x (m)
5m

0m
0.1s 0.2s 0.3s 0.4s 0.5s 0.6s 0.7s 0.8s t (s)
x (m)
5m

0m
1s 3s t (s)

-10m
 differential equations - 3.56

3.9 PRACTICE PROBLEM SOLUTIONS

1.
B
-----
– FM- – M t F FM-
x ( t ) = -----------
2
e – --- t + --------
2
B B B

2.
homogeneous: guess At · At ·· 2 At
yh = e y h = Ae yh = A e
2 At –2 At
A e + ( 100s )e = 0
2 –2 –1
A = –100s A = ± 10 js
y h = C 1 cos ( 10t + C 2 )

particular: guess · ··
yp = A yp = 0 yp = 0
–2 –2
( 0 ) + ( 100s )A = 9.81ms
–2 –2
( 100s )A = 9.81ms
–2
9.81ms -
A = ---------------------
–2
= 0.0981m
100s
y p = 0.0981m

initial conditions: y = y h + y p = C 1 cos ( 10t + C 2 ) + 0.0981m

y' = – 10C 1 sin ( 10t + C 2 )

for d/dt y0 = 0m:

0 = – 10C 1 sin ( 10 ( 0 ) + C 2 ) C2 = 0
for y0 = 0m:
0 = C 1 cos ( 10 ( 0 ) + ( 0 ) ) + 0.0981m
– 0.0981m = C 1 cos ( 0 ) C 1 = – 0.0981m

y ( t ) = ( – 0.0981m ) cos ( 10t ) + 0.0981m

 differential equations - 3.57

3.
– 5t
case 1: x ( t ) = – 0.0115e cos ( 8.66t – 0.524 ) + 0.010

– 10t – 10t
case 2: x ( t ) = – 0.010e – 0.10te + 0.010

case 3: – 6 – 37.32t – 2.679t

x ( t ) = 775 ⋅ 10 e – 0.0108e + 0.010

4.
– 0.6t
V o ( t ) = – 8.465e sin ( 1.960t + 1.274 ) + 8.095
or – 0.6t
V o ( t ) = – 8.465e cos ( 1.960t – 0.2971 ) + 8.095

5.

– 0.6t
V 0 ( t ) = – 8.331e cos ( 1.96t – 0.238 ) + 8.095

6.
a) b)
–t
x1 ( t ) = e – 1 – 0.5t
x ( t ) = – 12.485e cos ( 0.866t – 0.524 ) + 10.81
–t
x 2 ( t ) = 2e – 2
ζ = 0.5 ωn = 1
τ = 1

7.
case 1: – 5t
x ( t ) = – 0.00117e sin ( 8.66t – 1.061 ) + 0.0101 sin ( t – 0.101 )
case 2: –3 – 10t –3 – 10t –3
x ( t ) = ( 1.96 ⋅ 10 )e + ( 9.9 ⋅ 10 )te + ( 9.9 ⋅ 10 ) sin ( t + 0.20 )
case 3: –3 – 2.679t –6 – 37.32t –3
x ( t ) = ( 3.5 ⋅ 10 )e – ( 18 ⋅ 10 )e + ( 9.4 ⋅ 10 ) sin ( t + 0.382 )

8. 24.37 rad/sec

9. fn=3.77Hz, damp.=.575
 differential equations - 3.58

10.
·· ·
x + 8.048x + 77.88x = F ( t )

11.

Given s N
28N
K d = 6N ---- K s = 36 ----
M = ----------------- = 2.85kg
m m N
9.81 ------
kg
The typical transfer function for a mass-spring-damper systems is,

·· · K d Ks
x + x  ------ + x  ----- = -----
F
 M  M M
The second order parameters can be calculated from this.
·· · 2
x + x ( 2ζω n ) + x ( ω n ) = y ( t )

N kgm
36 ---- 36 ---------2-
Ks m = ms - = –2 rad
ωn = ----- = ---------------- ---------------- 12.63s = 3.55 --------- = 0.6Hz
M 2.85kg 2.85kg s
K d
 ----- s
- 6N ----
 M m
ξ = ------------ = --------------------------------------------- = 0.296
2ω n rad
2 ( 3.55 ) --------- 2.85kg
s
2 rad
ω d = ω n 1 – ξ = 3.39 ---------
s
If pulled and released the system would have a decaying oscillation about 0.52Hz

A critically damped system would require a damping coefficient of....

K -----d-
 M Kd Ns
ξ = ------------ = --------------------------------------------- = 1.00 K d = 20.2 ------
m
2ω n rad
2 ( 3.55 ) --------- 2.85kg
s

12.

– 6 – 39.50t 0.1383t – 0.3368t

θ ( t ) = – 66 ⋅ 10 e – 3.216e + 1.216e + 2.00
 differential equations - 3.59

13.
π
sin  12.74t + ---
– 2.6t – 2.6t
y ( t ) = 0.05e cos ( 12.74t ) = 0.05e
2

14.

x
4

t(s)
0 1 2 3 4
τ = 1
Given the equation form,
· 1
x + --- x = A
τ
The values at steady state will be
·
x = 0 x = 4
So the unknown ‘A’ can be calculated.
1 A = 4
0 + --- 4 = A
1
· 1
x + --- x = 4
1
·
x+x = 4

15.
Key points:

First-order:
find initial final values
find time constant with 63% or by slope
use these in standard equation

Second-order:
find damped frequency from graph
find time to first peak
use these in cosine equation
 differential equations - 3.60

16.
For the first peak:
b- – σt
----- = e p
∆x
π
2- – σ0.5 ω d ≈ ----
----- = e tp
10

ln  ------ = – σ0.5
2
 10
σ = – 2 ln  ------ = 3.219
2
 10 1
ξ = -----------------------------
For the damped frequency: π - 2
 ------- +1
2π  t p σ
ω d = ------ = 2π
1s
These values can be used to find the damping coefficient and natural frequency
3.219
σ = ξω n ω n = -------------
ξ
2
ωd = ωn 1 – ξ

3.219 2
2π = ------------- 1 – ξ
ξ
2
2π - 2
 ------------ 1 – ξ-
= -------------
 3.219 2
ξ
1
2π - 2
 ------------ 1 ξ = ------------------------------
- = 0.4560
+ 1 = ----2-
 3.219 2π - 2
 ------------
ξ  3.219 + 1

3.219 3.219
ω n = ------------- = ---------------- = 7.059
ξ 0.4560

This leads to the final equation using the steady state value of 10
·· · 2
x + 2ξωn x + ω n x = F
·· · 2
x + 2 ( 0.4560 ) ( 7.059 )x + ( 7.059 ) x = F
·· ·
x + 6.438x + 49.83x = F
( 0 ) + 6.438 ( 0 ) + 49.83 ( 10 ) = F F = 498.3
·· ·
x + 6.438x + 49.83x = 498.3
 differential equations - 3.61

17.
–t
 ----------
0.18 – 0.693 t
x(t) = 51 – e  x ( t ) = – 10e cos ( πt )
 

3.10 ASSIGNMENT PROBLEMS

1. A mass-spring-damper system has a mass of 10Kg and a spring coefficient of 1KN/m. Select a
damping coefficient so that the system will have an overshoot of 20% for a step input.
 numerical methods - 4.1

4. NUMERICAL ANALYSIS

Topics:
• State variable form for differential equations
• Numerical integration with software and calculators
• Numerical integration theory: first-order, Taylor series and Runge-Kutta
• Using tabular data
• A design case

Objectives:
• To be able to solve systems of differential equations using numerical methods.

4.1 INTRODUCTION

For engineering analysis it is always preferable to develop explicit equations that

include symbols, but this is not always practical. In cases where the equations are too
costly to develop, numerical methods can be used. As their name suggests, numerical
methods use numerical calculations (i.e., numbers not symbols) to develop a unique solu-
tion to a differential equation. The solution is often in the form of a set of numbers, or a
graph. This can then be used to analyze a particular design case. The solution is often
returned quickly so that trial and error design techniques may be used. But, without a sym-
bolic equation the system can be harder to understand and manipulate.

This chapter focuses on techniques that can be used for numerically integrating
systems of differential equations.

4.2 THE GENERAL METHOD

The general process of analyzing systems of differential equations involves first

putting the equations into standard form, and then integrating these with one of a number
of techniques. State variable equations are the most common standard equation form. In
this form all of the equations are reduced to first-order differential equations. These first-
order equations are then easily integrated to provide a solution for the system of equations.
 numerical methods - 4.2

4.2.1 State Variable Form

At any time a system can be said to have a state. Consider a car for example, the
state of the car is described by its position and velocity. Factors that are useful when iden-
tifying state variables are:

• The variables should describe energy storing elements (potential or kinetic).

• The variables must be independent.
• They should fully describe the system elements.

After the state variables of a system have been identified, they can be used to write
first-order state variable equations. The general form of state variable equations is shown
in Figure 4.1. Notice that the state variable equation is linear, and the value of x is used to
calculate the derivative. The output equation is not always required, but it can be used to
calculate new output values.

 ----
d-
state variable equation
 dt x = Ax + Bu
y = Cx + Du output equation
where,
x = state/output vector (variables such as position)
u = input vector (variables such as input forces)
A = transition matrix relating outputs/states
B = matrix relating inputs to outputs/states
y = non-state value that can be found directly (i.e. no integration)
C = transition matrix relating outputs/states
D = matrix relating inputs to outputs/states

Figure 4.1 The general state variable form

An example of a state variable equation is shown in Figure 4.2. As always, the

FBD is used to develop the differential equation. The resulting differential equation is sec-
ond-order, but this must be reduced to first-order. Using the velocity variable, ’v’ the sec-
ond-order differential equation can be reduced to a first-order equation. An equation is
also required to define the velocity as the first derivative of the position, ’x’. In the exam-
ple the two state equations are manipulated into a matrix form. This form can be useful,
and may be required for determining a solution. For example, HP calculators require the
matrix form, while TI calculators use the equation forms. Software such as Mathcad can
use either form. The main disadvantage of the matrix form is that it will only work for lin-
 numerical methods - 4.3

ear differential equations.

Given the FBD shown below, the differential equation for the system is,

F K d x·
M
x Ksx
+
∑ Fx = – F – K d x· – K s x = Mx··
– F – K d x· – K s x = Mx··
Mx··

The equation is second-order, so two state variables will be needed. One obvious
choice for a state variable in this equation is ’x’. The other choice can be the veloc-
ity, ’v’. Equation (1) defines the velocity variable. The velocity variable can then be
substituted into the differential equation for the system to reduce it to first-order.
x· = v (1)
Mx·· = – F – K d x· – K s x
Mv· = – F – K d v – K s x
–Ks –Kd
v· = x  --------- + v  ---------- +  ------
–F (2)
M M M
Equations (1) and (2) can also be put into a matrix form similar to that given in Fig-
ure 4.1.
0 1 0
d x = – K – K x + –F
-----
dt v --------s- ---------d- v ------
M
M M

Note: To have a set of differential equations that is solvable, there must be the
same number of state equations as variables. If there are too few equations,
then an additional equation must be developed using an unexploited relation-
ship. If there are too many equations, a redundancy or over constraint must be
eliminated.

Figure 4.2 A state variable equation example

 numerical methods - 4.4

Put the equation into state variable and matrix form. 2

∑ F = F = M  ----
d-
x
dt

· F
ans. x· = v v = -----
M
0
d- x 0 1 x
---- = + F
dt v 0 0 v -----
M

Figure 4.3 Drill problem: Put the equation in state variable form

Consider the two cart problem in Figure 4.4. The carts are separated from each
other and the wall by springs, and a force is applied to the left hand side. Free body dia-
grams are developed for each of the carts, and differential equations developed. For each
cart a velocity state variable is created. The equations are then manipulated to convert the
second-order differential equations to first-order state equations. The four resulting equa-
tions are then put into the state variable matrix form.
 numerical methods - 4.5

x1 x2

F K s1 K s2
M1 M2

··
F K s1 ( x 1 – x 2 ) +
∑ Fx = F – K s1 ( x 1 – x 2 ) = M 1 x 1
M1 ··
·· M 1 x 1 + K s1 x 1 – K s1 x 2 = F
M1 x1 ·
x1 = v1 (1)
·
M 1 v 1 + K s1 x 1 – K s1 x 2 = F
· F- – K s1 + K s1
v 1 = ------- ---------x 1 ---------x (2)
M1 M1 M1 2
··
K s1 ( x 1 – x 2 ) K s2 ( x 2 ) +
∑ Fx = K s1 ( x 1 – x 2 ) – K s2 ( x 2 ) = M 2 x 2
M2 ··
M 2 x 2 + ( K s1 + K s2 )x 2 – K s1 x 1 = 0
·· ·
M2 x2 x2 = v2 (3)
·
M 2 v 2 + ( K s1 + K s2 )x 2 – K s1 x 1 = 0
· K s1  K s1 + K s2
v 2 = ---------x –  -------------------------- x 2 (4)
M2 1  M2 
The state equations can now be combined in a matrix form.
0 1 0 0
x1 – K s1 K s1 x1 0
------------- 0 --------- 0 F-
d v1 M1 M1 v1 -------
----- = + M1
dt x 0 0 0 1 x
2 2 0
v2 K s1 – K s1 – K s2 v2
--------- 0 ----------------------------- 0 0
M2 M2

Figure 4.4 Two cart state equation example

 numerical methods - 4.6

Kd

M Ks

ans.
·
x = v
Kd Ks
v = v  ------ + x  -----
·
M M

Figure 4.5 Drill problem: Develop the state equations in matrix form
 numerical methods - 4.7

x1 x2

K d1
F K s2
M1 K s1 M2

ans. x· = v
1 1
·
x2 = v2
– K d1 – K s1 K d1 K s1
v 1 = v 1  ------------ + x 1  ----------- + v 2  --------- + x 2  -------- + -------
· F
M1 M1 M1 M1 M1
– K d1 – K s1 – K s2 K d1 K s1
v 2 = v 2  ------------ + x 2  --------------------------- + v 1  --------- + x 1  --------
·
 M2   M2   M2   M2 

Figure 4.6 Drill problem: Convert the system to state equations

 numerical methods - 4.8

In some cases we will develop differential equations that cannot be directly

reduced because they have more than one term at the highest order. For example, if a sec-
ond-order differential equation has two second derivatives it cannot be converted to a state
equation in the normal manner. In this case the two high order derivatives can be replaced
with a dummy variable. In mechanical systems this often happens when masses are
neglected. Consider the example problem in Figure 4.7, both ’y’ and ’u’ are first deriva-
tives. To solve this problem, the highest order terms (’y’ and ’u’) are moved to the left of
the equation. A dummy variable, ’q’, is then created to replace these two variables with a
single variable. This also creates an output equation as shown in Figure 4.1.

Given the equation,

3y· + 2y = 5u·
Step 1: put both the first-order derivatives on the left hand side,
3y· – 5u· = – 2y
Step 2: replace the left hand side with a dummy variable,
q = 3y – 5u q· = – 2y
Step 3: solve the equation using the dummy variable, then solve for y as an output eqn.
q + 5u
q· = – 2y y = ---------------
3

Figure 4.7 Using dummy variables for multiple high order terms

At other times it is possible to eliminate redundant terms through algebraic manip-

ulation, as shown in Figure 4.8. In this case the force on both sides of the damper is the
same, so it is substituted into the equation for the cart. But, the effects on the damper must
also be integrated, so a dummy variable is created for the integration. An output equation
was created to calculate the value for x1.
 numerical methods - 4.9

x1
x2
F Kd
M

The FBDs and equations are;

· ·
F Kd ( x1 – x2 )
· ·
+ ∑ Fx = F – Kd ( x1 – x2 ) = 0
· ·
Kd ( x 1 – x 2 ) = F (1)
q = x1 – x2
Kd ( q ) = F
F-
q· = ------ (2)
Kd
x1 = x2 + q (3)
· ·
Kd ( x1 – x2 ) · · ··
M + ∑ Fx = K d ( x 1 – x 2 ) = Mx2
··
F = Mx 2
·· ·
Mx 2 (4)
x2 = v2

· F- (5)
v 2 = ----
M
The state equations (2, 4, 5) can be put in matrix form. The output equation (2) can
also be put in matrix form.

F-
q q ------
0 0 0 Kd q
d x x
----- 2 = 0 0 1 2 +
0 x1 = 1 1 0 x2 + 0
dt
v2 0 0 0 v2 F- v2
----
M

Figure 4.8 A dummy variable example

 numerical methods - 4.10

4.3 NUMERICAL INTEGRATION

Repetitive calculations can be used to develop an approximate solution to a set of

differential equations. Starting from given initial conditions, the equation is solved with
small time steps. Smaller time steps result in a higher level of accuracy, while larger time
steps give a faster solution.

4.3.1 Numerical Integration With Tools

Numerical solutions can be developed with hand calculations, but this is a very
time consuming task. In this section we will explore some common tools for solving state
variable equations. The analysis process follows the basic steps listed below.

1. Generate the differential equations to model the process.

2. Select the state variables.
3. Rearrange the equations to state variable form.
4. Add additional equations as required.
5. Enter the equations into a computer or calculator and solve.

An example in Figure 4.9 shows the first four steps for a mass-spring-damper com-
bination. The FBD is used to develop the differential equations for the system. The state
variables are then selected, in this case the position, y, and velocity, v, of the block. The
equations are then rearranged into state equations. The state equations are also put into
matrix form, although this is not always necessary. At this point the equations are ready
for solution.
 numerical methods - 4.11

Step 1: Develop equations

KS Kd Ks y K d y·
 ----
d- d- 2
 ----
M ∑ y
F = – F – K d  dt y – K s y = M  dt y
y
d 2
F + K d  ----- y + K s y + M  ----- y = 0
M F d
My··  dt  dt
F
Step 2: We need to identify state variables. In this case the height is clearly a defining vari-
able. We will also need to use the vertical velocity, because the acceleration is a second
derivative (we can only have first derivatives). Using the height, y, and velocity, v, as state
variables we may now proceed to rewriting the equations. (Note: this is just an algebraic
trick, but essential when setting up these matrices.)
Step 3:  ----
d-
y = v
 dt
– F – Kd v – K s y
 ----
d-
v = ---------------------------------------
-
 dt M
Step 4: We put the equations into a state variable matrix form.

 ----
d-
0 1
 dt y F-
= – K – K y + 0 ----
---------- ---------d- v
S –1 M
 ----
d-
v M M
 dt

Figure 4.9 Dynamic system example

Figure 4.10 shows the method for solving state equations on a TI-86 graphing cal-
culator. (Note: this also works on other TI-8x calculators with minor modifications.) In the
example a sinusoidal input force, F, is used to make the solution more interesting. The
next step is to put the equation in the form expected by the calculator. When solving with
the TI calculator the state variables must be replaced with the predefined names Q1, Q2,
etc. The steps that follow describe the button sequences required to enter and analyze the
equations. The result is a graph that shows the solution of the equation. Points can then be
taken from the graph using the cursors. (Note: large solutions can sometimes take a few
minutes to solve.)
 numerical methods - 4.12

First, we select some parameter values for the equations of Figure 4.9. The input force
will be a decaying sine wave.

 ----
d-
y = vy
 dt

 ----
d- – F – Kd vy – Ks y – 0.5t
v y = ---------------------------------------- = – 4e sin ( t ) – 2v y – 5y
 dt M
Next, the calculator requires that the state variables be Q1, Q2, ..., Q9, so we replace y
with Q1 and v with Q2.
Q1' = Q2
– 0.5t
Q2' = – 4e sin ( t ) – 2Q2 – 5Q1

Now, we enter the equations into the calculator and solve. To do this roughly follow
the steps below. Look at the calculator manual for additional details.
1. Put the calculator in differential equation mode
[2nd][MODE][DifEq][ENTER]
2. Go to graph mode and enter the equations above [GRAPH][F1]
3. Set up the axis for the graph [GRAPH][F2] so that time and the x-
axis is from 0 to 10 with a time step of 0.5, and the y height is from
+3 to -3.
4. Enter the initial conditions for the system [GRAPH][F3] as Q1=0,
Q2=0
5. Set the axis [GRAPH][F4] as x=t and y=Q
6. (TI-86 only) Set up the format [GRAPH][MORE][F1][Fld-
Off][ENTER]
7. Draw the graph [GRAPH][F5]
8. Find points on the graph [GRAPH][MORE][F4]. Move the left/right
cursor to move along the trace, use the up/down cursor to move
between traces.

Figure 4.10 Solving state equations with a TI-85 calculator

 numerical methods - 4.13

First, we select some parameter values for the equations of Figure 4.9. The input force
will be a decaying sine wave.

 ----
d-
y = vy
 dt
– F – Kd vy – Ks y
 ----
d-
v y = ------------------------------------------ = – 4e
– 0.5t
sin ( t ) – 2v y – 5y
 dt M
Next, the calculator requires that the state variables be Q1, Q2, ..., Q9, so we replace y
with Q1 and v with Q2.
Q1' = Q2
– 0.5t
Q2' = – 4e sin ( t ) – 2Q2 – 5Q1

Now, we enter the equations into the calculator and solve. To do this roughly follow
the steps below. Look at the calculator manual for additional details.
1. Put the calculator in differential equation mode
[2nd][MODE][DifEq][ENTER]
2. Go to graph mode and enter the equations above [GRAPH][F1]

UPDATE FOR TI-89

3. Set up the axis for the graph [GRAPH][F2] so that time and the x-
axis is from 0 to 10 with a time step of 0.5, and the y height is from
+3 to -3.
4. Enter the initial conditions for the system [GRAPH][F3] as Q1=0,
Q2=0
5. Set the axis [GRAPH][F4] as x=t and y=Q
6. (TI-86 only) Set up the format [GRAPH][MORE][F1][Fld-
Off][ENTER]
7. Draw the graph [GRAPH][F5]
8. Find points on the graph [GRAPH][MORE][F4]. Move the left/right
cursor to move along the trace, use the up/down cursor to move
between traces.

Figure 4.11 Solving state equations with a TI-89 calculator

State equations can also be solved in Mathcad using built-in functions, as shown in
Figure 4.12. The first step is to enter the state equations as a function, ’D(t, Q)’, where ’t’
is the time and ’Q’ is the state variable vector. (Note: the equations are in a vector, but it is
not the matrix form.) The state variables in the vector ’Q’ replace the original state vari-
ables in the equations. The ’rkfixed’ function is then used to obtain a solution. The argu-
ments for the function, in sequence are; the state vector, the start time, the end time, the
number of steps, and the state equation function. In this case the 10 second time interval is
divided into 100 parts each 0.1s in duration. This time is chosen because of the general
 numerical methods - 4.14

response time for the system. If the time step is too large the solution may become unsta-
ble and go to infinity. A time step that is too small will increase the computation time mar-
ginally. When in doubt, run the calculator again using a smaller time step.

Figure 4.12 Solving state variable equations with Mathcad

Note: Notice that for the TI calculators the variables start at Q1, while in Mathcad the
arrays start at Q0. Many students encounter problems because they forget this.
 numerical methods - 4.15

4.3.2 Numerical Integration

The simplest form of numerical integration is Euler’s first-order method. Given the
current value of a function and the first derivative, we can estimate the function value a
short time later, as shown in Figure 4.13. (Note: Recall that the state equations allow us to
calculate first-order derivatives.) The equation shown is known as Euler’s equation. Basi-
cally, using a known position and first derivative we can calculate an approximate value a
short time, h, later. Notice that the function being integrated curves downward, creating an
error between the actual and estimated values at time ’t+h’. If the time step, h, were
smaller, the error would decrease.

y(t + h)
d- d
---- y( t) y ( t + h ) ≈ y ( t ) + h ----- y ( t )
dt dt
y(t)
Note: here the h value is the time step
between integrations points. A
smaller time step will increase the
accuracy.
t t+h

Figure 4.13 First-order numerical integration

The example in Figure 4.14 shows the solution of Newton’s equation using Euler’s
method. In this example we are determining velocity by integrating the acceleration
caused by a force. The acceleration is put directly into Euler’s equation. This is then used
to calculate values iteratively in the table. Notice that the values start before zero so that
initial conditions can be used. If the system was second-order we would need two previous
values for the calculations.
 numerical methods - 4.16

Given the differential equation,

F = M  ----- v
d
dt

we can create difference equations using simple methods.

 ----
d- F-
 dt v = ----
M
first rearrange equation

v ( t + h ) = v ( t ) + h  ----- v ( t )
d put this in the Euler equation
 dt
F(t)
v ( t + h ) = v ( t ) + h  ---------- finally substitute in known terms
 M 
We can now use the equation to estimate the system response. We will assume that the
system is initially at rest and that a force of 1N will be applied to the 1kg mass for
4 seconds. After this time the force will rise to 2N. A time step of 2 seconds will be
used.
i t (sec) F (N) d/dt vi vi

-1 -2 0 0 0
0 0 1 1 0
1 2 1 1 2
2 4 2 2 4
3 6 2 2 8
4 8 2 2 12
5 10 2 2 16
6 12 2 2 20
7 14 2 etc etc
8 16 2

Figure 4.14 First order numerical integration example

 numerical methods - 4.17

Use first-order integration to solve the differential equation from 0 ·

to 10 seconds with time steps of 1 second. x + 0.1x = 5

ans.
x ( t + h ) = x ( t ) + h ( – 0.1x ( t ) + 5 )
x ( 10 ) =

Figure 4.15 Drill problem: Numerically integrate the differential equation

An example of solving the previous example with a traditional programming lan-

guage is shown in Figure 4.16. In this example the results will be written to a text file
’out.txt’. The solution iteratively integrates from 0 to 10 seconds with time steps of 0.1s.
The force value is varied over the time period with ’if’ statements. The integration is done
with a separate function.
 numerical methods - 4.18

double step(double, double, double);

int main(){
double h = 0.1,
M = 1.0,
F;
FILE *fp;
double v,
t;
if( ( fp = fopen("out.txt", "w")) != NULL){
v = 0.0;
for( t = 0.0; t < 10.0; t += h ){
if((t >= 0.0) && (t < 4.0)) F = 1.0;
if(t > 4.0) F = 2.0;
v = step(v, h, F/M);
fprintf(fp, "%f, %f, %f\n", t, v, F, M);
}
}
fclose(fp);
}

double step(double v, double h, double slope){

double v_new;
v_new = v + h * slope;
return v_new;
}

Figure 4.16 Solving state variable equations with a C program

 numerical methods - 4.19

double step(double, double, double);

public class Integrate extends Object

public void main() {
double h = 0.1,
M = 1.0,
F;
FileOut fp = new FileOut("out.txt");
if(fp.writeStatus != fp.IO_EXCEPTION){
double v = 0.0;
for( double t = 0.0; t < 10.0; t += h ){
if((t >= 0.0) && (t < 4.0)) F = 1.0;
if(t > 4.0) F = 2.0;
v = step(v, h, F/M);
fp.printf(fp, "%f, %f, %f\n", t, v, F, M);
}
fp.close();
}
fclose(fp);
}

public double step(double v, double h, double slope){

double v_new;
v_new = v + h * slope;
return v_new;
}
}

Figure 4.17 Solving state variable equations with a Java program

The program below is for Scilab (a Matlab clone). The state variable function is
defined first. This is followed by a definition of the parameters to be used for the numeri-
cal integration. Finally the function is integrated with rectangular, trapezoidal and Simp-
son’s rule forms.
 numerical methods - 4.20

//
// first_order.sce
//
// A first order integration of an accelerating mass
//
// To run this in Scilab use 'File' then 'Exec'.
//
// by: H. Jack Sept., 16, 2002
//

// System component values

mass = 10;
force = 100;

x0 = 8; // initial conditions
v0 = 12;
X=[x0, v0];

// define the state matrix function

// the values returned are [x, v]
function foo=f(state,t)
foo = [ state($, 2), force/mass]; // d/dt x = v, d/dt v = F/M
endfunction

// Set the time length and step size for the integration
steps = 100;
t_start = 1;
t_end = 100;
h = (t_end - t_start) / steps;

Figure 4.18 First order integration with Scilab

 numerical methods - 4.21

//
// Loop for integration
//
for i=1:steps,
X = [X ; X($,:) + h*f(X, i*h)];
end
printf("The value at the end of first order integration is (x, v) = (%f, %f)\n", ...
X($,1), ...
X($,2));

//
// Explicit equation
//
function x=position(x0, v0, a0, t)
x = (0.5 * a0 * t^2) + (v0 * t) + x0;
endfunction

function v=velocity(v0, a0, t)

v = (a0 * t) + v0;
endfunction

printf("The value with integration is (x, v) = (%f, %f)\n", ...

position(x0, v0, force/mass, t_end), ...
velocity(v0, force/mass, t_end));

//
// The results should be
// first order integration = (49710, 1002)
// explicit = (51208, 1012)
//
// The difference is 1498 for position and 10 for velocity. This is relatively small, but
// shows a clear case of the innacuracy of the numerical solutions.
//
// Note: increasing the number of steps increases the accuracy
//

Figure 4.19 First order integration with Scilab (continued)

4.3.3 Taylor Series

First-order integration works well with smooth functions. But, when a highly
curved function is encountered we can use a higher order integration equation. The Taylor
series equation shown in Figure 4.20 for approximating a function. Notice that the first
part of the equation is identical to Euler’s equation, but the higher order terms add accu-
racy.
 numerical methods - 4.22

1 2 d 2 1 3 d 3 1 4 d 4
x ( t + h ) = x ( t ) + h  ----- x ( t ) + ----- h  ----- x ( t ) + ----- h  ----- x ( t ) + ----- h  ----- x ( t ) + …
d
dt 2! dt 3! dt 4! dt

Figure 4.20 The Taylor series

An example of the application of the Taylor series is shown in Figure 4.21. Given
the differential equation, we must first determine the derivatives and substitute these into
Taylor’s equation. The resulting equation is then used to iteratively calculate values.

– 20t 3
Given x· – x = 1 + e +t

We can write,  ----- x = 1 + e

d – 20t 3
+t +x
dt
d- 2
 ---- – 20t 2
 dt x = – e + 3t
d- 3
 ---- – 20t
x = e + 6t
 dt
In the Taylor series this becomes,
– 20t 3 1- h 2 ( – e – t + 3t 2 ) + ----
1- h 3 ( e – t + 6t
x(t + h) = x( t) + h(1 + e + t + x ) + ----
2! 3!

Thus x0 = 0 t (s) x(t)

h = 0.1 0 0
0.1 0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9

e.g., for t=0.1 1- ( 0.1 ) 2 ( – 1 ) + ----

1- ( 0.1 ) 3 ( 1 ) =
x ( 0 + 0.1 ) = 0 + 0.1 ( 2 ) + ----
2! 3!

Figure 4.21 Integration using the Taylor series method

 numerical methods - 4.23

Recall that the state variable equations are first-order equations. But, to obtain
accuracy the Taylor method also requires higher order derivatives, thus making is unsuit-
able for use with state variable equations.

4.3.4 Runge-Kutta Integration

First-order integration provides reasonable solutions to differential equations. That

accuracy can be improved by using higher order derivatives to compensate for function
curvature. The Runge-Kutta technique uses first-order differential equations (such as state
equations) to estimate the higher order derivatives, thus providing higher accuracy without
requiring other than first-order differential equations.

The equations in Figure 4.22 are for fourth order Runge-Kutta integration. The
function ’f()’ is the state equation or state equation vector. For each time step the values
’F1’ to ’F4’ are calculated in sequence and then used in the final equation to find the next
value. The ’F1’ to ’F4’ values are calculated at different time steps, and values from previ-
ous time steps are used to ’tweak’ the estimates of the later states. The final summation
equation has a remote similarity to the first order integration equation. Notice that the two
central values in time are more heavily weighted.

F 1 = hf ( t, x )
F Note: in this case the state equation function
F 2 = hf  t + h---, x + ------1 f(t, x) includes the state variables, x, and
2 2
F time, t. However, in simpler systems the
F 3 = hf  t + h---, x + ------2 state equations may not include time and
2 2 it could be replaced with f(x).

F 4 = hf ( t + h, x + F 3 )

x ( t + h ) = x ( t ) + 1--- ( F1 + 2F 2 + 2F 3 + F 4 )
6
where,
x = the state variables
f = the differential function or (d/dt) x
t = current point in time
h = the time step to the next integration point

Figure 4.22 Fourth order Runge-Kutta integration

 numerical methods - 4.24

An example of a Runge-Kutta integration calculation is shown in Figure 4.23. The

solution begins by putting the state equations in matrix form and defining initial condi-
tions. After this, the four integrating factors are calculated. Finally, these are combined to
get the final value after one time step. The number of calculations for a single time step
should make obvious the necessity of computers and calculators.

d- x = v
---- y = 2 (assumed input)
dt
v0 = 1
d- v = 3 + 4v + 5y
----
dt x0 = 3
d- x = 0 1 x + 0 0 y h = 0.1
----
dt v 0 4 v 5 3 1

For the first time step,

   
F 1 = 0.1  0 1 3 + 0 0 2  = 0.1  1 + 0  = 0.1
 0 4 1 5 3 1  4 13  1.7
 
 3 + 0.1
------- 
F 2 = 0.1  0 1 2 + 0 0 2  = 0.185
 
 0 4 1 + 1.7
------- 5 3 1 2.04
 2 
 
 3 + 0.185
------------- 

F 3 = 0.1  0 1 2 + 0 0 2  = 0.202

 0 4 1 + 2.04
---------- 5 3 1 2.108
 2 
 
F 4 = 0.1  0 1 3 + 0.202 + 0 0 2  = 0.3108
 0 4 1 + 2.108 5 3 1 2.5432
xi + 1 xi
= + 1--- ( F1 + 2F 2 + 2F 3 + F 4 )
vi + 1 vi 6

xi + 1  
= 3 + 1---  0.1 + 2 0.185 + 2 0.202 + 0.3108  = 3.1974667
vi + 1 1 6  1.7 2.04 2.108 2.5432  3.0898667

Figure 4.23 Runge-Kutta integration example

 numerical methods - 4.25

d 2
F = M  ----- x
use, x(0) = 1
dt v(0) = 2
h = 0.5 s
F = 10
M=1

Figure 4.24 Drill problem: Integrate the acceleration function

 numerical methods - 4.26

Leave the solution in variable form, and then change

values to see how the system response changes.
Kd
–1
K d = 10Nsm
–1
y K s = 10Nm
M
M = 1kg
·
y 0 = 1m y 0 = 0m
Ks

Figure 4.25 Drill problem: Integrate to find the system response

The program in Figure 4.26 and Figure 4.28 is used to perform a Runge-Kutta inte-
gration of a mass-spring-damper system. The ’main’ program loops through the time steps
and writes the value to a file. The ’step’ function performs one timestep integration for a
 numerical methods - 4.27

second order Runge-Kutta integration. It uses the functions ’add’ and ’multiply’ to manip-
ulate the state matrix. The ’derivative’ function updates the state matrix with the new
derivative values.

/* A program to do Runge Kutta integration of a mass spring damper system */

#include <stdio.h>

void multiply(double, double[], double[]);

void add(double[], double[], double[]);
void step(double, double, double[]);
void derivative(double, double[], double[]);

#define SIZE 2 /* the length of the state vector */

#define Ks 1000 /* the spring coefficient */
#define Kd 10000 /* the damping coefficient */
#define Mass 10 /* the mass coefficient */
#define Force 100 /* the applied force */

int main(){
FILE *fp;

double h = 0.001;
double t;
int j = 0;

double X[SIZE];// create state variable list

X[0] = 0;// set initial condition for x
X[1] = 0;// set initial condition for v

if( ( fp = fopen("out.txt", "w")) != NULL){

fprintf(fp, " t(s) x v \n\n");
for( t = 0.0; t < 50.0; t += h ){
step(t, h, X);
if(j == 0) fprintf(fp, "%9.5f %9.5f %9.5f\n", t, X[0], X[1]);
j++; if(j >= 10) j = 0;
}
}
fclose(fp);
}

Figure 4.26 Runge-Kutta integration C program

 numerical methods - 4.28

/* First order integration done here (could be replaced with runge kutta)*/
void step(double t, double h, double X[]){
double tmp[SIZE],
dX[SIZE],
F1[SIZE],
F2[SIZE],
F3[SIZE],
F4[SIZE];
/* Calculate F1 */
derivative(t, X, dX);
multiply(h, dX, F1);

/* Calculate F2 */
multiply(0.5, F1, tmp);
add(X, tmp, tmp);
derivative(t+h/2.0, tmp, dX);
multiply(h, dX, F2);

/* Calculate F3 */
multiply(0.5, F2, tmp);
add(X, tmp, tmp);
derivative(t+h/2.0, tmp, dX);
multiply(h, dX, F3);

/* Calculate F4 */
add(X, F3, tmp);
derivative(t+h, tmp, dX);
multiply(h, dX, F4);

/* Calculate the weighted sum */

add(F2, F3, tmp);
multiply(2.0, tmp, tmp);
add(F1, tmp, tmp);
add(F4, tmp, tmp);
multiply(1.0/6.0, tmp, tmp);
add(tmp, X, X);
}

/* State Equations Calculated Here */

void derivative(double t, double X[], double dX[]){
dX[0] = X[1];
dX[1] = (-Ks/Mass)*X[0] + (-Kd/Mass)*X[1] + (Force/Mass);
}

/* A subroutine to add vectors to simplify other equations */

void add(double X1[], double X2[], double R[]){
for(int i = 0; i < SIZE; i++) R[i] = X1[i] + X2[i];
}

/* A subroutine to multiply a vector by a scalar to simplify other equations*/

void multiply(double X, double V[], double R[]){
for(int i = 0; i < SIZE; i++) R[i] = X*V[i];
}

Figure 4.27 Runge-Kutta integration C program (cont’d)

A Scilab program is given in Figure 4.28 to perform a Runge Kutta integration.

 numerical methods - 4.29

// runge_kutta.sce
// A first order integration of an accelerating mass
// To run this in Scilab use 'File' then 'Exec'.
// by: H. Jack Sept., 15, 2003

// System component values

mass = 10;
force = 100;

x0 = 8; // initial conditions
v0 = 12;
X=[x0, v0];

// define the state matrix function

// the values returned are [x, v]
function foo=f(state,t)
foo = [ state($, 2), force/mass]; // d/dt x = v, d/dt v = F/M
endfunction

// Set the time length and step size for the integration
steps = 1000;
t_start = 0;
t_end = 100;
h = (t_end - t_start) / steps;
t = [t_start];

// Loop for integration

for i=1:steps,
t = [t ; t($,:) + h];
F1 = h * f(X($,:), t($,:));
F2 = h * f(X($,:) + F1/2.0, t($,:) + h/2.0);
F3 = h * f(X($,:) + F2/2.0, t($,:) + h/2.0);
F4 = h * f(X($,:) + F3, t($,:) + h);
X = [X ; X($,:) + (F1 + 2.0*F2 + 2.0*F3 + F4)/6.0];
end

// print some results to compare

printf("The value at the end of first order integration is (x, v) = (%f, %f)\n", ...
X($,1), ...
X($,2));
printf("The position (using an equation) should be %f\n", 0.5*force/mass*(t_end-
t_start)^2.0 + v0*(t_end - t_start) + x0);

// Graph the values

plot2d(t, X, [-2, -5], leg="position@velocity");
// leg - the legend titles
// style - draw lines with marks
// nax - grid lines for the graph
xtitle('Time (s)');

Figure 4.28 Runge-Kutta integration Scilab program

4.4 SYSTEM RESPONSE

In most cases the result of numerical analysis is graphical or tabular. In both cases
 numerical methods - 4.30

details such as time constants and damped frequencies can be obtained by the same meth-
ods used for experimental analysis. In addition to these methods there is a technique that
can determine the steady-state response of the system.

4.4.1 Steady-State Response

The state equations can be used to determine the steady-state response of a system
by setting the derivatives to zero, and then solving the equations. Consider the example in
Figure 4.29. The solution begins with a state variable matrix. (Note: this can also be done
without the matrix also.) The derivatives on the left hand side are set to zero, and the equa-
tions are rearranged and solved with Cramer’s rule.

Given the state variable form: 0 1 0

d x x +
----- = Ks K d F-
dt v – ------ – ------- v ----
M M M

Set the derivatives to zero 0 1 0

0 = x +
Ks Kd F-
0 – ------ – ------- v ----
M M M
Solve for x and v
0 1 0
x =
Ks Kd F
– ------ – ------- v – -----
M M M

0 1 0 0
F Kd  ----
F-
Ks F
– ----- – ------- – ------ – -----
M M  M F- M M 0 - = 0
x = ------------------------------ = ------------ = ------ v = ------------------------------ = -----------
K KS K
0 1  -----s- 0 1  -----s-
Ks Kd  M Ks Kd  M
– ------ – ------- – ------ – -------
M M M M

Figure 4.29 Finding the steady-state response

 numerical methods - 4.31

4.5 DIFFERENTIATION AND INTEGRATION OF EXPERIMENTAL

DATA

When doing experiments, data is often collected in the form of individual data
points (not as complete functions). It is often necessary to integrate or differentiate these
values. The basic equations for integrating and differentiating are shown in Figure 4.30.
Given data points, y, collected at given times, t, we can integrate and differentiate using
the given equations. The integral is basically the average height of the two points multi-
plied by the width to give an area, or integral. The first derivative is basically the slope
between two points. The second derivative is the change in slope values for three points.
In a computer based system the time points are often equally spaced in time, so the differ-
ence in time can be replaced with a sample period, T. Ideally the time steps would be as
small as possible to increase the accuracy of the estimates.

y(t)

yi + 1
yi
yi – 1

ti – 1 T T
ti ti + 1

ti yi + yi – 1
 ------------------------
- ( t – t T
∫t y ( t i ) d t ≈  2  i i – 1 ) = --2- ( y i + y i – 1 )
i–1
d-  y i – y i – 1  y i + 1 – y i
---- y ( t i ) ≈  ------------------------ =  ------------------------- = --1- ( y i – y i – 1 ) = --1- ( y i + 1 – y i )
dt  ti – t i – 1   ti + 1 – ti  T T

2 --1- ( y i + 1 – y i ) – --1- ( y i – y i – 1 ) –2 yi + yi – 1 + yi + 1
 ----
d-
y ( t ) ≈
T T
------------------------------------------------------------------------ = -----------------------------------------------------
 dt i T 2
T

Figure 4.30 Integration and differentiation using data points

 numerical methods - 4.32

An example of numerical integration using Scilab is given in Figure 4.31 and Fig-
ure 4.32.
//
// integrate.sce
//
// A simple program to integrate a function
//
// To run this in Scilab use 'File' then 'Exec'.
//
// by: H. Jack Sept., 9, 2002
//

// define the function

function foo=f(x)
foo = 5 * x + 2 * log(sin(x) / x + 2);
endfunction

// Set the time length and step size

steps = 10;
x_start = 1;
x_end = 10;
x_delta = (x_end - x_start) / steps;

//
// Loop for rectangular integration
//
total = 0; // set the initial sum to zero
for i=0:steps,
x = x_start + i * x_delta;
total = total + f(x);
end
total = total * x_delta;
printf("Rectangular integration value %f\n", total);

Figure 4.31 Integration with a Scilab program

 numerical methods - 4.33

// Loop for trapezoidal integration

//
total = 0; // set the initial sum to zero
for i=0:steps,
x = x_start + i * x_delta;
if i == 0 then
total = total + f(x);
elseif i == steps then
total = total + f(x);
else
total = total + 2 * f(x);
end
end
total = total * x_delta / 2;
printf("Trapezoidal integration value %f\n", total);

//
// Loop for Simpson's rule integration
//
total = 0; // set the initial sum to zero
even = 0;
for i=0:steps,
x = x_start + i * x_delta;
if i == 0 then
total = total + f(x);
elseif i == steps then
total = total + f(x);
else
even = even + 1;
if even > 1 then
total = total + 4 * f(x);
even = 0;
else
total = total + 2 * f(x);
end
end
end
total = total * x_delta / 3;
printf("Simpsons rule integration value %f\n", total);

Figure 4.32 Integration with a Scilab Program (cont’d)

4.6 ADVANCED TOPICS

4.6.1 Switching Functions

When analyzing a system, we may need to choose an input that is more complex
than inputs such as steps, ramps, sinusoids and parabolae. The easiest way to do this is to
use switching functions. Switching functions turn on (have a value of 1) when their argu-
 numerical methods - 4.34

ments are greater than or equal to zero, or off (a value of 0) when the argument is negative.
Examples of the use of switching functions are shown in Figure 4.33. By changing the val-
ues of the arguments we can change when a function turns on or off.

u(t)
1
t

u ( –t )
1
t

u(t – 1)
1
t
1
u(t + 1)

1 t
-1
u(t + 1) – u(t – 1)

1 t
-1 1

Figure 4.33 Switching function examples

These switching functions can be multiplied with other functions to create a com-
plex function by turning parts of the function on or off. An example of a curve created
with switching functions is shown in Figure 4.34.
 numerical methods - 4.35

f(t)
5

t
seconds
0 1 3 4

f ( t ) = 5t ( u ( t ) – ( t – 1 ) ) + 5 ( u ( t – 1 ) – u ( t – 3 ) ) + ( 20 – 5t ) ( u ( t – 3 ) – u ( t – 4 ) )
or
f ( t ) = 5tu ( t ) – 5 ( t – 1 )u ( t – 1 ) – 5 ( t – 3 )u ( t – 3 ) + 5 ( t – 4 )u ( t – 4 )

Figure 4.34 Switching functions to create a non-smooth function

The unit step switching function is available in Mathcad and makes creation of
complex functions relatively trivial. Step functions are also easy to implement when writ-
ing computer programs, as shown in Figure 4.35.

For the function f ( t ) = 5tu ( t ) – 5 ( t – 1 )u ( t – 1 ) – 5 ( t – 3 )u ( t – 3 ) + 5 ( t – 4 )u ( t – 4 )

double u(double t){
if(t >= 0) return 1.0;
return 0.0;
}

double function(double t){

double f;

f = 5.0 * t * u(t)
- 5.0 * (t - 1.0) * u(t - 1.0)
- 5.0 * (t - 3.0) * u(t - 3.0)
+ 5.0 * (t - 4.0) * u(t - 4.0);

return f;
}

Figure 4.35 A subroutine implementing the example in Figure 4.34

 numerical methods - 4.36

4.6.2 Interpolating Tabular Data

In some cases we are given tables of numbers instead of equations for a system
component. These can still be used to do numerical integration by calculating coefficient
values as required, in place of an equation.

Tabular data consists of separate data points as seen in Figure 4.36. But, we may
need values between the datapoints. A simple method for finding intermediate values is to
interpolate with the "lever law". (Note: it is called this because of its’ similarity to the
equation for a lever.) The table in the example only gives flow rates for a valve at 10
degree intervals, but we want flow rates at 46 and 23 degrees. A simple application of the
lever law gives approximate values for the flow rates.

valve angle flow rate

(deg.) (gpm)

0 0 Given a valve angle of 46 degrees the flow rate is,

10 0.1
Q = 2.0 + ( 2.3 – 2.0 )  ------------------ = 2.18
46 – 40
20 0.4 50 – 40
30 1.2
40 2.0 Given a valve angle of 23 degrees the flow rate is,
50 2.3
Q = 0.4 + ( 1.2 – 0.4 )  ------------------ = 0.64
60 2.4 23 – 20
70 2.4 30 – 20
80 2.4
90 2.4

Figure 4.36 Using tables of values to interpolate numerical values using the lever law

The subroutine in Figure 4.37 was written to return the numerical value for the
data table in Figure 4.36. In the subroutine the tabular data is examined to find the interval
that the flow rate value falls inside. Once this is found the valve angle is calculated as the
ratio between the two known values.
 numerical methods - 4.37

#define SIZE 10;

double data[SIZE][2] = {{0.0, 0.0},
{10.0, 0.1},
{20.0, 0.4},
{30.0, 1.2},
{40.0, 2.0},
{50.0, 2.3},
{60.0, 2.4},
{70.0, 2.4},
{80.0, 2.4},
{90.0, 2.4}};

double angle(double rate){

int i;
for(i = 0; i < SIZE-1; i++){
if((rate >= data[i][0]) && (rate <= data[i+1][0]){
return (data[i+1][1] - data[i][1])
* (rate - data[i][0]) / (data[i+1][0] - data[i][0])
+ data[i][1];
}
}
printf("ERROR: rate out of range\n");
exit(1);
}

Figure 4.37 A tabular interpolation subroutine example

4.6.3 Modeling Functions with Splines

When greater accuracy is required, smooth curves can be fitted to interpolate

points. These curves are known as splines. There are multiple methods for creating
splines, but the simplest is to use a polynomial fitted to a set of points.

The example in Figure 4.38 shows a spline curve being fitted for three data points.
In this case a second order polynomial is used. The three data points are written out as
equations, and then put into matrix form, using the coefficients as the unknown values.
The matrix is then solved to obtain the coefficient values for the final equation. This equa-
tion can then be used to build a mathematical model of the system.
 numerical methods - 4.38

The datapoints below might have been measured for the horsepower of an internal
combustion engine on a dynamometer.
S (RPM) P (HP)

1000 105
4000 205
6000 110

In this case there are three datapoints, so we can fit the curve with a second (3-1)
order polynomial. The major task is to calculate the coefficients so that the
curve passes through all of the given points.

2
P ( S ) = AS + BS + C
Data values can be substituted into the equation,
2
P ( 1000 ) = A1000 + B1000 + C = 105
2
P ( 4000 ) = A4000 + B4000 + C = 205
2
P ( 6000 ) = A6000 + B6000 + C = 110
This can then be put in matrix form to find the coefficients,
2 Note: in this example the
1000 1000 1 A 105
= inverse matrix is used, but
2
4000 4000 1 B 205 other methods for solving
2 C 110 systems of equations are
6000 6000 1 equally valid. If the equa-
–1 tions were simpler, substitu-
A 1000000 1000 1 105
B = 16000000 4000 1 205 tion might have been a
better approach.
C 36000000 6000 1 110
–8 –7 –7
A 6.667x10 – 1.667x10 1.000x10 105
B = –4 –3 – 4 205
– 6.667x10 1.167x10 – 5.000x10
C 110
1.600 –1.000 0.400
A –5
– 1.617x10
B = 0.114
C 7.000
–5 2
P ( S ) = ( – 1.617x10 )S + 0.114S + 7.000

Figure 4.38 A spline fitting example

 numerical methods - 4.39

The order of the polynomial should match the number of points. Although, as the
number of points increases, the shape of the curve will become less smooth. A common
way for dealing with this problem is to fit the spline to a smaller number of points and then
verify that it matches the remaining points, or use a least squares method to find the best
approximation.

4.6.4 Non-Linear Elements

Despite our deepest wishes for simplicity, most systems contain non-linear compo-
nents. In the last chapter we looked at dealing with non-linearities by linearizing them
locally. Numerical techniques will handle non-linearities easily, but smaller time steps are
required for accuracy.

Consider the mass and an applied force shown in Figure 4.39. As the mass moves
an aerodynamic resistance force is generated that is proportional to the square of the
velocity. This results in a non-linear differential equation. This equation can be numeri-
cally integrated using a technique such as Runge-Kutta. Note that the state equation matrix
form cannot be used because it requires linear equations.

x
2
20x· 2
F
M ∑ Fx = F – 20x· = Mx··

x· = v (1)

20- v2 + ----
v· = –-------- F- (2)
M M

Figure 4.39 Developing state equations for a non-linear system

4.7 CASE STUDY

Consider the simplified model of a car suspension shown in Figure 4.40. The
model distributes the vehicle weight over four tires with identical suspensions, so the mass
of the vehicle is divided by four. In this model the height of the road will change and drive
the tire up, or allow it to drop down. The tire acts as a stiff spring, with little deflection.
The upper spring and damper are the vibration isolation units. The damper has been
designed to stiffen as the damper is compressed. The given table shows how the damping
 numerical methods - 4.40

coefficient varies with the amount of compression.

yc
Mc N
K s1 = 50000 ---- L = yc – yt
m
N
K s2 = 200000 ---- L Kd
K s1 Kd m
(cm) (Ns/m)
M c = 400Kg
30 1000
M t = 20Kg
0 1600
Mt -20 2000
yt
-30 2250

K s2
yr

Figure 4.40 A model of a car suspension system

For our purposes we will focus only on the translation of the tire, and ignore its
rotational motion. The differential equations describing the system are developed in Fig-
ure 4.41.
 numerical methods - 4.41

·· yc
Mc yc Mc

· ·
K s1 ( y t – y c ) K d ( y t – y c )
· · ·
∑F = K s1 ( y t – y c ) + K d ( y t – y c ) – M c g = M c y c
·· · ·
M c y c = ( K s1 )y t + ( – K s1 )y c + ( K d )y t + ( – K d )y c – M c g
d- y = v
---- (1)
dt c c
d
----- y t = v t (2)
dt
·  K s1  – K s1  Kd  – K d
vc =  ---------y +  ------------
-  y +   v +  ---------- v + ( – g )
------- (3)
 M c  t  M c  c  M c t  M c  c

K s1 ( y t – y c ) · ·
Kd ( yt – yc )

··
Mt yt
Mt

K s2 ( y r – y t )

· · ··
∑F = K s2 ( y r – y t ) – K s1 ( y t – y c ) – K d ( y t – y c ) – M t g = M t y t
·· · ·
M t y t = ( K s2 )y r + ( –K s2 – K s1 )y t + ( K s1 )y c + ( – K d )y t + ( K d )y c – M t g

·  K s2  – K s2 – K s1  K s1  –K d  K d

v t =  --------- y r +  ----------------------------- y t +  --------- y c +  ---------- v t +  ------- v c + ( – g ) (4)
 Mt   Mt   Mt   Mt   M t

Figure 4.41 Differential and state equations for the car suspension system

The damping force must be converted from a tabular form to equation form. This
is done in Figure 4.42.
 numerical methods - 4.42

There are four data points, so a third order polynomial is L Kd

required. (cm) (Ns/m)

3 2 30 1000
K d ( L ) = AL + BL + CL + D 0 1600
-20 2000
-30 2250

The four data points can now be written in equation form, and then put into matrix form.

3 2
1000 = A ( 0.3 ) + B ( 0.3 ) + C ( 0.3 ) + D
3 2
1600 = A ( 0 ) + B ( 0 ) + C ( 0 ) + D
3 2
2000 = A ( – 0.2 ) + B ( – 0.2 ) + C ( – 0.2 ) + D
3 2
2250 = A ( – 0.3 ) + B ( – 0.3 ) + C ( – 0.3 ) + D
0.027 0.09 0.3 1 A 1000
0 0 0 1 B = 1600
– 0.008 0.04 – 0.2 1 C 2000
– 0.027 0.09 – 0.3 1 D 2250
The matrix can be solved to find the coefficients, and the final equation written.

A – 2778
B = 277.8
C – 1833
D 1600
3 2
K d ( L ) = ( –2778 )L + ( 277.8 )L + ( – 1833 )L + 1600

Figure 4.42 Fitting a spline to the damping values

The system is to be tested for overall deflection when exposed to obstacles on the
road. For the initial conditions we need to find the resting heights for the tire and car body.
This can be done by setting the accelerations and velocities to zero, and finding the result-
ing heights.
 numerical methods - 4.43

The initial accelerations and velocities are set to zero, assuming the car has settled to
a steady state height. This then yields equations that can be used to calculate the
initial deflections. Assume the road height is also zero to begin with.
K s1 – K s1 Kd –K d
0 =  -------- y t +  ----------- y c +  ------- 0 +  --------- 0 + ( – g )
Mc Mc Mc Mc
Mc
y c = y t – g --------
K s1
K s2 – K s2 – K s1 K s1 –Kd Kd
0 =  -------- 0 +  -------------------------- y t +  -------- y c +  --------- 0 +  ------ 0 + ( – g )
Mt Mt Mt Mt Mt
gM t = ( – K s2 – K s1 )y t + ( K s1 )y c
Mc Mc + Mt
gM t = ( – K s2 – K s1 )y t + ( K s1 )  y t – g -------- y t = – g --------------------
K s2
K s1
Mc + Mt Mc Mc + Mt Mc
y c = – g -------------------- – g -------- y c = – g  -------------------- + --------
K s2 K s1  K s2 K s1

Figure 4.43 Calculation of initial deflections

The resulting calculations can then be written in a computer program for analysis,
as shown in Figure 4.44.
 numerical methods - 4.44

#include <stdio.h>
#include <math.h>

#define SIZE 4 /* define state variables */

#define y_c 0
#define y_t 1
#define v_c 2
#define v_t 3

#define N_step 10000 // number of steps

#define h_step 0.001 // define step size

#define Ks1 50000.0 /* define component values */

#define Ks2 200000.0
#define Mc 400.0
#define Mt 20.0
#define grav 9.81

void integration_step(double h, double state[], double derivative[]){

int i;
for(i = 0; i < SIZE; i++) state[i] += h * derivative[i];
}

double damper(double L){

return (-2778*L*L*L + 277.8*L*L - 1833*L + 1600);
}

double y_r(double t){

/* return 0.0; /* a zero input to test the initial conditions */
/* return 0.2 * sin(t);/* a sinusoidal oscillation */
return 0.2; */ /* a step function */
/* return 0.2 * t; */ /* a ramp function */
}

Figure 4.44 Program for numerical analysis of suspension system

 numerical methods - 4.45

void d_dt(double t, double state[], double derivative[]){

double Kd;
Kd = damper(state[y_c] - state[y_t]);
derivative[y_c] = state[v_c];
derivative[y_t] = state[v_t];
derivative[v_c] = (Ks1/Mc)*state[y_t] - (Ks1/Mc)*state[y_c]
+ (Kd/Mc)*state[v_t] - (Kd/Mc)*state[v_c] - grav;
derivative[v_t] = (Ks2/Mt)*y_r(t) - ((Ks2+Ks1)/Mt)*state[y_t]
+ (Ks1/Mt)*state[y_c] - (Kd/Mt)*state[v_t]
+ (Kd/Mt)*state[v_c] - grav;
}

main(){
double state[SIZE];
double derivative[SIZE];
FILE *fp_out;
double t;
int i;

state[y_c] = - grav * ( (Mc/Ks1) + (Mt + Mc)/Ks2 ); /* initial values */

state[y_t] = - grav * (Mt + Mc) / Ks2;
state[v_c] = 0.0;
state[v_t] = 0.0;

if((fp_out = fopen("out.txt", "w")) != NULL){ /* open the file */

fprintf(fp_out, " t Yc Yt Vc Vt \n");
for(t = 0.0, i = 0; i < N_step; i++, t += h_step){
if((i % 100) == 0) fprintf(fp_out, "%f %f %f %f %f \n",
t, state[y_c], state[y_t], state[v_c], state[v_t]);
d_dt(t, state, derivative);
integration_step(h_step, state, derivative);
}
} else {
printf("ERROR: Could not open file \n");
}
fclose(fp_out);
}

Figure 4.45 Program for numerical analysis of suspension system (continued)

This program was then used to test various design cases by selecting input types
for changes in the road height, and then calculating how the tire and vehicle heights would
change as a result. Some of these results are seen in Figure 4.46. These results were
obtained by running the program, and then graphing the results in a spreadsheet program.
The input of zero for the road height was used to test the program. As shown the height of
the vehicle changes, indicating that the initial height calculations are correct, and the
model is stable. The step function shows some oscillations that settle out to a stable final
value. The oscillation is relatively slow, and is fully transmitted to the automobile. The
ramp function shows that the car follows the rise of the slope with small transient effects
at the start.
 numerical methods - 4.46

y r ( t ) = 0.0 y r ( t ) = 0.2m
0 0.3
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99
0.25
-0.02
0.2

-0.04 0.15

0.1
Series1 Series
-0.06
Series2 Series
0.05

-0.08 0
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99
-0.05
-0.1
-0.1

-0.12 -0.15

y r ( t ) = 0.2m sin ( t ) y r ( t ) = ( 0.2m )t

0.3
2.5

0.2
2

0.1
1.5

0
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 S eries1 Series
S eries2 1
Series
-0.1

0.5
-0.2

0
-0.3
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99

-0.4 -0.5

Figure 4.46 Graphs of simulation results

4.8 SUMMARY

• State variable equations are used to reduced to first order differential equations.
• First order equations can be integrated numerically.
• Higher order integration, such as Runge-Kutta increase the accuracy.
• Switching functions allow functions terms to be turned on and off to provide
more complex function.
• Tabular data can be used to get numerical values.
 numerical methods - 4.47

4.9 PRACTICE PROBLEMS

1. Convert the following differential equations to state variable form.

·· ·
x + 2x + 3x + 5y = 3
·· ·
y + y + 6y + 9x = sin ( t )

2. a) Put the differential equations given below in state variable form.

b) Put the state equations in matrix form
·· · · ·
y 1 + 2y 1 + 3y 1 + 4y 2 + 5y 2 + 6y 3 + t + F = 0
· · ··
y 1 + 7y 1 + 8y 2 + 9y 2 + 10y 3 + 11y 3 + 5 cos ( 5t ) = 0
· ·· ·
y 1 + 12y 2 + 13y 3 = 0 y 1, y 2, y 3 = outputs
F = input

3. Develop state equations for the mass-spring-damper system below.

x

Kd
F
M Ks

4. The system below is comprised of two masses. There is viscous damping between the masses
and between the bottom mass and the floor. The masses are also connected with a cable that is
run over a massless and frictionless pulley. Write the differential equations for the system, and
put them in state variable form.
x1
Ks
M1

B1
x2
F
M2

B2
 numerical methods - 4.48

5. Do a first order numerical integration of the derivative below from 0 to 10 seconds in one sec-
ond steps. Assume the system starts undeflected. Write the equation for each time step.
d- 2
---- x ( t ) = 5( t – 4 )
dt

6. Given the differential equation below integrate the values numerically for the first ten seconds
with 1 second steps. Assume the initial value of x is 1. You may use first order or Runge Kutta
integration.
·
x + 0.25x = 3

7. Write a Scilab program to calculate the area under the function below using a numerical
method, such as Simpson’s rule. Test it by using the range from x=1 to x=1.2. Assume the sine
function is in radians

f ( x ) = 5x + 2 ln  ----------
sin x
x

8. Write a computer program in a language of your choice (C/C++, Java, Scilab script, etc.) that
will numerically integrate the differential equation below.
·· ·
θ + 3θ + 9θ = 10

9. Given the following differential equation and initial conditions, draw a sketch of the first 5 sec-
onds of the output response. The input is a step function that turns on at t=0. Use at least two
different methods, and compare the results.
·· ·
0.5V o + 0.6V o + 2.1V o = 3V i + 2 initial conditions V i ( t ≥ 0 ) = 5V
V o ( 0 ) = 0V
·
V o ( 0 ) = 1V

10. a) For the mass-spring-damper system below solve the differential equation as a function of
time. Assume the system starts at rest and undeflected. b) Also solve the problem using your
calculator (and state equations) to verify your solution. Sketch the results.
x
K s = 10 Kd
K d = 10 F
M = 10 M Ks
F = 10

11. The mechanical system below is a mass-spring-damper system. A force ’F’ of 100N is applied
to the 10Kg cart at time t=0s. The motion is resisted by the spring and damper. The spring
 numerical methods - 4.49

coefficient is 1000N/m, and the damping coefficient is to be determined. Follow the steps
below to develop a solution to the problem. Assume the system always starts undeflected and
at rest.
a) Develop the differential equation for the system.
b) Solve the differential equation using damping coefficients of 100Ns/m and
10000Ns/m. Draw a graph of the results.
c) Develop the state equations for the system.
d) Solve the system with a first order numerical analysis using Scilab for damping
coefficients of 100Ns/m and 10000Ns/m. Draw a graph of the results.
e) Solve the system with a Runge Kutta numerical analysis using Scilab for damp-
ing coefficients of 100Ns/m and 10000Ns/m. Draw a graph of the results.
f) Write a computer program (in C, Java or Fortran) to do the Runge Kutta numer-
ical integration in step e). Draw a graph of the results.
g) Compare all of the solutions found in the previous steps.
h) Select a damper value to give an overall system damping coefficient of 1. Verify
the results by numerically integrating.
x

Kd
F
M Ks

12. For the mechanism illustrated in the figure below the values are Ks1=Ks2=100N/m,
M1=M2=1kg, F=1N. Assume that the system starts at rest, and the springs are undeformed ini-
tially.
x1 x2

F K s1 K s2
M1 M2

a. Derive the differential equations for the system.

b. Put the equations in state variable form.
c. Put the equations in state variable matrices.
d. Use a calculator to find values for x1 and x2 over the first 10 seconds. Provide
the results in a table in 1 second intervals.
e. Use Scilab to plot the values for the first 10 seconds.
f. Use a Scilab program and the Runge-Kutta method to produce a graph of the
first 10 seconds.
 numerical methods - 4.50

g. Repeat step g. using the first-order approximation method.

h. Use a C program to produce a graph of points for the first 10 seconds.

4.10 PRACTICE PROBLEM SOLUTIONS

1.
·
x = v
·
y = u
·
v = – 2v – 3x – 5y + 3
·
u = – u – 6y – 9x + sin t

x 0 0 1 0 x 0
d- y
---- = 0 0 0 1 y + 0
dt v –3 –5 –2 0 v 3
u –9 –6 0 –1 u sin t

2.
·
y1 = v1
·
y2 = v2
·
y3 = v3
·
v1 = – 2v 1 – 3y 1 – 4v 2 – 5y 2 – 6v 3 – t – F
· – v 13
v2 = -------1- – ------ v 3
12 12
· –v1 7 8 9 11 5
v3 = -------- – ------ y 1 – ------ v 2 – ------ y 2 – ------ y 3 – ------ cos ( 5t )
10 10 10 10 10 10
0 0 0 1 0 0
y1 y1 0
0 0 0 0 1 0
y2 y2 0
0 0 0 0 0 1
d- y 3 y3 0
---- = – 3 –5 0 –2 –4 – 6 + –t–F
dt v 1 13 v1
1
0 0 0 – ------ 0 – ------ 0
v2 12 12 v2
5
7 9 11 1 8 – ------ cos t
v3 – ------ – ------ – ------ – ------ – ------ 0 v3 10
10 10 10 10 10
 numerical methods - 4.51

3.
x

·
· x = v
F Kd x
–K d –Ks
v = v  --------- + x  --------- +  -----
Ks x · F
M
 M  M   M

4.
+y
+x

T Ks x 1 · · ··
M1
∑F = – T – Ks x1 – B1 ( x 1 – x2 ) = M1 x1
·
v1 = x1
· · ·
B1 ( x1 – x2 ) v2 = x2
–K –B B
v 1 = x 1  ---------s + v 1  --------1- + v 2  ------1- +  -------
· –T
 M1   M1   M 1  M 1
· ·
B1 ( x1 – x2 )
· · · ··
T
M2 ∑F = – T + B 1 ( x1 – x2 ) + F – B2 x2 = M2 x2
F
B1 –B1 – B 2
v 2 = v 1  ------- + v 2  --------------------- +  -----------------
· –T+F
· M2 M2 M2
B2 x2

5.

t x
h = 1
x ( t + h ) = x ( t ) + h5 ( t – 4 )
2 0 0
1 80
2 125
3 145
4 150
5 150
6 155
7 175
8 220
9 300
10 425
 numerical methods - 4.52

6.

t x x’
x' = 3 – 0.25x

x ( t + h ) = x ( t ) + h  ----- x ( t )
d 0 1 2.75
 dt  1 3.75 2.06
2 5.81 1.55
x ( t + h ) = x ( t ) + 1 ( 3 – 0.25x ( t ) ) 3 7.36 1.16
4 8.52 0.870
x ( t + h ) = 0.75x ( t ) + 3 5 9.39 0.652
6 10.0 0.489
7 10.5 0.367
8 10.9 0.275
9 11.2 0.206
10 11.4
 numerical methods - 4.53

7.
// integrate.sce - A simple program to integrate a function
// To run this in Scilab use 'File' then 'Exec'.
// by: H. Jack Sept., 9, 2002
// define the function
function foo=f(x)
foo = 5 * x + 2 * log(sin(x) / x);
endfunction
// Set the time length and step size
steps = 10;
x_start = 1;
x_end = 1.2;
x_delta = (x_end - x_start) / steps;
// Loop for rectangular integration
total = 0; // set the initial sum to zero
for i=0:steps,
x = x_start + i * x_delta;
total = total + f(x);
end
total = total * x_delta;
printf("Rectangular integration value %f\n", total);
// Loop for trapezoidal integration
total = 0; // set the initial sum to zero
for i=0:steps,
x = x_start + i * x_delta;
if i == 0 then
total = total + f(x);
elseif i == steps then
total = total + f(x);
else
total = total + 2 * f(x);
end
end
total = total * x_delta / 2;
printf("Trapezoidal integration value %f\n", total);
// Loop for Simpson's rule integration
total = 0; // set the initial sum to zero
even = 0;
for i=0:steps,
x = x_start + i * x_delta;
if i == 0 then
total = total + f(x);
elseif i == steps then
total = total + f(x);
else
even = even + 1;
if even > 1 then
total = total + 4 * f(x);
even = 0;
else
total = total + 2 * f(x);
end
end
end
total = total * x_delta / 3;
printf("Simpsons rule integration value %f\n", total);
 numerical methods - 4.54

8.
#include <stdio.h>

int main(){
int steps = 100,
i;
double theta,
omega,
step_t,
theta_last,
omega_last;
theta = 0.0;
omega = 0.0;
step_t = 1.0;
for(i = 0; i < steps; i++){
theta_last = theta;
omega_last = omega;
theta = theta_last + step_t * omega_last;
omega = omega_last + step_t*(-3 * omega_last - 9 * theta_last + 10);
printf("%f %f %f \n", i+step_t, theta, omega);
}
}
 numerical methods - 4.55

9.
– 0.6t
a) V o ( t ) = – 8.331e cos ( 1.960t – 0.238 ) + 8.095
·
b) Vo = Yo
·
Y o = – 1.2Y o – 4.2V o + 34
// assign4_1.sce
// by: H. Jack Sept., 23, 2003

v0 = 0; // initial conditions
y0 = 1;
X=[v0, y0];

// define the state matrix function

// the values returned are [x, v]
function foo=f(state,t)
foo = [ state($, 2), -1.2*state($,2) - 4.2*state($,1) + 34];
endfunction

// Set the time length and step size for the integration
steps = 1000;
t_start = 0;
t_end = 10;
h = (t_end - t_start) / steps;
t = [t_start];

//
// Loop for integration
//
for i=1:steps,
t = [t ; t($,:) + h];
F1 = h * f(X($,:), t($,:));
F2 = h * f(X($,:) + F1/2.0, t($,:) + h/2.0);
F3 = h * f(X($,:) + F2/2.0, t($,:) + h/2.0);
F4 = h * f(X($,:) + F3, t($,:) + h);
X = [X ; X($,:) + (F1 + 2.0*F2 + 2.0*F3 + F4)/6.0];
end

//
// Graph the values
//
plot2d(t, X, [-2, -5], leg="position@velocity");
xtitle('Time (s)');

//
// Generate points from the given function
//
XX = [v0];
for i=1:steps,
tt = i * h;
XX = [XX ; -8.331 * exp(-0.6 * tt) * cos( 1.960 * tt - 0.238) + 8.095];
end
plot2d(t, XX, [-4], leg="explicit");

c) The two curves produced by the scilab program overlap, so the results agree.
 numerical methods - 4.56

10.
·
Kd x
F · ··
M Ks x
∑F = F – K d x – K s x = Mx
·· ·
Mx + K d x + K s x = F
x ·· ·
10x + 10x + 10x = 10
·· ·
x+x+x = 1
homogeneous: ·· ·
x+x+x = 0
2
A +A+1 = 0
– 1 ± 1 – 4( 1 )( 1 ) 3
A = ---------------------------------------------- = – 0.5 ± j -------
2(1) 2
cos  ------- t + C 2
– 0.5t 3
xh = C1 e
2
particular: xp = B
0+0+B = 1
xp = 1
initial conditions:
cos  ------- t + C 2 + 1
– 0.5t 3
x = xh + xp = C1e
2

sin  ------- t + C 2 – 0.5C 1 e cos  ------- t + C 2

3 – 0.5t 3 – 0.5t 3
x' = – ------- C e
2 1 2 2
3
x' ( 0 ) = – ------- C sin ( C 2 ) – 0.5C 1 cos ( C 2 ) = 0
2 1
3
– ------- sin ( C 2 ) = 0.5 cos ( C 2 )
2

C 2 = atan  ------- = – 0.5236

–1 –1
tan ( C 2 ) = -------
3 3
x ( 0 ) = C 1 cos ( – 0.5236 ) + 1 = 0
–1
C 1 = --------------------------------- = – 1.155
cos ( –0.5236 )

sin  ------- t + 1.047 + 1

– 0.5t 3
x = – 1.155 e
2

First peak is at x=1.219, t=3.63s on graph

 numerical methods - 4.57

11.

·· K d · Ks
x +  ------ x +  ----- x = -----
F
a)
M M M
b) N – 5t
x 1 ( t ) = – 0.115e cos ( 5 3t – 0.524 ) + 0.10
K d = 100 ----
m
N x 1 ( t ) = – 0.1e
– 0.1t – 5 – 999.9t
+ 10 e + 0.10
K d = 10000 ----
m
c) 0 1
X0 X0 0
d
----- = –K –Kd + F-
dt X ---------s --------- X 1 ----
1 M
M M

g) For 100N/m: all solutions are underdamped and overshoot at;

b) 0.1163 at t = 0.363s
d) 0.1166 at t = 0.361s
e) 0.1163 at t = 0.363s
f) 0.1163 at t = 0.360s

For 10000N/m: all solutions are overdamped. The time to reach the time con-
stant (at 0.06321) is,
b) 0.06321 at t = 10.001s
d) 0.06321 at t = 10.000s
e) 0.06321 at t = 10.000s
f) 0.06321 at t = 10.0s

Ns
h) K d = 200 ------ (verify with numerical integration also)
m
 numerical methods - 4.58

12.

b) ·
x1 = v1
– K s1 K s1
x 1 = x 1  ----------- + x 2  -------- +  -------
· F
M1 M1 M1
·
x2 = v2
K s1 – K s1 – K s2
x 1 = x 1  -------- + x 2  ---------------------------
·
M2 M2
c)
0 1 0 0
x1 – K s1 K s1 x1 0
----------- 0 -------- 0 F
d- v 1 M1 M1 v1 -------
---- = + M1
dt x 0 0 0 1 x2
2
0
v2 K s1 – K s1 – K s2
------- - 0 v2
- 0 -------------------------- 0
M2 M2
 numerical methods - 4.59

// System component values

Ks1 = 100;
Ks2 = 100;
e), f), g) M1 = 1;
M2 = 1;
F = 1;

x0 = 0; // initial conditions
v0 = 0;
x1 = 0;
v1 = 0;
X=[x0, v0, x1, v1];

// define the state matrix function the values returned are [x, v]
function foo=f(state,t)
foo = [ state($,2), -Ks1/M1*state($,1)+Ks1/M1*state($,3)+F/M1, state($,4),
Ks1/M2*state($,1)-(Ks1+Ks2)/M2*state($,3)];
endfunction

// Set the time length and step size for the integration
steps = 10000;
t_start = 0;
t_end = 10;
h = (t_end - t_start) / steps;
t = [t_start];

// Loop for integration

for i=1:steps,
t = [t ; t($,:) + h];
F1 = h * f(X($,:), t($,:));
F2 = h * f(X($,:) + F1/2.0, t($,:) + h/2.0);
F3 = h * f(X($,:) + F2/2.0, t($,:) + h/2.0);
F4 = h * f(X($,:) + F3, t($,:) + h);
X = [X ; X($,:) + (F1 + 2.0*F2 + 2.0*F3 + F4)/6.0];
end

// Graph the values for part e)

plot2d(t, X, [-2, -5, -7, -9], leg="position1@velocity1@position2@velocity2");
xtitle('Time (s)');

// printf the values for part f)

printf("\n\nPart e output\n\n");
for time_count=0:20,
i = (time_count/2) / h + 1;
printf("Point at t=%f x1=%f, v1=%f, x2=%f, v2=%f \n", time_count/2, X(i, 1),
X(i, 2), X(i, 3), X(i, 4));
end

// First order integration for part h)

X=[x0, v0, x1, v1];
t = [t_start];
for i=1:steps,
t = [t ; t($,:) + h];
F1 = h * f(X($,:), t($,:));
X = [X ; X($,:) + F1 ];
end
printf("\n\nPart g output \n\n");
for time_count=0:20,
i = (time_count/2) / h + 1;
printf("Point at t=%f x1=%f, v1=%f, x2=%f, v2=%f \n", time_count/2, X(i, 1),
X(i, 2), X(i, 3), X(i, 4));
end
 numerical methods - 4.60

h) The following subroutine is used in place of the subroutine in

the program shown in Figure 4.26 and Figure 4.27.

//
// State Equations Calculated Here
//
void derivative(double t, double X[], double dX[]){
dX[0] = X[1];
dX[1] = -Ks1 / M1 * X[0] + Ks1 / M1 * X[2] + Force / M1;
dX[2] = X[3];
dX[3] = Ks1 / M2 * X[0] - (Ks1 + Ks2) / M2 * X[2];
}

4.11 ASSIGNMENT PROBLEMS

1. Write a Scilab program to implement the following equation to calculate the value of x.
i < 100

x = ∑ 5iu ( i – 20 )
i=0

where,
u(t) = 0 when t≤0
u(t) = 1 when t>0

2. Write a Scilab program to implement the following equation to calculate the value of x.
10
F( t)
v = ∫ ---------
M
- dt M = 10
0

where,
F ( t ) = 10 when t≤5
F(t) = 0 when t>5
 numerical methods - 4.61

3. Write a Scilab program to implement the following equation to calculate the value of x.
10

x = ∫ f ( t ) dt f ( t ) = 5 ln ( t )
0

4. Write a Scilab program to integrate the area under the function below using a numerical
method, such as Simpson’s rule. Find the area from 1 to 2.

f ( x ) = 5x + 2 ln  ----------
sin x
x

5. Numerically integrate one time step of the differential equation below using a) first order inte-
gration and b) Runge Kutta integration.
·· ·
θ + 3θ + 9θ = 10

6. Convert the third order differential equation below to state equation form. With a numerical
method of your choice, find the state of the system 1 second later. Show all calculations.
··· ·· ·
x + 4x + 2x + 5x = 10

7. The differential equation below describes a first order system that starts with an initial value of
k=20. Find the state at two milliseconds using a) explicit integration, b) first order numerical
integration and c) Runge-Kutta integration. For the numerical methods use a timestep of
h=0.001s. ----> The final results must be put in a table for easy comparison.

·
k + 10k = 5 k ( 0 ) = 20

8. For the mechanism shown in the figure below the values are Ks1=Ks2=100N/m, Kd1=10Nm/s,
M1=M2=1kg, F=1N. Assume that the system starts at rest, and the springs are undeformed ini-
 numerical methods - 4.62

tially.

x1 x2

K d1
F K s2
M1 K s1 M2

a. Derive the differential equations for the system.

b. Put the equations in state variable form.
c. Put the equations in state variable matrices.
d. Use a calculator or Scilab using first order integration to find values for x1 and
x2 over the first 10 seconds. Provide the results in a table in 1 second intervals.
e. Use Scilab to plot the values for the first 10 seconds using the values obtained in
part d.
f. Use a Scilab program and the Runge-Kutta method to produce a graph of the
first 10 seconds.
g. Use a C program to produce a list of points for the first 10 seconds.
h. Compare the results found in steps d, f and g in a table.

9. Explicitly solve the following differential equation. Verify the result numerically.
· 2
v + 20v = 200
 rotation - 5.1

5. ROTATION

Topics:
• Basic laws of motion
• Inertia, springs, dampers, levers, gears and belts
• Design cases
Objectives:
• To be able to develop and analyze differential equations for rotational systems.

5.1 INTRODUCTION

The equations of motion for a rotating mass are shown in Figure 5.1. Given the
angular position, the angular velocity can be found by differentiating once, the angular
acceleration can be found by differentiating again. The angular acceleration can be inte-
grated to find the angular velocity, the angular velocity can be integrated to find the angu-
lar position. The angular acceleration is proportional to an applied torque, but inversely
proportional to the mass moment of inertia.

equations of motion

ω =  ----- θ
d
θ (1)
dt
T
d 2
α =  ----- ω =  ----- θ
d (2)
dt dt
OR θ ( t ) = ∫ ω ( t ) dt = ∫ ∫ α ( t ) dt dt (3)

ω ( t ) = ∫ α ( t ) dt (4)
( t )-
α( t) = T
--------- (5)
JM
where,
θ, ω, α = position, velocity and acceleration
J M = second mass moment of inertia of the body
T = torque applied to body

Figure 5.1 Basic properties of rotation

 rotation - 5.2

Note: A ’torque’ and ’moment’ are equivalent in terms of calculations. The main differ-
ence is that ’torque’ normally refers to a rotating moment.

Given the initial state of a rotating mass, find the state 5 seconds later.

rad rad
θ 0 = 1rad ω 0 = 2 --------- α = 3 --------
2
-
s s

ans. θ ( 5 ) = 86rad
rad
ω ( 5 ) = 17 ---------
s

Figure 5.2 Drill problem: Find the position with the given conditions

5.2 MODELING

Free Body Diagrams (FBDs) are required when analyzing rotational systems, as
they were for translating systems. The force components normally considered in a rota-
tional system include,

• inertia - opposes acceleration and deceleration

• springs - resist deflection
• dampers - oppose velocity
• levers - rotate small angles
• gears and belts - change rotational speeds and torques
 rotation - 5.3

5.2.1 Inertia

When unbalanced torques are applied to a mass it will begin to accelerate, in rota-
tion. The sum of applied torques is equal to the inertia forces shown in Figure 5.3.

θ, ω, α
∑T = JM α (6)

J M = I xx + I yy (7)
2
I xx = ∫ y dM (8)
J
2
T I yy = ∫ x dM (9)

Note: The ’mass’ moment of inertia will be used when dealing with acceleration of a
mass. Later we will use the ’area’ moment of inertia for torsional springs.

Figure 5.3 Summing moments and angular inertia

The mass moment of inertia determines the resistance to acceleration. This can be
calculated using integration, or found in tables. When dealing with rotational acceleration
it is important to use the mass moment of inertia, not the area moment of inertia.

The center of rotation for free body rotation will be the centroid. Moment of inertia
values are typically calculated about the centroid. If the object is constrained to rotate
about some point, other than the centroid, the moment of inertia value must be recalcu-
lated. The parallel axis theorem provides the method to shift a moment of inertia from a
centroid to an arbitrary center of rotation, as shown in Figure 5.4.
 rotation - 5.4

J M = J˜M + Mr
2

where,
J M = mass moment about the new point
JM˜ = mass moment about the center of mass

M = mass of the object

r = distance from the centroid to the new point

Figure 5.4 Parallel axis theorem for shifting a mass moment of inertia

J A = J˜A + Ar
2

where,
J A = area moment about the new point
J˜A = area moment about the centroid

A = mass of the object

r = distance from the centroid to the new point

Figure 5.5 Parallel axis theorem for shifting a area moment of inertia

Aside: If forces do not pass through the center of an object, it will rotate. If the object
is made of a homogeneous material, the area and volume centroids can be used as
the center. If the object is made of different materials then the center of mass should
be used for the center. If the gravity varies over the length of the (very long) object
then the center of gravity should be used.

An example of calculating a mass moment of inertia is shown in Figure 5.6. In this

problem the density of the material is calculated for use in the integrals. The integrals are
then developed using slices for the integration element dM. The integrals for the moments
about the x and y axes, are then added to give the polar moment of inertia. This is then
shifted from the centroid to the new axis using the parallel axis theorem.
 rotation - 5.5

The rectangular shape to the right is constrained to rotate about

point A. The total mass of the object is 10kg. The given 4
dimensions are in meters. Find the mass moment of inertia. -2.5
-5 5
First find the density and calculate the moments of inertia about -1
the centroid.
-4
10Kg –2
ρ = -------------------------------- = 0.125Kgm
2 ( 5m )2 ( 4m )
4 4 3 4
–1 y
∫–4 y ∫–4 y
2 2
I xx = dM = ρ2 ( 5m ) dy = 1.25Kgm -----
3 –4
3 3
– 1 ( 4m ) ( – 4m )
∴ = 1.25Kgm  --------------- – ------------------ = 53.33Kgm
2
 3 3 
5
5 5 3
–1 x
∫–5 x ∫–5 x
2 2
I yy = dM = ρ2 ( 4m ) dx = 1Kgm -----
3 –5
3 3
– 1 ( 5m ) ( – 5m )
∴ = 1Kgm  --------------- – ------------------ = 83.33Kgm
2
3 3
2 2 2
J M = I xx + I yy = 53.33Kgm + 83.33Kgm = 136.67Kgm

The centroid can now be shifted to the center of rotation using the parallel axis theorem.

J M = J˜M + Mr = 136.67Kgm + ( 10Kg ) ( ( – 2.5m ) + ( – 1m ) ) = 209.2Kgm

2 2 2 2 2

Figure 5.6 Mass moment of inertia example

 rotation - 5.6

The rectangular shape to the right is constrained to rotate about

point A. The total mass of the object is 10kg. The given
dimensions are in meters. Find the mass moment of inertia 4
WITHOUT using the parallel axis theorem. -2.5
-5 5
-1

-4

ans. 2
I M x = 66.33Kgm
2
I M y = 145.8Kgm
2
J M = 209.2Kgm

Figure 5.7 Drill problem: Mass moment of inertia calculation

 rotation - 5.7

The 20cm diameter 10 kg cylinder to the left is sit-

ting in a depression that is effectively friction-
less. If a torque of 10 Nm is applied for 5
seconds, what will the angular velocity be?

ans. θ ( 5s ) = 312.5rad
rad
ω ( 5s ) = 125 ---------
s

Figure 5.8 Drill problem: Find the velocity of the rotating shaft

5.2.2 Springs

Twisting a rotational spring will produce an opposing torque. This torque increases
as the deformation increases. A simple example of a solid rod torsional spring is shown in
Figure 5.9. The angle of rotation is determined by the applied torque, T, the shear modu-
lus, G, the area moment of inertia, JA, and the length, L, of the rod. The constant parame-
ters can be lumped into a single spring coefficient similar to that used for translational
springs.
 rotation - 5.8

L
JAG
T =  ---------- θ (8)
 L 
θ
T = K S ( ∆θ ) (9)
T

Note: Remember to use radians for these calculations. In fact you are advised to use
radians for all calculations. Don’t forget to set your calculator to radians also.

Note: This calculation uses the area moment of inertia.

Figure 5.9 A solid torsional spring

The spring constant for a torsional spring will be relatively constant, unless the
material is deformed outside the linear elastic range, or the geometry of the spring changes
significantly.

When dealing with strength of material properties the area moment of inertia is
required. The calculation for the area moment of inertia is similar to that for the mass
moment of inertia. An example of calculating the area moment of inertia is shown in Fig-
ure 5.10, and based on the previous example in Figure 5.6. The calculations are similar to
those for the mass moments of inertia, except for the formulation of the integration ele-
ments. Note the difference between the mass moment of inertia and area moment of inertia
for the part. The area moment of inertia can be converted to a mass moment of inertia sim-
ply by multiplying by the density. Also note the units.
 rotation - 5.9

4
-2.5
-5 5
First, the area moment of inertia is calculated about the -1
centroid by integration. All dimensions are in m.
-4
4m
3  4m ) 3 ( – 4m ) 3
= 10m  (--------------
4m 2 4m 2 y-----
I xx = ∫–4m y d A = ∫–4m y 2 ( 5m ) d y = 10m
3  3
- – ------------------- = 426.7
3 
– 4m
5m
5m 2 5m 2 x
3  ( 5m ) 3 ( – 5m ) 3 4
I yy = ∫ x dA = ∫ x 2 ( 4m ) dx = 8m ----- = 8m  --------------- – ------------------- = 666.7m
–5m – 5m 3
– 5m  3 3 
4 4
J˜A = I xx + I yy = ( 426.7 + 666.7 )m = 1093.4m
Next, shift the area moment of inertia from the centroid to the other point of rotation.

2
J A = J˜A + Ar
4 2 2
∴ = 1093.4m + ( ( 4m – ( – 4m ) ) ( 5m – ( – 5m ) ) ) ( ( – 1m ) + ( – 2.5m ) )
4
∴ = 1673m

2
Note: The basic definitions for the area I xx = ∫ y dA (8)
moment of inertia are shown to the 2
right. I yy = ∫ x dA (9)

J A = I xx + I yy (10)
2
J A = J˜A + Ar (11)

Note: You may notice that when the area moment of inertia is multiplied by the den-
sity of the material, the mass moment of inertia is the result. Therefore if you have
a table of area moments of inertia, multiplying by density will yield the mass
moment of inertia. Keep track of units when doing this.

Figure 5.10 Area moment of inertia

 rotation - 5.10

For a 1/2" 1020 steel rod that is 1 yard long, find the torsional spring coefficient.

ans. Nm
K s = 215 ---------
rad

Figure 5.11 Drill problem: Find the torsional spring coefficient

An example problem with torsional springs is shown in Figure 5.12. There are
three torsional springs between two rotating masses. The right hand spring is anchored
solidly in a wall, and will not move. A torque is applied to the left hand spring. Because
the torsional spring is considered massless the torque will be the same at the other end of
the spring, at mass J1. FBDs are drawn for both of the masses, and forces are summed.
(Note: the similarity in the methods used for torsional, and for translational springs.)
These equations are then rearranged into state variable equations, and finally put in matrix
form.
 rotation - 5.11

J M1 J M2
τ K s1 K s2 K s3

Model the system above assuming that the center shaft is a torsional spring, and that a
torque is applied to the leftmost disk. Leave the results in state variable form.

θ1 ··
K s2 ( θ 1 – θ 2 ) + ∑M = τ – K s2 ( θ 1 – θ 2 ) = J M1 θ 1
··
J M1 θ 1 = – K s2 θ 1 + K s2 θ 2 + τ
·
θ1 = ω1 (1)
J M1
– K s2 K s2
ω 1 =  ----------- θ 1 +  -------- θ 2 + τ
·
τ JM JM
(2)
1 1

θ2 ··
K s2 ( θ 2 – θ 1 ) + ∑M = – K s2 ( θ 2 – θ 1 ) – K s3 θ 2 = J M2 θ 2
– K s3 – K s2 K s2
θ 2 =  --------------------------- θ 2 +  -------- θ 1
··
 J M2   J M 2
·
J M2 θ2 = ω2 (3)
– K s3 – K s2 K s2
K s3 θ 2 ω 2 =  --------------------------- θ 2 +  -------- θ 1
· (4)
JM 2
JM 2

0 1 0 0
θ1 – K s2 K s2 θ1 0
----------- 0 -------- 0
d- ω 1 J M1 J M1 ω1
---- = + τ
dt θ 0 0 0 1 θ2 0
2
ω2 - 0 ω2
K s2 – K s3 – K s2 0
-------- 0 --------------------------
J M2 J M2

Figure 5.12 A rotational spring example

 rotation - 5.12

5.2.3 Damping

Rotational damping is normally caused by viscous fluids, such as oils, used for
lubrication. It opposes angular velocity with the relationships shown in Figure 5.13. The
first equation is used for a system with one rotating and one stationary part. The second
equation is used for damping between two rotating parts.

T = Kd ω

T = Kd ( ω1 – ω2 )

Figure 5.13 The rotational damping equation

If a wheel (JM=5kg m2) is turning at 150 rpm and the damping coefficient is 1Nms/rad,
what is the deceleration?

ans.
·· rad-
θ = – 3.141 --------
2
s

Figure 5.14 Drill problem: Find the deceleration

The example in Figure 5.12 is extended to include damping in Figure 5.15. The
primary addition from the previous example is the addition of the damping forces to the
FBDs. In this case the damping coefficients are indicated with ’B’, but ’Kd’ could have
also been used. The state equations were developed in matrix form. Visual comparison of
the final matrices in this and the previous example reveal that the damping terms are the
 rotation - 5.13

only addition.

J M1 J M2
τ K s1 K s2 K s3

B1 B2

Model the system above assuming that the center shaft is a torsional spring, and that a
torque is applied to the leftmost disk. Leave the results in state variable form.

θ1 · ··
K s2 ( θ 1 – θ 2 ) + ∑M = τ – K s2 ( θ 1 – θ 2 ) – B 1 θ 1 = J M1 θ 1
·· ·
J M1 θ 1 = – B 1 θ 1 – K s2 θ 1 + K s2 θ 2 + τ
·
θ1 = ω1 (1)
J M1
–B – K s2 K s2 τ-
ω 1 =  --------1- ω 1 +  ----------- θ +  -------
·
τ - θ + -------
 JM   J M  1  J M  2 J M (2)
·
B1 θ1 1 1 1 1

θ2 · ··
K s2 ( θ 2 – θ 1 ) + ∑ M = – K s2 ( θ 2 – θ 1 ) – B 2 θ 2 – K s3 θ2 = J M 2 θ 2
–B2 · – K s3 – K s2 K s2
θ 2 =  --------- θ 2 +  --------------------------- θ 2 +  -------- θ 1
··
 JM   J M2   JM 
2 2
·
J M2 θ2 = ω2 (3)
– B2 – K s3 – K s2 K s2
K s3 θ 2 ω 2 =  --------- ω 2 +  --------------------------- θ 2 +  -------- θ 1
· (4)
·
B2 θ2 J M2 J M2 J M2

0 1 0 0
θ1 – K s2 –B1 K s2 θ1 0
----------- --------- -------- 0 τ-
d- ω 1 JM1 J M1 J M1 ω1 -------
---- = + J M1
dt θ 0 0 0 1 θ2
2
0
ω2 K s2 – K s3 – K s2 –B2 ω
2
-------- 0 --------------------------- --------- 0
JM2 J M2 J M2

Figure 5.15 A System Example

 rotation - 5.14

5.2.4 Levers

The lever shown in Figure 5.16 can be used to amplify forces or motion. Although
theoretically a lever arm could rotate fully, it typically has a limited range of motion. The
amplification is determined by the ratio of arm lengths to the left and right of the center.

F2
Note: As the lever rotates
d1 d2 the length ratio will be
maintained because of
δ2 similar triangles. This
allows the lever to work
F1
d1 F δ over a range of angles.
----- = -----2- = ----1- The lever above would
δ1 d2 F1 δ2 become ineffective if it
was vertical.

Note: The tip deflection can be related to the angle of δ1 δ

- = ----2-
rotation of the lever if the angle of rotation is small. θ = ----
d1 d2

Figure 5.16 Force transmission with a level

Given a lever set to lift a 1000 kg rock - if the lever is 2m long and the distance from the
fulcrum to the rock is 10cm, how much force is required to lift it?

ans.
F = 516.3N

Figure 5.17 Drill problem: Find the required force

 rotation - 5.15

5.2.5 Gears and Belts

While levers amplify forces and motions over limited ranges of motion, gears can
rotate indefinitely. Some of the basic gear forms are listed below.

Spur - Round gears with teeth parallel to the rotational axis.

Rack - A straight gear (used with a small round gear called a pinion).
Helical - The teeth follow a helix around the rotational axis.
Bevel - The gear has a conical shape, allowing forces to be transmitted at angles.

Gear teeth are carefully designed so that they will mesh smoothly as the gears
rotate. The forces on gears acts at a tangential distance from the center of rotation called
the pitch diameter. The ratio of motions and forces through a pair of gears is proportional
to their radii, as shown in Figure 5.18. The number of teeth on a gear is proportional to the
diameter. The gear ratio is used to measure the relative rotations of the shafts. For example
a gear ratio of 20:1 would mean the input shaft of the gear box would have to rotate 20
times for the output shaft to rotate once.

n1 n –T r n
T 1 = Fc r1 T2 = –Fc r2 ----- = ----2- --------1- = ----1 = ----1-
r1 r2 T2 r2 n2

r –ω –α –∆ θ n
Vc = ω1 r1 = –ω2 r2 ----2 = ---------1 = --------1- = ------------1 = ----2-
r1 ω2 α2 ∆θ 2 n1

where,
n = number of teeth on respective gears
r = radii of respective gears
Fc = force of contact between gear teeth
Vc = tangential velocity of gear teeth
T = torques on gears

Figure 5.18 Basic Gear Relationships

For lower gear ratios a simple gear box with two gears can be constructed. For
higher gear ratios more gears can be added. To do this, compound gear sets are required.
In a compound gear set two or more gears are connected on a single shaft, as shown in
Figure 5.19. In this example the gear ratio on the left is 4:1, and the ratio for the set on the
right is 4:1. Together they give a gear ratio of 16:1.
 rotation - 5.16

N 2 = 60
N2 N4
N 5 = 15 e = ------------- = 16
N3 N5

In this case the output shaft

Input shaft Output shaft turns 16 times faster
than the input shaft. If
we reversed directions
Compound gear set the output (former input)
would now turn 1/16 of
the input (former out-
put) shaft speed.
N 3 = 15
N 4 = 60

Figure 5.19 A compound gear set

A manual transmission is shown in Figure 5.20. In the transmission the gear ratio
is changed by sliding (left-right) some of the gears to change the sequence of gears trans-
mitting the force. Notice that when in reverse an additional compound gear set is added to
reverse the direction of rotation.
 rotation - 5.17

9
clutch 8
stem gear
2 7

Motor shaft
reverse idler

3 4

10
11

6
5

Speed (gear) Gear Train In this manual transmission the gear

1 2-3-6-9
2 2-3-5-8
3 2-3-4-7
4 bypass gear train
reverse 2-3-6-10-11-9
shifter will move the gears in and out of
contact. At this point all of the needed
gears will be meshed and turning. The
final step is to engage the last gear in
the gear train with the clutch (plate)
and this couples the gears to the wheels.

Figure 5.20 A manual transmission

Rack and pinion gear sets are used for converting rotation to translation. A rack is
a long straight gear that is driven by a small mating gear called a pinion. The basic rela-
tionships are shown in Figure 5.21.
 rotation - 5.18

T = Fr V c = ωr ∆l = r∆θ

where,
r = radius of pinion
F = force of contact between gear teeth
Vc = tangential velocity of gear teeth and velocity of rack
T = torque on pinion

Figure 5.21 Relationships for a rack and pinion gear set

Belt based systems can be analyzed with methods similar to gears (with the excep-
tion of teeth). A belt wound around a drum will act like a rack and pinion gear pair. A belt
around two or more pulleys will act like gears.

A gear train has an input gear with 20 teeth, a center gear that has 100 teeth, and an
output gear that has 40 teeth. If the input shaft is rotating at 5 rad/sec what is the
rotation speed of the output shaft?

What if the center gear is removed?

ans.
rad
case 1: ω 3 = 2.5 ---------
s
rad
case 2: ω 3 = – 2.5 ---------
s

Figure 5.22 Drill problem: Find the gear speed

 rotation - 5.19

5.2.6 Friction

Friction between rotating components is a major source of inefficiency in

machines. It is the result of contact surface materials and geometries. Calculating friction
values in rotating systems is more difficult than translating systems. Normally rotational
friction will be given as static and kinetic friction torques.

An example problem with rotational friction is shown in Figure 5.23. Basically

these problems require that the model be analyzed as if the friction surface is fixed. If the
friction force exceeds the maximum static friction the mechanism is then analyzed using
the dynamic friction torque. There is friction between the shaft and the hole in the wall.
The friction force is left as a variable for the derivation of the state equations. The friction
value must be calculated using the appropriate state equation. The result of this calculation
and the previous static or dynamic condition is then used to determine the new friction
value.
 rotation - 5.20

Model the system and consider

the static and kinetic friction
forces on the shaft on the right JM
hand side. τ Ks T s ≤ 10Nm
T k = 6Nm

θ ··
+ ∑M = τ – Ks θ – TF = JM θ
Ks θ
··
JM θ = τ – Ks θ – T F (1)
·
θ = ω (2)
JM
τ – TF Ks
ω =  -------------- +  ------ θ
τ · (3)
JM JM
FF
Next, the torque force must be calculated, and then used to determine the new torque force.
·
J M ω = τ – K s θ – T test
· (4)
T test = τ – K s θ – J M ω

cases:
Not slipping previously
T test ≤ 10Nm T F = T test
T test > 10Nm T F = 6Nm
Slipping previously
T test < 6Nm T F = T test
T test ≥ 6Nm T F = 6Nm

Figure 5.23 A friction system example

The friction example in Figure 5.23 can be analyzed using the C program in Figure
5.24. For the purposes of the example some component values are selected and the system
is assumed to be at rest initially. The program loops to integrate the state equations. Each
loop the friction conditions are checked and then used for a first-order solution to the state
equations.
 rotation - 5.21

int main(){
double h = 0.1, /* time step */
theta, w, /* the state variables */
acceleration, /* the acceleration */
TF, /* friction force */
Ttest, /* the friction test force */
J = 10, /* the moment of inertia (I picked the value) */
tau = 5, /* the applied torque (I picked the value) */
Ks = 10; /* the spring constant (I picked the value) */
int slip = 0; /* the system starts with no slip */
FILE *fp;

theta = 0; w = 0;/* the initial conditions - starting at rest here */

TF = 0.0; /* set the initial friction to 0.0; */
acceleration = 0.0;/* set the initial acceleration to zero also */
if( ( fp = fopen("out.txt", "w")) != NULL){/* open a file to write the results */
for( t = 0.0; t < 10.0; t += h ){/* loop */
Ttest = tau - Ks*theta - J*acceleration;
if(slip == 0){ /* not slipping */
if(Ttest >= 10){
TF = 6; slip = 1;
} else {
TF = Ttest;
}
} else { /* slipping */
if(Ttest < 6){ TF = Ttest; slip = 0;
} else {TF = 6;}
}
acceleration = (tau - TF + Ks*theta) / J;
w = w + h * acceleration;
theta = theta + h * w;
fprintf(fp, "%f, %f, %f\n", t, theta, w);
}
}
fclose(fp);
}

Figure 5.24 A C program for the friction example in Figure 5.23

 rotation - 5.22

5.2.7 Permanent Magnet Electric Motors

DC motors create a torque between the rotor (inside) and stator (outside) that is
related to the applied voltage or current. In a permanent magnet motor there are magnets
mounted on the stator, while the rotor consists of wound coils. When a voltage is applied
to the coils the motor will accelerate. The differential equation for a motor is shown in
Figure 5.25, and will be derived in a later chapter. The value of the constant ’K’ is a func-
tion of the motor design and will remain fixed. The value of the coil resistance ’R’ can be
directly measured from the motor. The moment of inertia ’J’ should include the motor
shaft, but when a load is added this should be added to the value of ’J’.

2 T load
∴ ----- ω + ω ------ = V s  ------ – ------------
  
d K K
 dt  JR  JR JM
where,

ω = the angular velocity of the motor

K = the motor speed constant
J M = the moment of inertia of the motor and attached loads
R = the resistance of the motor coils
T load = a torque applied to a motor shaft

Figure 5.25 Model of a permanent magnet DC motor

The speed response of a permanent magnet DC motor is first-order. The steady-

state velocity will be a straight line function of the torque applied to the motor, as shown
in Figure 5.26. In addition the line shifts outwards as the voltage applied to the motor
increases.

T voltage/current increases

ω ss

Figure 5.26 Torque speed curve for a permanent magnet DC motor

 rotation - 5.23

5.3 OTHER TOPICS

The energy and power relationships for rotational components are given in Figure
5.27. These can be useful when designing a system that will store and release energy.

E = E K + EP (5)

2
EK = JM ω (6)

E P = Tθ (7)

P = Tω (8)

Figure 5.27 Energy and power relations for rotation

Note: The units for various coefficient units

rotational quantities are
Nm-
listed to the right. They may Ks --------
rad
be used to check equations
by doing a unit balance. The Nms
K d, B -----------
unit ’rad’ should be ignored rad
as it appears/disappears spo-
2
radically. JM Kgm

5.4 DESIGN CASE

A large machine is to be driven by a permanent magnet electric motor. A 20:1 gear

box is used to reduce the speed and increase the torque of the motor. The motor drives a
10000kg mass in translation through a rack and pinion gear set. The pinion has a pitch
diameter of 6 inches. A 10 foot long shaft is required between the gear box and the rack
and pinion set. The mass moves on rails with static and dynamic coefficients of friction of
0.2 and 0.1 respectively. We want to select a shaft diameter that will keep the system criti-
cally damped when a voltage step input of 20V is applied to the motor.
 rotation - 5.24

To begin the analysis the velocity curve in Figure 5.28 was obtained experimen-
tally by applying a voltage of 15V to the motor with no load attached. In addition the resis-
tance of the motor coils was measured and found to be 40 ohms. The steady-state speed
and time constant were used to determine the constants for the motor.

rpm
2400

R = 40Ω

0.5s

 ----
d-  K2  K  T
 --------- load
ω + ω  ---------
-  = V s  J R- – -----------
-
 dt m m J R
 M  M JM
The steady-state velocity can be used to find the value of K.
rot  K 
2
( 0 ) +  2400 ---------  ---------- = 15V  ---------- – ( 0 )
K
min  J M R JMR
 2400 --------
rot- 1min ---------------- ( K ) = 15V
------------- 2πrad
 min 60s 1rot 
15V - – 3 Vs
K = ----------------------------
–1
= 39.8 × 10 ---------
120πrads rad
The time constant can be used to find the remaining parameters.
2
K - 1 –1
--------- = ---------- = 2s
JM R 0.5s
2
 39.8 × 10 –3 -------- Vs-
 rad
- = 0.198005 ×10-4 = 19.8 × 10 –6 Kgm 2
J = --------------------------------------------
–1
( 40Ω ) ( 2s )
 ----
d- –1 –1 –2 T load
ω m + ω m 2s = V s ( 50.3V s rad ) – -----------------------------------------
-
 dt –6 2
19.8 × 10 Kgm
θ m' = ω m (1)
– 1 –2 –1 –1 –2
ω m' = V s 50.3V s rad – ω m 2s – 50505Kg m T load (2)
 rotation - 5.25

Figure 5.28 Motor speed curve and the derived differential equation

The remaining equations describing the system are developed in Figure 5.29.
These calculations are done with the assumption that the inertial effects of the gears and
other components are insignificant.
 rotation - 5.26

The long shaft must now be analyzed. This will require that angles at both ends be
defined, and the shaft be considered as a spring.

θ gear, ω gear = angular position and velocity of the shaft at the gear box

θ pinion, ω pinion = angular position and velocity of the shaft at the pinion
1 1
θ gear = ------ θ m ω gear = ------ ω m
20 20

T shaft = K s ( θ gear – θ pinion )

The rotation of the pinion is related to the displacement of the rack through the circum-
ferential travel. This ratio can also be used to find the force applied to the mass.

x mass = θpinion π6in

T shaft = F mass  --------

6in
2
K s ( θ gear – θ pinion ) = F mass  --------
6in
 2 
··
∑ Fmass = Fmass = Mmass xmass
K s ( θ gear – θ pinion )
--------------------------------------------- = M mass θ·· pinion π6in
 6in
--------
 2 
·· –6 –2 –1
θ pinion = ( θ gear – θ pinion )1.768 × 10 in Kg K s

0.0254in 2
θ pinion =  ------ θ m – θ pinion 1.768 × 10 in Kg K s  ---------------------
·· 1 –6 – 2 –1
20 1.0m

θ pinion =  ------ θ m – θ pinion ( 1.141 × 10 )m Kg K s

·· 1 –9 –2 –1
20
·
θ pinion = ω pinion (3)

· – 12 – 2 –1 –9 –2 –1
ω pinion = 57.1 × 10 m Kg K s θ m – 1.141 × 10 m Kg K s θ pinion (4)

Figure 5.29 Additional equations to model the machine

If the gear box is assumed to have relatively small moment of inertia, then we can
say that the torque load on the motor is equal to the torque in the shaft. This then allows
 rotation - 5.27

the equation for the motor shaft to be put into a useful form, as shown in Figure 5.30. Hav-
ing this differential equation now allows the numerical analysis to proceed. The analysis
involves iteratively solving the equations and determining the point at which the system
begins to overshoot, indicating critical damping.

The Tload term is eliminated from equation (2)

· – 1 –2 –1 – 1 –2
ω m = V s 50.3V s rad – ω m 2s – 50505Kg m K s ( θgear – θ pinion )

ω m = V s 50.3V s rad – ω m 2s – 50505Kg m K s  ------ θm – θ pinion

· – 1 –2 –1 – 1 –2 1
 20 
· –1 –2 –1 – 2
ω m = ( V s 50.3V s rad ) + θ pinion ( 50505Kg m K s )
–1 –1 –2
+ ω m ( – 2 s ) + θ m ( – 2525 Kg m K s )
The state equations can then be put in matrix form for clarity. The units will be elimi-
nated for brevity, but acknowledging that they are consistent.

θm 0 1 0 0 θm 0
d- ω m –2525 K s –2 50505K s 0 ωm Vs 50.3
---- = +
dt θ 0 0 0 1 θpinion 0
pinion
– 12 –9
ω pinion 57.1 × 10 K s 0 – 1.141 × 10 K s 0 ω pinion 0

The state equations for the system are then analyzed using a computer for the parame-
ters below to find the Ks value that gives a response that approximates critical
damping for a step input from 0 to 10V.

Ks Overshoot
(rad/Nm) (rad)

100

Figure 5.30 Numerical analysis of system response

These results indicate that a spring value of XXX is required to have the system
 rotation - 5.28

behave as if it is critically damped. (Note: Clearly this system is not second order, but in
the absence of another characteristics we approximate it as second order.)

5.5 SUMMARY

• The basic equations of motion were discussed.

• Mass and area moment of inertia are used for inertia and springs.
• Rotational dampers and springs.
• A design case was presented.

5.6 PRACTICE PROBLEMS

1. Draw the FBDs and write the differential equations for the mechanism below. The right most
shaft is fixed in a wall.

Τ θ1 θ2

JM1 JM2
Ks1 Ks2

B
 rotation - 5.29

2. For the system pictured below a) write the differential equations (assume small angular deflec-
tions) and b) put the equations in state variable form.

A lever arm has a force on

one side, and a spring
Kd1
R1 R2 damper combination
on the other side with a
J1 suspended mass.
x2
F Ks1

M1 x1

3. Draw the FBDs and write the differential equations for the mechanism below.
θ2
R1

JM1
θ1
R2

JM2

Ks
Kd

x1

x2 M1

4. The system below consists of two masses hanging by a cable over mass ‘J’. There is a spring in
the cable near M2. The cable doesn’t slip on ‘J’.
a) Derive the differential equations for the following system.
 rotation - 5.30

b) Convert the differential equations to state variable equations

θ

R
JM
K s1

K s2

x1 M1

M2 x2

5. Write the state equations for the system to relate the applied force ’F’ to the displacement ’x’.
Note that the rotating mass also experiences a rotational damping force indicated with Kd1
K d1
F

r
JM θ

M
K s2
K s1 K s3

K d2

6. For the system pictured below a) write the differential equations (assume small angular deflec-
 rotation - 5.31

tions) and b) put the equations in state variable form.

A round drum with a slot. The slot
drives a lever arm with a sus-
pended mass. A force is applied
R1 R2 R3 R4 to a belt over the drum.

J1

Kd1 Ks1
θ1
F1
M1 x1

7. For the system pictured below a) write the differential equations (assume small angular deflec-
tions) and b) put the equations in state variable form.

Two gears with fixed

R1 R2
centers of rotation
J M1, N 1 x3 and lever arms.
x2 J M2, N 2
F1
Kd1 Ks1

M3 x1

8. For the system pictured below a) write the differential equations (assume small angular deflec-
 rotation - 5.32

tions) and b) put the equations in state variable form.

Kd1 A mass slides on a plane with

dry kinetic friction (0.3). It
is connected to a round
Ks1 mass that rolls and does
not slip.
M1 x1

Ks2

x2
J M2, R 2
θ3 M2 θ2

F2

9. For the system pictured below a) write the differential equations (assume small angular deflec-
tions) and b) put the equations in state variable form.

Kd3 Ks3
A pulley system has the bot-
tom pulley anchored. A
mass is hung in the middle
of the arrangement with M3 x3
springs and dampers on
either side. Assume that
the cable is always tight. Ks2 F1

R2,J2,M2
x2
θ1
R1,J1

10. For the system pictured below a) write the differential equations (assume small angular
 rotation - 5.33

deflections) and b) put the equations in state variable form.

M1

x1
Kd1
Ks1
R1 R2

x2
A mass is suspended over a Kd2 Ks2
lever arm. Forces are
applied to the lower side of
the moment arm through a
spring damper pair. F1 x3

11. For the system pictured below a) write the differential equations (assume small angular deflec-
tions) and b) put the equations in state variable form.
R2
R1
Two gears have a force on
Kd1 one side, and a mass
N1 N 2 on the other, both sus-
pended from moment
J1 J2 arms. There is a rota-
tional damping on one
F1 θ1 Ks1 of the gears.
θ2
M1 x1

12. Find the polar moments of inertia of area and mass for a round cross section with known
radius and mass per unit area. How are they related?

13. The rotational spring is connected between a mass ‘J’, and the wall where it is rigidly held.
The mass has an applied torque ‘T’, and also experiences damping ‘B’.
a) Derive the differential equation for the rotational system shown.
b) Put the equation in state variable form (using variables) and then plot the posi-
tion (not velocity) as a function of time for the first 5 seconds with your calcula-
 rotation - 5.34

tor using the parameters below. Assume the system starts at rest.
θ
Nm Nms
K s = 10 --------- B = 1 -----------
rad rad
T
J M = 1Kgm
2
T = 10Nm J
Ks

B
c) A differential equation for the rotating mass with a spring and damper is given
below. Solve the differential equation to get a function of time. Assume the sys-
tem starts at rest.
–1 –2 –2
θ'' + ( 1s )θ' + ( 10s )θ = 10s

14. Find the response as a function of time (i.e. solve the differential equation to get a function of
time.). Assume the system starts undeflected and at rest.
θ
τ = 10Nm
τ Ks J M = 1Kgm
2

Ns
JM K d = 3 ------
m
N
K s = 9 ----
m
Kd
 rotation - 5.35

5.7 PRACTICE PROBLEM SOLUTIONS

1.
θ1 θ2
T K s1 ( θ 1 – θ 2 )

J M1 J M2

K s1 ( θ 1 – θ 2 ) ·
Bθ 2
K s2 θ 2
··
+ ∑ M1 = T – K s1 ( θ1 – θ 2 ) = J M1 θ 1
K s1 – K s1
θ 1 + θ1  -------
- + θ2  ----------
·· T
- = --------
 JM   JM  J M1
1 1

· ··
+ ∑ M2 = K s1 ( θ 1 – θ 2 ) – Bθ 2 – K s2 θ2 = J M 2 θ 2
K s1 + K s2 – K s1
θ 2 + θ2  -------- + θ2  ----------------------- + θ 1  ----------- = 0
·· · B
JM2 J M2 J M2

2.
·
x1 = v1
– K s1 K s1
x 1 = x 1  ----------- + x 2  -------- + g
·
M1 M1
·
x2 = v2
2 2 2
·  – R 2 K d1  – R 2 K s1  R 2 K s1 – FR 1
v 2 = v 1  ------------------ + x 1  ------------------ + x 2  --------------- + -------------
 J1   J1   J1  J1
 rotation - 5.36

3.

T1
T2

M1 J M1
J M2
·
Kd x1
gM 1 T1 T2

–x 1 x1
if T1,T2,Kdx1 > 0 θ 1 = -------- θ 2 = ------ T1 = Ks ( x2 – x1 )
R2 R1

··
+ ∑ Fy = T 1 – gM 1 = – M 1 x 2
·· T1
x 2 = g – -------
M1
··
+ ∑ M1 = – T 1 R 1 + T 2 R 1 = – J M1 θ 2
·· T1 R1 – T2 R1
θ 2 = ------------------------------
J M1
· ··
+ ∑ M2 = T 2 R 2 – R 2 K d x 1 = – J M2 θ 1
· –R 2 Kd T2 R2
θ 1 + x 1  ---------------- = ------------
··
 JM  JM
2 2

6 equations, 6 unknowns
 rotation - 5.37

4.

θK s1 T K s2 ( x 2 – θR )
a)
JM
RT
M1 M2
RK s2 ( x 2 – θR )
F M1 g M2 g

·· if(T < 0) T=0

∑ FM1 = T – M1 g – F = –M1 x1 if(T >= 0) Rθ = – x 1
·· ··
T = – M 1 x 1 + M 1 g + F = M 1 Rθ + M 1 g + F
··
∑ MJ = – RT – θKs1 + RKs2 ( x2 – θR ) = JM θ
2 2 ··
– R ( M 1 g + F ) – θ ( K s1 + R K s2 ) + ( RK s2 )x 2 = ( J M + R M 1 )θ
2
··  K s1 + R K s2  – R K s2  –R ( M1 g + F )
θ + θ  ----------------------------- + x 2  --------------------------- = --------------------------------
- (1)
 JM + R2 M1   J M + R 2 M 1 JM + R M1
2

··
∑ FM2 = K s2 ( x 2 – θR ) – M 2 g = – M 2 x 2
K s2 –R K s2
x 2 + x 2  -------- + θ  --------------- = g
··
M2 M2 (2)

·
b) θ = ω
2
·  – K s1 – R K s2  RK s2   – RM 1 g – RF
ω = θ  ---------------------------------
- + x 2  --------------------------
- +  ---------------------------------
 JM + R2 M1   J M + R 2 M 1  J M + R 2 M 1 
·
x2 = v2
RK s2 – K s2
v 2 = θ  ------------ + x 2  ----------- + g
·
M2 M2
 rotation - 5.38

5.
·
θ = ω
2
 – K s1 – r K s2 – K d1 K s2 r
ω = θ  -------------------------------- + ω  ------------ + x  ----------- + ------
· Fr
 J M  J M J M J M
·
x = v
K s2 r – K d2 – K s2 – K s3
v = θ  ----------- + v  ------------ + x  ---------------------------
·
 M   M   M 

6.
·
x1 = v1
– K d1 – K s1 K d1 R 2 R 4 K s1 R2 R 4
v 1 = v 1  -----------
- + x 1  ----------
- + ω 1  ---------------------
- + θ1  --------------------
· - +g
 M1   M1   M1 R3   M1 R3 
·
θ1 = ω1
2 2 2 2
K d1 R 2 R 4 K s1 R 2 R 4  – K d1 R 2 R 4  – K s1 R2 R 4 – F 1 R 1
ω 1 = v 1  ---------------------- + x 1  --------------------- + ω 1  ------------------------
· - + θ 1  -----------------------
- + ---------------
J1 R3 J1 R3  J R2  1 3
 J R2  1 3
J1
 rotation - 5.39

7.
· ·
θ1 θ2 · · K d1 ( x 1 – x 3 )
R 2 K d1 ( x 1 – x 3 )
F1 R1 K s1 ( x 1 – x 3 )
··
JM1 JM2 J M2 θ 2

·· N2 M3
J M1 θ 1 τ τ ------
N1 R 2 K s1 ( x 1 – x 3 )
·· M3g
x M3 x1
θ1 N 1 = θ2 N2 θ 2 = -----3-
R2

·
x1 = v1

– K d1 – K s1 R2 K d1 R 2 K s1
v 1 = v 1  ------------ + x 1  ----------- + ω 2  --------------- + θ 1  --------------- – g
·
 M3   M3   M3   M3 

·
θ2 = ω2
–N2
v 1 ( R2 K d1 ) + x 1 ( R 2 K s1 ) + ω 2 ( – R 2 K d1 ) + θ 1 ( – R2 K s1 ) + F  --------- R 1
2 2
· N1
ω 2 = -----------------------------------------------------------------------------------------------------------------------------------------------------------------
2
-
N2
J M1 -----2- + J M2
N1
 rotation - 5.40

8.
·
K d1 x 1
K s2 ( x 1 – x 2 )
K s1 x 1
·· x2
x1 M2 x2 J M2, R 2
x1 M2 θ2
µ k N -------- M1
x1 K s2 ( x 1 – x 2 )
Fc F2
N = M 1 g cos ( θ 3 ) x M 2 g sin ( θ 3 )
θ 2 = -----2-
M 1 g sin ( θ 3 ) R2

·
x1 = v1
·
v1 =

·
x2 = v2

·
v2 =

9.
·
x2 = v2
2 2 2 2 2 2
·  R 1 R 2 ( – K s2 – K s3 )   R 1 R 2 K s2  R 1 R 2 ( 2F 1 + M 2 g )
v 2 = x 3  ----------------------------------------------------------- + x 2  ----------------------------------------------------------- + -----------------------------------------------------------
 4J· 1 R 2 + J 2 R 1 + R 1 R 2 M 2
2 2 2 2
 4J 1 R 2 + J 2 R 1 + R 1 R 2 M 2 4J 1 R 22 + J 2 R 21 + R 21 R 22 M 2
2 2 2 2
x3 = v3
– K d3 – K s2 – K s3 K s2
v 3 = v 3  ------------ + x 3  --------------------------- + x 2  -------- + g
·
M3 M3 M3
 rotation - 5.41

10.
state equations ·
x1 = v1
R1 + R 2
v 1 = F 1  ------------------
·
 R1 M1 
– K s2 K s2 F1
q = x 2  ----------- + x 3  --------- + ---------
·
K d2 K d2 K d1
– K s1 R 1 K s1 R1 + R2
p = x 1  ----------- + x 2  -------------------------------- + F1  ------------------
·
K d1 K d1 ( R 1 + R 2 ) R 1 K d1

R1 + R 2
output equations x 2 = ( x 1 – p )  ------------------
 R1 
x3 = – q + x2

11.
·
x1 = v1
– K s1 R 2 K s1 N 1
v 1 = x 1  ----------- + θ 1  --------------------- + g
·
M1 M1 N2
·
θ1 = ω1
2 2 2 2
·  R 2 K s1 N 1 N 2   – K d1 N 1   – R 2 K s1 N 1  F1 R1 N2
ω 1 = x 1  -----------------------------
- + ω 1  -----------------------------
- + θ 1  -----------------------------
- + -----------------------------
-
 J 1 N 2 + J 2 N 1
2 2
 J 1 N 2 + J 2 N 1
2 2
 J 1 N 2 + J 2 N 1 J 1 N 22 + J 2 N 21
2 2

12.

R R R R
For area: r
4
πR
4
= ∫ r dA = ∫ r ( 2πr dr ) = 2π ∫ r dr = 2π ----
2 2 3
J area = ---------
4 0
2
0 0 0

For mass: ρ = M M-
----- = --------
A 2
πR
R R R R
4 4 2
πR
= ρ  --------- = -----------
r MR
= ∫ r dM = ∫ r ( ρ2πr dr ) = 2πρ ∫ r dr = 2πρ ----
2 2 3
J mass
4  2  2
0
0 0 0
The mass moment can be found by multiplying the area moment by the area density.
 rotation - 5.42

13.
a) T · ··
+ ∑
M = T – K s θ – Bθ = J M θ
JM
·· ·
J M θ + Bθ + K s θ = T
· Ks θ Ks
·· · B T
Bθ θ + θ ------ + θ ------ = ------
JM JM JM

· T Ks B
b) θ' = ω ω = ------ – ------ θ – ------ ω
JM JM JM
Nm Nms
10 --------- 1 -----------
· 10Nm rad rad
ω = -----------------2 – -----------------2 θ – -----------------2 ω
2 1Kgm 1Kgm 1Kgm

ω = -------------2-  10 – --------- – --------- ω

· Nm 10θ s
1 rad rad
Kgm
 -----------
Kgm
-
0  2 m
-  10 – 10θ
· s s- 
5s ω = ---------------------
2 
--------- – -------- ω
Kgm rad rad

ω = s  10 – --------- – --------- ω
· –2 10θ s
 rad rad 
 rotation - 5.43

(c homogeneous:
·· –1 · –2
θ + ( 1s )θ + ( 10s )θ = 0
guess: θ = e At · At ·· 2 At
h θ h = Ae θh = A e
2 At –1 At –2 At
A e + ( 1s )Ae + ( 10s )e = 0
–1 –1 2 –2
2
A + ( 1s )A + 10s
–1 –2
= 0 – 1s ± ( 1s ) – 4 ( 1 ) ( 10s )
A = -------------------------------------------------------------------------------
2( 1)
–1 –2 –2
– 1s ± 1s – 40s –1
A = ------------------------------------------------------- = ( – 0.5 ± j3.123 )s
2(1)
–1
– 0.5s t –1
θh = C1 e cos ( 3.123s t + C 2 )
particular:
–1 –2 –2
θ'' + ( 1s )θ' + ( 10s )θ = 10s
· ··
guess: θp = A θp = 0 θp = 0 –2
–1 –2 –2 10s -
( 0 ) + ( 1s ) ( 0 ) + ( 10s ) ( A ) = 10s A = ------------
–2
= 1
10s
θp = 1
Initial conditions:
–1
– 0.5s t –1
θ ( t ) = C1e cos ( 3.123s t + C 2 ) + 1
–1
– 0.5s 0 –1
θ ( 0 ) = C1 e cos ( 3.123s 0 + C 2 ) + 1 = 0
C 1 cos ( C 2 ) + 1 = 0
–1 –1
–1 – 0.5s t –1 –1 – 0.5s t –1
θ' ( t ) = – 0.5s C 1 e cos ( 3.123s t + C 2 ) – 3.123s C 1 e sin ( 3.123s t + C 2 )
–1 –1
θ' ( 0 ) = – 0.5s C 1 ( 1 ) cos ( C 2 ) – 3.123s C 1 ( 1 ) sin ( C 2 ) = 0

– 0.5 cos ( C 2 ) – 3.123 sin ( C 2 ) = 0

sin ( C 2 ) – 0.5-
-------------------
- = ------------ = tan ( C 2 ) C 2 = – 0.159
cos ( C 2 ) 3.123
–1
C 1 cos ( – 0.159 ) + 1 = 0 C 1 = ------------------------------ = – 1.013
cos ( – 0.159 )
–1
– 0.5s t –1
θ ( t ) = – 1.013 e cos ( 3.123s t – 0.159 ) + 1

14.

cos  ---------- t – 0.524

10 – 1.5t 27
θ ( t ) = ------ + ( – 1.283 )e
9 2
 rotation - 5.44

5.8 ASSIGNMENT PROBLEMS

1. for the system pictured below a) write the differential equations (assume small angular deflec-
tions) and b) put the equations in state variable form.

R2
R1
θ2
N2 M1
N1
x1
J1 J2
x2 θ1 T1
Two gears have levers attached. On one
Kd1 Ks1 side is a mass, the other side a spring
damper pair. A torque is applied to
one gear. Assume the mass remains in
contact with the lever.

2. Draw FBDs for the following mechanical system containing two gears.

Ks1

J 2, R 2
J 1, R 1

Ks2 Kd2
 rotation - 5.45

3. Draw FBDs for the following mechanical system. Consider both friction cases.
Ks1 τ

J,R
x
Ks1
M
µ s, µ k

4. Draw FBDs for the following mechanical system.

J,R
Ks1

J,R

Ks2 Kd2

5. Develop a differential equation of motion for the system below assuming that the cable always
 rotation - 5.46

remains tight.

J,R
Ks1

J,R

Ks2 Kd2

6. Analyze the system pictured below assuming the rope remains tight.
J,R
F

F = 10N
N
M 1 = 1kg K s1 = 100 ----
M1 m
N
Ks1 y M 2 = 1kg K s2 = 100 ----
m
Kd1 µk=0.2 R = 0.1m
Ns
K d1 = 50 ------
2
m
M2 J = 10Kgm Ns
θ=45 K d2 = 50 ------
m

Kd2
Ks2

a) Draw FBDs and write the differential equations for the individual masses.
b) Combine the equations in input-output form with y as the output and F as the
input.
d) Write the equations in state variable matrix form.
c) Use Runge-Kutta to find the system state after 1 second.
 rotation - 5.47

7. Analyze the system pictured below assuming the rope remains tight.
J,R J,R
F = 10N
N
M 1 = 1kg K s1 = 100 ----
m
N
M 2 = 1kg K s2 = 100 ---- J,R
m F
Ns y
R = 0.1m K d1 = 50 ------
2
m
J = 10Kgm Ns
K d2 = 50 ------
m
M1 M2

Ks1 Kd1 Ks2 Kd2

a) Draw FBDs and write the differential equations for the individual masses.
b) Combine the equations in input-output form with y as the output and F as the
input.
c) Write the equations in state variable matrix form.
d) Use Runge-Kutta to find the system state after 1 second.
 input output equations - 6.1

6. INPUT-OUTPUT EQUATIONS

Topics:
• The differential operator, input-output equations
• Design case - vibration isolation
Objectives:
• To be able to develop input-output equations for mechanical systems.

6.1 INTRODUCTION

To solve a set of differential equations we have two choices, solve them numeri-
cally or symbolically. For a symbolic solution the system of differential equations must be
manipulated into a single differential equation. In this chapter we will look at methods for
manipulating differential equations into useful forms.

6.2 THE DIFFERENTIAL OPERATOR

The differential operator ’d/dt’ can be written in a number of forms. In this book
there have been two forms used thus far, d/dt x and x-dot. For convenience we will add a
third, ’D’. The basic definition of this operator, and related operations are shown in Figure
6.1. In basic terms the operator can be manipulated as if it is a normal variable. Multiply-
ing by ’D’ results in a derivative, dividing by ’D’ results in an integral. The first-order
axiom can be used to help solve a first-order differential equation.
 input output equations - 6.2

n
d n
d- x = Dx 1- x =
∫ x dt
---- -------- = D ---
basic definition n
dt dt D

algebraic manipulation Dx + Dy = D ( x + y )
Dx + Dy = Dy + Dx
Dx + ( Dy + Dz ) = ( Dx + Dy ) + Dz

n
D n–m
simplification ---------x- = D x
m
D
( D + a )- = x
x---------------------
(D + a)
x(t) -
= e  ∫ x ( t )e dt + C
first-order axiom – at at
------------------
(D + a)

Figure 6.1 General properties of the differential operator

Note: ·
x + Ax = y ( t ) xD + Ax = y ( t )
xD + Ax = y ( t ) x( D + A) = y( t)
y( t)
At At At x = --------------
e xD + e Ax = e y ( t ) D+A
At At
e xD = e y ( t )

At At
e xD = e y ( t )
d- at at d at
---- ( e x ) = e ----- x + ae x
At dt dt
∫e
At
e x = y ( t )dt + C
at at at
–A t 
De x = e Dx + ae x
At

 ∫ e y ( t )dt + C
x = e

y(t )-
= e  ∫ e y ( t )dt + C
–A t At
-------------
D+A

Figure 6.2 Proof of the first-order axiom

 input output equations - 6.3

Figure 6.3 contains an example of the manipulation of a differential equation using

the ’D’ operator. The solution begins by replacing the ’d/dt’ terms with the ’D’ operator.
After this the equation is rearranged to simplify the expression. Notice that the manipula-
tion follows the normal rules of algebra.

d- 2
 ---- d
x + ----- x + 5x = 5t
 dt dt
2
D x + Dx + 5x = 5t
2
x ( D + D + 5 ) = 5t
5t
x = --------------------------
2
D +D+5
x = t  --------------------------
5
 2 
D +D+5

Figure 6.3 An example of simplification with the differential operator

An example of the solution of a first-order differential equation is given in Figure

6.4. This begins by replacing the differential operator and rearranging the equation. The
first-order axiom is then used to obtain the solution. The initial conditions are then used to
calculate the coefficient values.
 input output equations - 6.4

d-
Given, ---- x + 5x = 3t x ( 0 ) = 10
dt

Dx + 5x = 3t d- 5t 5t
guess, ---- ( te – e )
3t dt
x = -------------
D+5 5t 5t 5t
= 5te + e – e
–5 t 
3tdt + C
5t
∫ e
x = e 5t
= 5te
d- 5t 5t 5t
–5 t 
5t 5t ---- ( te – e ) = 5e t
3  --------------------- + C
te – e dt
x = e
  5   5t 5t
te –e -
∫e
5t
-------------------- = t dt
Initial conditions, 5
(0)( 1) – ( 1)
x ( 0 ) = ( 1 )  3  ----------------------------- + C = 10
  5  

C = 10.6
5t 5t
–5 t  --------------------
te – e - 
x = e 3  + 10.6
5
–5t
x = 0.6 ( t – 1 ) + 10.6e
–5t
x = 0.6t – 0.6 + 10.6e

Figure 6.4 An example of a solution for a first-order system

6.3 INPUT-OUTPUT EQUATIONS

A typical system will be described by more than one differential equation. These
equations can be solved to find a single differential equation that can then be integrated.
The basic technique is to arrange the equations into an input-output form, such as that in
Figure 6.5. These equations will have only a single output variable, and these are always
shown on the left hand side. The input variables (there can be more than one) are all on the
right hand side of the equation, and act as the non-homogeneous forcing function.
 input output equations - 6.5

e.g., ··· ·· · · ··
2y 1 + y 1 + y 1 + 4y 1 = u 1 + u 1 + 3u 2 + u 3 + u 3
·· · ·· · ·
y 2 + 6y 2 + y 2 = u 1 + 3u 2 + u 2 + 0.5u 2 + u 3
where,
y = outputs
u = inputs

Figure 6.5 Developing input-output equations

A sample derivation of an input-output equation from a system of differential

equations is given in Figure 6.6. This begins by replacing the differential operator and
combining the equations to eliminate one of the output variables. The solution ends by
rearranging the equation to input-output form.

Given the differential equations,

· ·
y 1 = – 3y 1 + 2y 2 + u1 + 2u 2 (1)
· ·
y 2 = 2y 1 + y 2 + u1 (2)

Find the input-output equations.

(1) Dy 1 = – 3y 1 + 2y 2 + u 1 + 2Du 2
∴y 1 ( D + 3 ) = 2y 2 + u 1 + 2Du 2

∴y 2 = y 1  ------------- – 0.5u 1 – Du 2
D+3
2
(2) Dy 2 = 2y 1 + y 2 + Du 1

∴y 2 ( D – 1 ) = 2y 1 + Du 1

∴ y 1  ------------- – 0.5u 1 – Du2 ( D – 1 ) = 2y 1 + Du 1

D+3
 2 
2
∴y 1  ----------------------------- – 2 – 0.5Du 1 + 0.5u 1 – D u 2 + Du2 = Du 1
D + 2D – 3 2
2
2 2
∴0.5D y 1 + Dy 1 – 3.5y 1 = Du 1 + 0.5Du 1 – 0.5u 1 + D u 2 – Du 2

∴0.5y 1'' + y 1' – 3.5y 1 = u 1' + 0.5u1' – 0.5u 1 + u 2' – u 2'

Figure 6.6 An input output equation example

 input output equations - 6.6

Find the second equation for the example in Figure 6.6 for the output y2.

Figure 6.7 Drill problem: Find the second equation in the previous example

6.3.1 Converting Input-Output Equations to State Equations

In some instances we will want to numerically integrate an input-output equation.

The example starting in Figure 6.8 shows the development of an input-output equation for
two freely rolling masses joined by a spring. The final equation has a derivative on the
right hand side that would prevent it from being analyzed in many cases. In particular if
the input force ’F’ was a step function the first derivative would yield an undefined (infi-
nite) value that could not be integrated.
 input output equations - 6.7

x1 x2

F M1 Ks M2

Equations of motion can be derived for these masses.

Ks ( x2 – x 1 )
∑ Fx
2
F M1 = F + Ks ( x2 – x1 ) = M1 D x1
2
x1 ( M1 D + K s ) = F + Ks x2

Ks ( x2 – x 1 ) ∑ Fx
2
= –Ks ( x2 – x1 ) = M2 D x2
M2
2
Ks x1 = x 2 ( M2 D + Ks )
The equations can be combined to eliminate x2.
2  Ks x 1 
x 1 ( M 1 D + K s ) = F + K s  --------------------------
-
 M 2 D 2 + K s
2 2 2 2
x1 ( ( M1 D + K s ) ( M2 D + Ks ) – Ks ) = F ( M2 D + Ks )
4 2 2 2 2
x1 ( D M1 M2 + D Ks ( M1 + M2 ) + Ks – Ks ) = F ( M 2 D + Ks )
4 2 2
x1 ( D M1 M2 + D Ks ( M1 + M2 ) ) = F ( M2 D + Ks )
d- 4
 ---- d 2 d 2
x 1 M 1 M 2 +  ----- x 1 K s ( M 1 + M 2 ) =  ----- FM + FK s
 dt  dt  dt 2

Figure 6.8 Writing an input-output equation as a differential equation

The equation is then converted to state variable form, including a step to calculate
a second derivative of the input, as shown in Figure 6.9.
 input output equations - 6.8

This can then be written in state variable form by creating dummy variables for integrat-
ing the function ’F’.

 ----
d-
 dt x 1 = v 1
 ----
d-
v = a1
 dt 1
 ----
d-
 dt a 1 = d 1
 ----
d-
 dt d 1 M 1 M 2 + a 1 K s ( M 1 + M 2 ) = a F M 2 + FK s
 ----
d – K s ( M 1 + M 2 ) Ks 
- d 1 = a 1  ----------------------------------
- + aF  ------
1 -  --------------
-
+ F
 dt  M1 M2   M 1  M 1 M 2
The approximate value of the second derivative of a unit time step can be calculated using
the time step.
1 (when the timestep, step, it is turned on)
a F0 = -----2
T
1
a F 1 = – ----2- (after the first timestep)
T
v F = TaF 0 + Ta F1 = T  -----2 + T  – ----2- = 0
1 1
(to verify)
T T
2 2 2 2
x F = ----- a F + ----- aF = -----  -----2 + -----  -----2 = --- + --- = 1
T T T 1 T 1 1 1
2 2 2 T  2 T  2 2
These equations can then be written in matrix form.

x1 0 1 0 0 x 0 0
1
0 0 1 0 0 0
d- v 1 v1 aF
---- = 0 0 0 1 + 0 0
dt a a1 F
1 –Ks ( M1 + M2 ) Ks
1 - --------------
d1 0 0 ----------------------------------- 0 d ------ -
M1 M2 1 M1 M1 M2

Figure 6.9 Writing state equations for equations with derivatives

 input output equations - 6.9

6.3.2 Integrating Input-Output Equations

An input-output equation is already in a form suitable for normal integration tech-

niques, with the left hand side being the homogeneous part, and the right hand side is the
particular part. If the non homogeneous part includes derivatives, these determine the val-
ues of initial conditions. An example of explicitly solving such an equation is shown in
Figure 6.10 and Figure 6.11.

The solution begins by evaluating the homogeneous equation.

·· · ·
x + 3x + 2x = 4F + 5F
The solution begins by evaluating the homogeneous equation as normal.
·· ·
x + 3x + 2x = 0
2
A + 3A + 2 = 0
2
–3 ± 3 – 4 ( 2 )
A = ------------------------------------- = –1, – 2
2
–t – 2t
xh = C1 e + C 2 e
The particular solution can also be found as normal, assuming F is unit step function.

Guess, · ··
x = A x = 0 x = 0

0 + 3 ( 0 ) + 2A = 2 ( 0 ) + 5 ( 1 ) 5
A = ---
2
The derivative in the non-homogeneous solution must now be used to find the initial con-
ditions. However the initial position and velocity are known to be zero.
2
 ----
d-
x + 3  ----- x 0 + 2x 0 = 4  ----- F  + 5F 0
d d
 dt 0 dt dt 1
for a step function
2
 ----
d-  ----
d- 
 dt x 0 + 3 ( 0 ) + 2 ( 0 ) = 4  dt 1 + 5 ( 0 ) · 1
u ( t ) = -----
dt
2
 ----
d-  ----
d- 
 dt x 0 = 4  dt 1
d-
---- x = 4 Note: This will be used as an initial condition
dt 0

Figure 6.10 Integrating an input-output equation

 input output equations - 6.10

The initial conditions can then be used to find the values of the coefficients. It will be
assumed that the system starts undeflected and at rest.
–t – 2t 5
x ( t ) = C 1 e + C2 e + ---
2
–0 –0 5 5
x ( 0 ) = C1e + C2e + --- = 0 C 1 = – C 2 – ---
2 2
· –t – 2t
x ( t ) = – C 1 e – 2C 2 e
·
x ( 0 ) = – C 1 – 2C 2 = 4

–  – C 2 – --- – 2C 2 = 4
5 3
2 C 2 = – ---
2
C 1 = –  – --- – --- = – 1
3 5
 2 2
The final equation can then be written.

– t 3 – 2t 5
x ( t ) = – e – --- e + ---
2 2

Figure 6.11 Integrating an input-output equation (cont’d)

The example in Figure 6.10 is reconsidered with a sinusoidal input in Figure 6.12.
In this case the initial acceleration is found to be non-zero. In practical terms, this can be
ignored because only the initial position and velocity will be used to find the coefficients.
 input output equations - 6.11

Assume a sinusoidal input, with the system initially at rest.

·
F ( t ) = sin ( t ) F ( t ) = cos ( t )
·
F(0) = 0 F(0) = 1
d- 2
 ---- ·
x 0 + 3  ----- x 0 + 2x 0 = 4 ( F ( 0 ) ) + 5F ( 0 )
d
 dt dt
2
 ----
d-
 dt x 0 + 3 ( 0 ) + 2 ( 0 ) = 4 ( 1 ) + 5 ( 0 )
2
 ----
d- This indicates that the acceleration will have an initial value,
 dt x 0 = 4 but it will not affect the initial position or velocity.

Figure 6.12 Initial conditions for a sinusoidal input

6.4 DESIGN CASE

The classic mass-spring-damper system is shown in Figure 6.13. In this example

the forces are summed to provide an equation. The differential operator is replaced, and
the equation is manipulated into transfer function form. The transfer function is given in
two different forms because the system is reversible and the output could be either ’F’ or
’x’.
 input output equations - 6.12

2
dx- d x
∑ Fy = – F + Kd -----
dt
+ K s x = – M -------2-
dt
Ks 2
Kd d x dx
F = M -------2- + K d ------ + K s x
dt dt
M x 2
F = MD x + K d Dx + K s x
F 2
--- = MD x + K d Dx + K s x
F x
OR
--x- = -----------------------------------------
1 -
F 2
MD + K d D + K s

Aside: An important concept that is ubiquitous yet largely unrecognized is the use of
functional design. We look at parts of systems as self contained modules that use
inputs to produce outputs. Some systems (such a mechanisms) are reversible, others
are not (consider a internal combustion engine, turning the crank does not produce
gasoline). An input is typically something we can change, an output is the resulting
change in a system. For the example above ‘F’ over ‘x’ implies that we are chang-
ing the input ‘x’, and there is some change in ‘F’. We know this could easily be
reversed mathematically and practically.

Figure 6.13 A transfer function for a mechanical system

Aside: Keep in mind that the mathematical expression ‘F/x’ is a ratio between input
(displacement action) and output (reaction force). When shown with differentials it
is obvious that the ratio is not simple, and is a function of time. Also keep in mind
that if we were given a force applied to the system it would become the input (action
force) and the output would be the displacement (resulting motion). To do this all
we need to do is flip the numerators and denominators in the transfer function.

Mass-spring-damper systems are often used when doing vibration analysis and
design work. The first stage of such analysis involves finding the actual displacement for a
given displacement or force. A system experiencing a sinusoidal oscillating force is given
in Figure 6.14. Numerical values are substituted and the homogeneous solution to the
equation is found.
 input output equations - 6.13

Given the component values input force,

N Ns
M = 1Kg K s = 2 ---- K d = 0.5 ------ F = 5 sin ( 6t ) N
m m
The differential equation for the mass-spring damper system can be written.
2
1Kg 2 +  0.5 ------ ------ +  2 ---- x = 5 sin ( 6t ) N
d x-  Ns
------- dx N
dt m dt m
The homogeneous solution can be determined.
2
1Kg -------2- +  0.5 ------ ------ +  2 ---- x = 0
d x Ns dx N
 m dt   m
dt
Ns 2
– 0.5 ------ ±  0.5 ------ – 4 ( 1Kg )  2 ----
Ns N
m m m
A = -------------------------------------------------------------------------------------------
2 ( 1Kg )
–1
A = 0.5 ( – 0.5 ± 0.25 – 8 )s
–1
A = ( – 0.25 ± 1.392j )s
– 0.25t
xh = C1 e cos ( 1.392t + C2 )

Figure 6.14 Explicit analysis of a mechanical system

The solution continues in Figure 6.15 where the particular solution is found and
put in phase shift form.
 input output equations - 6.14

The particular solution can now be found with a guess.

2
1Kg -------2- +  0.5 ------ ------ +  2 ---- x = 5 sin ( 6t ) N
d x Ns dx N
dt m dt m
x p = A sin 6t + B cos 6t
x p' = 6A cos 6t – 6B sin 6t
x p'' = – 36 A sin 6t – 36B cos 6t
–36 A sin 6t – 36B cos 6t + 0.5 ( 6A cos 6t – 6B sin 6t ) + 2 ( A sin 6t + B cos 6t ) = 5 sin ( 6t )
34
– 36B + 3A + 2B = 0 A = ------ B
3
– 36A – 3B + 2A = 5
– 34  ------ B – 3B = 5
34
3
5 34
B = ----------------------------- = – 0.01288 A = ------ ( – 0.01288 ) = – 0.1460
34 ( 34 )-
–------------------- 3
–3
3
x p = ( – 0.1460 ) sin 6t + ( – 0.01288 ) cos 6t
2 2
( – 0.1460 ) + ( – 0.01288 )
x p = ------------------------------------------------------------------- ( ( – 0.1460 ) sin 6t + ( – 0.01288 ) cos 6t )
2 2
( – 0.1460 ) + ( – 0.01288 )

x p = 0.1466 ( – 0.9961 sin 6t –0.08788 cos 6t )

x p = 0.1466 sin  6t + atan  ---------------------- 

– 0.9961
– 0.08788
x p = 0.1466 sin ( 6t + 1.483 )

Figure 6.15 Explicit analysis of a mechanical system (continued)

The system is assumed to be at rest initially, and this is used to find the constants in
the homogeneous solution in Figure 6.16. Finally the displacement of the mass is used to
find the force exerted through the spring on the ground. In this case there are two force fre-
quency components at 1.392rad/s and 6rad/s. The steady-state force at 6rad/s will have a
magnitude of .2932N. The transient effects have a time constant of 4 seconds (1/0.25), and
should be negligible within a few seconds of starting the machine.
 input output equations - 6.15

The particular and homogeneous solutions can now be combined.

– 0.25t
x = xh + xp = C1 e cos ( 1.392t + C 2 ) + 0.1466 sin ( 6t + 1.483 )
– 0.25t – 0.25t
x' = – 0.25C 1 e cos ( 1.392t + C 2 ) – 1.392 ( C 1 e sin ( 1.392t + C 2 ) )
+ 6 ( 0.1466 cos ( 6t + 1.483 ) )

The initial conditions can be used to find the unknown constants.

0
0 = C 1 e cos ( 0 + C 2 ) + 0.1466 sin ( 0 + 1.483 )

C 1 cos ( C 2 ) = – 0.1460
– 0.1460
C 1 = --------------------
cos ( C 2 )
0 0
0 = – 0.25C 1 e cos ( 0 + C 2 ) – 1.392 ( C 1 e sin ( 0 + C 2 ) ) + 6 ( 0.1466 cos ( 0 + 1.483 ) )

0 = – 0.25C 1 cos ( C 2 ) – 1.392 ( C 1 sin ( C 2 ) ) + 0.07713

0 = – 0.25  -------------------- cos ( C 2 ) – 1.392  -------------------- sin ( C 2 ) + 0.07713

–0.1460 – 0.1460
cos ( C 2 ) cos ( C 2 )
0 = 0.0365 + ( 0.2032 ) tan ( C 2 ) + 0.07713

C 2 = atan  ------------------------------------------ = – 0.5099

0.0365 + 0.07713
– 0.2032
– 0.1460
C 1 = --------------------------------- = – 0.1673
cos ( –0.5099 )
– 0.25t
x = ( – 0.1673 e cos ( 1.392t – 0.5099 ) + 0.1466 sin ( 6t + 1.483 ) )m
The displacement can then be used to calculate the force transmitted to the ground,
assuming the spring is massless.
F = Ks x

F =  2 ---- ( – 0.1673 e
N – 0.25t
cos ( 1.392t – 0.5099 ) + 0.1466 sin ( 6t + 1.483 ) )m
m
– 0.25t
F = ( – 0.3346 e cos ( 1.392t – 0.5099 ) + 0.2932 sin ( 6t + 1.483 ) )N

Figure 6.16 Explicit analysis of a mechanical system (continued)

A decision has been made to reduce the vibration magnitude transmitted to the
ground to 0.1N. This can be done by adding a mass-spring isolator, as shown in Figure
6.17. In the figure the bottom mass-spring-damper combination is the original system. The
 input output equations - 6.16

mass and spring above have been added to reduce the vibration that will reach the ground.
Values must be selected for the mass and spring. The design begins by developing the dif-
ferential equations for both masses.

K s2

x2
M2

Kd K s1

M1 x1

· · ··
– K s2 x 2
∑F = –K s2 x 2 – K d ( x 2 – x 1 ) – K s1 ( x 2 – x 1 ) = M 2 x 2
2
M2 – K s2 x 2 – K d ( x 2 D – x1 D ) – K s1 ( x 2 – x 1 ) = M 2 x 2 D
2
· ·
Kd ( x2 – x1 ) K s1 ( x 2 – x 1 ) x 2 ( – K s2 – K d D – K s1 – M 2 D ) = x 1 ( – K d D – K s1 )
2
 K s2 + K d D + K s1 + M 2 D 
x 1 = x 2  ------------------------------------------------------------- (1)
 K d D + K s1 
· ·
Kd ( x2 – x1 ) K s1 ( x 2 – x 1 ) · · ··
∑F = K d ( x 2 – x 1 ) + K s1 ( x 2 – x 1 ) – F = M 1 x 1

2
M1 K d ( x 2 D – x 1 D ) + K s1 ( x 2 – x 1 ) – F = M 1 x 1 D
2
F x 1 ( – K d D – K s1 – M 1 D ) + x 2 ( K d D + K s1 ) = F (2)

Figure 6.17 Vibration isolation system

For the design we are only interested in the upper spring, as it determines the force
on the ground. An input-output equation for that spring is developed in Figure 6.18. The
 input output equations - 6.17

given values for the mass-spring-damper system are used. In addition a value for the upper
mass is selected. This is arbitrarily chosen to be the same as the lower mass. This choice
may need to be changed later if the resulting spring constant is not practical.

The solution begins by combining equations (1) and (2) and inserting the.numerical val-
ues for the lower mass, spring and damper. We can also limit the problem by selecting a
mass value for the upper mass.

2
 K s2 + K d D + K s1 + M 2 D  2
x 2  ------------------------------------------------------------- ( – K d D – K s1 – M 1 D ) + x 2 ( K d D + K s1 ) = F
 K d D + K s1 
N Ns
M 1 = 1Kg K s1 = 2 ---- K d = 0.5 ------ M 2 = 1Kg
m m
2
 K s2 + 0.5D + 2 + D  2
x 2  ------------------------------------------------- ( – 0.5D – 2 – D ) + x 2 ( 0.5D + 2 ) = F
 0.5D + 2 
2 2 2
x 2 ( D + 0.5D + 2 + K s2 ) ( D – 0.5D – 2 ) + x 2 ( 0.5D + 2 ) = F ( 0.5D + 2 )
4 2 1
x 2 ( D ( – 1 ) + D ( K s2 ) + D ( –0.5K s2 ) + ( – 2K s2 ) ) = F ( 0.5D + 2 )
This can now be converted back to a differential equation and combined with the force.
d 4 d 2
–  ----- x + K s2  ----- x 2 – 0.5K s2  ----- x 2 – 2K s2 x 2 = 0.5  ----- F + 2F
d d
dt 2  dt   dt   dt

d 4 d 2
–  ----- x + K s2  ----- x 2 – 0.5K s2  ----- x 2 – 2K s2 x 2 = 0.5  ----- 5 sin ( 6t ) + 2 ( 5 ) sin ( 6t )
d d
 dt 2  dt  dt  dt
d 4 d 2
–  ----- x + K s2  ----- x 2 – 0.5K s2  ----- x 2 – 2K s2 x 2 = 15 cos ( 6t ) + 10 sin ( 6t )
d
 dt 2  dt  dt

Figure 6.18 Developing an input output equation

This particular solution of the differential equation will yield the steady-state dis-
placement of the upper mass. This can then be used to find the needed spring coefficient.
 input output equations - 6.18

The particular solution begins with a guess.

d 4 d 2
–  ----- x + K s2  ----- x 2 – 0.5K s2  ----- x 2 – 2K s2 x 2 = 15 cos ( 6t ) + 10 sin ( 6t )
d
dt 2 dt dt
x p = A sin 6t + B cos 6t
 ----
d-
 dt x p = 6A cos 6t – 6B sin 6t
d- 2
 ---- x = – 36 A sin 6t – 36B cos 6t
 dt p
d- 3
 ---- x = – 216 A cos 6t + 216B sin 6t
 dt p
4
 ----
d-
 dt x p = 1296A sin 6t + 1296B cos 6t
sin ( 6t ) ( – 1296A – 36AK s2 + 0.5K s2 6B – 2K s2 A ) = 10 sin ( 6t )
A ( – 1296 – 38K s2 ) + B ( 3K s2 ) = 10
10 + A ( 1296 + 38K s2 )
B = ------------------------------------------------------
3K s2
cos ( 6t ) ( – 1296B – 36BK s2 + ( – 0.5 )K s2 6A – 2K s2 B ) = 15 cos ( 6t )
A ( – 3K s2 ) + B ( – 1296 – 38K s2 ) = 15
10 + A ( 1296 + 38K s2 )
A ( – 3K s2 ) + ------------------------------------------------------ ( – 1296 – 38K s2 ) = 15
3K s2
2
A ( – 9K s2 ) + ( 10 + A ( 1296 + 38K s2 ) ) ( – 1296 – 38K s2 ) = 45K s2
2
 – 9K s2  45K s2
A  ------------------------------------------ + A ( 1296 + 38K s2 ) = ------------------------------------------ – 10
 ( – 1296 – 38K s2 ) ( – 1296 – 38K s2 )
45K s2
-----------------------------------------
- – 10
( – 1296 – 38K s2 )
A = --------------------------------------------------------------------------------------
2
– 9K s2
------------------------------------------ + ( 1296 + 38K s2 )
( – 1296 – 38K s2 )
45K s2 + 10 ( 1296 + 38K s2 )
A = ------------------------------------------------------------------------------------------------------
2
-
– 9K s2 + ( 1296 + 38K s2 ) ( – 1296 – 38K s2 )
425K 2s + 12960
A = ---------------------------------------------------------------------------------
2
-
– 1453K 2s – 98496K 2s – 1679616

Figure 6.19 Finding the particular solution

 input output equations - 6.19

The value for B can then be found.

 425K 2s + 12960 
10 +  --------------------------------------------------------------------------------- - ( 1296 + 38K s2 )
 – 1453K 22s – 98496K 2s – 1679616
B = ------------------------------------------------------------------------------------------------------------------------------------------
3K s2
2
10 ( – 1453K 2s – 98496K 2s – 1679616 ) + ( 425K 2s + 12960 ) ( 1296 + 38K s2 )
B = -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
2
-
3K s2 ( – 1453K 2s – 98496K 2s – 1679616 )
2 8
1620K 2s + 2.097563 ×10 K 2s
B = --------------------------------------------------------------------------------------------------
2
-
3K s2 ( – 1453K 2s – 98496K 2s – 1679616 )
540K 2s + 69918768
B = ---------------------------------------------------------------------------------
2
-
– 1453K 2s – 98496K 2s – 1679616

Figure 6.20 Finding the particular solution (cont’d)

Finally the magnitude of the particular solution is calculated and set to the desired
amplitude of 0.1N. This is then used to calculate the spring coefficient.

2 2
amplitude = A +B
 425K 2s + 12960 2  540K 2s + 69918768 2
0.1 =  ---------------------------------------------------------------------------------
- +  ----------------------------------------------------------------------------------
 – 1453K 22s – 98496K 2s – 1679616  – 1453K 22s – 98496K 2s – 1679616

A value for the spring coefficient was then found using Mathcad to get a value of 662N/m.

Figure 6.21 Calculation of the spring coefficient

 input output equations - 6.20

6.5 SUMMARY

• The differential operator can be manipulated algebraically

• Equations can be manipulated into input-output forms and solved as normal dif-
ferential equations

6.6 PRACTICE PROBLEMS

1. Develop the input-output equation for the mechanical system below. There is viscous damping
between the block and the ground. A force is applied to cause the mass the accelerate.
x

F
M
B

2. Find the input-output form for the following equations.

·
y+y+x = 3
·
x+x+y = 0

3. Find the input-output form for the following equations.

·· · ·
x 1 + x 1 + 2x 1 – x 2 – x 2 = 0
· ·· ·
– x1 – x1 + x2 + x2 + x2 = F

4. The following differential equations were converted to the matrix form shown. Use Cramer’s
rule to find an input-output equation for ‘y’.
·· F
y + 2x = ------ 2 F
10 (D ) (2) y = ------
10
· ·· 2
7y + 4y + 9x + 3x = 0 ( 7D + 4 ) ( 9D + 3 ) x 0
 input output equations - 6.21

5. Find the input output equation for y2. Ignore the effects of gravity.

K s1

M1 y1

K s2

M2 y2

6. Find the input-output equations for the systems below. Here the input is the torque on the left
hand side.
θ1 θ2

J1 J2
τ K s1 K s2

B1 B2

7. Write the input-output equations for the mechanical system below. The input is force ‘F’, and
the output is ‘y’ or the angle theta (give both equations). Include the inertia of both masses, and
 input output equations - 6.22

gravity for mass ‘M’.

K
θ

R JM

K y

8. The applied force ‘F’ is the input to the system, and the output is the displacement ‘x’.

K1 = 500 N/m

x K2 = 1000 N/m

M = 10kg
w

a) Find x(t), given F(t) = 10N for t >= 0 seconds.

b) Using numerical methods, find the steady-state response for an applied force of
F(t) = 10cos(t + 1) N ?
a) Solve the differential equation to find the explicit response for an applied force
of F(t) = 10cos(t + 1) N ?
d) Set the acceleration to zero and find an approximate solution for an applied
force of F(t) = 10cos(t + 1) N. Compare the solution to the previous solutions.

6.7 PRACTICE PROBLEM SOLUTIONS

1.
 input output equations - 6.23

x + x  ----- = -----
·· · B F
 M M

2.
·· ·
x + 2x = –3
·· ·
y + 2y = 3

3.

d- 4
 ---- d 3 d 2
x + 2  ----- x + 3  ----- x +  ----- x + x 1 =  ----- F + F
d d
 dt 1 dt 1 dt 1  dt 1 dt
d- 4
 ---- d- 3
 ---- d- 2
 ----  ----
d- d- 2
 ----  ----
d-
x + 2 x + 3 x + x + x =
 dt 2  dt 2  dt 2  dt 2 2  dt F +  dt F + 2F

4.

F-
----- (2)
10
F- 2
0 ( 9D + 3 )
2 ----- ( 9D + 3 ) 2
10 F ( 0.9D + 0.3 ) -
y = -------------------------------------------------------- = --------------------------------------------------------------
2 2
= ---------------------------------------------------
4 2
2 D ( 9D + 3 ) – 2 ( 7D + 4 ) 9D + 3D – 14D – 8
(D ) (2)
2
( 7D + 4 ) ( 9D + 3 )
4 2 2
y ( 9D + 3D – 14D – 8 ) = F ( 0.9D + 0.3 )

d- 4
 ---- d- 2
 ---- d- 1
 ---- d- 2
 ----
( ) ( ) ( ) ( )  dt F ( 0.9 ) + F ( 0.3 )
y 9 + y 3 + y – 14 + y – 8 =
 dt  dt  dt

d- 4
 ---- d 2 1 d 1 – 14 d 2 1
y +  ----- y  --- +  ----- y  --------- + y  ------ =  ----- F  ------ + F  ---
–8 1
 dt dt 3 dt 9 9 dt 10 3

5.
d- 4
 ---- d 2 K s1 M 2 + K s2 M 2 + K s2 M 1 K s1 K s2
y 2 +  ----- y 2  ------------------------------------------------------------- + y 2  ----------------- =
 dt  dt  M1 M2   M1 M2 

M1 
d- 2  --------------
 ---- K s1 + K s2
F - + F  ----------------------
-
 dt  M 1 M 2  M1 M2 
 input output equations - 6.24

6.
d- 4
 ---- d- 3  J----------------------------
 ---- 1 B 2 + J 2 B 1 d- 2  J----------------------------------------------------
 ---- 1 K s2 + J 2 K s2 + B 1 B 2
θ + θ - + θ - +
 dt 2  dt 2  J1 J2   dt 2  J1 J2 
d   B 1 K s2 + B 2 K s2
 ---- K s2
- θ 2 ------------------------------------ = τ ---------
-
 dt  J1 J2  J1 J2

d- 4
 ---- d 3 J1 B2 + J2 B1 d 2 J 1 K s2 + J 2 K s2 + B1 B 2
θ 1 +  ----- θ 1  ----------------------------- +  ----- θ 1  ----------------------------------------------------- +
 dt  dt  J1 J2   dt  J1 J2 
d-  B
 ---- 1 K s2 + B 2 K s2 d- 2  ---- B2 K s2
- =  ---- τ - +  ----- τ  ---------- + τ  ----------
1 d
θ -----------------------------------
 dt  1 J1 J2   dt  J 1  dt  J 1 J 2  J 1 J 2
 input output equations - 6.25

7.

θ ··
∑M = – R ( K ( θR ) ) + R ( K ( – y – θR ) ) = J M θ
2 2 ··
JM R Kθ + RKy + R Kθ = –J M θ
2 2
R ( K ( θR ) )  2R K + J M D 
R ( K ( – y – θR ) ) θ  ---------------------------------- = y
 – RK 
K ( – y – θR ) y ··
∑F = K ( – y – θR ) – F – Mg = My
2
K ( y + θR ) + F + Mg = –M yD
M
2
θ ( KR ) + F + Mg = y ( – MD – K )
2
F+Mg θ ( KR ) + y ( MD + K ) = – F – Mg
for the theta output equation;
2 2
 2R K + J M D  2
θ ( KR ) + θ  ---------------------------------- ( MD + K ) = – F – Mg
 – RK 
2 2 2 2 2
θ ( – K R ) + θ ( 2R K + J M D ) ( MD + K ) = FKR + MgKR
2 2 2 2 2 2 4 2
θ ( – K R + 2R MKD + 2R K + J M MD + J M KD ) = FKR + MgKR
d- 4  J-----------
 ---- M M d- 4
θ - +  ---- 2 2 2
θ ( 2R M + J M ) + θ ( K R ) = FKR + MgKR
 dt  KR   dt
d- 4 4  2R 3 K KR  K 3 R 3 2 2 2 2
 ----  d  K R MgK R
θ + ----- θ  ------------- + -------- + θ  ------------ = F ------------ + ---------------------
 dt  dt  J M M  JM M  JMM JMM
for the y output equation;
 – RK 
- ( KR ) + y ( MD 2 + K ) = – F – Mg
y  ---------------------------------
 2R2 K + J M D 2
4 2
 2 2 JM D M 2 JM D 2 
y  2R MD + ------------------- + K2R + ------------- – R K =
 K K  2 2
 2 JM D   2 JMD 
– F  2R + ------------- – Mg  2R + -------------
 K   K 
 ----
4 J M
d-  ----------- 2 J
y M - +  ----- y  2R M + -----
d 2 M
- + y ( R2 K ) =
 dt  K   dt  K
d- 2  –---------
 ---- J M 2 2
F + F ( – 2 R ) + ( – 2 MgR )
 dt  K 
d- 2  2KR 1-  R 2 K 2
2
d- 4
 ---- 
 dt y  dt y  J M + M y  J M M 
+ ---- ------------- ---- + ------------ =

d- 2  – 1
 ----  – 2 KR 2  – 2 gKR2
F -----
- + F  ----------------- +  --------------------
 dt  M   JM M   JM 
 input output equations - 6.26

8.
K1 K 2 K2 K2
x + x  ----------------------------- = F  ----------------------------- + g  -------------------
··
M ( K1 + K 2 ) M ( K 1 + K2 ) K1 + K 2
a) x ( t ) = – 0.2168 cos ( 5.774t ) + 0.2162

b) x ( t ) = – 0.2168 cos ( 5.774t ) + 0.02 cos ( t + 1 ) + 0.1962

c) x ( t ) = 0.02 cos ( t + 1 ) + 0.1962

6.8 ASSGINMENT PROBLEMS

1. Find the input-output equations for the differential equations below if both ’x’ and ’y’ are out-
puts.
·
y + y + 5x = 3
·
x+x+y = 7

2. For the system pictured below find the output response as a function of time for y1 and y2 using
a) integration to find an explicit function, b) numerical analysis using Scilab or C.

N
K s1 K s1 = 100 ----
m
M 1 = 1Kg
M1 y1
N
K s2 = 1000 ----
m
K s2 M 2 = 0.1Kg

M2 y2

F
 input output equations - 6.27

6.9 REFERENCES

Irwin, J.D., and Graf, E.R., Industrial Noise and Vibration Control, Prentice Hall Publishers,
1979.

Close, C.M. and Frederick, D.K., “Modeling and Analysis of Dynamic Systems, second edition,
John Wiley and Sons, Inc., 1995.
 circuits - 7.1

7. ELECTRICAL SYSTEMS

Topics:
• Basic components; resistors, power sources, capacitors, inductors and op-amps
• Device impedance
• Example circuits

Objectives:
• To apply analysis techniques to circuits

7.1 INTRODUCTION

A voltage is a pull or push acting on electrons. The voltage will produce a current
when the electrons can flow through a conductor. The more freely the electrons can flow,
the lower the resistance of a material. Most electrical components are used to control this
flow.

7.2 MODELING

Kirchoff’s voltage and current laws are shown in Figure 7.1. The node current law
holds true because the current flow in and out of a node must total zero. If the sum of cur-
rents was not zero then electrons would be appearing and disappearing at that node, thus
violating the law of conservation of matter. The loop voltage law states that the sum of all
rises and drops around a loop must total zero.

∑ Inode = 0 node current

∑ Vloop = 0 loop voltage

Figure 7.1 Kirchoff’s laws

The simplest form of circuit analysis is for DC circuits, typically only requiring
algebraic manipulation. In AC circuit analysis we consider the steady-state response to a
 circuits - 7.2

sinusoidal input. Finally the most complex is transient analysis, often requiring integra-
tion, or similar techniques.

• DC (Direct Current) - find the response for a constant input.

• AC (Alternating Current) - find the steady-state response to an AC input.
• Transient - find the initial response to changes.

There is a wide range of components used in circuits. The simplest components are
passive, such as resistors, capacitors and inductors. Active components are capable of
changing their behaviors, such as op-amps and transistors. A list of components that will
be discussed in this chapter are listed below.

• resistors - reduce current flow as described with ohm’s law

• voltage/current sources - deliver power to a circuit
• capacitors - pass current based on current flow, these block DC currents
• inductors - resist changes in current flow, these block high frequencies
• op-amps - very high gain amplifiers useful in many forms

7.2.1 Resistors

Resistance is a natural phenomenon found in all materials except superconductors.

A resistor will oppose current flow as described by ohm’s law in Figure 7.2. The resis-
tance value is assumed to be linear, but in actuality it varies with conductor temperature.

+ V
I = ---
V R
R
- V = IR

Figure 7.2 Ohm’s law

The voltage divider example in Figure 7.3 illustrates the methods for analysis of
circuits using resistors. In this circuit an input voltage is supplied on the left hand side.
The output voltage on the right hand side will be some fraction of the input voltage. If the
output resistance is very large, no current will flow, and the ratio of output to input volt-
ages is determined by the ratio of the resistance between R1 and R2. To prove this the cur-
 circuits - 7.3

rents into the center node are summed and set equal to zero. The equations are then
manipulated to produce the final relationship.

I1 + I2 + I3 = 0
+
Vi – Vo 0 – Vo
I1 ∴ ----------------- + I 2 +  --------------- = 0
R1 R1 R2
Assume the output resistance is large, so
I2 is negligible.
Vi
Vi – V o – Vo
∴ ----------------- +  --------- = 0
I2 +
R1 R2
I3 Vo
R2
∴V o  ------ + ------ = V i  ------
1 1 1
-
- R1 R2 R1
R 1 + R 2
∴V o  -----------------
- = V i  -----
1-
 R 1 R2   R 1

Vo R2
∴------ = ------------------
Vi R 1 + R2

Figure 7.3 A voltage divider circuit

If two resistors are in parallel or series they can be replaced with a single equiva-
lent resistance, as shown in Figure 7.4.
 circuits - 7.4

R1

series resistors
R eq = R 1 + R 2

R2

1- 1 1
------- = ------ + ------
parallel resistors R eq R1 R2
R1 R2
1
R eq = -------------------
1 1-
------ + -----
R1 R2

R1 R 2
R eq = ------------------
R1 + R2

Figure 7.4 Equivalent resistances for resistors in parallel and series

7.2.2 Voltage and Current Sources

A voltage source will maintain a voltage in a circuit, by varying the current as

required. A current source will supply a current to a circuit, by varying the voltage as
required. The schematic symbols for voltage and current sources are shown in Figure 7.5.
The supplies with ’+’ and ’-’ symbols provide DC voltages, with the symbols indicating
polarity. The symbol with two horizontal lines is a battery. The circle with a sine wave is
an AC voltage supply. The last symbol with an arrow inside the circle is a current supply.
The arrow indicates the direction of positive current flow.
 circuits - 7.5

+ +
V+
V V I
- V
- -

Figure 7.5 Voltage and current sources

A circuit containing a voltage source and resistors is shown in Figure 7.6. The cir-
cuit is analyzed using the node voltage method.

Find the output voltage Vo. R1

+
Vi + R2 R3 Vo
-
-

Examining the circuit there are two loops, but only one node, so the node current meth-
ods is the most suitable for calculations. The currents into the upper right node, Vo,
will be solved.

Vo – Vi Vo Vo
∑ ----------------
I =
R1
- + ------ + ------ = 0
R2 R3
Aside: when doing node-cur-
rent methods, select cur-
rents out of a node as
V o  ------ + ------ + ------ = V i  -------
1 1 1 1
 R 1 R 2 R 3  R1 positive, and in as nega-
tive. This will reduce the
R2 R 3 + R1 R 3 + R 1 R2
V o  -------------------------------------------------
- = V i  ------
1- chances of careless mis-
 R 1 R2 R 3   R1 takes.
R2 R 3
V o = Vi  --------------------------------------------------
R 2 R 3 + R1 R 3 + R 1 R2

Figure 7.6 A circuit calculation

 circuits - 7.6

Evaluate the circuit in Figure 7.6 using the loop voltage method.

R2 R3
V o = V i  --------------------------------------------------
R1 R2 + R1 R3 + R2 R3

Figure 7.7 Drill problem: Mesh solution of voltage divider

 circuits - 7.7

Evaluate the voltage divider in Figure 7.3 using the loop current method. Hint: Put a volt-
age supply on the left, and an output resistor on the right. Remember that the output
resistance should be infinite.

Figure 7.8 Drill problem: Mesh solution of voltage divider

Dependant (variable) current and voltage sources are shown in Figure 7.9. The
voltage and current values of these supplies are determined by their relationship to some
other circuit voltage or current. The dependant voltage source will be accompanied by a
’+’ and ’-’ symbol, while the current source has an arrow inside.

V = f( )
+ I = f( )

Figure 7.9 Dependant voltage sources

 circuits - 7.8

2ohm + + 3ohm
Ii V1 Io
- - V =f(V )
2 1

Find the output current, Io, if Ii = 1A. f(V1) = 3V1

What if the input current is Ii = 1sin(2t)A?

Figure 7.10 Drill problem: Find the currents in the circuit above

7.2.3 Capacitors

Capacitors are composed of two isolated metal plates very close together. When a
voltage is applied across the capacitor, electrons will be forced into one plate, and forced
out of the other plate. Temporarily this creates a small current flow until the plates reach
equilibrium. So, any voltage change will result in some current flow. In practical terms
this means that the capacitor will block any DC voltages, except for transient effects. But,
high frequency AC currents will pass through the device. The equation for a capacitor and
schematic symbols are given in Figure 7.11.
 circuits - 7.9

+
+
I = C  ----- V = CDV
C d
C V I dt

Figure 7.11 Capacitors

The symbol on the left is for an electrolytic capacitor. These contain a special fluid
that increases the effective capacitance of the device but requires that the positive and neg-
ative sides must be observed in the circuit. (Warning: reversing the polarity on an electro-
lytic capacitor can make them leak, fail and possibly explode.) The other capacitor symbol
is for a regular capacitor, normally with values under a microfarad.

Find the current as a function of time.

+ C=1uF
-
V= I
5cos(10t)V

Figure 7.12 Drill problem: Current through a capacitor

 circuits - 7.10

7.2.4 Inductors

While a capacitor will block a DC current, an inductor will pass it. Inductors are
basically coils of wire. When a current flows through the coils, a magnetic field is gener-
ated. If the current through the inductor changes then the magnetic field must change, oth-
erwise the field is maintained without effort (i.e., no voltage). Therefore the inductor
resists changes in the current. The schematic symbol and relationship for an inductor are
shown in Figure 7.13.

+ I
V = L  ----- I = LDI
d
L
V dt

Figure 7.13 An inductor

An inductor is normally constructed by wrapping wire in loops about a core. The

core can be hollow, or be made of ferrite to increase the inductance. Inductors usually cost
more than capacitors. In addition, inductors are susceptible to interference when metals or
other objects disturb their magnetic fields. When possible, designers normally try to avoid
using inductors in circuits.
 circuits - 7.11

Find the current as a function of time.

+ L=1mH
-
V= I
5cos(10t)V

Figure 7.14 Drill problem: Current through an inductor

7.2.5 Op-Amps

The ideal model of an op-amp is shown in Figure 7.15. On the left hand side are
the inverting and non-inverting inputs. Both of these inputs are assumed to have infinite
impedance, and so no current will flow. Op-amp application circuits are designed so that
the inverting and non-inverting inputs are driven to the same voltage level. The output of
the op-amp is shown on the right. In circuits op-amps are used with feedback to perform
standard operations such as those listed below.

• adders, subtractors, multipliers, and dividers - simple analog math operations

• amplifiers - increase the amplitude of a signal
• impedance isolators - hide the resistance of a circuit while passing a voltage
 circuits - 7.12

V-
I- Note: for analysis use,
- Vo
I- = I+ = 0
I+ +
V+ V- = V+

Figure 7.15 An ideal op-amp

A simple op-amp example is given in Figure 7.16. As expected both of the op-amp
input voltages are the same. This is a function of the circuit design. (Note: most op-amp
circuits are designed to force both inputs to have the same voltage, so it is normally rea-
sonable to assume they are the same.) The non-inverting input is connected directly to
ground, so it will force both of the inputs to 0V. When the currents are summed at the
inverting input, an equation including the input and output voltages is obtained. The final
equation shows the system is a simple multiplier, or amplifier. The gain of the amplifier is
determined by the ratio of the input and feedback resistors.
 circuits - 7.13

R2
R1
-
+
Vi + +
- Vo
-

The voltage at the non-inverting input will be 0V, by design the voltage at the invert-
ing input will be the same.

V + = 0V

V - = V + = 0V

The currents at the inverting input can be summed.

V- – Vi V- – Vo
∑ I V- = ---------------- + ----------------- = 0
R1 R2
0 – Vi 0 – Vo
-------------- + --------------- = 0
R1 R2
–R 2 Vi
V o = --------------
R1
–R2
V o =  --------- Vi
 R1 

Figure 7.16 A simple inverting operational amplifier configuration

An op-amp circuit that can subtract signals is shown in Figure 7.17.

 circuits - 7.14

R2
Find the input/output ratio,

R1

+
-
+
Vi R5 +
- + R3 Vo
Vref -
R4
-

Figure 7.17 Op-amp example

For ideal op-amp problems the node voltage method is normally the best choice.
The equations for the circuit in Figure 7.17 and derived in Figure 7.18. The general
approach to this solution is to sum the currents into the inverting and non-inverting input
nodes. Notice that the current into the op-amp is assumed to be zero. Both the inverting
and non-inverting input voltages are then set to be equal. After that, algebraic manipula-
tion results in a final expression for the op-amp. Notice that if all of the resistor values are
the same then the circuit becomes a simple subtractor.
 circuits - 7.15

Note: normally node voltage methods work best with op-amp circuits, although others
can be used if the non-ideal op-amp model is used.

First sum the currents at the inverting and non-inverting op-amp terminals.
V+ – Vi V + – V o
∑ I V+ = ----------------- + ------------------ = 0
R1 R2

V +  ------ + ------ = V i  ------ + V o  ------

1 1 1 1
 R 1 R 2  R 1  R 2

R1 + R2
V +  ------------------ = V i  ------ + V o  ------
1 1
R1 R2 R1 R2
R2 R1
V + = V i  ------------------ + V o  ------------------ (1)
R1 + R2 R 1 + R2

V - – V ref V-
∑ V- -------------------
I =
R5
- + ------ = 0
R4

V -  ------ + ------ = V ref  ------

1 1 1
R4 R5 R5
R4
V - = V ref  ------------------ (2)
R4 + R5

Now the equations can be combined.

V- = V+

R4 R2 R1
V ref  ------------------ = V i  ------------------ + V o  ------------------
 R 4 + R 5  R 1 + R 2  R1 + R 2

R1 R2 R4
V o  ------------------ = Vi  ------------------ – V ref  ------------------
 R 1 + R 2  R 1 + R 2  R 4 + R 5

R R 4 ( R 1 + R 2 )
V o = Vi  -----2- – V ref  -----------------------------
- (3)
 R 1  R 1 ( R 4 + R 5 )

Figure 7.18 Op-amp example (continued)

An op-amp (operational amplifier) has an extremely high gain, typically 100,000

times. The gain is multiplied by the difference between the inverting and non-inverting
terminals to form an output. A typical op-amp will work for signals from DC up to about
 circuits - 7.16

100KHz. When the op-amp is being used for high frequencies or large gains, the model of
the op-amp in Figure 7.19 should be used. This model includes a large resistance between
the inverting and non-inverting inputs. The voltage difference drives a dependent voltage
source with a large gain. The output resistance will limit the maximum current that the
device can produce, normally less than 100mA.

V–
typically,
rn
+ Vo r n > 1M Ω
- 5
ro A > 10
V A ( V+ – V- ) r o < 100 Ω
+

Figure 7.19 A non-ideal op-amp model

7.3 IMPEDANCE

Circuit components can be represented in impedance form as shown in Figure

7.20. When represented this way the circuit solutions can focus on impedances, ’Z’,
instead of resistances, ’R’. Notice that the primary difference is that the differential opera-
tor has been replaced. In this form we can use impedances as if they are resistances.

Device Time domain Impedance

Note: Impedance is like
resistance, except that
Resistor V ( t ) = RI ( t ) Z = R it includes time variant
features also.
1 1
Capacitor V ( t ) = ---- ∫ I ( t ) dt Z = -------- V = ZI
C DC
d
Inductor V ( t ) = L ----- I ( t ) Z = LD
dt

Figure 7.20 Impedances for electrical components

 circuits - 7.17

When representing component values with impedances the circuit solution is done
as if all circuit components are resistors. An example of this is shown in Figure 7.21.
Notice that the two impedances at the right (resistor and capacitor) are equivalent to two
resistors in parallel, and the overall circuit is a voltage divider. The impedances are written
beside the circuit elements.

DL
t=0sec

50VDC +
+ R Vo
- 1/DC -

Find the equivalent for the capacitor and resistor in parallel.

DL

50VDC 1 1 R
+ ------------------------ = ------------------ = ----------------------
1 - --1- 1 RCD + 1
- ------------- + DC + ---
 -------
1 - R R
 DC

Treat the circuit as a voltage divider,

 ---------------------
R -
 1 + DCR
Vo = 50V ------------------------------------------ = 50V  ------------------------------------------
-
R
2 
DL +  ----------------------
R D RLC + DL + R
1 + DCR

Figure 7.21 A impedance example for a circuit

 circuits - 7.18

7.4 EXAMPLE SYSTEMS

The list of instructions below can be useful when approaching a circuits problem.
The most important concept to remember is that a minute of thinking about the solution
approach will save ten minutes of backtracking and fixing mistakes.

1. Look at the circuit to determine if it is a standard circuit type such as a voltage

divider, current divider or an op-amp inverting amplifier. If so, use the standard
solution to solve the problem.
2. Otherwise, consider the nodes and loops in the circuit. If the circuit contains
fewer loops, select the current loop method. If the circuit contains fewer nodes,
select the node voltage method. Before continuing, verify that the select method
can be used for the circuit.
3. For the node voltage method define node voltages and current directions. For
the current loop method define current loops and indicate voltage rises or drops
by adding ’+’ or ’-’ signs.
4. Write the equations for the loops or nodes.
5. Identify the desired value and eliminate unwanted values using algebra tech-
niques.
6. Use numerical values to find a final answer.

Note: The units for various coefficient units

electrical quantities are
As
listed to the right. They may C ------
V
be used to check equations
by doing a unit balance. Vs
L ------
A
V
R ---
A

The circuit in Figure 7.22 could be solved with two loops, or two nodes. An arbi-
trary decision is made to use the current loop method. The voltages around each loop are
summed to provide equations for each loop.
 circuits - 7.19

R1
C

+
V L
I2
- I1
+
R2 Vo
-

Note: when summing voltages in a loop remember to deal with sources

that increase the voltage by flipping the sign.

First, sum the voltages around the loops and then eliminate I1.

∑ VL1 = – V + R 1 I 1 + L ( DI 1 – DI 2 ) = 0
( R 1 + LD )I 1 = V + ( LD )I 2 (1)
I 1 = -------------------- +  -------------------- I 2
V LD
R 1 + LD R 1 + LD (2)
I2
∑ V L2 = L ( DI 2 – DI 1 ) + -------
CD
- + R2 I2 = 0
I2 (3)
L ( DI 2 – DI 1 ) + -------- + R 2 I 2 = 0
CD

( DL )I 1 =  LD + -------- + R2 I 2
1
 CD 
 1 R2 
I 1 =  1 + --------------2 + -------- I 2 (4)
 CLD LD

Figure 7.22 Example problem

The equations in Figure 7.22 are manipulated further in Figure 7.23 to develop an
input-output equation for the second current loop. This current can be used to find the cur-
rent through the output resistor R2. The output voltage can then be found by multiplying
the R2 and I2.
 circuits - 7.20

First, sum the voltages around the loops and then eliminate I1.
 R2 
I 1 = -------------------- +  -------------------- I 2 =  1 + --------------2 + -------- I 2
V LD 1
R 1 + LD  R 1 + LD  CLD LD
V -  1 R2 LD 
------------------- =  1 + --------------2 + -------- – -------------------- I 2
R 1 + LD  CLD LD R 1 + LD
2 2 3
V -  ( CLD + 1 + CDR2 ) ( R 1 + LD ) – CL D 
------------------- =  --------------------------------------------------------------------------------------------------- I2
R 1 + LD  CLD
2


 CLD
2 
I 2 =  ---------------------------------------------------------------------------------------------------------------------------------
- V
 ( R 1 + LD ) ( ( CLD 2 + 1 + CDR 2 ) ( R 1 + LD ) – CL 2 D 3 )

 CLD
2 
I 2 =  ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
- V
 CL ( R1 + R 2 )D 3 + L ( CR 21 + 2CR 1 R 2 + L )D 2 + R 1 ( CR1 R 2 + 2L )D + ( R 12 )
Convert it to a differential equation.

2
CL ( R 1 + R 2 )I 2''' + L ( CR1 + 2CR1 R 2 + L )I 2'' + R 1 ( CR 1 R 2 + 2L )I 2' + ( R 12 )I 2 = CLV''

Figure 7.23 Example problem (continued)

The equations can also be manipulated into state equations, as shown in Figure
7.24. In this case a dummy variable is required to replace the two first derivatives in the
first equation. The dummy variable is used in place of I1, which now becomes an output
variable. In the remaining state equations I1 is replaced by q1. In the final matrix form the
state equations are in one matrix, and the output variable must be calculated separately.
 circuits - 7.21

State equations can also be developed using equations (1) and (3).
· ·
(1) becomes R 1 I 1 + LI 1 = V + LI 2
· ·
LI 1 – LI 2 = V – R 1 I 1

· · V R1
I 1 – I 2 = --- – ------ I 1
L L
q1 = I1 – I2 (10)

· V R1
q 1 = --- – ------ I 1
L L
I1 = q1 + I2 (11)

· V R1
q 1 = --- – ------ ( q 1 + I 2 )
L L
R1 R1
q 1 = q 1  – ------ + I 2  – ------ + ---
· V (12)
L L L
·· ·· I ·
(3) becomes LI 2 – LI 1 + ---2- + R 2 I 2 = 0
C
I2 ·
V – R1 I 1 + ---- + R 2 I 2 = 0
C
I2 ·
V – R 1 ( q1 + I 2 ) + ---- = – R 2 I 2
C

I 2 = I 2  – R 1 + ---- + q 1 ( – R 1 ) + V
· 1 (13)
C
These can be put in matrix form,

R1 R1
d- q 1 – ------ – ------ q1 V
---
---- = L L + L q1
dt I I2 I1 = 1 1
2 1 V I2
– R 1 – R1 + ----
C

Figure 7.24 Example problem (continued)

 circuits - 7.22

Evaluate the circuit in Figure 7.22 using the node voltage method.

Figure 7.25 Drill problem: Use the node voltage method

 circuits - 7.23

Find the equation relating the output and input voltages,

Rf C

Ri
Vin
- Vout
+

Figure 7.26 Drill problem: Find the state equation

 circuits - 7.24

The circuit in Figure 7.27 can be evaluated as a voltage divider when the capacitor
is represented as an impedance. In this case the result is a first-order differential equation.

t=0
C
V s = 3 cos t +
+ R Vo
- -

As normal we relate the source voltage to the output voltage. The we find the values
for the various terms in the frequency domain.
ZR  1
V o = Vs  ------------------
- where, ZR = R Z C = --------
 Z R + Z C DC
Next, we may combine the equations, and convert it to a differential equation.

 
V o = Vs  ------------------
 R
1
 R + --------
DC

V o = Vs  ----------------------
CRD
 CRD + 1
V o ( CRD + 1 ) = V s ( CRD )
· ·
V o ( CR ) + V o = Vs ( CR )

V o + Vo  -------- = V s
· 1 ·
CR

Figure 7.27 Circuit solution using impedances

The first-order differential equation in Figure 7.27 is continued in Figure 7.28

where the equation is integrated. The solution is left in variable form, except for the sup-
ply voltage.
 circuits - 7.25

First, write the homogeneous solution using the known relationship.

t
– --------
V o + Vo  -------- = 0
· 1 yields CR
 CR V h = C 1 e
Next, the particular solution can be determined, starting with a guess.

V o + Vo  -------- =  ----- ( 3 cos t ) = – 3 sin t

· 1 d
CR dt
V p = A sin t + B cos t
V p' = A cos t – B sin t

( A cos t – B sin t ) + ( A sin t + B cos t )  -------- = – 3 sin t

1
CR
A + B  -------- = 0
1
 CR
A = B  --------
–1
 CR

– B + A  -------- = – 3
1
CR
– B + B  --------  -------- = – 3
–1 1
 CR  CR
B  ------------ 
1
 2 2 + 1 = 3
C R
2 2
3C R -  3C 2 R 2   – 1  – 3 CR -
B = ---------------------
2 2 A =  ---------------------
- -------- = ---------------------
2 2  CR
1+C R 1 + C R  1+C R
2 2

A + B sin  t + atan  --- 

2 2 B
Vp =
  A 
The homogeneous and particular solutions can now be combined. The system will be
assumed to be at rest initially.
t
– --------
+ A + B sin  t + atan  --- 
2 2
CR B
V 0 = Vh + V p = C1 e
  A 
0 = C 1 e + A + B sin  0 + atan  --- 
0 2 2 B
A
C 1 = – A + B sin  0 + atan  --- 
2 2 B
  A 

Figure 7.28 Circuit solution using impedances (continued)

 circuits - 7.26

7.5 ELECTROMECHANICAL SYSTEMS - MOTORS

7.5.1 Permanent Magnet DC Motors

DC motors apply a torque between the rotor and stator that is related to the applied
voltage/current. When a voltage is applied the torque will cause the rotor to accelerate. For
any voltage and load on the motor there will tend to be a final angular velocity due to fric-
tion and drag in the motor. And, for a given voltage the ratio between steady-state torque
and speed will be a straight line, as shown in Figure 7.29.

T voltage/current increases

Figure 7.29 Torque speed curve for a permanent magnet DC motor

The basic equivalent circuit model is shown in Figure 7.30, includes the rotational
inertia of the rotor and any attached loads. On the left hand side is the resistance of the
motor and the ’back emf’ dependent voltage source. On the right hand side the inertia
components are shown. The rotational inertia J1 is the motor rotor, and the second inertia
is an attached disk.
 circuits - 7.27

I R
J1 J2
+ T, ω
Voltage
Supply Vm +
Vs -
-

Because a motor is basically wires in a magnetic field, the electron flow (current) in
the wire will push against the magnetic field. And, the torque (force) generated
will be proportional to the current.
Tm
T m = KI ∴I = ------
K
Next, consider the power in the motor,
P = V m I = Tω = KIω ∴V m = Kω
Consider the dynamics of the rotating masses by summing moments.

= T m – T load = J  ----- ω ∴T m = J  ----- ω + T load

d d
∑M dt dt

Figure 7.30 The torque and inertia in a basic motor model

These basic equations can be manipulated into the first-order differential equation
in Figure 7.31.

The current-voltage relationship for the left hand side of the equation can be writ-
ten and manipulated to relate voltage and angular velocity.
Vs – V m
I = ------------------
R
Tm V s – Kω
∴------ = --------------------
K R
J  ----- ω + T load
d
 dt V s – Kω
∴------------------------------------ = -------------------
-
K R
2 T load
∴ ----- ω + ω  ------ = Vs  ------ – ------------
d K K
dt JR JR J

Figure 7.31 The first-order model of a motor

 circuits - 7.28

7.5.2 Induction Motors

AC induction motors are extremely common because of the low cost of construc-
tion, and compatibility with the power distribution system. The motors are constructed
with windings in the stator (outside of the motor). The rotor normally has windings, or a
squirrel cage. The motor does not have bushes to the rotor. The motor speed is close to, but
always less than the rotating AC fields. The rotating fields generate currents, and hence
opposing magnetic fields in the stator. The maximum motor speed is a function of the fre-
quency of the AC power, and the number of pole of the machine. For example, an induc-
tion motor with three poles being used with a 60Hz AC supply would have a maximum
speed of 2*(60Hz/3) = 40Hz = 2400RPM.

The equivalent circuit for an AC motor is given in Figure 7.32. The slip of the
motor determines the load current, IL. It is a function of the fraction, f, of full speed.

Ls Rs Lr Rr

IL RL
V Lm

1–f
R L = ---------- R r
f

Figure 7.32 Basic model of an induction motor

The torque relationship for AC motors is given in Figure 7.33. These can be com-
bined with the equivalent circuit model to determine the response of the motor to a load.
 circuits - 7.29

First the torques on the motor are summed,

d
∑M = T rotor – T load = J ----- ω
dt

CONTINUE TO STATE VARIABLE MODEL....

7.5.3 Brushless Servo Motors

Brushless servo motors are becoming very popular because of their low mainte-
nance requirements. The motors eliminate the need for brushes by using permanent mag-
nets on the rotor, with windings on the stator, as shown in Figure 7.33. The windings on
the stator are switched at a given frequency to produce a desired rotational speed, or held
static to provide a holding torque.

Va

windings on stator

Vb
N S
permanent
magnet rotor
Vc

Figure 7.33 The construction of a brushless servo motor

 circuits - 7.30

The basic relationships for brushless DC motors are given in Figure 7.34.

V t =  R m + ----- L I m + E
d
 dt 
E = Ke ω

T = Kt Im

where,
V t = terminal voltage across motor windings
R m = resistance of a motor winding
L = phase to phase inductance
I m = current in winding
E = back e.m.f. of motor
K e = motor speed constant
ω = motor speed
K t = motor torque constant
T = motor torque

Figure 7.34 Basic relationships for a brushless motor

 circuits - 7.31

V t =  R m + ----- L ----- + K e ω

d T
dt K t
d-
∑ M = T – Tload = J ---- dt
ω
d
T = J ----- ω + T load
dt
where,
J = combined moments of inertia for the rotor and external loads
T load = the applied torque in the system
d
J ----- ω + T load
V t =  R m + ----- L ------------------------------- + K e ω
d dt
dt Kt
JR m d LJ d 2 Rm
V t = ---------- ----- ω + ------  ----- ω + ------- T load + ----- ----- T load + K e ω
L d
K t dt K t dt   Kt K t dt
d 2
( LJ )  ----- ω + ( JR m ) ----- ω + K e K t ω = K t V t – L ----- T load – R m T load
d d
dt dt dt

d- 2 Rm d Ke Kt Kt V Rm T load
 ---- - ω = ----------t – 1--- ----
- ----- ω + -----------
ω + ------
d-
T load – -------------------
 dt L dt LJ LJ J dt LJ

Figure 7.35 An advanced model of a brushless servo motor

To rotate the motor at a constant velocity the waveform in Figure 7.36 would be
applied to each phase. Although each phase would be 120 degrees apart for a three pole
motor. A more sophisticated motor controller design would smooth the waves more to
approach a sinusoidal shape.

Vt

t(ms)
 circuits - 7.32

Figure 7.36 Typical supply voltages

7.6 FILTERS

Filters are useful when processing data signals. Low pass often used to eliminate
noise, high pass filters eliminate static signals and leave dynamic signals. Band pass filters
reject all frequencies outside a desired frequency band. A low pass filter is shown in Fig-
ure 7.37. At high frequencies the capacitor, C, has a very low impedance, and grounds the
input signal. At low frequencies the capacitor impedance is high, increasing the gain of the
op-amp circuit. This is easier to conceptualize if the R1-C pair are viewed as a voltage
divider.

Rf

Ri

-
+
R1 +
+ Vo
Vi -
C
-

Figure 7.37 Low-Pass Filter

A high pass filter is shown in Figure 7.38. In this case the voltage divider in the
previous circuit is reversed. In this circuit the gain will increase for signals with higher fre-
quencies.
 circuits - 7.33

Rf

Ri

-
+
C +
+ Vo
Vi -
R
-

Figure 7.38 High-Pass Filter

7.7 OTHER TOPICS

The relationships in Figure 7.39 can be used to calculate the power and energy in a
system. Notice that the power calculations focus on resistance, as resistances will dissipate
power in the form of heat. Other devices, such as inductors and capacitors, store energy,
but don’t dissipate it.

2
2V
P = IV = I R = ------ E = Pt
R

Figure 7.39 Electrical power and energy

7.8 SUMMARY

• Basic circuit components are resistors, capacitors, inductors op-amps.

• node and loop methods can be used to analyze circuits.
 circuits - 7.34

• Capacitor and inductor impedances can be used as resistors in calculations.

7.9 PRACTICE PROBLEMS

1. Derive the equations for combined values for resistors, capacitors and inductors in series and
parallel.

2. Find the output voltage as a function of input voltage.

R R
+
Vi
+
Vo
- R R
-

3. Write the differential equation for the following circuit.

L R

+
Vi
+ C Vo
-
-

4. Consider the following circuit.

L +
Vi + C
Vo
- R
-

a) Develop a differential equation for the circuit.

b) Put the equation in state variable matrix form.
 circuits - 7.35

5. Develop differential equations and the input-output equation for the electrical system below.

R1
+
Vi C
+ L R2 Vo
-
-

6. Consider the following circuit. Develop a differential equation for the circuit.

L +
C
Vi + R1
R2 R3 Vo
-
-

7. Find the input-output equation for the circuit below, and then find the natural frequency and
damping coefficient.

R1
+
Vi R2 Vo
+ L C
-
-

8. a) Find the differential equation for the circuit below where the input is Vi, and the output is Vo.

L
+
Vi
+ C R Vo
- -

b) Convert the equation to an input-output equation.

 circuits - 7.36

c) Solve the differential equation found in part b) using the numerical values given
below. Assume at time t=0, the circuit has the voltage Vo and the first derivative
shown below.
L = 10mH C = 1µF R = 1KΩ V i = 10V
V
at t=0s V o = 2V V o ′ = 3 ---
s

9.
a) Write the differential equations for the system pictured below.
b) Put the equations in input-output form.
R2
R1

Vi
- Vo

C +

10. Given the circuit below, find the ratio of the output over the input (this is also known as a
transfer function). Simplify the results.

R2

Vi R1
C
- Vo

+
 circuits - 7.37

11. Examine the following circuit and then derive the differential equation.
L R2

R1
+ -
+
+
Vi Vo
- -

12. Examine the following circuit and then derive the differential equation.

R2

R1
+ -
+
+
Vi Vo
- -

13.
a) Find the differential equation for the circuit below.
1000Ω

1µF

0.001H 1000Ω
-
+
+
Vi +
- 10Ω Vo
-
 circuits - 7.38

b) Put the differential equation in state variable form and a numerical method to
produce a detailed sketch of the output voltage Vo. Assume the system starts at
rest, and the input is Vi=5V.

14.
a) Write the differential equations for the system pictured below.
b) Put the equations in state variable form.
c) Use numerical methods to find the ratio between input and output voltages for a
range of frequencies. The general method is put in a voltage such as
Vi=1sin(___t), and see what the magnitude of the output is. Divide the magni-
tude of the output sine wave by the input magnitude. Note: This should act as a
high pass or low pass filter.
d) Plot a graph of gain against the frequency of the input.
R2
R1

Vi
- Vo

C +

C=1uF
R1=1K
R2=1K

7.10 PRACTICE PROBLEM SOLUTIONS

1.

R1 R 2
R parallel = ------------------ R series = R 1 + R 2
R1 + R2

C 1 C2
C series = ------------------- C parallel = C 1 + C 2
C1 + C 2

L1 L2
L parallel = ------------------ L series = L 1 + L 2
L1 + L2
 circuits - 7.39

2.
Vo
------ = 1---
Vi 5

3.

V o + V o  --- + V o  ------- = V i  -------

·· · R 1 1
L LC LC

4.

V o + Vo  -------- + V o  ------- = V i  -------

·· · 1 1 1
a)
RC LC LC

b) ·
V o = Xo

X o = Xo  -------- + V o  ------- + V i  -------

· –1 –1 1
RC LC LC

5.

V o + V o  ---------- + ---------- + V o  ------- = V i  ----------

·· · 1 1 1 · 1
 CR 1 CR 2  LC  CR 1

6.
· R2 R3 C + L R2 R2 R 3
V o + Vo  ------------------------------- + V o  ------------------------------- = V i  ---------------------------
·· ·
LC ( R 2 + R 3 ) LC ( R2 + R 3 ) L ( R2 + R3 )

7.

V o + Vo  ---------- + ---------- + V o  ------- = V i  ----------

·· · 1 1 1 · 1
a) CR 1 CR2 LC CR 1

1- L ( R1 + R2 )
b) ωn = ------ ζ = -------------------------------
LC 2 CR 1 R 2
 circuits - 7.40

8.
(a.
Sum currents at node Vo
( Vo – Vi ) ( Vo )
∑ IVo = ---------------------
DL
- + ( V o )DC + ----------
R
- = 0

2 V o DL
V o – V i + V o D LC + -------------- = 0
R
V o'' + Vo'  -------- + V o  ------- = V i  -------
1 1 1
CR LC LC

(b.
V o + V o  -------- + V o  ------- = V i  -------
·· · 1 1 1
 CR  LC  LC
 circuits - 7.41

(c.
·  10 
–2
V o ( LC ) + V o  --- + V o = V i
·· · L ·· – 2 –6
 R V o ( 10 10 ) + V o  ---------- + V o = 10
 10 3 
·· –8 · –5
V o ( 10 ) + V o ( 10 ) + V o = 10
·· · 3 8 9
V o + V o ( 10 ) + V o ( 10 ) = ( 10 )
homogeneous;
2 At At 3 At 8
A e + Ae ( 10 ) + e ( 10 ) = 0
3 3 2 8
2 3
A + A ( 10 ) + ( 10 ) = 0
8 – 10 ± ( 10 ) – 4 ( 10 )
A = -------------------------------------------------------------- = – 500 ± 9987j
2
– 500t
Vh = C1 e cos ( 9987t + C2 )
particular; guess
3 8 9
Vp = A ( 0 ) + ( 0 ) ( 10 ) + ( A ) ( 10 ) = ( 10 ) A = 10
V p = 10
for initial conditions,
– 500t
Vo = C1 e cos ( 9987t + C 2 ) + 10
– 500 ( 0 )
for t=0, Vo=2V 2 = C1 e cos ( 9987 ( 0 ) + C 2 ) + 10
– 8 = C 1 cos ( C 2 ) (1)
· – 500t – 500t
V o = –500C 1 e cos ( 9987t + C 2 ) – 9987C 1 e sin ( 9987t + C 2 )
for t=0, d/dt Vo=3V 3 = – 500C 1 cos ( C 2 ) – 9987C 1 sin ( C 2 )

3 = – 500C 1 cos ( C 2 ) – 9987C 1 sin ( C 2 )

3 = – 4000 – 9987  -------------------- sin ( C 2 )

–8
 cos ( C 2 )
– 3997 sin ( C 2 )
- = tan ( C 2 )
-------------------- = ------------------- C 2 = – 0.050
8 ( 9987 ) cos ( C 2 )
–8
C 1 = -------------------- = – 8.01
cos ( C 2 )
– 500t
V o = – 8.01 e cos ( 9987t – 0.050 ) + 10
 circuits - 7.42

9.
R1
V i + V i ( R 1 C ) + Vo  ------ = 0
·
a)
 R 2

–R 2
Vo = V i ( – CR 2 ) + Vi  ---------
·
b)
 R1 

10.
Vo – R2
------ = ---------------------------------
-
Vi R 1 + DR 1 R 2 C

11.
–R
V o = Vi  ------ + V i  --------2-
· –L
 R 1  R1 

12.
R2 · –R2
V o + Vo  ------ = V i  ---------
·
 L  R1 

13.
(a.
Create a node between the inductor and resistor Va, and use the node voltage method
( VA – Vi ) ( VA – V– ) V– = V = 0V
∑ IV A
= ---------------------- + -------------------------- = 0
0.001D 1000
+

1000000 ( V A – V i ) + V A D = 0
XXXXADD UNITSXXXXX
V A = V i  -------------------------------
1000000
1000000 + D
( V– – VA ) ( V– – Vo )
∑ – -------------------------
I V =
1000
- + ------------------------- + ( V – – V o ) ( 0.000001D ) = 0
1000
– 1 )-  1000000  (-------------
(----------- –Vo )
V i ------------------------------- + - + ( – V o ) ( 0.000001D ) = 0
1000  1000000 + D 1000

V o ( – 1 – 0.001D ) = V i  -------------------------------
1000000
 1000000 + D
2
V o ( – 1000000 – D – 1000D – 0.001D ) = 1000000V i
–9 –3
V o'' ( – 10 ) + V o' ( – 1.001 ( 10 ) ) + V o ( – 1 ) = V i
 circuits - 7.43

(b. d- · d- · ·
---- Vo = Vo ---- V o = – 1000000000V i – 1001000Vo – 1000000000V o
dt dt

14.
–R2
V o = Vi ( –CR 2 ) + V i  ---------
a) ·
 R1 
b) Not a state equation
c) assume Vi=1sin(wt)

1 + ( 1000ω ) sin  ωt + atan  ---------------- 

2 2 1
Vo =
1000ω
d)

7.11 ASSIGNMENT PROBLEMS

1. Write the differential equation for the following schematic.

R R

+ +
L Vo
Vi C
-
-
 circuits - 7.44

2. Write the differential equation for the following schematic.

L
R1
+ -
+
+
Vi Vo
- -

3. Write the differential equation for the following schematic.

L
R
+ -
+
C +
Vi Vo
- -

4. Develop the differential equation(s) for the system below, and use them to find the response to
the following inputs. Assume that the circuit is off initially.
R3

Vi - Vo
+ C
R1=R2=R3=R4=1Kohm - R4
C=1uF
a) V i = 5 sin ( 100t )
b) V i = 5 sin ( 1000000t ) R1
c) Vi = 5 R2
 circuits - 7.45

5. Write the input-output equation for the circuit below.

+
C1 C2

Vi
+ R1 R2
L Vo
-

6. Study the circuit below. Assume that for t<0s the circuit is discharged and off. Starting at t=0s
an input of Vi=5sin(100,000t) is applied.

1K
+
Vi -
+
1uF
Vo
-

a) Write a differential equation and then a transfer function describing the circuit.
b) Find the output of the circuit using explicit integration (i.e., homogeneous and
particular solutions).
 feedback control - 8.1

8. FEEDBACK CONTROL SYSTEMS

Topics:
• Transfer functions, block diagrams and simplification
• Feedback controllers
• Control system design

Objectives:
• To be able to represent a control system with block diagrams.
• To be able to select controller parameters to meet design objectives.

8.1 INTRODUCTION

Every engineered component has some function. A function can be described as a

transformation of inputs to outputs. For example it could be an amplifier that accepts a sig-
nal from a sensor and amplifies it. Or, consider a mechanical gear box with an input and
output shaft. A manual transmission has an input shaft from the motor and from the
shifter. When analyzing systems we will often use transfer functions that describe a sys-
tem as a ratio of output to input.

8.2 TRANSFER FUNCTIONS

Transfer functions are used for equations with one input and one output variable.
An example of a transfer function is shown below in Figure 8.1. The general form calls for
output over input on the left hand side. The right hand side is comprised of constants and
the ’D’ operator. In the example ’x’ is the output, while ’F’ is the input.

The general form An example

output x 4+D
----------------- = f ( D ) --- = ---------------------------------
input F 2
D + 4D + 16

Figure 8.1 A transfer function example

 feedback control - 8.2

If both sides of the example were inverted then the output would become ’F’, and
the input ’x’. This ability to invert a transfer function is called reversibility. In reality
many systems are not reversible.

There is a direct relationship between transfer functions and differential equations.

This is shown for the second-order differential equation in Figure 8.2. The homogeneous
equation (the left hand side) ends up as the denominator of the transfer function. The non-
homogeneous solution ends up as the numerator of the expression.

·· · 2 f
x + 2ζωn x + ω n x = -----
M
2 2 f
xD + 2ζω n xD + ω n x = -----
M
2 2 f
x ( D + 2ζω n D + ω n ) = -----
M
 ----
1-
x  M particular
-- = --------------------------------------------
f 2 2
D + 2ζω n D + ω n homogeneous

ωn Natural frequency of system - Approximate frequency of

control system oscillations.

ξ Damping factor of system - If < 1 then underdamped, and the system

will oscillate. If =1 critically damped. If < 1 overdamped, and never
any oscillation (more like a first-order system). As damping factor
approaches 0, the first peak becomes infinite in height.

Figure 8.2 The relationship between transfer functions and differential equations for a
mass-spring-damper example

The transfer function for a first-order differential equation is shown in Figure 8.3.
As before the homogeneous and non-homogeneous parts of the equation becomes the
denominator and the numerator of the transfer function.
 feedback control - 8.3

· 1
x + --- x = f
τ
1
xD + --- x = f
τ

x  D + --- = f
1
τ
x 1
-- = -------------
f 1
D + ---
τ

Figure 8.3 A first-order system response

8.3 CONTROL SYSTEMS

Figure 8.4 shows a transfer function block for a car. The input, or control variable
is the gas pedal angle. The system output, or result, is the velocity of the car. In standard
operation the gas pedal angle is controlled by the driver. When a cruise control system is
engaged the gas pedal must automatically be adjusted to maintain a desired velocity set-
point. To do this a control system is added, in this figure it is shown inside the dashed line.
In this control system the output velocity is subtracted from the setpoint to get a system
error. The subtraction occurs in the summation block (the circle on the left hand side).
This error is used by the controller function to adjust the control variable in the system.
Negative feedback is the term used for this type of controller.
 feedback control - 8.4

Control variable
INPUT OUTPUT
(e.g. θgas) SYSTEM
(e.g. velocity)
(e.g. a car)

vdesired verror control θgas car vactual

+ function
_

Note: The arrows in the diagram indicate directions so that outputs and inputs
are unambiguous. Each block in the diagram represents a transfer function.

Figure 8.4 An automotive cruise control system

There are two main types of feedback control systems: negative feedback and pos-
itive feedback. In a positive feedback control system the setpoint and output values are
added. In a negative feedback control the setpoint and output values are subtracted. As a
rule negative feedback systems are more stable than positive feedback systems. Negative
feedback also makes systems more immune to random variations in component values and
inputs.

The control function in Figure 8.4 can be defined many ways. A possible set of
rules for controlling the system is given in Figure 8.5. Recall that the system error is the
difference between the setpoint and actual output. When the system output matches the
setpoint the error is zero. Larger differences between the setpoint and output will result in
larger errors. For example if the desired velocity is 50mph and the actual velocity 60mph,
the error is -10mph, and the car should be slowed down. The rules in the figure give a gen-
eral idea of how a control function might work for a cruise control system.
 feedback control - 8.5

Human rules to control car (also like expert system/fuzzy logic):

1. If verror is not zero, and has been positive/negative for a while, increase/decrease θgas
2. If verror is very big/small increase/decrease θgas
3. If verror is near zero, keep θgas the same
4. If verror suddenly becomes bigger/smaller, then increase/decrease θgas.
5. etc.

Figure 8.5 Example control rules

In following sections we will examine mathematical control functions that are easy
to implement in actual control systems.

8.3.1 PID Control Systems

The Proportional Integral Derivative (PID) control function shown in Figure 8.6 is
the most popular choice in industry. In the equation given the ’e’ is the system error, and
there are three separate gain constants for the three terms. The result is a control variable
value.

u = K p e + K i ∫ edt + K d  ------
de
dt

Figure 8.6 A PID controller equation

Figure 8.7 shows a basic PID controller in block diagram form. In this case the
potentiometer on the left is used as a voltage divider, providing a setpoint voltage. At the
output the motor shaft drives a potentiometer, also used as a voltage divider. The voltages
from the setpoint and output are subtracted at the summation block to calculate the feed-
back error. The resulting error is used in the PID function. In the proportional branch the
error is multiplied by a constant, to provide a longterm output for the motor (a ballpark
guess). If an error is largely positive or negative for a while the integral branch value will
become large and push the system towards zero. When there is a sudden change occurs in
the error value the differential branch will give a quick response. The results of all three
branches are added together in the second summation block. This result is then amplified
to drive the motor. The overall performance of the system can be changed by adjusting the
gains in the three branches of the PID function.
 feedback control - 8.6

proportional PID function

Kp ( e )
V V
+ +V
integral
+ e u
Ki ( ∫ e ) amp motor
+
- derivative + -V
K d  ----- e
d
 dt 

Figure 8.7 A PID control system

There are other variations on the basic PID controller shown in Figure 8.8. A PI
controller results when the derivative gain is set to zero. (Recall the second order
response.) This controller is generally good for eliminating long term errors, but it is prone
to overshoot. In a P controller only the proportional gain in non-zero. This controller will
generally work, but often cannot eliminate errors. The PD controller does not deal with
longterm errors, but is very responsive to system changes.

For a PI Controller
θ gas = K p v error + K i ∫ v error dt

For a P Controller
θ gas = K p v error

For a PD Controller
dv error
θ gas = K p v error + K d  ----------------
 dt 

Figure 8.8 Some other control equations

 feedback control - 8.7

Aside: The manual process for tuning a PID controlled is to set all gains to zero. The pro-
portional gain is then adjusted until the system is responding to input changes without
excessive overshoot. After that the integral gain is increased until the longterm errors
disappear. The differential gain will be increased last to make the system respond
faster.

8.3.2 Manipulating Block Diagrams

A block diagram for a system is not unique, meaning that it may be manipulated
into new forms. Typically a block diagram will be developed for a system. The diagram
will then be simplified through a process that is both graphical and algebraic. For exam-
ple, equivalent blocks for a negative feedback loop are shown in Figure 8.9, along with an
algebraic proof.

c r
G( D) c G( D) r
------------------------------------
-
+
- is equal to 1 + G ( D )H ( D )

r = G ( D )e
H( D)
r = G ( D ) ( c – H ( D )r )

r ( 1 + G ( D )H ( D ) ) = G ( D )c
G( D)
r =  ------------------------------------- c
 1 + G ( D )H ( D )

Note: Units can be used to verify the feedback loop calculations.

Figure 8.9 A negative feedback block reduction

Other block diagram equivalencies are shown in Figure 8.10 to Figure 8.16. In all
cases these operations are reversible. Proofs are provided, except for the cases where the
equivalence is obvious.
 feedback control - 8.8

c r
G( D) c G( D) r
------------------------------------
-
+
+ is equal to 1 – G ( D )H ( D )

r = G ( D )e
H( D)
r = G ( D ) ( c + H ( D )r )

r ( 1 – G ( D )H ( D ) ) = G ( D )c
G( D)
r =  ------------------------------------- c
1 – G ( D )H ( D )

Figure 8.10 A positive feedback block reduction

r r 1 c
c G( D) is equal to -------------
G( D)
r = G ( D )c
c = r  -------------
1
G(D )

Figure 8.11 Reversal of function blocks

c r c r
G( D)
is equal to G( D)

G( D)
r r

Figure 8.12 Moving branches before blocks

 feedback control - 8.9

c x b is equal to
H( D) c b
G( D) G ( D )H ( D )

x = G ( D )c
b = H ( D )x
b = H ( D )G ( D ) )c

Figure 8.13 Combining sequential function blocks

c r c r
G( D) G( D)

is equal to 1
-------------
G( D)
c c

Figure 8.14 Moving branches after blocks

e c ± e
c ± G( D)
G( D)
is equal to
±
±
b 1 b
c -------------
G( D)
e = ± G ( D )c ± b

e = G ( D )  ± c ± -----------------
b
±G ( D )

Figure 8.15 Moving summation functions before blocks

 feedback control - 8.10

c r c r
G( D) G(D)

is equal to
d
d r = G ( D ) ( ±c ± d ) G( D)

r = ± cG ( D ) ± dG ( D )

Figure 8.16 Moving summation function past blocks

Recall the example of a cruise control system for an automobile presented in Fig-
ure 8.4. This example is extended in Figure 8.17 to include mathematical models for each
of the function blocks. This block diagram is first simplified by multiplying the blocks in
sequence. The feedback loop is then reduced to a single block. Notice that the feedback
line doesn’t have a function block on it, so by default the function is ’1’ - everything that
goes in, comes out.
 feedback control - 8.11

e.g. The block diagram of the car speed control system

vdesired verror Ki θgas F 1- vactual

K p + ----- + K d D 10 --------
+ D MD
-

vdesired verror Ki θgas 10- vactual

K p + ----- + K d D --------
+ D MD
-

= G(D)
vdesired verror vactual
K + K
-----i   --------
10-
+  p D + K d D  MD
-

H(D) = 1

K + K -----i   --------
10-
vdesired  p D + K d D  MD vactual
--------------------------------------------------------------------------
Ki
1 +  K p + ----- + K d D  --------- ( 1 )
10
D MD

Figure 8.17 An example of simplifying a block diagram

The function block is further simplified in Figure 8.18 to a final transfer function
for the whole system.
 feedback control - 8.12

 K + K -----i  
vdesired   p D + K d D  vactual
 -------------------------------------------------------
 Ms  K 
 ------- + K p + -----i + K d D 
10  D 

 2 
vdesired  D ( Kd ) + D ( K p ) + ( Ki )  vactual
 ----------------------------------------------------------------------
 D 2  -----M  
  10- + K d + D ( K p ) + ( K i )

 2 
v actual  D ( K d ) + D ( Kp ) + ( Ki ) 
----------------- =  ----------------------------------------------------------------------
v desired  D 2  -----M  
  10- + K d + D ( K p ) + ( K i )

Figure 8.18 An example of simplifying a block diagram (continued)

8.3.3 A Motor Control System Example

Consider the example of a DC servo motor controlled by a computer. The purpose

of the controller is to position the motor. The system in Figure 8.19 shows a reasonable
control system arrangement. Some elements such as power supplies and commons for
voltages are omitted for clarity.
 feedback control - 8.13

2.2K

Computer Running Labview 1K 12Vdc motor

PCI-1200 data acquisition card

-
gain Kp +

from National Instruments

LM675
X op-amp
shafts are coupled

-
+
+5V -5V
5K potentiometer

desired position
voltage Vd

Figure 8.19 A motor feedback control system

The feedback controller can be represented with the block diagram in Figure 8.20.

desired
position
voltage +
Vd gain Kp op-amp motor shaft

potentiometer

Figure 8.20 A block diagram for the feedback controller

 feedback control - 8.14

The transfer functions for each of the blocks are developed in Figure 8.21. Two of
the values must be provided by the system user. The op-amp is basically an inverting
amplifier with a fixed gain of -2.2 times. The potentiometer is connected as a voltage
divider and the equation relates angle to voltage. Finally the velocity of the shaft is inte-
grated to give position.

Given or selected values:

- desired potentiometer voltage Vd
- gain K
For the op-amp:
V+ – Vi V + – V o
∑ IV+ = ----------------- + ------------------ = 0
1K 2.2K
V + = V - = 0V

– Vi – V o
-------- + ----------- = 0
1K 2.2K
V
-----o- = – 2.2
Vi
For the potentiometer assume that the potentiometer has a range of 10 turns and 0
degrees is in the center of motion. So there are 5 turns in the negative and positive
direction.
θ
V o = 5V  --------------
5 ( 2π )
V
-----o- = 0.159Vrad –1
θ
For the shaft, it integrates the angular velocity into position:
d
ω = ----- θ
dt
θ- 1
--- = ----
ω D

Figure 8.21 Transfer functions for the power amplifier, potentiometer and motor shaft

The basic equation for the motor is derived in Figure 8.22 using experimental data.
In this case the motor was tested with the full inertia on the shaft, so there is no need to
calculate ’J’.
 feedback control - 8.15

For the motor use the differential equation and the speed curve when Vs=10V is applied:

1400 RPM 2
 ----
d-
ω +  ------ ω =  ------ V s
K K
 dt  JR  JR

1s 2s 3s
For steady-state
 ----
d-
ω = 0 ω = 1400RPM = 146.6rads
–1
 dt
2
0 +  ------ 146.6 =  ------ 10
K K
 JR  JR
K = 0.0682

τ ≈ 0.8s

1s
2
 -----
K -
= ---
1
 JR τ

0.0682  ------ = ----------

K 1
JR 0.8s
K-
----- = 18.328
JR
1
Dω + ------- ω = 18.33V s
0.8
ω- 18.33
---- = ---------------------
Vs D + 1.25

Figure 8.22 Transfer function for the motor

The individual transfer functions for the system are put into the system block dia-
gram in Figure 8.23. The block diagram is then simplified for the entire system to a single
transfer function relating the desired voltage (setpoint) to the angular position (output).
The transfer function contains the unknown gain value ’Kp’.
 feedback control - 8.16

Vd + θ
Kp -2.2 18.33 1-
--------------------- ---
D + 1.25 D
-

0.159

Vd + θ
– 40.33Kp -
-----------------------------
( D + 1.25 )D
-

0.159

Vd – 40.33Kp - θ
-----------------------------
( D + 1.25 )D
----------------------------------------------------------
-
– 40.33Kp
1 + ( 0.159 ) ------------------------------
( D + 1.25 )D

Vd θ
– 40.33Kp
---------------------------------------------------------
-
( D + 1.25 )D – 6.412Kp

Vd θ
– 40.33Kp
------------------------------------------------------
-
2
D + 1.25 D – 6.412Kp

Figure 8.23 The system block diagram, and simplification

The value of ’Kp’ can be selected to ’tune’ the system performance. In Figure 8.24
the gain value is calculated to give the system an overall damping factor of 1.0, or criti-
cally damped. This is done by recognizing that the bottom (homogeneous) part of the
transfer function is second-order and then extracting the damping factor and natural fre-
quency. The final result of ’Kp’ is negative, but this makes sense when the negative gain
on the op-amp is considered.
 feedback control - 8.17

We have specified, or been given the damping coefficient as a design objective.

ξ = 1.0
The denominator of the system transfer function can be compared to the standard
second-order response.
2 2
D + 1.25 D – 6.412K p = x'' + 2ξωn x' + ω n x
1.25 = 2ξω n
1.25 = 2 ( 1.0 )ω n
ω n = 0.625
2
– 6.412K p = ω n
2
– 6.412K p = 0.625
K p = – 0.0609

Figure 8.24 Calculating a gain Kp

8.3.4 System Error

System error is often used when designing control systems. The two common
types of error are system error and feedback error. The equations for calculating these
errors are shown in Figure 8.25. If the feedback function ’H’ has a value of ’1’ then these
errors will be the same.

e r
c
G System error, S = c–r
+
-
b e = c–b
Feedback error,
H

Figure 8.25 Controller errors

 feedback control - 8.18

An example of calculating these errors is shown in Figure 8.26. The system is a

simple integrator, with a unity feedback loop. The overall transfer function for the system
is calculated and then used to find the system response. The response is then compared to
the input to find the system error. In this case the error will go to zero as time approaches
infinity.

Given,
Kp
G ( D ) = ------ H( D) = 1
D
r G Kp
∴-- = ------------------ = ----------------
-
c 1 + GH D + Kp
·
∴r + K p c = K p c
c = A
The homogeneous solution is,
–Kp t
rh = C 1 e

The particular solution is found with a guess,

rp = C 2
·
rp = 0
0 + KpC2 = KpA C2 = A
The solutions can be combined and the remaining unknown found for the system at
rest initially.
–Kp t
r = rh + r p = C 1 e +A
0
0 = C1e + A

C1 = –A
–Kp t
r = –A e +A
The error can now be calculated.
S = c–r
–Kp t –Kp t
S = A – ( –A e + A ) = Ae

Figure 8.26 System error calculation example for a step input

 feedback control - 8.19

Solve the previous problem using a ramp input,

c = At

Figure 8.27 Drill problem: Calculate the system error for a ramp input
 feedback control - 8.20

Find the system error ’e’ for the given ramp input, R.

1
G ( D ) = -----------------------------
2
- H(D) = 5 c = 4t
D + 4D + 5

Figure 8.28 Drill problem: Calculate the errors

 feedback control - 8.21

8.3.5 Controller Transfer Functions

The PID controller, and simpler variations were discussed in earlier sections. A
more complete table is given in Figure 8.29.

Type Transfer Function

Proportional (P) Gc = K

G c = K  1 + -------
1
Proportional-Integral (PI) τD

Proportional-Derivative (PD) G c = K ( 1 + τD )

G c = K  1 + ------- + τD
1
Proportional-Integral-Derivative (PID)
τD

1 + ατD
Lead G c = K  -------------------- α>1
1 + τD
1 + τD
Lag G c = K  -------------------- α>1
1 + ατD
1 + τ 1 D   1 + ατ 2 D α>1
G c = K  ----------------------
- -----------------------
Lead-Lag  1 + ατ 1 D  1 + τ 2 D  τ1 > τ2

Figure 8.29 Standard controller types

8.3.6 Feedforward Controllers

When a model of a system is well known it can be used to improve the perfor-
mance of a control system by adding a feedforward function, as pictured in Figure 8.30.
The feedforward function is basically an inverse model of the process. When this is used
together with a more traditional feedback function the overall system can outperform
more traditional controllers function, such as the PID controller.
 feedback control - 8.22

feedforward
function

feedback process
function

Figure 8.30 A feed forward controller

8.3.7 State Equation Based Systems

State variable matrices were introduced before. These can also be used to form a
control system, as shown in Figure 8.31.

d-
---- X = AX + BU
dt
Y = CX + D

d-
+ ---- X X + Y
U dt 1-
B --- C
D
+ +

A D

Figure 8.31 A state variable control system

An example is shown in Figure 8.32 that implements a second order state equation.
The system uses two integrators to integrate the angular acceleration, then the angular
velocity, to get the position.
 feedback control - 8.23

d- d-
+ ---- ω ---- θ θ
τ dt 1- dt 1-
C --- ---
+ D D
d- +
---- θ = ω
dt A
d-
---- ω = Aω + Bθ + Cτ
dt
B

Figure 8.32 A second order state variable control system

The previous block diagrams are useful for simulating systems. These can then be
used in feedforward control systems to estimate system performance and then predict a
useful output value.

The model,
2
ω + ω  ------ = V s  ------
· K K
JR JR

V s = ω  ------ + ω ( K )
· JR
 K
State Based Controller
d-
ω ---- ω +
dt JR
A ------
K
+
K

+ K Plant
K P + -----i + DK d
D
-
Feedback
 feedback control - 8.24

Figure 8.33 A state based feedforward controller

8.3.8 Cascade Controllers

When controlling a multistep process a cascade controller can allow refined con-
trol of sub-loops within the larger control system. Most large processes will have some
form of cascade control. For example, the inner loop may be for a heating oven, while the
outer loop controls a conveyor feeding parts into the oven.

Gc1 Gc2 Gp1 Gp2

Figure 8.34 A cascade controller

8.4 SUMMARY

• Transfer functions can be used to model the ratio of input to output.

• Block diagrams can be used to describe and simplify systems.
• Controllers can be designed to meet criteria, such as damping ratio and natural
frequency.
• System errors can be used to determine the long term stability and accuracy of a
controlled system.
• Other control types are possible for more advanced systems.

8.5 PRACTICE PROBLEMS

1. Develop differential equations and then transfer functions for the mechanical system below.
There is viscous damping between the block and the ground. A force is applied to cause the
 feedback control - 8.25

mass the accelerate.

x

F
M

2. Develop a transfer function for the system below. The input is the force ‘F’ and the output is the
voltage ‘Vo’. The mass is suspended by a spring and a damper. When the spring is undeflected
y=0. The height is measured with an ultrasonic proximity sensor. When y = 0, the output
Vo=0V. If y=20cm then Vo=2V and if y=-20cm then Vo=-2V. Neglect gravity.

N s
K s = 10 ---- K d = 5N ---- M = 0.5Kg
m m
Ks Kd

Ultrasonic F y
Proximity
Sensor

3. Find the transfer functions for the systems below. Here the input is a torque, and the output is
the angle of the second mass.
θ1 θ2

J1 J2
τ K s1 K s2

B1 B2
 feedback control - 8.26

4. Find the transfer functions for the system below where Vi is the input and Vo is the output.
C L

+ +
Vi R1 R2 Vo
- -

5. Given the transfer function, G(s), determine the time response output Y(t) to a step input X(t).
4 Y
G = ------------- = ---
D+2 X X ( t ) = 20 When t >= 0

6. Given the transfer function below, develop a mechanical system that it could represent. (Hint:
Differential Equations).
x(D) 1 where: x = displacement
------------- = -----------------------
F(D ) 10 + 20D F = force

7. Given a mass supported by a spring and damper, find the displacement of the supported mass
over time if it is released from neutral at t=0sec, and gravity pulls it downward.
a) develop a transfer function for y/F.
b) find the input function F.
c) solve the input output equation to find an explicit equation of the position as a
function of time for Ks = 10N/m, Kd = 5Ns/m, M=10kg.
d) solve part c) numerically.

8. a) What is a Setpoint, and what is it used for? b) What does feedback do in control systems?

9. The block diagram below is for a servo motor position control system. The system uses a pro-
portional controller.
a) Draw a sketch of what the actual system might look like. Identify components.

θd V Ve Vm 10 - rad ω 1- θa
V- d K ------------ --------- ---
2 -------- D + 1 sV D
rad +
-

Va V
2 ---------
rad

b) Convert the system to a transfer function.

 feedback control - 8.27

10. Simplify the block diagram below to a transfer function.

A
- +
+ +
x B y
- +

11. Simplify the block diagram below. (Note: Vn is an input and cannot be combined in a transfer
function.)
+ +
A C
Vi Vo
- +

Vn

12. a) Develop an equation for the system below relating the two inputs to the output. Put it in
block diagram format. (Hint: think of a summation block.)

R R
Vf -
Ve
R
Vd +

b) Develop an equation for the system below relating the input to the output. Put
 feedback control - 8.28

the result in block diagram format.

1K 1K
-
Vs
1K
Ve +

RP

c) The equation below can be used to model a permanent magnet DC motor with
an applied torque. An equivalent block diagram is given. Prove that the block
diagram is equivalent to the equation.
2
 ----
d-  Km  Km T load
ω + ω  ------------- = V s  ------------- – ------------
 dt  J R R M  J R R M JR

Vs K
+
e RM ω
------m- ---------------------------------
-
RM 2
- JR RM D + Km

T load

d) Write the transfer function for the system below relating the input torque to the
output angle theta2. Then write the transfer function for the angular velocity of
mass 2.
θ1 θ2

J1 J2
τ K s1 K s2

e) The system below is a combination of previous components, and a tachometer

 feedback control - 8.29

for velocity feedback. Simplify the block diagram.

Vd K RM ω
2R P + ----------------------------------
-
+ --------------------
- ------m- 2
3 RM JM RM D + Km
R P + 10
- -

K s2
-----------------------------------------------------------------------------------------------------------------------
-
3 2
D ( J 1 J 2 ) + D ( BJ 1 ) + D ( Ks2 ( J 1 + J 2 ) ) + ( BK s2 )

KT

13. Find the system output, feedback error and system error when the input is a ramp with the
function c(t) = 0.5t. Sketch the system errors as a function of time.
C(D) + 5 - R( D)
----------------------
D( D + 1)
-

1 + 0.2D

14. Given the block diagram below, select a system gain K that will give the overall system a
damping ratio of 0.7 (for a step input). What is the resulting damped frequency of the system?

+ 2
K ------------------------------
2
D + 3D + 2
-

15. The following system is a feedback controller for an elevator. It uses a desired height ‘d’ pro-
vided by a user, and the actual height of the elevator ‘h’. The difference between these two is
called the error ‘e’. The PID controller will examine the value ‘e’ and then control the speed of
the lift motor with a control voltage ‘c’. The elevator and controller are described with transfer
functions, as shown below. all of these equations can be combined into a single system transfer
 feedback control - 8.30

equation as shown.
e = d–h error
K 2 h--- 10 - elevator
= ----------------
-r- = K p + -----i + K d D = 2D
+ 1 + D - PID controller
----------------------------- r D +D
2
e D D
combine the transfer functions
2 2
 -r-  h--- = h--- = -----------------------------
2D + 1 + D - ---------------- 10 - (D + 1) 10 10 ( D + 1 )
= --------------------- ----------------------- = ------------------------
 e  r  e D 2 D D(D + 1) 2
D +D D
h - 10 ( D + 1 ) eliminate ‘e’
----------- = ------------------------
d–h 2
D
( 1 )
h =  ------------------------
10 D +
(d – h)
 2 
D
10 ( D + 1 ) 10 ( D + 1 )
h  1 + ------------------------ =  ------------------------ (d)
 2   2 
D D
10 ( D + 1 ) 
 ------------------------
 2 
=  ---------------------------------- = -----------------------------------
h--- D 10D + 10 - system transfer function
d  10 ( D + 1 ) 2
 1 + ------------------------  D + 10D + 10
2
 D 
a) Find the response of the final equation to a step input. The system starts at rest
on the ground floor, and the input (desired height) changes to 20 as a step input.
b) Calculate the damping coefficient and natural frequency of the results in part a).

16. Study the circuit below. Assume that for t<0s the circuit is discharged and off. Starting at t=0s
an input of Vi=5sin(100,000t) is applied.

1K
+
Vi -
+
1uF
Vo
-

a) Write a differential equation and then a transfer function describing the circuit.
b) Find the output of the circuit using explicit integration (i.e., homogeneous and
particular solutions).

17. Write a C subroutine to implement the control system that is shown inside the dashed line. The
subroutine arguments are the setpoint, C, and the feedback value, F. the subroutine returns the
 feedback control - 8.31

new controller output value, U.

5D
+
C + e + U 1- R
Kp ------
5D
-

5
F

int controller(int C, int F){

8.6 PRACTICE PROBLEM SOLUTIONS

1.

--x- = ----------------------------
1
F D ( B – DM )

2.

10  ---- 10  ----
V V
Vo Vo y  m  m
------ = ------ --- = --------------------------------------------- = ---------------------------------------------------------------
F yF 2 2 Ns N
– MD – K d D – K s – 0.5kgD – 5 ------ D – 10 ----
m m

3.
θ2 K s2
----- = ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
τ 4 3 2
D ( J 1 J 2 ) + D ( B 1 J 2 + B 2 J 1 ) + D ( K s2 ( J 1 + J 2 ) + B1 B 2 ) + D ( K s2 ( B 1 + B 2 ) )

4.
Vo D ( R 1 R2 C )
------ = --------------------------------------------------------------------------------------------------
Vi 2
D ( R 1 LC ) + D ( L + CR 1 R 2 ) + ( R 1 + R2 )
 feedback control - 8.32

5.
– 2t
y ( t ) = – 40e + 40

6.

N
K s = 10 ----
m
F

Ns
K d = 20 ------
m

7.

--y- = -----------------------------------------
1
a) -
F 2
D M + DK d + K s
b) F = Mg
– 0.25t
c) y ( t ) = – 10.13e cos ( 0.968t – 0.253 ) + 9.81

d) y( t + h) = y(t) + h 0 1 y(t) + 0
v( t + h) v(t) – 1 – 0.5 v ( t ) 9.81

8.
a) A setpoint is a system input that correlates to a desired output for a system with
negative feedback control.
b) Feedback in a control system allows an out to be compared to an input. The differ-
ence, the error, is used to adjust the control variable and increase the accuracy.
 feedback control - 8.33

9.

a)
20K
----------------------------------
2
D + D + 20K

b)

10.

A+B
------------------------------------
-
1 + CB + A + B

11.

Vn --1-
A
+
Vo
+
CA
Vi -------------
1+A

12.

a) V + = V d  ------------- = 0.5V d
R
 R + R
V- – V f V - – V e
∑ IV -
= ---------------- + ----------------- = 0
R R

V e = 2V - – V f
Vd + Ve
V e = 2 ( 0.5V d ) – V f
-
Ve = Vd – Vf Vf
 feedback control - 8.34

b) V- – 0 V- – V s
∑ V- -------------
I =
10
3
- + ----------------- = 0
10
3

V - = 0.5V s (1)
V + – V e V+ – 0
∑ V+ -----------------
I =
10
3
- + --------------- = 0
RP
Ve
V+  -------3- + ------ = -------3-
1 1
 R P
10 10
substitute in (1)
 R P + 1K V
- = -------e-
0.5V s  -------------------
 R P ( 10 )3 3 Ve 2R P Vs
10 --------------------
-
3
V 2R P R P + 10
-----s = --------------------
-
Ve 3
R P + 10 note: a constant set by the variable resistor Rp

2
 ----
d-  Km  Km T load
c) ω + ω  ------------- = V s  ------------- – ------------
 dt  J R R M  J R R M JR
2
 Km  K m  T load
- = V s  ------------
ω  D + ------------ - – ------------
 J R R M  J R R M JR
 JR M  K m  T load
-  V s  ------------
ω =  --------------------------------- - – ------------
 J R R M D + K m 2   J R R M JR 

 RM  Km
ω =  ---------------------------------2-  V s  ------- – T load
 J R R M D + K m    R M 
 feedback control - 8.35

d)
K s2 ( θ 1 – θ 2 )
τ
∑M
2
= τ – K s2 ( θ 1 – θ 2 ) = J 1 θ 1 D
+
J1 2
τ – K s2 ( θ 1 – θ 2 ) = J 1 θ1 D
2
θ 1 ( J 1 D + K s2 ) + θ 2 ( – K s2 ) = τ (1)

K s2 ( θ 1 – θ 2 )
∑M
2
= K s2 ( θ 1 – θ 2 ) – Bθ2 D = J 2 θ 2 D
+
J2 2
K s2 ( θ 1 – θ 2 ) – Bθ2 D = J 2 θ 2 D
Bθ 2 D 2
θ 2 ( J 2 D + K s2 + BD ) = θ 1 ( K s2 )
2
 J 2 D + K s2 + BD
θ 1 = θ 2  ----------------------------------------- (2)
 K s2 
substitute (2) into (1)
2
 J 2 D + BD + K s2 2
θ 2  ----------------------------------------- ( J 1 D + K s2 ) + θ 2 ( – K s2 ) = τ
 K s2 
4 3 2
 D ( J 1 J 2 ) + D ( BJ 1 ) + D ( K s2 ( J 1 + J 2 ) ) + D ( BK s2 )
- = τ
θ 2  ------------------------------------------------------------------------------------------------------------------------------
 K s2 

θ2 K s2
----- = ------------------------------------------------------------------------------------------------------------------------------
-
τ 4 3 2
D ( J 1 J 2 ) + D ( BJ 1 ) + D ( K s2 ( J 1 + J 2 ) ) + D ( BK s2 )

ω 2 = Dθ 2

ω K s2
-----2- = -----------------------------------------------------------------------------------------------------------------------
-
τ 3 2
D ( J 1 J 2 ) + D ( BJ 1 ) + D ( K s2 ( J 1 + J 2 ) ) + ( BK s2 )

e)
3 2
ω-
----- D ( ) + D ( ) + D( ) + ( ) -
= ------------------------------------------------------------------------------------------
4 3 2
Vd D ( ) + D ( ) + D ( ) + D( ) + ( )
 feedback control - 8.36

13.

R ( t ) = – 0.224e cos  ---------- t – 0.464 – --- + -----

–t 15 1 1
2 5 2t

S ( t ) = 0.224e cos  ---------- t – 0.464 – ---

–t 15 1
2 5

e ( t ) = – 0.112e cos  ---------- t – 0.464 + ------

–t 15 1
2 10

14.
rad
ω d = 1.530 --------- K = 1.30
s

15.
– 8.873t – 1.127t
a) h ( t ) = 2.91e – 22.91e + 20

rad
b) ζ = 1.58 ω n = 3.162 ---------
s

16.

Vo 1000
a) ------ = ----------------------
Vi D + 1000

– 1000t 5
b) V o = 0.050e + 0.050 sin ( 10 t + 1.561 )

17.

int controller(){

}
 feedback control - 8.37

8.7 ASSIGNMENT PROBLEMS

1. Develop a transfer function for the circuit below.

Vi R
C
Vo
-

2. Given the transfer functions and input functions, F, use a numerical method to calculate the out-
put of the system as a function of time for 0 to 0.5 seconds in 0.05 second intervals. Record the
values in a table.
2
--x- = -----------------------------
D -
2
F = 5 sin ( 62.82t )
F ( D + 200π )
D ( D + 2π )-
--x- = ----------------------------- F = 5 sin ( 62.82t )
F 2
( D + 200π )
2
D ( D + 2π )-
--x- = ----------------------------- F = 5 sin ( 62.82t )
F 2
( D + 200π )

3. Simplify the following block diagram.

+ 1 -
------------
D+1
-

1 -
------------
D+2
 feedback control - 8.38

4. Simplify the following block diagram.

+
+ +
A B C

5. Simplify the following block diagram.

2
D + 3D + 5
+
+ +
K 1
---- ------------------------------
D 2
D + 3D + 5
-
10

6. Simplify the following block diagram.

2
D + 3D + 5
+
+ +
K- 1
--- ------------------------------
D 2
D + 3D + 5
-
10
 feedback control - 8.39

7. Simplify the following block diagram.

G1

-
C + + + R
G2 G3 G5

- -

G4 G6

8. Simplify the following block diagram for a crane gantry position control system. The system
also includes a negative feedback loop to control the sway of an attached load
encoder
Cp
K enc

-
Cd C pe C pc Vs θp
K pp K pwm G motor
+ +
- C Lc
CL VL θL
K sp K AD K pot G load

9. Write a C subroutine to implement the control system that is shown inside the dashed line. The
subroutine arguments are the setpoint, C, and the feedback value, F. the subroutine returns the
new controller output value, U.

C + e U 1 - R
Kp + KdD --------------------
2
(D + 1)
-
Ki
----- 5
D F

10. Write a C subroutine to implement the control system that is shown inside the dashed line. The
subroutine arguments are the setpoint, C, and the feedback value, F. the subroutine returns the
 feedback control - 8.40

new controller output value, U.

C + e Ki U 1 - R
K p + K d D + ----- --------------------
2
D (D + 1)
-

5
F
 frequency response - 9.1

9. PHASOR ANALYSIS

Topics:
• Phasor forms for steady state analysis
• Complex and polar calculation of steady state system responses
• Vibration analysis

Objectives:
• To be able to analyze steady state responses using the phasor transform

9.1 INTRODUCTION

When a system is stimulated by an input it will respond. Initially there is a substan-

tial transient response, that is eventually replaced by a steady state response. Techniques
for finding the combined steady state and transient responses were covered in earlier chap-
ters. These include the integration of differential equations, and numerical solutions. Pha-
sor analysis can be used to find the steady state response only. These techniques involve
using the phasor transform on the system transfer function, input and output.

9.2 PHASORS FOR STEADY-STATE ANALYSIS

When considering the differential operator we can think of it as a complex number,

as in Figure 9.1. The real component of the number corresponds to the natural decay (e-to-
the-t) of the system. But, the complex part corresponds to the oscillations of the system. In
other words the real part of the number will represent the transient effects of the system,
while the complex part will represent the sinusoidal steady-state. Therefore to do a steady-
state sinusoidal analysis we can replace the ’D’ operator with jω , this is the phasor trans-
form.

D = σ + jω where,
D = differential operator Phasor transform
σ = decay constant D = jω
ω = oscillation frequency

Figure 9.1 Transient and steady-state parts of the differential operator

 frequency response - 9.2

An example of the phasor transform is given in Figure 9.2. We start with a transfer
function for a mass-spring-damper system. In this example numerical values are assumed
to put the equation in a numerical form. The differential operator is replaced with jω, and
the equation is simplified to a complex number in the denominator. This equation then
described the overall response of the system to an input based upon the frequency of the
input. A generic form of sinusoidal input for the system is defined, and also converted to
phasor (complex) form. (Note: the frequency of the input does not show up in the complex
form of the input, but it will be used later.) The steady state response of the system is then
obtained by multiplying the transfer function by the input, to obtain the output.

A phasor transform can be applied to a transfer function for a mass-spring-damper

system. Some component values are assumed.
N Ns
M = 1000kg K s = 2000 ---- K d = 3000 ------
m m
x ( D ) 1 1
------------- = ------------------------------------------ = ------------------------------------------------------------
F( D) 2
MD + K d D + K s
2
1000D + 3000D + 2000
x(ω ) 1 1
------------ = -----------------------------------------------------------------
- = ----------------------------------------------------------------------
-
F( ω) 2 2
1000j ω + 3000jω + 2000 ( 2000 – 1000ω ) + j ( 3000ω )
2

A given input function can also be converted to phasor form.

F ( t ) = A sin ( ω input t + θ input ) Note: the frequency is not used when
F ( ω ) = A ( cos θ input + j sin θ input ) converting an oscillating signal to
complex form. But it is needed for the
F ( ω ) = A cos θ input + jA sin θ input transfer function.

The response of the steady state output ’x’ can now be found for the given input.
x(ω ) 1 - ( A cos θ input + jA sin θ input )
x ( ω ) = ------------ F ( ω ) = ----------------------------------------------------------------------
F( ω) ( 2000 – 1000ω ) + j ( 3000ω )
2

Figure 9.2 A phasor transform example

To continue the example in Figure 9.2 values for the sinusoidal input force are
assumed. After this the method only requires the simplification of the complex expression.
In particular having a complex denominator makes analysis difficult and is undesirable.
To simplify this expression it is multiplied by the complex conjugate. After this, the
expression is quickly reduced to a simple complex number. The complex number is then
converted to polar form, and then finally back into a function of time.
 frequency response - 9.3

Assume the input to the system is, F ( t ) = 10 sin ( 100t + 0.5 )N

rad
A = 10N ω = 100 --------- θinput = 0.5rad
s
These can be applied to find the steady state output response,
1
x ( ω ) = ------------------------------------------------------------------------------------------
2
- ( 10 cos 0.5 + j10 sin 0.5 )
( 2000 – 1000 ( 100 ) ) + j ( 3000 ( 100 ) )
1
x ( ω ) = ------------------------------------------------------------ ( 8.776 + j4.794 )
( – 9998000 ) + j ( 300000 )
8.776 + j4.794 ( – 9998000 – j300000 )
x ( ω ) = --------------------------------------------------- --------------------------------------------------------
– 9998000 + j300000 ( – 9998000 – j300000 )

Note: This is known as the complex conjugate;

1. The value is equivalent to 1 so it does not change the value of the expression
2. The complex component is now negative
3. Only the denominator is used top and bottom

( – 86304248 ) + j ( – 50563212 -)
x ( ω ) = -----------------------------------------------------------------------
14
1.0005 ×10
-6 -6
x ( ω ) = ( – 0.863 ×10 ) + j ( – 0.505 ×10 )
-6 2 -6 2  – 0.505 ×10-6
x(ω ) = ( – 0.863 ×10 ) + ( – 0.505 ×10 ) ∠ atan  -----------------------------
- + π
 – 0.863 ×10-6

Note: the signs of the components indicate that the angle is in the bottom left quad-
rant of the complex plane, so the angle should be between 180 and 270 degrees.
To correct for this pi radians are added to the result of the calculation.

-6
x ( ω ) = 0.9999 ×10 ∠3.671
This can then be converted to a function of time.
-6
x ( t ) = 0.9999 ×10 sin ( 100t + 3.671 )m

Figure 9.3 A phasor transform example (cont’d)

 frequency response - 9.4

Note: when dividing and multiplying complex numbers in polar form the magnitudes can
be multiplied or divided, and the angles added or subtracted.
Unfortunately when the numbers are only added or subtracted they need to be converted
back to cartesian form to perform the operations. This method eliminates the need to
multiply by the complex conjugate.

A + B ∠ atan  ---
2 2 B
 A 2 2
= ------------------------------------------------- = ----------------------- ∠ atan  --- – atan  ---- 
A + jB- A +B B D
---------------
C + jD A C
C + D ∠ atan  ----
2 2 D C +D
2 2
C

A ∠θ
-------------1- = A
--- ∠( θ 1 – θ 2 )
B ∠θ 2 B

( A ∠θ 1 ) ( B ∠θ 2 ) = AB ∠( θ 1 + θ2 )

For example,

( 1 + j )  --------------
2+j
3 + 4j

5 ∠0.4636
= 2 ∠0.7854 -------------------------------
25 ∠0.9273

2 5
= -------------- ∠0.7854 + 0.4636 – 0.9273
25
= 0.6325 ∠0.3217
= 0.6 + j0.2

Figure 9.4 Calculations in polar notation

The cartesian form of complex numbers seen in the last section are well suited to
operations where complex numbers are added and subtracted. But, when complex num-
bers are to be multiplied and divided these become tedious and bulky. The polar form for
complex numbers simplifies many calculations. The previous example started in Figure
9.2 is redone using polar notation in Figure 9.5. In this example the input is directly con-
verted to polar form, without the need for calculation. The input frequency is substituted
into the transfer function and it is then converted to polar form. After this the output is
found by multiplying the transfer function by the input. The calculations for magnitudes
involve simple multiplications. The angles are simply added. After this the polar form of
the result is converted directly back to a function of time.
 frequency response - 9.5

Consider the input function from the previous example in polar form it becomes,
F ( t ) = 10 sin ( 100t + 0.5 )N F ( ω ) = 10 ∠0.5
The transfer function can also be put in polar form.
x(ω) 1 1
------------ = ------------------------------------------------------------------------------------------
- = ---------------------------------------------------
F(ω) ( 2000 – 1000 ( 100 ) ) + j ( 3000 ( 100 ) )
2 – 9998000 + j300000
x(ω) 1 ∠0
------------ = --------------------------------------------------------------------------------------------------------------------------------
F(ω)
( – 9998000 ) + ( 300000 ) ∠ atan  ------------------------ + π
2 2 300000
  – 9998000 
x ( ω ) 1 ∠ 0 1
------------ = ------------------------------------------ = ------------------------ ∠( 0 – 3.112 ) = 0.9998 ×10-7 ∠– 3.112
F(ω) 10002500 ∠3.112 10002500
The output can now be calculated.
x( ω) -7
x ( ω ) = ------------ F ( ω ) = ( 0.9998 ×10 ∠– 3.112 ) ( 10 ∠0.5 )
F(ω)
-7 -6
x ( ω ) = 0.9998 ×10 ( 10 ) ∠( – 3.112 + 0.5 ) = 0.9998 ×10 ∠– 2.612
The output function can be written from this result.
-6
x ( ω ) = 0.9998 ×10 sin ( 100t – 2.612 )

Re imaginary
Note: recall that tan θ = ------ but the atan θ function in
Im
calculators and software only returns values between - +/- +/+
90 to 90 degrees. To compensate for this the sign of the real
real and imaginary components must be considered to
determine where the angle lies. If it lies beyond the -90 -/- -/+
to 90 degree range the correct angle can be obtained
by adding or subtracting 180 degrees.

Figure 9.5 Correcting quadrants for calculated angles

Consider the circuit analysis example in Figure 9.6. In this example the component
values are converted to their impedances, and the input voltage is converted to phasor
form. (Note: this is a useful point to convert all magnitudes to powers of 10.) After this the
three output impedances are combined to a single impedance. In this case the calculations
were simpler in the cartesian form.
 frequency response - 9.6

Given the circuit,

10K Ω
+
1mH 100nF 10K Ω
+ Vo
-
6 -
V i = 5 sin ( 10 t + 0.3 )

The impedances and input voltage can be written in phasor form.

4
10
–3 6 1 - 4 +
5 –0.3 10 10 j ----------------------- 10
+ –7 6 Vo
- 10 10 j
-

The three output impedances in parallel can then be combined.

4
10
+
5 ∠0.3 Z eq
+ Vo
-
-
1- 1
1 - -------------------------- 1-
------- = -------------------- + - + -------
10 10 j  -------------------- - 10
Z eq – 3 6 1 4
 –7 6 
10 10 j
1- –3 –1 –4 –4
------- = – 10 j + 10 j + 10 = ( 10 ) + j ( 0.099 )
Z eq
1
Z eq = -----------------------------------------
–4
-
( 10 ) + j ( 0.099 )

Figure 9.6 Phasor analysis of a circuit

The analysis continues in Figure 9.7 as the output is found using a voltage divider.
In this case a combination of cartesian and polar forms are used to simplify the calcula-
tions. The final result is then converted back from phasor form to a function of time.
 frequency response - 9.7

The output can be found using the voltage divider form.

 Z eq 
V o ( ω ) = ( 5 ∠0.3 )  ---------------------- 
 10 4 + Z eq
  -----------------------------------------
1 - 
  –4 
( 10 ) + j ( 0.099 )
V o ( ω ) = ( 5 ∠0.3 )  ---------------------------------------------------------------
 10 +  ----------------------------------------- - 
4 1

–4 
( 10 ) + j ( 0.099 ) 

V o ( ω ) = ( 5 ∠0.3 )  ---------------------------------------------------------------------
-
1
 4 –4 
( 10 ) ( ( 10 ) + j ( 0.099 ) ) + 1
5 ∠0.3 5 ∠0.3
V o ( ω ) = ( 5 ∠0.3 )  -------------------- = ---------------------------------------------------- = ---------------------------------------
1
2 + j990 2 2 990 990 ∠1.5687761
2 + 990 ∠ atan ---------
2
5 –3
V o ( ω ) = --------- ∠( 0.3 – 1.5687761 ) = 5.05 × 10 ∠– 1.269
990
Finally, the output voltage can be written.
6
V o ( t ) = 5.05 sin ( 10 t – 1.269 ) mV

Figure 9.7 Phasor analysis of a circuit (cont’d)

Phasor analysis is applicable to systems that are linear. This means that the princi-
ple of superposition applies. Therefore, if an input signal has more than one frequency
component then the system can be analyzed for each component, and then the results sim-
ply added. The example considered in Figure 9.2 is extended in Figure 9.8. In this exam-
ple the input has a static component, as well as frequencies at 0.5 and 20 rad/s. The
transfer function is analyzed for each of these frequencies components. The output com-
ponents are found by multiplying the inputs by the response at the corresponding fre-
quency. The results are then converted back to functions of time, and added together.
 frequency response - 9.8

Given the transfer function for the system,

x(ω) 1 Note: these are gains
------------ = ----------------------------------------------------------------------
-
F(ω) ( 2000 – 1000ω ) + j ( 3000ω )
2 and phase shifts
that will be used
and an input with multiple frequency components, heavily in Bode
F ( t ) = 1000 + 20 sin ( 20t ) + 10 sin ( ( 0.5t ) ( N ) ) plots later.
the transfer function for each frequency can be calculated,
x(0) 1 1 -
----------- = ------------------------------------------------------------------------------
- = ----------- = 0.0005 ∠0
F(0) ( 2000 – 1000 ( 0 ) ) + j ( 3000 ( 0 ) )
2 2000
x ( 20 ) 1 1 1 ∠0
-------------- = ------------------------------------------------------------------------------------
- = --------------------------------------------- = ------------------------------------
F ( 20 ) ( 2000 – 1000 ( 20 ) ) + j ( 3000 ( 20 ) )
2 – 398000 + j60000 402497 ∠2.992
x ( 0.5 )- 1 1 1 ∠0 -
--------------- = ---------------------------------------------------------------------------------------
- = ------------------------------- - = -----------------------------
F ( 0.5 ) ( 2000 – 1000 ( 0.5 ) ) + j ( 3000 ( 0.5 ) )
2 1750 + j1500 2305 ∠0.709

These can then be multiplied by the input components to find output components.
x ( 0 ) = ( 0.0005 ∠0 )1000 ∠0 = 0.5 ∠0
1 ∠0
x ( 20 ) =  ------------------------------------ 20 ∠0 = 0.497 ×10 ∠– 2.992
-4
402497 ∠2.992
1 ∠0
x ( 0.5 ) =  ------------------------------ 10 ∠0 = 4.34 × 10 ∠– 0.709
–3
2305 ∠0.709
Therefore the output is,
–6 –3
x ( t ) = 0.5 + 49.7 × 10 sin ( 20t – 2.992 ) + 4.34 × 10 sin ( 0.5t – 0.709 )

Figure 9.8 A example for a signal with multiple frequency components (based on the
example in Figure 9.2)

9.3 VIBRATIONS

Oscillating displacements and forces in mechanical systems will cause vibrations.

In some cases these become a nuisance, or possibly lead to premature wear and failure in
mechanisms. A common approach to dealing with these problems is to design vibration
isolators. The equations for transmissibility and isolation is shown in Figure 9.9. These
equations can compare the ratio of forces or displacements through an isolator. The calcu-
lation is easy to perform with a transfer function or Bode plot.
 frequency response - 9.9

Given a vibration force in, to a force out,

F out ( ω ) x out ( ω )
T = ------------------- = ------------------ The gain of a transfer function gives transmissibility
F in ( ω ) x in ( ω )

%I = ( 1 – T )100%

Figure 9.9 Transmissibility

Given the transfer function for a vibration isolator below, find a value of K that will
give 50% isolation for a 10Hz vibration.
x out 5D + 10
--------- = ----------------------------------------
-
x in 2
4D + 20KD + 4

Figure 9.10 Drill problem: Select a K value

 frequency response - 9.10

9.4 SUMMARY

• Phasor transforms and phasor representations can be used to find the steady state
response of a system to a given input.
• Vibration analysis determines frequency components in mechanical systems.

9.5 PRACTICE PROBLEMS

1. Develop a transfer function for the system pictured below and then find the response to an input
voltage of Vi = 10sin(1,000,000 t) using phasor transforms.
1mH
1KΩ
+

+ 1KΩ 1KΩ 1KΩ

Vi Vo
-

2. A single d.o.f. model with a weight of 1.2 kN and a stiffness of 340 N/m has a steady-state har-
monic excitation force applied at 95 rpm (revolutions per minute). What damper value will
give a vibration isolation of 92%?

3. Four helical compression springs are used, one at each corner of a piece of equipment. The
spring rate is 240 N/m for each spring and the vertical static deflection of the equipment is
10mm. Calculate the mass of the equipment and determine the amount of isolation the springs
would afford if the equipment operating frequency is twice the natural frequency of the system.

4. a) For the circuit below find the transfer function and the steady state response for an input of
 frequency response - 9.11

Vi=5sin(1000t)V.
1K

1uF
1mH
-
+ + + Vo
Vi
-
-

b) Verify the results in part a) by explicitly solving the differential equation.

 frequency response - 9.12

9.6 PRACTICE PROBLEM SOLUTIONS

1.
1KΩ Va
1mH +

+ 1KΩ 1KΩ 1KΩ

Vi Vo
-

Va – Vi Va Va V a – Vo
∑ I Va = ----------------
1KΩ
- + -----------
- +
1KΩ 1KΩ 0.001D
-----------
- + ------------------ = 0

Va  ------------ + ------------------ + Vi  ------------ = Vo  ------------------

3 1 –1 1
(1)
1KΩ 0.001D 1KΩ 0.001D
Vo – Va Vo
∑ IVo = -----------------
0.001D 1KΩ
- + ------------ = 0

V o  ------------------ + ------------ = V a  ------------------

1 1 1
0.001D 1KΩ 0.001D

V o  ------------------------------------ = V a
0.001D + 1KΩ
(2)
1KΩ
substitute (2) into (1)

V o  ------------------------------------  ------------ + ------------------ + Vi  ------------ = Vo  ------------------

0.001D + 1KΩ 3 1 –1 1
 1KΩ   1KΩ 0.001D   1KΩ   0.001D
1 -
Vo -----------
1KΩ
------ = -------------------------------------------------------------------------------------------------------------------
Vi  0.001D + 1KΩ  ----------- 3 - ----------------- 1 -  ----------------- 1 -
------------------------------------ + –
 1KΩ   1KΩ 0.001D  0.001D
 frequency response - 9.13

for the given input of Vi(t) = 10sin(1,000,000 t).

1 -
Vo -----------
1KΩ
----------------- = -----------------------------------------------------------------------------------------------------------------------------------------------
10 + 0j 6
 0.001j10 + 1KΩ-  -----------
------------------------------------------ 3 - ----------------------------
+
1 - –  ----------------------------
1 -
 1KΩ   1KΩ 6   6 
0.001 ( j10 ) 0.001 ( j10 )
Vo –3
10
----------------- = -----------------------------------------------------------------------------------
-
10 + 0j –3 –3 –3
( j + 1 ) ( 3 × 10 – j10 ) + j ( 10 )
Vo 1
----------------- = ---------------------------------------
10 + 0j ( j + 1)(3 – j) + j
Vo
- =  -------------
1 -  4-------------
– 3j-
----------------
10 + 0j  4 + 3j 4 – 3j 

V o = 10  -------------- = 2 ∠– 0.644
4 – 3j
 25 
6
V o ( t ) = 2.00 sin ( 10 t – 0.644 )V

2.
F
y
∑ Fy
2
M = – K s y – K d yD – F = MyD

--y- = -----------------------------------------
–1 -
F 2
D M + DK d + K s
Ks Kd
F floor = K s y + K d yD
F floor
------------
- = Ks + K d D
y
F floor F floor  y  –( Ks + Kd D )
------------- =  ------------
- --- = -----------------------------------------
-
F  y   F 2
D M + DK d + K s
F floor – ( K s + K d ωj )
------------- = ----------------------------------------------
F 2
– ω M + ωjK d + K s

2 2
F floor Ks + ( K d ω )
- = ------------------------------------------------------------
------------
F 2 2 2
( K s – ω M ) + ( ωK d )
 frequency response - 9.14

(cont’d
For 92% isolation, there is 100-92 = 8% transmission, at 95rpm.
N 1200N
K s = 340 ---- M = ----------------- = 122kg
m N
9.81 ------
kg

ω =  95 ---------  --------------  ---------------- = 9.95 ---------

rev 1min 2πrad rad
 min  60sec  rev  s
 340 --- N- 2  rad 
2
+ K 9.95 --------
-
 m  d s 
0.08 = ---------------------------------------------------------------------------------------------------------------------
2 2 2
 340 --- N- 
– 9.95 --------- 122kg +  9.95 rad
rad --------- K d
 m  s    s 
 340 --- N- 2  2
---------
rad
+ K 9.95
2 2 2  m  d s 
 340 ---
N- 
– 9.95 --------- 122kg +  9.95 rad
rad
--------- K d = ---------------------------------------------------------------
 m  s    s  2
0.08
2 2 2 2
N 8 rad - 2 115600 N 2 99.0025 rad
( 1.377878 ×10 ) -----2- + 99.0025 ----------
2
K d = ------------------ ------2 + K d ------------------- ----------
2
-
m s 0.0064 m 0.0064 s
8 2 2
 1.197253 ×10 - -----------------
N s - 2 Ns
--------------------------------- = Kd
 15370.138  2 2 K d = 88.3 ------
m rad m

3.
a) M = 0.979kg

b) I = 67%

4.

a) V o ( t ) = 3536 sin ( 1000t + 0.785 )

π
+ 3536 sin  1000t + ---
– 1000t
b) V o ( t ) = – 5000 + 2500e
4

9.7 ASSIGNMENT PROBLEMS

1. For the following transfer function,

 frequency response - 9.15

a) Draw the Bode plot on the attached semi-log graph paper.

b) Given an input of F=5sin(62.82t), find the output, x, using the Bode plot.
c) Given an input of F=5sin(62.82t), find the output, x, using phasors.
2
--x- = -----------------------------
D -
F 2
( D + 200π )

2. For the following transfer function,

a) Draw the Bode plot on the attached semi-log graph paper.
b) Given an input of F=5sin(62.82t), find the output, x, using the Bode plot.
c) Given an input of F=5sin(62.82t), find the output, x, using phasors.

D ( D + 2π )-
--x- = -----------------------------
F 2
( D + 200π )

3. For the following transfer function,

a) Draw the Bode plot on the attached semi-log graph paper.
b) Given an input of F=5sin(62.82t), find the output, x, using the Bode plot.
c) Given an input of F=5sin(62.82t), find the output, x, using phasors.
2
D ( D + 2π )-
--x- = -----------------------------
F 2
( D + 200π )
 bode plots - 10.1

10. BODE PLOTS

Topics:
• Bode plots

Objectives:
• To be able to describe the response of a system using Bode plots

10.1 INTRODUCTION

When a phasor transform is applied to a transfer function the result can be

expressed as a magnitude and angle that are functions of frequency. The magnitude is the
gain, and the angle is the phase shift. In the previous chapter these values were calculated
for a single frequency and then multiplied by the input values to get an output value. At
different frequencies the transfer function value will change. The transfer function gain
and phase angle can be plotted as a function of frequency to give an overall picture of sys-
tem response.
 bode plots - 10.2

Aside: Consider a ’graphic equalizer’ commonly found on home stereo equipment. The
spectrum can be adjusted so that high or low tones are emphasized or muted. The
position of the sliders adjusts the envelope that the audio signal is filtered through.
The sliders trace out a Bode gain plot. In theoretical terms the equalizer can be
described with a transfer function. As the slides are moved the transfer function is
changed, and the bode plot shifts. In the example below the slides are positioned to
pass more of the lower frequencies. The high frequencies would not be passed clearly,
and might sound somewhat muffled.
40-120 120-360 360-1K 1K-3.2K 3.2K-9.7K 9.7K-20K

gain

f(Hz)

Figure 10.1 Commonly seen Bode plot

The mass-spring-damper transfer function from the previous chapter is expanded

in Figure 10.2. In this example the transfer function is multiplied by the complex conju-
gate to eliminate the complex number in the denominator. The magnitude of the resulting
transfer function is the gain, and the phase shift is the angle. Note that to correct for the
quadrant of the phase shift angles pi radians is subtracted for certain frequency values.
 bode plots - 10.3

x(ω ) 1
------------ = ----------------------------------------------------------------------
-
F( ω) j ( 3000ω ) + ( 2000 – 1000ω )
2

2
x(ω ) 1 (------------------------------------------------------------------------------
– j ) ( 3000ω ) + ( 2000 – 1000ω )-
------------ = ----------------------------------------------------------------------
-
F( ω) j ( 3000ω ) + ( 2000 – 1000ω )
2
( – j ) ( 3000ω ) + ( 2000 – 1000ω )
2

2
x(ω ) ( 2000 – 1000ω ) – j ( 3000ω )-
------------ = -------------------------------------------------------------------------
F( ω) 2 2 2
( 2000 – 1000ω ) + ( 3000ω )
2 2 2
( 2000 – 1000ω ) + ( 3000ω ) -
x ( ω ) = ----------------------------------------------------------------------------- 1
------------ = ------------------------------------------------------------------------------
F( ω) ( 2000 – 1000ω ) + ( 3000ω )
2 2 2 2 2 2
( 2000 – 1000ω ) + ( 3000ω )

θ = atan  ----------------------------------------- = atan  ------------

– 3000ω –3ω-
 2   2 for ( ω ≤ 2 )
– 1000ω + 2000 2 –ω

θ = atan   ------------2- – π

– 3ω
for ( ω > 2 )
2 –ω

Figure 10.2 A phasor transform example

The results in Figure 10.2 are normally left in variable form so that they may be
analyzed for a range of frequencies. An example of this type of analysis is done in Figure
10.3. A set of frequencies is used for calculations. These need to be converted from Hz to
rad/s before use. For each one of these the gain and phase angle is calculated. The gain
gives a ratio between the input sine wave and output sine wave of the system. The magni-
tude of the output wave can be calculated by multiplying the input wave magnitude by the
gain. (Note: recall this example was used in the previous chapter) The phase angle can be
added to the input wave to get the phase of the output wave. Gain is normally converted to
’dB’ so that it may cover a larger range of values while still remaining similar numerically.
Also note that the frequencies are changed in multiples of tens, or magnitudes.
 bode plots - 10.4

f(Hz) (rad/sec) Gain Gain (dB) θ (rad.) θ (deg.)

0 0
0.001 0.006283
0.01 0.06283
0.1 0.6283
1 6.283
10 62.83
100 628.3
1000 6283

Note: the gain values will cover many magnitudes of Note: negative phase
values. To help keep the graphs rational we will angles mean that the
"compress" the values by converting them to dB mass motion lags the
(decibels) using the following formula. force.
gain db = 20 log ( gain )

Note: The frequencies chosen should be chosen to cover the points with the greatest
amount of change.

Figure 10.3 A phasor transform example (continued)

In this example gain is defined as x/F. Therefore F is the input to the system, and x
is the resulting output. The gain means that for each unit of F input to the system, there
will be gain*F=x output. If the input and output are sinusoidal, there is a difference in
phase between the input and output wave of θ (the phase angle). This is shown in Figure
10.4, where an input waveform is supplied with three sinusoidal components. For each of
the frequencies a gain and phase shift are calculated. These are then used to calculate the
resulting output wave, relative to the input wave. The resulting output represents the
steady-state response to the sinusoidal output.
 bode plots - 10.5

Assuming there is excitation from two sinusoidal sources in addition to the

static load, as defined by the equation,

F ( t ) = 1000 + 20 sin ( 20t ) + 10 sin ( 0.5t ) ( N )

The result for each component can be evaluated separately

x ( t ) = 1000 ( _________ ) sin  0t + ______  --------- 

· 2π
360

+ 20 ( _________ ) sin  20t + _______  --------- 

2π
  360 

+ 10 ( __________ ) sin  0.5t – ________  --------- 

2π
360
gain from phase angle from
calculations calculations
or Bode plot or Bode plot

x ( t ) = 0.5 + ______ sin ( 20t + _____ ) + _______ sin ( 0.5t + ______ )

Figure 10.4 A phasor transform example (continued)

10.2 BODE PLOTS

In the previous section we calculated a table of gains and phase angles over a range
of frequencies. Graphs of these values are called Bode plots. These plots are normally
done on semi-log graph paper, such as that seen in Figure 10.5. Along the longer axis of
this paper the scale is logarithmic (base 10). This means that if the paper started at 0.1 on
one side, the next major division would be 1, then 10, then 100, and finally 1000 on the
other side of the paper. The basic nature of logarithmic scales prevents the frequency from
being zero. Along the linear axis (the short one) the gains and phase angles are plotted,
normally with two graphs side-by-side on a single sheet of paper.
 1 2 3 4 5 6 789 1 2 3 4 5 6 789 1 2 3 4 5 6 789 1 2 3 4 5 6 789
bode plots - 10.6
 bode plots - 10.7

Figure 10.5 4 cycle semi-log graph paper

Plot the points from Figure 10.3 in the semi-log graph paper in Figure 10.5. The gen-
eral layout is pictured below.

gain
(dB)

-40dB/dec

phase
(deg)

0.1 1 10 100 1000

f(Hz)

Figure 10.6 Drill problem: Plot the points from Figure 10.3 on graph paper
 bode plots - 10.8

steps_per_dec = 6;
decades = 6;
start_freq = 0.1;

// the transfer function

function foo=G(w)
D = %i * w;
foo = (D + 5) / (D^2 + 100*D + 10000);
endfunction

// this section writes the values to a datafile that may be graphed in a spreadsheet
fd = mopen("data.txt", "w");
for step = 0:(steps_per_dec * decades),
f = start_freq * 10 ^ (step / steps_per_dec); // calculate the next frequency
w = f * 2 * %pi); // convert the frequency to radians
[gain, phase] = polar(G(w));// find the gain and convert it to mag and angle
gaindb = 20 * log10(gain);// convert magnitude to dB
phasedeg = 180 * phase / %pi;// convert to degrees
mfprint(fd, "%f, %f, %f \n", f, gaindb, phasedeg);
end
mclose(fd);

// to graph it directly the following is used

D = poly(0, ’D’);
h = syslin(’c’, (D + 5) / (D^2 + 100*D + 10000) ) ;
bode(h, 0.1, 1000, ’Sample Transfer Function’);

Figure 10.7 Aside: Bode Plot Example with Scilab

Use computer software, such as Mathcad or a spreadsheet, to calculate the points in

Figure 10.2, and then draw Bode plots. Most software will offer options for making
one axis use a log10 scale.

Figure 10.8 Drill problem: Plot the points from Figure 10.2 with a computer
 bode plots - 10.9

Draw the Bode plot for the transfer function by hand or D+3
-----------------------------------------------------
-
with computer. 2
D + 10000D + 10000

Figure 10.9 Drill problem: Draw the Bode plot for gain and phase
 bode plots - 10.10

An approximate technique for constructing a gain Bode plot is shown in Figure

10.10. This method involves looking at the transfer function and reducing it to roots in the
numerator and denominator. Once in that form, a straight line approximation for each term
can be drawn on the graph. An initial gain is also calculated to shift the results up or down.
When done, the straight line segments are added to produce a more complex straight line
curve. A smooth curve is then drawn over top of this curve.

Bode plots for transfer functions can be approximated with the following steps.
1. Plot the straight line pieces.
a) The gain at 0rad/sec is calculated and used to find an initial off-
set. For example this transfer function starts at 10(D+1)/
(D+1000)=10(0+1)/(0+1000)=0.01=-40dB.
b) Put the transfer function in root form to identify corner frequen-
cies. For example (D+1)/(D+1000) will have corner frequencies
at 1 and 1000 rad/sec.
c) Curves that turn up or down are drawn for each corner fre-
quency. At each corner frequency a numerator term causes the
graph to turn up, each term in the dominator causes the graph to
turn down. The slope up or down is generally +/- 20dB/decade
for each term. Also note that squared (second-order) terms would
have a slope of +/-40dB/decade.
2. The effect of each term is added up to give the resulting straight line
approximation.
3. When the smooth curve is drawn, there should commonly be a 3dB dif-
ference at the corner frequencies. In second-order systems the damping
coefficient make may the corner flatter or peaked.

Figure 10.10 The method for Bode graph straight line gain approximation
 bode plots - 10.11

Note: Some of the straight line approximation issues are discussed below.

Why is there 3dB between a first order corner and the smooth plot, and the phase
angle is 45 degrees of the way to +/- 90 degrees.
1
G ( jω ) = ------------------
ω c + jω
the initial gain is
1 1-
G ( 0 ) = ----------------- = -----
ω c + j0 ωc
at the corner frequency ωc = ω
1 1 1 π
G ( jω ) = --------------------- = ---------------------- = ------------- ∠– ---
ω c + jω c π ωc 2 4
ω c 2 ∠---
4
Therefore the difference is
G ( jω ) π π π
diff = --------------- = ------- ∠– --- = 20 log  ------- ∠– --- = – 3.01dB ∠– ---
1 1
G(0) 2 4  2 4 4
Why does a first order pole go down at 20dB/dec?

1
G ( jω ) = ------------------
ω c + jω
before the corner frequency, ω c > jω
1- = 20 log ( ω –1 ) = – 20 log ( ω )
G ( jω ) = ----- c c
ωc
after the corner frequency, ω c < jω

1- = 20 log ( ω – 1 ) = – 20 log ( ω )
G ( jω ) = -----
jω
each time the frequency increases by a multiple of 10, the log value
becomes 1 larger, thus resulting in a gain change of -20 dB.

Figure 10.11 Why the straight line method works

 bode plots - 10.12

Why does a pole make the phase angle move by -90deg after the corner frequency?

1 ∠0 ω
G ( jω ) = ------------------ = --------------------------------------------------- = ----------------------- ∠– atan  ------
1 1
ω c + jω ω ω
ω c + ω ∠ atan  ------
2 2 2 2
ωc + ω c
 ω c
before the corner frequency, ω c > jω
ω
angle ( G ( jω ) ) = – atan  ------ = – atan  ------ = 0
0
ωc ωc
after the corner frequency, ω c < jω
ω π
angle ( G ( jω ) ) = – atan  ------ = – atan ( ∞ ) = – ---
 ω c 2

Figure 10.12 Why the straight line method works (cont’d)

 bode plots - 10.13

Show that the second order transfer function below would result in a slope of +/- 40
dB/decade.
1
G ( jω ) = -------------------------2-
( ω c + jω )

Figure 10.13 Drill problem: Slope of second order transfer functions.

An example of the straight line plotting technique is shown in Figure 10.14. In this
example the transfer function is first put into a root form. In total there are three roots, 1,
10 and 100 rad/sec. The single root in the numerator will cause the curve to start upward
with a slope of 20dB/dec after 1rad/sec. The two roots will cause two curves downwards
at -20dB/dec starting at 10 and 100 rad/sec. The initial gain of the transfer function is also
calculated, and converted to decibels. The frequency axis is rad/sec by default, but if Hz
are used then it is necessary to convert the values.
 bode plots - 10.14

4
100D + 100 10 ( D + 1 ) -
G ( D ) = --------------------------------------------------
- = --------------------------------------------- (the equation is put in root form)
2
0.01D + 0.11D + 10 ( D + 10 ) ( D + 100 )

Step 1: Draw lines for each of the terms in the transfer function,

Gain (dB)
D+1
4
10 ( 1 )
GAIN 0 = ------------------------- = 10 = 20dB
( 10 ) ( 100 )
20dB/dec.

1 10 100
0dB freq.(rad/sec)
(could be Hz also)
-20dB/dec. -20dB/dec.

1 - 1 -
--------------- ------------------
D + 10 D + 100

Figure 10.14 An approximate gain plot example

The example is continued in Figure 10.15 where the straight line segments are
added to produce a combined straight line curve.
 bode plots - 10.15

Step 2: Sum the individual lines, and get the straight line approximation,
Gain (dB)

40dB

20dB

1 10 100
0dB freq.(rad/sec)

Figure 10.15 An approximate gain plot example (continued)

Finally a smooth curve is fitted to the straight line approximation. When drawing
the curve imagine that there are rubber bands at the corners that pull slightly and smooth
out. For a simple first-order term there is a 3dB gap between the sharp corner and the func-
tion. Higher order functions will be discussed later.

Step 3: Draw the smooth curve (leaving 3dB at the corners),

Gain (dB)

40dB 3dB

20dB

1 10 100
0dB freq.(rad/sec)

Figure 10.16 An approximate gain plot example (continued)

 bode plots - 10.16

The process for constructing phase plots is similar to that of gain plots, as seen in
Figure 10.18. The transfer function is put into root form, and then straight line phase shifts
are drawn for each of the terms. Each term in the numerator will cause a positive shift of
90 degrees, while terms in the denominator cause negative shifts of 90 degrees. The phase
shift occurs over two decades, meaning that for a center frequency of 100, the shift would
start at 10 and end at 1000. If there are any lone ’D’ terms on the top or bottom, they will
each shift the initial value by 90 degrees, otherwise the phase should start at 0degrees.

Gain plots for transfer functions can be approximated with the following steps.
1. Plot the straight line segments.
a) Put the transfer function in root form to identify center frequen-
cies. For example (D+1)/(D+1000) will have center frequencies
at 1 and 1000 rad/sec. This should have already been done for
the gain plot.
b) The phase at 0rad/sec is determined by looking for any individ-
ual D terms. Effectively they have a root of 0rad/sec. Each of
these in the numerator will shift the starting phase angle up by
90deg. Each in the denominator will shift the start down by 90
deg. For example the transfer function 10(D+1)/(D+1000)
would start at 0 deg while 10D(D+1)/(D+1000) would start at
+90deg.
c) Curves that turn up or down are drawn around each center fre-
quency. Again terms in the numerator cause the curve to go up 90
deg, terms in the denominator cause the curves to go down 90
deg. Curves begin to shift one decade before the center fre-
quency, and finish one decade after.
2. The effect of each term is added up to give the resulting straight line
approximation.
3. The smooth curve is drawn.

Figure 10.17 The method for Bode graph straight line gain approximation

The previous example started in Figure 10.14 is continued in Figure 10.18 to

develop a phase plot using the approximate technique. There are three roots for the trans-
fer function. None of these are zero, so the phase plot starts at zero degrees. The root in the
numerator causes a shift of positive 90 deg, starting one decade before 1rad/sec and end-
ing one decade later. The two roots in the denominator cause a shift downward.
 bode plots - 10.17

4
10 ( D + 1 )
G ( D ) = ----------------------------------------------
( D + 10 ) ( D + 100 )

Step 1: Draw lines for each of the terms in the transfer function

Phase angle (deg.)

D+1
+90

+45
0.1 1 10 100 1000
0 freq.(rad/sec)

-45 1 -
1 - ------------------
--------------- D + 100
D + 10
-90

Figure 10.18 An approximate phase plot example

The straight line segments for the phase plot are added in Figure 10.19 to produce
a straight line approximation of the final plot. A smooth line approximation is drawn using
the straight line as a guide. Again, the concept of an rubber band will smooth the curve.
 bode plots - 10.18

Phase angle (deg.)

+90

+45
0.1 1 10 100 1000
0 freq.(rad/sec)

-45

-90

Phase angle (deg.)

+90

+45
0.1 1 10 100 1000
0 freq.(rad/sec)

-45

-90

Figure 10.19 An approximate phase plot example (continued)

The previous example used a transfer function with real roots. In a second-order
system with double real roots (overdamped) the curve can be drawn with two overlapping
straight line approximations. If the roots for the transfer function are complex (under-
damped the corner frequencies will become peaked. This can be handled by determining
the damping coefficient and natural frequency as shown in Figure 10.20. The peak will
occur at the damped frequency. The peaking effect will become more pronounced as the
damping coefficient goes from 0.707 to 0 where the peak will be infinite.
 bode plots - 10.19

Gain
Mn Resonant Peak
A
-------------------------------------------------------
-
=0.1 2 2
D + ( 2ξω n )D + ( ω n )

=0.5 2
dB = A ωd = 1 – ξ ωn

ξ = 0.707

-40dB/decade
freq
ωd

Note: If ζ is less than 1 the roots become complex, and the Bode plots get a
peak. This can be seen mathematically because the roots of the transfer
function become complex.

Figure 10.20 Resonant peaks

The approximate techniques do decrease the accuracy of the final solution, but
they can be calculated quickly. In addition these curves provide an understanding of the
system that makes design easier. For example, a designer will often describe a system with
a Bode plot, and then convert this to a desired transfer function.
 bode plots - 10.20

Draw the straight line approximation for the transfer D+3

-----------------------------------------------------
-
function. 2
D + 10000D + 10000

Figure 10.21 Drill problem: Draw the straight line approximation

 bode plots - 10.21

10.3 SIGNAL SPECTRUMS

If a vibration signal is measured and displayed it might look like Figure 10.22. The
overall sinusoidal shape is visible, along with a significant amount of ’noise’. When this is
considered in greater detail it can be described with the given function. To determine the
function other tools are needed to determine the frequencies, and magnitudes of the fre-
quency components.

x(t)
x ( t ) = A sin  ---------- t + 0.3A sin  ------------- t + 0.01A sin  ---------------- t
displacement 2π 2π 2π
0.01 0.005 0.0005
A

5 t (ms)

frequency 1

freq. 3
freq. 2

Figure 10.22 A vibration signal as a function of time

A signal spectrum displays signal magnitude as a function of frequency, instead of

time. The time based signal in Figure 10.22 is shown in the spectrum in Figure 10.23. The
three frequency components are clearly identifiable spikes. The height of the peaks indi-
cates the relative signal magnitude.
 bode plots - 10.22

Amplitude
A

log(freq.)
(Hz or rads.)

1- 1 - 1 -
--------- ------------ ---------------
0.01 0.005 0.0005
=100Hz =200Hz =2000Hz

Figure 10.23 The spectrum for the signal in Figure 10.22

10.4 SUMMARY

• Bode plots show gain and phase angle as a function of frequency.

• Bode plots can be constructed by calculating point or with straight line approxi-
mations.
• A signal spectrum shows the relative strengths of components at different fre-
quencies.

10.5 PRACTICE PROBLEMS

1. Draw a Bode Plot for both of the transfer functions below using the phasor transform.

( D + 1 ) ( D + 1000 -) 5
--------------------------------------------- AND ------
2 2
( D + 100 ) D

2. Given the transfer function below,

( D )-
y----------- ( D + 10 ) ( D + 5 -)
= ---------------------------------------
x(D ) (D + 5)
2

a) draw the straight line approximation of the Bode (gain and phase shift) plots.
b) determine the steady-state output if the input is x(t) = 20 sin(9t+0.3) using the
 bode plots - 10.23

Bode plot.

3. Use the straightline approximation techniques to draw the Bode plot for the transfer function
below.
F D + 1000 -
G = --- = -----------------------------------
2
x D + 5D + 100

4. Given the transfer function below,

V 1000 -
-----o- = ---------------------
Vi D + 1000

a) Find the steady state response of the circuit using phasors (i.e., phasor trans-
forms) if the input is Vi=5sin(100,000t).
b) Draw an approximate Bode plot for the circuit.

5. For the transfer function,

D ( D + 2π )
----------------------------------------------------------
2 2
D + 300πD + 62500π
a) Use the straight line method and the attached log paper to draw an approximate
Bode plot.
b) Verify the Bode plot by calculating values at a few points.
c) Use the Bode plot to find the response to an input of 5sin(624t) + 1sin(6.2t).

6. The applied force ‘F’ is the input to the system, and the output is the displacement ‘x’. Neglect
the effects of gravity.
a) find the transfer function.
K1 = 500 N/m

x K2 = 1000 N/m

M = 10kg

b) What is the steady-state response for an applied force F(t) = 10cos(t + 1) N ?

c) Give the transfer function if ‘x’ is the input.
d) Draw the bode plots for the transfer function found in a).
e) Find x(t), given F(t) = 10N for t >= 0 seconds.
 bode plots - 10.24

f) Find x(t), given F(t) = 10N for t >= 0 seconds considering the effects of gravity.

7. The following differential equation is supplied, with initial conditions.

·· · ·
y + y + 7y = F y( 0) = 1 y(0) = 0

F ( t ) = 10 t>0
a) Write the equation in state variable form.
b) Solve the differential equation numerically.
c) Solve the differential equation using calculus techniques.
d) Find the frequency response (gain and phase) for the transfer function using the
phasor transform. Sketch the bode plots.

8. You are given the following differential equation for a spring damper pair.

5x +  ----- x = F F ( t ) = 10 sin ( 100t )

d
 dt
a) Write the transfer function for the differential equation if the input is F.
b) Apply the phasor transform to the transfer function to find magnitude and phase
as functions of frequency.
c) Draw a Bode plot for the system using either approximate or exact techniques.
d) Use the Bode plot to find the response to;
F ( t ) = 10 sin ( 100t )

e) Put the differential equation in state variable form and use a calculator to find
values in time for the given input.
F = 10 sin ( 100t ) t x

0.0
0.002
0.004
0.006
0.008
0.010
f) Give the expected ‘x’ response of this first-order system to a step function input
for force F = 1N for t > 0 if the system starts at rest. Hint: Use the canonical
form.
 bode plots - 10.25

10.6 PRACTICE PROBLEM SOLUTIONS

1.

20dB

0dB
-20dB 0dB -40dB/dec

1- 100 1000
----- --------- ------------
2π 2π 2π ------5-
2π

90deg 0deg

-45deg -180deg
 bode plots - 10.26

2.

( D )-
y----------- ( D + 10 ) ( D + 5 -) ( D + 10 )
= --------------------------------------- = ---------------------
x(D ) (D + 5)
2 (D + 5)

6dB
0dB
0.8 1.6
90

0.8 1.6

-90
x ( t ) = 20 cos ( 9t + 0.3 )

Aside: the numbers should be obtained from the graphs, but I have calculated them

( D + 10 )
= --------------------- =  -----------------  -------------- = -------------------------------- = 1.236 – 0.425j
y-- 9j + 10 5 – 9j 50 + 81 – 45j
x (D + 5)  9j + 5 5 – 9j    25 + 81

1.236 + 0.425 ∠ atan  ---------------- = 1.307 ∠– 0.3312rad

y-- 2 2 – 0.425
=
x 1.236
y--
= 2.33dB ∠– 9.49°
x

y ( t ) = 20 ( 1.307 ) sin ( 9t + 0.3 + ( – 0.3312 ) )

y ( t ) = 26.1 sin ( 9t – 0.031 )

Aside: This can also be done entirely with phasors in cartesian notation

( D + 10 )
= --------------------- =  -----------------  -------------- = -------------------------------- = 1.236 – 0.425j
y-- 9j + 10 5 – 9j 50 + 81 – 45j
x (D + 5)  9j + 5   5 – 9j 25 + 81
x = 20 ( cos ( 0.3rad ) + j sin ( 0.3rad ) ) = 19.1 + 5.91j
y - = ( 1.236 – 0.425j ) ( 19.1 + 5.91j ) = 26.1 – 0.813j = 26.1 ∠– 0.031
----------------------------
19.1 + 5.91j

y ( t ) = 26.11 sin ( 9t – 0.031 )

 bode plots - 10.27

(cont’d

Aside: This can also be done entirely with phasors in polar notation
( D + 10 ) 13.45 ∠0.733
= --------------------- =  ----------------- = ------------------------------- = ------------- ∠0.733 – 1.064 = 1.31 ∠– 0.331
y-- 9j + 10 13.45
x (D + 5) 9j + 5 10.30 ∠1.064 10.30
x = 20 ∠0.3
y -
----------------- = 1.31 ∠– 0.331
20 ∠0.3
y = 1.31 ( 20 ) ∠( – 0.331 + 0.3 ) = 26.2 ∠– 0.031
y ( t ) = 26.2 sin ( 9t – 0.031 )
 bode plots - 10.28

3.
for the numerator, (zero) rad
1000 --------- = 159Hz
s
for the denominator, (poles) 2 2 2
D + 2ω n ζD + ω n = D + 5D + 100
2
ω n = 100
rad
∴ω n = 10 --------- f n = 1.59Hz
s
2ω n ζ = 5
5
∴ζ = -------------- = 0.25 (underdamped)
2 ( 10 )
2
f d = f n 1 – ζ = 1.54Hz

for the initial gain 0 + 1000 -

G ( 0 ) = -------------------------------------
2
= 10 = 20dB
0 + 5 ( 0 ) + 100
Gain
(dB)
20
159 f(Hz)
0
0.1 1 1.54 10 100 1000

-40dB/dec

-60
-20dB/dec
-80
phase
(deg)
0 0.154 15.4,15.9 1590 f(Hz)
0.1 1 1.54 10 100 159 1000

-90

-180
 bode plots - 10.29

4.
a) 5
V o ( t ) = 0.050 sin ( 10 t – 1.561 )

b) 16Hz 159Hz 1.6KHz

-20dB/dec

0 deg

-90 deg

5.

a)

b)
–3
c) 8.89 sin ( 624t + 1.571 ) + 0.141 ×10 sin ( 6.2t + 2.356 )
 bode plots - 10.30

6.

--x- = ----------------------
a) 0.0667 - m ----
F 2
D + 33.3 N
b) ω = 1 ∴D = 1j

--x- = ----------------------------
0.0667 - –3 m
2
= 2.07 × 10 ∠0rad ----
F ( 1j ) + 33.3 N

F ( t ) = 10 cos ( t + 1 ) N
∴F ( ω ) = ( 10 ∠1rad )N
x - = 2.07 × 10 –3 ∠0rad m
------------------------------- ----
( 10 ∠1rad )N N
–3 m
∴x ( ω ) = 2.07 × 10 ∠0rad ---- ( 10 ∠1rad )N
–3
N
∴x ( ω ) = ( 10 )2.07 × 10 ∠( 0rad + 1rad )m
∴x ( ω ) = 0.0207 ∠1radm
∴x ( t ) = 0.0207 cos ( t + 1 ) m

2
F
--- = D
c) + 33.3- --- N-
----------------------
x 0.0667 m
 bode plots - 10.31

2 2 2
d) for the denominator, (poles) D + 2ω n ζD + ω n = D + 33.3

rad
ω n = 5.77 --------- f n = 0.918Hz
s
ζ = 0 (undamped)
2
f d = f n 1 – ζ = 0.918Hz

0.0667
for the initial gain G ( 0 ) = ---------------- = 2 = –54 dB
33.3

Gain
∞
(dB)
-54
f(Hz)
0
0.1 1 10 100
-94
-40dB/dec

-134
-20dB/dec

phase
(deg)
f(Hz)
0
0.1 1 10 100

-90

-180

e) x ( t ) = ( – 0.020 cos ( 5.77t ) + 0.020 )m

f) x ( t ) = ( – 0.2162 cos ( 5.77t ) + 0.2162 )m

 bode plots - 10.32

7.
·
a) y = v
·
v = – v – 7y + F

b)
yi + 1 yi 0 1 yi + 0
= +h
vi + 1 vi – 7 –1 v i F

given
y0
= 1 F = 10
v0 0

using h=0.001s

y1
= 1 + 0.001 0 1 1 + 0
v1 0 –7 –1 0 10

y 100 1.428m
etc.. until
= m
v 100 0.000 ----
s

c) – 0.5t
y ( t ) = ( – 0.437e cos ( 2.598t – 0.19 ) + 1.429 )m
 bode plots - 10.33

--y- = --------------------------
d) 1 1 1
2
= ----------------------------------
2
- = -------------------------------------
2
F D +D+7 ( jω ) + jω + 7 (7 – ω ) + j(ω)
1
--y- = ----------------------------------------
F 2 2 2
(7 – ω ) + ω
∠0 - = ∠( 0 – angle ( ω, 7 – ω 2 ) ) = ∠– angle ( ω, 7 – ω 2 )
∠θ = ----------------------------------------------
2
∠angle ( ω, 7 – ω )

-20dB

-40dB
0.01Hz 1Hz 100Hz

0deg
0.01Hz 1Hz 100Hz

-180deg
 bode plots - 10.34

8.
--x- = ------------
a) 1 -
F 5+D
1 ∠0 ω
= --------------- = ------------------------------------------------ = ---------------------- ∠– atan  ----
b)
--x- = ------------
1 - 1 1
F 5+D 5 + jω ω 5
5 + ω ∠ atan  ----
2 2
2 2
5 +ω
5
c)
initial gain = 20 log  ------------ = – 14dB
gain(dB) 1
5+0
-14dB 5
corner freq. = ------ = 0.8Hz
2π
-20dB/dec

phase(deg)
0°

– 45°

– 90°

0.08Hz 0.8Hz 8Hz

100
d) f = --------- = 16Hz
2π
From the Bode plot, gain = – 40dB = 0.01
π
phase = – 90° = – --- rad
2
π
x ( t ) = 10 ( 0.01 ) sin  100t – ---

2

Aside: verified by calculations,

= --------------------------- ∠– atan  --------- = 0.00999 ∠– 1.521
x 1 100
-------------
10 ∠0 2 2  5 
5 + 100
x = ( 10 ∠0 ) ( 0.00999 ∠– 1.521 ) = 0.0999 ∠–1.521
x ( t ) = 0.0999 sin ( 100t – 1.521 )
 bode plots - 10.35

t x
e) 0.2
0.186

0.0 0

Xi 0
0.002 9.98e-4

0.004 5.92e-3
− 0.099 0.2
0 0.5 1
0 i ⋅h 0.999 0.006 0.015

0.008 0.026

– 5t 0.010 0.041
1 e
f) x = --- – ---------
5 5

10.7 ASSIGNMENT PROBLEMS

1. For the transfer functions below, draw the bode plots using computer software.

( D + 2π )D ( D + 2π )
---------------------------------------------------------
- -------------------------------------------------------
( D + 200π ) ( D + 0.02π ) 2
D + 50πD + 10000π
2

2. Draw Bode plots for the following functions using straight line approximations.
1 - 1 - 1 1
------------ --------------- --------------------- -----------------------------
-
D+1 2 2 2
D +1 (D + 1) D + 2D + 2

3. Given the transfer function below,

( D )-
y----------- ( D + 10 ) ( D + 5 )
= ---------------------------------------
-
x(D ) (D + 5)
2

a) draw the straight line approximation of the bode and phase shift plots.
b) determine the steady state output if the input is x(s) = 20 cos(9t+.3) using the
striaght line plots.
c) use an exact method to verify part b).

4. a) Convert the following differential equation to a transfer function.

··
5x + 2x = 3F
 bode plots - 10.36

b) Apply a phasor (Fourier) transform to the differential equation and develop

equations for the system gain and phase shift as a function of input frequency.
c) Draw a Bode plot using the equations found in part b) on the attached log paper.
d) Draw a straight line approximation of the system transfer function on the
attached log paper.

10.8 LOG SCALE GRAPH PAPER

Please notice that there are a few sheets of 2 and 4 cycle log paper attached, make
additional copies if required, and if more cycles are required, sheets can be cut and pasted
together. Also note that better semi-log paper can be purchased at technical bookstores, as
well at most large office supply stores.
 bode plots - 10.37

1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9
 1 2 3 4 5 6 789 1 2 3 4 5 6 789 1 2 3 4 5 6 789 1 2 3 4 5 6 789
bode plots - 10.38
 root locus - 11.1

11. ROOT LOCUS ANALYSIS

Topics:
• Root-locus plots

Objectives:
• To be able to predict and control system stability.

11.1 INTRODUCTION

The system can also be checked for general stability when controller parameters
are varied using root-locus plots.

11.2 ROOT-LOCUS ANALYSIS

In a engineered system we may typically have one or more design parameters,

adjustments, or user settings. It is important to determine if any of these will make the sys-
tem unstable. This is generally undesirable and possibly unsafe. For example, think of a
washing machine that vibrates so much that it ‘walks’ across a floor, or a high speed air-
craft that fails due to resonant vibrations. Root-locus plots are used to plot the system roots
over the range of a variable to determine if the system will become unstable, or oscillate.

Recall the general solution to a homogeneous differential equation. Complex roots

will result in a sinusoidal oscillation. If the roots are real the result will be e-to-the-t terms.
If the real roots are negative then the terms will tend to decay to zero and be stable, while
positive roots will result in terms that grow exponentially and become unstable. Consider
the roots of a second-order homogeneous differential equation, as shown in Figure 11.1 to
Figure 11.7. These roots are shown on the complex planes on the left, and a time response
is shown to the right. Notice that in these figures (negative real) roots on the left hand side
of the complex plane cause the response to decrease while roots on the right hand side
cause it to increase. The rule is that any roots on the right hand side of the plane make a
system unstable. Also note that the complex roots cause some amount of oscillation.
 root locus - 11.2

R = – A, – B
jw x(t)

x0
-A -B
sigma

Figure 11.1 Negative real roots make a system stable

R = ± Aj
jw x(t)
A

sigma
-A
t

Figure 11.2 Complex roots make a system oscillate

R = – A ± Bj
jw x(t)
B

-A
sigma

-B t

Figure 11.3 Negative real and complex roots cause decaying oscillation
 root locus - 11.3

R = – A ± Bj
jw x(t)
B

-A
sigma

-B t

Figure 11.4 More negative real and complex roots cause a faster decaying oscillation

R = – A, – A
jw x(t)

-A,-A
sigma

Figure 11.5 Overlapped roots are possible

R = A, A
jw x(t)

A,A
sigma

Figure 11.6 Positive real roots cause exponential growth and are unstable
 root locus - 11.4

R = A ± Bj
jw x(t)
B

A sigma

-B t

Figure 11.7 Complex roots with positive real parts have growing oscillations and are
unstable

Next, recall that the denominator of a transfer function is the homogeneous equa-
tion. By analyzing the function in the denominator of a transfer function the general sys-
tem response can be found. An example of root-locus analysis for a mass-spring-damper
system is given in Figure 11.8. In this example the transfer function is found and the roots
of the equation are written with the quadratic equation. At this point there are three
unspecified values that can be manipulated to change the roots. The mass and damper val-
ues are fixed, and the spring value will be varied. The range of values for the spring coef-
ficient should be determined by practical and design limitations. For example, the spring
coefficient should not be zero or negative.
 root locus - 11.5

x ( D )- 1
------------ = -----------------------------------------
-
F(D) 2
MD + K d D + K s

Note: We want Ks Kd
the form below
A
--------------------------------------
- M
(D + B)(D + C)
1-
----
x ( D ) M
------------- = -------------------------------------
F(D) 2 Kd Ks Aside:
D + ------ D + -----
M M 2
ax + bx + c = ( x + A ) ( x + B )
2
–Kd K K
--------- ± ------d- – 4 -----s – b ± b – 4ac
2
M M
2 M A, B = --------------------------------------
B, C = ------------------------------------------
- 2a
2

M Kd Ks B C

0 0 0 0 0
100 100 100
100 100 1000
100 100 10000

Figure 11.8 A mass-spring-damper system equation

The roots of the equation can then be plotted to provide a root locus diagram.
These will show how the values of the roots change as the design parameter is varied. If
any of these roots pass into the right hand plane we will know that the system is unstable.
In addition complex roots will indicate oscillation.
 root locus - 11.6

Imaginary

Real

Figure 11.9 Drill problem: Plot the calculated roots on the axes above

A feedback controller with a variable control function gain is shown in Figure

11.10. The variable gain ’K’ necessitates the evaluation of controller stability over the
range of operating values. This analysis begins by developing a transfer function for the
overall system. The root of the denominator is then calculated and plotted for a range of
’K’ values. In this case all of the roots are on the left side of the plane, so the system is sta-
ble and doesn’t oscillate. Keep in mind that gain values near zero put the control system
close to the right hand plane. In real terms this will mean that the controller becomes unre-
sponsive, and the system can go where it pleases. It would be advisable to keep the system
gain greater than zero to avoid this region.
 root locus - 11.7

Note: This controller has adjustable gain. After this

design is built we must anticipate that all values of K
will be used. It is our responsibility to make sure that
none of the possible K values will lead to instability.

+ 1-
K ---
D
-

K H(D ) = 1
G ( D ) = ----
D
First, we must develop a transfer function for the entire control system.
 ---
K-
G( D)  D K
G S ( D ) = ------------------------------------- = --------------------------- = --------------
1 + G ( D )H ( D ) D+K
1 +  ---- ( 1 )
K
D
Next, we use the characteristic equation of the denominator to find the roots as
the value of K varies. These can then be plotted on a complex plane. Note:
the value of gain ’K’ is normally found from 0 to +infinity.

D+K = 0 K root
jω
0 -0
1 -1 K→∞ K = 0 σ
2 -2
3 -3
etc...

Note: This system will always be stable because all of the roots for all values of K are
negative real, and it will always have a damped response. Also, larger values of K,
make the system more stable.

Figure 11.10 Root-locus analysis in controller design

 root locus - 11.8

Aside: Scilab can be used to draw root locus plots for systems of the form below, where
there is a simple gain, K, multiplying the openloop gain, G(s).

+ 1-
K ---
D
-

1 H(D ) = 1
G ( D ) = ----
D
First, we must multiply G and H

G ( s )H ( s ) =  ---- ( 1 ) = ----
1 1
 D D
The numerator and denominator of this equation are then defined and plotted
using the ’evans’ function.
D = poly(0, ’D’); // define the differential operator
n = real(1.0); // define the numerator of GH
d = real(D); // define the denominator of GH
evans(n, d, 100); // plot for gains from K=0 to 100

Figure 11.11 Root-locus plotting in Scilab

 root locus - 11.9

Given the system elements (assume a negative feedback controller),

K H(D) = 1
G ( D ) = -----------------------------
2
-
D + 3D + 2
First, find the characteristic equation,. and an equation for the roots,

1 +  -----------------------------
- ( 1 ) = 0
K Note: For a negative feedback
2  controller the denominator is,
D + 3D + 2
2 1 + G ( D )H ( D )
D + 3D + 2 + K = 0

– 3 ± 9 – 4(2 + K) 1 – 4K
roots = ------------------------------------------------ = – 1.5 ± --------------------
2 2
Next, find values for the roots and plot the values,
K roots
jω
0
1
2
3

Figure 11.12 Drill problem: Complete the root-locus analysis

 root locus - 11.10

K(D + 5 )
G ( D )H ( D ) = ---------------------------------------
2
D ( D + 4D + 8 )

Figure 11.13 Drill problem: Draw a root locus plot

11.3 SUMMARY

• Root-locus plots show the roots of a transfer function denominator to determine

stability
 root locus - 11.11

11.4 PRACTICE PROBLEMS

1. Draw the root locus diagram for the system below. specify all points and values.
+ + 1
3.0 ---------------------
2
(D + 1)
- -

KdD

2. The block diagram below is for a motor position control system. The system has a proportional
controller with a variable gain K.

θd Vd + Ve Vs ω θa
2 K 100 - 1-
------------ ---
D+2 D
-
Va

a) Simplify the block diagram to a single transfer function.

b) Draw the Root-Locus diagram for the system (as K varies). Use either the
approximate or exact techniques.
c) Select a K value that will result in an overall damping coefficient of 1. State if
the Root-Locus diagram shows that the system is stable for the chosen K.

3. Given the system transfer function below.

θo 20K
----- = ---------------------------------
-
θd 2
D + D + 20K

a) Draw the root locus diagram and state what values of K are acceptable.
b) Select a gain value for K that has either a damping factor of 0.707 or a natural
frequency of 3 rad/sec.
c) Given a gain of K=10 find the steady-state response to an input step of 1 rad.
d) Given a gain of K=0.01 find the response of the system to an input step of
0.1rad.
 root locus - 11.12

4. A feedback control system is shown below. The system incorporates a PID controller. The
closed loop transfer function is given.
X + Y
Ki 3 -
K p + ----- + K d D ------------
D D+9
-

4 2
D ( 3K d ) + D ( 3K p ) + ( 3K i )
--Y- = ----------------------------------------------------------------------------------------------
-
X 2
D ( 12K d + 1 ) + D ( 9 + 12K p ) + ( 12K i )

a) Verify the close loop controller function given.

b) Draw a root locus plot for the controller if Kp=1 and Ki=1. Identify any values
of Kd that would leave the system unstable.
c) Draw a Bode plot for the feedback system if Kd=Kp=Ki=1.
d) Select controller values that will result in a natural frequency of 2 rad/sec and
damping coefficient of 0.5. Verify that the controller will be stable.
e) For the parameters found in the last step can the initial values be found?
f) If the values of Kd=1 and Ki=Kp=0, find the response to a unit ramp input as a
function of time.

5. Draw a root locus plot for the control system below and determine acceptable values of K,
including critical points.

X 0.1 Y
K  5 + ---- + D
+ 1 + --------------------------------------
-
2
D D + 10D + 100
- -

10
------
D
0.01

6. The feedback loop below is for controlling a DC motor with a PID controller.

Vd + e 2 Vs 100 -
------------------ ω 1-
--- θ
D + PD + 1 D + 100 D
------------------------------
D
-

Va
2
 root locus - 11.13

a) Find the transfer function for the system.

b) Draw a root locus diagram for the variable parameter ‘P’.
c) Find the response of the system in to a unit step input using explicit integration.
 root locus - 11.14

11.5 PRACTICE PROBLEM SOLUTIONS

1.

+ + 1 -
3.0 --------------------
2
(D + 1)
- -
KdD

+ 3.0
-------------------------------------
-
2
( D + 1 ) + Kd D
-

3.0
---------------------------------------------------
-
2
D + D ( K d + 2 ) + 4.0
2
2 – K d – 2 ± ( K d + 2 ) – 4 ( 4.0 )
D + D ( K d + 2 ) + 4.0 = 0 D = ---------------------------------------------------------------------------
2
2
Kd roots – K d – 2 ± K d + 4K d – 12
D = -----------------------------------------------------------------
2
0 -1 +/- 1.732j
1 -1.5 +/- 1.323j Critical points: (this is simple for a quadratic)
2 -2.000, -2.000
The roots becomes positive when
5 -0.628, -6.372
2
10 -0.343, -11.657 0 > – K d – 2 ± K d + 4K d – 12
100 -0.039, -102.0 2
1000 -0.004, -1000 2 + K d > ± K d + 4K d – 12
16 > 0
0 > – Kd – 2 Kd > –2

The roots becomes complex when

2
0 > K d + 4K d – 12

– 4 ± 16 – 4 ( – 12 ) K d = – 6, 2
K d = -----------------------------------------------
2

Gains larger than -2 will result in a stable system. Any gains

between -4 and -2 will result in oscillations.
 root locus - 11.15

2.
a) 200K
---------------------------------------
-
2
D + 2D + 200K

b) – 2 ± 4 – 4 ( 200K )
roots = ------------------------------------------------ = – 1 ± 1 – 200K
2
Im
K roots

0 0,-2 K=0.005
0.001 -0.1,-1.9
0.005 -1,-1 Re
0.1 etc. -2 -1
1
5
10

c) 2
D + 2D + 200K = D + 2ζω n D + ω n
2 2 ∴ω n = 1 ∴K = 0.005

From the root locus graph this value is critically stable.

 root locus - 11.16

3.
a) 2
D + D + 20K = 0
– 1 ± 1 – 4 ( 20K ) For complex roots
D = --------------------------------------------- 1
2 1 – 80K < 0 K > ------
80
K roots For negative real roots (stable)
1 ± 1 – 80K-
–------------------------------------
0 0.000, -1.000 <0
2
1/80 -0.500, -0.500
1 -0.5 +/- 4.444j ± 1 – 80K < 1 K>0
10 -0.5 +/- 14.13j
1000 -0.5 +/- 141.4j

b)
Matching the second order forms,
2
2ω n ξ = 1 ω n = 20K

The gain can only be used for the natural frequency

20 20
K = -----2- = -----2- = 2.22
ωn 3
 root locus - 11.17

θo 20 ( 10 )
c) ----- = ----------------------------------------
-
θd 2
D + D + 20 ( 10 )
·· ·
θ o + θd + θ d 200 = 200θd

Homogeneous:
2
A + A + 200 = 0

– 1 ± 1 – 4 ( 200 ) A = – 0.5 ± 14.1j

A = --------------------------------------------
2
– 0.5t
θo ( t ) = C1 e sin ( 14.1t + C 2 )

Particular:
θ = A
0 + 0 + A200 = 200 ( 1rad ) A = 1rad
θ o ( t ) = 1rad

Initial Conditions (assume at rest):

– 0.5t
θo ( t ) = C1 e sin ( 14.1t + C 2 ) + 1rad

θ o ( 0 ) = C 1 ( 1 ) sin ( 14.1 ( 0 ) + C 2 ) + 1rad = 0

C 1 sin ( C 2 ) = – 1rad (1)
– 0.5t – 0.5t
θ' o ( t ) = – 0.5C 1 e sin ( 14.1t + C 2 ) – 14.1C 1 e cos ( 14.1t + C 2 )
0 = – 0.5C 1 sin ( C 2 ) – 14.1C 1 cos ( C 2 )
14.1 cos ( C 2 ) = – 0.5 sin ( C 2 )
14.1-
--------- = tan ( C 2 ) C 2 = – 1.54
– 0.5
– 1rad – 1rad
C 1 = ------------------- = -------------------------- = 1.000rad
sin ( C 2 ) sin ( – 1.54 )

– 0.5t
θo ( t ) = ( e sin ( 14.1t – 1.54 ) + 1 ) ( rad )
 root locus - 11.18

θo 20 ( 0.01 )
d) ----- = --------------------------------------------
-
θd 2
D + D + 20 ( 0.01 )
·· ·
θ o + θd + θ d 0.2 = 0.2θd

Homogeneous:
2
A + A + 0.2 = 0

– 1 ± 1 – 4 ( 0.2 ) A = – 0.7236068, – 0.2763932

A = ------------------------------------------
2
– 0.724t – 0.276t
θo ( t ) = C1 e + C2 e

Particular:
θ = A
0 + 0 + A0.2 = 0.2 ( 1rad ) A = 1rad
θ o ( t ) = 1rad

Initial Conditions (assume at rest):

– 0.724t – 0.276t
θo ( t ) = C1 e + C2e + 1rad
– 0.724t – 0.276t
θo ( 0 ) = C1 e + C2 e + 1rad = 0
C 1 + C 2 = – 1rad (1)
– 0.724t – 0.276t
θ' o ( t ) = – 0.724 ( C 1 e ) – 0.276 ( C 2 e )

C 1 = – 0.381C 2

– 0.381C 2 + C 2 = – 1rad C 2 = – 1.616rad

C 1 = – 0.381 ( – 1.616rad ) = 0.616rad

– 0.724t – 0.276t
θ o ( t ) = ( 0.616 )e + ( – 1.616 )e + 1rad
 root locus - 11.19

4.
(ans. X + Kp D + K i + Kd D
2 Y
3 -
------------------------------------------- ------------
D D+9
-

X + 2 Y
3K p D + 3K i + 3K d D
-----------------------------------------------------
- D(D + 9)

X 2 Y
3K p D + 3K i + 3K d D
-------------------------------------------------------------------------------------------
2
D ( D + 9 ) + 12K p D + 12K i + 12K d D

X 2 Y
3K p D + 3K i + 3K d D
----------------------------------------------------------------------------------------------
-
2
D ( 12K d + 1 ) + D ( 9 + 12K p ) + ( 12K i )
 root locus - 11.20

b) 2
D ( 12K d + 1 ) + D ( 9 + 12K p ) + ( 12K i ) = 0
2
– 9 – 12K p ± ( 9 + 12K p ) – 4 ( 12K d + 1 )12K i
D = ------------------------------------------------------------------------------------------------------------------
-
2 ( 12K d + 1 )

Kd roots

-100 -0.092, 0.109

-10 -0.241, 0.418
-1 -0.46, 2.369
-0.1 -0.57, 105.6
0 -0.588, -20.41
1 -0.808 +/- 0.52j
10 -0.087 +/- 0.303j
100 -0.0087 +/- 0.1j

Stable for, 2
– 9 – 12K p ± ( 9 + 12K p ) – 4 ( 12K d + 1 )12K i < 0

2
± ( 9 + 12K p ) – 4 ( 12K d + 1 )12K i < 9 + 12K p
2 2
( 9 + 12K p ) – 4 ( 12K d + 1 )12K i < ( 9 + 12K p )

– 4 ( 12K d + 1 )12K i < 0

–1
K d > ------
12

Becomes complex at,

2
0 > ( 9 + 12K p ) – 4 ( 12K d + 1 )12K i
2
576K d K i > ( 9 + 12K p ) – 48K i
2
( 9 + 12K p ) – 48K i
K d > ----------------------------------------------- K d > 0.682
576K d K i
 root locus - 11.21

c) Kp = 1 Ki = 1 Kd = 1
2
3K p D + 3K i + 3K d D
--Y- = ----------------------------------------------------------------------------------------------
-
X 2
D ( 12K d + 1 ) + D ( 9 + 12K p ) + ( 12K i )

3-  -------------------------------------------------- 
2 2
--Y- = -----------------------------------------
3D + 3D + 3 -
=  ----- 
D +D+1 -
 13  2
D + D1.615 + 0.923
X 2
D 13 + D21 + 12

final gain = 20 log  ------ = – 12.7

3
 13
initial gain = 20 log  ------ = – 12.0
3
 12

for the numerator,

1
ωn = 1 = 1 ξ = --------- = 0.5
2ω n
2 2
ωd = ωn 1 – ξ = 1 – 0.5 = 0.866
for the denominator,
1.615
ωn = 0.923 = 0.961 ξ = ------------- = 0.840
2ω n
2 2
ω d = ω n 1 – ξ = 0.961 1 – 0.840 = 0.521

-12dB
 root locus - 11.22

2
3K p D + 3K i + 3K d D
--Y- = ----------------------------------------------------------------------------------------------
-
X 2
D ( 12K d + 1 ) + D ( 9 + 12K p ) + ( 12K i )
12K i
ωn = ---------------------
- = 2 12K i = 48K d + 4
12K d + 1

9 + 12K p
2ξω n = ---------------------- = 20.5 ( 2 ) 24K d = 7 + 12K p
12K d + 1

At this point there are two equations and two unknowns, one value
must be selected to continue, therefore,

K p = 10

24K d = 7 + 12K p = 7 + 12 ( 10 ) = 127 K d = 5.292

12K i = 48K d + 4 = 48 ( 5.292 ) + 4 = 258.0 K i = 21.5

Now to check for stability

2
D ( 12 ( 5.292 ) + 1 ) + D ( 9 + 12 ( 10 ) ) + ( 12 ( 21.5 ) ) = 0
2
64.504D + 129D + 258 = 0
2
– 129 ± 129 – 4 ( 64.5 )258
D = --------------------------------------------------------------------- = – 1 ± 1.73j
2 ( 64.5 )
 root locus - 11.23

e) Cannot be found without an assumed input and initial conditions

f) 2
3 ( 0 )D + 3 ( 0 ) + 3 ( 1 )D
--Y- = ---------------------------------------------------------------------------------------------------
-
X 2
D ( 12 ( 1 ) + 1 ) + D ( 9 + 12 ( 0 ) ) + ( 12 ( 0 ) )
2
--Y- = --------------------------
3D -
X 2
13D + 9D
2 2
Y ( 13D + 9D ) = X ( 3D )
·· · ·· · ··
Y 13 + Y 9 = X 3 X = t X = 1 X = 0
·· · 9
Y + Y ------ = 0
13
It is a first order system,
9
– ------ t
13
Y ( t ) = C1 e + C2

Y( 0) = 0 Y' ( 0 ) = 0 starts at rest/undeflected

0 = C11 + C2 C1 = –C2
9
– t ------
9- 13
Y' ( t ) = – C e
-----
13 1

9 C1 = 0
0 = – ------ C 1
13 1
C2 = 0 no response
 root locus - 11.24

5.

X 0.1 Y
K  5 + ---- + D
+ 1 + --------------------------------------
-
 D  2
D + 10D + 100
- -

10
------
D
0.01

X 2 0.1 Y
K  ------------------------------
+ 5D + 1 + D ----------------------------------------------------------------
-
 
D + 10D + 100 + 0.1  ------
D 2 10
 D
-

0.01

2
X + K ( 0.5D + 0.1 + 0.1D -) Y
-------------------------------------------------------
3 2
D + 10D + 100D + 1
-

0.01

X K ( 0.5D + 0.1 + 0.1D )

2 Y
-------------------------------------------------------------------------------------------------------------------------------------
-
3 2 2
D + 10D + 100D + 1 + 0.01 ( K ( 0.5D + 0.1 + 0.1D ) )

X K ( 0.5D + 0.1 + 0.1D )

2 Y
----------------------------------------------------------------------------------------------------------------------------------------------
-
3 2
D + D ( 10 + 0.001K ) + D ( 100 + 0.005K ) + ( 1 + 0.001K )
 root locus - 11.25

Given the homogeneous equation for the system,

3 2
D + D ( 10 + 0.001K ) + D ( 100 + 0.005K ) + ( 1 + 0.001K ) = 0
The roots can be found with a calculator, Mathcad, or equivalent.
K roots notes

-100,000 94.3, -3.992, -0.263

-1000 0, -4.5+/-8.65j roots become negative
-10 -0.0099, -4.99+/-8.66j
0 -0.01, -4.995+/-8.657j
10 -0.01, -5+/-8.66j
1000 -0.019, -5.49+/-8.64j
17165.12 -0.099, -13.52, -13.546 roots become real
100,000 -0.0174, -104.3, -5.572

6.
2
θ D ( 100 ) + D ( 100P ) + ( 100 ) -
a) ------ = ----------------------------------------------------------------------------------
Vd 3 2
D + D ( 300 ) + D ( 200P ) + ( 200 )
b)
c)

11.6 ASSIGNMENT PROBLEMS

1. The systems below have a variable spring coefficient. For each of the systems below,
 root locus - 11.26

a) Write the differential equation and convert it to a transfer function.

Kd1 = 1 Ns/m Kd = 1 Ns/m

Ks1

M = 1 kg
M = 1kg
y

F y
F

Ks

b) If the input force is a step function of magnitude 1N, calculate the time response
for ‘y’ by solving a differential equation for a Ks value of 10N/m.
c) Draw the poles for the transfer function on a real-complex plane.
d) Draw a Bode plot for Ks = 1N/m.

2. Draw a root locus diagram for the feedback system below given the variable parameter ‘P’.

Vd + e Vs 100 - ω 1-
--- θ
P ------------------
D + 100 D
- Va
2

3. For the transfer functions below, draw the root locus plots assuming there is unity feedback,
i.e., H(D) = 1. Draw an approximate time response for each for a step input.
1 - 1 - 1 - 1
G(s) = ------------ ---------------
2
--------------------
2
------------------------------
2
D+1 D +1 (D + 1) D + 2D + 2
 root locus - 11.27

4. Draw a root-locus plot for the following feedback control systems.

C + 1 R
K ---------------------
2
(D + 1)
-
2
------
2
D

C + 1 - R
K --------------------
2
(D + 1)
-
2
2D

C + 2 R
K (D + 1)

-
2
------
2
D
 nonlinear system - 12.1

12. NONLINEAR SYSTEMS

Topics:
•

Objectives:
•

12.1 INTRODUCTION

• how they are different from linear

- no transfer functions

- new elements needed in block diagrams

12.2 SOURCES OF NONLINEARITY

- important for analysis

- can be overcome by control strategies

12.2.1 Non-Linear Relationships

- input/output relationships are not linear

 nonlinear system - 12.2

θ
 – --π- 
Q ( θ ) = 8.5  e – 1
 

12.3 NON-LINEAR ELEMENTS

• If our models include a device that is non linear we will need to linearize the
model before we can proceed.

• A non-linear system can be approximated with a linear equation using the follow-
ing method.

1. Pick an operating point or range for the component.

2. Find a constant value that relates a change in the input to a change in the output.
3. Develop a linear equation.
4. Use the linear equation in the analysis (Laplace or other)

• Consider the example below,

In this case the relationship between pressure drop and flow are non-linear. We
need to develop an equation that approximates the local operating point.

∆p ≈ R ( q – q )
∆q
∆ ( ∆p ) q
R ≈ --------------- R = 2 -----2-
∆q K
q

∆p
∆ ( ∆p )
 nonlinear system - 12.3

Figure 12.1 Linearizing non-linear elements

12.3.1 Time Variant

- system parameters vary as a function of time.

Kd Ks
x + x  ----------------------------- + x  ----------------------------- = -----------------------------
·· · F
 M – 0.0001t  M – 0.0001t M – 0.0001t

12.3.2 Switching

- system components turned on/off

- cables in tension/compression

- show an example where input conditions change

·
y + 4y = f ( t ) f(t < 0) = 0

f ( 0 ≤ t < 5s ) = 2

f ( t ≥ 5s ) = 0

- give PWM (Pulse Width Modulation) example with ripple showing equivalent
voltage. PWM is used to generate analog voltage equivalents. Show for a system with first
 nonlinear system - 12.4

order response with tau = 0.1s for a frequency of 1KHz, 10Hz and 1Hz. Point out the rip-
ple and effective voltage.

- important to consider when doing system analysis

12.3.3 Deadband

- Friction in all components

- costs money to reduce friction, so it is better to compensate in software

- small actuation signals not large enough to overcome friction

- This effect is normally known as ’stiction’, a combination of the words static and
friction.

- Friction is common in less expensive motors, and when a motor is driving a

mechanical system.

- In systems there are two type of friction that must be considered.

- The static friction, ’stiction’, will prevent initial motion. If the systems breaks
free and starts turning, the kinetic friction will provide a roughly constant friction resis-
tance.

- relationship in figure below.

- the region where the applied voltage has no effect is called the deadband.
 nonlinear system - 12.5

-
ω
The kinetic
friction results
in a different curve
while slowing down

Vapplied
motor starts
deadband to turn as
friction is
overcome

Figure 12.2 Motor deadband for a bidirectional motor

- deadband compensation as shown in figure below.

Vadjusted

Vstick
Vwanted

-Vstick

Figure 12.3 Deadband approximation for a bidirectional motor

- equations for these are shown in figure

 nonlinear system - 12.6

Cadjusted
C max = 255

Cstick
C min = – 255 Cwanted
C max = 255
-Cstick if(Cwanted > 0)
C wanted
C adjusted = C stick +  ------------------ ( C max – C stick )
C min = 255  C max 
if(Cwanted < 0)
C wanted
C adjusted = – C stick +  ------------------ ( C min – C stick )
C min
if(Cwanted = 0)
C adjusted = 0

Figure 12.4 Deadband approximation for a bidirectional motor

- c-code below

#define c_stick_pos 100

#define c_stick_neg -110 /* make the value positive */
#define c_max 255
#define c_min -255 /* make the value positive */

int deadband(int c_wanted){ /* call this routine when updating */

int c_adjusted;

if(c_wanted == 0){ /* turn off the output */

c_adjusted = 0;
} else if(c_wanted > 0){ /* a positive output */
c_adjusted = c_stick_pos +
(c_max - c_stick_pos) * c_wanted / c_max;
if(c_adjusted > c_max) c_adjusted = c_max;
} else { /* the output must be negative */
c_adjusted = -c_stick_neg -
(c_min - c_stick_neg) * c_wanted / c_min;
if(c_adjusted < -c_min) c_adjusted = c_min;
}

return c_adjusted;

Figure 12.5 Deadband compensation subroutine

 nonlinear system - 12.7

12.3.4 Saturation and Clipping

• Some devices have natural maximum values, such as voltage or pressure limita-
tions caused by a regulated supply.

12.3.5 Hysteresis and Slip

- windup resulting from springiness and friction

- backlash

-
 nonlinear system - 12.8

- correct by tracking the previous motion direction and taking extra steps when
reversing direction

12.3.6 Delays and Lags

• Time delays are common in systems

• In the simplest form this is a period of time between when an event occurs and
when the effect occurs.

• If an output delay is larger than the control system step time it may be necessary
to predict future states and initiate outputs ahead of those.

• If an input delay is larger than the control system it might be necessary to slow
the control action, or build it into the control law.
 nonlinear system - 12.9

12.4 SUMMARY

12.5 PRACTICE PROBLEMS

1.

12.6 PRACTICE PROBLEM SOLUTIONS

12.7 ASIGNMENT PROBLEMS

 analog IO - 13.1

13. ANALOG INPUTS AND OUTPUTS

Topics:
• Analog inputs and outputs
• Sampling issues; aliasing, quantization error, resolution

Objectives:
• To understand the basics of conversion to and from analog values.

13.1 INTRODUCTION

An analog value is continuous, not discrete, as shown in Figure 13.1. In the previ-
ous chapters, techniques were discussed for designing continuos control systems. In this
chapter we will examine analog inputs and outputs so that we may design continuous con-
trol systems that use computers.

Voltage

logical

continuous

Figure 13.1 Logical and Continuous Values

Typical analog inputs and outputs for computers are listed below. Actuators and
sensors that can be used with analog inputs and outputs will be discussed in later chapters.

Inputs:
• oven temperature
• fluid pressure
• fluid flow rate
Outputs:
 analog IO - 13.2

• fluid valve position

• motor position
• motor velocity

A basic analog input is shown in Figure 13.2. In this type of system a physical
value is converted to a voltage, current or other value by a transducer. A signal conditioner
converts the signal from the transducer to a voltage or current that is read by the analog
input.

physical transducer signal analog integer

phenomenon (ie., sensor) conditioning input

Figure 13.2 Analog inputs

Analog to digital and digital to analog conversion uses integers within the com-
puter. Integers limit the resolution of the numbers to a discrete, or quantized range. The
effect of using integers is shown in Figure 13.3 where the desired or actual analog value is
continuous, but the possible integer values are quantified with a ’staircase’ set of values.
Consider when a continuous analog voltage is being read, it must be quantized into an
available integer value. Likewise, a desired analog output value is limited to available
quantized values. In general the difference between the analog and quantized integer value
is an error based upon the resolution of the analog I/O.

continuous
analog
quantized

integer
 analog IO - 13.3

Figure 13.3 Quantization error

13.2 ANALOG INPUTS

To input an analog voltage (into a computer) the continuous voltage value must be
sampled and then converted to a numerical value by an A/D (Analog to Digital) converter
(also known as ADC). Figure 13.4 shows a continuous voltage changing over time. There
are three samples shown on the figure. The process of sampling the data is not instanta-
neous, so each sample has a start and stop time. The time required to acquire the sample is
called the sampling time. A/D converters can only acquire a limited number of samples
per second. The time between samples is called the sampling period T, and the inverse of
the sampling period is the sampling frequency (also called sampling rate). The sampling
time is often much smaller than the sampling period. The sampling frequency is specified
when buying hardware, but a common sampling rate is 100KHz.

Voltage is sampled during these time periods

voltage

time

T = (Sampling Frequency)-1 Sampling time

Figure 13.4 Sampling an Analog Voltage

A more realistic drawing of sampled data is shown in Figure 13.5. This data is
noisier, and even between the start and end of the data sample there is a significant change
in the voltage value. The data value sampled will be somewhere between the voltage at the
start and end of the sample. The maximum (Vmax) and minimum (Vmin) voltages are a
function of the control hardware. These are often specified when purchasing hardware, but
reasonable ranges are;
 analog IO - 13.4

0V to 5V
0V to 10V
-5V to 5V
-10V to 10V

The number of bits of the A/D converter is the number of bits in the result word. If
the A/D converter is 8 bit then the result can read up to 256 different voltage levels. Most
A/D converters have 12 bits, 16 bit converters are used for precision measurements.

V(t)
V max

V ( t2 )

V ( t1 )

V min

t
τ
t1 t2
where,
V ( t ) = the actual voltage over time
τ = sample interval for A/D converter
t = time
t 1, t 2 = time at start,end of sample
V ( t 1 ), V ( t 2 ) = voltage at start, end of sample
V min, V max = input voltage range of A/D converter
N = number of bits in the A/D converter

Figure 13.5 Parameters for an A/D Conversion

The parameters defined in Figure 13.5 can be used to calculate values for A/D con-
verters. These equations are summarized in Figure 13.6. Equation 1 relates the number of
 analog IO - 13.5

bits of an A/D converter to the resolution. Equation 2 gives the error that can be expected
with an A/D converter given the range between the minimum and maximum voltages, and
the resolution (this is commonly called the quantization error). Equation 3 relates the volt-
age range and resolution to the voltage input to estimate the integer that the A/D converter
will record. Finally, equation 4 allows a conversion between the integer value from the A/
D converter, and a voltage in the computer.

N (1)
R = 2
V max – Vmin
V ERROR = ±  -----------------------------
(2)
2R

V in – V min 
V I = INT  ----------------------------
- (R – 1) (3)
 V max – V min

VI
V C =  ------------ ( V max – V min ) + Vmin (4)
R–1

where,
R = resolution of A/D converter
V I = the integer value representing the input voltage
V C = the voltage calculated from the integer value
V ERROR = the maximum quantization error

Figure 13.6 A/D Converter Equations

Consider a simple example, a 10 bit A/D converter can read voltages between -
10V and 10V. This gives a resolution of 1024, where 0 is -10V and 1023 is +10V. Because
there are only 1024 steps there is a maximum error of ±9.8mV. If a voltage of 4.564V is
input into the PLC, the A/D converter converts the voltage to an integer value of 746.
When we convert this back to a voltage the result is 4.570V. The resulting quantization
error is 4.570V-4.564V=+0.006V. This error can be reduced by selecting an A/D con-
verter with more bits. Each bit halves the quantization error.
 analog IO - 13.6

Given,
N = 10
V max = 10V
V min = – 10V
V in = 4.564V

Calculate,
N
R = 2 = 1024
V max – V min
V ERROR =  ----------------------------- = 0.0098V
 2R 

V in – V min 
V I = INT  ----------------------------
- R = 746
 V max – V min

VI
V C =  ----- ( V max – Vmin ) + V min = 4.570V
 R

Figure 13.7 Sample Calculation of A/D Values

If the voltage being sampled is changing too fast we may get false readings, as
shown in Figure 13.8. In the upper graph the waveform completes seven cycles, and 9
samples are taken. The bottom graph plots out the values read. The sampling frequency
was too low, so the signal read appears to be different that it actually is, this is called alias-
ing.
 analog IO - 13.7

Figure 13.8 Low Sampling Frequencies Cause Aliasing

Figure 13.9 Very Low Sampling Frequencies Produce Apparently Random

The Nyquist criterion specifies that sampling frequencies should be at least twice
the frequency of the signal being measured, otherwise aliasing will occur. The example in
 analog IO - 13.8

Figure 13.8 violated this principle, so the signal was aliased. If this happens in real appli-
cations the process will appear to operate erratically. In practice the sample frequency
should be 4 or more times faster than the system frequency.

f AD > 2f signal where,

f AD = sampling frequency
f signal = maximum frequency of the input

There are other practical details that should be considered when designing applica-
tions with analog inputs;

• Noise - Since the sampling window for a signal is short, noise will have added
effect on the signal read. For example, a momentary voltage spike might result
in a higher than normal reading. Shielded data cables are commonly used to
reduce the noise levels.
• Delay - When the sample is requested, a short period of time passes before the
final sample value is obtained.
• Multiplexing - Most analog input cards allow multiple inputs. These may share
the A/D converter using a technique called multiplexing. If there are 4 channels
using an A/D converter with a maximum sampling rate of 100Hz, the maximum
sampling rate per channel is 25Hz.
• Signal Conditioners - Signal conditioners are used to amplify, or filter signals
coming from transducers, before they are read by the A/D converter.
• Resistance - A/D converters normally have high input impedance (resistance), so
they affect circuits they are measuring.
• Single Ended Inputs - Voltage inputs to a PLC can use a single common for mul-
tiple inputs, these types of inputs are called single ended inputs. These tend to
be more prone to noise.
• Double Ended Inputs - Each double ended input has its own common. This
reduces problems with electrical noise, but also tends to reduce the number of
inputs by half.
• Sampling Rates - The maximum number of samples that can be read each sec-
ond. If reading multiple channels with a multiplexer, this may be reduced.
• Quantization Error - Analog IO is limited by the binary resolution of the con-
verter. This means that the output is at discrete levels, instead of continuous val-
ues.
• Triggers - often external digital signals are used to signal the start of data collec-
tion.
• Range - the typical voltages that the card can read. Typical voltage ranges are -
10V to 10V, 0V to 10V, 0V to 5V, 1V to 5V, -5V to 5V, 4mA to 20mA.
• DMA - a method to write large blocks of memory directly to computer memory.
This is normally used for high speed data captures.
 analog IO - 13.9

• Filters - some A/D input cards will provide built in functions to filter the incom-
ing data to remove high frequency noise components.
• Input impedance - most analog inputs have very high input resistances, in the
range of Mohms.

ASIDE: This device is an 8 bit A/D converter. The main concept behind this is the succes-
sive approximation logic. Once the reset is toggled the converter will start by setting
the most significant bit of the 8 bit number. This will be converted to a voltage Ve that is
a function of the +/-Vref values. The value of Ve is compared to Vin and a simple logic
check determines which is larger. If the value of Ve is larger the bit is turned off. The
logic then repeats similar steps from the most to least significant bits. Once the last bit
has been set on/off and checked the conversion will be complete, and a done bit can be
set to indicate a valid conversion value.

Vin above (+ve) or below (-ve) Ve

Vin
+
-
+Vref

successive 8
approximation D to A
clock converter
logic Ve

reset done

-Vref 8
data out

Quite often an A/D converter will multiplex between various inputs. As it switches the volt-
age will be sampled by a sample and hold circuit. This will then be converted to a digi-
tal value. The sample and hold circuits can be used before the multiplexer to collect
data values at the same instant in time.

Figure 13.10 A Successive Approximation A/D Converter

 analog IO - 13.10

13.3 ANALOG OUTPUTS

Analog outputs are much simpler than analog inputs. To set an analog output an
integer is converted to a voltage. This process is very fast, and does not experience the
timing problems with analog inputs. But, analog outputs are subject to quantization errors.
Figure 13.11 gives a summary of the important relationships. These relationships are
almost identical to those of the A/D converter.

N (5)
R = 2
V max – Vmin
V ERROR = ±  -----------------------------
(6)
2R

V desired – V min
VI = INT  ----------------------------------- R (7)
 Vmax – V min 

VI
V output =  ----- ( V max – V min ) + V min (8)
R

where,
R = resolution of A/D converter
V ERROR = the maximum quantization error
V I = the integer value representing the desired voltage
V output = the voltage output using the integer value
V desired = the desired output voltage

Figure 13.11 Analog Output Relationships

Assume we are using an 8 bit D/A converter that outputs values between 0V and
10V. We have a resolution of 256, where 0 results in an output of 0V and 255 results in
10V. The quantization error will be 20mV. If we want to output a voltage of 6.234V, we
would specify an output integer of 160, this would result in an output voltage of 6.250V.
The quantization error would be 6.250V-6.234V=0.016V.
 analog IO - 13.11

Given,
N = 8
V max = 10V
V min = 0V
V desired = 6.234V

Calculate,
N
R = 2 = 256
V max – V min
V ERROR =  ----------------------------- = 0.020V
 2R 

V in – V min 
V I = INT  ----------------------------
- R = 160
 V max – V min

VI
V C =  ----- ( V max – Vmin ) + V min = 6.250V
 R

The current output from a D/A converter is normally limited to a small value, typi-
cally less than 20mA. This is enough for instrumentation, but for high current loads, such
as motors, a current amplifier is needed. This type of interface will be discussed later. If
the current limit is exceeded for 5V output, the voltage will decrease (so don’t exceed the
rated voltage). If the current limit is exceeded for long periods of time the D/A output may
be damaged.
 analog IO - 13.12

ASIDE:
5K Ω

10K Ω V– V ss
MSB bit 3 -

V +
20K Ω + +
bit 2
0
Computer Vo
40K Ω
bit 1
-

80K Ω
LSB bit 0

First we write the obvious,

V + = 0 = V–

Next, sum the currents into the inverting input as a function of the output voltage and the
input voltages from the computer,
Vb3 V b2 Vb1 V b0 Vo
--------------
- + --------------
- + --------------
- + --------------
- = -----------
-
10KΩ 20KΩ 40KΩ 80KΩ 5KΩ
∴V o = 0.5V b3 + 0.25V b2 + 0.125V b1 + 0.0625Vb 0

Consider an example where the binary output is 1110, with 5V for on,
∴V o = 0.5 ( 5V ) + 0.25 ( 5V ) + 0.125 ( 5V ) + 0.625 ( 0V ) = 4.375V

Figure 13.12 A Digital-To-Analog Converter

13.4 NOISE REDUCTION

13.4.1 Shielding

When a changing magnetic field cuts across a conductor, it will induce a current
flow. The resistance in the circuits will convert this to a voltage. These unwanted voltages
 analog IO - 13.13

result in erroneous readings from sensors, and signal to outputs. Shielding will reduce the
effects of the interference. When shielding and grounding are done properly, the effects of
electrical noise will be negligible. Shielding is normally used for; all logical signals in
noisy environments, high speed counters or high speed circuitry, and all analog signals.

There are two major approaches to reducing noise; shielding and twisted pairs.
Shielding involves encasing conductors and electrical equipment with metal. As a result
electrical equipment is normally housed in metal cases. Wires are normally put in cables
with a metal sheath surrounding both wires. The metal sheath may be a thin film, or a
woven metal mesh. Shielded wires are connected at one end to "drain" the unwanted sig-
nals into the cases of the instruments. Figure 13.13 shows a thermocouple connected with
a thermocouple. The cross section of the wire contains two insulated conductors. Both of
the wires are covered with a metal foil, and final covering of insulation finishes the cable.
The wires are connected to the thermocouple as expected, but the shield is only connected
on the amplifier end to the case. The case is then connected to the shielding ground, shown
here as three diagonal lines.

Insulated wires
Two conductor
shielded cable
cross section Metal sheath

Insulating cover

Figure 13.13 Shielding for a Thermocouple

A twisted pair is shown in Figure 13.14. The two wires are twisted at regular inter-
vals, effectively forming small loops. In this case the small loops reverse every twist, so
any induced currents are cancel out for every two twists.
 analog IO - 13.14

1" or less typical

Figure 13.14 A Twisted Pair

When designing shielding, the following design points will reduce the effects of
electromagnetic interference.

• Avoid “noisy” equipment when possible.

• Choose a metal cabinet that will shield the control electronics.
• Use shielded cables and twisted pair wires.
• Separate high current, and AC/DC wires from each other when possible.
• Use current oriented methods such as sourcing and sinking for logical I/O.
• Use high frequency filters to eliminate high frequency noise.
• Use power line filters to eliminate noise from the power supply.

13.4.2 Grounding

- ground voltages are based upon the natural voltage level in the physical ground
(the earth under your feet). This will vary over a distance. Most buildings and electrical
systems use a ground reference for the building. Between different points on the same
building ground voltage levels may vary as much as a few hundred millivolts. This can
lead to significant problems with voltage readings and system safety.

- A signal can be floating, or connected to a ground

- if floating a system normally has a self contained power source, or self reference
such as a battery, strain gauge or thermocouple. These are usually read with double ended
outputs. The potential for floating voltage levels can be minimized by connecting larger
resistors (up to 100K) from the input to ground.

- a grounded system uses a single common (ground) for all signals. These are nor-
mally connected to a single ended inputs.

- the analog common can also be connected to the ground with a large resistor to
drain off induced voltages.
 analog IO - 13.15

- cable shields or grounds are normally only connected at one side to prevent
ground loops.

13.5 CASE STUDY

- data smoothing by averaging inputs

- conversion back to an input voltage

- use it to calculate an output voltage for control.

13.6 SUMMARY

• A/D conversion will convert a continuous value to an integer value.

• D/A conversion is easier and faster and will convert a digital value to an analog
value.
• Resolution limits the accuracy of A/D and D/A converters.
• Sampling too slowly will alias the real signal.
• Analog inputs are sensitive to noise.
• Analog shielding should be used to improve the quality of electrical signals.

13.7 PRACTICE PROBLEMS

1. What is the difference between an A/D input and D/A output?

2. An analog voltage that has a range of -10V to 10V and is to be read to a precision of +/-0.05V.
How is the minimum number of bits required for the A/D converter?

3. If given a 12 bit analog input with a range of -10V to 10V. If we put in 2.735V, what will the
integer value be after the A/D conversion? What is the error? What voltage can we calculate?

13.8 PRACTICE PROBLEM SOLUTIONS

1. an A/D converter will convert an analog input voltage (or current) to an integer value. A D/A
output will convert an integer value to an output voltage.
 analog IO - 13.16

2.

10V – ( – 10V ) 7 bits, R = 128

R = ---------------------------------- = 200
0.1V 8 bits, R = 256
The minimum number of bits is 8.

3.

N = 12 R = 4096 V min = – 10V V max = 10V V in = 2.735V

V in – V min 
V I = INT  ----------------------------
- ( R – 1 ) = 2607
 V max – V min
VI
V C =  ------------ ( V max – V min ) + Vmin = 2.733V
R–1
error = V c – V in = ( – 0.002 )V

13.9 ASSIGNMENT PROBLEMS

1. An analog output needs to be between -4V and 8V, in 0.005V intervals. How many bits are
required for the D/A converter?

2. Discuss methods for reducing electrical noise in analog inputs.

 continuous sensors - 14.1

14. CONTINUOUS SENSORS

Topics:
• Continuous sensor issues; accuracy, resolution, etc.
• Angular measurement; potentiometers, encoders and tachometers
• Linear measurement; potentiometers, LVDTs, Moire fringes and accelerometers
• Force measurement; strain gages and piezoelectric
• Liquid and fluid measurement; pressure and flow
• Temperature measurement; RTDs, thermocouples and thermistors
• Other sensors
• Continuous signal inputs and wiring
• Glossary

Objectives:
• To understand the common continuous sensor types.
• To understand interfacing issues.

14.1 INTRODUCTION

Continuous sensors convert physical phenomena to measurable signals, typically

voltages or currents. Consider a simple temperature measuring device, there will be an
increase in output voltage proportional to a temperature rise. A computer could measure
the voltage, and convert it to a temperature. The basic physical phenomena typically mea-
sured with sensors include;

- angular or linear position

- acceleration
- temperature
- pressure or flow rates
- stress, strain or force
- light intensity
- sound

Most of these sensors are based on subtle electrical properties of materials and
devices. As a result the signals often require signal conditioners. These are often amplifi-
ers that boost currents and voltages to larger voltages.

Sensors are also called transducers. This is because they convert an input phenom-
ena to an output in a different form. This transformation relies upon a manufactured
device with limitations and imperfection. As a result sensor limitations are often charac-
 continuous sensors - 14.2

terized with;

Accuracy - This is the maximum difference between the indicated and actual read-
ing. For example, if a sensor reads a force of 100N with a ±1% accuracy, then
the force could be anywhere from 99N to 101N.
Resolution - Used for systems that step through readings. This is the smallest
increment that the sensor can detect, this may also be incorporated into the
accuracy value. For example if a sensor measures up to 10 inches of linear dis-
placements, and it outputs a number between 0 and 100, then the resolution of
the device is 0.1 inches.
Repeatability - When a single sensor condition is made and repeated, there will be
a small variation for that particular reading. If we take a statistical range for
repeated readings (e.g., ±3 standard deviations) this will be the repeatability.
For example, if a flow rate sensor has a repeatability of 0.5cfm, readings for an
actual flow of 100cfm should rarely be outside 99.5cfm to 100.5cfm.
Linearity - In a linear sensor the input phenomenon has a linear relationship with
the output signal. In most sensors this is a desirable feature. When the relation-
ship is not linear, the conversion from the sensor output (e.g., voltage) to a cal-
culated quantity (e.g., force) becomes more complex.
Precision - This considers accuracy, resolution and repeatability or one device rel-
ative to another.
Range - Natural limits for the sensor. For example, a sensor for reading angular
rotation may only rotate 200 degrees.
Dynamic Response - The frequency range for regular operation of the sensor. Typ-
ically sensors will have an upper operation frequency, occasionally there will be
lower frequency limits. For example, our ears hear best between 10Hz and
16KHz.
Environmental - Sensors all have some limitations over factors such as tempera-
ture, humidity, dirt/oil, corrosives and pressures. For example many sensors
will work in relative humidities (RH) from 10% to 80%.
Calibration - When manufactured or installed, many sensors will need some cali-
bration to determine or set the relationship between the input phenomena, and
output. For example, a temperature reading sensor may need to be zeroed or
adjusted so that the measured temperature matches the actual temperature. This
may require special equipment, and need to be performed frequently.
Cost - Generally more precision costs more. Some sensors are very inexpensive,
but the signal conditioning equipment costs are significant.

14.2 INDUSTRIAL SENSORS

This section describes sensors that will be of use for industrial measurements. The
sections have been divided by the phenomena to be measured. Where possible details are
provided.
 continuous sensors - 14.3

14.2.1 Angular Displacement

14.2.1.1 - Potentiometers

Potentiometers measure the angular position of a shaft using a variable resistor. A

potentiometer is shown in Figure 14.1. The potentiometer is resistor, normally made with
a thin film of resistive material. A wiper can be moved along the surface of the resistive
film. As the wiper moves toward one end there will be a change in resistance proportional
to the distance moved. If a voltage is applied across the resistor, the voltage at the wiper
interpolate the voltages at the ends of the resistor.

V1

resistive
film V1
Vw
Vw
V2

wiper
V2 physical

schematic

Figure 14.1 A Potentiometer

The potentiometer in Figure 14.2 is being used as a voltage divider. As the wiper
rotates the output voltage will be proportional to the angle of rotation.
 continuous sensors - 14.4

V1

θ max θw
θw
V out = ( V 2 – V 1 )  ----------- + V 1
 θ max
V out

V2

Figure 14.2 A Potentiometer as a Voltage Divider

Potentiometers are popular because they are inexpensive, and don’t require special
signal conditioners. But, they have limited accuracy, normally in the range of 1% and they
are subject to mechanical wear.

Potentiometers measure absolute position, and they are calibrated by rotating them
in their mounting brackets, and then tightening them in place. The range of rotation is nor-
mally limited to less than 360 degrees or multiples of 360 degrees. Some potentiometers
can rotate without limits, and the wiper will jump from one end of the resistor to the other.

Faults in potentiometers can be detected by designing the potentiometer to never

reach the ends of the range of motion. If an output voltage from the potentiometer ever
reaches either end of the range, then a problem has occurred, and the machine can be shut
down. Two examples of problems that might cause this are wires that fall off, or the poten-
tiometer rotates in its mounting.

14.2.2 Encoders

Encoders use rotating disks with optical windows, as shown in Figure 14.3. The
encoder contains an optical disk with fine windows etched into it. Light from emitters
passes through the openings in the disk to detectors. As the encoder shaft is rotated, the
light beams are broken. The encoder shown here is a quadrature encode, and it will be dis-
cussed later.
 continuous sensors - 14.5

light light
emitters detectors

Shaft rotates

Note: this type of encoder is

commonly used in com-
puter mice with a roller
ball.

Figure 14.3 An Encoder Disk

There are two fundamental types of encoders; absolute and incremental. An abso-
lute encoder will measure the position of the shaft for a single rotation. The same shaft
angle will always produce the same reading. The output is normally a binary or grey code
number. An incremental (or relative) encoder will output two pulses that can be used to
determine displacement. Logic circuits or software is used to determine the direction of
rotation, and count pulses to determine the displacement. The velocity can be determined
by measuring the time between pulses.

Encoder disks are shown in Figure 14.4. The absolute encoder has two rings, the
outer ring is the most significant digit of the encoder, the inner ring is the least significant
digit. The relative encoder has two rings, with one ring rotated a few degrees ahead of the
other, but otherwise the same. Both rings detect position to a quarter of the disk. To add
accuracy to the absolute encoder more rings must be added to the disk, and more emitters
and detectors. To add accuracy to the relative encoder we only need to add more windows
to the existing two rings. Typical encoders will have from 2 to thousands of windows per
ring.
 continuous sensors - 14.6

sensors read across

a single radial line

relative encoder
(quadrature) absolute encoder

Figure 14.4 Encoder Disks

When using absolute encoders, the position during a single rotation is measured
directly. If the encoder rotates multiple times then the total number of rotations must be
counted separately.

When using a relative encoder, the distance of rotation is determined by counting

the pulses from one of the rings. If the encoder only rotates in one direction then a simple
count of pulses from one ring will determine the total distance. If the encoder can rotate
both directions a second ring must be used to determine when to subtract pulses. The
quadrature scheme, using two rings, is shown in Figure 14.5. The signals are set up so that
one is out of phase with the other. Notice that for different directions of rotation, input B
either leads or lags A.
 continuous sensors - 14.7

Quad input A

clockwise rotation

Quad Input B

Note the change total displacement can be determined

as direction by adding/subtracting pulse counts
is reversed (direction determines add/subtract)

Quad input A

counterclockwise rotation

Quad Input B

Note: To determine direction we can do a simple check. If both are off or on, the first to
change state determines direction. Consider a point in the graphs above where both
A and B are off. If A is the first input to turn on the encoder is rotating clockwise. If
B is the first to turn on the rotation is counterclockwise.

Aside: A circuit (or program) can be built for this circuit using an up/down counter. If
the positive edge of input A is used to trigger the clock, and input B is used to drive
the up/down count, the counter will keep track of the encoder position.

Figure 14.5 Quadrature Encoders

Interfaces for encoders are commonly available for PLCs and as purchased units.
Newer PLCs will also allow two normal inputs to be used to decode encoder inputs.
 continuous sensors - 14.8

Normally absolute and relative encoders require a calibration phase when a con-
troller is turned on. This normally involves moving an axis until it reaches a logical sensor
that marks the end of the range. The end of range is then used as the zero position.
Machines using encoders, and other relative sensors, are noticeable in that they normally
move to some extreme position before use.

14.2.2.1 - Tachometers

Tachometers measure the velocity of a rotating shaft. A common technique is to

mount a magnet to a rotating shaft. When the magnetic moves past a stationary pick-up
coil, current is induced. For each rotation of the shaft there is a pulse in the coil, as shown
in Figure 14.6. When the time between the pulses is measured the period for one rotation
can be found, and the frequency calculated. This technique often requires some signal
conditioning circuitry.

pickup Vout
rotating coil
shaft Vout
t
magnet

1/f

Figure 14.6 A Magnetic Tachometer

Another common technique uses a simple permanent magnet DC generator (note:

you can also use a small DC motor). The generator is hooked to the rotating shaft. The
rotation of a shaft will induce a voltage proportional to the angular velocity. This tech-
nique will introduce some drag into the system, and is used where efficiency is not an
issue.

Both of these techniques are common, and inexpensive.

14.2.3 Linear Position

14.2.3.1 - Potentiometers
 continuous sensors - 14.9

Rotational potentiometers were discussed before, but potentiometers are also

available in linear/sliding form. These are capable of measuring linear displacement over
long distances. Figure 14.7 shows the output voltage when using the potentiometer as a
voltage divider.

V2

Vout = V 1 + ( V 2 – V1 )  ---
a
 L
V out
L
a

V1

Figure 14.7 Linear Potentiometer

Linear/sliding potentiometers have the same general advantages and disadvantages

of rotating potentiometers.

14.2.3.2 - Linear Variable Differential Transformers (LVDT)

Linear Variable Differential Transformers (LVDTs) measure linear displacements

over a limited range. The basic device is shown in Figure 14.8. It consists of outer coils
with an inner moving magnetic core. High frequency alternating current (AC) is applied to
the center coil. This generates a magnetic field that induces a current in the two outside
coils. The core will pull the magnetic field towards it, so in the figure more current will be
induced in the left hand coil. The outside coils are wound in opposite directions so that
when the core is in the center the induced currents cancel, and the signal out is zero
(0Vac). The magnitude of the signal out voltage on either line indicates the position of the
core. Near the center of motion the change in voltage is proportional to the displacement.
But, further from the center the relationship becomes nonlinear.
 continuous sensors - 14.10

A rod drives
the sliding core

∆x

∆V = K∆x
where,
AC input
∆ V = output voltage
K = constant for device
∆ x = core displacement

signal out

Figure 14.8 An LVDT

Aside: The circuit below can be used to produce a voltage that is proportional to position.
The two diodes convert the AC wave to a half wave DC wave. The capacitor and resis-
tor values can be selected to act as a low pass filter. The final capacitor should be large
enough to smooth out the voltage ripple on the output.

Vac out Vdc out

Vac in LVDT

Figure 14.9 A Simple Signal Conditioner for an LVDT

These devices are more accurate than linear potentiometers, and have less friction.
Typical applications for these devices include measuring dimensions on parts for quality
 continuous sensors - 14.11

control. They are often used for pressure measurements with Bourdon tubes and bellows/
diaphragms. A major disadvantage of these sensors is the high cost, often in the thousands.

14.2.3.3 - Moire Fringes

High precision linear displacement measurements can be made with Moire

Fringes, as shown in Figure 14.10. Both of the strips are transparent (or reflective), with
black lines at measured intervals. The spacing of the lines determines the accuracy of the
position measurements. The stationary strip is offset at an angle so that the strips interfere
to give irregular patterns. As the moving strip travels by a stationary strip the patterns will
move up, or down, depending upon the speed and direction of motion.

Stationary
Moving

Note: you can recreate this effect with the strips below. Photocopy the pattern twice,
overlay the sheets and hold them up to the light. You will notice that shifting one sheet
will cause the stripes to move up or down.

Figure 14.10 The Moire Fringe Effect

A device to measure the motion of the moire fringes is shown in Figure 14.11. A
light source is collimated by passing it through a narrow slit to make it one slit width. This
is then passed through the fringes to be detected by light sensors. At least two light sensors
are needed to detect the bright and dark locations. Two sensors, close enough, can act as a
quadrature pair, and the same method used for quadrature encoders can be used to deter-
mine direction and distance of motion.
 continuous sensors - 14.12

on
off
on
off

Figure 14.11 Measuring Motion with Moire Fringes

These are used in high precision applications over long distances, often meters.
They can be purchased from a number of suppliers, but the cost will be high. Typical
applications include Coordinate Measuring Machines (CMMs).

14.2.3.4 - Accelerometers

Accelerometers measure acceleration using a mass suspended on a force sensor, as

shown in Figure 14.12. When the sensor accelerates, the inertial resistance of the mass
will cause the force sensor to deflect. By measuring the deflection the acceleration can be
determined. In this case the mass is cantilevered on the force sensor. A base and housing
enclose the sensor. A small mounting stud (a threaded shaft) is used to mount the acceler-
ometer.

Base

Mounting Force
Stud Sensor Mass Housing

Figure 14.12 A Cross Section of an Accelerometer

Accelerometers are dynamic sensors, typically used for measuring vibrations

 continuous sensors - 14.13

between 10Hz to 10KHz. Temperature variations will affect the accuracy of the sensors.
Standard accelerometers can be linear up to 100,000 m/s**2: high shock designs can be
used up to 1,000,000 m/s**2. There is often a trade-off between a wide frequency range
and device sensitivity (note: higher sensitivity requires a larger mass). Figure 14.13 shows
the sensitivity of two accelerometers with different resonant frequencies. A smaller reso-
nant frequency limits the maximum frequency for the reading. The smaller frequency
results in a smaller sensitivity. The units for sensitivity is charge per m/s**2.

resonant freq. (Hz) sensitivity

22 KHz 4.5 pC/(m/s**2)
180KHz .004

Figure 14.13 Piezoelectric Accelerometer Sensitivities

The force sensor is often a small piece of piezoelectric material (discussed later in
this chapter). The piezoelectic material can be used to measure the force in shear or com-
pression. Piezoelectric based accelerometers typically have parameters such as,

-100 to 250°C operating range

1mV/g to 30V/g sensitivity
operate well below one forth of the natural frequency

The accelerometer is mounted on the vibration source as shown in Figure 14.14.

The accelerometer is electrically isolated from the vibration source so that the sensor may
be grounded at the amplifier (to reduce electrical noise). Cables are fixed to the surface of
the vibration source, close to the accelerometer, and are fixed to the surface as often as
possible to prevent noise from the cable striking the surface. Background vibrations can be
detected by attaching control electrodes to non-vibrating surfaces. Each accelerometer is
different, but some general application guidelines are;

• The control vibrations should be less than 1/3 of the signal for the error to be less
than 12%).
• Mass of the accelerometers should be less than a tenth of the measurement mass.
• These devices can be calibrated with shakers, for example a 1g shaker will hit a
peak velocity of 9.81 m/s**2.
 continuous sensors - 14.14

Sealant to prevent moisture

hookup wire

accelerometer
isolated
stud isolated
wafer

surface

Figure 14.14 Mounting an Accelerometer

Equipment normally used when doing vibration testing is shown in Figure 14.15.
The sensor needs to be mounted on the equipment to be tested. A pre-amplifier normally
converts the charge generated by the accelerometer to a voltage. The voltage can then be
analyzed to determine the vibration frequencies.

Sensor
Source of vibrations,
pre- or site for vibration
amp measurement

signal processor/ control system

recorder

Figure 14.15 Typical Connection for Accelerometers

Accelerometers are commonly used for control systems that adjust speeds to
reduce vibration and noise. Computer Controlled Milling machines now use these sensors
to actively eliminate chatter, and detect tool failure. The signal from accelerometers can be
 continuous sensors - 14.15

integrated to find velocity and acceleration.

Currently accelerometers cost hundreds or thousands per channel. But, advances in

micromachining are already beginning to provide integrated circuit accelerometers at a
low cost. Their current use is for airbag deployment systems in automobiles.

14.2.4 Forces and Moments

14.2.4.1 - Strain Gages

Strain gages measure strain in materials using the change in resistance of a wire.
The wire is glued to the surface of a part, so that it undergoes the same strain as the part (at
the mount point). Figure 14.16 shows the basic properties of the undeformed wire. Basi-
cally, the resistance of the wire is a function of the resistivity, length, and cross sectional
area.

t
L
-

V L L
+ R = --- = ρ --- = ρ ------
I A wt
I
where,
R = resistance of wire
V, I = voltage and current
L = length of wire
w, t = width and thickness
A = cross sectional area of conductor
ρ = resistivity of material

Figure 14.16 The Electrical Properties of a Wire

 continuous sensors - 14.16

After the wire in Figure 14.16 has been deformed it will take on the new dimen-
sions and resistance shown in Figure 14.17. If a force is applied as shown, the wire will
become longer, as predicted by Young’s modulus. But, the cross sectional area will
decrease, as predicted by Poison’s ratio. The new length and cross sectional area can then
be used to find a new resistance.

w’

t’
L’

F F F
σ = --- = ------ = Eε ∴ε = ----------
A wt Ewt
L(1 + ε)
R' = ρ -------- = ρ  ----------------------------------------------
L'
w't' w ( 1 – νε )t ( 1 – νε )
F
(1 + ε)
∴∆R = R' – R = R ---------------------------------------- – 1
( 1 – νε ) ( 1 – νε )

where,
ν = poissons ratio for the material
F = applied force
E = Youngs modulus for the material
σ, ε = stress and strain of material

Aside: Gauge factor, as defined below, is a commonly used measure of stain gauge
sensitivity.

 ∆R
-------
 R
GF = -------------
ε

Figure 14.17 The Electrical and Mechanical Properties of the Deformed Wire
 continuous sensors - 14.17

Aside: Changes in strain gauge resistance are typically small (large values would require
strains that would cause the gauges to plastically deform). As a result, Wheatstone
bridges are used to amplify the small change. In this circuit the variable resistor R2
would be tuned until Vo = 0V. Then the resistance of the strain gage can be calculated
using the given equation.

R2 R1
V+ - when Vo = 0V
R strain = -------------
R3

R4
R2 R1

+
Vo

R3
Rstrain
R5

Figure 14.18 Measuring Strain with a Wheatstone Bridge

A strain gage must be small for accurate readings, so the wire is actually wound in
a uniaxial or rosette pattern, as shown in Figure 14.19. When using uniaxial gages the
direction is important, it must be placed in the direction of the normal stress. (Note: the
gages cannot read shear stress.) Rosette gages are less sensitive to direction, and if a shear
force is present the gage will measure the resulting normal force at 45 degrees. These
gauges are sold on thin films that are glued to the surface of a part. The process of mount-
ing strain gages involves surface cleaning. application of adhesives, and soldering leads to
the strain gages.
 continuous sensors - 14.18

direction
stress

uniaxial rosette

Figure 14.19 Wire Arrangements in Strain Gages

A design techniques using strain gages is to design a part with a narrowed neck to
mount the strain gage on, as shown in Figure 14.20. In the narrow neck the strain is pro-
portional to the load on the member, so it may be used to measure force. These parts are
often called load cells.

mounted in narrow section

to increase strain effect

F F

Figure 14.20 Using a Narrow to Increase Strain

Strain gauges are inexpensive, and can be used to measure a wide range of stresses
with accuracies under 1%. Gages require calibration before each use. This often involves
making a reading with no load, or a known load applied. An example application includes
using strain gages to measure die forces during stamping to estimate when maintenance is
needed.

14.2.4.2 - Piezoelectric

When a crystal undergoes strain it displaces a small amount of charge. In other

words, when the distance between atoms in the crystal lattice changes some electrons are
forced out or drawn in. This also changes the capacitance of the crystal. This is known as
 continuous sensors - 14.19

the Piezoelectric effect. Figure 14.21 shows the relationships for a crystal undergoing a
linear deformation. The charge generated is a function of the force applied, the strain in
the material, and a constant specific to the material. The change in capacitance is propor-
tional to the change in the thickness.

b
+
q
-

c a

F
εab d-
C = --------- i = εg F
----
c dt
where,
C = capacitance change
a, b, c = geometry of material
ε = dielectric constant (quartz typ. 4.06*10**-11 F/m)
i = current generated
F = force applied
g = constant for material (quartz typ. 50*10**-3 Vm/N)
E = Youngs modulus (quartz typ. 8.6*10**10 N/m**2)

Figure 14.21 The Piezoelectric Effect

These crystals are used for force sensors, but they are also used for applications
such as microphones and pressure sensors. Applying an electrical charge can induce
strain, allowing them to be used as actuators, such as audio speakers.

When using piezoelectric sensors charge amplifiers are needed to convert the small
amount of charge to a larger voltage. These sensors are best suited to dynamic measure-
ments, when used for static measurements they tend to drift or slowly lose charge, and the
signal value will change.
 continuous sensors - 14.20

14.2.5 Liquids and Gases

There are a number of factors to be considered when examining liquids and gasses.

• Flow velocity
• Density
• Viscosity
• Pressure

There are a number of differences factors to be considered when dealing with flu-
ids and gases. Normally a fluid is considered incompressible, while a gas normally fol-
lows the ideal gas law. Also, given sufficiently high enough temperatures, or low enough
pressures a fluid can be come a liquid.

PV = nRT

where,
P = the gas pressure
V = the volume of the gas
n = the number of moles of the gas
R = the ideal gas constant =
T = the gas temperature

When flowing, the flow may be smooth, or laminar. In case of high flow rates or
unrestricted flow, turbulence may result. The Reynold’s number is used to determine the
transition to turbulence. The equation below is for calculation the Reynold’s number for
fluid flow in a pipe. A value below 2000 will result in laminar flow. At a value of about
3000 the fluid flow will become uneven. At a value between 7000 and 8000 the flow will
become turbulent.
 continuous sensors - 14.21

VDρ
R = ------------
u
where,
R = Reynolds number
V = velocity
D = pipe diameter
ρ = fluid density
u = viscosity

14.2.5.1 - Pressure

Figure 14.22 shows different two mechanisms for pressure measurement. The
Bourdon tube uses a circular pressure tube. When the pressure inside is higher than the
surrounding air pressure (14.7psi approx.) the tube will straighten. A position sensor, con-
nected to the end of the tube, will be elongated when the pressure increases.

pressure pressure
increase increase
position sensor
position sensor

pressure pressure

a) Bourdon Tube b) Baffle

Figure 14.22 Pressure Transducers

 continuous sensors - 14.22

These sensors are very common and have typical accuracies of 0.5%.

14.2.5.2 - Venturi Valves

When a flowing fluid or gas passes through a narrow pipe section (neck) the pres-
sure drops. If there is no flow the pressure before and after the neck will be the same. The
faster the fluid flow, the greater the pressure difference before and after the neck. This is
known as a Venturi valve. Figure 14.23 shows a Venturi valve being used to measure a
fluid flow rate. The fluid flow rate will be proportional to the pressure difference before
and at the neck (or after the neck) of the valve.

differential
pressure
transducer

fluid flow

Figure 14.23 A Venturi Valve

 continuous sensors - 14.23

Aside: Bernoulli’s equation can be used to relate the pressure drop in a venturi valve.

2
--p- + ----
v-
+ gz = C
ρ 2
where,
p = pressure
ρ = density
v = velocity
g = gravitational constant
z = height above a reference
C = constant
Consider the centerline of the fluid flow through the valve. Assume the fluid is incompress-
ible, so the density does not change. And, assume that the center line of the valve does
not change. This gives us a simpler equation, as shown below, that relates the velocity
and pressure before and after it is compressed.
2 2
p before v before p after v after
---------------- + ------------------ + gz = C = ------------ + -------------- + gz
ρ 2 ρ 2
2 2
p before v before p after v after
---------------
- + ------------------ = -----------
- + --------------
ρ 2 ρ 2
2 2
 v after v before 
p before – p after = ρ  -------------- – ------------------
 2 2 

The flow velocity v in the valve will be larger than the velocity in the larger pipe sec-
tion before. So, the right hand side of the expression will be positive. This will mean
that the pressure before will always be higher than the pressure after, and the differ-
ence will be proportional to the velocity squared.

Figure 14.24 The Pressure Relationship for a Venturi Valve

Venturi valves allow pressures to be read without moving parts, which makes them
very reliable and durable. They work well for both fluids and gases. It is also common to
use Venturi valves to generate vacuums for actuators, such as suction cups.

14.2.5.3 - Coriolis Flow Meter

Fluid passes through thin tubes, causing them to vibrate. As the fluid approaches
the point of maximum vibration it accelerates. When leaving the point it decelerates. The
 continuous sensors - 14.24

result is a distributed force that causes a bending moment, and hence twisting of the pipe.
The amount of bending is proportional to the velocity of the fluid flow. These devices typ-
ically have a large constriction on the flow, and result is significant loses. Some of the
devices also use bent tubes to increase the sensitivity, but this also increases the flow resis-
tance. The typical accuracy for a Coriolis flowmeter is 0.1%.

14.2.5.4 - Magnetic Flow Meter

A magnetic sensor applies a magnetic field perpendicular to the flow of a conduc-

tive fluid. As the fluid moves, the electrons in the fluid experience an electromotive force.
The result is that a potential (voltage) can be measured perpendicular to the direction of
the flow and the magnetic field. The higher the flow rate, the greater the voltage. The typ-
ical accuracy for these sensors is 0.5%.

These flowmeters don’t oppose fluid flow, and so they don’t result in pressure
drops.

14.2.5.5 - Ultrasonic Flow Meter

A transmitter emits a high frequency sound at point on a tube. The signal must then
pass through the fluid to a detector where it is picked up. If the fluid is flowing in the same
direction as the sound it will arrive sooner. If the sound is against the flow it will take
longer to arrive. In a transit time flow meter two sounds are used, one traveling forward,
and the other in the opposite direction. The difference in travel time for the sounds is used
to determine the flow velocity.

A doppler flowmeter bounces a soundwave off particle in a flow. If the particle is

moving away from the emitter and detector pair, then the detected frequency will be low-
ered, if it is moving towards them the frequency will be higher.

The transmitter and receiver have a minimal impact on the fluid flow, and there-
fore don’t result in pressure drops.

14.2.5.6 - Vortex Flow Meter

Fluid flowing past a large (typically flat) obstacle will shed vortices. The fre-
quency of the vortices will be proportional to the flow rate. Measuring the frequency
allows an estimate of the flow rate. These sensors tend be low cost and are popular for low
accuracy applications.
 continuous sensors - 14.25

14.2.5.7 - Positive Displacement Meters

In some cases more precise readings of flow rates and volumes may be required.
These can be obtained by using a positive displacement meter. In effect these meters are
like pumps run in reverse. As the fluid is pushed through the meter it produces a measur-
able output, normally on a rotating shaft.

14.2.5.8 - Pitot Tubes

Gas flow rates can be measured using Pitot tubes, as shown in Figure 14.25. These
are small tubes that project into a flow. The diameter of the tube is small (typically less
than 1/8") so that it doesn’t affect the flow.

gas flow

pitot
tube

connecting hose pressure

sensor

Figure 14.25 Pitot Tubes for Measuring Gas Flow Rates

14.2.6 Temperature

Temperature measurements are very common with control systems. The tempera-
ture ranges are normally described with the following classifications.

very low temperatures <-60 deg C - e.g. superconductors in MRI units

low temperature measurement -60 to 0 deg C - e.g. freezer controls
fine temperature measurements 0 to 100 deg C - e.g. environmental controls
high temperature measurements <3000 deg F - e.g. metal refining/processing
 continuous sensors - 14.26

very high temperatures > 2000 deg C - e.g. plasma systems

14.2.6.1 - Resistive Temperature Detectors (RTDs)

When a metal wire is heated the resistance increases. So, a temperature can be
measured using the resistance of a wire. Resistive Temperature Detectors (RTDs) nor-
mally use a wire or film of platinum, nickel, copper or nickel-iron alloys. The metals are
wound or wrapped over an insulator, and covered for protection. The resistances of these
alloys are shown in Figure 14.26.

Material Temperature Typical

Range C (F) Resistance
(ohms)

Platinum -200 - 850 (-328 - 1562) 100

Nickel -80 - 300 (-112 - 572) 120
Copper -200 - 260 (-328 - 500) 10
Figure 14.26 RTD Properties

These devices have positive temperature coefficients that cause resistance to

increase linearly with temperature. A platinum RTD might have a resistance of 100 ohms
at 0C, that will increase by 0.4 ohms/°C. The total resistance of an RTD might double over
the temperature range.

A current must be passed through the RTD to measure the resistance. (Note: a volt-
age divider can be used to convert the resistance to a voltage.) The current through the
RTD should be kept to a minimum to prevent self heating. These devices are more linear
than thermocouples, and can have accuracies of 0.05%. But, they can be expensive

14.2.6.2 - Thermocouples

Each metal has a natural potential level, and when two different metals touch there
is a small potential difference, a voltage. (Note: when designing assemblies, dissimilar
metals should not touch, this will lead to corrosion.) Thermocouples use a junction of dis-
similar metals to generate a voltage proportional to temperature. This principle was dis-
covered by T.J. Seebeck.

The basic calculations for thermocouples are shown in Figure 14.27. This calcula-
tion provides the measured voltage using a reference temperature and a constant specific
 continuous sensors - 14.27

to the device. The equation can also be rearranged to provide a temperature given a volt-
age.

measuring +
device - V out

V out = α ( T – T ref )
V out
∴T = ---------- + T ref
α
where,
α = constant (V/C) 50 µ V
------- (typical)
°C
T, T ref = current and reference temperatures

Figure 14.27 Thermocouple Calculations

The list in Table 1 shows different junction types, and the normal temperature
ranges. Both thermocouples, and signal conditioners are commonly available, and rela-
tively inexpensive. For example, most PLC vendors sell thermocouple input cards that
will allow multiple inputs into the PLC.

Table 1: Thermocouple Types

Temperature
ANSI Voltage Range
Materials Range
Type (mV)
(°F)

T copper/constantan -200 to 400 -5.60 to 17.82

J iron/constantan 0 to 870 0 to 42.28
E chromel/constantan -200 to 900 -8.82 to 68.78
K chromel/aluminum -200 to 1250 -5.97 to 50.63
R platinum-13%rhodium/platinum 0 to 1450 0 to 16.74
S platinum-10%rhodium/platinum 0 to 1450 0 to 14.97
C tungsten-5%rhenium/tungsten-26%rhenium 0 to 2760 0 to 37.07
 continuous sensors - 14.28

mV
80
E

60 K

J C
40

20 T R
S

0
0 500 1000 1500 2000 2500
(°F)

Figure 14.28 Thermocouple Temperature Voltage Relationships (Approximate)

The junction where the thermocouple is connected to the measurement instrument

is normally cooled to reduce the thermocouple effects at those junctions. When using a
thermocouple for precision measurement, a second thermocouple can be kept at a known
temperature for reference. A series of thermocouples connected together in series pro-
duces a higher voltage and is called a thermopile. Readings can approach an accuracy of
0.5%.

14.2.6.3 - Thermistors

Thermistors are non-linear devices, their resistance will decrease with an increase
in temperature. (Note: this is because the extra heat reduces electron mobility in the semi-
conductor.) The resistance can change by more than 1000 times. The basic calculation is
shown in Figure 14.29.

often metal oxide semiconductors The calculation uses a reference temperature

and resistance, with a constant for the device, to predict the resistance at another tempera-
ture. The expression can be rearranged to calculate the temperature given the resistance.
 continuous sensors - 14.29

β  --- – -----
1 1
 T T o
Rt = Ro e
βT o
∴T = ---------------------------------
Rt
T o ln  ------ + β
 R o
where,
Ro, R t = resistances at reference and measured temps.
T o, T = reference and actual temperatures
β = constant for device

Figure 14.29 Thermistor Calculations

Aside: The circuit below can be used to convert the resistance of the thermistor to a volt-
age using a Wheatstone bridge and an inverting amplifier.

+V

R5
R1 R3

-
+
Vout
R2 R4

Figure 14.30 Thermistor Signal Conditioning Circuit

 continuous sensors - 14.30

Thermistors are small, inexpensive devices that are often made as beads, or metal-
lized surfaces. The devices respond quickly to temperature changes, and they have a
higher resistance, so junction effects are not an issue. Typical accuracies are 1%, but the
devices are not linear, have a limited temperature/resistance range and can be self heating.

14.2.6.4 - Other Sensors

IC sensors are becoming more popular. They output a digital reading and can have
accuracies better than 0.01%. But, they have limited temperature ranges, and require some
knowledge of interfacing methods for serial or parallel data.

Pyrometers are non-contact temperature measuring devices that use radiated heat.
These are normally used for high temperature applications, or for production lines where it
is not possible to mount other sensors to the material.

14.2.7 Light

14.2.7.1 - Light Dependant Resistors (LDR)

Light dependant resistors (LDRs) change from high resistance (>Mohms) in bright
light to low resistance (<Kohms) in the dark. The change in resistance is non-linear, and is
also relatively slow (ms).
 continuous sensors - 14.31

Aside: an LDR can be used in a voltage divider to convert the change in resistance to a
measurable voltage.

V high

These are common in low

cost night lights.

V out

V low

Figure 14.31 A Light Level Detector Circuit

14.2.8 Chemical

14.2.8.1 - pH

The pH of an ionic fluid can be measured over the range from a strong base (alka-
line) with pH=14, to a neutral value, pH=7, to a strong acid, pH=0. These measurements
are normally made with electrodes that are in direct contact with the fluids.

14.2.8.2 - Conductivity

Conductivity of a material, often a liquid is often used to detect impurities. This

can be measured directly be applying a voltage across two plates submerged in the liquid
and measuring the current. High frequency inductive fields is another alternative.
 continuous sensors - 14.32

14.2.9 Others

A number of other detectors/sensors are listed below,

Combustion - gases such as CO2 can be an indicator of combustion

Humidity - normally in gases
Dew Point - to determine when condensation will form

14.3 INPUT ISSUES

Signals from sensors are often not in a form that can be directly input to a control-
ler. In these cases it may be necessary to buy or build signal conditioners. Normally, a sig-
nal conditioner is an amplifier, but it may also include noise filters, and circuitry to
convert from current to voltage. This section will discuss the electrical and electronic
interfaces between sensors and controllers.

Analog signal are prone to electrical noise problems. This is often caused by elec-
tromagnetic fields on the factory floor inducing currents in exposed conductors. Some of
the techniques for dealing with electrical noise include;

twisted pairs - the wires are twisted to reduce the noise induced by magnetic fields.
shielding - shielding is used to reduce the effects of electromagnetic interference.
single/double ended inputs - shared or isolated reference voltages (commons).

When a signal is transmitted through a wire, it must return along another path. If
the wires have an area between them the magnetic flux enclosed in the loop can induce
current flow and voltages. If the wires are twisted, a few times per inch, then the amount
of noise induced is reduced. This technique is common in signal wires and network cables.

A shielded cable has a metal sheath, as shown in Figure 14.32. This sheath needs
to be connected to the measuring device to allow induced currents to be passed to ground.
This prevents electromagnetic waves to induce voltages in the signal wires.
 continuous sensors - 14.33

A Shield is a metal sheath that Analog Input

surrounds the wires

Analog voltage source

+ IN1
- REF1
SHLD

Figure 14.32 Cable Shielding

When connecting analog voltage sources to a controller the common, or reference

voltage can be connected different ways, as shown in Figure 14.33. The least expensive
method uses one shared common for all analog signals, this is called single ended. The
more accurate method is to use separate commons for each signal, this is called double
ended. Most analog input cards allow a choice between one or the other. But, when double
ended inputs are used the number of available inputs is halved. Most analog output cards
are double ended.
 continuous sensors - 14.34

device + device +
#1 Ain 0 #1 Ain 0
- -
Ain 1 Ain 0
device + device +
#1 Ain 2 #1 Ain 1
- -
Ain 3 Ain 1

Ain 4 Ain 2

Ain 5 Ain 2

Ain 6 Ain 3

Ain 7 Ain 3

COM

Single ended - with this arrangement the Double ended - with this arrangement
signal quality can be poorer, but more the signal quality can be better, but
inputs are available. fewer inputs are available.

Figure 14.33 Single and Double Ended Inputs

Signals from transducers are typically too small to be read by a normal analog
input card. Amplifiers are used to increase the magnitude of these signals. An example of
a single ended signal amplifier is shown in Figure 14.34. The amplifier is in an inverting
configuration, so the output will have an opposite sign from the input. Adjustments are
provided for gain and offset adjustments.

Note: op-amps are used in this section to implement the amplifiers because they are
inexpensive, common, and well suited to simple design and construction projects.
When purchasing a commercial signal conditioner, the circuitry will be more com-
plex, and include other circuitry for other factors such as temperature compensation.
 continuous sensors - 14.35

+V

Ro offset Rf Rg

-V gain

-
+
Ri
Vin Vout

Rf + Rg
V out =  ----------------- V in + offset
 Ri 

Figure 14.34 A Single Ended Signal Amplifier

A differential amplifier with a current input is shown in Figure 14.35. Note that Rc
converts a current to a voltage. The voltage is then amplified to a larger voltage.

R1 Rf
Rc -
Iin +
R2
Vout

R3

R4

Figure 14.35 A Current Amplifier

 continuous sensors - 14.36

The circuit in Figure 14.36 will convert a differential (double ended) signal to a
single ended signal. The two input op-amps are used as unity gain followers, to create a
high input impedance. The following amplifier amplifies the voltage difference.

-
+

-
Vin +
Vout

-
+

CMRR
adjust

Figure 14.36 A Differential Input to Single Ended Output Amplifier

The Wheatstone bridge can be used to convert a resistance to a voltage output, as

shown in Figure 14.37. If the resistor values are all made the same (and close to the value
of R3) then the equation can be simplified.
 continuous sensors - 14.37

+V

R5
R1 R3

-
+
Vout
R2 R4

R2
V out = V ( R 5 )   ------------------  ------ + ------ + ------ – ------
1 1 1 1
 R 1 + R 2 R 3 R 4 R 5 R 3

or if R = R 1 = R 2 = R 4 = R 5

V out = V  ---------
R
2R 3

Figure 14.37 A Resistance to Voltage Amplifier

14.4 SENSOR GLOSSARY

Ammeter - A meter to indicate electrical current. It is normally part of a DMM

Bellows - This is a flexible volumed that will expand or contract with a pressure
change. This often looks like a cylinder with a large radius (typ. 2") but it is
very thin (type 1/4"). It can be set up so that when pressure changes, the dis-
placement of one side can be measured to determine pressure.
Bourdon tube - Widely used industrial gage to measure pressure and vacuum. It
resembles a crescent moon. When the pressure inside changes the moon shape
will tend to straighten out. By measuring the displacement of the tip the pres-
sure can be measured.
Chromatographic instruments - laboratory-type instruments used to analyze chem-
ical compounds and gases.
Inductance-coil pulse generator - transducer used to measure rotational speed. Out-
 continuous sensors - 14.38

put is pulse train.

Interferometers - These use the interference of light waves 180 degrees out of
phase to determine distances. Typical sources of the monochromatic light
required are lasers.
Linear-Variable-Differential transformer (LVDT) electromechanical transducer
used to measure angular or linear displacement. Output is Voltage
Manometer - liquid column gage used widely in industry to measure pressure.
Ohmmeter - meter to indicate electrical resistance
Optical Pyrometer - device to measure temperature of an object at high tempera-
tures by sensing the brightness of an objects surface.
Orifice Plate - widely used flowmeter to indicate fluid flow rates
Photometric Transducers - a class of transducers used to sense light, including
phototubes, photodiodes, phototransistors, and photoconductors.
Piezoelectric Accelerometer - Transducer used to measure vibration. Output is
emf.
Pitot Tube - Laboratory device used to measure flow.
Positive displacement Flowmeter - Variety of transducers used to measure flow.
Typical output is pulse train.
Potentiometer - instrument used to measure voltage
Pressure Transducers - A class of transducers used to measure pressure. Typical
output is voltage. Operation of the transducer can be based on strain gages or
other devices.
Radiation pyrometer - device to measure temperature by sensing the thermal radia-
tion emitted from the object.
Resolver - this device is similar to an incremental encoder, except that it uses coils
to generate magnetic fields. This is like a rotary transformer.
Strain Gage - Widely used to indicate torque, force, pressure, and other variables.
Output is change in resistance due to strain, which can be converted into volt-
age.
Thermistor - Also called a resistance thermometer; an instrument used to measure
temperature. Operation is based on change in resistance as a function of temper-
ature.
Thermocouple - widely used temperature transducer based on the Seebeck effect,
in which a junction of two dissimilar metals emits emf related to temperature.
Turbine Flowmeter - transducer to measure flow rate. Output is pulse train.
Venturi Tube - device used to measure flow rates.

14.5 SUMMARY

• Selection of continuous sensors must include issues such as accuracy and resolu-
tion.
• Angular positions can be measured with potentiometers and encoders (more
accurate).
• Tachometers are useful for measuring angular velocity.
 continuous sensors - 14.39

• Linear positions can be measured with potentiometers (limited accuracy), LVDTs

(limited range), moire fringes (high accuracy).
• Accelerometers measure acceleration of masses.
• Strain gauges and piezoelectric elements measure force.
• Pressure can be measured indirectly with bellows and Bourdon tubes.
• Flow rates can be measured with Venturi valves and pitot tubes.
• Temperatures can be measured with RTDs, thermocouples, and thermistors.
• Input signals can be single ended for more inputs or double ended for more accu-
racy.

14.6 REFERENCES

Bryan, L.A. and Bryan, E.A., Programmable Controllers; Theory and Implementation, Industrial
Text Co., 1988.

Swainston, F., A Systems Approach to Programmable Controllers, Delmar Publishers Inc., 1992.

14.7 PRACTICE PROBLEMS

1. Name two types of inputs that would be analog input values (versus a digital value).

2. Search the web for common sensor manufacturers for 5 different types of continuous sensors. If
possible identify prices for the units. Sensor manufacturers include (hyde park, banner, allen
bradley, omron, etc.)

3. What is the resolution of an absolute optical encoder that has six binary tracks? nine tracks?
twelve tracks?

4. Suggest a couple of methods for collecting data on the factory floor

5. If a thermocouple generates a voltage of 30mV at 800F and 40mV at 1000F, what voltage will
be generated at 1200F?

6. A potentiometer is to be used to measure the position of a rotating robot link (as a voltage
divider). The power supply connected across the potentiometer is 5.0 V, and the total wiper
travel is 300 degrees. The wiper arm is directly connected to the rotational joint so that a given
rotation of the joint corresponds to an equal rotation of the wiper arm.
a) If the joint is at 42 degrees, what voltage will be output from the potentiometer?
b) If the joint has been moved, and the potentiometer output is 2.765V, what is the
position of the potentiometer?

7. A motor has an encoder mounted on it. The motor is driving a reducing gear box with a 50:1
 continuous sensors - 14.40

ratio. If the position of the geared down shaft needs to be positioned to 0.1 degrees, what is the
minimum resolution of the incremental encoder?

8. What is the difference between a strain gauge and an accelerometer? How do they work?

9. Use the equations for a permanent magnet DC motor to explain how it can be used as a tachom-
eter.

10. What are the trade-offs between encoders and potentiometers?

11. A potentiometer is connected to a PLC analog input card. The potentiometer can rotate 300
degrees, and the voltage supply for the potentiometer is +/-10V. Write a ladder logic program
to read the voltage from the potentiometer and convert it to an angle in radians stored in F8:0.

14.8 PRACTICE PROBLEM SOLUTIONS

1. Temperature and displacement

2. Sensors can be found at www.ab.com, www.omron.com, etc

3. 360°/64steps, 360°/512steps, 360°/4096steps

4. data bucket, smart machines, PLCs with analog inputs and network connections

5.
V out = α ( T – T ref ) 0.030 = α ( 800 – T ref ) 0.040 = α ( 1000 – T ref )

1 800 – T ref 1000 – T ref

--- = ------------------------ = ---------------------------
α 0.030 0.040
800 – T ref = 750 – 0.75T ref
0.040 50µV
50 = 0.25T ref T ref = 200F α = --------------------------- = -------------
1000 – 200 F
V out = 0.00005 ( 1200 – 200 ) = 0.050V
 continuous sensors - 14.41

6.
θw
V out = ( V2 – V 1 )  ----------- + V 1 = ( 5V – 0V )  ------------------ + 0V = 0.7V
42deg
a)
θ max 300deg
θw
b) 2.765V = ( 5V – 0V )  ------------------ + 0V
300deg
θw
2.765V = ( 5V – 0V )  ------------------ + 0V
300deg
θ w = 165.9deg

7.
θ input
deg
θ output = 0.1 -------------- - = 50
--------------- ------ θ input = 50  0.1 -------------- = 5 --------------
deg deg
count θ output 1  count count

deg
360 ---------
rot count
R = ------------------ = 72 --------------
deg rot
5 --------------
count

8.
strain gauge measures strain in a material using a stretching wire that increases resis-
tance - accelerometers measure acceleration with a cantilevered mass on a piezoelec-
tric element.

9.
R
When the motor shaft is turned by
another torque source a voltage is gener-
ated that is proportional to the angular
DMM V = Ks ω +
velocity. This is the reverse emf. A dmm,
- or other high impedance instrument can
be used to measure this, thus minizing
the loses in resistor R.
2
ω + ω  ------ = V s  ------
· K K
JR  JR

V s = ω ( K ) + ω  ------
· JR
K
 continuous sensors - 14.42

10.
encoders cost more but can have higher resolutions. Potentiometers have limited
ranges of motion

11.

FS BTW
Rack: 0
Group: 0
Module: 0
BT Array: BT9:0
Data File: N7:0
Length: 37
Continuous: no
BT9:0/EN BT9:1/EN
BTR
Rack: 0
Group: 0
Module: 0
BT Array: BT9:1
Data File: N7:37
Length: 20
Continuous: no
BT9:1/DN
CPT
Dest F8:0
Expression
"20.0 * N7:41 / 4095.0 - 10"
CPT
Dest F8:0
Expression
"300.0 * (F8:0 + 10) / 20"
RAD
Source F8:0
Dest F8:1

14.9 ASSIGNMENT PROBLEMS

1. Write a simple C program to read incremental encoder inputs (A and B) to determine the cur-
rent position of the encoder. Note: use the quadrature encoding to determine the position of the
motor.
 continuous sensors - 14.43

2. A high precision potentiometer has an accuracy of +/- 0.1% and can rotate 300degrees and is
used as a voltage divider with a of 0V and 5V. The output voltage is being read by an A/D con-
verter with a 0V to 10V input range. How many bits does the A/D converter need to accommodate
the accuracy of the potentiometer?

3. The table of position and voltage values below were measured for an inexpensive potentiome-
ter. Write a C subroutine that will accept a voltage value and interpolate the position value.
theta (deg) V

0 0.1
67 0.6
145 1.6
195 2.4
213 3.4
296 4.2
315 5.0
 continuous actuators - 15.1

15. CONTINUOUS ACTUATORS

Topics:
• Servo Motors; AC and DC
• Stepper motors
• Single axis motion control
• Hydraulic actuators

Objectives:
• To understand the main differences between continuous actuators
• Be able to select a continuous actuator
• To be able to plan a motion for a single servo actuator

15.1 INTRODUCTION

Continuous actuators allow a system to position or adjust outputs over a wide

range of values. Even in their simplest form, continuous actuators tend to be mechanically
complex devices. For example, a linear slide system might be composed of a motor with
an electronic controller driving a mechanical slide with a ball screw. The cost for such
actuators can easily be higher than for the control system itself. These actuators also
require sophisticated control techniques that will be discussed in later chapters. In general,
when there is a choice, it is better to use discrete actuators to reduce costs and complexity.

15.2 ELECTRIC MOTORS

An electric motor is composed of a rotating center, called the rotor, and a station-
ary outside, called the stator. These motors use the attraction and repulsion of magnetic
fields to induce forces, and hence motion. Typical electric motors use at least one electro-
magnetic coil, and sometimes permanent magnets to set up opposing fields. When a volt-
age is applied to these coils the result is a torque and rotation of an output shaft. There are
a variety of motor configuration the yields motors suitable for different applications. Most
notably, as the voltages supplied to the motors will vary the speeds and torques that they
will provide.
 continuous actuators - 15.2

• Motor Categories
• AC motors - rotate with relatively constant speeds proportional to the fre-
quency of the supply power
induction motors - squirrel cage, wound rotor - inexpensive, effi-
cient.
synchronous - fixed speed, efficient

• DC motors - have large torque and speed ranges

permanent magnet - variable speed
wound rotor and stator - series, shunt and compound (universal)
• Hybrid
brushless permanent magnet -
stepper motors

• Contactors are used to switch motor power on/off

• Drives can be used to vary motor speeds electrically. This can also be done with
mechanical or hydraulic machines.

• Popular drive categories

• Variable Frequency Drives (VFD) - vary the frequency of the power
delivered to the motor to vary speed.
• DC motor controllers - variable voltage or current to vary the motor speed
• Eddy Current Clutches for AC motors - low efficiency, uses a moving
iron drum and windings
• Wound rotor AC motor controllers - low efficiency, uses variable resistors
to adjust the winding currents

A control system is required when a motor is used for an application that requires
continuous position or velocity. A typical controller is shown in Figure 15.1. In any con-
trolled system a command generator is required to specify a desired position. The control-
ler will compare the feedback from the encoder to the desired position or velocity to
determine the system error. The controller with then generate an output, based on the sys-
tem error. The output is then passed through a power amplifier, which in turn drives the
motor. The encoder is connected directly to the motor shaft to provide feedback of posi-
tion.
 continuous actuators - 15.3

command
generator
(e.g., PLC)

desired position
or velocity amplified encoder
voltage/ voltage/
current power current motor
controller
amp

Figure 15.1 A Typical Feedback Motor Controller

15.2.1 Basic Brushed DC Motors

In a DC motor there is normally a set of coils on the rotor that turn inside a stator
populated with permanent magnets. Figure 15.2 shows a simplified model of a motor. The
magnetics provide a permanent magnetic field for the rotor to push against. When current
is run through the wire loop it creates a magnetic field.

magnetic
field

axis of
rotation ω
 continuous actuators - 15.4

Figure 15.2 A Simplified Rotor

The power is delivered to the rotor using a commutator and brushes, as shown in
Figure 15.3. In the figure the power is supplied to the rotor through graphite brushes rub-
bing against the commutator. The commutator is split so that every half revolution the
polarity of the voltage on the rotor, and the induced magnetic field reverses to push against
the permanent magnets.

motor
shaft brushes

Top split commutator

Front split commutator

motor
shaft

brushes

V+ power V-
supply

Figure 15.3 A Split Ring Commutator

The direction of rotation will be determined by the polarity of the applied voltage,
and the speed is proportional to the voltage. A feedback controller is used with these
motors to provide motor positioning and velocity control.

These motors are losing popularity to brushless motors. The brushes are subject to
 continuous actuators - 15.5

wear, which increases maintenance costs. In addition, the use of brushes increases resis-
tance, and lowers the motors efficiency.

ASIDE: The controller to drive a servo motor normally uses a Pulse Width Modulated
(PWM) signal. As shown below the signal produces an effective voltage that is rela-
tive to the time that the signal is on. The percentage of time that the signal is on is
called the duty cycle. When the voltage is on all the time the effective voltage deliv-
ered is the maximum voltage. So, if the voltage is only on half the time, the effective
voltage is half the maximum voltage. This method is popular because it can pro-
duce a variable effective voltage efficiently. The frequency of these waves is nor-
mally above 20KHz, above the range of human hearing.

50% duty cycle

V max
50
Veff = --------- V max
0 100
t
20% duty cycle
V max
20
Veff = --------- V max
0 100
t
100% duty cycle
V max
100
Veff = --------- V max
0 100
t
0% duty cycle
V max
0
Veff = --------- V max
0 100
t

Figure 15.4 Pulse Width Modulation (PWM) For Control

 continuous actuators - 15.6

ASIDE: A PWM signal can be used to drive a motor with the circuit shown below. The
PWM signal switches the NPN transistor, thus switching power to the motor. In this
case the voltage polarity on the motor will always be the same direction, so the
motor may only turn in one direction.

power
V+ supply
V+ V-
signal
source DC motor

com

Figure 15.5 PWM Unidirectional Motor Control Circuit

 continuous actuators - 15.7

+Vs
ASIDE: When a motor is to be con-
trolled with PWM in two directions
the H-bridge circuit (shown below)
is a popular choice. These can be
built with individual components, or Va Vb
purchased as integrated circuits for
smaller motors. To turn the motor in
one direction the PWM signal is
applied to the Va inputs, while the
Vb inputs are held low. In this
arrangement the positive voltage is
at the left side of the motor. To
reverse the direction the PWM sig- Vb Va
nal is applied to the Vb inputs, while
the Va inputs are held low. This
applies the positive voltage to the
right side of the motor.
-Vs

Figure 15.6 PWM Bidirectional Motor Control Circuit

15.2.2 AC Motors

• Power is normally generated as 3-phase AC, so using this increases the efficiency
of electrical drives.

• In AC motors the AC current is used to create changing fields in the motor.

• Typically AC motors have windings on the stator with multiple poles. Each pole
is a pair of windings. As the AC current reverses, the magnetic field in the rotor appears to
rotate.
 continuous actuators - 15.8

L1

stator windings
rotor

Neut.

Figure 15.7 A 2 Pole Single Phase AC Motor

 continuous actuators - 15.9

L2

L1

L3

Neut.

Neut.

Neut.

Figure 15.8 A 6 Pole 3-Phase AC Motor

• The number of windings (poles) can be an integer multiple of the number of

phases of power. More poles results in a lower rotation of the motor.

• Rotor types for induction motors are listed below. Their function is to intersect
changing magnetic fields from the stator. The changing field induces currents in the rotor.
These currents in turn set up magnetic fields that oppose fields from the stator, generating
a torque.

Squirrel cage - has the shape of a wheel with end caps and bars
Wound Rotor - the rotor has coils wound. These may be connected to exter-
nal contacts via commutator

• Induction motors require slip. If the motor turns at the precise speed of the stator
field, it will not see a changing magnetic field. The result would be a collapse of the rotor
magnetic field. As a result an induction motor always turns slightly slower than the stator
field. The difference is called the slip. This is typically a few percent. As the motor is
loaded the slip will increase until the motor stalls.
 continuous actuators - 15.10

An induction motor has the windings on the stator. The rotor is normally a squirrel
cage design. The squirrel cage is a cast aluminum core that when exposed to a changing
magnetic field will set up an opposing field. When an AC voltage is applied to the stator
coils an AC magnetic field is created, the squirrel cage sets up an opposing magnetic field
and the resulting torque causes the motor to turn.

The motor will turn at a frequency close to that of the applied voltage, but there is
always some slip. It is possible to control the speed of the motor by controlling the fre-
quency of the AC voltage. Synchronous motor drives control the speed of the motors by
synthesizing a variable frequency AC waveform, as shown in Figure 15.9.

Controller

Figure 15.9 AC Motor Speed Control

These drives should be used for applications that only require a single rotational
direction. The torque speed curve for a typical induction motor is shown in Figure 15.10.
When the motor is used with a fixed frequency AC source the synchronous speed of the
motor will be the frequency of AC voltage divided by the number of poles in the motor.
The motor actually has the maximum torque below the synchronous speed. For example a
motor 2 pole motor might have a synchronous speed of (2*60*60/2) 3600 RPM, but be
rated for 3520 RPM. When a feedback controller is used the issue of slip becomes insig-
nificant.
 continuous actuators - 15.11

torque operating range

synchronous speed

speed

Figure 15.10 Torque Speed Curve for an Induction Motor

torque
Class A

speed

torque
Class B

speed

torque
Class C

speed

torque
Class D

speed

Figure 15.11 NEMA Squirrel Cage Torque Speed Curves

 continuous actuators - 15.12

• Wound rotor induction motors use external resistors. varying the resistance
allows the motors torque speed curve to vary. As the resistance value is increased the
motor torque speed curve shifts from the Class A to Class D shapes.

• The figure below shows the relationship between the motor speed and applied
power, slip, and number of poles. An ideal motor with no load would have a slip of 0%.

RPM = -----------  1 – --------------

f120 S
p 100%
where,
f = power frequency (60Hz typ.)

p = number of poles (2, 4, 6, etc...)

RPM = motor speed in rotations per minute

S = motor slip

• Single phase AC motors can run in either direction. To compensate for this a
shading pole is used on the stator windings. It basically acts as an inductor to one side of
the field which slows the filed buildup and collapse. The result is that the field strength
seems to naturally rotate.

• Thermal protection is normally used in motors to prevent overheating.

• Universal motors were presented earlier for DC applications, but they can also be
used for AC power sources. This is because the field polarity in the rotor and stator both
reverse as the AC current reverses.

• Synchronous motors are different from induction motors in that they are designed
to rotate at the frequency of the fields, in other words there is no slip.

• Synchronous motors use generated fields in the rotor to oppose the stators field.

• Starting AC motors can be hard because of the low torque at low speeds. To deal
with this a switching arrangement is often used. At low speeds other coils or capacitors are
connected into the circuits. At higher speeds centrifugal switches disconnect these and the
motor behavior switches.
 continuous actuators - 15.13

• Single phase induction motors are typically used for loads under 1HP. Various
types (based upon their starting and running modes) are,

- split phase - there are two windings on the motor. A starting winding is
used to provide torque at lower speeds.
- capacitor run -
- capacitor start
- capacitor start and run
- shaded pole - these motors use a small offset coil (such as a single copper
winding) to encourage the field buildup to occur asymmetrically. These
motors are for low torque applications much less than 1HP.
- universal motors (also used with DC) have a wound rotor and stator that
are connected in series.
 continuous actuators - 15.14

Split Winding

Vin running squirrel cage

winding rotor

starting winding

starting capacitor

Capacitor Start

Vin running squirrel cage

winding rotor

starting winding

capacitor

Capacitor Run

Vin running squirrel cage

winding rotor

capacitor winding

Figure 15.12 Single Phase Motor Configurations

 continuous actuators - 15.15

running capacitor

Capacitor Start and Capacitor Run Motor

starting capacitor

Vin running squirrel cage

winding rotor

starting winding

Figure 15.13 Single Phase Motor Configurations

15.2.3 Brushless DC Motors

Brushless motors use a permanent magnet on the rotor, and use windings on the
stator. Therefore there is no need to use brushes and a commutator to switch the polarity of
the voltage on the coil. The lack of brushes means that these motors require less mainte-
nance than the brushed DC motors.

A typical Brushless DC motor could have three poles, each corresponding to one
power input, as shown in Figure 15.14. Each of coils is separately controlled. The coils are
switched on to attract or repel the permanent magnet rotor.
 continuous actuators - 15.16

V1

N
V2

V3

Figure 15.14 A Brushless DC Motor

To continuously rotate these motors the current in the stator coils must alternate
continuously. If the power supplied to the coils was a 3-phase AC sinusoidal waveform,
the motor will rotate continuously. The applied voltage can also be trapezoidal, which will
give a similar effect. The changing waveforms are controller using position feedback from
the motor to select switching times. The speed of the motor is proportional to the fre-
quency of the signal.

A typical torque speed curve for a brushless motor is shown in Figure 15.15.
 continuous actuators - 15.17

torque

speed

Figure 15.15 Torque Speed Curve for a Brushless DC Motor

15.2.4 Stepper Motors

Stepper motors are designed for positioning. They move one step at a time with a
typical step size of 1.8 degrees giving 200 steps per revolution. Other motors are designed
for step sizes of 1.8, 2.0, 2.5, 5, 15 and 30 degrees.

There are two basic types of stepper motors, unipolar and bipolar, as shown in Fig-
ure 15.16. The unipolar uses center tapped windings and can use a single power supply.
The bipolar motor is simpler but requires a positive and negative supply and more com-
plex switching circuitry.

a b 1a 1b

a b 2a 2b

2
unipolar bipolar
 continuous actuators - 15.18

Figure 15.16 Unipolar and Bipolar Stepper Motor Windings

The motors are turned by applying different voltages at the motor terminals. The
voltage change patterns for a unipolar motor are shown in Figure 15.17. For example,
when the motor is turned on we might apply the voltages as shown in line 1. To rotate the
motor we would then output the voltages on line 2, then 3, then 4, then 1, etc. Reversing
the sequence causes the motor to turn in the opposite direction. The dynamics of the motor
and load limit the maximum speed of switching, this is normally a few thousand steps per
second. When not turning the output voltages are held to keep the motor in position.

1a Step 1a 2a 1b 2b
2a stepper
controller 1 1 0 1 0
1b motor
2 0 1 1 0
2b 3 0 1 0 1
4 1 0 0 1

To turn the motor the phases are stepped through 1, 2, 3, 4, and then back to 1.
To reverse the direction of the motor the sequence of steps can be reversed,
eg. 4, 3, 2, 1, 4, ..... If a set of outputs is kept on constantly the motor will be
held in position.

Figure 15.17 Stepper Motor Control Sequence for a Unipolar Motor

Stepper motors do not require feedback except when used in high reliability appli-
cations and when the dynamic conditions could lead to slip. A stepper motor slips when
the holding torque is overcome, or it is accelerated too fast. When the motor slips it will
move a number of degrees from the current position. The slip cannot be detected without
position feedback.

Stepper motors are relatively weak compared to other motor types. The torque
speed curve for the motors is shown in Figure 15.18. In addition they have different static
and dynamic holding torques. These motors are also prone to resonant conditions because
of the stepped motion control.
 continuous actuators - 15.19

torque

speed

Figure 15.18 Stepper Motor Torque Speed Curve

The motors are used with controllers that perform many of the basic control func-
tions. At the minimum a translator controller will take care of switching the coil voltages.
A more sophisticated indexing controller will accept motion parameters, such as distance,
and convert them to individual steps. Other types of controllers also provide finer step res-
olutions with a process known as microstepping. This effectively divides the logical steps
described in Figure 15.17 and converts them to sinusoidal steps.

translators - the user indicates maximum velocity and acceleration and a distance
to move
indexer - the user indicates direction and number of steps to take
microstepping - each step is subdivided into smaller steps to give more resolution

15.2.5 Wound Field Motors

• Uses DC power on the rotor and stator to generate the magnetic field (i.e., no per-
manent magnets)

• Shunt motors

- have the rotor and stator coils connected in parallel.

- when the load on these motors is reduced the current flow increases
slightly, increasing the field, and slowing the motor.
- these motors have a relatively small variation in speed as they are varied,
and are considered to have a relatively constant speed.
- the speed of the motor can be controlled by changing the supply voltage,
or by putting a rheostat/resistor in series with the stator windings.
 continuous actuators - 15.20

Va
I a = ------
Ra
T = Kt Ia φ

where,
I a, V a, R a = Armature current, voltage and resistance

T = Torque on motor shaft

K t = Motor speed constant
φ = motor field flux

operating
range

• Series motors\

- have the rotor and stator coils connected in series.

- as the motor speed increases the current increases, the motor can theoreti-
 continuous actuators - 15.21

cally accelerate to infinite speeds if unloaded. This makes the dangerous

when used in applications where they are potentially unloaded.
- these motors typically have greater starting torques that shunt motors

Va
I a = -----------------
Ra + Rf
2
T = Kt Ia φ = Kt Ia

where,
I a, V a = Armature current, voltage

R a, R f = Armature and field coil resistance

T = Torque on motor shaft

K t = Motor speed constant
φ = motor field flux

stall torque

T
 continuous actuators - 15.22

The XXXXXXX

e f = r a i a + Dla i a + e m

e m = K e θD

T = KT ia

e a = ( r a + l a D )i a + K e Dθ

e a = ( r a + l a D )  ------ + K e Dθ
T
 K T

Figure 15.19 Equations for an armature controlled DC motor

• Compound motors\

- have the rotor and stator coils connected in series.

- differential compound motors have the shunt and series winding field
aligned so that they oppose each other.
- cumulative compound motors have the shunt and series winding fields
aligned so that they add

cumulative

differential

T
 continuous actuators - 15.23

ef = rf if + l f if D

T = KT if

T
--- = JD 2 + BD
θ

--θ- = -----------------------
1 -
T 2
JD + BD

--- = --θ- T
θ KT
--- = -----------------------
-
if T if 2
JD + BD
θ θi  KT   1 
---- = --- ----f =  -----------------------
- ------------------
ef if ef  JD 2 + BD  r f + l f D
i
--- ----f = K T  -----------------
1 -
---- = T
T
ef if ef  r f + l f D

Figure 15.20 Equations for a controlled field motor

15.3 HYDRAULICS

Hydraulic systems are used in applications requiring a large amount of force and
slow speeds. When used for continuous actuation they are mainly used with position feed-
back. An example system is shown in Figure 15.21. The controller examines the position
of the hydraulic system, and drivers a servo valve. This controls the flow of fluid to the
actuator. The remainder of the provides the hydraulic power to drive the system.
 continuous actuators - 15.24

valve hydraulic
controller power
supply

position position
sensor hydraulic
actuator

sump

Figure 15.21 Hydraulic Servo System

The valve used in a hydraulic system is typically a solenoid controlled valve that is
simply opened or closed. Newer, more expensive, valve designs use a scheme like pulse
with modulation (PWM) which open/close the valve quickly to adjust the flow rate.

15.4 OTHER SYSTEMS

The continuous actuators discussed earlier in the chapter are the more common
types. For the purposes of completeness additional actuators are listed and described
briefly below.

Heaters - to control a heater with a continuous temperature a PWM scheme can be

used to limit a DC voltage, or an SCR can be used to supply part of an AC
waveform.
Pneumatics - air controlled systems can be used for positioning with suitable feed-
back. Velocities can also be controlled using fast acting valves.
Linear Motors - a linear motor works on the same principles as a normal rotary
motor. The primary difference is that they have a limited travel and their cost is
typically much higher than other linear actuators.
Ball Screws - rotation is converted to linear motion using balls screws. These are
low friction screws that drive nuts filled with ball bearings. These are normally
used with slides to bear mechanical loads.
 continuous actuators - 15.25

15.5 SUMMARY

• AC motors work at higher speeds

• DC motors work over a range of speeds
• Motion control introduces velocity and acceleration limits to servo control
• Hydraulics make positioning easy

15.6 PRACTICE PROBLEMS

1. A stepping motor is to be used to drive each of the three linear axes of a cartesian coordinate
robot. The motor output shaft will be connected to a screw thread with a screw pitch of 0.125”.
It is desired that the control resolution of each of the axes be 0.025”
a) to achieve this control resolution how many step angles are required on the step-
per motor?
b) What is the corresponding step angle?
c) Determine the pulse rate that will be required to drive a given joint at a velocity
of 3.0”/sec.

2. For the stepper motor in the previous question, a pulse train is to be generated by the robot con-
troller.
a) How many pulses are required to rotate the motor through three complete revo-
lutions?
b) If it is desired to rotate the motor at a speed of 25 rev/min, what pulse rate must
be generated by the robot controller?

3. Explain the differences between stepper motors, variable frequency induction motors and DC
motors using tables.
 continuous actuators - 15.26

15.7 PRACTICE PROBLEM SOLUTIONS

1.
P = 0.125  -------
in in
a) R = 0.025 ----------
rot step
in
0.025 ---------- 1 step
R step rot
θ = --- = --------------------------- = 0.2 ---------- Thus ------------------
rot
= 5 ----------
rot
P step
0.125  -------
in 0.2 ----------
rot step

rot deg
b) θ = 0.2 ---------- = 72 ----------
step step
in
c) 3 -----
s steps
PPS = ------------------------ = 120 -------------
in s
0.025 ----------
step

2.

a)
pulses = ( 3rot )  5 ---------- = 15steps
step
 rot 

---------------- =  25 --------
rot-  step
5 ---------- = 125 ------------- = 125  ------------- ------------- = 2.08 ----------
b) pulses steps 1min steps step
s  min   rot  min  60s min  s

3.
speed torque
stepper motor very low speeds low torque
vfd limited speed range good at rated speed
dc motor wide range decreases at higher speeds

15.8 ASSIGNMENT PROBLEMS

1. A stepper motor is to be used to actuate one joint of a robot arm in a light duty pick and place
application. The step angle of the motor is 10 degrees. For each pulse received from the pulse
train source the motor rotates through a distance of one step angle.
a) What is the resolution of the stepper motor?
b) Relate this value to the definitions of control resolution, spatial resolution, and
accuracy, as discussed in class.
c) For the stepper motor, a pulse train is to be generated by a motion controller.
 continuous actuators - 15.27

How many pulses are required to rotate the motor through three complete revo-
lutions? If it is desired to rotate the motor at a speed of 25 rev/min, what pulse
rate must be generated by the robot controller?

2. Describe the voltage ripple that would occur when using a permanent magnet DC motor as a
tachometer. Hint: consider the use of the commutator to switch the polarity of the coil.

3. Compare the advantages/disadvantages of DC permanent magnet motors and AC induction

motors.
 multiaxis motion - 16.1

16. MOTION CONTROL

Topics:
• Motion controllers
• Motion profiles, trapezoidal and smooth
• Gain schedulers

Objectives:
• To understand single and multi axis motion control systems.

16.1 INTRODUCTION

A system with a feedback controller will attempt to drive the system to a state
described by the desired input, such as a velocity. In earlier chapters we simply chose step
inputs, ramp inputs and other simple inputs to determine the system response. In practical
applications this setpoint needs to be generated automatically. A simple motion control
system is used to generate setpoints over time.

An example of a motion control system is shown in Figure 16.1. The motion con-
troller will accept commands or other inputs to generate a motion profile using parameters
such as distance to move, maximum acceleration and maximum velocity. The motion pro-
file is then used to generate a set of setpoints, and times they should be output. The set-
point scheduler will then use a real-time clock to output these setpoints to the motor drive.
 multiaxis motion - 16.2

motion controller
motion Motor
commands Setpoint Setpoint Servo
Generator Scheduler Drive

setpoint schedule

motion profile t (s) setpoint

ω
0 0.0
0.4 0.1 0.2
t 0.2 0.4
0.3 0.4
0 0.2 0.6 0.8 0.4 0.4
..... ....

Figure 16.1 A motion controller

The combination of a motion controller, drive and actuator is called an axis. When
there is more than one drive and actuator the system is said to have multiple axes. Com-
plex motion control systems such as computer controlled milling machines (CNC) and
robots have 3 to 6 axes which must be moved in coordination.

16.2 MOTION PROFILES

16.2.1 Velocity Profiles

A simple example of a velocity profile for a point-to-point motion is shown in Fig-

ure 16.2. In this example the motion starts at 20 deg and ends at 100 deg. (Note: in motion
controllers it is more common to used encoder pulses, instead of degrees, for positions
velocities, etc.) For position control we want a motion that has a velocity of zero at the
start and end of the motion, and accelerates and decelerates smoothly.
 multiaxis motion - 16.3

θ ( deg )

100

20
t (s)

Figure 16.2 An example of a desired motion (position)

A trapezoidal velocity profile is shown in Figure 16.3. The area under the curve is
the total distance moved. The slope of the initial and final ramp are the maximum acceler-
ation and deceleration. The top level of the trapezoid is the maximum velocity. Some con-
trollers allow the user to use the acceleration and deceleration times instead of the
maximum acceleration and deceleration. This profile gives a continuous acceleration, but
there will be a jerk (third order derivative) at the four sharp corners.

ω ( deg )
ω max

– α max
α max
t (s)
0

0 t acc t max t dec t total

where,
ω max = the maximum velocity
α max = the maximum acceleration
t acc, t dec = the acceleration and deceleration times

t max = the times at the maximum velocity

t total = the total motion time

Figure 16.3 An example of a velocity profile

 multiaxis motion - 16.4

The basic relationships for these variables are shown in Figure 16.4. The equations
can be used to find the acceleration and deceleration times. These equations can also be
used to find the time at the maximum velocity. If this time is negative it indicates that the
axis will not reach the maximum velocity, and the acceleration and deceleration times
must be decreased. The resulting velocity profile will be a triangle.

ω max
t acc = t dec = ------------ (1)
α max
t total = t acc + t max + t dec (2)

t acc t dec
θ = --- t acc ω max + t max ω max + --- t dec ω max = ω max  -------- + t max + --------
1 1 (3)
2 2  2 2 
θ t acc t dec
(4)
t max = ------------ – ----------- – -----------
ω max 2 2

Note: if the time calculated in equation 4 is negative then the axis never reaches
maximum velocity, and the velocity profile becomes a triangle.

Figure 16.4 Velocity profile basic relationships

For the example in Figure 16.5 the move starts at 100deg and ends at 20 deg. The
acceleration and decelerations are completed in half a second. The system moves for 7.5
seconds at the maximum velocity.
 multiaxis motion - 16.5

Given, θ start = 100deg θ end = 20deg

deg-
deg
ω max = 10 --------- α max = 20 --------
2
s s
The times can be calculated as,
deg
ω max 10 ---------
s
t acc = t dec = ------------ = --------------- = 0.5s
α max deg-
20 --------2
s
θ = θ end – θ start = 20deg – 100deg = – 80deg

θ t acc t dec 80deg 0.5s 0.5s

t max = ------------ – ----------
- – ----------
- = --------------- – ---------- – ---------- = 7.5s
ω max 2 2 deg 2 2
10 ---------
s
t total = t acc + t max + t dec = 0.5s + 7.5s + 0.5s = 8.5s

Figure 16.5 Velocity profile example

The motion example in Figure 16.6 is so short the axis never reaches the maximum
velocity. This is made obvious by the negative time at maximum velocity. In place of this
the acceleration and deceleration times can be calculated by using the basic acceleration
position relationship. The result in this example is a motion that accelerates for 0.316s and
then decelerates for the same time.
 multiaxis motion - 16.6

Given, θ start = 20deg θ end = 22deg

deg-
deg
ω max = 10 --------- α max = 20 --------
2
s s
The times can be calculated as,
deg
ω max 10 ---------
s
t acc = t dec = ------------ = --------------- = 0.5s
α max deg-
20 --------2
s
θ = θ end – θ start = 22deg – 20deg = 2deg

θ t acc t dec 2deg 0.5s 0.5s

t max = ------------ – ----------- – ----------- = --------------- – ---------- – ---------- = – 0.3 s
ω max 2 2 deg 2 2
10 ---------
s
The time was negative so the acceleration and deceleration times become,
θ
--- = 1--- α max t 2acc
2 2
θ - = 2deg = 2
t acc = ----------- --------------- 0.1s = 0.316s
αmax deg-
20 --------2
s
t max = 0s

Figure 16.6 Velocity profile example without reaching maximum velocity

Given the parameters calculated for the motion, the setpoints for motion can be
calculated with the equations in Figure 16.7.
 multiaxis motion - 16.7

Assuming the motion starts at 0s,

0s ≤ t < t acc
1 2
θ ( t ) = --- α max t + θ start
2
t acc ≤ t < t acc + t max

1 2
θ ( t ) = --- α max t acc + ω max ( t – t acc ) + θ start
2
t acc + t max ≤ t < t acc + t max + t dec
1 2 1 2
θ ( t ) = --- α max t acc + ω max t max + --- α max ( t – t max – t acc ) + θ start
2 2
t acc + t max + t dec ≤ t

θ ( t ) = θ end

Figure 16.7 Generating points given motion parameters

A subroutine that implements these is shown in Figure 16.8. In this subroutine the
time is looped with fixed time steps. The position setpoint values are put into the setpoint
array, which will then be used elsewhere to guide the mechanism.
 multiaxis motion - 16.8

void generate_setpoint_table(
double t_acc, double t_max, double t_step,
double vel_max, double acc_max,
double theta_start, double theta_end,
double setpoint[], int *count){
double t, t_1, t_2, t_total;
t_1 = t_acc;
t_2 = t_acc + t_max;
t_total = t_acc + t_max + t_acc;
*count = 0;
for(t = 0.0; t <= t_total; t += t_step){
if( t < t_1){
setpoint[*count] = 0.5*acc_max*t*t + theta_start;
} else if ( (t >= t_1) && (t < t_2)){
setpoint[*count] = 0.5*acc_max*t_acc*t_acc
+ vel_max*(t - t_1) + theta_start;
} else if ( (t >= t_2) && (t < t_total)){
setpoint[*count] = 0.5*acc_max*t_acc*t_acc
+ vel_max*(t_max)
+ 0.5*acc_max*(t-t_2)*(t-t_2) + theta_start;
} else {
setpoint[*count] = theta_end;
}
*count++;
}
setpoint[*count] = theta_end;
*count++;
}

Figure 16.8 Subroutine for calculating motion setpoints

In some cases the jerk should be minimized. This can be achieved by replacing the
acceleration ramps with a smooth polynomial, as shown in Figure 16.9. In this case two
quadratic polynomials will be used for the acceleration, and another two for the decelera-
tion.
 multiaxis motion - 16.9

2
ω ( deg ) ω ( t ) = At + Bt + C
ω max

α max – α max
t (s)
0

0 t acc t max t dec t total

where,
ω max = the maximum velocity
α max = the maximum acceleration
t acc, t dec = the acceleration and deceleration times

t max = the times at the maximum velocity

t total = the total motion time

Figure 16.9 A smooth velocity profile

An example of calculating the polynomial coefficients is given in Figure 16.10.

The curve found is for the first half of the acceleration. It can then be used for the three
other required curves.
 multiaxis motion - 16.10

Given, θ start θ end ω max α max

The constraints for the polynomial are,

t acc ω max
ω( 0) = 0 ω  -------- = ------------
 2  2
d t
d-  -------
acc
----- ω ( 0 ) = 0 ---- ω - = α max
dt dt  2 
These can be used to calculate the polynomial coefficients,
2
0 = A0 + B0 + C ∴C = 0
0 = 2A0 + B ∴B = 0
2 ω max
ω max = At acc A = -----------2-
t acc
α max
α max = 2At acc A = ------------
2t acc

ω max α max 2ω max

A = -----------
- = -----------
- t acc = --------------
-
t acc
2 2t acc αmax
2
α max α max α max
A = ------------ = ------------------------ = ---------------
2t acc 2ω max 4ω max
2  ---------------
 α max 

The equation for the first segment is,

2
α max 2 t acc
ω ( t ) = --------------- t 0 ≤ t < --------
4ω max 2
The equation for the second segment can be found using the first segment,
2
α max 2
ω ( t ) = ω max – --------------- ( t acc – t )
4ω max
2 t acc
α max 2 2 - ≤ t < t acc
-------
ω ( t ) = ω max – --------------- ( t – 2t acc t + t acc ) 2
4ω max

Figure 16.10 A smooth velocity profile example

 multiaxis motion - 16.11

The distance covered during acceleration, the area under the curves, is,
t acc 2 2
-------- α max 2 t acc  α max 2 
θ acc = ∫0 - t dt + ∫  ω max – --------------
- ( t – 2t acc t + t 2acc ) dt
2
--------------
--------  
4ω max tacc 4ω max
t acc 2
-------- t acc
2
2 2
α max 3  α max  t 3 
+  ω max t – ---------------  ---- – t acc t + t acc t 
2 2
θ acc = ------------------ t
12ω max
0
 4ω max 3 
t acc
-------
-
2
2 3 2 3 2 3 3 3
α max t acc αmax  t acc 3  t acc α max  t acc t acc t acc
acc = ----------------- -  -------- – t acc + t 3acc – ω max -------
- -------- + ω max t acc – -------------- - + ---------------  -------- – -------- + -------
-
12ω max 8 4ω max  3  2 4ω max  24 4 2 
2 2 2
αmax 3 ω max t acc α max 3 7α max 3
θ acc = ------------------ t acc + --------------------- – ------------------ t acc – ------------------ t acc
96ω max 2 12ω max 96ω max
2
– 14αmax 3 ω max t acc
θ acc = --------------------- t acc + ---------------------
96ω max 2

so the time required at the maximum velocity is,

( θ – 2θ acc )
t max = ---------------------------
ω max

Figure 16.11 A smooth velocity profile example (cont’d)

16.2.2 Position Profiles

A motion can be described using position points along a path. These methods are
normally used when a controller does not have any velocity or acceleration limits. The
method shown in Figure 16.12 controls motion using a parametric function ’p(u)’. The
function value varies from 0 to 1 as the parameter ’u’ varies from 0 to 1. However, the
parameters of the function are selected so that the motion starts and stops with a velocity
of zero. In this case the final polynomial equation, (3), is fairly simple. This equation can
then be used in equation (1) to generate a smooth motion path between any arbitrary start
 multiaxis motion - 16.12

and end point, with arbitrary start and end times.

t – t start
θ ( t ) = θ start + ( θ end – θstart )p  -------------------------- (1)
t end – t start
where,
θ start, θ end = start and end positions of motion
t start, t end = start and end times for the motion
3 2
p ( u ) = Au + Bu + Cu + D (2)
The constraints for the polynomial are,

p(0 ) = 0 p(1 ) = 1

d- d-
---- p(0) = 0 ---- p(1 ) = 0
dt dt

These can be used to calculate the polynomial coefficients,

3
0 = A0 + B0 + C0 + D
2 ∴D = 0

0 = 3A0 + 2B0 + C
2 ∴C = 0

∴B =  – --- A
2
3
0 = 3A1 + 2B1 2
3 2
1 = A1 + B1 + ( 0 )0 + 0 ∴A = – 2

∴B = 3
3 2
p ( u ) = – 2 u + 3u (3)

Figure 16.12 Generating smooth motion paths

The example in Figure 16.13 shows the use of a trigonometric function, instead of
a polynomial. This function was used to generate the points in the following sample pro-
gram in Figure 16.14.
 multiaxis motion - 16.13

where,
p ( u ) = A sin ( Bt + C ) + D
The coefficients can be calculated using the conditions used previously,
d-
---- p ( 0 ) = AB cos ( B ( 0 ) + C ) = 0
dt π
∴ cos ( C ) = 0 ∴C = – ---
2
d-
---- p ( 1 ) = AB cos ( B ( 1 ) + C ) = 0
dt
∴ cos ( B + C ) = 0
π ∴B = π
∴B + C = ---
2
π
p ( 0 ) = A sin  B ( 0 ) – --- + D = 0 ∴A = D
 2
π
p ( 1 ) = A sin  π ( 1 ) – --- + D = 1
 2
1
∴A ( 1 ) + A = 1 ∴A = ---
2
The final relationship is,
π
p ( u ) = --- sin  πt – --- + ---
1 1
2 2 2

Figure 16.13 Generating smooth motion paths

The program in Figure 16.14 generates a motion table that can then be used to
update setpoints. The function ’table_init()’ must be called once when the program starts
to set up global time and table values. When a new target position has been specified the
’table_generate()’ function is called to generate the setpoint table. The ’table_update()’
function is called once every interrupt scan to check the setpoint table, and update the glo-
bal setpoint variable, ’point_current’ at scheduled times. This function also includes a
simple clock to keep track of the system time.
 multiaxis motion - 16.14

#define TABLE_SIZE 11
int point_master[TABLE_SIZE] = {0, 24, 95, 206, 345, 500, 655, 794, 905, 976, 1000};
int point_position[TABLE_SIZE];
int point_time[TABLE_SIZE];
int point_start_time;
int point_index;

int ticks; /* variables to keep a system clock count */

int point_current; /* a global variable to track position */

int table_init(){ /* initialize the setpoint table */

ticks = 0; /* set the clock to zero */
point_current = 0; /* start the system at zero */
point_index = TABLE_SIZE; /* mark the table as empty */
}

void table_generate(int start, int end, int duration_sec){

unsigned i;

point_time[0] = ticks + 10; /* delay the start slightly */

point_position[0] = start;

for(i = 1; i < TABLE_SIZE; i++){

point_time[i] = point_time[0] +
(unsigned long)i * duration_sec * 250 / (TABLE_SIZE - 1);
point_position[i] = start + (long int)(end - start) * point_master[i] / 1000;
}
point_index = 0;
}

int table_update(){/* interrupt driven encoder update */

ticks++; /* update the clock */

if(point_index < TABLE_SIZE){

if(point_time[point_index] == ticks){
point_current = point_position[point_index++];
outint16(point_current);
putch("\n");
}
}
return point_current;
}

Figure 16.14 Subroutines for motion profile generation and use

16.3 MULTI AXIS MOTION

In a machine with multiple axes the motions of individual axes must often be coor-
dinated. A simple example would be a robot that needs to move two joints to reach a new
position. We could extend the motion of the slower joints so that the motion of each joint
would begin and end together.
 multiaxis motion - 16.15

16.3.1 Slew Motion

When the individual axis of a machine are not coordinated this is known as slew
motion. Each of the axes will start moving at the same time, but finish at separate times.
Consider the example in Figure 16.15. A three axis motion is required from the starting
angles of (40, 80, -40)deg, and must end at (120, 0, 0)deg. The maximum absolute acceler-
ations and decelerations are (50, 100, 150) degrees/sec/sec, and the maximum velocities
are (20, 40, 50) degrees/sec.

Joint angle (degrees)

180

90

time(sec)

θ3
θ2
-90 θ1

Joint velocity (degrees/sec)

ω max – α max

t acc t max t dec

Figure 16.15 Multi-axis slew motion

The calculations for the motion parameters are shown in Figure 16.16. These are
done in vector format for simplicity. All of the joints reach the maximum acceleration.
The fastest motion is complete in 1.13s, while the longest motion takes 4.4s.
 multiaxis motion - 16.16

The area under the velocity curve is the distance (angle in this case) travelled. First
we can determine the distance covered during acceleration, and deceleration and
the time during acceleration, and deceleration.

ω max
t acc = t dec = ------------ =  ------, ---------, --------- = ( 0.4, 0.4, 0.333 )sec.
20 40 50
αmax  50 100 150
t acc ω max.vel. 0.4 ( 20 ) 0.4 ( 40 ) 0.333 ( 50 )
θ acc. = θ dec. = ---------------------------- =  ------------------, ------------------, ------------------------ = ( 4, 8, 8.33 )deg.
2  2 2 2 

The next step is to examine the moves specified,

θ move = θ end – θ start = ( 120 – 40, 0 – 80, 0 – ( – 40 ) ) = ( 80, – 80, 40 )deg.

Remove the angles covered during accel./deccel., and find the travel time at maximum
velocity.
θ move – 2θ acc 80 – 2 ( 4 ) 80 – 2 ( 8 ) 40 – 2 ( 8.333 )
t max = ----------------------------------- =  -----------------------, -----------------------, ---------------------------------
ω max 20 40 50

t max = ( 3.6, 1.6, 0.46668 )sec.

t total = t acc + t max + t dec = ( 4.4, 2.4, 1.13 )s

Figure 16.16 Calculated times for the slew motion

16.3.1.1 - Interpolated Motion

In interpolated motion the faster joints are slowed so that they finish in coordina-
tion with the slowest. This is essential in devices such as CNC milling machines. If this
did not occur a straight line cut in the x-y plane would actually be two straight lines. The
slew motion example can be extended to be slew motion where all joints finish their
motion at 4.4s. This can be done by accelerating at the maximum acceleration, but setting
a new maximum velocity. This is shown in the example in Figure 16.17 using the results
from the example in Figure 16.16.
 multiaxis motion - 16.17

The longest motion time is 4.4s for joint 1, and this can be used to prolong the other
motions. The calculation begins by rewriting the velocity/position relationship using
a new maximum velocity.
1 ω max' ω max' 1 ω max'
∆θ = --- ------------- ω max' +  t total – 2 ------------- ω max' + --- ------------- ω max'
2 α max  α max  2 α max
ω max'
∴∆ θ = ω max'  t total – -------------
 α max 
2
∴( ω max' ) + ω max' ( – t total α max ) + ∆θα max = 0
2 2
t total α max ± t total α max – 4∆θαmax
∴ω max' = --------------------------------------------------------------------------------------
2
ω max'
t acc' = t acc' = -------------
α max
A new maximum velocity can be calculated for joint 2 using this equation.

2 2
4.4 ( 100 ) ± ( 4.4 ) ( 100 ) – 4 ( 80 ) ( 100 )
∴ω max' = --------------------------------------------------------------------------------------------------
2
440 ± 401.99502
∴ω max' = ---------------------------------------- = 421, 19.0
2
19.0
t acc' = t acc' = ---------- = 0.19s
100
A new maximum velocity can be calculated for joint 3 using this equation.

2 2
4.4 ( 150 ) ± ( 4.4 ) ( 150 ) – 4 ( 40 ) ( 150 )
∴ω max' = --------------------------------------------------------------------------------------------------
2
∴ω max' = 651, 9.22
9.22
t acc' = t acc' = ---------- = 0.092s
100

Figure 16.17 Interpolated motion based upon Figure 16.16

16.3.2 Motion Scheduling

After the setpoint schedule has been developed, it is executed by the setpoint
scheduler. The setpoint scheduler will use a clock to determine when an output setpoint
 multiaxis motion - 16.18

should be updated. A diagram of a scheduler is shown in Figure 16.18. In this system the
setpoint scheduler is an interrupt driven subroutine that compares the system clock to the
total motion time. When enough time has elapsed the routine will move to the next value
in the setpoint table. The frequency of the interrupt clock should be smaller than or equal
to the time steps used to calculate the setpoints. The servo drive is implemented with an
algorithm such a PID control.

Time based interrupt Servo motor routine runs

routine for each axis

Setpoint Choose new

table point from
trajectory table
Read θ desired

Compute
θ desired error

Interrupt
Clock
Output actuator
signal
Return

Figure 16.18 A setpoint scheduler

The output from the scheduler updates every time step. This then leads to a situa-
tion where the axis is always chasing the target value. This leads to small errors, as shown
in Figure 16.19.
 multiaxis motion - 16.19

speed
actual position

trajectory table

time
trajectory table time step

position
required

actual

time

Figure 16.19 Errors in path following

16.4 PATH PLANNING

• When we have simple features, paths are easy to generate. These features include,
- steps
- pockets
- holes
- etc.

• Typically paths for these will repeat as shown below,

 multiaxis motion - 16.20

• For complex surfaces we want to contour appropriately. These surfaces will almost always
be represented with spline patches.

• Recall that a spline patch can be represented parametrically

 multiaxis motion - 16.21

(u=0,v=1) (u=1,v=1)

xp
p ( u, v ) = y p
zp

(u=0,v=0) (u=1,v=0)

• A simple algorithm to cut the surface is shown below.

dirn_flag = 1; a direction flag

n=10 ; number of passes to cut the surface
step=1.0/n ; step sizes for u and v directions
start=step/2 ; the start offset in the u and v directions
[xp,yp,zp] = p(start,start) ; calculate the start position
print(“G00 X”,xp,” Y”,yp,” Z”,zp+0.2) ; move the tool to above the start position
for i=0 to (n-1) ; will increment in the u direction
for j=0 to (n-1) ; will increment in the v direction
; calculate next point
if dirn_flag=-1 then [xp,yp,zp]=p(start+i*step,start+j*step)
if dirn_flag=1 then [xp,yp,zp]=p(start+i*step,start+(n-j)*step)
print(“G01 X”,xp,” Y”,yp,” Z”,zp) ; instruction to cut to next point
next j ; make next step in v direction until done
dirn_flag = -dirn_flag ; reverse direction to cut in opposite direction
next i ; move to next cut line in the u direction
print(“G00 Z”,zp+0.2) ; move the tool to above the end position

16.5 CASE STUDIES

- the controller described in the block diagram below uses a model for a DC per-
manent magnet DC motor to estimate a voltage based upon a predicted velocity and posi-
tion.

- the desired position and velocity are given in figure xxx based upon the motion
 multiaxis motion - 16.22

position control derived in the earlier section

t – t start
θ ( t ) = θ start + ( θ end – θ start )p  --------------------------
t end – t start

3 2
p ( u ) = – 2 u + 3u

· t – t start
ω ( t ) = ( θ end – θ start )p  --------------------------
t end – t start

· 2
p ( u ) = – 6 u + 6u

Figure 16.20 Motion position control equation

- the controller that uses the desired position and velocity is shown in figure xxxx

real-time control loop

ωd
2
JRD + K
------------------------
K physical system

θd + θa
+ K
P ------------------------
2
+ JRD + K
-

1-
---------- K enc
K enc

Figure 16.21 Feedforward trajectory controller

 multiaxis motion - 16.23

16.6 SUMMARY

• Axis limits can be used to calculate motion profiles.

• Trapezoidal and smooth motion profiles were presented.
• Motion profiles can be used to generate setpoint tables.
• Values from the setpoints can then be output by a scheduler to drive an axis.

16.7 PRACTICE PROBLEMS

1. a) Develop a motion profile for a joint that moves from -100 degrees to 100 degrees with a
maximum velocity of 20 deg/s and a maximum acceleration of 100deg/s/s. b) Develop a set-
point table that has values for positions every 0.5 seconds for the entire motion.

2. Consider a basic servo controller with encoder feedback. The motor will start at a position
count of 100, and end the motion at a count of 3000. The motion is to have a maximum accel-
eration of 300 counts/s/s, and a maximum velocity of 100 counts/s. Find a motion profile that
satisfies these constraints. Generate a table of setpoints for the desired position every 2 sec-
onds.
 multiaxis motion - 16.24

16.8 PRACTICE PROBLEM SOLUTIONS

1.
Given,
deg deg-
ω max = 20 --------- α max = 100 -------- 2
∆θ = 100 – ( – 100 ) = 200°
s s
The motion times can be calculated.
deg
ω max 20 ---------
s
t acc = t dec = ------------ = ------------------ = 0.2s
α max deg
100 -------- 2
-
s
deg
∆θ – ω max t acc 200deg – 20 --------- 0.2s
s
t max = ---------------------------------- = --------------------------------------------------- = 9.8s
ω max deg
20 ---------
s
t total = t acc + t max + t dec = 0.2s + 9.8s + 0.2s = 10.2s

t (s) angle
(deg)
0.0 -100 1 deg- 2 deg
θ 0.5s = --- 100 -------- ( 0.2s ) + 20 --------- ( 0.5s – 0.2s ) – 100deg
0.5 -92 2 s
2 s
1.0 -82
1.5 -72 θ0.5s = – 92deg
2.0 -62
deg
2.5 -52 θ 1.0s = θ 0.5s + 20 --------- ( 0.5s ) = – 92deg + 10deg
s
3.0 -42
3.5 -32
4.0 -22
4.5 -12
5.0 -2
5.5 8
6.0 18
6.5 28
7.0 38
7.5 48
8.0 58
8.5 68
9.0 78
9.5 88
10.0 98
10.5 100
11.0 100
 multiaxis motion - 16.25

2.
t counts
(s)
0 100
2 283
4 483
6 683
8 883
10 1083
12 1283
14 1483
16 1683
18 1883
20 2083
22 2283
24 2483
26 2683
28 2883
30 3000

16.9 ASSIGNMENT PROBLEMS

1. Find a smooth path for a robot joint that will turn from θ= 75° to θ = -35° in 10 seconds. Do this
by developing an equation then calculating points every 1.0 seconds along the path for a total
motion time of 10 seconds. Do not assume a maximum velocity.

2. Paths are to be planned for a three axis motion controller. Each of the joints has a maximum
velocity of 20 deg/s, and a maximum acceleration of 30 deg/s/s. Assuming all of the joints start
at an angle of 0 degrees. Joints 1, 2 and 3 move to 40 deg, 100deg and -50deg respectively.
Develop the motion profiles assuming,
a) slew motion
b) interpolated motion

3. Develop a smooth velocity profile for moving a cutting tool that starts at 1000 inches and
moves to -1000 inches. The maximum velocity is 100 in/s and the maximum acceleration is
50in/s/s.

4. An axis has a maximum velocity of 2.0 rad/s and a maximum acceleration/deceleration of 0.5
rad/s^2. Sketch a trapezoidal motion profile and find the motion time for a -4.0 rad motion.

5. An axis has a maximum velocity of 5m/s and an acceleration/deceleration time of 1s. Sketch a
trapezoidal motion profile and find the motion time for a 25m motion.

6. Develop a smooth velocity profile for moving a cutting tool that starts at 1000 inches and
 multiaxis motion - 16.26

moves to -1000 inches. The maximum velocity is 100 in/s and the maximum acceleration is
50in/s/s.
 numerical methods - 17.1

17. LAPLACE TRANSFORMS

Topics:
• Laplace transforms
• Using tables to do Laplace transforms
• Using the s-domain to find outputs
• Solving Partial Fractions

Objectives:
• To be able to find time responses of linear systems using Laplace transforms.

17.1 INTRODUCTION

Laplace transforms provide a method for representing and analyzing linear sys-
tems using algebraic methods. In systems that begin undeflected and at rest the Laplace ’s’
can directly replace the d/dt operator in differential equations. It is a superset of the phasor
representation in that it has both a complex part, for the steady state response, but also a
real part, representing the transient part. As with the other representations the Laplace s is
related to the rate of change in the system.

D = s (if the initial conditions/derivatives are all zero at t=0s)

s = σ + jω

Figure 17.1 The Laplace s

The basic definition of the Laplace transform is shown in Figure 17.2. The normal
convention is to show the function of time with a lower case letter, while the same func-
tion in the s-domain is shown in upper case. Another useful observation is that the trans-
form starts at t=0s. Examples of the application of the transform are shown in Figure 17.3
for a step function and in Figure 17.4 for a first order derivative.
 numerical methods - 17.2

∞
∫0 f ( t )e
– st
F(s) = dt

where,
f ( t ) = the function in terms of time t
F ( s ) = the function in terms of the Laplace s

Figure 17.2 The Laplace transform

Aside: Proof of the step function transform.

For f(t) = 5,
∞ ∞ ∞ – s0
5 –st 5 –s∞ 5e 5
∫0 ∫0
– st – st
F(s) = f ( t )e dt = 5e dt = – --- e = – --- e – – ------------ = ---
s 0
s s s

Figure 17.3 Proof of the step function transform

 numerical methods - 17.3

Aside: Proof of the first order derivative transform

Given the derivative of a function g(t)=df(t)/dt,
d ∞
G ( s ) = L [ g ( t ) ] = L ----- f ( t ) = ∫ (d/dt)f ( t )e dt
– st
dt 0
we can use integration by parts to go backwards,
b b
∫a u dv = uv a – ∫ v du
b
a
∞
∫0 (d/dt)f ( t )e
– st
dt

therefore,
– st
du = df ( t ) v = e

– st
u = f(t) dv = – se dt

∞ – st ∞ ∞
∴∫ f ( t ) ( – s )e – ∫ (d/dt)f ( t )e
– st – st
dt = f ( t )e 0
dt
0 0

∞ – ∞s ∞
∴∫ (d/dt)f ( t )e ] + s ∫ f ( t )e
– st – 0s – st
dt = [ f ( t )e – f ( t )e dt
0 0

d
∴L ----- f ( t ) = – f ( 0 ) + sL [ f ( t ) ]
dt

Figure 17.4 Proof of the first order derivative transform

The previous proofs were presented to establish the theoretical basis for this
method, however tables of values will be presented in a later section for the most popular
transforms.

17.2 APPLYING LAPLACE TRANSFORMS

The process of applying Laplace transforms to analyze a linear system involves the
basic steps listed below.
 numerical methods - 17.4

1. Convert the system transfer function, or differential equation, to the s-domain by

replacing ’D’ with ’s’. (Note: If any of the initial conditions are non-zero these
must be also be added.)
2. Convert the input function(s) to the s-domain using the transform tables.
3. Algebraically combine the input and transfer function to find an output function.
4. Use partial fractions to reduce the output function to simpler components.
5. Convert the output equation back to the time-domain using the tables.

17.2.1 A Few Transform Tables

Laplace transform tables are shown in Figure 17.5, Figure 17.7 and Figure 17.8.
These are commonly used when analyzing systems with Laplace transforms. The trans-
forms shown in Figure 17.5 are general properties normally used for manipulating equa-
tions, and for converting them to/from the s-domain.
 numerical methods - 17.5

TIME DOMAIN FREQUENCY DOMAIN

f(t ) f(s )

Kf ( t ) KL [ f ( t ) ]

f1 ( t ) + f2 ( t ) –f3 ( t ) + … f 1 ( s ) + f 2 ( s ) –f 3 ( s ) + …

df ( t ) –
----------- sL [ f ( t ) ] – f ( 0 )
dt
2 –
d f(t ) 2 df ( 0 )
–
--------------
2
s L [ f ( t ) ] – sf ( 0 ) – ------------------
dt dt
n – n–1 –
f ( t )- n – 2 df ( 0
------------------) – … – -------------------------
d------------- n n–1 – d f(0
n
s L[f(t ) ] – s f(0 ) – s n
dt dt dt
t
[ f ( t ) -]
∫0 f ( t ) dt L
----------------
s
– as
f ( t – a )u ( t – a ), a > 0 e L[ f( t )]
– at
e f(t) f(s – a)

f ( at ), a > 0 1---  --s-

f
a  a

df ( s -)
–--------------
tf ( t )
ds
n
n n d f(s)
t f(t) ( – 1 ) --------------
n
-
ds
∞
( t )-
f-------
t ∫ f ( u ) du
s

Figure 17.5 Laplace transform tables

 numerical methods - 17.6

·· · where,
L [ x + 7x + 8x = 9 ] = ··
x(0) = 1
·
x(0) = 2
x(0) = 3

Figure 17.6 Drill Problem: Converting a differential equation to s-domain

The Laplace transform tables shown in Figure 17.7 and Figure 17.8 are normally
used for converting to/from the time/s-domain.
 numerical methods - 17.7

TIME DOMAIN FREQUENCY DOMAIN

δ(t) unit impulse 1

step A
A ---
s
ramp 1
t ----
2
s
2 2
t ----
3
s
n n!
t ,n>0 -----------
n+1
s
– at exponential decay 1-
e ----------
s+a
sin ( ωt ) ω
-----------------
2 2
s +ω
s
cos ( ωt ) -----------------
2 2
s +ω
– at 1
te -------------------
2
(s + a)
2 – at 2! -
t e ------------------
3
(s + a)

Figure 17.7 Laplace transform tables (continued)

 numerical methods - 17.8

TIME DOMAIN FREQUENCY DOMAIN

– at ω
e sin ( ωt ) -------------------------------
2
-
2
(s + a) + ω
– at s+a -
e cos ( ωt ) -------------------------------
2 2
(s + a) + ω
– at ω
e sin ( ωt ) -------------------------------
2
-
2
(s + a) + ω

B cos ωt +  ----------------- sin ωt

– at C – aB Bs + C
e --------------------------------
 ω  2 2
(s + a) + ω
complex conjugate
– αt A A
2Ae cos ( βt + θ ) ----------------------- + --------------------------------------
s + α – βj s + α + βj
complex conjugate
– αt A A
2t A e cos ( βt + θ ) ------------------------------
- + -------------------------------------
-
2 2
( s + α – βj ) ( s + α + βj )
– at – bt
( c – a )e – ( c – b )e s+c
--------------------------------------------------------- --------------------------------
-
b–a ( s + a)(s + b)
– at – bt
e –e 1
------------------------ --------------------------------
-
b–a ( s + a)(s + b)

Figure 17.8 Laplace transform tables (continued)

 numerical methods - 17.9

f ( t ) = 5 sin ( 5t + 8 )

f(s ) = L[ f( t) ] =

Figure 17.9 Drill Problem: Converting from the time to s-domain

5 6
f ( s ) = --- + -----------
s s+7
–1
f(t ) = L [ f(s )] =

Figure 17.10 Drill Problem: Converting from the s-domain to time domain

17.3 MODELING TRANSFER FUNCTIONS IN THE s-DOMAIN

In previous chapters differential equations, and then transfer functions, were

derived for mechanical and electrical systems. These can be converted to the s-domain, as
 numerical methods - 17.10

shown in the mass-spring-damper example in Figure 17.11. In this case we assume the
system starts undeflected and at rest, so the ’D’ operator may be directly replaced with the
Laplace ’s’. If the system did not start at rest and undeflected, the ’D’ operator would be
replaced with a more complex expression that includes the initial conditions.

2
F = MD x + K d Dx + K s x
Ks
Kd

F ( t )- 2
--------- = MD + K d D + K s
M x x( t)

F(t) F( s) 2
L ---------- = ----------- = Ms + K d s + K s
F x( t) x(s )

ASIDE: Here ’D’ is simply replaced with ’s’. Although this is very convenient, it is
only valid if the initial conditions are zero, otherwise the more complex form,
shown below, must be used.
n – n –
d f ( t -) n n–1 – n – 2 df ( 0 ) d f ( 0 )-
L -------------
n
= s L[f(t )] – s f(0 ) – s ------------------ – … – -------------------
n
dt dt dt

Figure 17.11 A mass-spring-damper example

Impedances in the s-domain are shown in Figure 17.12. As before these assume
that the system starts undeflected and at rest.

Device Time domain s-domain Impedance

Resistor V ( t ) = RI ( t ) V ( s ) = RI ( s ) Z = R
1 I(s)
V ( s ) =  ---- ---------
1 1
Capacitor V ( t ) = ---- ∫ I ( t ) dt Z = ------
C  C s sC
d
Inductor V ( t ) = L ----- I ( t ) V ( s ) = LsI ( s ) Z = Ls
dt

Figure 17.12 Impedances of electrical components

 numerical methods - 17.11

Figure 17.13 shows an example of circuit analysis using Laplace transforms. The
circuit is analyzed as a voltage divider, using the impedances of the devices. The switch
that closes at t=0s ensures that the circuit starts at rest. The calculation result is a transfer
function.

L
t=0sec

Vi +
+ R Vo
- C
-

Treat the circuit as a voltage divider,

 
Vi  ------------------
1
 1 V i  ----------------------
R
 DC + ---  1 + DCR
V o = --------------------------------------- = ---------------------------------------------- = V i  -------------------------------------------------
R R
   2 2 
 ----------------------
R D R LC + DLR + R
DL +  ------------------
1 DLR +  1 + DCR
1
 DC + ---
R
Vo
------ =  ---------------------------------------------
R
Vi  2 2 
s R LC + sLR + R

Figure 17.13 A circuit example

At this point two transfer functions have been derived. To state the obvious, these
relate an output and an input. To find an output response, an input is needed.

17.4 FINDING OUTPUT EQUATIONS

An input to a system is normally expressed as a function of time that can be con-

verted to the s-domain. An example of this conversion for a step function is shown in Fig-
ure 17.14.
 numerical methods - 17.12

Apply a constant force of A, starting at time t=0 sec.

(*Note: a force applied instantly is impossible but assumed)
F(t) = 0 for t < 0
Perform Laplace transform using tables
= A for t >= 0
A
F ( s ) = L [ F ( t ) ] = ---
s

Figure 17.14 An input function

In the previous section we converted differential equations, for systems, to transfer

functions in the s-domain. These transfer functions are a ratio of output divided by input.
If the transfer function is multiplied by the input function, both in the s-domain, the result
is the system output in the s-domain.

Given, x ( s )- 1
---------- = -------------------------------------
-
F( s) 2
Ms + K d s + K s
A
F ( s ) = ---
s
Therefore,
x( s)  
x ( s ) =  ----------- F ( s ) =  -------------------------------------
1 - A---
 F ( s )  Ms + K d s + Ks s
2

Assume, Ns
K d = 3000 ------
m
N
K s = 2000 ----
m
M = 1000kg
A = 1000N
1
∴x ( s ) = ---------------------------------
2
( s + 3s + 2 )s

Figure 17.15 A transfer function multiplied by the input function

Output functions normally have complex forms that are not found directly in trans-
form tables. It is often necessary to simplify the output function before it can be converted
back to the time domain. Partial fraction methods allow the functions to be broken into
 numerical methods - 17.13

smaller, simpler components. The previous example in Figure 17.15 is continued in Figure
17.16 using a partial fraction expansion. In this example the roots of the third order
denominator polynomial, are calculated. These provide three partial fraction terms. The
residues (numerators) of the partial fraction terms must still be calculated. The example
shows a method for finding resides by multiplying the output function by a root term, and
then finding the limit as s approaches the root.

1 1 A B C
x ( s ) = --------------------------------- = ------------------------------------ = --- + ----------- + -----------
2
( s + 3s + 2 )s ( s + 1 ) ( s + 2 )s s s+1 s+2

A = lim s  ------------------------------------
1 1
= ---
s→0 ( s + 1 ) ( s + 2 )s 2

B = lim ( s + 1 )  ------------------------------------
1
= –1
s → –1 ( s + 1 ) ( s + 2 )s

C = lim ( s + 2 )  ------------------------------------
1 1
= ---
s → –2 ( s + 1 ) ( s + 2 )s 2

Aside: the short cut above can reduce time for simple partial fraction
expansions. A simple proof for finding ‘B’ above is given in this box.
1 - = A B - ----------
C-
----------------------------------- --- + ---------- +
( s + 1 ) ( s + 2 )s s s+1 s+2
1 A B C
( s + 1 ) ------------------------------------ = ( s + 1 ) --- + ( s + 1 ) ----------- + ( s + 1 ) -----------
( s + 1 ) ( s + 2 )s s s+1 s+2
1 - A C
------------------ = ( s + 1 ) --- + B + ( s + 1 ) -----------
( s + 2 )s s s+2

1 A C
lim ------------------- = lim ( s + 1 ) --- + lim B + lim ( s + 1 ) -----------
s → –1 ( s + 2 )s s → –1 s s → –1 s → –1 s +2

1
lim ------------------- = lim B = B
s → – 1 ( s + 2 )s s → –1

1 0.5 –1 0.5
x ( s ) = ---------------------------------
2
= ------- + ----------- + -----------
( s + 3s + 2 )s s s + 1 s +2

Figure 17.16 Partial fractions to reduce an output function

After simplification with partial fraction expansion, the output function is easily
 numerical methods - 17.14

converted back to a function of time as shown in Figure 17.17.

–1 –1 0.5 – 1 - ----------
0.5-
x(t) = L [ x( s)] = L ------- + ---------- +
s s+1 s+2
–1 0.5 –1 - – 1 0.5
x(t) = L ------- + L –1 ---------- + L -----------
s s+1 s+2
–t – 2t
x ( t ) = [ 0.5 ] + [ ( – 1 )e ] + [ ( 0.5 )e ]
–t – 2t
x ( t ) = 0.5 – e + 0.5e

Figure 17.17 Partial fractions to reduce an output function (continued)

17.5 INVERSE TRANSFORMS AND PARTIAL FRACTIONS

The flowchart in Figure 17.18 shows the general procedure for converting a func-
tion from the s-domain to a function of time. In some cases the function is simple enough
to immediately use a transfer function table. Otherwise, partial fraction expansion is nor-
mally used to reduce the complexity of the function.
 numerical methods - 17.15

Start with a function of ’s’.

NOTE: This does not apply
for transfer functions.

is the
function in
simplify the
yes the transform
function
tables?
no

can the
function be
simplified? yes

no

Use partial fractions to

break the function into
smaller parts

Match the function(s) to

the form in the table
and convert to a time
function

Done

Figure 17.18 The methodology for doing an inverse transform of an output function

Figure 17.19 shows the basic procedure for partial fraction expansion. In cases
where the numerator is greater than the denominator, the overall order of the expression
can be reduced by long division. After this the denominator can be reduced from a polyno-
mial to multiplied roots. Calculators or computers are normally used when the order of the
polynomial is greater than second order. This results in a number of terms with unknown
resides that can be found using a limit or algebra based technique.
 numerical methods - 17.16

start with a function that

has a polynomial numerator
and denominator

is the
order of the use long division to
numerator >= yes reduce the order of the
denominator? numerator

no
Find roots of the denominator
and break the equation into
partial fraction form with
unknown values

OR
use limits technique. use algebra technique
If there are higher order
roots (repeated terms)
then derivatives will be
required to find solutions

Done

Figure 17.19 The methodology for solving partial fractions

Figure 17.20 shows an example where the order of the numerator is greater than
the denominator. Long division of the numerator is used to reduce the order of the term
until it is low enough to apply partial fraction techniques. This method is used infrequently
because this type of output function normally occurs in systems with extremely fast
response rates that are infeasible in practice.
 numerical methods - 17.17

3 2
5s + 3s + 8s + 6-
x ( s ) = -------------------------------------------
2
s +4
This cannot be solved using partial fractions because the numerator is 3rd order
and the denominator is only 2nd order. Therefore long division can be used to
reduce the order of the equation.

5s + 3

2 3 2
s +4 5s + 3s + 8s + 6
3
5s + 20s
2
3s – 12s + 6
2
3s + 12
– 12s – 6
This can now be used to write a new function that has a reduced portion that can be
solved with partial fractions.

– 12s – 6- –---------------------
12s – 6- A B
x ( s ) = 5s + 3 + ---------------------
2
solve
2
= ------------- + -------------
s +4 s +4 s + 2j s – 2j

Figure 17.20 Partial fractions when the numerator is larger than the denominator

Partial fraction expansion of a third order polynomial is shown in Figure 17.21.

The s-squared term requires special treatment. Here it produces partial two partial fraction
terms divided by s and s-squared. This pattern is used whenever there is a root to an expo-
nent.
 numerical methods - 17.18

1 - A B C
x ( s ) = --------------------
2
= ---2- + --- + -----------
s (s + 1) s s s+1

C = lim ( s + 1 )  --------------------
1 -
= 1
s → –1  2 
s (s + 1)

A = lim s  --------------------
2 1 - 1
= lim ----------- = 1
s→0  2  s→0 s + 1
s (s + 1)

B = lim ----- s  --------------------

1 -
= lim -----  -----------
d 2 d 1 –2
= lim [ – ( s + 1 ) ] = – 1
s→0 ds  2  s→0 ds  s + 1 s→0
s (s + 1)

Figure 17.21 A partial fraction example

Figure 17.22 shows another example with a root to an exponent. In this case each
of the repeated roots is given with the highest order exponent, down to the lowest order
exponent. The reader will note that the order of the denominator is fifth order, so the
resulting partial fraction expansion has five first order terms.

5
F ( s ) = -----------------------3-
2
s (s + 1)
5 - A B C D E
----------------------- = ----2 + --- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) (s + 1)
s (s + 1) s

Figure 17.22 Partial fractions with repeated roots

Algebra techniques are a reasonable alternative for finding partial fraction resi-
dues. The example in Figure 17.23 extends the example begun in Figure 17.22. The equiv-
alent forms are simplified algebraically, until the point where an inverse matrix solution is
used to find the residues.
 numerical methods - 17.19

5 -
----------------------- A B C D E
= ----2 + --- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) ( s + 1)
s (s + 1) s
3 3 2 2 2 2
A ( s + 1 ) + Bs ( s + 1 ) + Cs + Ds ( s + 1 ) + Es ( s + 1 ) -
= -----------------------------------------------------------------------------------------------------------------------------------------
3
2
s (s + 1)
4 3 2
s ( B + E ) + s ( A + 3B + D + 2E ) + s ( 3A + 3B + C + D + E ) + s ( 3A + B ) + ( A )-
= --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
3
2
s (s + 1)
–1
0 1 0 0 1 A 0 A 0 1 0 0 1 0 5
1 3 0 1 2 B 0 B 1 3 0 1 2 0 – 15
3 3 1 1 1 C = 0 C = 3 3 1 1 1 0 = 5
3 1 0 0 0 D 0 D 3 1 0 0 0 0 10
1 0 0 0 0 E 5 E 1 0 0 0 0 5 15

5 - 5 – 15 5 10 15
----------------------- = ---2- + --------- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) ( s + 1)
s (s + 1) s

Figure 17.23 Solving partial fractions algebraically

For contrast, the example in Figure 17.23 is redone in Figure 17.24 using the limit
techniques. In this case the use of repeated roots required the differentiation of the output
function. In these cases the algebra techniques become more attractive, despite the need to
solve simultaneous equations.
 numerical methods - 17.20

5 -
----------------------- A B C D E
= ---2- + --- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) ( s + 1)
s (s + 1) s

A = lim  -----------------------3- s = lim ------------------3- = 5

5 2 5
s→0  2  s → 0 (s + 1)
s (s + 1)

5( –3 )
B = lim -----  -----------------------3- s = lim -----  ------------------3-
d 5 2 d 5
= lim ------------------4- = – 15
s→0 ds 2 s→0 ds (s + 1) s → 0 (s + 1)
s (s + 1)

C = lim  -----------------------3- ( s + 1 ) = lim ----2 = 5

5 3 5
s → –1 2 s → –1 s
s (s + 1)
1 – 2 ( 5 -)
D = lim ----- -----  -----------------------3- ( s + 1 ) = lim ----- ----- ---2- = lim ----- -------------
1 d 5 3 1 d 5
= 10
s → – 1 1! ds 2 s → – 1 1! ds s s → – 1 1! s 3
s (s + 1)

E = lim ----- -----  -----------------------3- ( s + 1 ) = lim ----- ----- ----2 = lim ----- -----4- = 15
1 d2 5 3 1 d25 1 30
s → –1 2! ds  2  s → –1 2! ds s → –1 2!
s (s + 1) s s
5 - 5 – 15 5 10 15
----------------------- = ---2- + --------- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) (s + 1)
s (s + 1) s

Figure 17.24 Solving partial fractions with limits

An inductive proof for the limit method of solving partial fractions is shown in
Figure 17.25.

5 - A B C D E
----------------------- = ---2- + --- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) (s + 1)
s (s + 1) s
5 A B C D E
lim -----------------------3- = ---2- + --- + ------------------3- + ------------------2- + ----------------
s → –1 2 s (s + 1) (s + 1) (s + 1)
s (s + 1) s

lim ( s + 1 )  -----------------------3- = ---2- + --- + ------------------3- + ------------------2- + ----------------

3 5 A B C D E
s → –1  2 s (s + 1) ( s + 1 )
s (s + 1) s (s + 1)
3 3
5 A(s + 1) (s + 1)-
-+B
lim ---2- = ---------------------- ---------------------- + C + D(s + 1) + E(s + 1)
2
s → –1 s 2 s
s
 numerical methods - 17.21

For C, evaluate now,

3 3
5 A ( – 1 + 1 ) - ---------------------------
B(– 1 + 1 ) - 2
------------- = ---------------------------
2 2
+ + C + D ( – 1 + 1 ) + E ( – 1 + 1 )
( –1 ) ( –1 ) –1
3 3
5 A ( 0 ) - -------------
B( 0) - 2
------------- = -------------
2 2
+ + C + D(0) + E(0) C = 5
( –1 ) ( –1 ) – 1

For D, differentiate once, then evaluate

d  5 A( s + 1) - B 2
3 3
(s + 1)-
lim -----  ----2 = ---------------------- + ---------------------- + C + D ( s + 1 ) + E ( s + 1 ) 
s → – 1 dt  s s
2 s 

2 ( 5 )-
–-------------  2 ( s + 1 ) 3 3 ( s + 1 ) 2  ( s + 1 ) 3 3 ( s + 1 ) 2
lim = A– ---------------------
- + ---------------------
- - + ---------------------- + D + 2E ( s + 1 )
 + B  – ------------------
s → –1 s
3
 s
3
s
2
  s
2 s 
2 ( 5 )-
–-------------
3
= D = 10
( –1 )
For E, differentiate twice, then evaluate (the terms for A and B will be ignored to save
space, but these will drop out anyway).
2 5 ( )
3
( )
3 
 d  A s + 1 B s + 1
- + ----------------------- + C + D ( s + 1 ) + E ( s + 1 )2
lim  -----  ----2 = ----------------------
s → – 1 dt  s s
2 s 
d – 2 ( 5 -)
lim  -----  ------------- = A ( … ) + B ( … ) + D + 2E ( s + 1 )
s → – 1  dt  s 3 

–3 ( –2 ( 5 ) )
lim -------------------------
4
= A ( … ) + B ( … ) + 2E
s → –1 s
3 ( –2 ( 5 ) )
–-------------------------
4
= A ( 0 ) + B ( 0 ) + 2E E = 15
( –1 )

Figure 17.25 A proof of the need for differentiation for repeated roots

17.6 EXAMPLES

17.6.1 Mass-Spring-Damper Vibration

A mass-spring-damper system is shown in Figure 17.26 with a sinusoidal input.

 numerical methods - 17.22

Given,
1-
----
x(s) M
∴----------- = ---------------------------------
F( s) 2 K K
s + -----d- s + -----s
M M
Component values are,
N Ns
M = 1kg K s = 2 ---- K d = 0.5 ------
m m
The sinusoidal input is converted to the s-domain,

F ( t ) = 5 cos ( 6t )N
5s -
∴F ( s ) = ---------------
2 2
s +6
This can be combined with the transfer function to obtain the output function,
1---
 
x(s)
x ( s ) = F ( s )  ----------- =  ---------------
5s -  -----------------------------
-
1
 F ( s )  2 2  2 
s + 6  s + 0.5s + 2
5s
∴x ( s ) = ---------------------------------------------------------
2 2
-
( s + 36 ) ( s + 0.5s + 2 )

A B C D
∴x ( s ) = ------------- + ------------- + ------------------------------------- + -------------------------------------
s + 6j s – 6j s – 0.25 + 1.39j s – 0.25 – 1.39j

Figure 17.26 A mass-spring-damper example

The residues for the partial fraction in Figure 17.26 are calculated and converted to
a function of time in Figure 17.27. In this case the roots of the denominator are complex,
so the result has a sinusoidal component.
 numerical methods - 17.23

( s + 6j ) ( 5s )
------------------------------------------------------------------------ – 30j
- = ----------------------------------------------
-
A = lim
s → – 6j ( s – 6j ) ( s + 6j ) ( s + 0.5s + 2 )
2 ( – 12j ) ( 36 – 3j + 2 )
–3
∴A = 73.2 × 10 ∠3.05
* –3
∴B = A = 73.2 × 10 ∠– 3.05
Continue on to find C, D same way
–3 –3
73.2 × 10 ∠3.05 73.2 × 10 ∠– 3.05
∴x ( s ) = ------------------------------------------- + ---------------------------------------------- + …
s + 6j s – 6j

Do inverse Laplace transform

–3 – 0t
x ( t ) = 2 ( 73.2 × 10 )e cos ( 6t – 3.05 ) + …

Figure 17.27 A mass-spring-damper example (continued)

17.6.2 Circuits

It is not necessary to develop a transfer functions for a system. The equation for the
voltage divider is shown in Figure 17.28. Impedance values and the input voltage are con-
verted to the s-domain and written in the equation. The resulting output function is manip-
ulated into partial fraction form and the residues calculated. An inverse Laplace transform
is used to convert the equation into a function of time using the tables.
 numerical methods - 17.24

3
R = 10 Ω

t=0

V s = 3 cos t +
+ Vo
- C = 10 F
–3
-

As normal, relate the source voltage to the output voltage using component values in
the s-domain.
ZC 1
V o = Vs  -------------------
3s
V s ( s ) = -------------
2
- ZR = R Z C = ------
sC
ZR + ZC s +1
Next, equations are combined. The numerator of resulting output function must be
reduced by long division.
 ----- 1- 
3s  sC  3s 3s
-  ---------------- = -------------------------------------------
V o = ------------- - = ------------------------------------------------------- -
s + 1  R + ----- 1-
2 2 2 3 –3
( s + 1 ) ( 1 + sRC ) ( s + 1 ) ( s10 10 + 1 )
 sC
The output function can be converted to a partial fraction form and the residues calcu-
lated.
2 2
3s As + B- ---------- C- As + As + Bs + B + Cs + C-
V o = -----------------------------------
2
= ---------------
2
+ = ---------------------------------------------------------------------
2
(s + 1 )(s + 1) s +1 s+1 (s + 1)( s + 1)
2
3s s ( A + C ) + s ( A + B ) + ( B + C )-
V o = -----------------------------------
2
= ----------------------------------------------------------------------------
2
(s + 1)(s + 1) (s + 1)(s + 1)
B+C = 0 ∴B = – C
A+C = 0 ∴A = – C
A+B = 3 ∴– C – C = 3 ∴C = – 1.5 ∴A = 1.5 ∴B = 1.5
1.5s + 1.5- ---------- – 1.5-
V o = -----------------------
2
+
s +1 s+1

Figure 17.28 A circuit example

 numerical methods - 17.25

The output function can be converted to a function of time using the transform tables,
as shown below.

–1 –1 1.5s + 1.5- ----------

– 1.5- – 1 1.5s + 1.5 – 1.5-
Vo ( t ) = L [ V o ( s ) ] = L ----------------------- + - + L –1 ----------
= L -----------------------
2 s + 1 2 s +1
s +1 s +1
–1 s - –1 1 - –t
∴V o ( t ) = 1.5L -------------
2
+ 1.5L ------------- 2
– 1.5e
s +1 s +1
–t
∴V o ( t ) = 1.5 cos t + 1.5 sin t – 1.5e

1.5 + 1.5 sin  t + atan  -------  – 1.5e

2 2 1.5 –t
∴V o ( t ) =
  1.5 
π
∴V o ( t ) = 2.121 sin  t + --- – 1.5e
–t
4

Figure 17.29 A circuit example (continued)

17.7 ADVANCED TOPICS

17.7.1 Input Functions

In some cases a system input function is comprised of many different functions, as

shown in Figure 17.30. The step function can be used to switch function on and off to cre-
ate a piecewise function. This is easily converted to the s-domain using the e-to-the-s
functions.
 numerical methods - 17.26

f(t)

t
seconds
0 1 3 4

f ( t ) = 5tu ( t ) – 5 ( t – 1 )u ( t – 1 ) – 5 ( t – 3 )u ( t – 3 ) + 5 ( t – 4 )u ( t – 4 )

–s – 3s – 4s
5 5e - 5e 5e
f ( s ) = ---2- – ---------
2
– ------------
2
+ ------------
2
s s s s

Figure 17.30 Switching on and off function parts

17.7.2 Initial and Final Value Theorems

The initial and final values an output function can be calculated using the theorems
shown in Figure 17.31.

x ( t → ∞ ) = lim [ sx ( s ) ] Final value theorem

s→0

1s 1
- = ------------------------------------- = 1---
1
∴x ( t → ∞ ) = lim ---------------------------------
2
= lim -------------------------
2 2
s → 0 ( s + 3s + 2 )s s → 0 s + 3s + 2 (0 ) + 3(0) + 2 2

x ( t → 0 ) = lim [ sx ( s ) ] Initial value theorem

s→∞

1(s) 1 1-
∴x ( t → 0 ) = lim --------------------------------- = -------------------------------------------
- = --- = 0
2
s → ∞ ( s + 3s + 2 )s 2
( ( ∞ ) + 3( ∞) + 2 ) ∞

Figure 17.31 Final and initial values theorems

 numerical methods - 17.27

17.8 A MAP OF TECHNIQUES FOR LAPLACE ANALYSIS

The following map is to be used to generally identify the use of the various topics
covered in the course.

System Model Some input

(differential equations) disturbance in
terms of time
Laplace Transform

Transfer
Function Input described with
Laplace Equation

We can figure
out from plots
Substitute
Experiments Root-Locus
for stability
Output Function
(Laplace form)
Bode and Phase Plots Phasor Transform
- Straight line
- Exact plot
Output Function
Equations for gain and phase (Laplace terms)
at different frequencies Partial Fraction
Approximate equations for Inverse Laplace
steady state vibrations

Steady state Time based

frequency response response
equation

Figure 17.32 A map of Laplace analysis techniques

 numerical methods - 17.28

17.9 SUMMARY

• Transfer and input functions can be converted to the s-domain

• Output functions can be calculated using input and transfer functions
• Output functions can be converted back to the time domain using partial frac-
tions.

17.10 PRACTICE PROBLEMS

1. Convert the following functions from time to laplace functions using the tables.

·· · ·
a) L[5] o) L [ x + 5x + 3x ], x ( 0 ) = 8, x ( 0 ) = 7
– 3t
b) L[ e ] p) d
L ----- sin ( 6t )
dt
– 3t d 32
L [ 5e ] L  ----- t

c) q)
dt
– 3t
d) L [ 5te ] t
r) L ∫0 y dt
e) L [ 5t ] 3 – 5t
s) L [ 3t ( t – 1 ) + e ]
2
f) L [ 4t ] t) L[u( t – 1) – u(t – 2)]
g) L [ cos ( 5t ) ] – 2t
u) L[e u( t – 2) ]
–( t + 1)
h) L[ 3( t – 1 ) + e ] –( t – 3 )
v) L[e u( t – 1) ]
– 3t
i) L [ 5e cos ( 5t ) ] – 3t
w) L [ 5e + u(t – 1) – u(t – 2)]
– 3t
j) L [ 5e cos ( 5t + 1 ) ] t–3
x) L [ cos ( 7t + 2 ) + e ]
k) L [ sin ( 5t ) ] y) L [ cos ( 5t + 1 ) ]
l) L [ sinh ( 3t ) ] – 2.7t
z) L [ 6e cos ( 9.2t + 3 ) ]
2
m) L [ t sin ( 2t ) ]
aa)
n) d 2 – 3t
L ----- t e
dt
 numerical methods - 17.29

2. Convert the following functions below from the laplace to time domains using the tables.
–1 1- –1 5--- – 4.5s
a) L ----------
s+1
g) L (1 – e )
s
–1 5- –1 4 + 3j - ----------------------
4 – 3j -
b) L ---------- h) L --------------------- +
s+1 s + 1 – 2j s + 1 + 2j
–1 6- –1 6- -------------6
c) L --- i) L --- + 2 -
2 4
s s s +9
–1 6- –1 6
d) L --- j) L -------------------------
-
3 2
s s + 5s + 6
–1 s+2 - –1 6
e) L -------------------------------- k) L ----------------------------------
-
( s + 3)( s + 4) 2
4s + 20s + 24
–1 6 -
-------------
f) L 2
s +6

3. Convert the following functions below from the laplace to time domains using partial fractions
and the tables.
3 2
–1 s+2 - –1 s----------------------------------------
+ 9s + 6s + 3
a) L -------------------------------- g) L
( s + 3)( s + 4) 3
s + 5s + 4s + 6
2

–1 –1 9s + 4-
b) L [ ] h) L ------------------
3
(s + 3)
–1 –1 9s + 4 -
c) L [ ] i) L -----------------------
3 3
s (s + 3)
2
–1 –1 s-------------------------
+ 2s + 1-
d) L [ ] j) L 2
s + 3s + 2
2
–1 6 –1 s-------------------------
+ 3s + 5-
e) L ---------------- k) L 2
2
s + 5s 6s + 6
2 2
–1 9s + 6s + 3 –1 s-------------------------
+ 2s + 3-
f) L ---------------------------------------- l) L 2
3 2
s + 5s + 4s + 6 s + 2s + 1

4. Convert the output function below Y(s) to the time domain Y(t) using the tables.
5 12 - ---------------------
3 - ----------------------
3 -
Y ( s ) = --- + ------------- + +
s s 2 + 4 s + 2 – 3j s + 2 + 3j
 numerical methods - 17.30

5. Convert the following differential equations to transfer functions.

·· ·
a) 5x + 6x + 2x = 5F
·
b) y + 8y = 3x
·
c) y – y + 5x = 0

6. Given the transfer function, G(s), determine the time response output Y(t) to a step input X(t).
4 Y( s)
G ( s ) = ----------- = ----------- X ( t ) = 20 When t >= 0
s+2 X(s)

7. Given the following input functions and transfer functions, find the response in time.
Transfer Function Input
x ( s )-
= ---------------------------------  ----
s+2 m
a) ---------- F ( t ) = 5N
F(s) ( s + 3 ) ( s + 4 ) N 
x ( s )-
= ---------------------------------  ----
s+2 m
b) ---------- x ( t ) = 5m
F(s) (s + 3)(s + 4) N

8. Do the following conversions as indicated.

–4t
a) L [ 5e cos ( 3t + 2 ) ] =
– 2t
b) L[ e + 5t ( u ( t – 2 ) – u ( t ) ) ] =

d 3 y0 = 1 y 0' = 2
L  ----- y + 2  ----- y + y =
c) d where at t=0
dt dt
y 0'' = 3 y 0''' = 4

d) –1 1 + j - ---------------------
1–j -
L ---------------------- + =
s + 3 + 4j s + 3 – 4j
–1 1 3
e) L s + ----------- + ----------------------------
2
- =
s + 2 s + 4s + 40

9. Convert the output function to functions of time.

a) 3 2
s----------------------------------------
+ 4s + 4s + 4
3
s + 4s
b) 2
s +4
------------------------------------------------------------------
-
4 3 2
s + 10s + 35s + 50s + 24

10. Solve the differential equation using Laplace transforms. Assume the system starts unde-
flected and at rest.
··· ·· ·
θ + 40θ + 20θ + 2θ = 4
 numerical methods - 17.31

17.11 PRACTICE PROBLEM SOLUTIONS

1.
5--- 0.5- ----------
0.5-
a) l) ---------- –
s s–3 s+3
2 2
–4 16s - 12s – 16
1- m) --------------------- + -------------------- = ----------------------3-
b) ---------- 2 2 2 3 2
s+3 (s + 4) (s + 4) (s + 4)

5- n) 2s -
c) ---------- ------------------
3
s+3 (s + 3)
5 -
------------------ o) 2
( s x – 7s – 8 ) + 5 ( sx – 7 ) + 3x
d) 2
(s + 3)
6s
5 p) -----------------
e) ---- 2
2 s + 36
s
f) 8
---- q) 2
3
s
s r) y--
g) -----------------
2 s
s + 25
------ – 18
72 1 -
–1 s) ------ + ----------
h) 5 4 s+5
---- – 3--- + ----------
3 e - s s
2 s s+1
s –s – 2s
5(s + 3) t) e------- e---------
i) ------------------------------ –
2 2 s s
(s + 3) + 5
4 – 2s
j) 2.5 ∠1 - ----------------------
2.5 ∠– 1 - u) e
-------------
--------------------- +
s + 3 – 5j s + 3 + 5j s+2
s( ) + e -
2–s
= ----------------------------
2
- v) ----------
s + 6s + 34 s+1
–s – 2s
5 -
---------------- 5 - e------- e---------
k) w) ---------- + –
2 s+3 s s
s + 25
–3 –3
x) cos ( 2 ) s – sin ( 2 ) 7 e – 0.416s – 6.37 e
-------------------------------------------- + ----------- = ------------------------------------- + -----------
2 s–1 2 s–1
s + 49 s + 49
s----------------------------------
cos 1 – 5 sin 1
y) 2
s + 25
3 ∠3 - -------------------------------
3 ∠– 3 - s( ) +
------------------------------ + = ----------------------------------------
-
z) s + 2.7 – 9.2j s + 2.7 + 9.2j 2
s + 5.4s + 91.93
 numerical methods - 17.32

2.
–t
a) e
g) 5 – 5u ( t – 4.5 )
–t
b) 5e
cos  2t + atan  – --- 
– ( 1 )t 3
h) 2 ( 5 )e
4
c) 6t 3
i) t + 2 sin ( 3t )
2
d) 3t
j) – 2t – 3t
6e – 6e
– 3t – 4t
e) – e + 2e
– 2t – 3t
k) 1.5e – 1.5e
f) 6 sin ( 6t )

3.

– 3t – 4t
a) –e + 2e g)

b) h)

c) i)
– 2t
d) j) δ(t) – e
– 5t δ( t)
e) 1.2 – 1.2e k) --------- + 0.834 cos ( t + 0.927 )
6
– 4.4t – 0.3t –t
f) 8.34e + 2 ( 0.99 )e cos ( 1.13t + 1.23 ) l) δ ( t ) + 2te

4.
– 2t
y ( t ) = 5 + 6 sin ( 2t ) + 2 ( 3 )e cos ( 3t – 0 )

5.
--x- = ----------------------------
5
a) -
F 2
5s + 6s + 2
y-- 3
b) = -----------
x s+8
y-- –5
c) = -----------
x s–1
 numerical methods - 17.33

6.
– 2t
y ( t ) = 40 – 40e

7.
5--- 5--- – 3t 5--- – 4t
a) + e – e
6 3 2
– 2t
b) ( 5δ ( t ) + 30 – 5e )N

8.
a)
L [ 5e
–4 t
cos ( 3t + 2 ) ] = L [ 2 A e
– αt
cos ( βt + θ ) ] α = 4 β = 3
A = 2.5 θ = 2
A = 2.5 cos 2 + 2.5j sin 2 = – 1.040 + 2.273j
complex conjugate
A A – 1.040 + 2.273j – 1.040 – 2.273j
----------------------- + -------------------------------------- = --------------------------------------- + ---------------------------------------
s + α – βj s + α + βj s + 4 – 3j s + 4 + 3j

b)
– 2t – 2t
L[e + 5t ( u ( t – 2 ) – u ( t ) ) ] = L [ e ] + L [ 5tu ( t – 2 ) ] – L [ 5tu ( t ) ]
1 5 1 5
= ----------- + 5L [ tu ( t – 2 ) ] – ----2 = ----------- + 5L [ ( t – 2 )u ( t – 2 ) + 2u ( t – 2 ) ] – ----2
s+2 s s+2 s
1 5
= ----------- + 5L [ ( t – 2 )u ( t – 2 ) ] + 10L [ u ( t – 2 ) ] – ---2-
s+2 s
1 – 2s – 2s 5
= ----------- + 5e L [ t ] + 10e L [ 1 ] – ----2
s+2 s
– 2s – 2s
1 5e 10e 5
= ----------- + ------------
2
+ --------------- – ----2
s+2 s s s

c) 3
 ----
d 3 2 1
- y = s y + 1s + 2s + 3s
0
 dt
 ----
d- 1 0
 dt y
= s y+s 1

d 3
L  ----- y + 2  ----- y + y = ( s y + 1s + 2s + 3 ) + ( sy + 1 ) + ( y )
d 3 2
dt dt
3 2
= y ( s + s + 1 ) + ( s + 2s + 4 )
 numerical methods - 17.34

d) complex conjugate
–1 1 + j - ---------------------
1–j - –1 A A
L ---------------------- + = L ----------------------- + --------------------------------------
s + 3 + 4j s + 3 – 4j s + α – βj s + α + βj

π
θ = atan  ------ = – ---
–1
A =
2
1 + 1 = 1.141
2 α = 3 β = 4
1 4
π
cos  4t – ---
– αt – 3t
= 2Ae cos ( βt + θ ) = 2.282e
4

e) –1 1 3 1-
- = L [ s ] + L ---------- 3
L s + ----------- + ----------------------------
2
+ L ----------------------------
2
-
s + 2 s + 4s + 40 s+2 s + 4s + 40
d – 2t 3 d- – 2t 6
= ----- δ ( t ) + e + L ------------------------------
2
- = ---- δ ( t ) + e + 0.5L ------------------------------
2
-
dt ( s + 2 ) + 36 dt ( s + 2 ) + 36
d – 2t – 2t
= ----- δ ( t ) + e + 0.5e sin ( 6t )
dt

9.
a) 3 2 1
s + 4s + 4s + 4
---------------------------------------- 3 3 2
3
s + 4s s + 4s s + 4s + 4s + 4
3
– ( s + 4s )
2
4s + 4
2 2 A = 1
4s + 4- A Bs + C- s ( A + B ) + s ( C ) + ( 4A )
= 1 + ---------------- = 1 + --- + --------------- = 1 + -----------------------------------------------------------
3
s + 4s s s +4 2 3
s + 4s C = 0
B = 3
1 3s -
= 1 + --- + -------------
s s2 + 4
= δ ( t ) + 1 + 3 cos ( 2t )
 numerical methods - 17.35

b) 2
s +4
------------------------------------------------------------------ A - ----------
- = ---------- B - ----------
C - ----------
D-
4 3 2
+ + +
s + 10s + 35s + 50s + 24 s+1 s+2 s+3 s+4
2
A = lim  -------------------------------------------------- = ---
s +4 5
s → –1 ( s + 2 ) ( s + 3 ) ( s + 4 ) 6

2
B = lim  -------------------------------------------------- = ------
s +4 8
s → –2 ( s + 1 ) ( s + 3 ) ( s + 4 ) –2

2
C = lim  -------------------------------------------------- = ------
s +4 13
s → –3 ( s + 1 ) ( s + 2 ) ( s + 4 ) 2

2
D = lim  -------------------------------------------------- = ------
s +4 20
s → –4 ( s + 1 ) ( s + 2 ) ( s + 3 ) –6
5--- –t – 2t 13 – 3t 10 – 4t
e – 4e + ------ e – ------ e
6 2 3

10.
– 6 – 39.50t 0.1383t – 0.3368t
θ ( t ) = – 66 ⋅ 10 e – 3.216e + 1.216e + 2.00

17.12 ASSIGNMENT PROBLEMS

1. Prove the following relationships.

a) L f  ---
t
= aF ( as ) d) lim f ( t ) = lim sF ( s )
a t→∞ s→0

L [ f ( at ) ] = --- F  --- lim f ( t ) = lim sF ( s )

1 s e)
b)
a  a t→∞ s→0

d
c) L[ e
– at
f(t )] = F(s + a) f) L [ tf ( t ) ] = – ----- F ( s )
dt
 numerical methods - 17.36

2. The applied force ‘F’ is the input to the system, and the output is the displacement ‘x’.
a) find the transfer function.
K1 = 500 N/m

x K2 = 1000 N/m

M = 10kg

F
b) What is the steady state response for an applied force F(t) = 10cos(t + 1) N ?
c) Give the transfer function if ‘x’ is the input.
d) Find x(t), given F(t) = 10N for t >= 0 seconds using Laplace methods.

3. The following differential equation is supplied, with initial conditions.

·· · ·
y + y + 7y = F y( 0) = 1 y(0) = 0

F ( t ) = 10 t>0
a) Solve the differential equation using calculus techniques.
b) Write the equation in state variable form and solve it numerically.
c) Find the frequency response (gain and phase) for the transfer function using the
phasor transform. Roughly sketch the bode plots.
d) Convert the differential equation to the Laplace domain, including initial condi-
tions. Solve to find the time response.

4. Given the transfer functions and input functions, F, use Laplace transforms to find the output of
the system as a function of time. Indicate the transient and steady state parts of the solution.
2
--x- = -----------------------------
D -
2
F = 5 sin ( 62.82t )
F ( D + 200π )

D ( D + 2π )-
--x- = ----------------------------- F = 5 sin ( 62.82t )
F 2
( D + 200π )
 numerical methods - 17.37

2
D ( D + 2π )-
--x- = ----------------------------- F = 5 sin ( 62.82t )
F 2
( D + 200π )

17.13 REFERENCES

Irwin, J.D., and Graf, E.R., Industrial Noise and Vibration Control, Prentice Hall Publishers,
1979.

Close, C.M. and Frederick, D.K., “Modeling and Analysis of Dynamic Systems, second edition,
John Wiley and Sons, Inc., 1995.
 linear feedback controller analysis - 18.1

18. CONTROL SYSTEM ANALYSIS

Topics:

Objectives:

18.1 INTRODUCTION

18.2 CONTROL SYSTEMS

• Control systems use some output state of a system and a desired state to make
control decisions.

• In general we use negative feedback systems because,

- they typically become more stable

- they become less sensitive to variation in component values
- it makes systems more immune to noise

• Consider the system below, and how it is enhanced by the addition of a control
system.
 linear feedback controller analysis - 18.2

Control variable

INPUT OUTPUT
(e.g. θgas) SYSTEM
(e.g. velocity)
(e.g. a car)

vdesired verror Driver or θgas vactual

car
+ cruise control
_

The control system is in the box and could be a driver or a cruise control
(this type is known as a feedback control system)

Figure 18.1 An example of a feedback controller

Human rules to control car (also like expert system/fuzzy logic):

1. If verror is not zero, and has been positive/negative for a while, increase/decrease θgas
2. If verror is very big/small increase/decrease θgas
3. If verror is near zero, keep θgas the same
4. If verror suddenly becomes bigger/smaller, then increase/decrease θgas.
5. etc.

Figure 18.2 Rules for a feedback controller

 linear feedback controller analysis - 18.3

• Some of the things we do naturally (like the rules above) can be done with math-
ematics

18.2.1 PID Control Systems

• The basic equation for a PID controller is shown below. This function will try to
compensate for error in a controlled system (the difference between desired and actual
output values).

u = K c e + K i ∫ edt + K d  ------
de
 dt 

Figure 18.3 The PID control equation

• The figure below shows a basic PID controller in block diagram form.

proportional PID Controller

Kp ( e )
V V
+ +V
integral
+ e u
Ki ( ∫ e ) amp motor
+
- derivative + -V
K d  ----- e
d
dt

Figure 18.4 A block diagram of a feedback controller

 linear feedback controller analysis - 18.4

e.g.

dv error
θ gas = K c v error + K i ∫ v error dt + K d  ----------------
 dt 

Rule 4
Rules 2 & 3
(Immediate error)
(general difference)
Rule 1
(Long term error)

Kc
Ki Relative weights of components
Kd

This is a PID Controller

Proportional
Integral
Derivative

For a PI Controller
θ gas = K c v error + K i ∫ v error dt

For a P Controller
θ gas = K c v error

For a PD Controller
dv error
θ gas = K c v error + K d  ----------------
 dt 

• The PID controller is the most common controller on the market.

 linear feedback controller analysis - 18.5

18.2.2 Analysis of PID Controlled Systems With Laplace Transforms

1. We can rewrite the control equation as a ratio of output to input.

dv error
θ gas = K c v error + K i ∫ v error dt + K d  ----------------
 dt 

θ gas
- = K c + K i ∫ dt + K d  ----
d-
------------
v error  dt

Then do a Laplace transform

d- dx
---- →s ------ → sx
dt dt
1 x
∫ dt = ---
s ∫ xdt = --
s

θ gas K The transfer function

- = K c + -----i + K d s
L ------------
v error s
 linear feedback controller analysis - 18.6

2. We can also develop a transfer function for the car.

F = Aθ gas = 10θ gas

F- Transfer function for engine and transmission. (Laplace

--------- = 10
θ gas transform would be the same as initial value.)

2
d x dv
F = Ma = M -------2- = M ------
dt dt

F d-
--- = M ----
v dt

v 1
L --- = ------- Transfer function for acceleration of car mass
F Ms

3. We want to draw the system model for the car.

θgas F 1- vactual
10 ------
Ms

• The ‘system model’ is shown above.

• If θgas is specified directly, this is called ‘open loop control’. This is not
desirable, but much simpler.
• The two blocks above can be replaced with a single one.

θgas 10- vactual

------
Ms
 linear feedback controller analysis - 18.7

4. If we have an objective speed, and an actual speed, the difference is the

‘system error’

v error = v desired – v actual

‘set-point’ - desired system operating point

5. Finally, knowing the error is verror, and we can control θgas (the control variable),
we can select a control system.

verror θgas
Controller

*The coefficients can be calculated using

θ gas Ki classical techniques, but they are more
L ------------- = K c + ----- + K d s
v error s commonly approximated by trial and
error.

6. For all the components we can now draw a ‘block diagram’

vdesired verror Ki θgas F 1- vactual

K c + ----- + K d s 10 ------
+ s Ms
-

• A ‘negative feedback loop

that is the fundamental part of this
‘closed loop control system’
 linear feedback controller analysis - 18.8

18.2.3 Finding The System Response To An Input

• Even though the transfer function uses the Laplace ‘s’, it is still a ratio of input to
output.

• Find an input in terms of the Laplace ‘s’

vdesired
100 step
vdesired(t) = 100 for t >= 0 sec
ramp
vdesired(t) = 50t for t >= 0 sec

t(sec)
0

Input type Time function Laplace function

A
STEP f ( t ) = Au ( t ) f ( s ) = ---
s

A
RAMP f ( t ) = Atu ( t ) f ( s ) = ----2
s
2
Aω
SINUSOID f ( t ) = A sin ( ωt )u ( t ) f ( s ) = -----------------
2 2
s +ω

PULSE f ( t ) = A ( u ( t ) – u ( t – t1 ) ) f( s) =

etc......
 linear feedback controller analysis - 18.9

Therefore to continue the car example, lets assume the input below,

v desired ( t ) = 100 t ≥ 0 sec

100
v desired ( s ) = L [ v desired ( t ) ] = ---------
s

Next, lets use the input, and transfer function to find the output of the system.

v actual
v actual =  ----------------- v desired
v desired

 2 
 s ( K d ) + s ( K c ) + K i   100
v actual =  -------------------------------------------------------------  ---------
 s 2  -----
M   s
  10- + K d + s ( K c ) + K i

To go further, some numbers will be selected for the values.

Kd = 10000
Kc = 10000
Ki = 1000
M = 1000

 s 2 ( 10000 ) + s ( 10000 ) + 1000  100

v actual =  -----------------------------------------------------------------------
- ---------
 s 2 ( 10100 ) + s ( 10000 ) + 1000  s 
 linear feedback controller analysis - 18.10

At this point we have the output function, but not in terms of time yet. To do this
we break up the function into partial fractions, and then find inverse Laplace
transforms for each.

2 
2
s + s + 0.1
v actual = 10  -------------------------------------------------
-
 s ( s ( 1.01 ) + s + 0.1 )
2

Aside: We must find the roots of the equation, before we

can continue with the partial fraction expansion.

recall the quadratic formula,

2
2 – b ± b – 4ac
ax + bx + c = 0 x = --------------------------------------
2a

2
– 1 ± 1 – 4 ( 1.01 ) ( 0.1 )
∴x = ------------------------------------------------------------ = – 0.113, – 0.877
2 ( 1.01 )

2 2
v actual = ----------  ----------------------------------------------------
10 s + s + 0.1
1.01 s ( s + 0.113 )s + 0.877

A B C
v actual = --- + ---------------------- + ----------------------
s s + 0.114 s + 0.795
 linear feedback controller analysis - 18.11

2 2 2
A = lim s  ----------  ---------------------------------------------------------  = ----------  -------------------------------------
10 s + s + 0.1 10 0.1
s→0 1.01 s ( s + 0.113 ) ( s + 0.877 ) 1.01 ( 0.113 ) ( 0.877 )

∴A = 99.9

2 2
B = lim  ---------
10 -  --------------------------------------------------------
s + s + 0.1 -  ( s + 0.113 )
s → – 0.113  1.01  s ( s + 0.113 ) ( s + 0.877 ) 

2 2
10 ( – 0.113 ) + ( – 0.113 ) + 0.1
∴B =  ----------  -----------------------------------------------------------------  = 0.264
1.01 ( – 0.113 ) ( – 0.113 + 0.877 )

2 2
C = lim  ---------
10  s + s + 0.1 -  ( s + 0.877 )
- --------------------------------------------------------
s → – 0.877  1.01  s ( s + 0.113 ) ( s + 0.877 ) 

2 2
10 ( – 0.877 ) + ( – 0.877 ) + 0.1
∴C =  ----------  -----------------------------------------------------------------  = – 1.16
1.01 ( – 0.877 ) ( – 0.877 + 0.113 )

99.9 0.264 1.16

v actual = ---------- + ---------------------- – ----------------------
s s + 0.113 s + 0.877
 linear feedback controller analysis - 18.12

Next we use a list of forward/inverse transforms to replace the terms in the

partial fraction expansion.

f(t ) f(s )

A
---
A
s

A
At ----
2
s

– αt A -
Ae -----------
s+α
Aω
A sin ( ωt ) -----------------
2 2
s +ω

2
– ξω n t 2 ωn 1 – ξ for ( ξ < 1 )
e sin ( ω n t 1 – ξ ) ---------------------------------------
2
-
2
s + 2ξω n s + ω n

etc.

To finish the problem, we simply convert each term of the partial fraction back to
the time domain.

99.9 0.264 1.16

v actual = ---------- + ---------------------- – ----------------------
s s + 0.113 s + 0.877

– 0.113t – 0.877t
v actual = 99.9 + 0.264e – 1.16e
 linear feedback controller analysis - 18.13

18.2.4 Controller Transfer Functions

• The table below is for typical control system types,

Type Transfer Function

Proportional (P) Gc = K

G c = K  1 + -----
1
Proportional-Integral (PI)  τs

Proportional-Derivative (PD) G c = K ( 1 + τs )

G c = K  1 + ----- + τs
1
Proportional-Integral-Derivative (PID)
 τs 

1 + ατs
G c = K  ------------------ α>1
1 + τs
Lead
1 + τs
G c = K  ------------------ α>1
Lag 1 + ατs
1 + τ 1 s 1 + ατ 2 s α>1
G c = K  ---------------------  ---------------------
Lead-Lag 1 + ατ 1 s 1 + τ 2 s τ1 > τ2

18.3 ROOT-LOCUS PLOTS

• Consider the basic transform tables. A superficial examination will show that the
denominator (bottom terms) are the main factor in determining the final form of the solu-
tion. To explore this further, consider that the roots of the denominator directly impact the
partial fraction expansion and the following inverse Laplace transfer.

• When designing a controller with variable parameters (typically variable gain),

we need to determine if any of the adjustable gains will lead to an unstable system.

• Root locus plots allow us to determine instabilities (poles on the right hand side
of the plane), overdamped systems (negative real roots) and oscillations (complex roots).
 linear feedback controller analysis - 18.14

• Note: this procedure can take some time to do, but the results are very important
when designing a control system.

• Consider the example below,

 linear feedback controller analysis - 18.15

Note: This controller has adjustable gain. After this design

is built we must anticipate that all values of K will be
used. It is our responsibility to make sure that none of
the possible K values will lead to instability.

+ 1---
K
s
-

K H(s) = 1
G ( s ) = ----
s
First, we must develop a transfer function for the entire control system.
K ----
G ( s )  s K
G S ( s ) = --------------------------------- = --------------------------- = ------------
1 + G ( s )H ( s ) s+K
1 +  ---- ( 1 )
K
s
Next, we use the characteristic equation of the denominator to find the roots as
the value of K varies. These can then be plotted on a complex plane. Note:
the value of gain ’K’ is normally found from 0 to +infinity.

s+K = 0 K root
jω
0
1 K→∞ K = 0 σ
2
3
etc..

Note: because all of the roots for all values of K are real negative this system will
always be stable, and it will always tend to have a damped response. The large the
value of K, the more stable the system becomes.

• Consider the previous example, the transfer function for the whole system was
found, but then only the denominator was used to determine stability. So in general we do
not need to find the transfer function for the whole system.
 linear feedback controller analysis - 18.16

Consider the general form for a negative feedback system.

G(s) Note: two assumptions that are not often clearly

G S ( s ) = ---------------------------------
1 + G ( s )H ( s ) stated are that we are assuming that the control
system is a negative feedback controller, and
that when not given the feedback gain is 1.

The system response is a function of the denominator, and it’s roots.

1 + G ( s )H ( s ) = 0
It is typical, (especially in textbook problems) to be given only G(s) or G(s)H(s).

The transfer function values will often be supplied in a pole zero form.
K ( s + z0 ) ( s + z1 )… ( s + z m )
G ( s )H ( s ) = ------------------------------------------------------------------
-
( s + p 0 ) ( s + p 1 )… ( s + p n )

• Consider the example,

 linear feedback controller analysis - 18.17

Given the system elements (you should assume negative feedback),

K H(s) = 1
G ( s ) = -------------------------
2
-
s + 3s + 2

First, find the characteristic equation,. and an equation for the roots,

1 +  -------------------------
- ( 1 ) = 0
K
2 
s + 3s + 2
2
s + 3s + 2 + K = 0

– 3 ± 9 – 4(2 + K) 1 – 4K
roots = ------------------------------------------------ = – 1.5 ± --------------------
2 2
Next, find values for the roots and plot the values,
K root
jω
0 0
1 -1 σ
2 -2
3 -3

******CALCULATE AND PUT IN NUMBERS

18.3.1 Approximate Plotting Techniques

• The basic procedure for creating root locus plots is,

1. write the characteristic equation. This includes writing the poles and zeros of the
 linear feedback controller analysis - 18.18

equation.

( s + z 1 ) ( s + z 2 )… ( s + z m )
1 + G ( s )H ( s ) = 1 + K ---------------------------------------------------------------- = 0
( s + p1 ) ( s + p 2 )… ( s + pn )

2. count the number of poles and zeros. The difference (n-m) will indicate how
many root loci lines end at infinity (used later).
3. plot the root loci that lie on the real axis. Points will be on a root locus line if
they have an odd number of poles and zeros to the right. Draw these lines in.
4. determine the asymptotes for the loci that go to infinity using the formula below.
Next, determine where the asymptotes intersect the real axis using the second
formula. Finally, draw the asymptotes on the graph.

± 180° ( 2k + 1 )
β ( k ) = ----------------------------------- k ∈ [ 0, n – m – 1 ]
n–m

( p1 + p2 + … + pn ) ( z 1 + z2 + … + z m )
σ = --------------------------------------------------------------------------------------------
n–m
5. the breakaway and breakin points are found next. Breakaway points exist
between two poles on the real axis. Breakin points exist between zeros. to cal-
culate these the following polynomial must be solved. The resulting roots are
the breakin/breakout points.

A = ( s + p 1 ) ( s + p 2 )… ( s + p n ) B = ( s + z 1 ) ( s + z 2 )… ( s + z m )

 ----
d- 
A B – A  ----- B = 0
d
 ds   ds 

6. Find the points where the loci lines intersect the imaginary axis. To do this sub-
stitute the phasor for the laplace variable, and solve for the frequencies. Plot the
asymptotic curves to pass through the imaginary axis at this point.

( jω + z 1 ) ( jω + z 2 )… ( jω + z m )
1 + K --------------------------------------------------------------------------- = 0
( jω + p 1 ) ( jω + p2 )… ( jω + p n )

• Consider the example in the previous section,

 linear feedback controller analysis - 18.19

Given the system elements (you should assume negative feedback),

K H(s) = 1
G ( s ) = -------------------------
2
-
s + 3s + 2

Step 1: (put equation in standard form)

1 + G ( s )H ( s ) = 1 +  -------------------------
- ( 1 ) = 1 + K --------------------------------
K 1 -
2  ( s + 1 ) (s + 2)
s + 3s + 2
Step 2: (find loci ending at infinity)

m = 0 n = 2 (from the poles and zeros of the previous step)

n–m = 2 (loci end at infinity)
Step 3: (plot roots)
jω

σ
-2 -1

Step 4: (find asymptotes angles and real axis intersection)

180° ( 2k + 1 )
β ( k ) = -------------------------------- k ∈ I [ 0, 1 ] jω
2
180° ( 2 ( 0 ) + 1 )
β ( 0 ) = ------------------------------------- = 90°
2
180° ( 2 ( 1 ) + 1 ) σ
β ( 1 ) = ------------------------------------- = 270°
2 -2 -1
( 0 )( – 1 – 2)
σ = ----------------------------- = 0
2
asymptotes
 linear feedback controller analysis - 18.20

Step 5: (find the breakout points for the roots)

2 jω
A = 1 B = s + 3s + 2

d- d-
---- A = 0 ---- B = 2s + 3
ds ds
σ
A  ----- B – B  ----- A = 0
d d
 ds   ds  -2 -1
2
1 ( 2s + 3 ) – ( s + 3s + 2 ) ( 0 ) = 0 -1.5

2s + 3 = 0
s = – 1.5

Note: because the loci do not intersect the imaginary axis, we know the system will be
stable, so step 6 is not necessary, but we it will be done for illustrative purposes.

Step 6: (find the imaginary intercepts)

1 + G ( s )H ( s ) = 0
1
1 + K -------------------------
2
- = 0
s + 3s + 2
2
s + 3s + 2 + K = 0
2
( jω ) + 3 ( jω ) + 2 + K = 0
2
– ω + 3jω + 2 + K = 0
2
ω + ω ( – 3j ) + ( – 2 – K ) = 0
2
3j ± ( – 3j ) – 4 ( – 2 – K ) 3j ± – 9 + 8 + 4K 3j ± 4K – 1
ω = -------------------------------------------------------------- = ---------------------------------------------- = -------------------------------
2 2 2
In this case the frequency has an imaginary value. This means that there will be no
frequency that will intercept the imaginary axis.

• Plot the root locus diagram for the function below,

 linear feedback controller analysis - 18.21

K( s + 5 )
G ( s )H ( s ) = ---------------------------------
2
s ( s + 4s + 8 )

18.4 DESIGN OF CONTINUOUS CONTROLLERS

18.5 SUMMARY

•
 linear feedback controller analysis - 18.22

18.6 PRACTICE PROBLEMS

3. Given the transfer function below, and the input ‘x(s)’, find the output ‘y(t)’ as a function of
time.
( s )-
y--------- 5
= ----------- x( t) = 5 t ≥ 0 sec
x( s) s+2

8. Draw a detailed root locus diagram for the transfer function below. Be careful to specify angles
of departure, ranges for breakout/breakin points, and gains and frequency at stability limits.
2
2K ( s + 0.5 ) ( s + 2s + 2 )-
G ( s ) = -----------------------------------------------------------
3
s (s + 1)(s + 2)

10. Draw the root locus diagram for the transfer function below,
2
K(s + 4)-
G ( s ) = ----------------------
2
s (s + 1)

11. Draw the root locus diagram for the transfer function below,
K(s + 1)(s + 2)
G ( s ) = -------------------------------------
3
s

12. The block diagram below is for a motor position control system. The system has a propor-
tional controller with a variable gain K.

θd Vd + Ve Vs ω θa
2 K 100- 1---
----------
s+2 s
-
Va

a) Simplify the block diagram to a single transfer function.

ans. 200K
-----------------------------------
-
2
s + 2s + 200K
b) Draw the Root-Locus diagram for the system (as K varies). Use either the
approximate or exact techniques.
 linear feedback controller analysis - 18.23

ans. – 2 ± 4 – 4 ( 200K )
roots = ------------------------------------------------ = – 1 ± 1 – 200K
2
Im
K r
o
o K=0.005
ts
Re
0 -2 -1

0
0

c) Select a K value that will result in an overall damping coefficient of 1. State if the Root-Locus
diagram shows that the system is stable for the chosen K.

(ans. 2
s + 2s + 200K = s + 2ζωn s + ω n
2 2 ∴ω n = 1 ∴K = 0.005

From the root locus graph this value is critically stable.

13. Draw a Bode Plot for either one of the two transfer functions below.

( s + 1 ) ( s + 1000 -) 5-
----------------------------------------- OR ---
2 2
( s + 100 ) s

15. Given the system transfer function below.

θ 20K -
----o- = -----------------------------
θd 2
s + s + 20K
a) Draw the root locus diagram and state what values of K are acceptable.
b) Select a gain value for K that has either a damping factor of 0.707 or a natural
frequency of 3 rad/sec.
c) Given a gain of K=10 find the steady state response to an input step of 1 rad.
d) Given a gain of K=10 find the response of the system as

17. The equation below describes a dynamic system. The input is ‘F’ and the output is ‘V’. It has
the initial values specified. The following questions ask you to find the system response to a
 linear feedback controller analysis - 18.24

unit step input using various techniques.

V'' + 10V' + 20V = – 20F V(0) = 1 V' ( 0 ) = 2
a) Find the response using Laplace transforms.
b) Find the response using the homogenous and particular solutions.
c) Put the equation is state variable form, and solve it using your calculator. Sketch
the result accurately below.

18. A feedback control system is shown below. The system incorporates a PID controller. The
closed loop transfer function is given.
X + Y
Ki 3-
K p + ----- + K d s ----------
s s+9
-

4 2
s ( 3K d ) + s ( 3K p ) + ( 3K i )
--Y- = ------------------------------------------------------------------------------------
-
X 2
s ( 4 + 3K d ) + s ( 36 + 3K p ) + ( 3K i )
a) Verify the close loop controller function given.
b) Draw a root locus plot for the controller if Kp=1 and Ki=1. Identify any values
of Kd that would leave the system unstable.
c) Draw a Bode plot for the feedback system if Kd=Kp=Ki=1.
d) Select controller values that will result in a natural frequency of 2 rad/sec and
damping coefficient of 0.5. Verify that the controller will be stable.
e) For the parameters found in the last step find the initial and final values.
f) If the values of Kd=1 and Ki=Kd=0, find the response to a ramp input as a func-
tion of time.

19. The following system is a feedback controller for an elevator. It uses a desired heigh ‘d’ pro-
vided by a user, and the actual height of the elevator ‘h’. The difference between these two is
called the error ‘e’. The PID controller will examine the value ‘e’ and then control the speed of
the lift motor with a control voltage ‘c’. The elevator and controller are described with transfer
functions, as shown below. all of these equations can be combined into a single system transfer
 linear feedback controller analysis - 18.25

equation as shown.
e = d–h error
Ki 2 h--- 10 - elevator
c-- 2s + 1 + s PID controller = ------------
= K p + ----- + K d s = -------------------------- c 2
s +s
e s s
combine the transfer functions
2 2
 c--  h--- = h--- = -------------------------
2s + 1 + s - ------------ 10 - (s + 1) 10 10 ( s + 1 )
= ------------------- ------------------- = ----------------------
 e  c e s 2 s s(s + 1 ) 2
s +s s
h - 10 ( s + 1 ) eliminate ‘e’
----------- = ----------------------
d–h 2
s
( 1 )
h =  ----------------------
10 s +
(d – h)
 2 
s
10 ( s + 1 ) 10 ( s + 1 )
h  1 + ---------------------- =  ---------------------- (d)
 2   2 
s s
10 ( s + 1 ) 
 ----------------------
 2 
=  -------------------------------- = -------------------------------
h--- s 10s + 10 - system transfer function
d  10 ( s + 1 ) 2
 1 + ----------------------  s + 10s + 10
2
 s 
a) Find the response of the final equation to a step input. The system starts at rest
on the ground floor, and the input (desired height) changes to 20 as a step input.
b) Write find the damping coefficient and natural frequency of the results in part
a).
c) verify the solution using the initial and final value theorems.
 linear feedback controller analysis - 18.26

(ans. a) h 10s + 10 - 20
--- = ------------------------------- d ( t ) = 20u ( t ) d ( s ) = ------
d 2 s
s + 10s + 10

h =  -------------------------------
10s + 10 - 20 ------ = A B
--- + -------------------------------- C
- + --------------------------------
-
 2 
s + 10s + 10 s s s + 5 – 3.873j s + 5 + 3.873j

A = lim  -------------------------------
200s + 200 -
= 20
s → 0 s + 10s + 10
2

B = lim  ----------------------------------------
200s + 200 -
= – 2.5 – 22.6j
s → – 5 + 3.873j s ( s + 5 + 3.873j )


C = – 2.5 + 22.6j
2 2
–1 – 2.5 – 22.6j- --------------------------------
– 2.5 + 22.6j- A = 2.5 + 22.6 = 22.73
h ( t ) = 20 + L -------------------------------- +
s + 5 – 3.873j s + 5 + 3.873j
θ = 4.602
α = 5
β = 3.873
– 5t
h ( t ) = 20 + 2 ( 22.73 )e cos ( 3.873t + 4.602 )

5
b) –5 = – ωn ζ ζ = ------
ωn
2 25
3.873 = 1 – ζ ωn = 1 – -----2- ω n
ωn

 25  2 2 ωn = 35 = 5.916
15 =  1 – -----2- ω n = ω n – 25
 ω n
5
ζ = ---------- = 0.845
35

c)
h ( 0 ) = lim s  -------------------------------
10s + 10 - 20 ------ = lim  10s
-------- 20 = 0
s → ∞  s 2 + 10s + 10 s s → ∞  s2 

------ = (----------------
10 )20-
h ( ∞ ) = lim s  -------------------------------
10s + 10 - 20
= 20
s→0  2  s 10
s + 10s + 10
 linear feedback controller analysis - 18.27

18.7 PRACTICE PROBLEM SOLUTIONS

18.8 ASSIGNMENT PROBLEMS

1. a) The block diagram below is for an angular positioning system. The setpoint is a desired
angle, which is converted to a desired voltage. This is compared to a feedback voltage from a
potentiometer. A PID controller is used to generate an output voltage to drive a DC motor.
Simplify the block diagram.
PID Controller Motor
θd Vd e Ki Vs 100 - ω θa
1- K p + ----- + sK d --------------- 1---
--------- s s + 0.1
0.25 + s
-
Potentiometer
Va
1-
---------
0.25

b) Given the transfer function below, select values for Kp, Ki and Kd that will result in a second
order response that has a damping coefficient of 0.125 and a natural frequency of 10rad/s.
(Hint: eliminate Ki).
2
θa s ( 400K d ) + s ( 400K p ) + ( 400K i )
----- = ----------------------------------------------------------------------------------------------------------
θd 3 2
s + s ( 0.1 + 400K d ) + s ( 400K p ) + ( 400K i )

c) The function below has a step input of magnitude 1.0. Find the output as a function of time
using numerical methods. Give the results in a table OR graph.
θa s ( 0.9 ) + ( 4 ) -
----- = -----------------------------------
θd 2
s + 2.5s + 100

d) The function below has a step input of magnitude 1. Find the output as a function of time by
integrating the differential equation (i.e., using the homogeneous and particular solutions).
θa s ( 0.9 ) + ( 4 )
----- = ------------------------------------
θd 2
s + 2.5s + 100
 linear feedback controller analysis - 18.28

e) The function below has a step input of magnitude 1. Find the output as a function of time using
Laplace transforms.
θ s ( 0.9 ) + ( 4 ) -
----a- = -----------------------------------
θd 2
s + 2.5s + 100

f) Given the transfer function below; a) apply a phasor/Fourier transform and express the gain and
phase angle as a function of frequency, b) calculate a set of values and present them in a table,
c) use the values calculated in step b) to develop a frequency response plot on semi-log paper,
d) draw a straight line approximation of the Bode plot on semi-log paper.
θa ( 0.9 ) + ( 4 )-
----- = s----------------------------
θd 2
s +s+4

2. Select a controller transfer function, Gc, that will reduce the system to a first order system with
a time constant of 0.5s, as shown below.
Controller Motor
θd Vd e Vs 100 - ω θa
1- Gc --------------- 1---
--------- s + 0.1
0.25 + s
-
Potentiometer
Va
1-
---------
0.25

θa 1
desired transfer function ----- = -----------
θd s+2
 convolution - 19.1

19. CONVOLUTION

Topics:

Objectives:

19.1 INTRODUCTION

• Can be used to find normal input reponses for linear systems

• It is most useful for finding an output response for a system given an arbitrary
input function

• It is also the basis for other methods that come later for system analysis.

19.2 UNIT IMPULSE FUNCTIONS

• A unit impulse function

 convolution - 19.2

Unit impulse P(t)

Area = ε  --- = 1
1 1---
ε ε t

ε
Dirac delta function
δ ( t ) = lim P ( t ) This function is theoretically
ε→0
undefined, it goes to infinity
∞
for an instant, but it does
Area = ∫–∞ δ ( t ) dt = 1
have an area of 1.

• For a unit step function,

u(t < 0) = 0
u(t ≥ 0) = 1
d-
---- u( t) = δ ( t )
dt

• If we look at an input signal (force here) we can break it into very small segments
in time. As the time becomes small we can approximate it with a set of impulses.
 convolution - 19.3

F(t) ∆t → 0

An impulse

Figure 19.1 An impulse as a brief duration (instant) pulse

19.3 IMPULSE RESPONSE

• If we put an impulse into a system the output will be an impulse response.

F(t) x( t)

∫ g ( t )F ( t ) dt

t t

Figure 19.2 Response of the system to a single pulse

• If we add all of the impulse responses together we will get a total system
response. This operation is called convolution.
 convolution - 19.4

F( t) x(t)
sum of responses
t

∫ g ( t – τ )F ( τ ) dτ
0

t t
impulse responses
The convolution integral
t

c(t) = ∫ g ( t – τ )r ( τ ) dτ
0

Figure 19.3 A set of pulses for a system gives summed responses to give the output

• Consider the unit impulse of a system with the given differential equation. Note:
This method is only valid for trivial differential equations with only one homogeneous
term. The preferred method is shown later.
 convolution - 19.5

The following first order differential equation has an input ’F’. The input can be
replaced with a unit impulse function.
·
x + 0.5x = 2F
·
X + 0.5X = 2δ ( t )
The homogeneous solution can be found.
· guess, At · At
X + 0.5X = 0 Xh = e X h = Ae
A + 0.5 = 0
– 0.5 t
X h = Ce
The particular solution is found.
· ·
X + 0.5X = 2δ ( t ) guess, Xp = A Xp = 0
0 + 0.5A = 2 ( 0 )

Xp = A = 0
The initial condition caused by the impulse function found, assuming a zero initial con-
dition.

 ----
1-  ----
1-
 dt X0 + 0.5 ( 0 ) = 2  dt
X0 = 2
The initial condition caused by the impulse function found, assuming a zero initial con-
dition.

– 0.5 t
X ( t ) = Ce
0
X ( 0 ) = 2 = Ce
– 0.5 t
X ( t ) = 2e

• Note that the derivation of the unit impulse function assumed zero initial condi-
tions, so the process of convolution must also assume systems start at rest and undeflected.

19.4 CONVOLUTION

• The following example shows the use of the convolution integral to find a num-
 convolution - 19.6

ber of responses.

The unit impulse response to a step input can be calculated using the convolution inte-
gral.
t t t t
0.5τ
– 0.5 ( t – τ ) – 0.5t e
∫ X ( t – τ )F ( τ ) dτ ∫ ( 2e ∫e
– 0.5t 0.5τ
x(t) = = ) ( 1 ) dτ = 2e dτ = 2e ----------
0.5 0
0 0 0
0.5t 0.5 ( 0 )
– 0.5t  e
x ( t ) = 2e ---------- e-------------- = 4e –0.5t ( e 0.5t – 1 ) = 4 ( 1 – e –0.5t )
 0.5 – 0.5 

The unit impulse response to a unit ramp input can be calculated using the convolution
integral.
t t t
– 0.5 ( t – τ )
∫ X ( t – τ )F ( τ ) dτ ∫ 2e ∫e
– 0.5t 0.5τ
x(t) = = τ dτ = 2e τ dτ
0 0 0

x(t) = FINISH THE INTEGRAL

The unit impulse response to a sinusoidal input can be calculated using the convolution
integral.
t t t
– 0.5 ( t – τ )
∫ X ( t – τ )F ( τ ) dτ ∫ 2e ∫e
– 0.5t 0.5τ
x(t) = = sin ( τ ) dτ = 2e sin ( τ ) dτ
0 0 0

x(t) = FINISH THE INTEGRAL

19.5 NUMERICAL CONVOLUTION

• The convolution integral can also be solved numerically. This is particularly use-
ful for systems with arbitrary inputs.
 convolution - 19.7

n–1

x ( t ) = h ∑ X ( t – ( i + 0.5 )h )F ( ( i + 0.5 )h )
i=0

where,
t
n = --- = number of steps
h
h = step size (s)

X ( t ) = the unit impulse response function

F ( t ) = the input function

• This can be applied to the previous example for a unit step input to find the sys-
tem position at 10 seconds, with a 2 second time step.
 convolution - 19.8

n–1
– 0.5 ( t – ( i + 0.5 )h )
x ( t ) = h ∑ 2e F ( ( i + 0.5 )h )
i=0
10
n = ------ = 5
2
5–1
– 0.5 ( 10 – 2 ( i + 0.5 ) )
x ( 10 ) = 2 ∑ 2e (1)
i=0
4

x ( 10 ) = 4 ∑ e
– 4.5 + i

i=0

– 4.5 + 0 – 4.5 + 1 – 4.5 + 2 – 4.5 + 3 – 4.5 + 4

x ( 10 ) = 4 ( e +e +e +e +e )

– 4.5 – 3.5 – 2.5 – 1.5 – 0.5

x ( 10 ) = 4 ( e +e +e +e +e )

x ( 10 ) = 4 ( 0.01111 + 0.03020 + 0.08208 + 0.22313 + 0.6065 )

x ( 10 ) = 3.81208

This value can be compared to the exact value calculated below. The accuracy of the
numerical value would increase substantially if the step size were decreased.
– 0.5 ( 10 )
x ( 10 ) = 4 ( 1 – e ) = 4 ( 1 – 0.006738 ) = 3.973048

• A Scilab program to perform the previous calculation numerically is shown

below.
 convolution - 19.9

// A numerical convolution example

function foo=X(t) // The impulse response function

foo = 2 * exp(-0.5*t);
endfunction

function foo=F(t) // The input function

foo = 1; // a step function
endfunction

// define the variables

function foo = convolution(t, h) // The integration function

n = t / h;
foo = 0;
for i=0:n-1
foo = foo + h * X(t - (i+0.5)*h) * F((i+0.5)*h);
end
endfunction

h = 1; // the time step for the integration

t = 10; // the time point to calculate
printf("The estimated value x(%fs) = %f \n", t, convolution(t, h));
x_calc = 4*(1-exp(-0.5*t));
printf("The actual value is x(%fs) = %f \n", t, x_calc);

19.6 LAPLACE IMPULSE FUNCTIONS

• The convolution integral can be difficult to deal with because of the time shift.
But, the Laplace transform for the convolution integral turns it into a simple multiplica-
tion.
 convolution - 19.10

c(t) = ∫ g ( t – τ )r ( τ ) dτ
0

C ( s ) = G ( s )R ( s )

Figure 19.4 The convolution integral in the Laplace s-domain

19.7 SUMMARY

19.8 PRACTICE PROBLEMS

19.9 PRACTICE PROBLEM SOLUTIONS

19.10 ASSIGNMENT PROBLEMS

1.
 state space analysis - 20.1

20. STATE SPACE ANALYSIS

Topics:

Objectives:

20.1 INTRODUCTION

- state equations can be converted to transfer functions. The derivation follows.

 state space analysis - 20.2

State equations as functions of time

·
x = Ax + Bu
y = Cx + Du
In the s-domain
sX – X 0 = AX + BU

X ( sI – A ) = BU + X 0
–1 –1
X = ( sI – A ) BU + ( sI – A ) X 0
Y = CX + DU
–1 –1
Y = C ( ( sI – A ) BU + ( sI – A ) X 0 ) + DU
–1 –1
Y = ( C ( sI – A ) B + D )U + C ( sI – A ) X 0
Assuming the system starts at rest,
–1
Y = ( C ( sI – A ) B + D )U
Y- –1
--- = ( C ( sI – A ) B + D ) (the transfer function)
U

- state equation coefficient matrices can be transformed to another equivalent for,

if the state vector is rearranged.
 state space analysis - 20.3

State equations as functions of time

·
x = Ax + Bu
y = Cx + Du
Map one state vector to another one
x = Tz

This can be used to calculate new coefficient matrices

· –1
z = T ( Ax + Bu )
· –1
z = T ( ATz + Bu )
· –1 –1
z = T ATz + T Bu
· –1 –1 –1 –1
z = T ATz + T Bu A = T AT B = T B
·
z = Az + Bu

The output equation becomes

y = Cx + Du
y = CTz + Du
y = CTz + Du C = CT D = D
y = Cz + Du

The equivalent set of state equations is,

· –1 –1
z = Az + Bu A = T AT B = T B

y = Cz + Du C = CT D = D

•
 state space analysis - 20.4

The two following transfer functions are equivalents

–1
G = C ( sI – A ) B + D
–1
G = C ( sI – A ) B + D

This can be shown by applying the transformation matrix

–1 –1 –1 –1
G = CTT ( sI – A ) TT B + TT D
–1 –1 –1
G = ( CT ) ( T ( sI – A ) T ) ( T B ) + D
–1 –1
G = ( CT ) ( ( sI – A ) ) ( T B ) + D
–1
G = C ( ( sI – A ) )B + D = G

• The transfer function form can be put into a matrix form. In this case the denom-
inator is the characteristic equation.

The transfer function can be said to be equivalant to the determinants of the matrix form.
( sI – A ) – B

------------------------------------ = (---------------------------------------------
C D sI – A )D – ( – B )C-
= ( sI – A )D + BC ( sI – A )
sI – A sI – A

( sI – A ) – B
–1 C D poles
G = C ( sI – A ) B + D = ------------------------------------ = --------------
sI – A zeros

sI – A = characteristic equation = homogeneous

• The free (homogeneous) response of a system can be used to find the state transi-
tion matrix.
 state space analysis - 20.5

The homogeneous equation can be written in the s-domain, and then converted to time.
X h = sI – A X0
–1
x h ( t ) = L [ sI – A X 0 ]
2 3
-I + A
--- + A
------ + A
–1
xh ( t ) = L ------ + … X 0
s s s2 s3

At At
xh ( t ) = e x0 e = transition matrix

aside: This expansion is a McLaurin (Taylor) series.

= I + At +  ----- A t +  ----- A t + …
At 1 22 1 33
e
2! 3!

• The forced response (particular) response of the system can be found using con-
volution,

The homogeneous equation can be written in the s-domain, and then converted to time.
·
x = Ax + Bu
t
A(t – τ)
x(t) = e x( 0) + ∫ e
At
Bu ( τ ) dτ
0

y = Cx + Du
t
A(t – τ)
y ( t ) = Ce x ( 0 ) + ∫ Ce
At
Bu ( τ ) dτ + Du ( t )
0

initial impulse
response response

• As an example the homogeneous/free response of the system is shown below.

 state space analysis - 20.6

Given a second order differential equation, the state matrix can be found,
··
F = Mx
·
x = v 0
d- x
---- = 01 x + 1 F A = 0 1
· F dt v 00 v ----
-
v = ----- M 0 0
M
This can be used to find the inverse matrix,
s 1 s ---1- 1--- 1
–1 ---- ----
–1  1 0 0 1  s – 1
–1
0 s s
2 2
s s s2
( sI – A ) =  s –  = -------------
= 2 = =
 0 1 0 0 0 s s –0 0 s 1
---- ---- 0 ---
2 2 s
s s
The function of time can be found assuming an initial position of 10 and velocity of 5.

= L [ ( sI – A ) ] = 1 t
At –1 –1
e
0 1

x h ( t ) = x = e 10 = 1 t 10 = 10 + 5t
At

v 5 0 1 5 5

• The forced/particular solution is shown below,

Given a second order differential equation, the state matrix can be found,
··
F = Mx

• If a matrix is diagonizable, the diagonal matrix can be found with the following
technique. This can be used for more advanced analysis techniques to create diagonal (and
 state space analysis - 20.7

separable) system matrices.

Given a matrix A, a purely diagonal matrix may be found,

λ1 …
–1 –1
T AT = Λ A = TΛT Λ = … …
λn
This may also be written as,
At Λt – 1
e = Te T λ t
e 1 …
Λt
e = … …
λn t
e
This can also be applied to the
At 1 22
e = I + At + ----- A t + …
2!
At –1 1 –1 2 2
e = I + TΛT t + ----- ( TΛT ) t + …
2!
Eigenvalues can be found using the following relationship. If any of the Eigen-
values are repeated the Jordan normal form is required.

λI – A = 0 Note: the Eigenvalues are the values

found in the characteristic/homoge-
neous equation solution.

The Eigenvalues are then used to find the Eigenvectors,

( λI – A )v = 0 Av = λv

The Eigenvectors are then combined into a single matrix, this is the transition matrix.

• example,
 state space analysis - 20.8

Given the state matrix

A = –1 2
–3 4
The eigenvalues can be calculated,

λI – A = λ 1 0 – – 1 2 = λ + 1 –2
0 1 –3 4 3 λ–4
2
∴ = ( λ + 1 ) ( λ – 4 ) – ( – 6 ) = λ – 3λ + 2 = ( λ – 2 ) ( λ – 1 )
Therefore the Eigenvalues are, λ1 = 2 λ2 = 1
An eigenvector can be calculated using the Eigenvalue 2.

– 1 2 u 11 = λ u 11 = 2 u 11
1
– 3 4 u 21 u 21 u 21
u 11
– u 11 + 2u 21 = 2u 11 ∴2 u 21 = 3u 11 v1 = = 2
u 21 3
Another eigenvector can be calculated using the Eigenvalue 1.

– 1 2 u 12 = λ u 12 = 1 u 12
2
– 3 4 u 22 u 22 u 22
u 12
– u 12 + 2u 22 = u 12 ∴u 22 = u 12 v2 = = 1
u 22 1
The Eigenvectors can be combined to a single matrix.

T = v1 v2 = 2 1
31
To check, Av i = λ i v i

u11 u 12
A ,A = 2 2 ,1 1
u 21 u 22 3 1

–1 2 2 , – 1 2 1 = 4 , 1
–3 4 3 – 3 4 1 6 1

4, 1 = 4, 1 verified
6 1 6 1

• If there are repeated Eigenvalues in the system the Jordan Form can be used.
 state space analysis - 20.9

λ1 1 0 … 0 0
0 λ2 1 … 0 0
0 0 λ3 … 0 0
J = Jordan block
… … … … … …
0 0 0 … λn – 1 1
0 0 0 … 0 λn

Consider the example below with repeated eigenvalues.

–2 1 1
A = 1 –2 1 Find a matrix, that has repeated eigenvalues, and is not singular
1 1 –2

λ – 2 –1 – 1
2
λI – A = –1 λ – 2 –1 = (λ + 3) (λ – 0)
– 1 –1 λ – 2
 state space analysis - 20.10

Solve for the Jordan block one column at a time,

T 11 – 2 1 1 T 11 – 2 T 11 + T 21 + T 31
1 T 21 = 1 –2 1 T 21 = T 11 – 2T 21 + T 31 FIX
T 31 1 1 – 2 T 31 T 11 + T 21 – 2T 31

T 11 = 2T 11 + T 21 + T 31 T 21 =

T 21 = T 11 + 2T 21 + T 31 T 21 =

T 31 = T 11 + T 21 + 2T 31 T 31 =

T 12 – 2 1 1 T 12 – 2 T 12 + T 22 + T 32
1 T 22 = 1 –2 1 T 22 = T 12 – 2T 22 + T 32
T 32 1 1 – 2 T 32 T 12 + T 22 – 2T 32

T 12 = 2T 12 + T 22 + T 32

T 22 = T 12 + 2T 22 + T 32

T 32 = T 12 + T 22 + 2T 32

T 13 – 2 1 1 T 13 – 2 T 13 + T 23 + T 33
0 T 23 = 1 –2 1 T 23 = T 13 – 2T 23 + T 33
T 33 1 1 – 2 T 33 T 13 + T 23 – 2T 33

4T 13 = 2T 13 + T 23 + T 33

4T 23 = T 13 + 2T 23 + T 33

4T 33 = T 13 + T 23 + 2T 33
 state space analysis - 20.11

• The Eigenvectors can be used to calculate the system response.

Given the following Eigen values and vectors the free response is,

λ1 = 2 λ2 = 1 v1 = 2 v2 = 1
3 1
n
λi t 2t t
∑ = 2 e + 1 e = 2e + e
2t 1t
xh ( t ) = vi e
3 1 2t t
i=1 3e + e

• zeros of state space functions can be found using the state matrices.
 state space analysis - 20.12

The following relationship locates the zero frequencies in state equations.

s0 I – A –B
= 0
C D
Consider the example beginning with state equations,
·
x = v – 5F
·
v = – 3x – 4v + 5F
d- x
---- = 0 1 x + –5 F x = 1 0 x + 0F
dt v –3 – 4 v 5 v

A = 0 1 B = –5 C = 1 0 D = 0
– 3 –4 5
This can be put into the matrix form,

 1 0 
 s0 – 0 1  – –5
 0 1 –3 –4  5 = 0

1 0 0

s0 –1 5
3 s0 + 4 – 5 = 0
1 0 0
s 0 ( 0 ) – ( – 1 ) ( – 4.5 ) + 5 ( – ( s 0 + 4 ) ) = 0
– 24.5 = 5s 0
s 0 = – 4.9
There is one zero at -4.9 rad/sec. The order of the polynomial determines the
number of zeros. If the polynomial has no ’s’ terms, then there are no zeros.
 state space analysis - 20.13

20.2 OBSERVABILITY

- a system is observable iff the system state x(t) can be found by observing the
input u and output y over a period of time from x(t) to x(t+h).

- If an input is to be observable it must be detectable in the output. For example

consider the following state equations.

d- x
---- = a b x + e u
dt v c d v f

y = g h x + j u
v k
if the parameters c, d or f were not in the final expression for y, the second state equa-
tion would be said to be not observable. This will happen when an output parameter
is zero in the C matrix and an equation is decoupled in the A matrix, i.e. the same
variable column is zero.
For example,

d- x
---- = 1 0 x + 3 u
dt v 0 –a v b

y = 5 0 x + 6 u
v
–1
 
G = C ( sI – A ) B + D = 5 0  s 0 – 1 0  3 +
–1
6
 0 s 0 –a  b

 s+a 0 
 
–1  0 s – 1  3
G = 5 0  s–1 0  3 +  --------------------------------
6 = 5 0  ( s – 1 ) ( s + a -) + 6
 0 s+a  b   b
 
1- 3
---------- 0 -----------
s–1 3 15
G = 5 0 + 6 = 5 0 s – 1 + 6 = ----------- + 6
1- b b - s–1
0 ---------- ----------
s+a s+a

Note: the variables ’a’ and ’b’ both dissapeared because this system is not
observable.
 state space analysis - 20.14

• This often happens when a system has elements that are decoupled, or when a
pole and zero cancel each other.

• Observability can be verified formally for an LTI system with the following rela-
tionship.

The system is unobservable if the equation below is true for any state,

At * 1 0 0
Ce x = 0 t≥0
x = 0 , 1 ,… 0
*

… … …
0 0 1

One example,

A = 1 2 C = 5 6
3 4
At * e e
t 2t  1 0   1 0 
Ce x = 5 6  ,  = ( 5e t + 6e 3t ) ( 5e 2t + 6e 4t )  , 
e
3t
e
4t  0 1   0 1 
 
Ce x =  ( 5e t + 6e 3t ) ( 5e 2t + 6e 4t ) 1 , ( 5e t + 6e 3t ) ( 5e 2t + 6e 4t ) 0 
At *

 0 1 
At * t 3t 2t 4t
Ce x = { ( 5e + 6e ), ( 5e + 6e ) }
Both values are non-zero, so both states are observable.

One example,

A = 1 0 C = 3 0
0 2
t    
Ce x = 3 0 e 0  1 , 0  = 3e t 0  1 , 0  = { 3e , 0 }
At * t

0 e  0 1   0 1 
2t

The second state is not observable.

• Another theorem for testing observability is given below. If any of the states sat-
isfies the equation it is unobservable.
 state space analysis - 20.15

C
CA *
x = 0
…
n–1
CA

• Yet another test for observability is,

C
CA
rank ( M o ) = rank = n
…
n–1
CA

• If a system in unobservable, it is possible to make it observable by changing the

model.

• A pole-zero cancellation is often the cause of the loss of observability.

• If all unstable modes are observable, the system is detectable.

20.3 CONTROLLABILITY

- a system is controllable iff there is an input u(t) that will cause the system to go
from any initial state to any final state in a finite time.

- stabilizable if it is controllable or if the uncontrollable nodes are stable.

- If an input is to be observable it must be detectable in the output. For example

consider the following state equations.
 state space analysis - 20.16

d- x
---- = a b x + e u
dt v c d v f

y = g h x + j u
v k
if the parameters c, d or f were not in the final expression for y, the second state equa-
tion would be said to be not uncontrollable. This will happen when an output
parameter is zero in the C matrix and an equation is decoupled in the A matrix, i.e.
the same variable column is zero.
For example,

d- x
---- = 1 0 x + 2 u
dt v a –b v 0

y = 3 4 x + 5 u
v
–1
 
G = C ( sI – A ) B + D = 3 4  s 0 – 1 0  2 +
–1
5
 0 s a –b  0

s+B 0
–1
0 s–1 2
G = 3 4 s–1 0 2 +
5 = 3 4 --------------------------------
- + 5
0 s+b 0 (s – 1)( s + b) 0

1-
---------- 0
s–1 2 + 3 4 2 + 6
G = 3 4 5 = ----------- ----------- 5 = ----------
s – 1
-+5
1- 0 s – 1 s + b 0
0 ----------
s+b

Note: the variables ’a’ and ’b’ both dissapeared because this system is not con-
trollable.

• This can be verified with the

 state space analysis - 20.17

The basic relationship is,

* T At
(x ) e B = 0 t≥0
An example is,

A = 1 2 B = 5
3 4 6

1 2t
T
 1 0  3 4 5 =   e 1t e 2t 5
 ,  e ,
 1 0 0 1 
 0 1  6   e 3t e 4t 6

  5e 1t + 6e 2t 1t 2t 3t 4t
 1 0, 0 1  = { 5e + 6e , 5e + 6e }
  5e 3t + 6e 4t

The results are both non-zero, so both states are controllable.

An example is,

A = 1 0 B = 0
0 2 3
1 0t
T
 1 0  0 2 0 =   e 1t 0 0
 ,  e ,
 1 0 0 1 
 0 1  3   0 e 2t 3

  0 2t
 1 0 , 0 1  2t = { 0, 3e }
  3e

The first state has a zero, so it in not controllable.

• Another test for controllability is,

* T
( x ) B AB … A n – 1 B

• Yet another test for controllability is,

 state space analysis - 20.18

rank ( M c ) = rank B AB … A n – 1 B = n

• For a system to be controllable, all of the states must be controllable.

• If a system in uncontrollable, it is possible to make it observable by changing the

model.

• A pole-zero cancellation is often the cause of the loss of observability.

• if all unstable modes are controllable, the systems is said to be stabilizable.

• The principle of Duality requires that a system be completely observable to be

controllable.

20.4 OBSERVERS

• Observers estimate system state variables when not all of the variables are
directly observable.

• Observers use a limited set of system states that are available to identify other
system states that are not observable.

• The separation principle ensures that the observer cannot effect the stability of
the system it is observing.

20.5 SUMMARY

•
 state space analysis - 20.19

20.6 PRACTICE PROBLEMS

20.7 PRACTICE PROBLEM SOLUTIONS

20.8 ASSIGNMENT PROBLEMS

1.

20.9 BIBLIOGRAPHY

How, J, "16.31 Feedback Control Course Notes", MIT Opencourseware website.

 state space control - 21.1

21. STATE SPACE CONTROLLERS

Topics:

Objectives:

21.1 INTRODUCTION

• There are a few terms to be defined,

identification - the process of determining a state space (or other) model of

a process.
parameter identification - used to estimate model parameter values.
estimatior - this tries to determine state variable values given available
data.
observer - a special type of estimator that uses available input and output
data to estimate a system state given unknown initial conditions. This is
simalar to running a system simulation in parallel with the real system.
compensator - a feedback control system designed to drive a system to an
arbitrary input.
regulator - a feedback control system designed to drive a system to a given
operating point. A special case of a compensator with a zero input.
 state space control - 21.2

21.2 FULL STATE FEEDBACK

• The basic definition of a feedback controller is below,

The system state equations, with D=0,

·
x = Ax + Bu
·
y = Cx + 0u

The Eigenvalues of A give the system poles

u = r – Kx r = setpoint
u = control output/plant input
K = proportional gain values
·
x = Ax + B ( r – Kx )
·
x = ( A – BK )x + Br
·
x = A c x + Br
·
y = Cx

• An example of a controller design follows,

 state space control - 21.3

The controller design involves calculating K values to meet control objectives.

· 0 1 x+ 0 u ·
x = y = 1 x
– 2 –3 4 0
Check for controlability,

M c = B AB = 0 0 1 0 = 0 4
4 – 2 –3 4 4 – 12
The rank is 2, and the order of the matrix A is 2, so the system is controllable.

Check for the stability of the system without a controller,

sI – A = s 0 – 0 1 = s –1
0 s – 2 –3 –2 s – 3
2
sI – A = s ( s – 3 ) – 2 = s – 3s – 2 = ( s – ) ( s – )
Therefore, the uncompensated system is.......
Select an input vector with variables,
u = – Kx = – k 1 k 2 x

Ac = 0 1 – 0 0 1 – 0 0 = 0 1
k1 k2 =
– 2 –3 4 –2 – 3 4k 1 4k 2 ( – 2 – 4k 1 ) ( – 3 – 4k 2 )

s 0 – 0 1 s –1
sI – A c = =
0 s ( – 2 – 4k 1 ) ( – 3 – 4k 2 ) s + 2 + 4k 1 s + 3 + 4k 2
2
sI – A c = s ( s + 3 + 4k 2 ) + ( s + 2 + 4k 1 ) = s + s ( 4 + 4k 2 ) + ( 2 + 4k 1 )
The system is clearly second order, so the second order methods may be used to select
parameters. However for variety, why don’t we select a system that has poles at,
p = 1 + 2j, 1 – 2j
2
s + s ( 4 + 4k 2 ) + ( 2 + 4k 1 ) = ( s + 1 + 2j ) ( s + 1 – 2j )
2 2
s + s ( 4 + 4k 2 ) + ( 2 + 4k 1 ) = s + s ( 1 + 2j + 1 – 2j ) + ( ( 1 + 2j ) ( 1 – 2j ) )
2 2
s + s ( 4 + 4k 2 ) + ( 2 + 4k 1 ) = s + s ( 2 ) + ( 5 )

4 + 4k 2 = 2 k 2 = – 0.5
2 + 4k 1 = 5 k 1 = 0.75
 state space control - 21.4

·
r u x x y
1---
B C
+ + s
- +
A

Figure 21.1 Block diagram of a state based controller

• The value of K in the previous example could also be picked using Ackermann’s
formula.

–1
K = 0 0 … 1 Mc Φd ( A )
where,
Φ d ( s ) = the desired response

• Consider the previous example,

 state space control - 21.5

The state equations and the desired response of the system are,

· 0 1 x+ 0 u ·
x = y = 1 x
– 2 –3 4 0
2
Φd ( s ) = s + s ( 2 ) + ( 5 )
The Markov (?) parameter can be calculated first.

Mc = 0 0 1 0 = 0 4
4 –2 –3 4 4 – 12

–1
 0 1 2

K = 0 1 0 4  + 0 1 ( 2 ) + ( 5 )I
4 – 12  – 2 –3 –2 –3 

– 12 –4
–4 0  
K = 0 1 ----------------------  – 2 – 3 + 0 2 + 5 0 
0 – 16  6 7 –4 –6 0 5

K = 0 1 – 0.75 –0.25 3 – 1 = – 0.25 0 3 – 1 = – 0.75 0.25

– 0.25 0 2 6 2 6

NOTE: DOESN’T MATCH, TRACK DOWN ALGEBRA PROB

21.3 OBSERVERS

• Observers are used to estimate the next system state. They can also be used to
predict future state values.

• As an example, an observer might be used for a targeting system that is tracking a

moving object. The estimator can be used to direct a missle to an estimated location so that
the missle and target arrive at the same time.

• An observer is shown in block diagram form

 state space control - 21.6

Actual Plant
(and assumed state equations)
U
d- Y
---- X = AX + BU
dt
Y = CX

Observer
d- ˆ ˆˆ ˆ ˆ ˆ
---- X = AX + BU + K( Y – Y ) Y
dt
ˆ ˆ ˆ
Y = CX X

Figure 21.2 Block diagram of an observer

 state space control - 21.7

Aside: For an open loop estimator a simulation is run, while the system is running. The
simulation values are compared to the actual system responses, and the model is
adjusted. It is assumed that the initial value of the state is unknown. If A is stable then
the estimation will eventually converge (this can be checked using the Eigenvalues).
However, there is no guarantee that it will converge quickly.

The actual system

d- y = Cx C≠I
---- x = Ax + Bu
dt
The state estimator model
d- ˆ
---- x = Axˆ + Bu
dt

The estimation error

x˜ = x – xˆ
d- d- ˜
---- ( x – xˆ ) = A ( x – xˆ ) ---- x = Ax˜
dt dt
At
x˜ ( t ) = e x˜ ( 0 )

• A closed loop estimatior is shown below. It uses a learning matrix L to adjust the
convergence rate. A larger L value will result in faster convergence, but the exponent must
 state space control - 21.8

now be checked to ensure observability.

The system and estimator
y = Cx yˆ = Cxˆ

The estimator error

y˜ = y – yˆ = Cx˜
The estimator
d- ˆ
---- x = Axˆ ( t ) + Bu ( t ) + L ( y ( t ) – y˜ ( t ) )
dt
yˆ ( t ) = Cxˆ ( t )
where,
L = a gain matrix
The convergence rate,
d- ˜
---- x = ( A – LC )x̃
dt
( A – LC )t ˜
x˜ ( t ) = e x(0)

• Estimator gain selection

If a system is controllable the Eigenvalues of A-BK can be placed anywhere,

rank ( A ) = rank B AB … A n – 1 B = n

If a system is observable the Eigenvalues of A-LC can be placed anywhere,

C
CA
rank ( A ) = rank = n
…
n–1
CA

• Consider the example below, for a second order system. The learning rate values
are selected to determine how quickly the model converges.
 state space control - 21.9

Consider the state equations,

d- x 1 –a1 1 x1 b x1
---- = + 1 u y = 1 0
dt x –a2 0 x2 b2 x2
2

Find the convergence rate and then the Eigenvalues,

– a1 1 l1 – a1 – l1 1
A – LC = – 1 0 =
– a2 0 l2 – a2 – l2 0

s + a1 + l1 –1
sI – ( A – LC ) = = s ( s + a 1 + l 1 ) – ( –1 ) ( a 2 + l2 )
a2 + l2 s
2
= s ( 1 ) + s ( a1 + l 1 ) + ( a2 + l2 )

• This can also be done with Ackermann’s formula,

0
T
L = K e = Φ e ( s )M o
–1 …
0
1
 state space control - 21.10

Consider the state equations for a mass spring damper system (y is the position).
0 1 0
d
----- x = –K x + y = 1 0 x + 0 F
dt v –K d 1- F
---------s --------- v ---- v
M M M
0 1 0
A = –K –Kd B = 1 C = 1 0 D = 0
---------s --------- -----
M
M M
First test for observability, (note: it is observable)

1 0
1 0 1 0
rank C = rank 0 1 = rank = rank = 2
CA
1 0 –Ks –K d 0 1 0 1
--------- ---------
M M

Consider the open loop estimator

d- ˆ
---- x = Axˆ + Bu yˆ = Cxˆ
dt

ˆ 0 1 0
d
----- x = –K xˆ + yˆ = 1 0 x
ˆ
dt ˆ –Kd 1- F
v ---------s --------- vˆ ---- vˆ
M M M

• The open loop estimator will normally ignore the response to initial conditions
and will have long term errors resulting from modeling errors and random disturbances in
the system.

Consider the closed loop estimator with a learning rate.

d- ˆ
---- x = Axˆ ( t ) + Bu ( t ) + Ly˜ ( t ) yˆ ( t ) = Cxˆ ( t )
dt

ˆ 0 1 0 ˆ
d
----- x = –K xˆ + l
F + 1 y˜ yˆ ( t ) = 1 0 x
–Kd 1
dt ˆ
v ---------s --------- vˆ ----- l2 vˆ
M M M

These are then simulated numerically using initial conditions of zero.

 state space control - 21.11

XXXXXXXXXX Example of learning parameter estimator to obtain desired

responses.

21.4 SUPPLEMENTAL OBSERVERS

• In many cases some state variables can be measured directly, but others cannot.
In these cases it is more efficient to only estimate the unmeasurable variable values [Ero-
nini].

• This can be

W = KX
–1
X = C
---- Y- = M –1 ----
Y- M = C (Note: the inverse must exist)
---- ----
K W W K

Y- = MX
----
W

d- Y d –1 Y
---- ----- = ----- MX = M ( AX + BU ) = MAX + MBU = MAM ----
- + MBU
dt W dt W
The matrices can be partitioned to isolate the unmeasurable state variables. Note that the
Eigenvalues of Hw must be relatively small.

–1 … … …
MAM = MB =
Hy Hw Bw

d- ˆ ˆ
---- W = H y Y + H w W + Bw U
dt

21.5 REGULATED CONTROL WITH OBSERVERS

• For a regulated system the control output ’u’ can be estimated using the desired
 state space control - 21.12

system state.
ˆ
u = Kx = K y

• The system state is often difficult to measure directly, so an estimator can be

used. .
u = – Kx = – Kxˆ

• A regulator can use an estimate of the system state to

CL
+ d- ˆ
- ---- d- y
x ---- x x
dt 1--- xˆ u dt 1---
–K B C
s s
+
+
A A

system
B

estimator and regulator

 state space control - 21.13

For a regulator assume the state variables go to zero, so....

u = – Kxˆ
A closed loop estimator
d- ˆ
---- x = Axˆ + Bu + L ( y – yˆ )
dt

yˆ = Cxˆ
The state model for the system is,
d-
---- x = Ax + Bu
dt
y = Cx
These can be combined to obtain,
d- ˆ
---- x = Axˆ + Bu + L ( y – yˆ ) = xˆ ( A – BK – LC ) + x ( LC )
dt
d-
---- x = Ax + Bu = x ( A ) + xˆ ( – BK )
dt
Resulting in the matrix,

d- x x
---- = A – BK
dt x ˆ LC A – BK – LC xˆ
 state space control - 21.14

An error variable can be introduced, and written in matrix form,

x˜ = x – xˆ

x x
= I 0 = Tx cl ≈ x˜cl T = I 0 = T
–1

x˜ I – I x˜ I –I
An error variable can be introduced, and written in matrix form,
–1
A cl = TAcl T ≈ A cl
The previous matrix can be revised

A cl = I 0 A – BK I 0
I – I LC A – BK – LC I – I

A cl = I 0 A – BK BK
I – I A – BK – A + BK + LC

A cl = A – BK BK
0 A – LC
The new poles for this system can be written,
2
sI – A cl = ( A – BK ) ( A – LC ) – BK ( 0 ) = A – ALC – ABK + BKLC

• The estimators and regulators can be designed separately and then combined.
This is the separation principle.

• The simplified block diagram for the system is,

d- ˆ d-
---- x ---- x
dt xˆ dt
+ x
1--- 1---
– BK
s s
+
+ +
A – BK – LC A

LC
 state space control - 21.15

• The compensator,

The previous estimator equations can be combined with a compensator

u = – Kxˆ
d- ˆ
---- x = Axˆ + Bu + L ( y – yˆ ) = Axˆ – BKxˆ + L ( y – Cxˆ ) = ( A – BK – LC )x̂ + ( L )y
dt
These can be used to define a new compenator state,
xˆ = x c
u = –Cc xc
d-
---- x = Ac xc + Bc y
dt c
where,
A c = A – BK – LC
Bc = L
Cc = K

• This can be converted to a transfer function for the regulator.

These can be used to define a new compenator state,
sx c = A c x c + B c y
x c ( sI – Ac ) = B c y
xc Bc
---- = ---------------
-
y sI – A c
u = –Cc xc
u
---- = –C c
xc
u--- u xc –C c B c
= ---- ---- = ----------------
y xc y sI – A c

• The system can also be set up for a compensator by assuming the setpoint input is
not zero,
 state space control - 21.16

First a system error is defined.

e = r–y
The previously defined compensator can be used here,
e = r–y
u = –Cc xc
u
---- = –C c
xc
d-
---- x = Ac xc + Bc e
dt c
xc Bc
---- = ---------------
-
e sI – A c
u--- u xc –C c Bc
= ---- ---- = ----------------
e xc e sI – A c
The system can also be modelled,
y = Cx
y--
= C
x
d-
---- x = Ax + Bu
dt
--x- = -------------
B -
u sI – A

--y- = y-- --x- = -------------

CB -
u xu sI – A
The feedforward and feedback loop can be simplified,
y-- uy – C c B c CB
= --- --- = -----------------------------------------
e eu ( sI – A c ) ( sI – A )
–C c B c CB
----------------------------------------
-
y-- ( sI – A c ) ( sI – A ) – C c B c CB
= -------------------------------------------------- = -------------------------------------------------------------------
r – C c B c CB ( sI – A c ) ( sI – A ) – C c B c CB
1 + -----------------------------------------
( sI – A c ) ( sI – A )

• Consider an example of a regulator/estimator design,

 state space control - 21.17

Consider the state equations for a mass-spring-damper system.

0 1 0
d x
----- = –K x + y = 1 0 x + 0 F
dt v –Kd 1- F
---------s --------- v ---- v
M M M
0 1 0
A = –K –Kd B = 1 C = 1 0 D = 0
---------s --------- -----
M M M
The regulator will be designed to have a response time of 0.5s and damped frequency of 2
rad/s.
1
s = ------- + j ( 2 ) = – 2 ± 2j
0.5
λ i ( A – BK ) = – 2 ± 2j
 0 1   0 1 
 0   
= λ i  –K – K – 1 k1 k2  = λ i  – K – k – K – k 
 ---------s --------d- -----   --------------------
s 1 d 2
- ---------------------
- 
 M M M   M M 
0 1 s –1
= s 0 – =
– Ks – k1 – K d – k2 K s + k 1 
 ---------------- K d + k 2
0 s --------------------- ---------------------- - s + -----------------
-
M M  M   M 
K d + k 2  K s + k 1 K d + k 2  K s + k 1
= s  s + -----------------
- + ----------------- = s  s + -----------------
- + -----------------
 M   M   M   M 
Kd + k2 Kd + k2 2 Ks + k1
–  ------------------ ±  ------------------ – 4  -----------------
M M M
s = -------------------------------------------------------------------------------------------------
2
Kd + k2
– ------------------
M k 2 = 4M – Kd
– 2 = ---------------------
2
K d + k 2 2
 ----------------- K s + k 1
- – 4  ---------------- -
 M   M 
2 = --------------------------------------------------------------
2
2
16M = ( K d + k 2 ) – 4 ( K s + k 1 )
2
4 ( K s + k 1 ) = ( K d + 4M – K d ) – 16M
2
k 1 = 4M – 4M – K s
K = 4M 2 – 4M – K 4M – K
s d
 state space control - 21.18

To select the regulator values, component values must be selected.

N Ns
M = 1kg K s = 10 ---- K d = 1 ------
m m
K = 4M 2 – 4M – K 4M – K
s d

K = 4 ( 1 ) 2 – 4 ( 1 ) – 10 4 ( 1 ) – ( 1 )

K = – 10 3
The desired response time for the system will be a time response of approximately
0.2sec.

x 1
G ( s ) = --- = -------------------------------------
2
-
F Ms + K d s + K s
0 1 0 0 1
A cl = A – BK = – K –Kd – 1- k 1 k 2 = – Ks – k1 – Kd – k2
---------s --------- ---- --------------------- ----------------------
M M M M M
0 1 s 1
Φ e ( s ) = sI – Acl = s 0 – =
– Ks – k1 – Kd – k2 K s + k 1 Kd + k2
0 s --------------------
- ---------------------- ----------------- ------------------ + s
M M M M
K d + k 2  – K s – k 1
= s + s  -----------------
2
- + -------------------- = ( s + 0.2 ) 2
 M   M 
–Ks – k 1
-------------------- = ( 0.2 ) 2 k 1 = – 10.04
M
Kd + k2
------------------ = 2 ( 0.2 ) k 2 = – 0.6
M
 state space control - 21.19

The estimator may then be designed using Ackermann’s formula.

–1
L = Φe ( A ) C 0
CA 1

 0 
2 1 0
 1 
=  – K – K + 0.2 0
0 1
 ---------s --------d-  1
 M M  1 0 –Ks
---------
–K d
---------
M M
 0 1
2
0 1 
 2  1 0 0
=  –K – K + 2 ( 0.2 ) –K – K d + 0.2 I
 ---------s --------d- ---------s ---------  1
 M M M M  1 1

 –Ks –Kd 
 --------- --------- 0 0.4 
 M M 0.04 0  1 0 0
=  2
+ – 0.4K – 0.4K + 
 K s Kd –K s Kd -----------------s ----------------d- 0 0.04  1
 ------------ --------- + ------- M M  1 1
 M
2 M M2 

–Ks –Kd
--------- + 0.04 --------- + 0.4
M M 1 0 0
= 2
K s K d 0.4K s – K s Kd
0.4K d 1 1 1
------------ – ------------- --------- + – -------
-------------- + 0.04
2 M M M M2
M
–Ks –Kd
--------- + 0.04 --------- + 0.4
M M 0
= 2
K s K d 0.4K s – K s Kd
0.4K d 1
------------ – ------------- --------- + – -------
-------------- + 0.04
2 M M M M2
M

–Kd
--------- + 0.4
M – 0.6
= 2
=
–K s Kd
0.4K d – 9.36
--------- + -------
– -------------- + 0.04
M M M 2
 state space control - 21.20

The compensator coefficients may then be calculated.

A c = A – BK – LC

0
Ac = 0 1 – – 0.6
1 – 10 3 – 1 0
– 10 –1 ----- – 9.36
M

Ac = 0 1 – 0 0 – – 0.6 0 = 0.6 1
– 10 –1 – 10 3 – 9.36 0 9.36 –4

Bc = L = – 0.6
– 9.36

C c = K = – 10 3

The compensator transfer function can then be written,

–1
G c ( s ) = C c ( sI – A c ) B c
–1
  –1
= –10 3  s 0 – 0.6 1  – 0.6 =
– 10 3
s – 0.6 –1 – 0.6
 0 s 9.36 – 4  – 9.36 – 9.36 s – 4 – 9.36
T
s – 0.6 – 1 –0.6 s – 4 9.36 – 0.6
– 10 3 – 10 3
– 9.36 s – 4 –9.36 1 s – 0.6 – 9.36
= ------------------------------------------------------------------------------ = ---------------------------------------------------------------------------
( s – 0.6 ) ( s – 4 ) – ( – 1 ) ( – 9.36 )
s – 0.6 – 1
– 9.36 s – 4
s–4 1 – 0.6 – 0.6s – 6.96
– 10 3 –10 3
9.36 s – 0.6 – 9.36 – 9.36s
= ------------------------------------------------------------------------
2
- = -------------------------------------------------------
2
1s – 4.6s – 6.96 1s – 4.6s – 6.96
– 22.08s + 69.6
= ------------------------------------
2
-
s – 4.6s – 6.96

• This system can be implemented with a C program as shown below,

 state space control - 21.21

#define PERIOD 0.001 // period of the interrupt

interrupt_loop(){ // this routine is called at a regular interval PERIOD

double analog_in(){

d- ˆ d-
---- x ---- x
dt xˆ dt
+ x
1 1
--- – BK ---
s s
+
+ +
A – BK – LC A

LC
 state space control - 21.22

21.6 LQR

- Linear Quadratic Regulator (LQR) allows the design of a regulating system that
tends to zero, but also considers the control effort.

- The basic function for evaluating the controller is defined below. The integral is
evaluated to reduce the value. Ideally the state value will go to zero (it is a regulator), and
the plant input will be minimized. It is assumed that D=0. Here a new state ’z’ is defined.
It should reflect the system state values that should be regulated to zero.

∫ [ (y(t) )
T 2
J LQR = y ( t ) + r ( u ( t ) ) ] dt
0
T
( y ( t ) ) y ( t ) = state cost
2
( u ( t ) ) = control cost
r = control penalty
z = Cz x
∞

∫ [ (x(t) )
T T 2
J LQR = C z C z x ( t ) + r ( u ( t ) ) ] dt
0

- the design process involves adjusting the r value to obtain the desired result.

-
 state space control - 21.23

Assume a simple controller form,

u ( t ) = –K LQR x ( t )

Solve using an Algebraic Riccati Equation

Consider a simple system

·
x = Ax + Bu y = Cx

The equivalent transfer function can be written, where a(s) is the characteristic eqn.
–1 b(s) b(s)
C ( sI – A ) B = ---------- = -----------------
a(s) sI – A

This is followed by.......

UNKNOWN STUFF

The desired charactersitic equation

n can be used to find the controller gain values.

sI – A + BK LQR = ∏ ( s – pi )
i=1
–1
∆ ( s ) = a ( s )a ( – s ) + r b ( s )b ( – s )
The poles are then found using Ackermann’s method.

2
( u ( t ) ) = control cost
r = control penalty
∞

∫ [ (x(t) )
T T 2
J LQR = C Cx ( t ) + r ( u ( t ) ) ] dt
0
∞

∫ [ (x(t) )
T T 2
J LQR = C Cx ( t ) + r ( u ( t ) ) ] dt
0
 state space control - 21.24

21.7 LINEAR QUADRATIC GAUSSIAN (LQG) COMPENSATORS

• Uses an optimal regulator and estimator.

• Stability is guaranteed if the model is accurate. Verification is needed

• The design focuses on,

A cost function for the system state, the design tries to minimize this
The relative weighting for the process and sensor noise

21.8 VERIFYING CONTROL SYSTEM STABILITY

• Introduce random errors to the system model and test for stability

• Use the Nyquist Stability Theorem,

 state space control - 21.25

Z = N+P
where,

P = poles of G(s)Gc(s) in the right hand plane

Z = closed loop poles in the right hand plane
N = number of clockwise loops in the Nyquist Diagram about -1

Z = 0, N = – P For stability

Note: Larger loops indicate a more stable system. Tight loops are very sensitive to
parameter variations.

21.8.1 Stability

• The stability function is defined using the Nyquist plot.

 state space control - 21.26

Define the points for the Nyquist plot

L ( s ) = G ( s )G c ( s )

L N ( s ) = G N ( s )G c ( s )

L A ( s ) = G A ( s )G c ( s )
The distance, ’d’, from the point at -1 to the calculated point is found as a function.

d ( jω ) = L N ( jω ) – ( – 1 + 0j )
The sensistivity ’S’ is defined as the inverse of the distance from the point at -1.

1 1
S ( jω ) = -------------- = ---------------------------
d ( jω ) 1 + L N ( jω )

21.8.2 Bounded Gain

• In practical terms system gain is limited by real components. As a result it is

important to ensure that the design will not exceed the maximum values available.
 state space control - 21.27

The boundedness of the system can be tested by ensuring that the sensitivity always
remains less than some given value ’gamma’ for all frequencies. Note: this requires
that D must be smaller than gamma.

S ( jω ) < γ

–1
C ( jωI – A ) B + D < γ
The Hamiltonian matrix is defined below. It does not have purely complex Eigenvalues.

2 T –1 T 2 T –1 T
H = A + B( γ I – D D) D C B(γ I – D D) B
T 2 T –1 T T T 2 T –1 T
– C ( I + D ( γ I – D D ) D )C – A – C D ( γ I – D D ) B

Or, if D=0,

T
BB
A ----------
H = 2
γ
T T
–C C –A
This matrix can then be tested, if there are no purely complex eigenvalues, the system is
bounded.

T T
BB - BB -
λ 0 – A --------- λ – A – ---------
λI – H = γ
2 = γ
2
0 λ
T T T T
–C C –A C C λ+A

2 T  BB T C T C
= λ ( 1 ) + λ ( – A + A ) +  – ---------------------
 γ
2


etc....
 state space control - 21.28

The algebraic Ricatti equation can also be used

T
T T XBB X
A X + XA + C C + -----------------
2
- = 0
γ

21.9 ADAPTIVE CONTROLLERS

• There are two basic types of adaptive control,

Self Tuining Regulators

Model Reference Adaptive Control (MRAC)

• A self tuning regulator

Parameter Regulator
Estimation Design

Regulator Plant
 state space control - 21.29

• Parameter Adaptive MRAC

Adaptation
Reference Algorithm
Model

Controller Plant

• Signal Synthesis MRAC

Adaptation
Reference Algorithm
Model
+
Controller Plant
+

• Parallel MRAC

Adaptation
Reference Algorithm
Model

Adjustable
Plant
 state space control - 21.30

• Series MRAC

Adaptation
Algorithm

Reference Adjustable
Model Plant

• The basic MRAC system is defined below,

The reference model is developed from the fundamentals of the system. It should be
controllable and stable.
d-
---- X = Am Xm + Bm r
dt m

Parameters can be adpated using,

e = X m – Xp
d-
---- X = A p X p + Bp r
dt p
Where,

A p, B p = Adjustable parameter matrices

A m, B m = Reference model matrices

d-
---- X = Ap Xp + Bp ( r + ua )
dt p
d-
---- e = Am Xm + Bm r – Ap Xp – Bp ( r + ua ) = Am ( e + Xp ) + Bm r – Ap Xp – Bp ( r + ua )
dt
d-
---- e = e ( Am ) + Xp ( A m – A p ) + r ( B m – B p ) + u a ( –B p )
dt
d-
---- e = e ( Am ) + f f = Xp ( Am – Ap ) + r ( Bm – Bp ) + ua ( –Bp )
dt
 state space control - 21.31

A Lyapunov can be defined for the error

T
V = e Qe

V =  ----- e  Qe + e Q  ----- e = ( e ( A m ) + f )Qe + e Q ( e ( A m ) + f )

d- d T T d T T T T
----
dt  dt   dt 
d- T T T
---- V = e ( A m Q + QA m )e + 2e Qf
dt
d- T T T
---- V = – e We + 2e Qf W = – ( A m Q + QAm )
dt

21.10 OTHER METHODS

H-infinity and mu methods are designed to shape the sensitivity plot

 state space control - 21.32

21.10.1 Kalman Filtering

• Kalman Filter helps settle estimators using estimates of snesor and process noise.

d-
---- x = Ax + Bu + B w w
dt
y = Cx + v

where,
B w = process noise weighting matrix

w = process noise matrix

v = sensor noise
–1 N(s)
G yw ( s ) = C ( sI – A ) B w = -----------
D( s)
R
D ( s )D ( – s ) ± ------w N ( s )N ( – s ) = 0
Rv

where,
R w, R v = factors related to process/sensors noise

21.11 SUMMARY

•
 state space control - 21.33

21.12 PRACTICE PROBLEMS

21.13 PRACTICE PROBLEM SOLUTIONS

21.14 ASSIGNMENT PROBLEMS

1.
 state space control - 22.1

22. SYSTEM IDENTIFICATION

Topics:

Objectives:

22.1 INTRODUCTION

• A simple example

d-
---- X = AX out + BX in
dt out
d-
---- Y = CX out + DXin
dt out
 state space control - 22.2

Given a system model for a force applied to a mass,

d-
---- x = v
dt
d- F
---- v = -----
dt M
0
d- x
---- = 0 1 x + F
dt v 0 0 v -----
M
 state space control - 22.3

For an unknown second order linear system,

d- x
---- = a b x + e F
dt v c d v f

d- x i x
---- = a b i + e F
dt v c d vi f
i

The state variables can be replaced with backwards difference equations,

xi – xi – 2
--------------------- xi – 1
T a b e F
= xi – xi – 2 +
x i – 2x i – 1 + x i – 2 c d --------------------- f
----------------------------------------- T
2
T
The equations can then be written for three different set of points to yield six equations.
xi – xi – 2 xi – xi – 2
--------------------
- = ax i – 1 + b -------------------- - + eF
T T
xi – xi – 2 xi – xi – 2
a ( x i – 1 ) + b  --------------------- + c ( 0 ) + d ( 0 ) + e ( F ) + f ( 0 ) = ---------------------
 T  T
xi – 3 – xi – 5 xi – 3 – xi – 5
a ( x i – 4 ) + b  --------------------------- + c ( 0 ) + d ( 0 ) + e ( F ) + f ( 0 ) = ---------------------------
T T
xi – 6 – xi – 8 xi – 6 – xi – 8
a ( x i – 7 ) + b  --------------------------- + c ( 0 ) + d ( 0 ) + e ( F ) + f ( 0 ) = ---------------------------
 T  T
x i – 2x i – 1 + x i – 2 xi – xi – 2
----------------------------------------- = cx i – 1 + d --------------------
- + fF
2 T
T
xi – xi – 2 x i – 2x i – 1 + x i – 2
a ( 0 ) + b ( 0 ) + c ( x i – 1 ) + d  --------------------- + e ( 0 ) + f ( F ) = ----------------------------------------
2
-
T T
xi – 3 – xi – 5 x i – 3 – 2x i – 4 + x i – 5
a ( 0 ) + b ( 0 ) + c ( x i – 4 ) + d  --------------------------- + e ( 0 ) + f ( F ) = -----------------------------------------------
2
-
T T
xi – 6 – xi – 7 x i – 6 – 2x i – 7 + x i – 8
a ( 0 ) + b ( 0 ) + c ( x i – 7 ) + d  --------------------------- + e ( 0 ) + f ( F ) = -----------------------------------------------
2
-
T T
 state space control - 22.4

The equations can be rewritten in matrix form,

xi – xi – 2
xi – xi – 2 ---------------------
xi – 1 --------------------- 0 0 F 0 T
T
xi – 3 – xi – 5
xi – 3 – xi – 5 ---------------------------
xi – 4 --------------------------
- 0 0 F 0 T
T a
xi – 6 – xi – 8
xi – 6 – xi – 8 b ---------------------------
xi – 7 --------------------------
- 0 0 F 0 T
T c =
x i – 2x i – 1 + x i – 2
xi – xi – 2 d -----------------------------------------
0 0 xi – 1 --------------------- 0 F 2
T e T
xi – 3 – xi – 5 f x i – 3 – 2x i – 4 + x i – 5
0 0 xi – 4 --------------------------- 0 F ------------------------------------------------
T 2
T
xi – 6 – xi – 7 x i – 6 – 2x i – 7 + x i – 8
0 0 xi – 7 --------------------------- 0 F ------------------------------------------------
T 2
T

Some sample data points can be calculated for F=1 and M=1,
1 2 1F 2
x = --- at = --- ----- t T = 1 F = 1 M = 1
2 2M

x(0) = 0 xi – 8 = 0
x ( 1 ) = 0.5 x i – 7 = 0.5
x(2) = 2 xi – 6 = 2
x ( 3 ) = 4.5 x i – 5 = 4.5

x(4) = 8 xi – 4 = 8

x ( 5 ) = 12.5 x i – 3 = 12.5

x ( 6 ) = 18 x i – 2 = 18

x ( 7 ) = 24.5 x i – 1 = 24.5

x ( 8 ) = 32 x i = 32
 state space control - 22.5

The values can be substituted into the matrix, and the matrix solved

32 – 18-
32 – 18 -----------------
24.5 ------------------ 0 0 1 0 1
1
12.5 – 4.5-
12.5 – 4.5 -----------------------
8 ------------------------ 0 0 1 0 a 1
1
b 2-----------
– 0-
2-----------
– 0-
0.5 0 0 1 0 1
1 c =
32 – 2 ( 24.5 ) + 18-
32 – 18 d ------------------------------------------
0 0 24.5 ------------------ 0 1 1
2
1 e
12.5 – 4.5 12.5 – 2 ( 8 ) + 4.5-
0 0 8 ------------------------ 0 1 f ----------------------------------------
2
1 1
0 0
2 – 0.5
0.5 ---------------- 0 1 2 – 2 ( 0.5 ) + 0
----------------------------------
1 2
1
–1
a 24.5 14 0 0 1 0 14
a 0
b 8 8 0 0 1 0 8
b 1
c = 0.5 2 0 0 1 0 2
c = 0
d 0 0 24.5 14 0 1 1
d 0
e 0 0 8 8 0 1 1
e 0
f 0 0 0.5 1.5 0 1 1
f 1

Substituting these values back into the state equation results in the following
expression, which matches the exact state equation.

d- x i x
---- = a b i + e F state equation
dt v c d vi f
i

xi x
= 10 i + 0 0 F output equation
0 0 0 vi 0 0 1

This can be incorporated as a realtime algorithm to calculate the data parameters

based upon realtime updates.
 state space control - 22.6

–1 xi – xi – 2
xi – xi – 2 ---------------------
xi – 1 --------------------- 0 0 F 0 T
T
xi – 3 – xi – 5
xi – 3 – xi – 5 ---------------------------
xi – 4 --------------------------- 0 0 F 0 T
a T
xi – 6 – xi – 8
b xi – 6 – xi – 8 ---------------------------
x --------------------------- 0 0 F 0 T
c = i–7 T
x i – 2x i – 1 + x i – 2
d xi – xi – 2 -----------------------------------------
0 0 x i – 1 --------------------- 0 F T
2
e T
f xi – 3 – xi – 5 x i – 3 – 2x i – 4 + x i – 5
0 0 x i – 4 --------------------------- 0 F ------------------------------------------------
T 2
T
xi – 6 – xi – 7 x i – 6 – 2x i – 7 + x i – 8
0 0 x i – 7 --------------------------- 0 F ------------------------------------------------
T 2
T

This can then be used as a model to calculate a transfer function as shown below
using the previous example.
 state space control - 22.7

The state equation variables are,

d- x i x A = a b B = e
---- = a b i + e F
dt v c d vi f c d f
i

xi x
= 10 i + 0 0 F C = 1 0 D = 0 0
0 0 0 v i 0 0 1 0 0 0 0
these can be used to calculate a SISO system transfer function, and its inverse.
–1
 
G = C ( sI – A ) B + D = 1 0  s 1 0 – a b  e + 0 0
–1

0 0  01 c d f 0 0

 s–d b 
 
–1  c s–a 
G = 1 0  s – a –b  e = 1 0  --------------------------------------------
- e
0 0  –c s – d  f 
0 0  ( s – a ) ( s – d ) + bc  f

 
 s–d b 
 
 0 0 ( s – d )e + bf -
 e = --------------------------------------------
G = ---------------------------------------------
 ( s – a ) ( s – d ) + bc f ( s – a ) ( s – d ) + bc
 
 
–1 (--------------------------------------------
s – a ) ( s – d ) + bc-
G =
( s – d )e + bf

A block diagram for the system may then be written,

–1
G
+
+ + F xa
plant
xd controller

1
 state space control - 22.8

Consider the controller with the previous example.

The identified coefficients for the controller are substituted into the inverse transfer func.

a 0
b 1
c = 0 –1 ( s – a ) ( s – d ) + bc (s – 0)(s – 0) + 1(0 ) 2
G = --------------------------------------------- = -------------------------------------------------- = s
( s – d )e + bf ( s – 0 )0 + 1 ( 1 )
d 0
e 0
f 1
The identified coefficients for the controller are substituted into the inverse transfer func.
d- 2
 ---- F F
x = ----- = ---
 dt M 1

--x- = ----
1
F 2
s
A state diagram for the system may then be written using a proportional controller,
2
s
+
+ + F xa
1-
xd P ---
2
s
-

1
 state space control - 22.9

The state diagram is simplified,

--1-
2
s
P
+
+ + xa
1-
xd P ---
2
s
-
1

2
s-
---
P
+
+ xa
P-
xd ---
2
s
- -
1
2
s-
---
P
+
xa
+ P
xd --------------
2
s +P

xa
2 P
xd s---- --------------
+1 2
P s +P

2
xd
+ P  --------------
 s-------------- P 
xa
 P  2 
s +P

xd 1 xa

Note: This suggests that the result is an ideal controller, however in practice there are
errors, such as sensor noise, imperfect actuators, and changing system conditions
such as mass. This also assumes that the force actuator can generate any requested
force.
 state space control - 22.10

22.2 SUMMARY

22.3 PRACTICE PROBLEMS

22.4 PRACTICE PROBLEM SOLUTIONS

22.5 ASSIGNMENT PROBLEMS

1.
 electromagnetics - 23.1

23. ELECTROMECHANICAL SYSTEMS

Topics:

Objectives:

23.1 INTRODUCTION

• Magnetic fields and forces are extremely useful. The fields can allow energy stor-
age, or transmit forces.

23.2 MATHEMATICAL PROPERTIES

• Magnetic fields have direction. As a result we must pay special attention to direc-
tions, and vector calculations.

23.2.1 Induction

• Magnetic fields pass through space.

 electromagnetics - 23.2

• resistivity of materials decreases with temperature

• Amperes Circuit Law

I =
°∫ Hdl
where,
I = current flowing along a line (A)

H = magnetic field intensity (A/m)

l = A perpendicular path around the current flow (m)

For example, at a fixed radius (r) around a wire,

2π
I = ∫0 Hr dθ

∴I = 2πHr
I
∴H = ---------
2πr

• Flux density can be calculated for low H values. As the value climbs the relation-
ship becomes non-linear.

B = µH = µ r µ 0 H

where,
B = Flux density (Wb/m*2 or T)
µ = permeability of material

– 7 Henry
µ 0 = permeability of free space = 4π ×10 ----------------
m
µ r = relative permeability

• Permeability,
 electromagnetics - 23.3

B
µ = ----
H
–7 H
µ 0 = 4π10 ----
m
µ = µr µ0

where,

H = Magnetic field intensity  ----

A
 m
µ 0 = permeability of free space
µ r = relative permeability of a material
µ = permeability of a material

• Permeability is approximately linear for smaller electric fields, but with larger
magnetic fields the materials saturate and the value of B reaches a maximum value.

1.5

linear saturation
H

4000

Figure 23.1 Saturation for a mild steel (approximately)

 electromagnetics - 23.4

1.5

silicon sheet steel

cast steel

1.0

B (T)

0.5
cast iron

0
0 500 1000
H (At/m)

Figure 23.2 Magnetization curves (Sen, 1989)

• Flux density about a wire

 electromagnetics - 23.5

I
B = --------- For an infinitely long straight conductor
2πr
where,
B = Flux density  -------2- or Tesla
Wb
 
m
I = Current in the conductor (A)
r = radial distance from the conductor

• Flux and flux density,

Φ = ∫ B dA
where,
Φ = Flux density ( Wb )

A = Cross section area perpendicular to flux

• When a material is used out of the saturation region the permeabilities may be
written as reluctances,

L
R = -------
µA
where,
R = reluctance of a magnetic path
L = length of a magnetic path
A = cross section area of a magnetic path
 electromagnetics - 23.6

• Electric circuit analogy

Ni F
Φ = ------------- = ---
F  ------
l - R
 µA

• Example,

lc
I R c = -----------
µc Ac

lg lg
R g = ------------
µg Ag

lc
Ni
Φ = ------------------
Rc + R g

• Faraday’s law,
 electromagnetics - 23.7

d
e = N ----- Φ For a coil
dt
where,
e = the potential voltage across the coil
N = the number of turns in the coil

• Field energy,

2 2
B B
W = ------ V = ------ Al
2µ 2µ

• Force can be derived from the energy,

2
B
F = ------ A
2µ

• The basic property of induction is that it will (in the presence of a magnetic field)
convert a changing current flowing in a conductor to a force or convert a force to a current
flow from a change in the current or the path.
 electromagnetics - 23.8

F = ( I × B )L B
NOTE: As with all cross prod-
F = ( L × B )I ucts we can use the right
hand rule here.
I, L
F
where,
L = conductor length
F = force (N)
The FBD/schematic equivalent is,

I F
M

Figure 23.3 The current and force relationship

• We will also experience an induced current caused by a conductor moving in a

magnetic field. This is also called emf (Electro-Motive Force)
 electromagnetics - 23.9

B
e m = ( v × B )L

where,
v
e m = electromotive force (V)
e
φ = magnetic flux (Wb - webers)
v = velocity of conductor
The FBD/schematic equivalent

e
+
M
-

Figure 23.4 Electromagnetically induced voltage

• Hysteresis

23.3 EXAMPLE SYSTEMS

• These systems are very common, take for example a DC motor. The simplest
motor has a square conductor loop rotating in a magnetic field. By applying voltage the
wires push back against the magnetic field.
 electromagnetics - 23.10

y
I
B
x
1 I
z

2 5

4
b 3
a

axis of
rotation ω

Figure 23.5 A motor winding in a magnetic field

 electromagnetics - 23.11

For wire 3,

r cos ( ωt )
– rω sin ( ωt ) 0
P 3 = r sin ( ωt ) V3 = rω cos ( ωt ) B = –B
b b
– --- to --- 0 0
2 2

de m3 = ( V × B )dL

b
--- – rω sin ( ωt ) 0
∫–b---
2
e m3 = rω cos ( ωt ) × – B dr
2 0 0

b 0
---
∫–b---
2
e m3 = 0 dr
2 Brω sin ( ωt )
b
---
2
2
b 2
= Bω sin ( ωt )  --- –  – --- = 0
r b
e m3 = B ---- ω sin ( ωt )
2 b
2 2
– ---
2
For wires 1 and 5,

By symmetry, the two wires together will act like wire 3. Therefore they both
have an emf (voltage) of 0V.

e m1 = e m5 = 0V

Figure 23.6 Calculation of the motor torque

 electromagnetics - 23.12

For wire 2 (and 4 by symmetry),

b--- b
cos ( ωt ) – --- ω sin ( ωt )
2 2 0
P2 = b V2 = b B = –B
--- sin ( ωt ) --- ω cos ( ωt )
2 2 0
0toa 0
de m2 = ( V × B )dL

b
– --- ω sin ( ωt )
a 2 0
e m2 = ∫0 b---
ω cos ( ωt )
× – B dl
2 0
0

0
a b
∫0
e m2 = 0 dl = aB --- cos ( ωt )
b 2
B --- ω cos ( ωt )
2

e m4 = e m2

For the total loop,

e m = e m1 + e m2 + e m3 + e m4 + e m5

b b
e m = 0 + aB --- ω cos ( ωt ) + 0 + aB --- ω cos ( ωt ) + 0
2 2
e m = aBbω cos ( ωt )

Figure 23.7 Calculation of the motor torque (continued)

• As can be seen in the previous equation, as the loop is rotated a voltage will be
generated (a generator), or a given voltage will cause the loop to rotate (motor).

• In this arrangement we have to change the polarity on the coil every 180 deg of
rotation. If we didn’t do this the loop the torque on the loop would reverse for half the
motion. The result would be that the motor would swing back and forth, but not rotate
fully. To make the torque push consistently in the same direction we need to reverse the
 electromagnetics - 23.13

applied voltage for half the cycle. The device that does this is called a commutator. It is
basically a split ring with brushes.

e m = aBbω cos ( ωt )

• Real motors also have more than a single winding (loop of wire). To add this into
the equation we only need to multiply by the number of loops in the winding.

e m = NaBbω cos ( ωt )

• As with most devices the motor is coupled. This means that one change, say in
torque/force will change the velocity and hence the voltage. But a change in voltage will
also change the current in the windings, and hence the force, etc.

• Consider a motor that is braked with a constant friction load of Tf.

F w = ( I × B )L = IBa

T w = 2  F × --- = Fb = IBab
b
 2

∑M = T w – T f = Jα

IBab – T f = Jα

Figure 23.8 Calculation of the motor torque (continued)

• We still need to relate the voltage and current on the motor. The equivalent circuit
for a motor shows the related components.
 electromagnetics - 23.14

IA
RA LA

em
eA +
-
-

where,

I A, e A = voltage and current applied to the armature (motor supply)

R A, L A = equivalent resistance and inductance of windings
d
∑V = e A – I A R A – L A ----- I A – e m = 0
dt
we can now add in the other equations,

d
e A – I A RA – L A ----- I A – NaBbω cos ( ωt ) = 0
dt
and recall the previous equation,

IBab – T f = Jα

Figure 23.9 Calculation of the motor torque (continued)

• Practice problem,
 electromagnetics - 23.15

Write the transfer function relating the displacement ’x’ to the

current ’I’

N Ks
M

Kd
S
R
x

Figure 23.10 Drill problem: Electromotive force

• Consider a motor with a separately excited magnetic field (instead of a perma-

nent magnet there is a coil that needs a voltage to create a magnetic field). The model is
similar to the previous motor models, but the second coil makes the model highly nonlin-
ear.
 electromagnetics - 23.16

23.4 SUMMARY

23.5 PRACTICE PROBLEMS

1.

23.6 PRACTICE PROBLEM SOLUTIONS

23.7 ASSIGNMENT PROBLEMS

 fluids - 24.1

24. FLUID SYSTEMS

Topics:

Objectives:

24.1 SUMMARY

• Fluids are a popular method for transmitting power (hydraulics). Basically, by

applying a pressure at one point, we can induce flow through a pipe/value/orifice.

24.2 MATHEMATICAL PROPERTIES

• Fluids do work when we have a differential pressure on a surface. The pressure

may be expressed as an absolute value. More correctly we should consider atmospheric
pressure (gauge pressure).

• When we deal with fluids we approximate them as incompressible.

• Fluids observe some basic laws,

w = flow rate
p = pressure
 fluids - 24.2

24.2.1 Resistance

• If fluid flows freely, we say it is without resistance. In reality, every fluid flow
experiences some resistance. Even a simple pipe has resistance. Of similar interest is the
resistance of a value.

w = K ∆p

where,
w = flow rate through the pipe
∆p = pressure difference across valve/pipe
K = a constant specific to the pipe/valve/orifice

NOTE: In this case the relationship between pressure drop and flow are non-linear.
We have two choices if we want to analyze this system.
1. We can do a non-linear analysis (e.g. integration)
2. We can approximate the equation with a linear equation. This is only good for
operation near the chosen valve position. As the flow rate changes significantly
the accuracy of the equation will decrease.
w

∆p ≈ R ( w – w )
∆w
∆ ( ∆p ) w
R ≈ --------------- R = 2 -----2-
∆w K
w

∆p
∆ ( ∆p )

Figure 24.1 Fluid flow resistance

• Resistance may also result from valves. Valves usually restrict flow by reducing
an area for fluid to flow through.

• A simple form of valve is a sliding plunger. The valve below is called a two way
 fluids - 24.3

valve because it will allow fluid to flow in or out.

NOTE: For this type of valve the

fluid in is typically under pres-
Fluid in sure. When the valve slide is up
the pressure is applied to the cyl-
inder, and causes it to actuate. The
to cylinder fluid out is typically not under
pressure, and just returns fluid to a
reservoir tank. When used this
way a cylinder needs a spring, or
some other force to return it and
Fluid out force the fluid out.

Figure 24.2 Fluid servo valve

• Four way valve allow fluid force to be applied in both directions of a cylinder
motion.
 fluids - 24.4

to
reservoir

fluid in fluid in

to
reservoir

retract
no motion
advance

Figure 24.3 Fluid flow control valves

24.2.2 Capacitance

• Fluids are often stored in reservoirs or tanks. In a tank we have little pressure
near the top, but at the bottom the mass of the fluid above creates a hydrostatic pressure.
Other factors also affect the pressure, such as the shape of the tank, or whether or not the
top of the tank is open.

• To calculate the pressure we need to integrate over the height of the fluid.
 fluids - 24.5

A
Fp

h
F p + Fg Fp + Fg
Fg

F p + Fg
Fg
P = ------
A
F g = ρV = ρgAh

ρgAh
P = -------------- = ρgh
A

Figure 24.4 Pressures on fluid elements

• Consider a tank as a capacitor. As fluid is added the height of the fluid rises, and
the hydrostatic pressure increases. Hence we can pump fluid into a tank to store energy,
and letting fluid out recovers the energy. A very common application of this principle is a
municipal water tower. Water is pumped into these tanks. As consumers draw water
through the system these tanks provide pressure to the system. When designing these
tanks we should be careful to keep the cross section constant (e.g. a cylinder). If the cross
section varies then the fluid pressure will not drop at a linear rate and you won’t be able to
use linear analysis techniques (eg., Laplace).

• The mathematical equations for a constant cross section tank are,

A
C = ------
ρg
d
w = C ----- p
dt
 fluids - 24.6

24.2.3 Power Sources

• As with most systems we need power sources. In hydraulics these are pumps that
will provide pressure and/or flow to the system.

• One type of pump uses a piston.

Fluid In
Normally closed
inlet valve

Fluid Out

piston
Normally
Closed Valve

In this common form of piston pump, the piston rod is drawn

back creating suction that holds the valve closed, and pulls
fluid into the chamber. When the cylinder is full of fluid
the piston motion is reversed, creating a pressure, and forcing
the inlet vale closed, and the outlet valve open, and the
fluid is pumped out. The fluid volume can be controlled by
using the cylinder size, and piston strokes

Figure 24.5 A piston driven hydraulic pump

• A geared hydraulic pump is pictured below. Other types use vanes and pistons.
 fluids - 24.7

a geared hydraulic pump

Figure 24.6 A gear driven hydraulic pump

• Vane based pumps can be used to create fluid flow. As the pump rotates the vanes
move to keep a good seal with the outer pump wall. The displacement on the advance and
return sides are unequal (aided by the sliding vanes). The relative displacement across the
pump determines the fluid flow.

Fluid in
As the center core of the pump rotates, the vanes
will slide in and out. Fluid is trapped between
vanes. On the inlet side the volume enclosed by
the vanes expands, drawing fluid in. On the outlet
side the volume between vanes decreases, forcing
fluid out.

Fluid out
 fluids - 24.8

Figure 24.7 A vaned hydraulic pump

• As with the resistance of valves, these are not linear devices. It is essential that
we linearize the devices. To do this we look at the pressure flow curves. (Note: most
motors and engines have this problem)

∆p ∆ ( ∆p )
K = ---------------
∆p ∆w
1
w ≈ ---- ( ∆p – ∆p )
K
∆ ( ∆p )

w
∆w

Figure 24.8 Linearizing a hydraulic valve ?????

24.3 EXAMPLE SYSTEMS

• We can model a simple hydraulic system using the elements from before. Con-
sider the example below,

pump valve tank

Figure 24.9 A hydraulic system example

 fluids - 24.9

∆p
For the pump the relationship is
shown in the graph, and it will
operate at the point marked.

w
For the valve the relationship is
given below

∆p

w
The pipes are all equal length and
have the relationships shown
below.

∆p

Figure 24.10 A hydraulic system example (continued)

 fluids - 24.10

24.4 SUMMARY

24.5 PRACTICE PROBLEMS

1.

24.6 PRACTICE PROBLEMS SOLUTIONS

24.7 ASSIGNMENT PROBLEMS

 thermal systems - 25.1

25. THERMAL SYSTEMS

Topics:

Objectives:

25.1 INTRODUCTION

• Energy can be stored and transferred in materials in a number of forms. Thermal

energy (heat) is stored and transmitted through all forms of matter.

25.2 MATHEMATICAL PROPERTIES

25.2.1 Resistance

• A universal property of energy is that it constantly strives for equilibrium. This

means that a concentration of energy will tend to dissipate. When material separates
regions of different temperatures it will conduct heat energy at a rate proportional to the
difference. Materials have a measurable conductivity or resistance (note: this is different
than electrical conductivity and resistance.)
 thermal systems - 25.2

------ = --1-  dθ
------
1 dq
q = --- ( θj – θ i )
R dl R  dl 
where,

q = heat flow rate from j to i (J/s or watts)

R = thermal resistance
θ = temperature

d
R = -------
Aα

d = length of thermal conductor

A = cross section area of thermal conductor

α = thermal conductivity (W/mK)

Figure 25.1 Thermal resistance

• When dealing with thermal resistance there are many parallels to electrical resis-
tance. The flow of heat (current) is proportional to the thermal difference (voltage). If we
have thermal systems in parallel or series they add as normal resistors do.

R1
R1 R2
R2

1- 1 1
----- = ------ + ------ RT = R1 + R2
RT R1 R2

Figure 25.2 Parallel and series thermal resistance

 thermal systems - 25.3

25.2.2 Capacitance

• Heat energy is absorbed at different rates in different materials - this rate is

referred to as thermal capacitance. And as long as the material has no major changes in
structure (i.e., gas-solid or phase transition) this number is relatively constant.

1 dθ 1- dq
∆θ = ---- ( q in – q out ) ------ = --- ------
C dt C dt
where,
C = thermal capacitance

C = Mσ

where,
M = mass of thermal body
σ = specific heat of material in mass

Figure 25.3 Thermal capacitance

• One consideration when dealing with heating capacitance is that the heat will not
instantly disperse throughout the mass. When we want to increase the rate of heat absorp-
tion we can use a mixer (with a gas or fluid). A mixer is shown in the figure below. This
mixer is just a rotating propeller that will cause the liquid to circulate.

Figure 25.4 A mixer to prevent uneven temperature distribution in thermal storage

 thermal systems - 25.4

25.2.3 Sources

• If we plan to design a thermal system we will need some sort of heat source. One
popular heating source is an electrical heating element.

• heating coils are normally made out of some high resistance metal/alloy such as
nichrome (nickel and chrome) or tungsten.

• If we run a current through the wire the resistance will generate heat. As these
metals get hotter the resistance rises, hence the temperature is self regulating. If we control
the voltage and current we can control the amount of heat delivered by the coil.

resistance
I

+
V
-

temp

P = IV

where,
P = power generated as heat
I = current into coil
V = voltage across coil

Figure 25.5 Electrical heating elements

25.3 EXAMPLE SYSTEMS

• We know that as we heat materials at one point, they do not instantly heat at all
points, there is some delay (consider a spoon in a hot bowl of soup). This delay is a func-
tion of heat capacitance (yes there is a parallel to the electrical description).
 thermal systems - 25.5

• Consider an insulated sealed chamber with a resistive barrier between sides, and
a heating element in one side.

qi

θ 1, C 1 R θ 2, C 2

Given
q 0 = heat flow into the left chamber from the heating element

θ 1, C 1 = initial temperature and heat capacitance of left side

θ 2, C 2 = initial temperature and heat capacitance of right side

Next,
q 1 = heat flow through the center barrier
1
q 1 = --- ( θ 1 – θ 2 )
R
d- 1 d- 1
---- θ 1 = ------ ( q0 – q 1 ) ---- θ 2 = ------ ( q 1 )
dt C1 dt C2

sθ 1 = ------  q 0 – --- ( θ 1 – θ 2 )
1 1 1
C1  R  sθ2 = ---------- ( θ 1 – θ 2 )
RC 2
Rq o – sRC 1 θ 1 = θ 1 – θ 2 θ1 = θ 2 + sRC 2 θ 2

θ 1 ( 1 + sRC 1 ) = θ2 + Rq o
θ2 + Rq o
θ 1 = ---------------------
-
1 + sRC 1

Figure 25.6 A two chamber thermal system

 thermal systems - 25.6

θ 2 + Rq o
θ 1 = ---------------------- = θ 2 + sRC 2 θ 2
1 + sRC 1
θ 2 + Rq o
------------------------------
- = 1 + sRC 1
θ 2 + sRC 2 θ 2
Rq o
 ---------
1  θ2 
---------------------- + ---------------------- = 1 + sRC 1
1 + sRC 2 1 + sRC 2

Rq o
1 +  --------- = 1 + s ( RC 1 + RC 2 ) + s R C 1 C 2
2 2
θ2
Rq o
--------- = s ( RC 1 + RC 2 ) + s 2 R 2 C 1 C 2
θ2
q
----o- = s ( C1 + C 2 ) + s 2 RC 1 C 2
θ2
qo 
 -----------------
 RC 1 C 2 A B
θ2 = -------------------------------------- = --- + ----------------------------
C + C s C 1 + C2
s  ------------------ 2
1
- + s2 s + ------------------ -
 RC 1 C 2  RC 1 C 2

  qo   qo 
 ----------------
  RC C ----------------
-   RC 1 C 2
-
qo
 1 2 
A = lim -------------------------------------- s = --------------------------------------- = -------------------
s → 0  C 1 + C 2  C 1 + C 2
 ------------------ C1 + C2
 s ------------------- + s 2  - + (0)
  RC 1 C 2    RC 1 C 2 

  qo   qo 
 ----------------
  ----------------
RC 1 C 2
-
C 1 + C 2    RC 1 C 2
-
– qo
B =  
lim  --------------------------------------  s + -------------------  = ---------------------------- = -------------------
C 1 + C 2  C 1 + C 2 RC 1 C 2 C 1 + C 2 C1 + C2
s → – ------------------
RC 1 C 2  RC C 
 s ------------------
- + s
2
 – ------------------
-
 1 2   RC 1 C 2 

qo –qo
------------------- ------------------
-
C1 + C2 C1 + C2
θ 2 = ------------------- + ----------------------------
s C 1 + C2
s + ------------------ -
RC 1 C 2

Figure 25.7 A two chamber thermal system (continued)

 thermal systems - 25.7

–1 qo –1 1 –1 1
L ( θ 2 ) = ------------------- L --- – L ----------------------------
C 1 + C2 s C1 + C2
s + -------------------
RC 1 C 2
 C1 + C 2
–  ------------------ t
qo RC 1 C 2
θ 2 ( t ) = ------------------- 1 – e
C1 + C 2

Figure 25.8 A two chamber thermal system (continued)

25.4 SUMMARY

25.5 PRACTICE PROBLEMS

1.

25.6 PRACTICE PROBLEM SOLUTIONS

25.7 ASSIGNMENT PROBLEMS

 multiaxis motion - 26.1

26. OPTIMIZATION

Topics:
•

Objectives:
•

26.1 INTRODUCTION

• For simple examples use derivatives and find the maxima/minima

• Consider the example below to find the

L =

• local and global optimums

 multiaxis motion - 26.2

• search functions, gradient descent, random walk, simplex

26.2 OBJECTIVES AND CONSTRAINTS

• Objectives are those things we want to minimize, or maximize.

- money
- time
- mass
- volume
- power consumption
- some combination of factors

• Expressed as a function of variables that provides a value

• Consider the example of building a fenced pasture. In this case when the area
becomes too large, there is a reduced value. We want to maximize the value of V.
 multiaxis motion - 26.3

C fence = 20  ----   ( 2w + 2d )
$
 m
where,

C fence = cost to construct the fence ($)

w = width of the pasture (m)

d = depth of the pasture (m)

= 0.35  -----2-
$
C land
 
m
where,
C land = cost for the land

R = 0.05  -----------
$ - 2
2 
wd wd < 300m
m yr

R = 0.01  -----------
$ - 2 2
2 
( wd – 300m ) + 60$ wd ≥ 300m
m yr

where,
R = revenue generated by pasture land

V = R – C fence – C land

where,
V = total value

Figure 26.1 Example cost function for building a fence around a pasture

• The cost function can be written as..

 multiaxis motion - 26.4

double cost(double w, double d){

double value;
double cfence, cland, R;

Cfence = 40*(w + d);

Cland = 0.05*w*d;
if (w*d < 300){
R = 0.05 * w * d;
} else {
R = 0.01 * (w * d - 300) + 60;
}
value = R - Cfence - Cland

return value;
}

Figure 26.2 A subroutine for cost function calculation

• Constraints are boundaries that cannot be crossed.

• Example of constraints, the pasture cannot be larger than one 1600m be 1600m
beacuse of the constraints of an existing road system.

w ≤ 1600m

d ≤ 1600m

Figure 26.3 Example constraint functions for a pasture

• The cost function can be written as..

 multiaxis motion - 26.5

double cost(double w, double d){

double value;
double cfence, cland, R;

Cfence = 40*(w + d);

Cland = 0.05*w*d;
if (w*d < 300){
R = 0.05 * w * d;
} else {
R = 0.01 * (w * d - 300) + 60;
}
value = R - Cfence - Cland

if(w > 1600) value = 1000000;

if(d > 1600) value = 1000000;

return value;
}

Figure 26.4 A subroutine for cost function calculation

• Slack variables allow constraints to be considered as part of the cost function.

Helps with a system with many local minimum.

4 4
C penaly =  ---------------- +  ----------------
w d
 1600m  1600m

V = R – C fence – C land – C penaly

Figure 26.5 Example of slack variables for including constraints

• The cost function can be written as..

 multiaxis motion - 26.6

double cost(double w, double d){

double value;
double cfence, cland, R;
double slack;

Cfence = 40*(w + d);

Cland = 0.05*w*d;
if (w*d < 300){
R = 0.05 * w * d;
} else {
R = 0.01 * (w * d - 300) + 60;
}
// slack variable
slack = pow((w/1600), 4) + pow((d/1600), 4);
value = R - Cfence - Cland - slack;

return value;
}

Figure 26.6 A subroutine for cost function calculation

26.3 SEARCHING FOR THE OPTIMUM

• Local Search Space

• A topographical map shows the relationship between search parameters and cost
values.
 multiaxis motion - 26.7

Maximum (or minimum) Search path

Objective Space

Constraints

Figure 26.7 Local searches

• Global Search space. In this case the system becomes ’stuck’ in a local mimina.
 multiaxis motion - 26.8

Local Optima Global Optima

Figure 26.8 Global searches

• Global Search space. In this case the system searches all maxim.
 multiaxis motion - 26.9

Local Optima Global Optima

Figure 26.9 A Global Search

26.4 OPTIMIZATION ALGORITHMS

• The search algorithms change system parameters and try to lower system param-
eters.

• The main question is how to change the system paramters to minimize the system
value.

26.4.1 Random Walk

 multiaxis motion - 26.10

26.4.2 Gradient Decent

26.4.3 Simplex

26.5 SUMMARY

26.6 PRACTICE PROBLEMS

1.

26.7 PRACTICE PROBLEM SOLUTIONS

1.

26.8 ASSIGNMENT PROBLEMS

1.
 state space analysis - 27.1

27. FINITE ELEMENT ANALYSIS (FEA)

Topics:

Objectives:

27.1 INTRODUCTION
 state space analysis - 27.2

27.2 FINITE ELEMENT MODELS

• Consider a central node ’i’ connected to neighboring nodes with springs.

nb

y
Ks
x

na Ks ni Ks

nc

Fi Ks

nd

• Equations can be written to relate the position of node i to the surrounding nodes
 state space analysis - 27.3

and the applied force.

Ks ( y b – yi )
+
∑ Fx = Fix – Ks ( xi – xa ) + K s ( xc – x i ) = 0
Ks ( x i – x a )
Ks ( x c – xi ) x i ( 2K s ) + x a ( – K s ) + x c ( – K s ) = F i x

F i = ( F i x, F i y ) + ∑ Fx = F iy – K s ( y i – y d ) + K s ( y b – y i ) = 0

Ks ( yi – yd ) y i ( 2K s ) + y b ( – K s ) + y d ( – K s ) = F i y

xi
yi
xa
ya
Fix 2K s 0 – K s 0 0 0 – K s 0 0 0 xb
=
Fiy 0 2K s 0 0 0 – K s 0 0 0 – K s y b
xc
yc
xd
yd
 state space analysis - 27.4

27.3 FINITE ELEMENT MODELS

• Consider a central node ’i’ connected to neighboring nodes with springs.

na nb
y
Ks
Fa x
Fb Ks
Ks

Ks
nd nc

Fd Fc

• Equations can be written to relate the position of node i to the surrounding nodes
and the applied force. The ’x’ and ’y’ values are deflections from the unloaded state. The
’Ks’ value is based on the material stiffness and the geometry of the elements.
+
Ks ( xb – xa ) ∑ Fx = F ax + K s ( x b – x a ) = 0
( F a x, F a y )
∴F ax = x a ( K s ) + x b ( – K s )
Ks ( ya – y d )
+ ∑ Fx = F ay – K s ( y a – y d ) = 0

∴F ay = y a ( K s ) + y d ( – K s )

+
Ks ( xb – xa )
∑ Fx = Fbx – Ks ( xb – xa ) = 0

∴F bx = x a ( – K s ) + x b ( K s )
( Fb x, F by )

+ ∑ Fx = Fby – Ks ( yb – yc ) = 0
Ks ( y b – yc )
∴F b y = y b ( K s ) + y c ( – K s )
 state space analysis - 27.5

Ks ( yb – yc ) +
∑ Fx = F cx – Ks ( x c – x d ) = 0
Ks ( xc – xd )
∴F cx = x c ( K s ) + x d ( – K s )

( F cx, F cy ) + ∑ Fx = F cy + Ks ( y b – y c ) = 0

∴F cy = y b ( – K s ) + y c ( K s )

Ks ( ya – yd ) +
∑ Fx = Fdx + Ks ( xc – x d ) = 0

Ks ( x c – xd ) ∴F d x = x c ( – K s ) + x d ( K s )

( F d x, F d y ) + ∑ Fx = F dy + K s ( y a – y d ) = 0

∴F dy = y a ( – K s ) + y d ( K s )

Fax Ks 0 –Ks 0 0 0 0 0 xa
Fay 0 Ks 0 0 0 0 0 –K s ya
Fbx –Ks 0 Ks 0 0 0 0 0 xb
Fby 0 0 0 Ks 0 –Ks 0 0 yb
=
Fcx 0 0 0 0 Ks 0 –Ks 0 xc
Fcy 0 0 0 –Ks 0 Ks 0 0 yc
Fdx 0 0 0 0 –Ks 0 Ks 0 xd
Fdy 0 –Ks 0 0 0 0 0 Ks yd

local stiffness matrix

• This can be combined into a more

 state space analysis - 27.6

• A four element mesh

n2 n3
n1
Ks Ks

Ks #1 Ks #2 Ks

Ks Ks
n5 n6
n4

Ks #3 Ks #4 Ks

Ks Ks
n7 n8 n9
 state space analysis - 27.7

For element #1
F 1x Ks 0 –K s 0 0 0 0 0 x1
Local stiffness matrix
F 1y 0 Ks 0 0 0 0 0 –Ks y1
F 2x –Ks 0 Ks 0 0 0 0 0 x2
F 2y 0 0 0 Ks 0 –K s 0 0 y2
=
F 4x 0 0 0 0 Ks 0 –Ks 0 x4
F 4y 0 0 0 –Ks 0 Ks 0 0 y4
F 5x 0 0 0 0 –Ks 0 Ks 0 x5
F 5y 0 –Ks 0 0 0 0 0 Ks y5

Global stiffness matrix

F 1x x1
F 1y y1
Ks 0 –Ks 0 0 0 0 0 0 0 0 0 0 0 0 0
F 2x x2
0 Ks 0 0 0 0 0 0 0 –Ks 0 0 0 0 0 0
F 2y y2
–K s 0 Ks 0 0 0 0 0 0 0 0 0 0 0 0 0
F 3x x3
0 0 0 Ks 0 0 0 –K s 0 0 0 0 0 0 0 0
F 3y y3
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
F 4x x
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4
F 4y 0 0 0 0 0 0 Ks 0 –K s 0 0 0 0 0 0 0 y4
F 5x 0 0 0 –Ks 0 0 0 K 0 0 0 0 0 0 0 0 x5
= s
F 5y 0 0 0 0 0 0 –Ks 0 Ks 0 0 0 0 0 0 0 y5
F 6x 0 –Ks 0 0 0 0 0 0 0 K 0 0 0 0 0 0 x6
s
F 6y 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 y6
F 7x 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 x7
F 7y 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 y7
F 8x 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 x8
F 8y 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 y8
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
F 9x x9
F 9y y9
 state space analysis - 27.8

For element #2 F2x Ks 0 –Ks 0 0 0 0 0 x2

F2y 0 Ks 0 0 0 0 0 –K s y2
local stiffness matrix F3x –Ks 0 Ks 0 0 0 0 0 x3
F3y 0 0 0 Ks 0 –Ks 0 0 y3
=
F5x 0 0 0 0 Ks 0 –Ks 0 x5
F5y 0 0 0 –Ks 0 Ks 0 0 y5
F6x 0 0 0 0 –K s 0 Ks 0 x6
F6y 0 –Ks 0 0 0 0 0 Ks y6

added to global matrix

F 1x x1
F 1y Ks 0 –K s 0 0 0 0 0 0 0 0 0 0 0 0 y1
0
F 2x 0 Ks 0 0 0 0 0 0 0 –K s 0 0 0 0 0 0 x2
F 2y – K s 0 2K s 0 – K s 0 0 0 0 0 0 0 0 0 0 0 y2
F 3x 0 0 0 2K s 0 0 0 –Ks 0 0 0 0 0 0 0 0 x3
F 3y 0 0 –K s 0 Ks 0 0 0 0 0 0 0 0 0 0 0 y3
F 4x 0 0 0 0 0 Ks 0 0 0 0 0 0 0 0 0 0 x4
F 4y 0 0 0 0 0 0 Ks 0 –Ks 0 0 0 0 0 0 0 y4
F 5x 0 0 0 –Ks 0 0 0 Ks 0 0 0 0 0 0 0 0 x5
=
F 5y 0 0 0 0 0 0 – K s 0 2K s 0 0 0 0 0 0 0 y5
F 6x 0 –Ks 0 0 0 0 0 0 0 2K s 0 0 0 0 0 0 x 6
F 6y 0 0 0 0 0 0 0 0 0 0 Ks 0 0 0 0 0 y6
F 7x 0 0 0 0 0 0 0 0 0 0 0 K 0 0 0 0 x7 s
F 7y 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 y7
F 8x 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 x8
F 8y 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 y8
F 9x 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 x9
F 9y y9
 state space analysis - 27.9

For element #3 F4x Ks 0 –Ks 0 0 0 0 0 x4

F4y 0 Ks 0 0 0 0 0 –K s y4
F5x –Ks 0 Ks 0 0 0 0 0 x5
F5y 0 0 0 Ks 0 –Ks 0 0 y5
=
F7x 0 0 0 0 Ks 0 –Ks 0 x7
F7y 0 0 0 –Ks 0 Ks 0 0 y7
F8x 0 0 0 0 –Ks 0 Ks 0 x8
F8y 0 –Ks 0 0 0 0 0 Ks y8
 state space analysis - 27.10

For element #4 F5x Ks 0 –Ks 0 0 0 0 0 x5

F5y 0 Ks 0 0 0 0 0 –K s y5
F6x –Ks 0 Ks 0 0 0 0 0 x6
F6y 0 0 0 Ks 0 –Ks 0 0 y6
=
F8x 0 0 0 0 Ks 0 –Ks 0 x8
F8y 0 0 0 –Ks 0 Ks 0 0 y8
F9x 0 0 0 0 –K s 0 Ks 0 x9
F9y 0 –K s 0 0 0 0 0 Ks y9

• If the input forces are known, then the resulting displacements of the nodes can
be calculated by inverting the matrix. Consider the matrix for a single node.
 state space analysis - 27.11

–1 Fax
xa Ks 0 –Ks 0 0 0 0 0
ya 0 Ks 0 0 0 0 0 –Ks Fay
xb –Ks 0 Ks 0 0 0 0 0 Fbx
yb 0 0 0 Ks 0 –Ks 0 0 Fby
=
xc 0 0 0 0 Ks 0 –Ks 0 F cx
yc 0 0 0 –Ks 0 Ks 0 0 F cy
xd 0 0 0 0 –Ks 0 Ks 0 Fdx
yd 0 –K s 0 0 0 0 0 Ks Fdy

• If we assume that node ’c’ is fixed in the ’x’ and ’y’ directions, the matrix can
reflect this by setting the appropriate matrix rows to zero.
–1
xa Ks 0 –Ks 0 0 0 F ax
ya 0 Ks 0 0 0 –K s F ay
xb –Ks 0 Ks 0 0 0 F bx
=
yb 0 0 0 Ks 0 0 F by
xd 0 0 0 0 Ks 0 F dx
yd 0 –K s 0 0 0 Ks F dy

• The displacements can then be found by selecting values for the coefficents and
solving the matrix. We can select a value of 1000 for the stiffness, and a force of 10 will be
 state space analysis - 27.12

applied at point ’a’ in the positive ’x’ direction.

–1
3 3
xa 10 0 – 10 0 0 0
3 3
10
ya 0 10 0 0 0 –10 0
xb 3 3
= – 10 0 10 0 0 0 0
yb 3 0
0 0 0 10 0 0
xd 3 0
0 0 0 0 10 0
yd 0
3 3
0 – 10 0 0 0 10

• This method generates very large matrices, easily into the millions.

• To reduce the matrix size techniques such as symmetry are commonly used.

• Strains can be found by calculating the relative displacements of neighboring

points. These can then be used to calculate the stresses.

• More complex elements are commonly used depending upon the stress condi-
tions, part geometry and other factors.

• There are a variety of finite element element methods and applications

Computation Fluid Dynamics

Nonlinear deformation

27.4 SUMMARY

•
 state space analysis - 27.13

27.5 PRACTICE PROBLEMS

27.6 PRACTICE PROBLEM SOLUTIONS

27.7 ASSIGNMENT PROBLEMS

1.

27.8 BIBLIOGRAPHY

How, J, "16.31 Feedback Control Course Notes", MIT Opencourseware website.

 Fuzzy logic - 28.1

28. FUZZY LOGIC

<TODO - Find an implementation platform and add section>

Topics:
• Fuzzy logic theory; sets, rules and solving
•

Objectives:
• To understand fuzzy logic control.
• Be able to implement a fuzzy logic controller.

28.1 INTRODUCTION

Fuzzy logic is well suited to implementing control rules that can only be expressed
verbally, or systems that cannot be modelled with linear differential equations. Rules and
membership sets are used to make a decision. A simple verbal rule set is shown in Figure
28.1. These rules concern how fast to fill a bucket, based upon how full it is.

1. If (bucket is full) then (stop filling)

2. If (bucket is half full) then (fill slowly)
3. If (bucket is empty) then (fill quickly)

Figure 28.1 A Fuzzy Logic Rule Set

The outstanding question is "What does it mean when the bucket is empty, half
full, or full?" And, what is meant by filling the bucket slowly or quickly. We can define
sets that indicate when something is true (1), false (0), or a bit of both (0-1), as shown in
Figure 28.2. Consider the bucket is full set. When the height is 0, the set membership is 0,
so nobody would think the bucket is full. As the height increases more people think the
bucket is full until they all think it is full. There is no definite line stating that the bucket is
full. The other bucket states have similar functions. Notice that the angle function relates
the valve angle to the fill rate. The sets are shifted to the right. In reality this would proba-
bly mean that the valve would have to be turned a large angle before flow begins, but after
that it increases quickly.
 Fuzzy logic - 28.2

1 1
bucket is full stop filling
0 0

1 1
bucket is half full fill slowly
0 0

1 1
bucket is empty fill quickly
0 0
height angle

Figure 28.2 Fuzzy Sets

Now, if we are given a height we can examine the rules, and find output values, as
shown in Figure 28.3. This begins be comparing the bucket height to find the membership
for bucket is full at 0.75, bucket is half full at 1.0 and bucket is empty at 0. Rule 3 is
ignored because the membership was 0. The result for rule 1 is 0.75, so the 0.75 member-
ship value is found on the stop filling and a value of a1 is found for the valve angle. For
rule 2 the result was 1.0, so the fill slowly set is examined to find a value. In this case there
is a range where fill slowly is 1.0, so the center point is chosen to get angle a2. These two
results can then be combined with a weighted average to get
0.75 ( a1 ) + 1.0 ( a2 )
angle = ---------------------------------------------- .
0.75 + 1.0
 Fuzzy logic - 28.3

1. If (bucket is full) then (stop filling)

1 1
bucket is full stop filling
0 0
height angle
a1
2. If (bucket is half full) then (fill slowly)
1 1
bucket is half full fill slowly
0 0
height a2 angle

3. If (bucket is empty) then (fill quickly)

1 1
bucket is empty fill quickly
0 0
height angle

Figure 28.3 Fuzzy Rule Solving

An example of a fuzzy logic controller for controlling a servomotor is shown in

Figure 28.4 [Lee and Lau, 1988]. This controller rules examines the system error, and the
rate of error change to select a motor voltage. In this example the set memberships are
defined with straight lines, but this will have a minimal effect on the controller perfor-
mance.
 Fuzzy logic - 28.4

Fuzzy Vmotor Motor vactual

vdesired verror Imotor
Logic Power Servo
+ Motor
- Controller Amplifier

The rules for the fuzzy logic controller are;

1. If verror is LP and d/dtverror is any then Vmotor is LP.
2. If verror is SP and d/dtverror is SP or ZE then Vmotor is SP.
3. If verror is ZE and d/dtverror is SP then Vmotor is ZE.
4. If verror is ZE and d/dtverror is SN then Vmotor is SN.
5. If verror is SN and d/dtverror is SN then Vmotor is SN.
6. If verror is LN and d/dtverror is any then Vmotor is LN.

The sets for verror, d/dtverror, and Vmotor are;

d/
verror dtverror Vmotor

1 1 1

LN 0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

1 1 1

SN 0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

1 1 1
ZE 0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

1 1 1
SP 0
rps
0
rps/s
0
V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

1 1 1
LP
0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

Figure 28.4 A Fuzzy Logic Servo Motor Controller

Consider the case where verror = 30 rps and d/dt verror = 1 rps/s. Rule 1to 6 are calcu-
lated in Figure 28.5.
 Fuzzy logic - 28.5

1. If verror is LP and d/dtverror is any then Vmotor is LP.

1 1 1

0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24
ANY VALUE
30rps (so ignore) 17V
(could also
This has about 0.6 (out of 1) membership have chosen
some value
above 17V)

2. If verror is SP and d/dtverror is SP or ZE then Vmotor is SP.

the AND means take the
the OR means take the lowest of the two
highest of the two memberships
memberships

1 1 1 1
rps rps/s V
0 0 0 rps/s 0
-100-50 0 50 100 -6 -3 0 3 6 -6 -3 0 3 6 0 6 12 18 24
1rps/s
30rps 1rps/s 14V

This has about 0.4 (out of 1) membership

3. If verror is ZE and d/dtverror is SP then Vmotor is ZE.

1 1 1
rps/s
0 rps 0 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24
the lowest results in 0 set
30rps 1rps/s membership
This has about 0.0 (out of 1) membership
 Fuzzy logic - 28.6

4. If verror is ZE and d/dtverror is SN then Vmotor is SN.

1 1 1

0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24
1rps/s the lowest results in 0 set
30rps
membership
This has about 0.0 (out of 1) membership

5. If verror is SN and d/dtverror is SN then Vmotor is SN.

1 1 1

0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24
1rps
30rps
This has about 0.0 (out of 1) membership

6. If verror is LN and d/dtverror is any then Vmotor is LN.

1 1 1

0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

ANY VALUE
30rps
This has about 0 (out of 1) membership

Figure 28.5 Rule Calculation

The results from the individual rules can be combined using the calculation in Fig-
ure 28.6. In this case only two of the rules matched, so only two terms are used, to give a
 Fuzzy logic - 28.7

final motor control voltage of 15.8V.

∑ ( Vmotor ) ( membershipi )
i

i---------------------------------------------------------------------
=1 -
V motor = n

∑ ( membershipi )
i=1

0.6 ( 17V ) + 0.4 ( 14V )

V motor = --------------------------------------------------- = 15.8V
0.6 + 0.4

Figure 28.6 Rule Results Calculation

28.2 COMMERCIAL CONTROLLERS

At the time of writing Allen Bradley did not offer any Fuzzy Logic systems for
their PLCs. But, other vendors such as Omron offer commercial controllers. Their control-
ler has 8 inputs and 2 outputs. It will accept up to 128 rules that operate on sets defined
with polygons with up to 7 points.

It is also possible to implement a fuzzy logic controller manually, possible in struc-

tured text.

28.3 REFERENCES

Li, Y.F., and Lau, C.C., “Application of Fuzzy Control for Servo Systems”, IEEE International
Conference on Robotics and Automation, Philadelphia, 1988, pp. 1511-1519.

28.4 SUMMARY

• Fuzzy rules can be developed verbally to describe a controller.

• Fuzzy sets can be developed statistically or by opinion.
• Solving fuzzy logic involves finding fuzzy set values and then calculating a value
for each rule. These values for each rule are combined with a weighted average.
 Fuzzy logic - 28.8

28.5 PRACTICE PROBLEMS

28.6 PRACTICE PROBLEM SOLUTIONS

28.7 ASSIGNMENT PROBLEMS

1. Find products that include fuzzy logic controllers in their designs.

2. Suggest 5 control problems that might be suitable for fuzzy logic control.

3. Two fuzzy rules, and the sets they use are given below. If verror = 30rps, and d/dtverror = 3rps/s,
find Vmotor.

1. If (verror is ZE) and (d/dtverror is ZE) then (Vmotor is ZE).

2. If (verror is SP) or (d/dtverror is SP) then (Vmotor is SP).
d/
verror dtverror Vmotor

1 1 1

SN 0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

1 1 1
ZE 0 rps 0 rps/s 0 V
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

1 1 1
SP rps rps/s V
0 0 0
-100 -50 0 50 100 -6 -3 0 3 6 0 6 12 18 24

4. Develop a set of fuzzy control rules adjusting the water temperature in a sink.

5. Develop a fuzzy logic control algorithm and implement it in structured text. The fuzzy rule set
below is to be used to control the speed of a motor. When the error (difference between desired
and actual speeds) is large the system will respond faster. When the difference is smaller the
 Fuzzy logic - 28.9

response will be smaller. Calculate the outputs for the system given errors of 5, 20 and 40.

100%

Big Error

error
0%
10 30

100%

Small Error

error
0%
10 30
20

Big Output

error
50

Small Output
5
error
50
if (big error) then (big output)
if (small error) then (small output)
29. NEURAL NETWORKS

Topics:
• Basic laws of motion
• Gravity, inertia, springs, dampers, cables and pulleys, drag, friction, FBDs
• System analysis techniques
• Design case

Objectives:
• To be able to develop differential equations that describe translating systems.

I. Introduction

Over the past decade, the artificial intelligence community has undergone a resurgence of interest in the
research and development of artificial neural networks. An artificial neural network is an attempt to simulate the man-
ner in which the brain interprets information as determined by the current knowledge of biology, physiology, and psy-
chology [14]. Artificial neural networks behave in much the same manner as biological neural networks, giving many
of the same benefits. Artificial neural nets are fault tolerant, exhibit the ability to learn and adapt to new situations,
and have the ability to generalize based on a limited set of data. This arises because of the structure which allows
neural nets to process information simultaneously, as opposed to the serial nature of traditional digital computers. This
parallel nature, inherent in neural networks, achieves the increase in speed by distributing the calculations among many
neurons. The network structure provides a level of fault tolerance, which allows the network to withstand component
failures without having the entire network fail.

This paper focuses on a popular feedforward model of neural networks. In this model a set of inputs are
applied to the network, and multiplied by a set of connection weights. All of the weighted inputs to the neuron are
then summed and an activation function is applied to the summed value. This activation level becomes the neuron’s
output and can be either an input for other neurons, or an output for the network. Learning in this network is done by
adjusting the connection weights based upon training vectors (input and corresponding desired output). When a train-
ing vector is presented to a neural net, the connection weights are adjusted to minimize the difference between the de-
sired and actual output. After a network is trained with a set of training vectors, the network should produce a good
output match for the inputs.

Artificial neural networks are mainly used in two areas --- pattern recognition, and pattern matching.
Pattern recognition is performed by classifying an unknown pattern through comparisons with previously learned pat-
 translation - 29.2

terns. This ability is termed associative recall. An example of this is the use of neural networks to recognize hand-
written digits [4]. In pattern recognition, when a particular pattern is noisy or distorted, the network can generalize
and choose the closest match [3][4][10][13]. Pattern matching uses continuous input patterns to evoke continuous
output patterns (i.e. one-to-one mappings). An example is the use of a neural network as a basic controller for a plant.
The controller would accept the plant conditions as the inputs, and a set of control outputs would drive the current man-
ufacturing process. This approach is discussed in more detail in [11]. This paper uses the pattern matching strategy
for the inverse kinematics of a typical three link manipulator.

This paper focuses on the kinematic control of a three link manipulator working in three dimensional
space, within a quarter of the robot workspace. When manipulators interact with the environment, the position of the
end effector must be described. Human users are more suited to working in cartesian coordinates than the joint coor-
dinates of robots; hence, there is a need for conversion between the two coordinate systems. The conversion from
world coordinates to robot joint angles is called inverse kinematics. In the past, the problem of solving the inverse
kinematics has been done through the application of various methods. These methods take one of two basic forms (i)
Closed form equations or, (ii) Iterative. A closed form solution is found by decoupling the joints responsible for po-
sition and those responsible for orientation. These solution types are more desirable since explicit equations are avail-
able. Iterative techniques, such as, Newton-Raphson, Newton-Gauss and Linearization are available when no closed
form solutions can be obtained. Many of these solutions require that the kinematics be performed off-line because of
the computational demands of the algorithms. A brief survey of these various algorithms is given in [2] and [15]. Both
of the forms are generally rigid and do not account for uncontrollable variables such as manufacture tolerances, cali-
bration error, and wear. A neural network approach could adapt to changes in the robot due to wear from extended
use over time. Such algebraic and algorithmic procedures cannot incorporate these situations.

Previous research in the application of neural networks to kinematic control was performed for a to a two-
link planar manipulator (working in a plane), and a five degree of freedom robot working in three dimensional space.
The manipulators were controlled with a high accuracy; however, the training region for the two link robot was a small
square in the centre of the workspace, and the five link robot was limited to a small wedge volume within the robot
workspace [5][6][7]. This paper explains research extended to investigate the performance throughout the entire work
space of the robot. Investigations of a more complete representation of inverse kinematics throughout the entire work-
space is provided.

The paper begins with a general introduction to the artificial neuron, a brief description of feedforward
neural networks, and a description of the backpropagation learning algorithm. The inverse kinematics problem is de-
scribed, and the application of a two-layer neural network is discussed. The use of compensation networks is explored
and the results of the experimental work are presented. The implications of the neural network approach to the kine-
matic control problem are presented at the conclusion of this paper.

II. The Basic Artificial Neuron

The basic building block of an artificial neural network is the neuron. In a popular model which will be
 translation - 29.3

used in this paper, the connection weights between neurons are adjusted. The neuron receives inputs opi from neuron

ui while the network is exposed to input pattern p. Each input is multiplied by a connection weight wij, where wij is

the connection between neurons ui and uj. The connection weights correspond to the strength of the influence of each
of the preceding neurons. After the inputs have been multiplied by the connection weights for input pattern p, their
values are summed, netpj. Included in the summation is a bias value θj to offset the basic level of the input to the

activation function, f(netpj), which gives the output opj . Figure 1 shows the structure of the basic neuron.

wij Neuron uj

opi activation function

opj
1
Σ netpj netpj
opj = f(netpj)

netpj = Σ wijopi + θj
where, wij ≡ synaptic weight, or connection weight between neuron ui and neuron uj
opi ≡ output of neuron ui (input of neuron uj)
θj ≡ bias value of neuron uj
opj ≡ output of neuron uj (function of the summation f(netpj))

Figure 1 Basic structure of an artificial neuron.

In order to establish a bias value θj, the bias term can appear as an input from a separate neuron with a
fixed value (a value of +1 is common). Each neuron requiring a bias value will be connected to the same bias neuron.
The bias values are then self-adjusted as the other neurons learn, without the need for extra considerations.

In calculating the output of the neuron, the activation function may be in the form of a threshold function,
in which the output of the neuron is +1 if a threshold level is reached and 0 otherwise. Squashing functions limit the
linear output between a maximum and minimum value. These linear functions, however, do not take advantage of
multi-layer networks [14]. Hyperbolic tangents and the sigmoidal functions are similar to real neural responses; how-
ever, the hyperbolic tangent is unbounded and hard to implement in hardware. In this paper, the Sigmoidal function
is used because of its ability to produce continuous non-linear functions, which can be implemented in hardware in
future research areas. Figure 2 shows some commonly used activation functions.
 translation - 29.4

linear sigmoidal logistic hyperbolic tangent Squashing Function linear threshold

1 1 1 1

-1
1
f(netpj) = - net pj
1-e

Figure 2 Common activation functions.

For more in depth information on neural networks, please refer to introductions to the subject in [3], [8], [14] and [16].

III. Neural Network Architectures in Feedforward Systems

A single neuron can only simulate 14 of the 16 basic boolean logic functions. It cannot emulate the X-OR,
or the X-NOR gates [9]. These limitations require that more than a single neuron be used, and thus, the architecture
becomes an important consideration. In Feedforward systems, networks propagate the inputs forward through the neu-
ral net. Each neuron has inputs that only come from neurons in preceding layers. A one-layer neural net consists of a
layer of input neurons and a layer of output neurons. A multi-layer network consists of an input layer of neurons, one
or more hidden layers, and an output layer.

The circular nodes in figure 3 represent basic neurons, as described in the previous section. The input neu-
rons are shown as squares because they only act as terminal points (i.e. opi = input). The input layer does not process

information; thus, it is not considered to be a part of the structure and is numbered layer 0. A simple one-layer network
can effectively map many sets of inputs to outputs. In pattern recognition problems, this depends upon the linear sep-
arability of the problem domain. In pattern matching problems, this depends upon continuity, and topography of the
function. If such a set of connection weights cannot be found using one-layered networks, then multi-layered networks
must be considered.
 translation - 29.5

input output input hidden output

layer layer layer layer layer

(a) one-layer network (b) two-layer network

Figure 3 Simple feedforward artificial neural networks.

IV. Backpropagation of Errors

Learning methods may be either supervised, or unsupervised. Supervised learning is required for pattern
matching, and backpropagation is the most popular of the supervised learning techniques. In the backpropagation
Training Paradigm, the connection weights are adjusted to reduce the output error. In the initial state, the network has
a random set of connection weights. When a system starts with all connection weights equal, the network begins at a
sort of local optimum, and will not converge to the global solution. In order for the network to learn, a set of inputs
are presented to the system and a set of outputs are calculated. A difference between the actual outputs and desired
outputs is calculated and the connection weights are modified to reduce this difference [14].

After inputs have been applied and the network output solution has been calculated, the estimated error
contribution, δ, by each neuron must be calculated. The calculations begin at the output layer of the network. The
delta value for any output neuron is computed as,

δpj = (tpj - opj) f ´(netpj) (1)

where, tpj is the desired output of output neuron uj, and opj is the actual output, and f ´(netpj) is the first derivative of

the activation function with respect to the total input (netpj) for the given input pattern p evaluated at neuron uj.

For the output layer, the change in connection weights can easily be calculated since the δ of the output
neurons is easily determined. With the introduction of hidden layers, the desired outputs of these hidden neurons be-
comes more difficult to estimate. In order to estimate the delta of a hidden neuron, the error signal from the output
 translation - 29.6

layer must be propagated backwards to all preceding layers. The delta, δpj , for the hidden neurons is calculated as,

δpj = f ´(netpj) Σδpkwjk (2)

k

where, f ´(netpj) is the first derivative of the activation function with respect to netpj at hidden neuron uj, δpk is the

delta value for the subsequent neuron uk, and wjk is the connection weight for the link between hidden neuron uj and
subsequent neuron uk. This process of calculating the δs is performed for all layers until the input layer is reached

After all the error terms have been calculated for all the neurons, the weights may be adjusted. The esti-
mated weight change ∆pwij(n+1) is calculated for each input connection to neuron uj from neuron ui by,

∆pwij(n+1) = η δpj opi + α ∆wij(n) (3)

new delta momentum
term

where η is the learning rate, α is the smoothing term, δpj is the delta of neuron uj, opi is the output of preceding neuron

ui, and ∆wij(n) is the weight change in the previous training interval. The estimated errors are used to estimate the

weight changes with the least mean squares method.

The backpropagation method is essentially a gradient descent approach which minimizes the errors. In
order to adjust the connection weights, the gradient descent approach requires that steps be taken in weight space. If
the steps are too large, then the weights will overshoot their optimum values. If the steps are too small, it may take
longer to converge and may become caught in local minima. The step size is then a function of the learning rate. The
learning rate, which varies between 0 and 1, should be large enough to come up with a solution in a reasonable amount
of time without having the solution oscillate within weight space. By introducing the momentum term, which varies
between 0 and 1, the learning rate can be increased while preventing oscillation. The momentum term is introduced
by adding to the weight change some percentage of the last weight change. Common values for η range from 0.1 to
0.9, and common values of α range from 0.2 to 0.8

The training set is comprised of points which have inputs and desired outputs, which may or may not be
in a random order. In this paper, the training points were chosen at random from an ordered table. This allows a weight
correction after each point is trained. This method helps avoid local optimum points in weight space, by creating a
random walk type of weight optimization, and gives a much faster global convergence. This process is then repeatedly
performed for all training examples until a satisfactory output error is achieved. For an in-depth derivation of the back
propagation method, the reader is suggested to consult Rumelhart and McClelland [14].

V. A Typical Three Link Manipulator

In order for a robotic manipulator to perform tasks within space, a user must specify a location in three
dimensional space. The robotic controller must then determine the correct joint coordinates to locate the manipulator.
This area of robotics, Inverse Kinematics, has been well researched and many good solutions exist. In all cases the

* note: The subscript notation (i, j, k) means that neuron ui precedes neuron uj, which precedes uk, etc.
 translation - 29.7

resultant solutions are highly specific to a particular robot configuration, with exact dimensions. Such explicit solu-
tions do not tolerate changes over time. These changes may be caused by poor tolerances in manufacture, wear over
extended periods of operation, damage to the robot, and poor calibration. Computations are often complex, can be
quite slow, and require a computer capable of performing complex mathematical functions.

An example of a robot which may be used with an artificial neural network control system is the three link
manipulator as shown in figure 4. The forward and inverse kinematic equations, used by the authors in testing, are
given in the equations (4) and (5).

Z0
θ1
a2

Table 1 Link parameter table. θ2

a3
link variable α a d
Z1
θ3
1 θ1 90o 0 0
2 θ2 0o a2 0
3 θ3 0o a3 0 Z2
(px, py, pz)

X
Denavit-Hartenburg
Transformations

c1c23 -c1s23 s1 c1(a3c23 + a2c2)

s1c23 -s1c23 -c1 s1(a3c23 + a2c2)
T3 = A1A2A3 = s23 c23 0 a3s23 + a2s2

0 0 0 1

Figure 4 A typical three link manipulator.

The ± sign in θ2 indicates the general configuration of the robot arm. The positive sign (+ve) corresponds to an elbow
down solution, and the negative sign (-ve) corresponds to an elbow up solution.
 translation - 29.8

px = c1(a3c23 + a2c2)
py = s1(a3c23 + a2c2) forward kinematic equations
pz = a3s23 + a2s2 (4)

py
θ1 = tan-1
px

β α
θ2 = tan-1 - tan-1 inverse kinematic equations
± √
pz2 + α2 - β2 pz
(5)

pz - a2s2
θ3 = tan-1 - θ2
α - a 2 c2

s1 = sin θ1
c1 = cos θ1
α = + √px2 + py 2
where and
c12 = cos (θ1 + θ2) α2 + pz2 + a22 - a32
β =
2a2
s12 = sin (θ1 + θ2)

It should be noted that the inverse kinematics is not determinate. The sources of indeterminacy for this
manipulator occur:

• for Elbow up and Elbow Down Configurations, and

• when the end of the arm is directly over the origin, θ1 is undefined.

Other examples of indeterminacy occur in PUMA style robots where left and right arm configurations would give in-
determinacy. These sources of indeterminacy may all be overcome by constraining the inverse kinematics to a partic-
ular case.

VI. Training a Neural Network for Inverse Kinematics

Neural Network solutions have the benefit of having faster processing times since information is processed
in parallel. The solutions may be adaptive and still be implemented in hardware by using specialized electronics. Neu-
ral systems can generalize to approximate solutions from small training sets. Neural systems are fault tolerant and
robust. The network will not fail if a few neurons are damaged, and the solutions may still retain accuracy. When
implementing a neural network approach, complex computers are not essential and robot controllers need not be spe-
 translation - 29.9

cific to any one manipulator.

The nature of neural networks require that a set of training points be chosen which represent the nature of
the inputs (x, y, z) and the corresponding outputs (θ , θ and θ ). If the training points chosen tend to be clustered,
1 2 3
then the network will be very accurate when dealing with points near the cluster. In this case, the robot should be fa-
miliar with points throughout the entire workspace; thus, points should be evenly distributed. The order in which
points are presented to the network also affect the speed and quality of convergence. If the points are not presented in
a random order, each training update will tend to train the network for the current section of space which the points are
from.

The neural network architecture chosen for this problem is a simple network with one hidden layer. The
network has three input neurons for the desired position, and three output neurons for the estimated joint angles. A
bias neuron (with a value of +1) is attached to all of the neurons in the hidden layer, and to all of the output neurons.
The number of neurons in the hidden layer varied from 10, 20, and 40. This network architecture is pictured in figure 5.

input hidden output

layer layer layer

x θ1

y θ2

z θ3

+1
bias
neuron

Figure 5 Neural architecture for solving the inverse kinematic problem.

For simplicity, the network is shown with a small amount of arrows;
however, the network is actually fully connected.

29.1 SUMMARY

•
 translation - 29.10

29.2 PRACTICE PROBLEMS

29.3 PRACTICE PROBLEM SOLUTIONS

29.4 ASSIGNMENT PROBLEMS

29.5 REFERENCES
translation - 29.11
translation - 29.12
 analog IO - 30.1

30. EMBEDDED CONTROL SYSTEM

Topics:
•

Objectives:
•

30.1 INTRODUCTION

• Elements of embedded systems (with program examples of each)

NOTE: emphasize a top-down program structure with subroutines for each
one.
logical IO - digital inputs and outputs
analog outputs - immediate
analog inputs - delayed in (show use of nops and loop to wait), use potenti-
ometer
timers including PWM to control transisor/h-bridge for motor control,
show sound generation also.
counters including encoder decoding and tachometer decoding
serial IO
other peripherals such as displays, sounds, etc.
• Control system fundamentals
a simple PID feedback loop using nops and time calcs to ensure time
a multiple step process that waits for an input and does a task with the PID
loop. Show an executive subroutine and calls to
dealing with events
event types; asynchronous, delayed
polling
interrupts
• Concurrent processes
single thread vs concurrent processing
how to implement a single control thread
how to create multiple processes
real-time
the need for multiple processes
non-time critical
time critical; regular updates and minimum time between runs
 analog IO - 30.2

priority levels
hard vs. soft

• Structured Design
General systems design topics
show the general structure of a program with an executive routine
that calls task routines.
show a program that mixes an asynchronous GUI mixed with a
realtime routine
Modal System design
show the use of global mode bits to track the mode of the system
show how the mode bits changes the flow of execution in the exec-
utive routine.
Flowcharts
map flowchart structures to program structure
using a register value to track the location in the flowchart
State Diagrams
show the use of state diagrams to model a process and then how the
program is written for it.
Failure Analysis
show basic probability theory
show parallel vs serial failures
show failure estimation using theory for single and chained failure
modes
show the methods for categorizing failures as a hazard, danger, etc.

• Communication
Receiving and sending strings
String Handling
Parsing strings
Composing strings
Command and response structures
Error Checking
Non byte oriented data
Networked structured, ie destination address header
A full feedback control system
• Other topics
Keyboard multiplexing
output refreshing, LEDs
 analog IO - 30.3

30.2 CASE STUDY

30.3 SUMMARY

30.4 PRACTICE PROBLEMS

1.

30.5 PRACTICE PROBLEM SOLUTIONS

1.

30.6 ASSIGNMENT PROBLEMS

1.
 writing guide - 31.1

31. WRITING

Reports are the most common form of document written by engineering. Report
writing is an art that we often overlook, but in many cases can make a dramatic impact on
how our work is perceived. Your reports are most likely to find their way to a superiors
desk than you are to meet the individual.

31.1 FORGET WHAT YOU WERE TAUGHT BEFORE

In my experience, most engineering students have been prepared for persuasive

writing, but many of these skills become a negative when writing reports. The items listed
below are the most common problems that engineering encounter when writing.

Writing is the last thing you do - Students have been required to do rough drafts to
clarify their thoughts and determine what they want to say. To put it simply, if
you don’t know what you will say you are not ready to write. When you are
ready to write it should already be clear. An author should not start to write the
report until all
Don’t find ’creative ways to say things’ - many student have been taught that they
should not repeat themselves and find multiple ways to say things. When this is
done in technical documents it leads to confusion. Authors should use precise
terms (as many times as needed) and avoid trying to generate creative word
choices. For example, we could increase confusion by also describing transla-
tion as motion, movement, sliding, displacing, etc.
Keep it simple - In an attempt to increase the ’prestige’ of a document many
authors will use ’large’ words. This often leads to confusion and should be
avoided. In some cases when an author is unsure they will respond by making
their writing style more complex but most readers recognize this. For example,
“Electronic computer based digital readings can provided a highly accurate data
source to improve the quality of the ascertained data.” could be replaced with
“Computer based data collection is more accurate.”
Clear, concise and to the point - In the past many students have been required to
write reports with a minimum number of words. This encourages writers to
break ideas into smaller pieces. But, readers appreciate shorter documents that
get to the point. For example, “Readings of the pressure, as the probe was
ascending up the chimney towards the top, were taken.” is better put “Pressure
probe readings were taken as the probe was inserted”.
There is no great opening paragraph - Over half of all student authors spent a large
amount of time on the opening paragraph to set the tone for the report. In my
experience the longer a student tries to write the opening paragraph, the worse it
is. In most cases it can be removed entirely from the document without any neg-
ative impact. Ironically, the writing of these students often improves once they
 writing guide - 31.2

get beyond the first paragraph, but they have already lost the interest of the
reader.
Transitions are not that important - Students are often coached to create clean tran-
sitions between sentences and paragraphs. As a result they often add unneces-
sary sentences and words. Words that are often warning signs are also and then.
You don’t need to keep the good stuff to the end - Many authors try to write the
report so that there is a ’climax’. It can be very frustrating for a reader who
reads 90% of a report before they encounter some discussion of the results.
(note: a report is not like a mystery novel.)
Saying it more than once is acceptable- Over 3/4 of students feel that it is unac-
ceptable to state facts more than once. In truth you want to state fact as many
times as necessary to make a technical point. In the case of very important
details, this will be in the abstract, introduction discussion and conclusion.

A better procedure for writing engineering reports is detailed below. The method
leaves writing to one of the last stages, but it becomes much easier when done this way.

1. Plan and do the work as normal. Regardless of what the report requires this will
often include drawings, sketches, graphs/charts of collected data, pictures, etc.
2. Do the analysis (preferably on computer) of the data and results. These should
be organized into a logical sequence.
3. Review the results to ensure they make sense and follow a logical flow.If neces-
sary add figures to help clarify.
4. Review the materials to verify that they make sense without the text.
5. Put in point form notes to lay out the document.
6. Write the text for the report.
7. Verify that the report conforms to guidelines.
8. Proofread.

31.2 WHY WRITE REPORTS?

Reports are written for a number of reasons, as listed below. In all cases they are
designed to communicate technical information. Report readers are often looking for clear
details that lead to a clear set of conclusions. By contrast, they are not looking for content
that is highly philosophical, or riddled with artistic diversions.

- (lets forget about this one) as a student you must do them to get marks
- to let other engineers know the results of an experiment
- to leave a record of work done so that others may continue on
- as a record you may use yourself if you must do work again some time later
- they are required for legal reasons (contract or legislation)
- they bring closure to the project
 writing guide - 31.3

31.3 THE TECHNICAL DEPTH OF THE REPORT

Knowing the level of reader is very important when writing. Engineers often write
reports for other engineers, or technically aware readers. As a result an engineer will often
write at their own knowledge level. When necessary, appendices can be added to accom-
modate readers with variable backgrounds. The depth of detail is another common issue.
The author should provide enough detail to allow others to follow and reproduce the
results. The two questions below should form a good basis for evaluation.

“What happens if I am hit by a car; could somebody pick up my report and con-
tinue?” if the answer to this question is no the report requires more detail.

“Would I skip or skim this sentence/paragraph/section/report because it is so obvi-

ous?”. If the answer is yes section can be condensed, or moved to and appendix.

31.4 TYPES OF REPORTS

The list below shows many of the reports written by engineers. In general, reports
give necessary background, discuss details and then draw conclusions from them. The
credibility of a report is based upon the evidence it contains. These then support the con-
clusions drawn by the author.

Laboratory - Theses ‘Lab Reports’ describe one or more experiments, the results,
and the conclusions drawn from them.
Consulting - A summary of details, test results, observations, and a set of conclu-
sions. Typically they will also contain a recommendation.
Project - A description of work done in a project to inform other engineers who
may be asked to take up further work on the project.
Research - A summary of current advances in a topic. This should end with some
comparison of alternatives.
Interim - A report to apprise supervisors and others as to the progress of a project
or other major undertaking.
Executive - A brief summary of the report, and any implications for decision mak-
ing at the management levels.
Proposals - A technical document describing a project for the purposes of obtain-
ing permission to proceed.

31.5 LABORATORY REPORTS

• Purpose: These reports should outline your procedure and results in detail. They
 writing guide - 31.4

should also contain the analysis and conclusions. The completeness of detail allows you
(and others) to review these and verify the correctness of what has been done. These have
been historically used for hundreds of years and are accepted as a form of scientific and
legal evidence. It is completely unacceptable to make incorrect entries or leave out impor-
tant steps or data.

• Standard Format:

1. Title, Author, Date - these make it clear what the labs contain, who did the work,
and when it was done. If the report is done for a course it should also include
the course name/number, the lab instructor and the lab section.
2. Purpose - a brief one line statement that allows a quick overview of what the
experiment is about. This is best written in the form of a scientific goal using
the scientific methods.
3. Theory - a review of applicable theory and calculations necessary. Any design
work is done at this stage
4. Equipment - a list of the required equipment will help anybody trying to repli-
cate the procedure. Specific identifying numbers should be listed when possi-
ble. If there are problems in the data, or an instrument is found to be out of
calibration, we can track the problems to specific sets of data and equipment.
5. Procedure - these are sequential operations that describe what was done during
the experiment. The level of detail should be enough that somebody else could
replicate the procedure. We want to use this as a scientific protocol.
6. Results (Note: sometimes procedure and results are mixed) - the results are
recorded in tables, graphs, etc. as appropriate. It will also be very helpful to note
other events that occur (e.g. power loss, high humidity, etc.)
7. Discussion - At this stage the results are reviewed for trends and other observa-
tions. At this point we want to consider the scientific method.
8. Conclusions - To conclude we will summarize the significant results, and make
general statements either upholding or rejecting our purpose.

• Style: These are meant to be written AS the work is done. As a result the work
should be past tense

• Laboratory reports should have one or more hypotheses that are to be tested. If
testing designs these are the specifications. Examples might be,

- what is the thermal capacity of a material?

- what is the bandwidth of an amplifier?
- will the counter increment/decrement between 0 to 9?

• NOTE: These reports are much easier to write if you prepare all of the calcula-
tions, graphs, etc. before you start to write. If you sit down and decide to do things as you
 writing guide - 31.5

write it will take twice as long and get you half the marks...... believe me, I have written
many in the past and I mark them now.

31.5.0.1 - An Example First Draft of a Report

Grand Valley State University

Padnos School of Engineering

EGR 345 Dynamics Systems Modeling and Control

Laboratory Exercise 7

Title: The Cooling of Coffee

Author: I. M. Wyred

Date: Dec., 23, 1998

Purpose: To derive a theoretical model of the rate at which coffee cools and exper-
imentally verify the model and find coefficients.

Theory:

When coffee is heated kinetic energy is added, when coffee is cooled kinetic
energy is removed. In a typical use, coffee cools as heat is lost through convection and
conduction to the air and solids in contact. The factors involved in this convection/conduc-
tion can be difficult to measure directly, but we can approximate them with a simple ther-
mal resistance. Consider the temperature difference between the coffee and the ambient
temperature. The greater the temperature difference, the higher the rate of heat flow out of
the coffee. This relationship can be seen formally in the equation below. We can also
assume that the atmosphere is so large that the heat transfer will not change the tempera-
ture.
 writing guide - 31.6

1
q = --- ( θ coffee – θ air )
R
where,

q = heat flow rate from coffee to air (J/s)

R = thermal resistance between air and coffee
θ = temperatures in the coffee and air

We can also consider that coffee has a certain thermal capacity for the heat energy.
As the amount of energy rises, there will be a corresponding temperature increase. This is
known as the thermal capacitance, and this value is unique for every material. The basic
relationships are given below. I will assume that the energy flow rate into the coffee is
negligible.

1 –1 dθ coffee –1 -
∆θ coffee = ---------------- ( q in – qout ) = ---------------- q -----------------
- = --------------- q
C coffee C coffee dt C coffee
where,
C coffee = thermal capacitance
C coffee = M coffee σ coffee
where,
M coffee = mass of thermal body
σ coffee = specific heat of material in mass

The temperatures can be found by consider that the energy flowing out of the cup,
and into the atmosphere is governed by the resistance. And, the temperature in the coffee
and air are governed by the two capacitances. We will make two assumptions, that the
thermal capacitance of the atmosphere is infinite, and that there is no energy flowing into
the coffee.
 writing guide - 31.7

dθ coffee –1
-----------------
- = ---------------
-q
dt C coffee

dθ coffee
= ---------------------------------  --- ( θ coffee – θ air )
–1 1
∴------------------
dt M coffee σ coffee  R 

dθ coffee
∴------------------ +  ------------------------------------- θ coffee =  ------------------------------------- θair
1 1
dt M coffee σ coffee R M coffee σ coffee R

This differential equation can then be solved to find the temperature as a function
of time.

Guess Ct d- Ct
θ = A + Be ---- θ = BCe
dt

+  ------------------------------------- ( A + Be ) =  ------------------------------------- θ air

Ct 1 Ct 1
∴BCe
 M coffee σ coffee R  M coffee σ coffee R

∴e  BC + ------------------------------------- +  ------------------------------------- +  ------------------------------------- θ air = 0

Ct B A –1
M coffee σ coffee R M coffee σ coffee R M coffee σ coffee R

B –1
BC + ------------------------------------- = 0 C = -------------------------------------
M coffee σ coffee R M coffee σ coffee R

------------------------------------- +  ------------------------------------- θ air = 0

A –1
A = θ air
M coffee σ coffee R  M coffee σ coffee R

To find B, the initial temperature of the coffee should be used,

C(0)
θ 0 = A + Be = θ air + B B = θ 0 – θair
The final equation is,
–t
----------------------------------
-
M coffee σ coffee R
θ = θ air + ( θ 0 – θ air )e

The time constant of this problem can be taken from the differential equation
 writing guide - 31.8

above.

τ = M coffee σ coffee R

Equipment:

1 ceramic coffee cup (14 oz.)

2 oz. ground coffee
1 coffee maker - Proctor Silex Model 1234A
1 thermocouple (gvsu #632357)
1 temperature meter (gvsu #234364)
1 thermometer
2 quarts of tap water
1 standard #2 coffee filter
1 clock with second hand
1 small scale (gvsu# 63424)

Procedure and Results:

1. The coffee pot was filled with water and this was put into the coffee maker. The
coffee filter and grounds were put into the machine, and the machine was turned on. After
five minutes approximately the coffee was done, and the pot was full.

2. The mass of the empty coffee cup was measured on the scale and found to be
214g.

3. The air temperature in the room was measured with the thermometer and found
to be 24C. The temperature of the coffee in the pot was measured using the thermocouple
and temperature meter and found to be 70C.

4. Coffee was poured into the cup and, after allowing 1 minute for the temperature
to equalize, the temperature was measured again. The temperature was 65C. Readings of
the coffee temperature were taken every 10 minutes for the next 60 minutes. These values
were recorded in Table 1 below. During this period the cup was left on a table top and
allowed to cool in the ambient air temperature. During this period the mass of the full cof-
fee cup was measured and found to be 478g.
 writing guide - 31.9

Table 1: Coffee temperatures at 10 minute intervals

time (min) temperature (deg C)

0 65
10 53
20 43
30 35
40 30
50 28
60 26

Discussion:

The difference between the temperature of the coffee in the pot and in the cup was
5C. This indicates that some of the heat energy in the coffee was lost to heating the cup.
This change is significant, but I will assume that the heating of the cup was complete
within the first minute, and this will have no effect on the data collected afterwards.

The readings for temperature over time are graphed in Figure 1 below. These show
the first-order response as expected, and from these we can graphically estimate the time
constant at approximately 32 minutes.

temp
(deg C)

60

40

24
20 t (min)

0 20 40 60
τ ≈ 32min

Figure 1 - A graph of coffee temperature measured at 10 minute intervals

 writing guide - 31.10

We can compare the theoretical and experimental models by using plotting both on
the same graph. The graph clearly shows that there is good agreement between the two
curves, except for the point at 30 minutes, where there is a difference of 3.5 degrees C.

temp
(deg C)
experimental data
60 mathematical model

max. difference
40 of 3.5 deg. C

20 t (min)

0 20 40 60

Figure 2 - Comparison of experimental and modelled curves

This gives an overall error of 8.5% between these two curves, compared to the
total range of the data.

3.5
error = ------------------ 100 = 8.5%
65 – 24

Finally, the results can be used to calculate a thermal resistance. If we know the
mass of the coffee and assume that the coffee has the same specific heat as water, and have
the time constant, the thermal resistance is found to be 1731sC/J.

τ = M coffee σ coffee R M coffee = 478g – 214g = 0.264Kg

C
σ coffee = 4.2 ----------
τ sC KgJ
R = --------------------------------- = 1731 ------
M coffee σ coffee J τ = 32min = 1920s
 writing guide - 31.11

Conclusion:

In general the models agreed well, except for a single data point. This error was
relatively small, only being 8.5% of the entire data range. This error was most likely
caused by a single measurement error. The error value is greater than the theoretical value,
which suggests that the temperature might have been read at a "hot spot". In the procedure
the temperature measuring location was not fixed, which probably resulted in a variation
in measurement location.

31.5.0.2 - An Example Final Draft of a Report

• A final draft of the report is available on the course website in Mathcad format,
and it will be distributed in the lab.

31.6 RESEARCH

• Purpose: After looking at a technical field we use these reports to condense the
important details and differences. After reading a research report another reader should be
able to discuss advanced topics in general terms.

• Strategy A:

1. Clearly define the objectives for the report

2. Outline what you know on a word processor in point form and find the ‘holes’
3. Do research to find the missing information
4. Incorporate the new and old information (still in point form)
5. Rearrange the points into a logical structure
6. Convert point form into full text
7. Proof read and edit

31.7 DRAFT REPORTS

- draft reports will be edited for contect. As a result they must be double spaced to
make room for editing comments.
 writing guide - 31.12

31.8 PROJECT REPORT

XXXXX UPDATE USING PROJECT DESCRIPTION

- do exhibits first, then write

Basic elements are,

Title page
- title of project
- Department/Company (i.e., unviersity) name
- course/name
- author list
- date of submission
Executive Summary/Overview
- the purpose is to let people know if they want ro read the report
- write it last after all the details are done
- briefly describe the purpose of the work
- what were the results, be detailed
- given numbers
- keep it short
Nomenclature
- list all variables used in the report
- put in alphaetical order, with upper case first
- list greek and other variables separately
Table of contents
- list all major sections, and minor sections to a reasonable level
- list page numbers
- right justify the numbers
Introduction
- if you must include this section, it should be brief
- the introduction is expected to outline details to be discussed in
the report.
Body............

Title page
 writing guide - 31.13

• Purpose: These reports allow the developer or team to document all of the design
decisions made during the course of the project. This report should also mention avenues
not taken. Quite often the projects that we start will be handed off to others after a period
of time. In many cases they will not have the opportunity to talk to us, or we may not have
the time. These reports serve as a well known, central document that gathers all relevant
information.

• Strategy A:

1. Define the goals for the project clearly in point form

2. Examine available options and also add these in point form
3. Start to examine engineering aspects of the options
4. Make engineering decisions, and add point form to the document
5. As work continues on the project add notes and figures
6. When the project is complete, convert the point form to full text.
7. Proof read and edit

31.9 OTHER REPORT TYPES

31.9.1 Executive

Purpose: These reports condense long topics into a very brief document, typically
less than one page in length. Basically these save a manager from having to
read a complete report to find the details that interest him/her.

31.9.2 Consulting

Purpose: These reports are typically commissioned by an independent third party

to review a difficult problem. The consultant will review the details of the prob-
lem, do tests as required, and summarize the results. The report typically ends
with conclusions, suggestions or recommendations.

31.9.3 Memo(randum)

Purpose: An internal business letter designed to convey a business policy or infor-

 writing guide - 31.14

mation.
- begin with Date, From, To, Subject:, cc:

31.9.4 Interim

Purpose: This report is normally a formal report to track the progress of a project.
When a project is initially planned, it will be given a timeline to follow. The
interim report will indicate progress relative to the initial timeline, as well as
major achievements and problems.

31.9.5 Poster

Purpose: A presentation format that conveys the key concept visually so that a
spectator would have a concept of the project in a glance, and be able to review
the key concepts in under one minute.

- use a minmimum of 16pt, or higher font.

- abundant usage of figures
- put on foan core backing board, or worst, a purchased cardboard backing board.
- The format uses bullets (still full sentences) to abreviate the presentation.
- the poster should be self-explanatory
- be neat
- use color in a tasteful way.

Typical sections include;

purpose/motivation - a brief summary of the purpose of the work. A sum-
mary of the major work detials and outcomes.
approach - the work that was done
conclusions - the results of the work that was assigned
acknowledgements - indicate others who played a significant role in the
results.

31.9.6 Progress Report

Purpose: To provide details of project progress to a supervisor or customer.

 writing guide - 31.15

31.9.7 Oral

Purpose: Work may be presented to an audience in a formal presentation.

- This can be a very effective method of comminucation when done well, but it can
also be very cruel because an audience may be forced to sit through an uncom-
fortable experience.

- use a minmimum of 16pt, or higher font.

- abundant usage of figures
- The format uses bullets (still full sentences) to abreviate the presentation.
- the presentation should be self-explanatory
- use color, visuals and sounds in a tasteful way.
- present concepts, by introducing, presenting then reviewing.

31.9.8 Patent

Purpose: A document to describe a design for legal purposes.

- a patent documents that an individual/company owns the idea.

- it gives them the right to sue those who use the idea and ’infringe’
- the patent is submitted to the patent office and reviewed. If the reviewers are sat-
isfied, then the patent is granted, and a patent number is assigned.
- a patent only gives the holder the right to sue.
- Normally a patent lists the inventor (normally an engineer), but is assigned to the
employer. If the inventor is an employee, they assign the rights to the employer
in exchange for $1.
- Patents are publically available after they are granted to encourage innovation.
- Patents have a life of 20 years from the date of application.
- A patent is only available for ideas that can be patented.
- the patent is structured into the follwing sections;
Abstract - a brief description of the patent
References - A list of other patents, or other publications that are related.
These are expected as most patents are built upon previous concepts.
Claims - An itemized list of features that make the design unique from pre-
vious ideas. Normally the are at least a dozen claims. These are the items
that are the legal basis for the patent.
Description - A techibcally detailed description of the patentable
work.Tihis normally includes sections such as Summary of the Inven-
tion, Description of the Drawings, Detailed Description of the Invention.
It is common for the description to include drawings and equations.
 writing guide - 31.16

31.10 LAB BOOKS

(This is adapted from a document developed by Dr. P. Anyalebechi)

Each student is required to maintain a well-kept and up-to-date laboratory note-

book. This notebook should be used to record all of the details of the various experiments
and tests that are performed no matter how trivial they may seem at the time. The entries
should include the following:

(i) Date the test was performed

(ii) Objective of the test or experiment
(iii) A brief description of the experimental or test apparatus or machine and the
procedure used
(iv) Results obtained - including visual observations and computer data file names
(if the results were obtained with a data acquisition system)
(v) Preliminary thoughts on the significance of the results

At the end of each lab period, the notebook should be witnessed and signed by
either the instructor or a laboratory assistant. The notebook should be submitted to your
instructor for grading within 24 hour of the end of the lab.

For team studies, every member of the team is expected to keep an independent
record (in his/her own laboratory notebook) of the experimental procedure, results, and
preliminary interpretation of the results. This will facilitate team discussions.

Well-kept laboratory notebooks are very important. They facilitate the statistical
analysis and interpretation of the data and the eventual writing of a final technical report.
They provide a legal record of events and often make the difference between winning or
losing legal cases on patents and on a company's right to use a certain technology in their
manufacturing facilities. Losing one of these cases could bankrupt a company or cost it
millions of dollars in royalty!

31.11 REPORT ELEMENTS

• In reports we must back up our opinions with data, equations, drawings, etc. As a
result we use a number of common items,
 writing guide - 31.17

- figures
- tables
- equations
-

• When these elements are included, there MUST be a mention of them in the writ-
ten text.

• These days it is common to cut and paste figures in software. Make sure

- the resolution is appropriate

- the colors print properly in the final form or print well as black and white
- the smallest features are visible
- scanned drawings are clean and cropped to size
- scanned photographs are clear and cropped to size
- digital photographs should be properly lit, and cropped to size
- screen captures are clipped to include only relevant data

31.11.1 Figures

• Figures include drawings, schematics, graphs, charts, etc.

• They should be labelled underneath sequentially and given a brief title to distin-
guish it from other graphs. For example “Figure 1 - Voltage and currents for 50 ohm resis-
tor”

• In the body of the report the reference may be shortened to ‘Fig. 1’

• The figures do not need to immediately follow the reference, but they should be
kept in sequence. We will often move figures to make the type setting work out better.

• If drawing graphs by computer,

- if fitting a line/curve to the points indicate the method used (e.g. linear regres-
sion)
- try not to use more than 5 curves on the same graph
- use legends that can be seen in black and white
- clearly label units and scales
- label axes with descriptive term. For example “Hardness (RHC)” instead of
“RHC”
 writing guide - 31.18

- scale the curve to make good use of the graph

- avoid overly busy graphs
- titles should clearly, and disctinct, indicate why the content of the figure is signif-
icant.

Figure 2 - Various Techniques for Making a Sphere with AMP

31.11.2 Graphs

• Units on axes

• Should have points drawn and connected with straight (or no) lines if experimen-
tal.

• Smooth lines are drawn for functions or fitted curves. If a curve has been fitted
the fitting method should be described.

• if using graphing software don’t put a title on the graph.

- titles should clearly, and disctinct, indicate why the content of the figure is signif-
 writing guide - 31.19

icant.

31.11.3 Tables

• Tables are often treated as figures.

• They allow dense information presentation, typically numerical in nature.

Table 1: A Comparison of Toy Vehicle Properties

Description Number Color Shape Material

car 3 red rectangular die cast

truck 6 blue long polyprop.
motorcycle 2 green small wood

• Legends can be added to tables to help condense size.

• table columns should have units as appropriate

- titles should clearly, and disctinct, indicate why the content of the figure is signif-
icant.

31.11.4 Equations

• When presenting equations, use a good equation editor, and watch to make sure
subscripts, etc are visible.

• Number equations that are referred to in the text.

• Box in equations of great significance.

• left justify equations

 writing guide - 31.20

• express results in engineering notation

• use care with subscripts

+
∑ Fx = – T 1 sin 60° + F R sin θ R = 0 (1)

+ ∑ Fy = – T 1 – T 1 cos ( 60° ) + F R cos θ R = 0 (2)

T 1 sin 60° T 1 + T 1 cos 60° sub (1) into (2)

∴F R = ---------------------- = ------------------------------------
sin θ R cos θ R

sin 60° sin θ R

∴--------------------------- = --------------- = tan θ R
1 + cos 60° cos θ R

0.866
∴ tan θ R = ----------------
1 + 0.5
∴θ R = 30°

98 sin 60° = F R sin 30°

∴F R = 170N

31.11.5 Experimental Data

• When analyzing the results from an experiment there are a few basic methods
that may be used,

Absolute difference - For example if the theoretical value is 105.8 and the mea-
sured value is 104.0 then you would say that "... the measured value was 1.8
below the theoretical value."
Percent difference - Avoid using this, but if must remember to assign a sign to the
percentage value. For example if the theoretical value is 100 and the measured
value is 50 then you would say that "... the measured value was 50% below the
theoretical value." OR "... the theoretical value was 200% of the experimental
value."
Mean and standard deviation - Useful for multiple data readings.
Point by point - look at each point, and compare it to another.
 writing guide - 31.21

• After data is presented, it should be discussed. In general a few good strategies

are,

- Discuss the fit between theory and measured. Explain differences in terms of
hard facts and with numbers.
- Don’t make vague references to human or experimental error.
- Try to develop a conclusion about the quality of the experimental data.
- Make a recommendation about changes that would improve that data quality.

31.11.6 Result Summary

A results summary allows both the author and the reader a chance to review the
resuls before a final conclusion. These should include the important results presented else-
where in the report. It is best to use a graphical format such as a table or list.

31.11.7 References

• References help provide direction to the sources of information when the infor-
mation may be questioned, or the reader may want to get additional detail.

• Reference formats vary between publication sources. But, the best rule is be con-
sistent.

• One popular method for references is to number them. The numbers are used in
the body of the paper (eg, [14]), and the references are listed numerically at the end.

• Another method is to list the author name and year (eg, [Yackish, 1997]) and then
list the references at the end of the report.

• Footnotes are not commonly used in engineering works.

• References on the internet should include a working web address, a description of

the page (normally the page title), a date viewed and the page author if available.

31.11.8 Acknowledgments

• When others have contributed to the work but are not listed as authors we may
 writing guide - 31.22

choose to recognize them.

• Acknowledgments are brief statements that indicate who has contributed to a

work.

31.11.9 Abstracts

An abstract is a brief summary of the results of the project. It us used by potential

readers to determine if they are interested in reading a report. It should repeat a number of
key details. An abstract should not omit key findings, as if it were an advertisement for a
movie.

31.11.10 Appendices

Reports often include appendices to reduce bulk from the body. When putting
material in an appendix it should stand alone and be referred to in the body of the report. It
is customary to provide a summary of the results in the appendix, unless it is tutorial in
nature. Examples of common appendices are given below.

sample calculations - these are redundant numerical calculations, or a prolonged

derivation of equations. The body of the report has a summary of key assump-
tions,sample calculations and results. The calculations are often provided so
that the reader may verify the calculations.
long tables of data - tables of numerical data are often put in appendices. Typically
a sample of the table is included in the body for discussion purposes. These are
often provided for the reader who wants to use the data beyond the uses in the
report.
program listings - long listings of computer programs are often put in appendices.
They are referenced in the body on the report near the algorithm/calculation/
method they implement. These listings are provided for readers who want to
use the program.
multiple data graphs - multiple sets of data graphs are often put in appendices and
summarized in a report body. The graphs are often provided so that the reader
may use the graphs for verification or further analysis.
reviews of basic theory - these are often referenced in the body of the report for
readers who may not have seen a topic previously. These are uncommon in stu-
dent reports.
 writing guide - 31.23

31.11.11 Page Numbering

- number all pages sequentially,

roman numerals starting from ‘i)’ on the first page
arabic numerals starting from ‘1’ on the
- or, number pages by section. This is very useful for multi part manuals
for example ‘4-7’ would be the 7th page in the 4th section
- if pages are blank label them ‘this page left blank’
- number sections sequentially with roman or Arabic numerals

31.11.12 Numbers and Units

- use engineering notation (move exponents 3 places) so that units are always
micro, milli, kilo, mega, giga, etc. Avoid number formats such as ’0.00000456’
- use significant figures to round the numbers. For example a length of 0.345432in
for a dimension measured with a ruler is ridiculous.
- units are required always
- greek symbols or the unit name spelled out
- take care to distinguish between frequency in hertz and radians/sec
- put a ’0’ before a point, 0.5 not .5
-

31.11.13 Engineering Drawings

The Padnos School of Engineering basic drawing requirements are summarized

below.

General Requirements
- a standard border
- a title block with a part name, employer/client/etc, designers name, date
of creation, units used, tolerances (unless noted on each dimension).
Orthographic views
- three views unless axial symmetry allows fewer.
- the front view should be the most descriptive
- aligned correctly
- any hole/arc over 180 degrees must have center lines and marks
- blind holes made by drilling must have a drill point shown
- all parts must be manufacturable
Dimensioning
- The location and size of each feature must be clearly defined
- critical assembly dimensions must be directly readable and not require
 writing guide - 31.24

addition.
- holes that form patterns must be dimensioned relative to each other and
relative to a major feature
- smaller dimensions should be closer to the part
- chained dimensions must be aligned
- hole sizes and dimensions should be on the profile view
- arcs/circles more than 180 degrees are sized by diameter, otherwise radius
is used.
- redundant dimensions should be eliminated

Tolerancing
- Tolerances must be reasonable for manufacturing
- Tolerances must ensure proper assembly and operation at maximum/min-
imum material conditions
- mating parts should not have identical dimensions, they should be free
running or press fits.
- Smaller tolerances should be used for mating parts
- a general part tolerance should be defined for the part, and smaller toler-
ances indicated for critical dimensions to reduce clutter

31.11.14 Discussions

- discussions are expected to state all of the details. Sometines these might be obvi-
ous when the data is observed, but they must be stated.

31.11.15 Conclusions

- This is the section to restate the key numbers that support the purpose of the lab.
- clear, concise recommendations are needed.
- You must not make statements such as ’this lab was a wonderful learning experi-
ence’ or ’all students should enjoy this lab’.

31.11.16 Recomendations

- This is the section to restate the key numbers that support the purpose of the lab.
- clear, concise recommendations are needed.
-
 writing guide - 31.25

31.11.17 Appendices

- Some material is too bulky for a report body, these are normally moved to an
appendix.
- when material is placed in an appendix, it must be summarized in the body of the
report.
- The report must briefly summarize (ussually a figure or equation or more) and
then refer to the appendix.
- it is expected that there will be some duplication between the appendix and the
summary in the appendix.

31.11.18 Units

- Put a apce between numbers an units

- verify that units match the numerical results]
- radians are one of the units that may not observe norma; conventions.

31.12 GENERAL WRITING ISSUES

This list below indicates some of the general problems students encounter when
writing technical reports and some strategies for fixing these problems.

Basic Spelling - A document should always be checked for spelling. Considering

that utilities for checking spelling are available in most software this is
expected. Be aware that ’smell checkers’ will only point out mis-spelled words,
not words used inappropriately, so you should also proofread.
Technical Spelling - Many technical terms are not in the dictionaries used for
checking spelling. You may add these terms to the dictionary, or visually verify.
Be very careful when using the ’autoreplace’ options in software.
Basic Grammar - ’Grammar checkers’ can be used to look for obvious problems.
Using simple sentence structures will reduce problems and speed the writing
process. Grammar checking software should not be used as a replacement for
proofreading.
Technical Grammar - Normally grammar checking will reject text written in pas-
sive voice, but the software can often be reconfigured. This software will also
be confused by the interchangeable use of nouns and verbs common in techni-
cal english, such as input.
Jargon and Acronyms - A number of technical terms and acronyms have been
 writing guide - 31.26

developed for efficiency and clarity. Examples include DMM, HTTP, kitted,
parted, etc. All acronyms should be defined at their first use.
Colloquialisms - Avoid informal language in technical reports. Use of informal
language such as ’show me the beef’ will look unprofessional, confuse some
readers and make material easily dated.

31.13 WRITERS BLOCK

Everybody has suffered from writers block before. The obvious cause is not being
sure what to write. This can be caused by a lack of understanding about the topic, or possi-
bly how to express the point.

Knowledge - In the case you are not sure what you are writing about you should
spend time clarifying your knowledge before returning to writing. Reorganizing
the material often helps to create clarity.
Lack of knowledge - Current knowledge is based upon knowledge discovered and
used before. This means that no matter how simple something apparently is, it
has more layersw of knowledge than could be known by any one person. In the
cases where you don’t know everything you should define what you do/don’t
know.
Skip that great opening paragraph - It is quite acceptable to start by writing central
sections of a report. Many authors will write the abstract, introduction and con-
clusions last.
It doesn’t need to sound impressive - Simply write what you mean to say. If you
are having trouble saying it skip it and come back later, or leave it out.

31.14 TECHNICAL ENGLISH

The table below indicates some good and poor choices for words and terms to use
in technical papers.

Table 2: Technical Words and Phrases

Good Bad

calculated found,
measured read
 writing guide - 31.27

Table 2: Technical Words and Phrases

Good Bad

chose optimized
parallax error human error

Table 3: Plain English

Good Bad

was it became obvious that

came in at
representing
 writing guide - 31.28

Table 3: Plain English

Good Bad

Generally bad phrases,

in order to be
needed to be ----> is
needed to be used
so as to
decided to be
can be located
found to have
found through
it was found that
 writing guide - 31.29

implementation of
important
precise
exact
perfect
noted to be
invoved
allowed for it
was found to be
was looked thorough ---> reviewed
along with - with
also
then
decided on - selected
found - measured, calculated

31.15 EVALUATION FORMS

The forms on the following pages are examples of forms that may be used to eval-
uate written reports during the semester. The

Lab Report Form - used to grade weekly lab reports

Design Report Form - used to evaluate the final reports
Peer Evaluation Form - this is used by students for team mates
 writing guide - 31.30

EGR 345 - Laboratory Report Evaluation

Student:

Lab Number/Title:

Technical:
technically accurate
theory section is complete
complete procedure and collected data
data analysis
theory is verified with data
conclusions based on data
adequate use of numbers

A = _____ / 10

Communication:
clear and concise
good use of figures, graphs and tables
use of english
professional style
passive voice/past tense
suitable use of formatting

B = _____ / 10

Overall: A * B = ______ / 100

 writing guide - 31.31

EGR 345 - Design Project Evaluation Sheet

Team:
no yes
Follows defined formats and requirements

Writing style is clear and concise and has a logical flow

The paper is technically accurate and complete

The writing is professional

Overall writing quality and style

Comments:

31.16 PATENTS

United States Patent 6,368,227

Olson April 9, 2002
Method of swinging on a swing

Abstract
 writing guide - 31.32

A method of swing on a swing is disclosed, in which a user positioned on a standard swing sus-
pended by two chains from a substantially horizontal tree branch induces side to side motion by
pulling alternately on one chain and then the other.
Inventors: Olson; Steven (337 Otis Ave., St. Paul, MN 55104)
Appl. No.: 715198
Filed: November 17, 2000

Current U.S. Class: 472/118

Intern'l Class: A63G 009/00
Field of Search: 472/118,119,120,121,122,123,125
References Cited [Referenced By]
U.S. Patent Documents
242601Jun., 1881Clement472/118.
5413298May., 1995Perreault248/228.

Primary Examiner: Nguyen; Kien T.

Attorney, Agent or Firm: Olson; Peter Lowell
Claims

I claim:

1. A method of swinging on a swing, the method comprising the steps of:

a) suspending a seat for supporting a user between only two chains that are hung from a tree
branch;

b) positioning a user on the seat so that the user is facing a direction perpendicular to the tree
branch;

c) having the user pull alternately on one chain to induce movement of the user and the swing
toward one side, and then on the other chain to induce movement of the user and the swing toward
the other side; and

d) repeating step c) to create side-to-side swinging motion, relative to the user, that is parallel to
the tree branch.

2. The method of claim 1, wherein the method is practiced independently by the user to create the
side-to-side motion from an initial dead stop.

3. The method of claim 1, wherein the method further comprises the step of:

e) inducing a component of forward and back motion into the swinging motion, resulting in a
swinging path that is generally shaped as an oval.
 writing guide - 31.33

4. The method of claim 3, wherein the magnitude of the component of forward and back motion is
less than the component of side-to-side motion.
Description

TECHNICAL FIELD

The present invention relates to a method of swinging on a swing.

BACKGROUND OF THE INVENTION

A few basic types of swings have been around for generations. Perhaps the most common is one
that includes a seat suspended between two ropes or chains that are hung from a tree branch or
other substantially horizontal support. These swings are often found in side-by-side sets of two or
three or more on, for example, a school playground.

Young children often need help to climb onto a swing, and may need a push (sometimes even an
"underdog" push) to begin swinging. Others may be able to begin the swinging movement on
their own by pushing with their feet against the ground, and once moving may coordinate the
motion of their legs and body in what may be called "pumping" to sustain the movement of the
swing. When swinging in this manner, the user travels along a path as generally shown in the
cross-section of FIG. 1. Another method of swinging on a swing involves twisting the seat around
repeatedly so that the chains or ropes are wound in a double helix. When allowed to unwind, the
swing spins quickly, which can be entertaining for the user.

These methods of swinging on a swing, although of considerable interest to some people, can lose
their appeal with age and experience. A new method of swinging on a swing would therefore rep-
resent an advance of great significance and value.

SUMMARY OF THE INVENTION

In accordance with one embodiment of the present invention, a method is provided for swinging
on a swing. The swing comprises a seat for supporting a user that is suspended between two
chains that are hung from a substantially horizontal tree branch. The method comprises the steps
of: a) positioning a user on the seat; and b) having the user pull alternately on one chain to induce
movement of the user and the swing toward one side, and then on the other chain to induce move-
ment of the user and the swing toward the other side, to create side-to-side motion. In another
embodiment of the invention, the swinging method may be practiced independently by the user to
create the side-to-side motion from an initial dead stop. These and other features of the invention
are described in greater detail below.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic top view of the swinging path of a swing used in accordance with conven-
tional swinging methods.
 writing guide - 31.34

FIG. 2 is a front view of a swinging path of a swing used in accordance with one embodiment of
the swinging method of the present invention.

FIG. 3 is a schematic top view of a swinging path of a swing used in accordance with a second
embodiment of the swinging method of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The present inventor has created, through experimentation on a standard swing, a new and
improved method of swinging. The swing is of the type described above, in which a seat is sus-
pended between two chains that are hung from a substantially horizontal tree branch. As is appar-
ent to those of ordinary skill in the area of swinging, the chains could be replaced with ropes,
cables, or the like, or the tree branch could be replaced with another substantially horizontal sup-
port such as a metal bar or pole.

The standard swing should be a single swing that is suspended sufficiently far away from obstruc-
tions to make the practice of the inventive swinging method completely safe. That is, the swing
should be suspended a sufficient distance away from the trunk of the tree from which it sus-
pended, and from any other swing, building, support, overhead wire, or other obstruction or threat
to safety that may be present.

The standard method of swinging on a swing is defined by oscillatory motion of the swing and the
user along an axis that is substantially perpendicular to the axis of the tree branch from which the
swing is suspended. This "forward and back" movement has been known for generations, and is
illustrated in FIG. 1. In contrast to the conventional method of swinging, the present inventor has
discovered that much greater satisfaction can be obtained by alternately pulling on one chain to
move the swing and the user toward that side, and then pulling on the other chain to move the
swing and the user toward that side. This side-to-side oscillatory motion of the swing and the user
is thus along an axis that is substantially parallel to the axis of the tree branch from which the
swing is suspended, and is illustrated in FIG. 2. This side to side swinging method has the added
benefit that it can be continued for long periods of time simply by alternately pulling on one chain
and then the other. The importance of sufficient clearance between the swing and any obstructions
or threats to the user's safety is apparent.

The present inventor has discovered certain other improvements in the art of swinging on a swing,
either or both of which can be used in conjunction with the swinging method described immedi-
ately above. The first is that the inventive swinging method can be initiated from a dead stop with-
out pushing, and without the user having to contact the ground. That is, the user can climb onto
the swing, and begin from an initial dead stop to pull first on one chain, and then on the other
chain, alternately until the user and the swing have begun to swing side-to-side in accordance with
the inventive swinging method described herein. This enables even young users to swing indepen-
dently and joyously, which is of great benefit to all.

Another improvement on the swinging method described above is the induction into the side-to-
side swinging movement of a component of forward-and-back motion. That is, by skillful manip-
ulation of the body, the present inventor has found it possible to add a relatively minor component
 writing guide - 31.35

of forward-and-back motion to the side-to-side swinging motion, resulting in a swinging path that
is generally shaped like an oval, as is shown in FIG. 3. It is preferred that the magnitude of the for-
ward-and back motion (shown in FIG. 3 as being along the Y axis) be less than the magnitude of
the side-to side motion (shown in FIG. 3 as being along the X axis), so that the latter predomi-
nates. In this manner, the motion can be more easily continued simply by alternately pulling on
one chain and then the other in the manner described.

Lastly, it should be noted that because pulling alternately on one chain and then the other resem-
bles in some measure the movements one would use to swing from vines in a dense jungle forest,
the swinging method of the present invention may be referred to by the present inventor and his
sister as "Tarzan" swinging. The user may even choose to produce a Tarzan-type yell while swing-
ing in the manner described, which more accurately replicates swinging on vines in a dense jungle
forest. Actual jungle forestry is not required.

Licenses are available from the inventor upon request.

 writing guide - 32.1

32. PROJECTS

32.1

The project is intended to emphasize proper project management and team skills to
produce a complex engineering system from concept to completion.

32.2 OVERVIEW

Cranes are used to move large loads in industrial environments. To do this they
normally lift a load and then move it to a new position. At the end of the motion
the load is put down. If the load is swaying, the operator may have to pause
before putting down the load. This delay is undesirable because it increases the
total cycle time for the crane.

This design project will endeavor to build a proof-of-concept prototype for an anti-
sway system for a gantry crane as shown in Figure 32.1. The crane will ride on
a wooden beam constructed with 2x4 lumber (actually smaller). An electric
 writing guide - 32.2

motor will drive the cart left or right. The mass will hang below the cart.

motor driven cart

pivot point fixed rail (wood 2x4)

1Kg mass

Figure 32.1 The equipment

The system will use active compensation to move the cart between positions, and
to move the cart to compensate for swaying of the suspended mass. The overall
system performance will be evaluated on how quickly the crane can move the
load, and then eliminate sway. Additional factors used in assessment of perfor-
mance will be the mass, cost and overall design quality.

32.2.1 The Objectives and Constraints

The design will be mounted on 2x4 lumber and controlled through a wire harness
to external electronics and power supplies. The apparatus must be easily
mounted and removed without damaging the experimental setup. It will be
expected to carry a 1Kg mass, suspended 40cm below the top of the 2x4. The
2x4 will be oriented so that the longer dimension is vertical. The mass will be
mounted with a carriage bolt that is at least 3 inches long, with a 1/4" diameter.
The arm must be able to swing freely, as it is meant to be equivalent to a cable
suspending the mass. The mass will be provided with a rod that will be between
1/4" and 1/2". The mass will be a cylinder that is free to rotate to reduce
dynamic effects rotation. The width of the mass will be less than or equal to the
 writing guide - 32.3

width of the 2x4. None of the cart, except for the arms, should be less than 2cm
below the bottom of the 2x4.

The objectives for the design are itemized below. The following equation will be
used to assess the overall score for the design.
- The equipment cost should be minimized. The goal of $150 or less has
been specified as very important.
- Mass should be minimized, the target value is 0.2Kg.
- A clearly justified and supported design process. This will be judged
before the competition.
- The settling time, ts, to less than a +/- 0.5 inch margin around the com-
manded position. This time will be determined using the best ts/d for two
arbitrary target motions.
- The test will start with the mass stationary at one position. The gantry sys-
tem must move the mass to another position once started. The total
movement range will be 2 to 20 inches in 2 inch increments.

C M
 t s 2 --------
200
-
B T
-------
0.2
score = --
- ( 4 ) ( 10 ) ( 10 ) ( 2 )
 d
where,
t s = the time to settle (s)
C = total cost of part ($)
d = distance moved in test (m)
B = build quality score assigned by judges (0=best, 1=worst)
T = theory quality score assigned by judges (0=best, 1 = poor)
M = mass of the apparatus (Kg)

32.3 MANAGEMENT

The management details for the project are outlined below. These may evolve as
the semester progresses.

32.3.1 Timeline - Tentative

Sept 12 - Teams assigned

Sept 12 - Teams matched with EGR 101 students
Sept 14 - EGR 345 Teams visit EGR 101 lab - EC616 (6pm) to formalize contracts
Sept 16 - Contract submitted
 writing guide - 32.4

Oct 4 - Preliminary design concept submitted with specifications

materials list estimate
budget estimate
Gantt charts
Oct 13 - Design concept approved
Oct 24 - Proposal submitted
detailed drawings (CAD)
materials list
budget
calculations/simulations
Oct 29 - Proposals approved, 345/101 building begins
Nov 12 - First test
Nov 19 - Second test
Nov 20 - First draft of report posted to the web used to determine the score for
competition
Nov 25 - Competition and judging
Dec 4 - Final report draft posted to the web

32.3.2 Teams

The teams are composed of up to 4 students from EGR 345 and one student from
EGR 101. The EGR 345 teams are grouped using the skills survey attached in
the appendices.

Teams are expected to divided tasks for members to work in parallel. It is also
expected that team members will review the work of others to ensure accuracy
and completeness. This is particularly true of calculations, materials lists draw-
ings and budgets.

All team members are expected to work in a professional manner. The general
rules of conduct in a team are,
- treat others as you want to be treated
- communicate expectations and problems clearly
- be polite and accommodating
- when problems arise, help to solve them, even if they are not your fault.
Don’t lay the blame for problems on others.

Personal conflicts must be resolved by team members in a professional manner.

The performance of the team will be assessed using peer evaluations on a regu-
lar basis and may impact individual grades. In the case of non-participation the
penalty may be up to 100% of the project grade.

In the event that team members cannot resolve differences with a team member,
the team may ’fire’ the team member by a vote (it must be unanimous, except
 writing guide - 32.5

for the member in question). In this case the ’fired’ student is responsible for
finding another team that will accept (or ’hire’) them. In the event they cannot
find another team to join, they will be expected to perform all of the work them-
self. The firing mechanism is intended to deal with individuals who ’harm’ the
progress of the team, not for non-participants.

32.4 DELIVERABLES

The following items are to be submitted at various stages of the project.

32.4.1 Conceptual Design

Concepts are normally communicated with sketches that make the overall design
clear. Components that would be expected for this type of design are,
- Sketches should be done using normal drafting practices. A good set of
sketches will include 2-D, isometric and pictorial views.
- Calculations should be provided that support the design concept.
- Electrical schematics to describe the control system.
- Block diagram to describe the control system.
- Lists of components and budgets to indicate the major parts of the system.
- Other items, such as flowcharts, are often required to clarify the design
concept.

32.4.2 EGR 345/101 Contract

This project involves participants from EGR 101 and 345. Although the students
are at different points in their academic careers they are all considered peers,
and should have equal voices in the team. To help clarify issues of differences
between the EGR 101 and 345 students, a ’contract’ will be written that formal-
izes the responsibilities of the students in EGR 101. As with most contracts, it
outlines a process that is mutually beneficial. In this case both parties are
expected to benefit in an educational sense. The contract will be written in such
a manner that if all requirements are satisfied work will be marked ’as-is’. In
the event of a failure, marks will be deducted to hold the responsible party(s)
accountable.
 writing guide - 32.6

32.4.3 Progress Reports

Teams are expected to submit progress reports on a weekly basis. These reports
will include the following elements divided into sections with a heading for
each. Point form is preferred, but complete sentences must be used. (Note:
’Parts purchased’ should be ’Parts were purchased for the cart assembly.’) Each
section should include items completed since the last report, and current action
items. If there is nothing to be said about a category use ’no changes’, ’nothing
done’, or ’complete’ as appropriate.

Cover Page - a cover sheet indicating the course, project and tema num-
bers. The names of all team members should be listed on the cover,
including the EGR 101 team members.
Gantt chart - updated on a weekly basis and included each time.
Budget - when changes are made, include an updated budget.
Mass Table - when changes are made, include an updated mass table.
Design - Design changes should be indicated. Appropriate drawings, sche-
matics, or equivalent should be included. When appropriate, these should
normally be accompanied by a new set of models, calculations and/or
simulations to verify the new design.
Software - The current status of software development should be indicated,
including major accomplishments and issues.
Fabrication - The status of items being built/assembled should be indicated.
Purchasing - The status of ordered items should be indicated.
Testing - The testing progress should be indicated, including any numerical
results when available.
Other Issues - Items that may impact the success of the team should be
indicated.
Performance - A prediction of performance, including the overall perfor-
mance equation.

Early in the semester, other items will be requested, such as a combined timetable
for all team members, and a skills inventory. These should only be included in the reports
the weeks they are requested.

32.4.4 Design Proposal

The design proposal is used to present all of the design details in a single docu-
ment. The focus of this document is a MINIMAL AMOUNT OF TEXT, but a thorough
presentation of the design details. Typical elements are listed below in a typical sequence;

- a cover page indicating all of the team members and all other pertinent informa-
 writing guide - 32.7

tion.
- a table of contents
- three view drawings of each significant part
- block diagram of the control system
- block diagrams showing the system architecture
- circuit schematic
- if a motion profile will be used, it should be documented
- an assembly drawing of the mechanism, including a BOM
- a budget listing each of the parts that must be purchased/acquired. Catalog pages
and quotes can be used to validate the budget. In the final report, copies of
receipts, or catalog pages will be required.
- a weight inventory, itemized by each part of the design
- additional calculations for mechanical design issues, such as stress that may
result in failure. Normally these result in a factor of safety.
- the equations of motion for the system
- a Scilab program that verifies the operation of the system using the equations of
motion.
- a C program that implements the controller as designed.

32.4.5 The Final Report

The final design will follow the same structure as the Design Proposal, with the
addition of the following elements.

- test results
- the drawings, calculations, etc. should be based upon the final design. It is rea-
sonable to write a page or less about the modifications that were required, but it
is a minimal/optional part of the report.

The report should concisely and clearly describe the design, as shown in the dia-
grams, drawings, calculations, etc. In general the format of the report is as outlined below.
Sections and subsections should be numbered.

Cover page
Executive summary - one page or less that summarizes the design and results.
Table of contents
Design description - this section should describe the mechanical, electrical and
software design aspects. Subsections should focus on the following elements;
Drawing summary - selected isometric and assembly drawings
System block diagrams
Description of control scheme, such as the motion profile
Schematics
 writing guide - 32.8

Calculations - FBDs and differential equations

Project budget and BOM
Weight inventory
Test results - this section should describe
Simulation results
The tests that were done to describe the overall performance. There should
be a comparison of the device with and without sway compensation.
The results of formal tests should also be described.
A comparison of overall score estimates.
Conclusions - A brief description of the overall results indicating what the
strengths and weaknesses of the design.
Reccomendations - Suggestions for improvement.

Appendix - Drawings
Appendix - Stress and other similar calculations
Appendix - Controller C program
Appendix - Simulation program
Appendix - Receipts and cost evidence

Final reports will be evaluated on numerous factors inculding the clarity for the
design documentation (i.e., how clear is what has been done?), theory to backup the
design (does the theory match the actual design?), did the theory and actual match?

32.5 REPORT ELEMENTS

32.5.1 Gantt Charts

Gantt charts are used to track major project tasks, including sequencing. a brief
tutorial on project management with Microsoft Project is available at http://
www.stylusinc.net/ms_project_tutorial/project_management.shtml. You may
also download free project management software from http://www.smart-
works.us/htm/downloads.htm. In general a Gantt chart should include,
- Most project parts should have a development, and then a review stage by
another team member. For example, if one task is preparing mechanical
drawings, this should be followed by a review task - done by another
team member.
- Most tasks that stretch more than a week should be broken into smaller
sub tasks.
- A Gantt chart should be presented on one single sheet. If it is small
enough this can be one page, otherwise a larger sheet can be used.
- On a weekly basis a Gantt chart should be updated to include the comple-
tion of tasks.
 writing guide - 32.9

32.5.2 Drawings

All drawings will observe the standards used in EGR 101 (see the EGR 101 or 345
course pages). This includes dimensions and tolerances that can be produced
using the available equipment and materials. Please note that ’sketch’ means
that it is done by hand, approximately, while drawing means it is done formally
in a CAD package. Normally you should create solid models, and then generate
multiview drawings. Note: All drawings must have a title block.

Shaded views have very little value and should be avoided, wireframe drawings
are much more useful. If there is a definite need to include a shaded drawing,
change the background to change the quantity of toner used.

32.5.3 Budgets and Bills of Material

Budgets should list all substantial components. Consumables, such as bolts are
normally listed under a ’miscellaneous’ heading. However, all other compo-
nents should be listed, and prices provided. If the components have been drawn
from the engineering stores, similar devices can be identified from catalogs and
those prices may be used. If your design calls for parts not commonly issued to
engineering students, you may be required to purchase these yourself. An excel-
lent local source of small parts is hobby stores, such as Ryder’s Hobbies on 28th
St. The budget should also list the quantity of parts/material, price, source/sup-
plier and status (eg., not ordered, received, due 2 weeks, late 1 week).

Don’t forget to include cost of the controller and other components used. Assume
the power supply is provided as part of the crane system and therefore has no
cost is associated. Some of the commonly available components are listed
below. List simple commodity items such as wires, bolts, etc., under a miscella-
neous category with a general cost estimate.
68HC11 Axiom board $89
Grayhill encoder est. $8
Power Supply, list it but do not give a price
Gearhead motors assume $10

A Bill of Materials (BOM) lists all of the parts required to produce or assemble
some other device. This is different from parts listed in a budget in that some of
the parts will be work in process. In other words, the original material has been
worked on to produce new parts. A BOM is normally found on assembly draw-
ings.

Notes;
- there are different type of plastic, some are more brittle, others are
tougher.
 writing guide - 32.10

32.5.4 Calculations

Calculations are required to justify the design work. These should follow the con-
ventions used in EGR 345. When computer programs are written, they should
be commented and included.

32.6 APPENDICES

32.6.1 Appendix A - Sample System

The example system in Figure 32.2 is to control a cart with a hanging pendulum.
The cart is driven by a motor. The motor has an attached encoder to measure the motor
position. As the cart moves the mass underneath may sway. The position of the hanging
mass is measured with a potentiometer.

H-bridge V s Motor ω cart x hanging θ mass

68HC11 PWM mass

quadrature 1-
---
input D
Digital in
encoder
motor position θ motor

Vθ
Analog in potentiometer
mass angle

Figure 32.2 System Architecture

A block diagram for the system is shown in Figure 32.3. A motion generator is
used to generate a smooth path for the mass. There are two feedback control loops. The
first loop adjusts the motor for general positioning of the system. The inner loop adjusts
the output to reduce sway of the cart. The two control signals are added to produce a com-
mand signal the drives the load to the final position and then reduces vibrations.
 writing guide - 32.11

Cp generate update Cd
setpoint setpoints
table

where,
C p = commanded position
C d = desired position setpoint

Cp encoder
K enc

-
Cd C pe C pc Cc Vs θp θL
cart and
K pp PWM motor mass
+ +
- C Lc
CL VL potentiometer
K sp A/D
K pot θ L – V zero

where,
C pe = the position error
K pp = the proportional gain for position control
C p = the position count from the encoder
θ p = the rotational position of the motor
C pc = output command for position control
C c = the combined motor control output
V s = effective voltage to the motor
θ p = position of the motor (cart)
θ L = angle of the load from vertical
V L = voltage proportional to angle of load
C L = integer value for load angle
C Lc = output command for sway control

Figure 32.3 System Block Diagram

- for a stationary cart

 writing guide - 32.12

··
x
··
M p lθ L
θL Fp

x
l
Fw ·· · 2
Mc x θL Mp l
Mc
Mp
M p g cos θ L
Fp
M p g sin θ L
θL ··
··
x sin θL M p
x cos θ L M p
where,
M c = mass of cart
M p = mass of payload
F p = force in suspension arm
F w = force from wheels
θ L = angle of payload from vertical
l = length of suspension arm
r w = radius of cart wheel
·· ··
∑ F = – Mp g sin θL – Mp lθL – x cos θL Mp = 0
·· ··
lθ L = – g sin θ L – x cos θL (1)
· 2 ··
∑ F = F p – M p g cos θ L – θ L M p l + x sin θ L M p = 0

· 2 ··
F p = M p g cos θ L + θ L M p l – x sin θ L M p
··
∑ Fx = – M c x + F p sin θ L + F w = 0
··
– M c x + F p sin θ L + F w = 0

·· ·2 ·· (2)
– M c x + ( M p g cos θL + θ L M p l – x sin θ L M p ) sin θ L + F w = 0

Figure 32.4 FBDs for the suspended mass and the compensator
 writing guide - 32.13

2 Fw rw
ω w + ω w  ------ = V s  ------ – ------------

· K  K
JR JR J

 K  ·  J   K2 
∴F w = V s --------- – ω w ----- – ω w  ---------
 r w R  r w  r w R
given x = θw rw

·· J  ·  K -
2
∴Fw = V s  --------- – x  ----2- – x  --------
K
 r w R 
 r w  r 2w R
(1) --->
·· J  ·  K -
2
– M c x + ( M p g cos θ L + θ L M p l – x sin θ L M p ) sin θL + V s  --------- – x  ----2- – x  --------
·· · 2 ·· K
 r w R  = 0
 r w  r 2w R
··  ·  K -
2
– x  M c + ----2- + ( sin θL ) M p + cos θ L sin θ L M p g + θ L ( M p l sin θ L ) + V s  --------- – x  --------
J 2 ·2 K
 r w R  = 0
 rw   r 2w R
2 2 2
M c r w + J + ( sin θL ) M p r w
·· -------------------------------------------------------------- · 2  K - ·  -------- K -
2
x -  = ( cos θ L sin θ L )M p g + θ L ( M p l sin θ L ) + V s  -------- – x 2 
 rw
2
 r w R  r w R
2 2
 M p gr w  2 M p l sin θ L r w 
·· - + θ·L  --------------------------------------------------------------
x = ( cos θL sin θL )  -------------------------------------------------------------- - +
 M c r 2w + J + ( sin θL ) 2 M p r 2w  M c r 2w + J + ( sin θ L )2 M p r 2w
 Kr w   –K
2 
- + x·  -----------------------------------------------------------------------
V s  ----------------------------------------------------------------------- -
 R ( M c r 2w + J + ( sin θ L ) 2 M p r 2w )  R ( M c r 2w + J + ( sin θ L ) 2 M p r 2w )

Figure 32.5 Adding the differential equation for the motor

 writing guide - 32.14

State Equations:
·
x = v
2
·  –K
2   M p gr w 
- + ( cos θ L sin θ L )  --------------------------------------------------------------
v = v  ----------------------------------------------------------------------- -
 R ( M c r w + J + ( sin θ L ) M p r w )
2 2 2
 M c r w + J + ( sin θ L ) M p r w
2 2 2

2
2 M p l sin θ L r w   Kr w 
+ ω L  --------------------------------------------------------------
- + V s  -----------------------------------------------------------------------
-
 M c r 2w + J + ( sin θ L ) 2 M p r 2w  R ( Mc r 2w + J + ( sin θ L ) 2 M p r 2w )
·
θL = ωL
·
· –-----g- v cos θ L
ωL = sin θ L – -----------------
l l

Figure 32.6 State equations for the system

A Scilab program that simulates the given system is given in Figure 32.7.
 writing guide - 32.15

// contest.sce

// System component values

l = 0.4; // 40cm
Mp = 1.0; // 1kg
Mc = 0.2; // 200g
g = 9.81; // good old gravity
rw = 0.03; // 6cm dia tires
K = 0.5; // motor speed constant
R = 7; // motor resistance
J = 0.005; // rotor inertia

// System state
x0 = 0; // initial conditions for position
v0 = 0;
theta0 = 0.0;// the initial position for the load
omega0 = 0.0;
X=[x0, v0, theta0, omega0];

// The controller definition

PI = 3.14159;
ppr = 16;// encoder pulses per revolution;
Kpot = 1.72;// the angle voltage ratio
Vzero = 2.5;// the voltage when the pendulum is vertical
Vadmax = 5;// the A/D voltage range
Vadmin = 0;
Cadmax = 255;// the A/D converter output limits
Cadmin = 0;
tolerance = 0.5;// the tolerance for the system to settle
Kpp = 20;// position feedback gain
Ksp = 2;// sway feedback gain
Vpwmmax = 12; // PWM output limitations in V
Cpwmmax = 255; // PWM input range
Cdeadpos = 100;// deadband limits
Cdeadneg = 95;

function foo=control(Cdesired)
Cp = ppr * X($, 1)/(2*PI*rw);
Cpe = Cdesired - Cp;
Cpc = Kpp * Cpe;
VL = Kpot * X($,3) + 2.5; // assume the zero angle is 2.5V
CL = ((VL - Vadmin) / (Vadmax - Vadmin)) * (Cadmax - Cadmin);
if CL > Cadmax then CL = Cadmax; end// check for voltages over limits
if CL < Cadmin then CL = Cadmin; end
CLc = Ksp * (CL - (Cadmax + Cadmin) / 2);
Cc = Cpc + CLc;
Cpwm = 0;
if Cc > 0.5 then// deadband compensation
Cpwm = Cdeadpos + (Cc/Cpwmmax)*(Cpwmmax - Cdeadpos);
end
if Cc <= -0.5 then
Cpwm = -Cdeadneg + (Cc/Cpwmmax)*(Cpwmmax - Cdeadneg);
end

foo = Vpwmmax * (Cpwm / Cpwmmax) ; // PWM output

if foo > Vpwmmax then foo = Vpwmmax; end // clip voltage if too large
if foo < -Vpwmmax then foo = -Vpwmmax; end
endfunction

Figure 32.7 A Scilab program to simulate the swaying mass on the crane
 writing guide - 32.16

// The motion profile generator

function foo=motion(t_start, t_end, t_now, C_start, C_end)
if t_now < t_start then
foo = C_start;
elseif t_now > t_end then
foo = C_end;
else
foo = C_start + (C_end - C_start) * (t_now - t_start ) / (t_end - t_start);
end
endfunction

// define the state matrix function

term1 = Mc*rw^2 + J;// Precalculate these terms to save time
term2 = R*term1;
Xd = 10;// the setpoint 10 turns == 160 pulses
Cd = ppr * Xd / (2 * PI * rw) ;
function foo=derivative(state,t, h)
Vs=control(motion(0, 6, t($, 1), 0, Cd));
// Vs=control(Cd);
term3 = cos(state($,3));// precalculate some repeated terms to save time
term4 = sin(state($,3));
term5 = term1 + Mp * (term4 * rw)^2;

//printf("%f %f \n", Cd, Vs);

v_dot = -state($,2)*(K^2) / (term5 * R) ...// d/dt v
+ term3*term4*Mp*g*rw^2 / term5 ...
+ state($,4)^2 * Mp*l*term4*rw^2 / term5 ...
+ Vs*K*rw / term5;

foo = [ state($,2), ...// d/dt x = v

v_dot, ...
state($, 4), ...// d/dt theta
-g * term4 / l - state($, 2) * term3 / l ...// d/dt omega
];
endfunction

// Integration Set the time length and step size for the integration
steps = 5000;// The number of steps to use
t_start = 0;// the start time - normally use zero
t_end = 10;// the end time
h = (t_end - t_start) / steps;// the step size
t = [t_start];// an array to store time values
for i=1:steps,
t = [t ; t($,:) + h];
X = [X ; X($,:) + h * derivative(X($,:), t($,:), h)];// first order
end

Figure 32.8 A Scilab program to simulate the swaying mass on the crane
 writing guide - 32.17

// Graph the values for part e)

plot2d(t, [X(:,1) + l*sin(X(:,3))], [-2], leg="mass position");
//plot2d(t, [X(:,1), 10*X(:, 3)], [-2, -5], leg="position@theta (X 10)");
xtitle('Time (s)');

// printf the values over time

intervals = 20;
for time_count=1:intervals,
i = int((time_count - 1)/intervals * steps + 1);
xmass = X(i,1) + l*sin(X(i,3));
printf("Mass location at t=%f x=%f \n", i * h, xmass);
// printf("Point at t=%f x=%f, v=%f, theta=%f, omega=%f \n", i * h, X(i, 1), X(i, 2),
X(i, 3), X(i, 4));
end

// find the settling time in the results array

ts = 0;
for i = 1:steps,
xmass = X(i,1) + l*sin(X(i,3));
if xmass > (Cd + tolerance) then ts = i*h + t_start;
end
if xmass < (Cd - tolerance) then ts = i*h + t_start;
end
end
printf("\nTheoretical settling time %f \n", ts);

Figure 32.9 A Scilab program to simulate the swaying mass on the crane
 writing guide - 32.18

32.6.2 Appendix B - EGR 345/101 Contract

The text below outlines the general form of a contract between students in EGR
101 and 345.

This contract has been entered into this date by the parties of the first part Joe Junior, Pete
Zaa, Anne Nyther and Robert Sochs, to be referred to as ’345 students’, with Virve Meurte, to be
referred to as ’101 student(s)’.

Articles:
1. The 101 student is to participate in the design and construction of a cart as outlined below.
The 345 students are to prepare a design and construct a multicomponent system that uses the cart
as described below. The result must be a fully functional systems that meets the published design
objectives.

2. The 345 students are expected to prepare a functional design for an anti-sway system for a
crane. This design will include a cart that is designed in coordination with the 101 student. The
345 students will be required to do all calculation including system dynamics and strength of
materials. The 101 student will be responsible for all other design details related to the cart
including the geometry, mass, budget bill of materials and construction. This design will be
documented fully by the 101 student using accepted CAD practices and ProE. The design work will
result in the submission of a Formal Proposal, as shown on the Schedule of Actions.

3. Dr. Jack and/or Farris will comment on the Formal Proposal. Based upon these comments the 101
and 345 students will revise the design and agree upon a design for the cart. This will be labelled
Cart Build Approval. This will be signed by all parties and submitted to Dr. Farris by the
Scheduled date.

4. The cart will be build according to the Cart Build Approval before the date specified on the
Schedule of Actions. At the end of this period the design must be fully documented in ProE and be
ready for inclusion in the Design Report Draft.

5. Both the 101 and 345 students will participate in the first tests to verify the operation of the
system and develop a First List of Deficiencies. This list will be finalized and signed, according
to the date on the Schedule of Actions. The First List of Deficiencies will include a list of
remedies to be performed by the 101 and 345 students.

6. The 345 students will prepare the Design Report Draft using the ProE drawing submitted by the
101 student. They are responsible for submitting the report by the date in the Schedule of Actions.

7. Both the 101 and 345 students will participate in the final tests to verify the operation of the
system and develop a Final List of Deficiencies. This list will be finalized and signed, according
to the date on the Schedule of Actions. The Final List of Deficiencies will include a list of
remedies to be performed by the 101 and 345 students. Any changes made to the design must be
updated and submitted to the 345 students for inclusion in the Design Report.

8. Both the 101 and 345 students will participate in the Competition listed in the Schedule of
Actions.

9. The 345 students are to submit the final report with all necessary changes by the date listed in
the Schedule of Actions.

10. The 101 student is expected to produce a cart that is built to professional standards. All
drawings are expected to observe professional standards. When communicating drawings, generally
accessible files formats should be used.

11. The 345 students are to, at all times, maintain a functional design concept. They must ensure
that this will lead to a system that functions within the rules of the competition.

12. In the event of a dispute, 101 and 345 students are expected to resolve any conflicts
informally and mitigate any losses. In the event that one or both parties fundamentally breach the
 writing guide - 32.19

contract Dr. Farris and Dr. Jack will acts as arbiters. If this occurs, one or both of the parties
will be penalized. This may involve actions as severe as receiving a failing grade in the project.

Exhibits:

1. Schedule of Actions
Oct 15-24 - Cart designs are developed by 101 and 345 students
resulting in submission of the Formal Proposal
Oct 30, 2003 - Cart Build Approval submitted
Nov 11 - Initial build completed
Nov 12 - First test completed and First List of Deficiencies submitted
Nov 15 - Design Report Draft submitted for review
Nov 19 - Final test completed and Final List of Deficiencies submitted
Nov 25 - Competition
Dec 4 - Submit Final Report

32.6.3 Appendix C - Forms

Skills Self Evaluation

Peer Evaluations
 writing guide - 32.20

Skills Self Evaluation

Your Name:

none proficient
Hands-on Mechanical:
The ability to build components with wood, plastic, metal or other materials.
1 2 3 4 5
Hands-on Electrical: 1 2 3 4 5
Basic wiring skills, soldering, etc.

Hands-on Computer Usage: 1 2 3 4 5

CAD, Spreadsheets, creating web pages, etc.

Hands-on Computer Application: 1 2 3 4 5

Programming and computer interfacing

Mathematical Problem Solving: 1 2 3 4 5

Ability to formulate and solve complex problems

Writing:
Layout and write complex documents
1 2 3 4 5
Teamwork Skills:
The ability to work with others in a team environment.
1 2 3 4 5
Leadership Skills:
The ability to act as a role model that teammates will follow.
1 2 3 4 5
Design Skills:
Work in unstructured/semistructured problem solving.
1 2 3 4 5
Personal/Technical Strengths:
Personal/Technical Weaknesses:

People you would like to work with:

People you would NOT like to work with:
Other Commitments (courses, work, etc. - give hours for each)

Other Items of Interest:

 writing guide - 32.21

EGR 101 / 345 Project Peer Evaluation

Your Name:

Your Class: EGR 101 / 345

Person Being Evaluated:

good poor
Communicates well: 1 2 3 4 5
Did the teammate return e-mails and other forms of communication promptly? Could the teammate
understand, explain and evaluate the technical aspects of the project in a clear concise manner?

Works in team environment: 1 2 3 4 5

Did your teammate come to meetings on time? Did the teammate participate in all aspects of the
project? How much did the teammate’s efforts contribute to the overall success of the project?

Meets deadlines: 1 2 3 4 5
Did you teammate complete individual tasks on time? Did the teammate keep the project progressing forward in a timely
manner with a consistent effort throughout the project or was the teammate only available when the team was in trouble?

Quality of work: 1 2 3 4 5
Was you r teammate willing to accept and carry out individual tasks on time? How well were these
individual tasks carried out? Did your teammate do his or her fair share of the work?

Overall: 1 2 3 4 5
Would you be happy working with the person again? Would you give this person a job reference?

Would you hire this person: yes / no

Other Comments:
 writing guide - 32.22

Competition Check Sheet

Team Number:

Arms Swing Freely:

arms should swing freely, with the exception of measurement appa- YES / NO
ratus

Height of Payload:
40cm below the beam = 40cm

Mass of Cart:
not including the harness
M = Kg
Cost:
proof of costs
C = $
Build Quality:
assigned by judges. 0 is the best
C = [ 0, 1 ]
Technical Analysis:
0 is the best
T = [ 0, 1 ]
Test Distance:
the distance specified in the test
d = m
Settling Times:
The time required to move within +/- 0.5in. from the target.
ts = s

Overall Score:
based upon other values
C M SCORE =
t s 2 --------
- -------
score =  --- ( 4 ) ( 10 ) ( 10 ) ( 2 )
200 B T 0.2
d

Comments:
 writing guide - 32.23

EGR 101 / 345 Project Judging

Your Name:

Team Number: Work Quality:

excellent average poor

1 0 1 2 3 4 5 6 7 8 9 10

2 0 1 2 3 4 5 6 7 8 9 10

3 0 1 2 3 4 5 6 7 8 9 10

4 0 1 2 3 4 5 6 7 8 9 10

5 0 1 2 3 4 5 6 7 8 9 10

6 0 1 2 3 4 5 6 7 8 9 10

7 0 1 2 3 4 5 6 7 8 9 10

8 0 1 2 3 4 5 6 7 8 9 10

9 0 1 2 3 4 5 6 7 8 9 10

10 0 1 2 3 4 5 6 7 8 9 10

11 0 1 2 3 4 5 6 7 8 9 10

12 0 1 2 3 4 5 6 7 8 9 10
 writing guide - 33.1

33. ENGINEERING PROBLEM SOLVING

- solving problems is important.....

33.1 BASIC RULES OF STYLE

The basic rules are listed below.

- Always starts mechanics/circuits problems with FBDs or schematics

- Create temporary variables when necessary, but eliminate them
- Eliminate redundant variables
- Always simply the results of calculations
- Box final answers
- Show derivations for all problems
- Use green engineering paper and write your name, problem number and
date on the top of each sheet.
- Note: a calculator is not a method
- Problems should be solved in a logical order from the top to the bottom of
the page (not left to right).
- Use engineering notation, not scientific notation
- Don’t cram things into one page, many problems will take less than one
page.
- Don’t start a long problem at the bottom of a page.
- Don’t just look at the final grade for an assignment, read the comments.
Ussually comments will include elements that should be fixed. In later
assignments these items normally result in reduced marks.
- Variable subscripts are very important, don’t be careless with these.

33.2 EXPECTED ELEMENTS

Students in the junior year are expected to,

- Compare the results of calculations when asked to do so. This is also

expected when different calculations should lead to the same result.
- When comparing results, use different methods
- verify that units make sense
- verify that numerical magnitudes are reasonable.
- Use significant figures as appropriate
 writing guide - 33.2

- If answers are provided, solutions should match normally. This means if

answers don’t match, students should check them.

33.3 SEPCIAL ELEMENTS

33.3.1 Graphs

- Label all axis, including units

- Provide a legend for surves with multiple traces
- Equations are plotted as smooth functions.
- Data is plotted as discrete points.

33.3.2 EGR 345 Specific

- Frequency responses should be in phase angle forms.

33.4 SCILAB

Basic ruules are,

- when solving problems using computational tools, such as C programs, or

numerical simulations, provide copies of the programs.
- when used to solve problems, include program listings
 writing guide - 33.3

33.5 TERMINOLOGY

The terminology is significant,

graph: a plot,
numbers: a value
equation: a function, a relationship, an expression,
proof: a derivation
 math guide - 34.1

34. MATHEMATICAL TOOLS

***** This contains additions and sections by Dr. Andrew Sterian.

• We use math in almost every problem we solve. As a result the more relevant top-
ics of mathematics are summarized here.

• This is not intended for learning, but for reference.

34.1 INTRODUCTION

• This section has been greatly enhanced, and tailored to meet our engineering
requirements.

• The section outlined here is not intended to teach the elements of mathematics,
but it is designed to be a quick reference guide to support the engineer required to use
techniques that may not have been used recently.

• For those planning to write the first ABET Fundamentals of Engineering exam,
the following topics are commonly on the exam.

- quadratic equation
- straight line equations - slop and perpendicular
- conics, circles, ellipses, etc.
- matrices, determinants, adjoint, inverse, cofactors, multiplication
- limits, L’Hospital’s rule, small angle approximation
- integration of areas
- complex numbers, polar form, conjugate, addition of polar forms
- maxima, minima and inflection points
- first-order differential equations - guessing and separation
- second-order differential equation - linear, homogeneous, non-homogeneous,
second-order
- triangles, sine, cosine, etc.
- integration - by parts and separation
- solving equations using inverse matrices, Cramer’s rule, substitution
- eigenvalues, eigenvectors
- dot and cross products, areas of parallelograms, angles and triple product
- divergence and curl - solenoidal and conservative fields
- centroids
- integration of volumes
 math guide - 34.2

- integration using Laplace transforms

- probability - permutations and combinations
- mean, standard deviation, mode, etc.
- log properties
- taylor series
- partial fractions
- basic coordinate transformations - cartesian, cylindrical, spherical
- trig identities
- derivative - basics, natural log, small angles approx., chain rule, partial fractions

34.1.1 Constants and Other Stuff

• A good place to start a short list of mathematical relationships is with greek let-
ters

name
lower case upper case

α Α alpha
β Β beta
γ Γ gamma
δ ∆ delta
ε Ε epsilon
ζ Ζ zeta
η Η eta
θ Θ theta
ι Ι iota
κ Κ kappa
λ Λ lambda
µ Μ mu
ν Ν nu
ξ Ξ xi
ο Ο omicron
π Π pi
ρ Ρ rho
σ Σ sigma
τ Τ tau
υ Υ upsilon
φ Φ phi
χ Χ chi
ψ Ψ psi
ω Ω omega
 math guide - 34.3

Figure 34.1 The greek alphabet

• The constants listed are amount some of the main ones, other values can be
derived through calculation using modern calculators or computers. The values are typi-
cally given with more than 15 places of accuracy so that they can be used for double preci-
sion calculations.

1 n
e = 2.7182818 = lim  1 + --- = natural logarithm base
n→∞ n

π = 3.1415927 = pi

γ = 0.57721566 = Eulers constant

1radian = 57.29578°

Figure 34.2 Some universal constants

34.1.2 Basic Operations

• These operations are generally universal, and are described in sufficient detail for
our use.

• Basic properties include,

commutative a+b = b+a

distributive a ( b + c ) = ab + ac

associative a ( bc ) = ( ab )c a + (b + c) = (a + b) + c

Figure 34.3 Basic algebra properties

 math guide - 34.4

34.1.2.1 - Factorial

• A compact representation of a series of increasing multiples.

n! = 1 ⋅ 2 ⋅ 3 ⋅ 4 ⋅ … ⋅ n

0! = 1

Figure 34.4 The basic factorial operator

34.1.3 Exponents and Logarithms

• The basic properties of exponents are so important they demand some sort of
mention

1
---
n m n+m 0 n
(x )( x ) = x x = 1 , if x is not 0 x = n x
m
–p 1 ----
n
(x ) x = ----p- n n m
---------- n–m x = x
m
= x x
(x )
n n n x-- = n------x-
n m n⋅m
( xy ) = ( x ) ( y ) n
(x ) = x y n y

Figure 34.5 Properties of exponents

• Logarithms also have a few basic properties of use,

 math guide - 34.5

The basic base 10 logarithm:

y
log x = y x = 10
The basic base n logarithm:
y
log n x = y x = n
The basic natural logarithm (e is a constant with a value found near the start of this section:
y
ln x = log e x = y x = e

Figure 34.6 Definitions of logarithms

• All logarithms observe a basic set of rules for their application,

logn ( xy ) = log n ( x ) + log n ( y ) logn ( n ) = 1

logn ( 1 ) = 0
logn  -- = log n ( x ) – logn ( y )
x
y

y
log n ( x ) = ylogn ( x )

log m ( x )
log n ( x ) = -------------------
logm ( n )

ln ( A ∠θ ) = ln ( A ) + ( θ + 2πk )j k∈I

Figure 34.7 Properties of logarithms

34.1.4 Polynomial Expansions

• Binomial expansion for polynomials,

 math guide - 34.6

n n n–1 n( n – 1 ) n – 2 2 n
( a + x ) = a + na x + -------------------- a x +…+x
2!

Figure 34.8 A general expansion of a polynomial

34.1.5 Practice Problems

1. Are the following expressions equivalent?

a) A ( 5 + B ) – C = 5A + B – C
A + B- A B
b) ------------- = ---- + ----
C+D C D
c) log ( ab ) = log ( a ) + log ( b )
4 5
d) 5(5 ) = 5
e) 3 log ( 4 ) = log ( 16 )
2
f) ( x + 6 ) ( x – 6 ) = x + 36
log ( 5 ) 10
g) 10 = ------
5
6
h) (------------------
x + 1 ) - = x 2 + 2x + 1
2
(x + 1)

2. Simplify the following expressions.

2
a) x ( x + 2 ) – 3x 5---  8---
h)
3  9
2
(--------------------------------------
x + 3 ) ( x + 1 )x -
b) 2
(x + 1) x i)  5---
3
c) log ( x )  4
64 --------
d) ------ 5
16 3
j) (y + 4) (y – 2)
15 3-
e) ------ + -----
21 28 2
2 3 4 k) x y
f) (x y )
2 4 x-----------
+ 1-
g) 4x – 8y l) = 4
x+2
 math guide - 34.7

3. Simplify the following expressions.

a) A + B-
------------
AB
AB -
b) ------------
A+B
4 5 3
c)  (--------------
x y )-
 2 
 x 
d) log ( x 5 ) + log ( x 3 )

(ans.

a) A + B-
------------ A B 1 1
= ------- + ------- = --- + ---
AB AB AB B A
AB -
b) ------------
A + B cannot be simplified
3
c)  (--------------
x y )
4 5
2 5 3 6 15
 2 - = (x y ) = x y
 x 

d) log ( x 5 ) + log ( x 3 ) = 5 log ( x ) + 3 log ( x ) = 8 log ( x )

4. Simplify the following expressions.

2 3 4
a) n log ( x ) + m log ( x ) – log ( x )

(ans.
a) 2 3 4
n log ( x ) + m log ( x ) – log ( x )
2n log ( x ) + 3m log ( x ) – 4 log ( x )
( 2n + 3m – 4 ) log ( x )
( 2n + 3m – 4 ) log ( x )
 math guide - 34.8

5. Rearrange the following equation so that only ‘y’ is on the left hand side.
y----------
+ x-
= x+2
y+z

(ans. y----------
+ x-
= x+2
y+z
y + x = (x + 2)(y + z)
y + x = xy + xz + 2y + 2z
y – xy – 2y = xz + 2z – x
y ( – x – 1 ) = xz + 2z – x
xz + 2z – x
y = --------------------------
–x–1

6. Find the limits below.

a)  t3 + 5 
lim  ----------------
t → 0 5t 3 + 1

 t3 + 5 
b) lim  ----------------
t → ∞ 5t 3 + 1

(ans.  t3 + 5  3
0 +5 -
a) lim  ---------------- = ---------------------- = 5
t → 0 5t 3 + 1 5(0) + 1
3

b)  t3 + 5  ∞
3
+ 5 ∞
3
lim  ---------------- = -----------------------
- = -------------- = 0.2
t → ∞ 5t 3 + 1 5(∞) + 1
3
5(∞)
3
 math guide - 34.9

34.2 FUNCTIONS

34.2.1 Discrete and Continuous Probability Distributions

Binomial

∑  t  p q
n t n–t
q, p ∈ [ 0, 1 ]
P(m) = q = 1–p
t≤m

Poisson
t –λ
λe
P(m) = ∑ ------------
t!
λ>0
t≤m

Hypergeometric
 r  s 
 t   n – t
P ( m ) = ∑ -----------------------
-
t≤m
 r + s
 n 
Normal
1 x –t2
P ( x ) = ---------- ∫ e dt
2π –∞

Figure 34.9 Distribution functions

34.2.2 Basic Polynomials

• The quadratic equation appears in almost every engineering discipline, therefore

is of great importance.

2
2
ax + bx + c = 0 = a ( x – r 1 ) ( x – r 2 ) – b ± b – 4ac
r 1, r 2 = --------------------------------------
2a

Figure 34.10 Quadratic equation

 math guide - 34.10

• Cubic equations also appear on a regular basic, and as a result should also be con-
sidered.

3 2
x + ax + bx + c = 0 = ( x – r 1 ) ( x – r 2 ) ( x – r 3 )

First, calculate,
2 3
3b – a 9ab – 27c – 2a S =
3
R+ Q +R
3 2
T =
3
R– Q +R
3 2
Q = ----------------- R = ---------------------------------------
9 54
Then the roots,

a S+T a j 3 S+T a j 3
r 1 = S + T – --- r 2 = ------------ – --- + --------- ( S – T ) r 3 = ------------ – --- – --------- ( S – T )
3 2 3 2 2 3 2

Figure 34.11 Cubic equations

• On a few occasions a quartic equation will also have to be solved. This can be
done by first reducing the equation to a quadratic,

4 3 2
x + ax + bx + cx + d = 0 = ( x – r 1 ) ( x – r 2 ) ( x – r 3 ) ( x – r 4 )

First, solve the equation below to get a real root (call it ‘y’),
3 2 2 2
y – by + ( ac – 4d )y + ( 4bd – c – a d ) = 0

Next, find the roots of the 2 equations below,

 a + a 2 – 4b + 4y
2  y + y2 – 4d
r 1, r 2 = z +  ------------------------------------------- z +  ------------------------------ = 0
 2   2 

2  a – a – 4b + 4y  y – y2 – 4d
2
r 3, r 4 = z +  ------------------------------------------- z +  ------------------------------ = 0
 2   2 

Figure 34.12 Quartic equations

 math guide - 34.11

34.2.3 Partial Fractions

• The next is a flowchart for partial fraction expansions.

start with a function that

has a polynomial numerator
and denominator

is the
order of the use long division to
numerator >= yes reduce the order of the
denominator? numerator

no
Find roots of the denominator
and break the equation into
partial fraction form with
unknown values

OR
use limits technique. use algebra technique
If there are higher order
roots (repeated terms)
then derivatives will be
required to find solutions

Done

Figure 34.13 The methodolgy for solving partial fractions

• The partial fraction expansion for,

 math guide - 34.12

1 - A B C
x ( s ) = --------------------
2
= ----2 + --- + -----------
s (s + 1) s s s+1

C = lim ( s + 1 )  --------------------
1 -
2  = 1
s → –1 s (s + 1)

A = lim s  --------------------
2 1 - 1
2  = lim ----------- = 1
s→0 s (s + 1) s→0 s + 1

B = lim ----- s  --------------------

1 -
= lim -----  -----------
d 2 d 1 –2
2  = lim [ – ( s + 1 ) ] = – 1
s → 0 ds s (s + 1) s → 0 ds s + 1 s→0

Figure 34.14 A partial fraction example

• Consider the example below where the order of the numerator is larger than the
denominator.
 math guide - 34.13

3 2
5s + 3s + 8s + 6-
x ( s ) = -------------------------------------------
2
s +4
This cannot be solved using partial fractions because the numerator is 3rd order
and the denominator is only 2nd order. Therefore long division can be used to
reduce the order of the equation.

5s + 3

2 3 2
s +4 5s + 3s + 8s + 6
3
5s + 20s
2
3s – 12s + 6
2
3s + 12
– 12s – 6
This can now be used to write a new function that has a reduced portion that can be
solved with partial fractions.

– 12s – 6- –---------------------
12s – 6- A B
x ( s ) = 5s + 3 + ---------------------
2
solve
2
= ------------- + -------------
s +4 s +4 s + 2j s – 2j

Figure 34.15 Solving partial fractions when the numerator order is greater than the
denominator

• When the order of the denominator terms is greater than 1 it requires an expanded
partial fraction form, as shown below.

5
F ( s ) = -----------------------3-
2
s (s + 1)
5 - A B C D E
----------------------- = ----2 + --- + ------------------3- + ------------------2- + ----------------
2 3 s (s + 1) (s + 1) (s + 1)
s (s + 1) s

Figure 34.16 Partial fractions with repeated roots

 math guide - 34.14

• We can solve the previous problem using the algebra technique.

5 A B C D E
------------------------ = ---- + --- + ------------------3- + ------------------2- + ----------------
2 3 2 s (s + 1) (s + 1) ( s + 1)
s (s + 1) s
3 3 2 2 2 2
A ( s + 1 ) + Bs ( s + 1 ) + Cs + Ds ( s + 1 ) + Es ( s + 1 ) -
= -----------------------------------------------------------------------------------------------------------------------------------------
3
2
s (s + 1)
4 3 2
s ( B + E ) + s ( A + 3B + D + 2E ) + s ( 3A + 3B + C + D + E ) + s ( 3A + B ) + ( A )-
= --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
3
2
s (s + 1)
–1
0 1 0 0 1 A 0 A 0 1 0 0 1 0 5
1 3 0 1 2 B 0 B 1 3 0 1 2 0 – 15
3 3 1 1 1 C = 0 C = 3 3 1 1 1 0 = 5
3 1 0 0 0 D 0 D 3 1 0 0 0 0 10
1 0 0 0 0 E 5 E 1 0 0 0 0 5 15

5 5 – 15 5 10 15
------------------------ = ---- + --------- + ------------------- + ------------------- + ----------------
3 2 s 3 2 (s + 1)
2
s (s + 1) s (s + 1) (s + 1)

Figure 34.17 An algebra solution to partial fractions

34.2.4 Summation and Series

• The notation ∑ xi is equivalent to x a + x a + 1 + x a + 2 + … + x b assuming a

i=a
and b are integers and b ≥ a . The index variable i is a placeholder whose name does not
matter.

• Operations on summations:

b a

∑ xi = ∑ xi
i=a i=b
 math guide - 34.15

b b

∑ αxi = α ∑ xi
i=a i=a
b b b

∑ xi + ∑ yj = ∑ ( xi + yi )
i=a j=a i=a
b c c

∑ xi + ∑ xi = ∑ xi
i=a i = b+1 i=a
 b  
d b d
   y =
 ∑ i  ∑ j ∑ ∑ xi yj
x
i = a  j = c  i=a j=c

• Some common summations:

N
1
∑i = --- N ( N + 1 )
2
i=1

N–1  1 – αN
 ----------------, α ≠ 1
∑ α =  1 – α
i
for both real and complex α .
i=0  N, α = 1
∞
1
∑α
i
= ------------, α < 1 for both real and complex α . For α ≥ 1 , the summation
1–α
i=0
does not converge.
 math guide - 34.16

34.2.5 Practice Problems

1. Convert the following polynomials to multiplied terms as shown in the example.

e.g., 2
x + 2x + 1 = ( x + 1 ) ( x + 1 )
2 2
a) x – 2x + 1 d) x + x + 10
2
b) x –1 e)
2
c) x +1 f)

2. Solve the following equation to find ‘x’.

2
2x + 8x = –8

(ans. 2
2x + 8x = – 8
2
x + 4x + 4 = 0
2
(x + 2) = 0
x = – 2, – 2

3. Reduce the following expression to partial fraction form.

x + 4-
----------------
2
x + 8x

(ans. x + 4-
---------------- x+4
------------------- A B-
--- + ----------- Ax + 8A + Bx-
--------------------------------
= = =
2
x + 8x x(x + 8) x x+8 2
x + 8x
( 1 )x + 4 = ( A + B )x + 8A

8A = 4 A = 0.5

1 = A+B B = 0.5

----------------- = 0.5
x+4 0.5 -
------- + -----------
2 x x +8
x + 8x
 math guide - 34.17

34.3 SPATIAL RELATIONSHIPS

34.3.1 Trigonometry

• The basic trigonometry functions are,

y 1
sin θ = -- = -----------
r csc θ
x 1
cos θ = -- = ----------- r
r sec θ
y
y 1 sin θ
tan θ = -- = ----------- = ------------
x cot θ cos θ

Pythagorean Formula:
2
r = x +y
2 2 θ

• Graphs of these functions are given below,

Sine - sin

-270° -180° -90° 0° 90° 180° 270° 360° 450°

-1
 math guide - 34.18

Cosine - cos

-270° -180° -90° 0° 90° 180° 270° 360° 450°

-1

Tangent - tan

-270° -180° -90° 0° 90° 180° 270° 360° 450°

-1

Cosecant - csc

-270° -180° -90° 0° 90° 180° 270° 360° 450°

-1
 math guide - 34.19

Secant - sec

-270° -180° -90° 0° 90° 180° 270° 360° 450°

-1

Cotangent - cot

-270° -180° -90° 0° 90° 180° 270° 360° 450°

-1

• NOTE: Keep in mind when finding these trig values, that any value that does not
lie in the right hand quadrants of cartesian space, may need additions of ±90° or ±180°.
 math guide - 34.20

Cosine Law:
2 2 2
c = a + b – 2ab cos θ c
c
θA
b
Sine Law:

a -
------------- b c
= -------------- = --------------
sin θ A sin θ B sin θ C

θC
θB

• Now a group of trigonometric relationships will be given. These are often best
used when attempting to manipulate equations.
 math guide - 34.21

sin ( – θ ) = – sin θ cos ( – θ ) = cos θ tan ( – θ ) = – tan θ

sin θ = cos ( θ – 90° ) = cos ( 90° – θ ) = etc.

sin ( θ 1 ± θ 2 ) = sin θ 1 cos θ 2 ± cos θ 1 sin θ 2 OR sin ( 2θ ) = 2 sin θ cos θ

− sin θ sin θ
cos ( θ1 ± θ 2 ) = cos θ 1 cos θ 2 + OR
2
cos ( 2θ ) = ( cos θ ) + ( sin θ )
2
1 2

tan θ 1 ± tan θ 2
tan ( θ 1 ± θ 2 ) = ------------------------------------
-
1− + tan θ 1 tan θ 2

cot θ 1 cot θ2 − +1
cot ( θ 1 ± θ 2 ) = -------------------------------------
tan θ 2 ± tan θ 1

θ – cos θ-
sin --- = ± 1-------------------- -ve if in left hand quadrants
2 2
θ + cos θ-
cos --- = ± 1--------------------
2 2
θ sin θ 1 – cos θ
tan --- = --------------------- = ---------------------
2 1 + cos θ sin θ
2 2
( cos θ ) + ( sin θ ) = 1

• Numerical values for these functions are given below.

θ (deg) sin θ cos θ tan θ

-90 -1.0 0.0 -infinity

-60 -0.866 0.5
-45 -0.707 0.707 -1
-30 -0.5 0.866
0 0 1 0
30 0.5 0.866
45 0.707 0.707 1
60 0.866 0.5
90 1.0 0.0 infinity
 math guide - 34.22

• These can also be related to complex exponents,

jθ –jθ jθ –j θ
e +e e –e
cos θ = ---------------------- sin θ = ---------------------
2 2j

34.3.2 Hyperbolic Functions

• The basic definitions are given below,

x –x
e –e
sinh ( x ) = ------------------ = hyperbolic sine of x
2

x –x
e +e
cosh ( x ) = ------------------ = hyperbolic cosine of x
2
x –x
sinh ( x ) e –e
tanh ( x ) = ------------------- = -----------------
- = hyperbolic tangent of x
cosh ( x ) x
e +e
–x

1 2 -
csch ( x ) = ------------------ = ----------------- = hyperbolic cosecant of x
sinh ( x ) x
e –e
–x

1 2
sech ( x ) = ------------------- = -----------------
- = hyperbolic secant of x
cosh ( x ) x
e +e
–x

x –x
cosh ( x ) e +e -
coth ( x ) = ------------------- = ----------------- = hyperbolic cotangent of x
sinh ( x ) x
e –e
–x

• some of the basic relationships are,

 math guide - 34.23

sinh ( – x ) = – sinh ( x )

cosh ( –x ) = cosh ( x )

tanh ( – x ) = – tanh ( x )

csch ( – x ) = – csch ( x )

sech ( – x ) = sech ( x )

coth ( –x ) = – coth ( x )

• Some of the more advanced relationships are,

2 2 2 2 2 2
( cosh x ) – ( sinh x ) = ( sech x ) + ( tanh x ) = ( coth x ) – ( csch x ) = 1
sinh ( x ± y ) = sinh ( x ) cosh ( y ) ± cosh ( x ) sinh ( y )
cosh ( x ± y ) = cosh ( x ) cosh ( y ) ± sinh ( x ) sinh ( y )
tanh ( x ) ± tanh ( y )
tanh ( x ± y ) = -----------------------------------------------
1 ± tanh ( x ) tanh ( y )

• Some of the relationships between the hyperbolic, and normal trigonometry func-
tions are,

sin ( jx ) = j sinh ( x ) j sin ( x ) = sinh ( jx )

cos ( jx ) = cosh ( x ) cos ( x ) = cosh ( jx )

tan ( jx ) = j tanh ( x ) j tan ( x ) = tanh ( jx )

34.3.2.1 - Practice Problems

 math guide - 34.24

1. Find all of the missing side lengths and corner angles on the two triangles below.

5 5 3
3

10°

2. Simplify the following expressions.

cos θ cos θ – sin θ sin θ = ( s + 3j ) ( s – 3j ) ( s + 2j ) =
2

(ans.
cos θ cos θ – sin θ sin θ = cos ( θ + θ ) = cos ( 2θ )
2 2 2 2 2
( s + 3j ) ( s – 3j ) ( s + 2j ) = ( s – 9j ) ( s + 4js + 4j )
4 3 2 2 2 2 2 2 2
s + 4js + 4j s – 9j s – 9j 4js – 9j 4j
4 3 2
s + ( 4j )s + ( 5 )s + ( 36j )s + ( – 36 )

3. Solve the following partial fraction

4 1 -
Note: there was a typo here, so ---------------
-------------------------- =
2
x + 3x + 2 an acceptable answer is also. x + 0.5

(ans. 4 A B Ax + 2A + Bx + B- ( 2A + B ) + ( A + B )x
-------------------------- = ------------ + ------------ = ------------------------------------------ = -------------------------------------------------
2 x + 1 x + 2 2 2
x + 3x + 2 x + 3x + 2 x + 3x + 2
A+B = 0 A = –B
4 - -----------
–4 -
----------- +
2A + B = 4 = – 2B + B = –B B = –4 A = 4 x+1 x+2

34.3.3 Geometry

• A set of the basic 2D and 3D geometric primitives are given, and the notation
used is described below,
 math guide - 34.25

A = contained area

P = perimeter distance

V = contained volume

S = surface area

x, y, z = centre of mass

x, y, z = centroid
I x, I y, I z = moment of inertia of area (or second moment of inertia)

AREA PROPERTIES:

∫y
2
Ix = dA = the second moment of inertia about the y-axis
A

∫x
2
Iy = dA = the second moment of inertia about the x-axis
A

I xy = ∫ xy dA = the product of inertia

A

∫r ∫ (x
2 2 2
JO = dA = + y ) dA = I x + I y = The polar moment of inertia
A A

∫ x dA
A
x = ------------
- = centroid location along the x-axis
∫ dA
A

∫ y dA
A
y = ------------
- = centroid location along the y-axis
∫ dA
A
 math guide - 34.26

y
Rectangle/Square:

A = ab
P = 2a + 2b a

b
Centroid: Moment of Inertia Moment of Inertia
(about centroid axes): (about origin axes):
3 3
b ba ba
x = --- I x = -------- I x = --------
2 12 3
3 3
b a b a
a I y = -------- I y = --------
y = --- 12 3
2
2 2
b a
I xy = 0 I xy = -----------
4
 math guide - 34.27

Triangle: y
bh
A = ------
2 a
P = h
θ
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
a+b
x = ------------ 3 3
3 bh bh
I x = -------- I x = --------
36 12
h
y = ---
3 bh 2 2 bh 2 2
I y = ------ ( a + b – ab ) I y = ------ ( a + b – ab )
36 12
2 2
bh bh
I xy = -------- ( 2a – b ) I xy = -------- ( 2a – b )
72 24

Circle: y
2
r
A = πr

P = 2πr

Centroid: Moment of Inertia Moment of Inertia Mass Moment of Inertia

(about centroid axes): (about origin axes): (about centroid):
4
x = r πr Mr
2
I x = -------- Ix = J z = ----------
4 2
4
πr
y = r I y = -------- Iy =
4
I xy = 0 I xy =
 math guide - 34.28

Half Circle: y
2
πr
A = --------
2
P = πr + 2r r
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
π 8 4
I x =  --- – ------ r
4
x = r πr
8 9π I x = --------
8
4r 4
y = ------ πr πr
4
3π I y = -------- I y = --------
8 8
I xy = 0 I xy = 0

Quarter Circle: y
2
πr
A = --------
4
πr
P = ----- + 2r
2 r
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
4r πr
4
x = ------ 4
Ix = -------
-
3π I x = 0.05488r
16
4
4r 4 πr
y = ------ I y = 0.05488r I y = --------
3π 16
4 4
I xy = – 0.01647 r r
I xy = ----
8
 math guide - 34.29

Circular Arc: y
2
θr
A = --------
2
P = θr + 2r r
x
θ

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
θ r
4
2r sin --- I x = ---- ( θ – sin θ )
2 Ix =
x = ----------------- 8
3θ 4
r
Iy = I y = ---- ( θ + sin θ )
y = 0 8

I xy = I xy = 0
 math guide - 34.30

Ellipse: y
r1
A = πr 1 r 2 r2

π
--- 2 2
r1 + r 2
P = 4r 1 ∫
2 2
1 – -------------------- ( sin θ ) dθ x
0 a
2 2
r 1 + r2
P ≈ 2π ---------------
-
2

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
3
x = r2 πr 1 r 2
I x = ------------- Ix =
4
3
πr 1 r 2
y = r1 I y = ------------- Iy =
4
I xy = I xy =
 math guide - 34.31

Half Ellipse: y
πr 1 r 2
A = ------------
-
2
r1
π 2 2
--- r1 +
r2 r2
P = 2r 1 ∫
2
2
1 – -------------------- ( sin θ ) dθ + 2r 2 x
0 a
2 2
r 1 + r2
P ≈ π ---------------
- + 2r 2
2

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
3
x = r2 πr 2 r 1
I x = 0.05488r 2 r 1
3 I x = ------------
-
16
3
4r 1 3 πr 2 r 1
y = -------- I y = 0.05488r 2 r 1 Iy = ------------
-
3π 16
2 2
2 2 r1 r2
I xy = – 0.01647r 1 r 2 I xy = ---------
8
 math guide - 34.32

Quarter Ellipse: y
πr 1 r 2
A = -------------
4
r1
π
--- 2 2
r1 + r 2
r1 ∫
2 2
P = 1 – -------------------- ( sin θ ) dθ + 2r 2 r2 x
0 a
2 2
π r 1 + r2
P ≈ --- ---------------
- + 2r 2
2 2

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):

4r 2 3
x = -------- Ix = I x = πr 2 r 1
3π
4r 1 3
y = -------- Iy = I y = πr 2 r 1
3π 2 2
r2 r1
I xy = I xy = ---------
8

Parabola: y

2
A = --- ab
3
a
b + 16a
2 2
b
2  4a + b + 16a 2 2
P = --------------------------- + ------ ln  ---------------------------------------- x
2 8a  b 
b
Centroid: Moment of Inertia Moment of Inertia
(about centroid axes): (about origin axes):
b Ix = Ix =
x = ---
2

2a Iy = Iy =
y = ------
5
I xy = I xy =
 math guide - 34.33

Half Parabola: y

ab
A = ------
3
a
b + 16a
2 2
b
2  4a + b2 + 16a 2
P = --------------------------- + --------- ln  ---------------------------------------- x
4 16a  b 
b

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
3 3
3b 8ba 2ba
x = ------ I x = ------------ I x = ------------
8 175 7
3 3
19b a 2b a
2a I y = --------------- I y = ------------
y = ------ 480 15
5 2 2 2 2
b a b a
I xy = ----------- I xy = -----------
60 6

• A general class of geometries are conics. This for is shown below, and can be
used to represent many of the simple shapes represented by a polynomial.
 math guide - 34.34

2 2
Ax + 2Bxy + Cy + 2Dx + 2Ey + F = 0

Conditions A = B = C = 0 straight line

B = 0, A = C circle
2
B – AC < 0 ellipse
2 parabola
B – AC = 0
2
B – AC > 0 hyperbola
 math guide - 34.35

VOLUME PROPERTIES:

∫ rx
2
Ix = dV = the moment of inertia about the x-axis
V

∫ ry
2
Iy = dV = the moment of inertia about the y-axis
V

∫ rz
2
Iz = dV = the moment of inertia about the z-axis
V

∫ x dV
V
x = ------------
- = centroid location along the x-axis
∫ dV
V

∫ y dV
V
y = ------------
- = centroid location along the y-axis
∫ dV
V

∫ z dV
V
z = ------------
- = centroid location along the z-axis
∫ dV
V
 math guide - 34.36

Parallelepiped (box): y

V = abc c
z
S = 2 ( ab + ac + bc )
b
x

Centroid: Moment of Inertia Moment of Inertia Mass Moment of Inertia

(about centroid axes): (about origin axes): (about centroid):
a M(a + b )
2 2
x = --- I x = --------------------------- Ix = Jx =
2 12
2 2
b M(a + c )
y = --- I y = --------------------------- Iy = Jy =
2 12
2 2
c M(b + a )
z = --- I z = --------------------------- Iz = Jz =
2 12

Sphere: y
r
4 3
V = --- πr
3
z
2
S = 4πr
x

Centroid: Moment of Inertia Moment of Inertia Mass Moment of Inertia

(about centroid axes): (about origin axes): (about centroid):
2 2
2Mr 2Mr
x = r I x = ------------- Ix = J x = -------------
5 5
2 2
2Mr 2Mr
I y = ------------- Iy = J y = -------------
y = r 5 5
2 2
2Mr 2Mr
I z = ------------- Iz = J z = -------------
z = r 5 5
 math guide - 34.37

Hemisphere: y

2 3
V = --- πr
3
z
S = r
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
83 2
x = r I x = --------- Mr Ix =
320
2
2Mr
3r I y = ------------- Iy =
y = ----- 5
8
83 2
I z = --------- Mr Iz =
z = r 320

Cap of a Sphere: y

1 2 h
V = --- πh ( 3r – h )
3
z
S = 2πrh r
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):

x = r Ix = Ix =

Iy = Iy =
y =

Iz = Iz =
z = r
 math guide - 34.38

Cylinder: y
r

2
V = hπr h
2 z
S = 2πrh + 2πr
x

Centroid: Moment of Inertia Moment of Inertia Mass Moment of Inertia

(about centroid axis): (about origin axis): (about centroid):
2 2 2 2 2 2
M ( 3r + h )
I x = M  ------ + ---- I x = M  ----- + ----
x = r h r h r
J x = ------------------------------
12 4 3 4 12
2 2
h Mr Mr
y = --- I y = ---------- Iy = J y = ----------
2 2 2
2 2 2 2 2 2
M ( 3r + h )
I z = M  ------ + ---- I z = M  ----- + ----
h r h r
J z = ------------------------------
z = r  12 4   3 4 12

Cone: y
1 2
V = --- πr h
3
2 2 h
S = πr r + h z
r
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):
3 2
I x = M  -------- + --------
x = r 3h 3r
Ix =
80 20
2
h 3Mr
y = --- I y = ------------- Iy =
4 10
3 2
I z = M  -------- + --------
3h 3r
z = r Iz =
80 20
 math guide - 34.39

Tetrahedron: y
z
1
V = --- Ah
3
h

A x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):

x = Ix = Ix =

h Iy = Iy =
y = ---
4
Iz = Iz =
z =

Torus: y
r2
1 2 2
V = --- π ( r 1 + r 2 ) ( r 2 – r 1 )
4 r1
z
2 2 2
S = π ( r2 – r1 ) x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):

x = r2 Ix = Ix =

r 2 – r 1
y =  --------------
- Iy = Iy =
 2 

Iz = Iz =
z = r2
 math guide - 34.40

Ellipsoid: y
r2 r3
4
V = --- πr 1 r 2 r 3
3
z r1
S =
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):

x = r1 Ix = Ix =

Iy = Iy =
y = r2

Iz = Iz =
z = r3

Paraboloid: y

1 2
V = --- πr h h
2
z
S = r
x

Centroid: Moment of Inertia Moment of Inertia

(about centroid axes): (about origin axes):

x = r Ix = Ix =

Iy = Iy =
y =

Iz = Iz =
z = r
 math guide - 34.41

34.3.4 Planes, Lines, etc.

• The most fundamental mathematical geometry is a line. The basic relationships

are given below,

y = mx + b defined with a slope and intercept

1
m perpendicular = ---- a slope perpendicular to a line
m
y2 – y1
m = ---------------- the slope using two points
x2 – x1

--x- + --y- = 1 as defined by two intercepts

a b

• If we assume a line is between two points in space, and that at one end we have a
local reference frame, there are some basic relationships that can be derived.
 math guide - 34.42

θβ d ( x 2, y 2, z 2 )

( x 1, y 1, z 1 )

2 2 2 θγ θα
d = ( x2 – x1 ) + ( y2 – y1 ) + ( z2 – z1 )

x
( x 0, y 0, z 0 )
z

The direction cosines of the angles are,

x2 – x1 y2 – y1 z2 – z1
θ α = acos  ---------------- θ β = acos  ---------------- θ γ = acos  ---------------
d d d
2 2 2
( cos θ α ) + ( cos θ β ) + ( cos θ γ ) = 1
The equation of the line is,
x – x1 y – y1 z–z Explicit
--------------- - = --------------1-
- = ---------------
x2 – x1 y2 – y1 z2 – z1

( x, y, z ) = ( x 1, y 1, z 1 ) + t ( ( x 2, y 2, z 2 ) – ( x 1, y 1, z 1 ) ) Parametric t=[0,1]

• The relationships for a plane are,

 math guide - 34.43

The explicit equation for a plane is, b

Ax + By + Cz + D = 0
P 1 = ( x 1, y 1, z 1 )

where the coefficients defined by the intercepts are,

N
1 1 1
A = --- B = --- C = --- D = –1 P 2 = ( x 2, y 2, z 2 )
a b c
P 3 = ( x 3, y 3, z 3 )
a x

z c

The determinant can also be used,

x – x1 y – y1 z – z1
det x – x 2 y – y 2 z – z 2 = 0
x – x3 y – y3 z – z3

y2 – y1 z2 – z1 z2 – z1 x2 – x1
∴det ( x – x 1 ) + det ( y – y1 )
y3 – y1 z3 – z1 z3 – z1 x3 – x1

x2 – x1 y2 – y1
+ det ( z – z1 ) = 0
x3 – x1 y3 – y1

The normal to the plane (through the origin) is,

( x, y, z ) = t ( A , B , C )

34.3.5 Practice Problems

1. What is the circumferenece of a circle? What is the area? What is the ratio of the area to the cir-
cumference?
 math guide - 34.44

2. What is the equation of a line that passes through the points below?
a) (0, 0) to (2, 2)
b) (1, 0) to (0, 1)
c) (3, 4) to (2. 9)

3. Find a line that is perpendicular to the line through the points (2, 1) and (1, 2). The perpendicu-
lar line passes through (3, 5).

4. Manipulate the following equations to solve for ‘x’.

a) x 2 + 3x = – 2
b) sin x = cos x

(ans.

a) x 2 + 3x = – 2
2
x + 3x + 2 = 0
2
– 3 ± 3 – 4( 1)( 2 ) –3± 9–8 –3±1
x = ------------------------------------------------ = ------------------------------ = ---------------- = – 1, – 2
2(1 ) 2 2
b) sin x = cos x
sin x-
---------- = 1
cos x
tan x = 1
x = atan 1
x = …, –135°, 45°, 225°, …

5. Simplify the following expressions.

2
a) ( cos 2θ )
sin 2θ  ---------------------- + sin 2θ
 sin 2θ 

(ans.
2
a) ( cos 2θ )
sin 2θ  ---------------------- + sin 2θ = ( cos 2θ ) + ( sin 2θ ) = 1
2 2
sin 2θ
 math guide - 34.45

6. A line that passes through the point (1, 2) and has a slope of 2. Find the equation for the line,
and for a line perpendicular to it.

(ans. given,
y = mx + b
m = 2
x = 1
y = 2
substituting
2 = 2(1) + b b = 0
y = 2x
perpendicular
1
m perp = – ---- = – 0.5
m
y = – 0.5 x

34.4 COORDINATE SYSTEMS

34.4.1 Complex Numbers

• In this section, as in all others, ‘j’ will be the preferred notation for the complex
number, this is to help minimize confusion with the ‘i’ used for current in electrical engi-
neering.

• The basic algebraic properties of these numbers are,

 math guide - 34.46

The Complex Number:

2
j = –1 j = –1

Complex Numbers:
a + bj where,
a and b are both real numbers

Complex Conjugates (denoted by adding an asterisk ‘*’ the variable):

N = a + bj N* = a – bj
Basic Properties:
( a + bj ) + ( c + dj ) = ( a + c ) + ( b + d )j

( a + bj ) – ( c + dj ) = ( a – c ) + ( b – d )j

( a + bj ) ⋅ ( c + dj ) = ( ac – bd ) + ( ad + bc )j

= -------------- = -----  ------- =  --------------  ------------- = -----------------

N- a + bj N N* a + bj c – dj ac + bd  bc – ad
---- - + ------------------ j
M c + dj M N* c + dj c – dj 2 2  2 2
c +d c +d

• We can also show complex numbers graphically. These representations lead to

alternative representations. If it in not obvious above, please consider the notation above
uses a cartesian notation, but a polar notation can also be very useful when doing large cal-
culations.
 math guide - 34.47

CARTESIAN FORM
Ij
imaginary (j)
N = R + Ij

R
real

2 2
A = R +I R = A cos θ

θ = atan  --- I = A sin θ

I
R
POLAR FORM
Ij
imaginary (j)
N = A ∠θ

θ
R
real
A = amplitude

θ = phase angle

• We can also do calculations using polar notation (this is well suited to multiplica-
tion and division, whereas cartesian notation is easier for addition and subtraction),
 math guide - 34.48

jθ
A ∠θ = A ( cos θ + j sin θ ) = Ae

( A 1 ∠θ 1 ) ( A 2 ∠θ 2 ) = ( A 1 A 2 ) ∠( θ 1 + θ 2 )

( A 1 ∠θ 1 ) A
--------------------- =  -----1- ∠( θ 1 – θ 2 )
( A 2 ∠θ 2 )  A 2

n n
( A ∠θ ) = ( A ) ∠( nθ ) (DeMoivre’s theorem)

• Note that DeMoivre’s theorem can be used to find exponents (including roots) of
complex numbers

jθ
• Euler’s formula: e = cos θ + j sin θ

• From the above, the following useful identities arise:

jθ – jθ
e +e
cos θ = ----------------------
2
jθ – jθ
e –e
sin θ = ---------------------
2j

34.4.2 Cylindrical Coordinates

• Basically, these coordinates appear as if the cartesian box has been replaced with
a cylinder,
 math guide - 34.49

z z
( x, y, z ) ↔ ( r, θ, z )

z
z

y
r
x
x y
θ

2 2
x = r cos θ r = x +y

θ = atan  --
y
y = r sin θ x

34.4.3 Spherical Coordinates

• This system replaces the cartesian box with a sphere,

 math guide - 34.50

z z
( x, y, z ) ↔ ( r, θ, φ ) φ

r
z

y
x
x y
θ

2 2 2
x = r sin θ cos φ r = x +y +z

y = r sin θ sin φ θ = atan  --

y
x
z = r cos θ
φ = acos  - 
z
r

34.4.4 Practice Problems

1. Simplify the following expressions.

a) 16 -
--------------------
2
( 4j + 4 )
b) 3j + 5
---------------------
2
( 4j + 3 )
c) ( 3 + 5j )4j where, j = –1
 math guide - 34.51

(ans. a) 16 16 16-
--------------------- = ------------------------------------
- = ------- = – 0.5j
2 – 16 + 32j + 16 32j
( 4j + 4 )

= ---------------------  --------------------- = --------------------------------------- = -----------------------

3j + 5
b) -------------------- 3j + 5 - 3j + 5 – 7 – 24j – 35 – 141j + 72 37 – 141j
- = ---------------------------------
2 – 16 + 24j + 9 – 7 + 24j – 7 – 24j 49 + 576 625
( 4j + 3 )

c) ( 3 + 5j )4j = 12j + 20j 2 = 12j – 20

2. For the shape defined below,

y 2
y = (x + 2)
a) find the area of the shape.
b) find the centroid of the shape.
c) find the moment of inertia of the shape
about the centroid.

34.5 MATRICES AND VECTORS

34.5.1 Vectors

• Vectors are often drawn with arrows, as shown below,

head
terminus

A vector is said to have magnitude (length or

strength) and direction.
origin
tail
 math guide - 34.52

• Cartesian notation is also a common form of usage.

y j

x i
z k
becomes
y j

z x k i

• Vectors can be added and subtracted, numerically and graphically,

A = ( 2, 3, 4 ) A + B = ( 2 + 7, 3 + 8, 4 + 9 )
B = ( 7, 8, 9 ) A – B = ( 2 – 7, 3 – 8, 4 – 9 )

B
Parallelogram Law
A+B
A A
B

34.5.2 Dot (Scalar) Product

• We can use a dot product to find the angle between two vectors
 math guide - 34.53

y F2 = 5i + 3j
F1 • F2
cos θ = -----------------
-
F1 F2
θ F1 = 2i + 4j
 ( 2 ) ( 5 ) + ( 4 )( 3) 
∴θ = acos  -------------------------------------------
 2 2 + 4 2 5 2 + 3 2

∴θ = acos  ----------------------- = 32.5°

22
( 4.47 ) ( 6 ) x

• We can use a dot product to project one vector onto another vector.

z
F 1 = ( – 3i + 4j + 5k )N
We want to find the component of
force F1 that projects onto the
vector V. To do this we first con-
vert V to a unit vector, if we do
not, the component we find will V = 1j + 1k
be multiplied by the magnitude y
of V.

F1
V 1j + 1k
λV = ------ = --------------------- = 0.707j + 0.707k
V 2 2
1 +1
F 1V = λ V • F 1 = ( 0.707j + 0.707k ) • ( – 3i + 4j + 5k )N

∴F 1V = ( 0 ) ( – 3 ) + ( 0.707 ) ( 4 ) + ( 0.707 ) ( 5 ) = 6N V

F 1V

• We can consider the basic properties of the dot product and units vectors.
 math guide - 34.54

Unit vectors are useful when breaking up vector magnitudes and direction. As an exam-
ple consider the vector, and the displaced x-y axes shown below as x’-y’.

y F = 10N
y’ x’

45°

60°
x
We could write out 5 vectors here, relative to the x-y axis,

x axis = 2i
y axis = 3j
x‘ axis = 1i + 1j
y‘ axis = – 1i + 1j
F = 10N ∠60° = ( 10 cos 60° )i + ( 10 sin 60° )j
None of these vectors has a magnitude of 1, and hence they are not unit vectors. But, if
we find the equivalent vectors with a magnitude of one we can simplify many tasks.
In particular if we want to find the x and y components of F relative to the x-y axis we
can use the dot product.

λ x = 1i + 0j (unit vector for the x-axis)

F x = λ x • F = ( 1i + 0j ) • [ ( 10 cos 60° )i + ( 10 sin 60° )j ]

∴ = ( 1 ) ( 10 cos 60° ) + ( 0 ) ( 10 sin 60° ) = 10N cos 60°

This result is obvious, but consider the other obvious case where we want to project a
vector onto itself,
 math guide - 34.55

F 10 cos 60°i + 10 sin 60°j

λ F = ------ = --------------------------------------------------------- = cos 60°i + sin 60°j
F 10

Incorrect - Not using a unit vector

FF = F • F
= ( ( 10 cos 60° )i + ( 10 sin 60° )j ) • ( ( 10 cos 60° )i + ( 10 sin 60° )j )
= ( 10 cos 60° ) ( 10 cos 60° ) + ( 10 sin 60° ) ( 10 sin 60° )
2 2
= 100 ( ( cos 60° ) + ( sin 60° ) ) = 100

Using a unit vector

FF = F • λF
= ( ( 10 cos 60° )i + ( 10 sin 60° )j ) • ( ( cos 60° )i + ( sin 60° )j )
= ( 10 cos 60° ) ( cos 60° ) + ( 10 sin 60° ) ( sin 60° )
2 2
= 10 ( ( cos 60° ) + ( sin 60° ) ) = 10 Correct
Now consider the case where we find the component of F in the x’ direction. Again,
this can be done using the dot product to project F onto a unit vector.

u x' = cos 45°i + sin 45°j

F x' = F • λ x' = ( ( 10 cos 60° )i + ( 10 sin 60° )j ) • ( ( cos 45° )i + ( sin 45° )j )
= ( 10 cos 60° ) ( cos 45° ) + ( 10 sin 60° ) ( sin 45° )
= 10 ( cos 60° cos 45° + sin 60° sin 45° ) = 10 ( cos ( 60° – 45° ) )

Here we see a few cases where the dot product has been applied to find the vector pro-
jected onto a unit vector. Now finally consider the more general case,
 math guide - 34.56

y
V2

V1

θ2 V 2V 1
θ1
x
First, by inspection, we can see that the component of V2 (projected) in the direction
of V1 will be,

V 2V 1 = V 2 cos ( θ 2 – θ1 )

Next, we can manipulate this expression into the dot product form,

= V 2 ( cos θ 1 cos θ2 + sin θ 1 sin θ2 )

= V 2 [ ( cos θ 1 i + sin θ 1 j ) • ( cos θ 2 i + sin θ 2 j ) ]

V1 V2 V 1 • V2 V1 • V2
= V 2 --------- • --------- = V 2 ------------------ = ------------------ = V2 • λ V1
V1 V2 V1 V2 V1

Or more generally,

V1 • V 2
V2 V 1 = V 2 cos ( θ2 – θ 1 ) = V 2 ------------------
V1 V2

V1 • V2
∴ V 2 cos ( θ 2 – θ1 ) = V 2 ------------------
V1 V2

V1 • V2
∴ cos ( θ 2 – θ 1 ) = ------------------
V 1 V2

*Note that the dot product also works in 3D, and similar proofs are used.
 math guide - 34.57

34.5.3 Cross Product

• First, consider an example,

F = ( – 6.43i + 7.66j + 0k )N
NOTE: note that the cross prod-
d = ( 2i + 0j + 0k )m uct here is for the right hand
rule coordinates. If the left
handed coordinate system is
i j k used F and d should be
M = d×F = 2m 0m 0m reversed.
– 6.43N 7.66N 0N

∴M = ( 0m0N – 0m ( 7.66N ) )i – ( 2m0N – 0m ( – 6.43N ) ) j +

( 2m ( 7.66N ) – 0m ( –6.43N ) )k = 15.3k ( mN )

NOTE: there are two things to note about the solution. First, it is a vector. Here
there is only a z component because this vector points out of the page, and a
rotation about this vector would rotate on the plane of the page. Second, this
result is positive, because the positive sense is defined by the vector system.
In this right handed system find the positive rotation by pointing your right
hand thumb towards the positive axis (the ‘k’ means that the vector is about
the z-axis here), and curl your fingers, that is the positive direction.

• The basic properties of the cross product are,

 math guide - 34.58

The cross (or vector) product of two vectors will yield a new vector perpendicular
to both vectors, with a magnitude that is a product of the two magnitudes.

V1 × V2
V1

V2
V1 × V2 = ( x 1 i + y1 j + z1 k ) × ( x2 i + y 2 j + z 2 k )

i j k
V1 × V2 = x1 y1 z1
x2 y2 z2
V 1 × V 2 = ( y 1 z 2 – z 1 y 2 )i + ( z 1 x 2 – x 1 z 2 )j + ( x 1 y 2 – y 1 x 2 )k

We can also find a unit vector normal ‘n’ to the vectors ‘V1’ and ‘V2’ using
a cross product, divided by the magnitude.

V1 × V2
λ n = ---------------------
V1 × V2

• When using a left/right handed coordinate system,

The positive orientation of angles and moments about an axis can be determined
by pointing the thumb of the right hand along the axis of rotation. The fingers
curl in the positive direction.

y x z

x z y
+ + +

z y x

• The properties of the cross products are,

 math guide - 34.59

The cross product is distributive, but not associative. This allows us to collect terms
in a cross product operation, but we cannot change the order of the cross product.

r 1 × F + r 2 × F = ( r 1 + r2 ) × F DISTRIBUTIVE

r×F≠F×r NOT ASSOCIATIVE

but
r × F = –( F × r )

34.5.4 Triple Product

When we want to do a cross product, followed by a dot product (called the mixed triple
product), we can do both steps in one operation by finding the determinant of the follow-
ing. An example of a problem that would use this shortcut is when a moment is found
about one point on a pipe, and then the moment component twisting the pipe is found
using the dot product.

ux uy uz
( d × F) • u = dx dy dz
F x F y Fz

34.5.5 Matrices

• Matrices allow simple equations that drive a large number of repetitive calcula-
tions - as a result they are found in many computer applications.

• A matrix has the form seen below,

 math guide - 34.60

n columns
If n=m then the matrix is said to be square.
a 11 a21 … a n1 Many applications require square matrices.
We may also represent a matrix as a 1-by-3
a 12 a22 … a n2 m rows for a vector.
… … … …
a 1m a 2m … a nm

• Matrix operations are available for many of the basic algebraic expressions,
examples are given below. There are also many restrictions - many of these are indicated.

3 4 5 12 13 14 21
A = 2 B = 6 7 8 C = 15 16 17 D = 22 E = 24 25 26
9 10 11 18 19 20 23

Addition/Subtraction 3 + 12 4 + 13 5 + 14
3+2 4+2 5+2
A+B = 6+2 7+2 8+2 B + C = 6 + 15 7 + 16 8 + 17
9 + 2 10 + 2 11 + 2 9 + 18 10 + 19 11 + 20
B + D = not valid

3–2 4–2 5–2 3 – 12 4 – 13 5 – 14

B–A = 6–2 7–2 8–2 B + C = 6 – 15 7 – 16 8 – 17
9 – 2 10 – 2 11 – 2 9 – 18 10 – 19 11 – 20
B – D = not valid
 math guide - 34.61

Multiplication/Division
3--- 4--- 5---
3(2 ) 4(2 ) 5(2) 2 2 2
A ⋅ B = 6(2 ) 7(2 ) 8(2)
B
--- = 6--- 7--- 8---
9 ( 2 ) 10 ( 2 ) 11 ( 2 ) A 2 2 2
9--- 10 11
------ ------
2 2 2
( 3 ⋅ 21 + 4 ⋅ 22 + 5 ⋅ 23 )
B⋅D = ( 6 ⋅ 21 + 7 ⋅ 22 + 8 ⋅ 23 ) D ⋅ E = 21 ⋅ 24 + 22 ⋅ 25 + 23 ⋅ 26
( 9 ⋅ 21 + 10 ⋅ 22 + 11 ⋅ 23 )

( 3 ⋅ 12 + 4 ⋅ 15 + 5 ⋅ 18 ) ( 3 ⋅ 13 + 4 ⋅ 16 + 5 ⋅ 19 ) ( 3 ⋅ 14 + 4 ⋅ 17 + 5 ⋅ 20 )
B⋅C = ( 6 ⋅ 12 + 7 ⋅ 15 + 8 ⋅ 18 ) ( 6 ⋅ 13 + 7 ⋅ 16 + 8 ⋅ 19 ) ( 6 ⋅ 14 + 7 ⋅ 17 + 8 ⋅ 20 )
( 9 ⋅ 12 + 10 ⋅ 15 + 11 ⋅ 18 ) ( 9 ⋅ 13 + 10 ⋅ 16 + 11 ⋅ 19 ) ( 9 ⋅ 14 + 10 ⋅ 17 + 11 ⋅ 20 )
B- ---
B D
--- , -, ---- , etc = not allowed (see inverse)
C D B
Note: To multiply matrices, the first matrix must have the same number
of columns as the second matrix has rows.
 math guide - 34.62

Determinant

B = 3⋅ 7 8 –4⋅ 6 8 +5⋅ 6 7 = 3 ⋅ ( – 3 ) – 4 ⋅ ( –6 ) + 5 ⋅ ( – 3 ) = 0
10 11 9 11 9 10

7 8 = ( 7 ⋅ 11 ) – ( 8 ⋅ 10 ) = – 3
10 11

6 8 = ( 6 ⋅ 11 ) – ( 8 ⋅ 9 ) = – 6
9 11

6 7 = ( 6 ⋅ 10 ) – ( 7 ⋅ 9 ) = – 3
9 10

D , E = not valid (matrices not square)

Transpose
3 6 9
T T 24
B = 4 7 10 D = 21 22 23 T
E = 25
5 8 11
26
 math guide - 34.63

Adjoint T
7 8 – 6 8 6 7
10 11 10 11 9 10 The matrix of determinant to
the left is made up by get-
ting rid of the row and col-
B = – 4 5 3 5 – 3 4 umn of the element, and
10 11 9 11 9 10
then finding the determi-
4 5 nant of what is left. Note
– 3 5 3 4
the sign changes on alter-
7 8 6 8 6 7 nating elements.

D = invalid (must be square)

Inverse To solve this equation for

x x,y,z we need to move x
–1 –1
D = B⋅ y B to the left hand side. B D = B ⋅B⋅ y
To do this we use the
z z
inverse.
x x
–1
B D = I⋅ y = y
z z

–1 B
B = ---------
B In this case B is singular, so the
inverse is undetermined, and the
matrix is indeterminate.

–1
D = invalid (must be square)
 math guide - 34.64

Identity Matrix
This is a square matrix that is the matrix equivalent to ‘1’.

B⋅I = I⋅B = B
D⋅I = I⋅D = D
–1
B ⋅B = I

1 0 0
, 1 0,
1 0 1 0 , etc=I
0 1
0 0 1

• The eigenvalue of a matrix is found using,

A – λI = 0

34.5.6 Solving Linear Equations with Matrices

• We can solve systems of equations using the inverse matrix,

Given,

2⋅x+4⋅y+3⋅z = 5
9⋅x+6⋅y+8⋅z = 7

11 ⋅ x + 13 ⋅ y + 10 ⋅ z = 12
Write down the matrix, then rearrange, and solve.
–1
2 4 3 x 5 x 2 4 3 5
9 6 8 y = 7 ∴ y = 9 6 8 7 =
11 13 10 z 12 z 11 13 10 12
 math guide - 34.65

• We can solve systems of equations using Cramer’s rule (with determinants),

Given,

2⋅x+4⋅y+3⋅z = 5
9⋅x+6⋅y+8⋅z = 7

11 ⋅ x + 13 ⋅ y + 10 ⋅ z = 12
Write down the coefficient and parameter matrices,

2 4 3 5
A = 9 6 8 B = 7
11 13 10 12
Calculate the determinant for A (this will be reused), and calculate the determinants
for matrices below. Note: when trying to find the first parameter ‘x’ we replace the
first column in A with B.

5 4 3
7 6 8
12 13 10
x = ------------------------------ =
A

2 5 3
9 7 8
11 12 10
y = ------------------------------ =
A

2 4 5
9 6 7
11 13 12
z = ------------------------------ =
A

34.5.7 Practice Problems

1. Perform the matrix operations below.

 math guide - 34.66

Multiply
ANS.
1 2 3 10 1 2 3 10 68
4 5 6 11 = 4 5 6 11 = 167
7 8 9 12 7 8 9 12 266
Determinant

12 3 1 2 3
42 6 = 4 2 6 = 36
78 9 7 8 9

Inverse –1
–1 1 2 3 – 0.833 0.167 0.167
12 3 =
4 2 6 0.167 – 0.333 0.167
42 6 =
7 8 9 0.5 0.167 –0.167
78 9

2. Perform the vector operations below,

ANS.
1 6 Cross Product A×B =
A = 2 B = 2 A × B = ( – 4, 17, – 10 )
3 1 A•B =
Dot Product A • B = 13

4. Solve the following equations using any technique,

5x – 2y + 4z = – 1 ANS.
6x + 7y + 5z = – 2 x= 0.273
y= -0.072
2x – 3y + 6z = – 3
z= -0.627

5. Solve the following set of equations using a) Cramer’s rule and b) an inverse matrix.
2x + 3y = 4
5x + 1y = 0
 math guide - 34.67

(ans. a) 4 3 2 4
2 3 x = 4
0 1 4 5 0 – 20 20
5 1 y 0 x = ---------------- = --------- y = ---------------- = --------- = ------
– 13 – 13 13
2 3 2 3
5 1 5 1

b) T
1 –( 5 ) 1 –3 4
–1
– ( 3 ) 2 – 5 2 – ------
x = 2 3 4 = -------------------------------- 4 = ------------------- 4 =  – -----
1- 4
= 13
y 5 1 0 0 – 13 0  13  – 20 20
2 3 ------
5 1 13

6. Perform the following matrix calculation. Show all work.

T
A B C X L
D E F Y + M
GH I Z N

(ans.
T T T
A B C X L AX + BY + CZ L AX + BY + CZ + L
D E F Y + M = DX + EY + FZ + M = DX + EY + FZ + M
GH I Z N GX + YH + IZ N GX + YH + IZ + N
= AX + BY + CZ + L DX + EY + FZ + M GX + YH + IZ + N

7. Perform the matrix calculations given below.

a) A B C X
D E F Y =
GH I Z

b) A B C
D E F X Y Z =
GH I
 math guide - 34.68

8. Find the dot product, and the cross product, of the vectors A and B below.

x p
A = y B = q
z r

(ans.
A • B = xp + yq + zr

i j k yr – zq
A×B = x y z = i ( yr – zq ) + j ( zp – xr ) + k ( xq – yp ) = zp – xr
p q r xq – yp

9. Perform the following matrix calculations.

a) T
a d e f b) c) –1
a b a b
b g h k c d
c d
c mnp

(ans.
a) T
a d e f d e f
b g h k = a b c g h k = ( ad + bg + cm ) ( ae + bh + cn ) ( af + bk + cp )
c mnp mnp

b)
a b = ad – bc
c d

adj
c) a b d –c d - -----------------
–c -
–1
c d –b a -----------------
a b = ------------------- = ------------------- = ad – bc ad – bc
c d ad – bc – b - -----------------
a -
a b -----------------
c d ad – bc ad – bc
 math guide - 34.69

10. Find the value of ‘x’ for the following system of equations.
x + 2y + 3z = 5
x + 4y + 8z = 0

4x + 2y + z = 1

(ans.
12 3 x 5
14 8 y = 0
42 1 z 1
5 2 3
0 4 8
1 2 1 5 ( 4 – 16 ) + 2 ( 8 – 0 ) + 3 ( 0 – 4 ) – 60 + 16 – 12 – 56
x = ------------------ = ---------------------------------------------------------------------------------- = ---------------------------------- = --------- = – 7
1 ( 4 – 16 ) + 2 ( 32 – 1 ) + 3 ( 2 – 16 ) – 12 + 62 – 42 8
1 2 3
1 4 8
4 2 1

11. Perform the matrix calculations given below.

a) A B C X
D E F Y =
GH I Z

b) A B C
D E F X Y Z =
GH I
c)
det A B =
C D

d) A
A B =
C D

e) –1
A B =
C D
 math guide - 34.70

12. Solve the following set of equations with the specified methods.
3x + 5y = 7
a) Inverse matrix
4x – 6y = 2
b) Cramer’s rule
c) Gauss-Jordan row reduction
d) Substitution

34.6 CALCULUS

• NOTE: Calculus is very useful when looking at real systems. Many students are
turned off by the topic because they "don’t get it". But, the secret to calculus is to remem-
ber that there is no single "truth" - it is more a loose collection of tricks and techniques.
Each one has to be learned separately, and when needed you must remember it, or know
where to look.

34.6.1 Single Variable Functions

34.6.1.1 - Differentiation

• The basic principles of differentiation are,

 math guide - 34.71

Both u, v and w are functions of x, but this is not shown for brevity.

Also note that C is used as a constant, and all angles are in radians.

d-
----- (C) = 0
dx

d- d
----- ( Cu ) = ( C ) ------ ( u )
dx dx

d- d d
----- ( u + v + … ) = ------ ( u ) + ------ ( v ) + …
dx dx dx

d- n n–1 d
----- ( u ) = ( nu ) ------ ( u )
dx dx

d- d d
----- ( uv ) = ( u ) ------ ( v ) + ( v ) ------ ( u )
dx dx dx

d-  u---  ----
v d
- ------ ( u ) –  ----
u- -----
d-
 2 dx ( v )
----- =
dx  v  2 dx
v v

d- d d d
----- ( uvw ) = ( uv ) ------ ( w ) + ( uw ) ------ ( v ) + ( vw ) ------ ( u )
dx dx dx dx

d- d d
----- ( y ) = ------ ( y ) ------ ( u ) = chain rule
dx du dx

d- 1
----- ( u ) = --------------
dx d-
----- ( x)
du

d-
----- (y)
d du
------ ( y ) = --------------
dx d-
----- (x)
du

• Differentiation rules specific to basic trigonometry and logarithm functions

 math guide - 34.72

d- 2 d
d- d ----- ( cot u ) = ( – csc u ) ------ ( u )
----- ( sin u ) = ( cos u ) ------ ( u ) dx dx
dx dx
d- d
d- d ----- ( sec u ) = ( tan u sec u ) ------ ( u )
----- ( cos u ) = ( – sin u ) ------ ( u ) dx dx
dx dx

1 2d
( tan u ) =  ----------- ------ ( u )
d- d- d
----- ----- ( csc u ) = ( – csc u cot u ) ------ ( u )
dx cos u dx dx dx

d- u u d
----- ( e ) = ( e ) ------ ( u ) d- d
dx dx ----- ( sinh u ) = ( cosh u ) ------ ( u )
dx dx
d- 1
----- ( ln x ) = --- d- d
dx x ----- ( cosh u ) = ( sinh u ) ------ ( u )
dx dx

d- 2 d
----- ( tanh u ) = ( sech u ) ------ ( u )
dx dx

• L’Hospital’s rule can be used when evaluating limits that go to infinity.

  ----
d- d- 2
  ----
f(x)  f( x) 
( )   dt    dt 
lim  ---------- = lim  --------------------- = lim  -----------------------
f x - = …
x → a g(x) x → a  d   x → a  d  2
----
  dt- g ( x ) - g ( x )
  ----
dt

• Some techniques used for finding derivatives are,

 math guide - 34.73

Leibnitz’s Rule, (notice the form is similar to the binomial equation) can be used for
finding the derivatives of multiplied functions.

d- n
 ----- d- 0  -----
 ----- d- n d- 1  -----
 n  ----- d- n – 1
( uv ) = ( u ) ( v ) + ( u ) (v)
 dx  dx  dx  1  dx  dx

d- 2  -----
 n  ----- d n–2 n d n d 0
( u )  - ( v ) + … +    ------ ( u )  ------ ( v )
 2  dx dx  n dx dx

34.6.1.2 - Integration

• Some basic properties of integrals include,

In the following expressions, u, v, and w are functions of x. in addition to this, C is a

constant. and all angles are radians.

∫ C dx = ax + C

∫ Cf ( x ) dx = C ∫ f ( x ) dx

∫ ( u + v + w + … ) dx = ∫ u dx + ∫ v dx + ∫ w dx + …
∫ u dv = uv – ∫ v du = integration by parts
1
∫ f ( Cx ) dx = ---- ∫ f ( u ) du
C
u = Cx

d- F(u) u = f(x)
∫ F ( f ( x ) ) dx = ∫ F ( u ) -----
du
( x ) du = ∫ ----------- du
f' ( x )
n+1
x - 1
∫ x dx = ----------- ∫ --x- dx
n
+C = ln x + C
n+1
x
a-
∫ a dx = ------- ∫ e dx
x x x
+C = e +C
ln a

• Some of the trigonometric integrals are,

 math guide - 34.74

3x sin 2x sin 4x
∫ sin x dx ∫ ( cos x )
4
= – cos x + C dx = ------ + ------------- + ------------- + C
8 4 32
n+1
( sin x ) -
∫ cos x dx ∫ cos x ( sin x ) dx = ------------------------
n
= sin x + C +C
n+1
sin x cos x + x
∫ ( sin x ) ∫ sinh x dx
2
dx = – ------------------------------- + C = cosh x + C
2
sin x cos x + x
∫ ( cos x ) ∫ cosh x dx
2
dx = ------------------------------- + C = sinh x + C
2
2
cos x ( ( sin x ) + 2 )-
∫ ( sin x ) dx = – ------------------------------------------- ∫ tanh x dx
3
+C = ln ( cosh x ) + C
3
2
sin x ( ( cos x ) + 2 )-
∫ ( cos x ) dx = -------------------------------------------
3
+C
3

cos ( ax -) --x-
∫ x cos ( ax ) dx = ------------------
a
2
+ sin ( ax ) + C
a

2 2
2x cos ( ax -) -------------------
a x –2
∫ x cos ( ax ) dx = -------------------------
2
2
+ 3
sin ( ax ) + C
a a

• Some other integrals of use that are basically functions of x are,

 math guide - 34.75

n+1
x
∫ x dx = -----------
n
-+C
n+1
ln a + bx
∫ ( a + bx )
–1
dx = ----------------------- + C
b

) dx = ---------------------- ln  --------------------------- + C, a > 0, b < 0

2 –1 1 a + 2 –b
∫ ( a + bx
2 ( – b )a a – x –b
2
ln ( bx + a -)
2 –1
∫ x ( a + bx ) dx = ---------------------------
2b
+C

( a + bx ) dx = --- – ------------- atan  ------------- + C

2 –1 x a x ab
∫x
2
b b ab a

– x ) dx = ------ ln  ------------ + C, a > x

2 –1 1 a+x
∫ (a
2 2 2
2a a–x

2 a + bx
∫ ( a + bx )
–1
dx = ---------------------- + C
b CORRECT??
1
– ---
2
∫ x(x
2 2 2 2
± a ) dx = x ±a +C

2 –1 1 b
∫
2
( a + bx + cx ) dx = ------ ln a + bx + cx + x c + ---------- + C, c > 0
c 2 c

2 –1 1 – 2cx – b
∫ ( a + bx + cx ) dx = ---------- asin ------------------------- + C, c < 0
–c 2
b – 4ac
 math guide - 34.76

1 3
--- ---
2-
∫ ( a + bx ) dx = -----
2 2
( a + bx )
3b
1--- 3---
2-
∫ ( a + bx ) dx = -----
2 2
( a + bx )
3b
3
1 ---
--- 2
2 ( 2a – 3bx ) ( a + bx ) -
∫ x ( a + bx )
2
dx = – ----------------------------------------------------
2
15b
1
 ---
  1
ln  x +  ----2- + x  
2 2
1
--- a
2 2 2  
1---
x ( 1 + a x ) + ---------------------------------------------
2 a
∫ ( 1 + a x ) dx = ----------------------------------------------------------------------------------
2 2
-
2
3---

a  ----2- + x
1 2 2
1
--- 
2 2 2 a
∫ x ( 1 + a x ) dx = ---------------------------
3
-
1
 ---
  1
----- 2 2
1 3
--- 1 ln  x +  2 + x  
--- --- a
2 2 2 ax  ----
1- 2 2 8 2 2 2  
∫
2
x ( 1 + a x ) d x = -----
- + x – -------- x ( 1 + a x ) – --------------------------------------------
-
4 a 2 
8a
2
8a
3

1--- 1---
2 2 21--- 2 2 2 asin ( ax )
∫ ( 1 – a x ) d x =
2
x ( 1 – a x ) + ---------------------
a
1--- 3---

∫ x ( 1 – a x ) dx = – --3-  a----2- – x 
2 2 2 a 1 2 2

1 3 1
--- --- ---
) + a asin  ---
x- 2 2 2 1---
2 2 2 2 x
∫
2 2 2 2
x ( a – x ) d x = – -- ( a – x ) + x ( a – x
4 8  a
1 1
– --- ---

∫ ( 1 + a x ) dx = --a- ln x +  a----2- + x 
2 2 1 2 1 2 2

1
– ---
1 2 1
∫(1 – a
2 2
x ) dx = --- asin ( ax ) = – --- acos ( ax )
a a
 math guide - 34.77

• Integrals using the natural logarithm base ‘e’,

ax
e -
∫ e dx = ------
ax
+C
a

ax
e -
∫ xe dx = ------
ax
2
( ax – 1 ) + C
a

34.6.2 Vector Calculus

• When dealing with large and/or time varying objects or phenomenon we must be
able to describe the state at locations, and as a whole. To do this vectors are a very useful
tool.

• Consider a basic function and how it may be represented with partial derivatives.
 math guide - 34.78

y = f ( x, y, z )
We can write this in differential form, but the right hand side must contain partial
derivatives. If we separate the operators from the function, we get a simpler form.
We can then look at them as the result of a dot product, and divide it into two vec-
tors.

∂ ∂ ∂
( d )y =    f ( x, y, z ) dx +    f ( x, y, z ) dy +    f ( x, y, z ) dz
∂x ∂y ∂z
∂ ∂ ∂
( d )y =   dx +   dy +   dz f ( x, y, z )
∂x ∂y ∂z
∂ ∂ ∂
( d )y =  i + j + k • ( dxi + dyj + dzk ) f ( x, y, z )
∂x ∂y ∂z
We then replace these vectors with the operators below. In this form we can manipu-
late the equation easily (whereas the previous form was very awkward).

( d )y = [ ∇ • dX ]f ( x, y, z )
( d )y = ∇f ( x, y, z ) • dX
( d )y = ∇f ( x, y, z ) dX cos θ
In summary,

∂ ∂ ∂ ∇ • F = the divergence of function F

∇ = i+ j+ k
∂x ∂y ∂z
∇ × F = the curl of function F
F = F x i + F y j + Fz k

• Gauss’s or Green’s or divergence theorem is given below. Both sides give the
flux across a surface, or out of a volume. This is very useful for dealing with magnetic
fields.

∫ ( ∇ • F ) dV =
°∫ FdA
V A
where,
V, A = a volume V enclosed by a surface area A
F = a field or vector value over a volume
 math guide - 34.79

• Stoke’s theorem is given below. Both sides give the flux across a surface, or out
of a volume. This is very useful for dealing with magnetic fields.

∫ ( ∇ × F ) dA =
°∫ FdL
A L
where,
A, L = A surface area A, with a bounding parimeter of length L
F = a field or vector value over a volume

34.6.3 Differential Equations

• Solving differential equations is not very challenging, but there are a number of
forms that need to be remembered.

• Another complication that often occurs is that the solution of the equations may
vary depending upon boundary or initial conditions. An example of this is a mass spring
combination. If they are initially at rest then they will stay at rest, but if there is some dis-
turbance, then they will oscillate indefinitely.

• We can judge the order of these equations by the highest order derivative in the
equation.

• Note: These equations are typically shown with derivatives only, when integrals
occur they are typically eliminated by taking derivatives of the entire equation.

• Some of the terms used when describing differential equations are,

ordinary differential equations - if all the derivatives are of a single variable. In the
example below ’x’ is the variable with derivatives.

d- 2
 ---- x +  ----- x = y
d
e.g.,
 dt dt
first-order differential equations - have only first-order derivatives,

e.g.,  ----
d-  ----
d-
 dt x +  dt y = 2
 math guide - 34.80

second-order differential equations - have at least on second derivative,

2
e.g.,  ----
d-
x +  ----- y = 2
d
 dt  dt
higher order differential equations - have at least one derivative that is higher than
second-order.
partial differential equations - these equations have partial derivatives

• Note: when solving these equations it is common to hit blocks. In these cases
backtrack and try another approach.

• linearity of a differential equation is determined by looking at the dependant vari-

ables in the equation. The equation is linear if they appear with an exponent other than 1.

eg. y'' + y' + 2 = 5x linear

2
( y'' ) + y' + 2 = 5x non-linear
3
y'' + ( y' ) + 2 = 5x non-linear

y'' + sin ( y' ) + 2 = 5x non-linear

34.6.3.1 - First-order Differential Equations

• These systems tend to have a relaxed or passive nature in real applications.

• Examples of these equations are given below,

2 3
y' + 2xy – 4x = 0
y' – 2y = 0

• Typical methods for solving these equations include,

guessing then testing

separation
 math guide - 34.81

homogeneous

34.6.3.1.1 - Guessing

• In this technique we guess at a function that will satisfy the equation, and test it to
see if it works.

y' + y = 0 the given equation

–t
y = Ce the guess
now try to substitute into the equation
–t
y' = –Ce
–t –t
y' + y = – Ce + Ce = 0 therefore the guess worked - it is correct
–t
y = Ce

• The previous example showed a general solution (i.e., the value of ’C’ was not
found). We can also find a particular solution.

–t
y = Ce a general solution
–t
y = 5e a particular solution

34.6.3.1.2 - Separable Equations

• In a separable equation the differential can be split so that it is on both sides of

the equation. We then integrate to get the solution. This typically means there is only a sin-
gle derivative term.
 math guide - 34.82

dx
e.g., ------ + y 2 + 2y + 3 = 0
dy
2
∴dx = ( – y – 2y – 3 )dy
3
–y 2
∴x = -------- – y – 3y + C
3

dx
e.g., ------ + x = 0
dy

∴ – --- dx = dy
1
x

∴ ln ( – x ) = y

34.6.3.1.3 - Homogeneous Equations and Substitution

• These techniques depend upon finding some combination of the variables in the
equation that can be replaced with another variable to simplify the equation. This tech-
nique requires a bit of guessing about what to substitute for, and when it is to be applied.
 math guide - 34.83

dy
e.g., ------ = y-- – 1 the equation given
dx x
y the substitution chosen
u = --
x
Put the substitution in and solve the differential equation,
dy
------ = u – 1
dx
du
∴u + x ------ = u – 1
dx
du –1
∴------ = ------
dx x
du 1
∴– ------ = ---
dx x

∴– u = ln ( x ) + C
Substitute the results back into the original substitution equation to get rid of ’u’,
y
– -- = ln ( x ) + C
x
∴y = – x ln ( x ) – Cx

34.6.3.2 - Second-order Differential Equations

• These equations have at least one second-order derivative.

• In engineering we will encounter a number of forms,

- homogeneous
- nonhomogeneous

34.6.3.2.1 - Linear Homogeneous

• These equations will have a standard form,

 math guide - 34.84

d- 2
 ----  ----
d-
y + A
 dt  dt y + By = 0

• An example of a solution is,

d- 2
 ----  ----
d-
e.g.,  dt y + 6  dt y + 3y = 0
Guess,
Bt
y = e
 ----
d- Bt
 dt y = Be
2
 ----
d- 2 Bt
 dt y = B e

substitute and solve for B,

2 Bt Bt Bt
B e + 6Be + 3e = 0
2
B + 6B + 3 = 0

B = – 3 + 2.449j, – 3 – 2.449j
substitute and solve for B,
( – 3 + 2.449j )t
y = e
– 3 t 2.449jt
y = e e
–3 t
y = e ( cos ( 2.449t ) + j sin ( 2.449t ) )

Note: if both the roots are the same,

Bt Bt
y = C1e + C 2 te

34.6.3.2.2 - Nonhomogeneous Linear Equations

• These equations have the general form,

 math guide - 34.85

d- 2
 ----  ----
d-
y + A
 dt  dt y + By = Cx

• to solve these equations we need to find the homogeneous and particular solu-
tions and then add the two solutions.

y = yh + yp

to find yh solve,
d- 2
 ----  ----
d
-
 dt y + A  dt y + B = 0

to find yp guess at a value of y and then test for validity, A good table of guesses is,
Cx form Guess
A C
Ax + B Cx + D
Ax Ax Ax
e Ce Cxe
B sin ( Ax ) or B cos ( Ax ) C sin ( Ax ) + D cos ( Ax )
or Cx sin ( Ax ) + xD cos ( Ax )

• Consider the example below,

 math guide - 34.86

d- 2
 ----  ----
d- – 2x
y + y – 6y = e
 dt  dt
First solve for the homogeneous part,
d- 2
 ----  ----
d- y = e
Bx
 dt y +  dt y – 6y = 0 try
 ----
d-
y = Be
Bx
 dt
2
 ----
d-
y = B
2 Bx
e
 dt
2 Bx Bx Bx
B e + Be – 6e = 0
2
B +B–6 = 0
B = – 3, 2
– 3x 2x
yh = e +e
Next, solve for the particular part. We will guess the function below.
– 2x
y = Ce
 ----
d-
y = – 2C e
– 2x
 dt
2
 ----
d- – 2x
 dt y = 4Ce
– 2x – 2x – 2x – 2x
4Ce + – 2C e – 6Ce = e

4C – 2C – 6C = 1
C = 0.25
– 2x
y p = 0.25e
Finally,
– 3x 2x – 2x
y = e +e + 0.25e

34.6.3.3 - Higher Order Differential Equations

34.6.3.4 - Partial Differential Equations

• Partial difference equations become critical with many engineering applications

involving flows, etc.
 math guide - 34.87

34.6.4 Other Calculus Stuff

• The Taylor series expansion can be used to find polynomial approximations of

functions.

2 (n – 1) n–1
f'' ( a ) ( x – a ) f ( a)(x – a)
f ( x ) = f ( a ) + f' ( a ) ( x – a ) + ------------------------------- + … + ------------------------------------------------
2! ( n – 1 )!

34.6.5 Practice Problems

1. Find the derivative of the function below with respect to time.

3t
------------------ + e 2t
2
(2 + t)

(ans.  d   3t
----- ------------------ + e 2t =  ----
d-
( 3t ( 2 + t ) ) +  ----- ( e ) = 3 ( 2 + t ) – 6t ( 2 + t ) + 2e
–2 d 2t –2 –3 2t
 dt  2   dt  dt
(2 + t)

2. Solve the following differential equation, given the initial conditions at t=0s.
x'' + 4x' + 4x = 5t x0 = 0 x' 0 = 0
 math guide - 34.88

(ans. homogeneous solution:

x'' + 4x' + 4x = 0
guess: At At 2 At
xh = e x h' = Ae x h'' = A e
2 At At At 2
A e + 4Ae + 4e = 0 = A + 4A + 4 = ( A + 2 ) ( A + 2 )
– 2t – 2t
xh = C1 e + C 2 te
particular solution:
guess: x p = At + B x p' = A x p'' = 0
( 0 ) + 4 ( A ) + 4 ( At + B ) = 5t
( 4A + 4B ) + ( 4A )t = ( 0 ) + ( 5 )t
5
A = --- = 1.25 4 ( 1.25 ) + 4B = 0 B = – 1.25
4
x p = 1.25t – 1.25
combine and solve for constants:
– 2t – 2t
x ( t ) = xh + xp = C1 e + C 2 te + 1.25t – 1.25
for x(0) = 0 –2 ( 0 ) –2 ( 0 )
0 = C1 e + C 2 ( 0 )e + 1.25 ( 0 ) – 1.25 = C 1 – 1.25
C 1 = 1.25
–2 ( 0 ) –2( 0) –2 ( 0 )
for (d/dt)x(0) = 0 0 = ( – 2 )C 1 e + ( – 2 )C 2 ( 0 )e + C2 e + 1.25
– 2 C 1 + C 2 + 1.25 = 0 = –2 ( 1.25 ) + C 2 + 1.25 C 2 = 1.25
2t 2t
x ( t ) = 1.25e + 1.25te + 1.25t – 1.25

3. Find the following derivatives.

d
a) ----- ( sin t + cos t )
dt
d –2
b) ----- ( ( t + 2 ) )
dt
d 8t
c) ----- ( 5te )
dt
d
d) ----- ( 5 ln t )
dt
 math guide - 34.89

(ans.
d d d
a) ----- ( sin t + cos t ) = ----- ( sin t ) + ----- ( cos t ) = cos t – sin t
dt dt dt
d –2 –3
b) ----- ( ( t + 2 ) ) = –2 ( t + 2 )
dt
d 8t 8t 8t
c) ----- ( 5te ) = 5e + 40te
dt
d 5
d) ----- ( 5 ln t ) = ---
dt t

4. Find the following integrals

∫ 6 t dt
2
a)
b) ∫ 14 e dt
7t

c) ∫ sin ( 0.5t ) dt
5
d)
∫ --x- dx

(ans.
3
a) ∫ 6 t dt = 6  ---- = 2t + C
2 t 3
3
7t
b) ∫ 14 e dt = 14  ------ + C = 2e + C
7t e 7t
7
– cos ( 0.5t )
c) ∫ sin ( 0.5t ) dt = -------------------------- + C = – 2 cos ( 0.5t ) + C
0.5
5
d)
∫ --x- dx = 5 ln ( x ) + C

5. Find the following derivative.

d- 4t –3
---- ( 5te + ( t + 4 ) )
dt
 math guide - 34.90

6. Find the following derivatives.

d-  -----------
----- 1 -
a)
dx  x + 1

d- –t
b) ---- ( e sin ( 2t – 4 ) )
dt

(ans. d-  -----------
1 - –1
a) ----- = ------------------2-

dx x + 1 
(x + 1)

d
b) ----- ( e –t sin ( 2t – 4 ) ) = – e –t sin ( 2t – 4 ) + 2e –t cos ( 2t – 4 )
dt

7. Solve the following integrals.

∫e
a) 2t
dt

b)
∫ ( sin θ + cos 3θ ) dθ

(ans.
∫e
a) 2t 2t
dt = 0.5e + C

1
b)
∫ ( sin θ + cos 3θ ) dθ = – cos θ + --- sin 3θ + C
3

8. Solve the following differential equation.

x'' + 5x' + 3x = 3 x( 0) = 1
x' ( 0 ) = 1
 math guide - 34.91

(ans.
x'' + 5x' + 3x = 3 x( 0) = 1 x' ( 0 ) = 1

Homogeneous:
2
A + 5A + 4 = 0
– 5 ± 25 – 16 –5±3
A = ------------------------------------ = ---------------- = – 4, – 1
2 2
– 4t –t
xh = C1 e + C2e
Particular:
xp = A x'p = 0 x'' p = 0

0 + 5 ( 0 ) + 3A = 3 A = 1

xp = 1
Initial values:
– 4t –t
x = x h + xp = C1 e + C2 e + 1

1 = C1( 1 ) + C2( 1 ) + 1 C1 = –C2

– 4t –t
x' = – 4C 1 e – C2e
1 = – 4C 1 ( 1 ) – C 2 ( 1 )
1
4 ( – C2 ) + C2 = –1 C 2 = ---
3
1
C 1 = – ---
1 – 4t 1 –t 3
x = – --- e + --- e + 1
3 3

9. Set up an integral and solve it to find the volume inside the shape below. The shape is basically
a cone with the top cut off.
 math guide - 34.92

z
y
b

a
c

(ans. a
V = ∫ dV = ∫0 A dx
r = b +  ----------- x
c–b
a
2 2
A = πr = π  b +  ----------- x = πb + 2π  ----------- x + π  ----------- x
2 c–b 2 c–b c–b 2
  a    a   a 
a 2
∫0  πb + 2π  ----------
c – b-
x + π  ----------- x  dx
2 c–b 2
V =
a  a
a
c–b 2 π c–b 3 2
V = πb x + π  ----------- x + ---  ----------- x
2
a 3 a
0

c–b 2 π c–b 2 3
V = πb a + π  ----------- a + ---  ----------- a
2
a 3 a
2 π 2 2
V = πb a + π ( c – b )a + --- ( c – b ) a
3

10. Solve the first order non-homogeneous differential equation below. Assume the system starts
at rest.
2x' + 4x = 5 sin 4t

11. Solve the second order non-homogeneous differential equation below.

2x'' + 4x' + 2x = 5 where, x( 0) = 2

x' ( 0 ) = 0
 math guide - 34.93

34.7 NUMERICAL METHODS

• These techniques approximate system responses without doing integrations, etc.

34.7.1 Approximation of Integrals and Derivatives from Sampled Data

• This form of integration is done numerically - this means by doing repeated cal-
culations to solve the equation. Numerical techniques are not as elegant as solving differ-
ential equations, and will result in small errors. But these techniques make it possible to
solve complex problems much faster.

• This method uses forward/backward differences to estimate derivatives or inte-

grals from measured data.

y(t)

yi + 1
yi
yi – 1

ti – 1 T T
ti ti + 1

ti
 y--------------------
i + y i – 1 T
∫ti – 1 i  2 - ( ti – ti – 1 ) = --2- ( y i + yi – 1 )
y ( t ) ≈

y i – y i – 1 y i + 1 – y i
y ( t i ) ≈  --------------------
- =  --------------------
d-
---- - = --1- ( y i – y i – 1 ) = --1- ( y i + 1 – y i )
dt  ti – ti – 1   t i + 1 – ti  T T

--1- ( y i + 1 – y i ) – --- ( y i – y i – 1 )
1
 ----
d- 2 T T –2 yi + yi – 1 + y i + 1
 dt y ( t i ) ≈ ----------------------------------------------------------------- = ---------------------------------------------
2
T T
 math guide - 34.94

34.7.2 Euler First-order Integration

• We can also estimate the change resulting from a derivative using Euler’s equa-
tion for a first-order difference equation.

d
y ( t + h ) ≈ y ( t ) + h ----- y ( t )
dt

34.7.3 Taylor Series Integration

• Recall the basic Taylor series,

1 2 d 2 1 3 d 3 1 4 d 4
x ( t + h ) = x ( t ) + h  ----- x ( t ) + ----- h  ----- x ( t ) + ----- h  ----- x ( t ) + ----- h  ----- x ( t ) + …
d
 dt 2!  dt 3!  dt 4!  dt

• When h=0 this is called a MacLaurin series.

• We can integrate a function by,

 ----
d- 2 3  ----
d-
 dt x 0
x = 1+x +t = 0
 dt
x0 = 0

t (s) x(t) d/dt x(t) h = 0.1

0 0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
 math guide - 34.95

34.7.4 Runge-Kutta Integration

• The equations below are for calculating a fourth order Runge-Kutta integration.

1
x ( t + h ) = x ( t ) + --- ( F 1 + 2F 2 + 2F 3 + F4 )
6
F 1 = hf ( t, x )

F1
F 2 = hf  t + ---, x + ------
h
2 2
F2
F 3 = hf  t + ---, x + ------
h
 2 2

F 4 = hf ( t + h, x + F 3 )

where,

x = the state variables

f = the differential function
t = current point in time
h = the time step to the next integration point

34.7.5 Newton-Raphson to Find Roots

• When given an equation where an algebraic solution is not feasible, a numerical

solution may be required. One simple technique uses an instantaneous slope of the func-
tion, and takes iterative steps towards a solution.

f ( xi )
x i + 1 = x i – -----------------------
 -----
d-
f ( x )
 dx i 

• The function f(x) is supplied by the user.

 math guide - 34.96

• This method can become divergent if the function has an inflection point near the
root.

• The technique is also sensitive to the initial guess.

• This calculation should be repeated until the final solution is found.

34.8 LAPLACE TRANSFORMS

• The Laplace transform allows us to reverse time. And, as you recall from before
the inverse of time is frequency. Because we are normally concerned with response, the
Laplace transform is much more useful in system analysis.

• The basic Laplace transform equations is shown below,

∞
∫0 f ( t )e
– st
F(s) = dt

where,
f ( t ) = the function in terms of time t
F ( s ) = the function in terms of the Laplace s

34.8.1 Laplace Transform Tables

• Basic Laplace Transforms for operational transformations are given below,

 math guide - 34.97

TIME DOMAIN FREQUENCY DOMAIN

Kf ( t ) Kf ( s )

f1 ( t ) + f2 ( t ) –f3 ( t ) + … f1 ( s ) + f2 ( s ) –f 3 ( s ) + …

df (t) –
----------- sf ( s ) – f ( 0 )
dt
2 –
f ( t )-
d------------- 2 – df ( 0 )
2
s f ( s ) – sf ( 0 ) – ------------------
dt dt
n – n –
f ( t )- n – 2 df ( 0
-----------------)- – … – d-------------------
d------------- n n–1 – f ( 0 )-
n
s f( s) – s f(0 ) – s n
dt dt dt
t
( s -)
∫0 f ( t ) dt f--------
s
– as
f ( t – a )u ( t – a ), a > 0 e f( s)
– at
e f(t) f(s – a)

f ( at ), a > 0 1  s
--- f ---
a  a

df ( s )-
–--------------
tf ( t )
ds
n
n f ( s -)
nd
t f(t) ( – 1 ) --------------
n
ds
∞
( t )-
f-------
t ∫ f ( u ) du
s

• A set of useful functional Laplace transforms are given below,

 math guide - 34.98

TIME DOMAIN FREQUENCY DOMAIN

A
A ---
s
1
t ----
2
s
– at 1-
e ----------
s+a
sin ( ωt ) ω
-----------------
2 2
s +ω
s
cos ( ωt ) -----------------
2 2
s +ω
– at 1
te -------------------
2
(s + a)
– at ω
e sin ( ωt ) -------------------------------
2 2
-
(s + a) + ω
– at s+a -
e cos ( ωt ) -------------------------------
2 2
(s + a) + ω
– at A-
Ae ----------
s–a
– at A -
Ate -----------------
2
(s – a)

– αt complex conjugate
2Ae cos ( βt + θ ) A A
----------------------- + --------------------------------------
s + α – βj s + α + βj

– αt complex conjugate
2t A e cos ( βt + θ ) A
------------------------------ A
- + -------------------------------------
-
2 2
( s + α – βj ) ( s + α + βj )

• Laplace transforms can be used to solve differential equations.

 math guide - 34.99

34.9 z-TRANSFORMS

• For a discrete-time signal x [ n ] , the two-sided z-transform is defined by

∞ ∞

∑ ∑ x [ n ]z
–n –n
X(z) = x [ n ]z . The one-sided z-transform is defined by X ( z ) = . In
n = –∞ n=0
both cases, the z-transform is a polynomial in the complex variable z .

• The inverse z-transform is obtained by contour integration in the complex plane

1
x [ n ] = -------- ∫ X ( z )z
n–1
j2π°
dz . This is usually avoided by partial fraction inversion tech-
niques, similar to the Laplace transform.

• Along with a z-transform we associate its region of convergence (or ROC). These
are the values of z for which X ( z ) is bounded (i.e., of finite magnitude).
 math guide - 34.100

• Some common z-transforms are shown below.

Table 1: Common z-transforms

Signal z-Transform
ROC
x[ n] X(z)

δ[n] 1 All z

1 -
u[n] ---------------
–1 z >1
1–z
–1
z
nu [ n ] ----------------------
- z >1
–1 2
(1 – z )
–1 –1
( 1 + z )-
z----------------------------
2
n u[n ] z >1
–1 3
(1 – z )

1 -
n
a u[n ] ------------------
–1 z > a
1 – az
–1
az
n
na u [ n ] -------------------------- z > a
–1 2
( 1 – az )

1 -
n
( – a )u [ – n – 1 ] ------------------
–1 z < a
1 – az
–1
az
n
( – na )u [ – n – 1 ] -------------------------- z < a
–1 2
( 1 – az )
–1
1 – z cos ω 0
cos ( ω 0 n ) u [ n ] ------------------------------------------------
- z >1
–1 –2
1 – 2z cos ω 0 + z

–1
z sin ω 0
sin ( ω 0 n ) u [ n ] ------------------------------------------------
- z >1
–1 –2
1 – 2z cos ω 0 + z

–1
1 – az cos ω 0
n
a cos ( ω 0 n ) u [ n ] ---------------------------------------------------------
–1 2 –2
- z > a
1 – 2az cos ω 0 + a z
 math guide - 34.101

Table 1: Common z-transforms

Signal z-Transform
ROC
x[ n] X(z)

–1
az sin ω 0
n
a sin ( ω 0 n ) u [ n ] ---------------------------------------------------------
–1 2 –2
- z > a
1 – 2az cos ω 0 + a z

–k
n! - z
---------------------- u[n] -----------------------------
- z >1
k! ( n – k )! (1 – z )
–1 k + 1

• The z-transform also has various properties that are useful. The table below lists
properties for the two-sided z-transform. The one-sided z-transform properties can be
derived from the ones below by considering the signal x [ n ]u [ n ] instead of simply x [ n ] .

Table 2: Two-sided z-Transform Properties

Property Time Domain z-Domain ROC

Notation x[ n] X(z) r2 < z < r 1

x1 [ n ] X1 ( z ) ROC 1
x2 [ n ] X2 ( z ) ROC 2

Linearity αx 1 [ n ] + βx 2 [ n ] αX 1 ( z ) + βX 2 ( z ) At least the intersec-

tion of ROC 1 and
ROC 2

Time Shifting x[ n – k] –k
z X( z) That of X ( z ) , except
z = 0 if k > 0 and
z = ∞ if k < 0
z-Domain Scaling n
a x[ n] X(a z)
–1 a r2 < z < a r1

Time Reversal x [ –n ] X(z )

–1 1 1
---- < z < ----
r1 r2

z-Domain nx [ n ] dX ( z ) r2 < z < r 1

Differentiation – z --------------
dz
 math guide - 34.102

Table 2: Two-sided z-Transform Properties

Property Time Domain z-Domain ROC

Convolution x 1 [ n ]*x 2 [ n ] X 1 ( z )X 2 ( z ) At least the intersec-

tion of ROC 1 and
ROC 2

Multiplication x 1 [ n ]x 2 [ n ] At least
X 1 ( v )X 2  -- v dv
1- z –1
j2π°
------- ∫  v r 1l r 2l < z < r 1u r 2u

Initial value theo- x [ n ] causal x [ 0 ] = lim X ( z )

rem z→∞

34.10 FOURIER SERIES

• These series describe functions by their frequency spectrum content. For example
a square wave can be approximated with a sum of a series of sine waves with varying
magnitudes.

• The basic definition of the Fourier series is given below.

∞
a0
a n cos  --------- + b n sin  ---------
nπx nπx
f ( x ) = ----- +
2 ∑  L   L 
n=1

1 L 1 L
a n = --- ∫ f ( x ) cos  --------- dx b n = --- ∫ f ( x ) sin  --------- dx
nπx nπx
L –L L L –L L

34.11 TOPICS NOT COVERED (YET)

• To ensure that the omissions are obvious, I provide a list of topics not covered
below. Some of these may be added later if their need becomes obvious.

• Frequency domain - Fourier, Bessel

 math guide - 34.103

34.12 REFERENCES/BIBLIOGRAPHY

Spiegel, M. R., Mathematical Handbook of Formulas and Tables, Schaum’s Outline Series,
McGraw-Hill Book Company, 1968.
 C programming - 35.1

35. A BASIC INTRODUCTION TO ‘C’

35.1 WHY USE ‘C’?

• ‘C’ is commonly used to produce operating systems and commercial software.

Some examples of these are UNIX, Lotus-123, dBase, and some ‘C’ compilers.

• Machine Portable, which means that it requires only small changes to run on
other computers.

• Very Fast, almost as fast as assembler.

• Emphasizes structured programming, by focusing on functions and subroutines.

• You may easily customize ’C’ to your own needs.

• Suited to Large and Complex Programs.

 C programming - 35.2

• Very Flexible, allows you to create your own functions.

35.2 BACKGROUND

• Developed at Bell Laboratories in the Early 70’s, and commercial compilers

became available in the Late 70’s.Recnetly has become more popular because of its ties to
UNIX. Over 90% of UNIX is written in ‘C’. AT&T originally developed ‘C’ with the
intention of making it an in-house standard.

35.3 PROGRAM PARTS

• /* is the start of a comment.

• */ is the end of comment.

• The main program is treated like a function, and thus it has the name main().

• lower/UPPER case is crucial, and can never be ignored.

 C programming - 35.3

• Statements are separated by semi-colons ‘;’

• Statements consist of one operation, or a set of statements between curly brackets

{, }

• There are no line numbers.

Program to Add two Numbers:

/* A simple program to add two numbers and print the results */

main()
{
int x, y = 2, z; /* define three variables and give one a value */
x = 3; /* give another variable a value */
z = x + y; /* add the two variables */
printf(“%d + %d = %d\n”, x, y, z); /*print the results */
}

Results (output):

• lines may be of any length.

• A very common function in ‘C’ is printf(). This function will do a formatted

print. The format is the first thing which appears between the brackets. In this case the for-
mat says print an integer %d followed by a space then a ‘+’ then another space, another
integer, another space, ‘=’, and another space, another integer, then a line feed ‘\n’. All
variables that follow the format statement are those to be printed. x, y, and z are the three
integers to be printed, in their respective orders.
 C programming - 35.4

• Major Data Types for variables and functions are (for IBM PC):

int (2 byte integer),

short (1 byte integer),
long (4 byte integer),
char (1 byte integer),
float (4 byte IEEE floating point standard),
double (8 byte IEEE floating point standard).

•int, short, long, char can be modified by the addition of unsigned, and register.
An unsigned integer will not use 1 bit for number sign. A register variable will use a data
register in the microprocessor, if possible, and it will speed things up (this is only available
for integers).

Example of Defining Different Data Types:

main()
{
unsigned int i;
register j;
short k;
char l;
double m;
etc

• A function consists of a sub-routine or program, which has been assigned a name.

This function is capable of accepting an argument list, and returning a single value. The
function must be defined before it is called from within the program. (e.g. sin() and
read()).
 C programming - 35.5

Program to add numbers with a function:

/* A simple program to add two numbers and print the results */

int add(); /* Declare a integer function called ‘add’ */

main()
{
int x = 3, y = 2, z; /* define three variables and give values */
z = add(x, y); /* pass the two values to ‘add’ and get the sum*/
printf(“%d + %d = %d\n”, x, y, z); /*print the results */
}

int add(a, b) /* define function and variable list */

int a, b; /* describe types of variable lists */
{
int c; /* define a work integer */
c = a + b; /* add the numbers */
return(c); /* Return the number to the calling program */

• Every variable has a scope. This determines which functions are able to use that
variable. If a variable is global, then it may be used by any function. These can be modi-
fied by the addition of static, extern and auto. If a variable is defined in a function, then it
will be local to that function, and is not used by any other function. If the variable needs to
be initialized every time the subroutine is called, this is an auto type. static variables can
be used for a variable that must keep the value it had the last time the function was called.
Using extern will allow the variable types from other parts of the program to be used in a
function.
 C programming - 35.6

Program example using global variables:

/* A simple program to add two numbers and print the results */

int x = 3, /* Define global x and y values */

y = 2,
add(); /* Declare an integer function called ‘add’ */

main()
{
printf(“%d + %d = %d\n”, x, y, add()); /*print the results */
}

int add() /* define function */

{
return(x + y); /* Return the sumto the calling program */

• Other variable types of variables are union, enum, struct, etc.

• Some basic control flow statements are while(), do-while(), for(), switch(), and
if(). A couple of example programs are given below which demonstrate all the ’C’ flow
statements.

Program example with a for loop:

/* A simple program toprint numbers from 1 to 5*/

main()
{
int i;
for(i = 1; i <= 5; i = i + 1){
printf(“number %d \n”, i); /*print the number */
}
}
 C programming - 35.7

or example with a while loop:

main()
{
int i = 1;
while(i <= 5){
printf(“number %d \n”, i);
i = i + 1;
}
}

or example with a do while loop

main()
{
int i = 1;
do{
printf(“number %d \n”, i);
i = i + 1;
}while(i <= 5)
}

or example with a do until loop:

main()
{
int i = 1;
do{
printf(“number %d \n”, i);
i = i + 1;
}until(i > 5)
}
 C programming - 35.8

Example Program with an if else:

main()
{
int x = 2, y = 3;
if(x > y){
printf(“Maximum is %d \n”, x);
} else if(y > x){
printf(“Maximum is %d \n”, y);
} else {
printf(“Both values are %d \n”, x);
}
}

Example Program using switch-case:

main()
{
int x = 3; /* Number of People in Family */
switch(x){ /* choose the numerical switch */
case 0: /* Nobody */
printf(“There is no family \n”);
break;
case 1: /* Only one person, but a start */
printf(“There is one parent\n”);
break;
case 2: /* You need two to start something */
printf(“There are two parents\n”);
break;
default: /* critical mass */
printf(“There are two parents and %d kids\n”, x-2);
break;
}
}
 C programming - 35.9

• #include <filename.h> will insert the file named filename.h into the program.
The *.h extension is used to indicate a header file which contains ‘C’ code to define func-
tions and constants. This almost always includes “stdio.h”. As we saw before, a function
must be defined (as with the ‘add’ function). We did not define printf() before we used it,
this is normally done by using #include <stdio.h> at the top of your programs. “stdio.h”
contains a line which says ‘int printf();’. If we needed to use a math function like y =
sin(x) we would have to also use #include <math.h>, or else the compiler would not know
what type of value that sin() is supposed to return.

•#define CONSTANT TEXT will do a direct replacement of CONSTANT in the pro-

gram with TEXT, before compilation. #undef CONSTANT will undefine the CONSTANT.

A Sample Program to Print Some sin() values

(using defined constatnts)

#include “stdio.h”
#include “math.h”
#define TWO_PI 6.283185307
#define STEPS 5

main()
{
double x; /* Current x value*/

for(x = 0.0; x <= TWO_PI; x = x + (TWO_PI / STEPS)){

printf(“%f = sin(%f) \n”, sin(x), x);
}
}

• #ifdef, #ifndef, #if, #else and #else can be used to conditionally include parts of a
program. This is use for including and eliminating debugging lines in a program.

• #define, #include, #ifdef, #ifndef, #if, #else, /* and */ are all handled by the Pre-
processor, before the compiler touches the program.
 C programming - 35.10

• Matrices are defined as shown in the example. In ‘C’ there are no limits to the
matrix size, or dimensions. Arrays may be any data type. Strings are stored as arrays of
characters.

• i++ is the same as i = i + 1.

A Sample Program to Get a String

Then Print its ASCII Values (with matrix)

#include “stdio.h”
#define STRING_LENGTH 5

main()
{
int i;
char string[STRING_LENGTH]; /* character array */
gets(string); /* Input string from keyboard */
for(i = 0; i < STRING_LENGTH; i++){
printf(“pos %d, char %c, ASCII %d \n”, i, string[i], string[i]);
}
}

INPUT:
HUGH<return>

OUTPUT:
pos 0, char H, ASCII 72
pos 0, char U, ASCII 85
pos 0, char G, ASCII 71
pos 0, char H, ASCII 72
pos 0, char , ASCII 0
 C programming - 35.11

• Pointers are a very unique feature of ‘C’. First recall that each variable uses a real
location in memory. The computer remembers where the location of that variable is, this
memory of location is called a pointer. This pointer is always hidden from the program-
mer, and uses it only in the background. In ‘C’, the pointer to a variable may be used. We
may use some of the operations of ‘C’ to get the variable that the pointer, points to. This
allows us to deal with variables in a very powerful way.

A Sample Program to Get a String

Then Print its ASCII Values (with pointers):

#include “stdio.h”

main()
{
int i;
char *string; /* character pointer */
gets(string); /* Input string from keyboard */
for(i = 0; string[i] != 0; i++){
printf(“ pos %d, char %c, ASCII %d \n”, i, string[i], string[i]);
}
}

INPUT:
HUGH<return>

OUTPUT:
pos 0, char H, ASCII 72
pos 0, char U, ASCII 85
pos 0, char G, ASCII 71
pos 0, char H, ASCII 72

35.4 HOW A ‘C’ COMPILER WORKS

• A ‘C’ compiler has three basic components: Preprocessor, First and Second Pass
 C programming - 35.12

Compiler, and Linker.

Source code “filename.c”

The Preprocessor

Will remove comments, replace strings

#include files which have a defined value, include programs,
(like “stdio.h”) and remove unneeded characters.

ASCII Text Code

The First and Second Pass

The compiler will parse the program and

check the syntax. TheSecond Pass produces
some simple machine language, which performs
the basic functions of the program.

Object Code (*.obj)

The Linker

The compiler will combine the basic

machine language from the first pass and com-
Library files bine it with the pieces of machine language in
(*.lib) the compiler libraries. An optimization opera-
tion may be used to reduce execution time.

Executable Code
(*.exe)
 C programming - 35.13

35.5 STRUCTURED ‘C’ CODE

• A key to well designed and understandable programs.

• Use indents, spaces and blank lines, to make the program look less cluttered, and
give it a block style.

• Comments are essential to clarify various program parts.

• Descriptive variable names, and defined constants make the purpose of the vari-
able obvious.

• All declarations for the program should be made at the top of the program listing.
 C programming - 35.14

A Sample of a Bad Program Structure:

main(){int i;for(;i<10;i++)printf(“age:%d\n”,i);}

A Good Example of the same Program:

#include <stdio.h>
#define COUNT 10 /* Number of counts in loop */

main()
{
int i; /* counter */
for(i = 0; i < COUNT; i++){ /* loop to print numbers */
printf(“age:%d\n”, i);
}
exit(0);
}

35.6 ARCHITECTURE OF ‘C’ PROGRAMS (TOP-DOWN)

35.6.1 How?

• A program should be broken into fundamental parts (using functions for each
part) and then assembled using functions. Each function consists of programs written
using the previous simpler functions.
 C programming - 35.15

Example with a Car Frame is like one subroutine

which also calls other
subroutines like Suspension
Frame

Suspension Body Engine

Wheel

• A Clear division should be maintained between program levels.

• Never use goto’s, they are a major source of logic errors. Functions are much eas-
ier to use, once written.

• Try to isolate machine specific commands (like graphics) into a few functions.

35.6.2 Why?

• A top-down design allows modules to be tested as they are completed. It is much

easier to find an error in a few lines of code, than in a complete program.

• When programs are complete, errors tend to be associated with modules, and are
thus much easier to locate.
 C programming - 35.16

• Updates to programs are much easier, when we only need to change one function.

• It is just as easy to change the overall flow of a program, as it is to change a func-

tion.

Application of ‘C’ to a CAD Program

35.7 CREATING TOP DOWN PROGRAMS

1. Define Objectives - Make a written description of what the program is expected

to do.

2. Define Problem - Write out the relevant theory. This description should include
variables, calculations and figures, which are necessary for a complete solution to the
problem. From this we make a list of required data (inputs) and necessary results (output).

3. Design User Interface - The layout of the screen(s) must be done on paper. The
method of data entry must also be considered. User options and help are also considered
here. (There are numerous factors to be considered at this stage, as outlined in the course
notes.)

4. Write Flow Program - This is the main code that decides when general opera-
tions occur. This is the most abstract part of the program, and is written calling dummy
‘program stubs’.
 C programming - 35.17

5. Expand Program - The dummy ‘stubs’ are now individually written as func-
tions. These functions will call another set of dummy ‘program stubs’. This continues
until all of the stubs are completed. After the completion of any new function, the program
is compiled, tested and debugged.

6. Testing and Debugging- The program operation is tested, and checked to make
sure that it meets the objectives. If any bugs are encountered, then the program is revised,
and then retested.

7. Document - At this stage, the operation of the program is formally described.

For Programmers, a top-down diagram can be drawn, and a written description of func-
tions should also be given.

Golden Rule: If you are unsure how to proceed when writing a program, then work
out the problem on paper, before you commit yourself to your programmed solution.

Note: Always consider the basic elements of Software Engineering, as outlined in

the ES488 course notes.

35.8 HOW THE BEAMCAD PROGRAM WAS DESIGNED

 C programming - 35.18

35.8.1 Objectives:

• The program is expected to aid the design of beams by taking basic information
about beam geometry and material, and then providing immediate feedback. The beam
will be simply supported, and be under a single point load. The program should also pro-
vide a printed report on the beam.

35.8.2 Problem Definition:

• The basic theory for beam design is available in any good mechanical design
textbook. In this example it will not be given.

• The inputs were determined to be few in number: Beam Type, Beam Material,
Beam Thickness, Beam Width, Beam Height, Beam Length, Load Position, Load Force.

• The possible outputs are Cross Section Area, Weight, Axial Stiffness, Bending
Stiffness, and Beam Deflection, a visual display of Beam Geometry, a display of Beam
Deflection.

35.8.3 User Interface:

35.8.3.1 - Screen Layout (also see figure):

 C programming - 35.19

• The small number of inputs and outputs could all be displayed, and updated, on a
single screen.

• The left side of the screen was for inputs, the right side for outputs.

• The screen is divided into regions for input(2), input display and prompts(1),
Beam Cross section(3), Numerical Results(4), and Beam Deflection(5).

35.8.3.2 - Input:

• Current Inputs were indicated by placing a box around the item on the display(1).

• In a separate Prompt Box(2), this input could be made.

• The cursor keys could be used to cursor the input selector up or down.

• Single keystroke operation.

• Keys required: UP/DOWN Cursors, F1, F2, F4, numbers from ‘0’ to ‘9’, ‘.’, ‘-’,
 C programming - 35.20

and <RETURN>. In the spirit of robustness it was decided to screen all other keys.

35.8.3.3 - Output:

• Equations, calculations, material types, and other relevant information were

obtained from a text.

• Proper textual descriptions were used to ensure clarity for the user.

• For a printed report, screen information would be printed to a printer, with the
prompt area replaced with the date and time.

35.8.3.4 - Help:

• A special set of help information was needed. It was decided to ensure that the
screen always displays all information necessary(2).

35.8.3.5 - Error Checking:

• Reject any input which violates the input limits.

• A default design was given, which the user could modify.

 C programming - 35.21

• An error checking program was created, which gives error messages.

35.8.3.6 - Miscellaneous:

• The screen was expressed in normalized coordinates by most sub-routines.

• Colors were used to draw attention, and highlight areas.

 C programming - 35.22

35.8.4 Flow Program:

main()
/*
* EXECUTIVE CONTROL LEVEL
*
* This is the main terminal point between the
* various stages of setup, input, revision
* and termination.
*
* January 29th, 1989.
*/
{
static int error;

if((error = setup()) != ERROR) {

screen(NEW);
screen(UPDATE);
while((error = input()) != DONE) {
if(error == REVISED) {

screen(NEW);

screen(UPDATE);
}
}
error = NO_ERROR;
}
kill();
if(error == ERROR) {
printf(“EGA Graphics Driver Not Installed”);

35.8.5 Expand Program:

 C programming - 35.23

• The routines were written in a top down fashion, in a time of about 30 hours.
These routines are listed below.

Routines Used In Package:

• main() - to be used as the main program junction.

• setup() - to set up graphics mode and printer.

• screen() - A function to draw, or refresh part of the screen. In the interest of

program speed, this function uses some low level commands.

• calculations() - perform the calculations of outputs from the inputs

• picture() - draws the beam cross section and deflection of beam. For the sake
of speed, this section will use low level commands.

• input() - A function which controls the main input loop for numbers, con-

• Condition and error flags were used to skip unnecessary operations, and thus
speed up response. A response of more than 0.5 seconds will result in loss of attention by
the user.
 C programming - 35.24

High Level Low Level

Executive Subroutine Specific Subroutines

Top Down

main() screen() box() setup()

input() printer() kill()

picture()
In this case we see that most of the routines
are at the bottom of the design tree. This structure
shows a clear division of tasks, to their basic parts. enter()
On the above diagram, none of the functions calls
any of the functions to the left of it, only to the
right. In this case main() will call setup(), screen(), draw_line()
input() and kill() directly.
text()

calculations()

printes()
Machine Dependence Increases
Consideration of detail
Consideration of flow

35.8.6 Testing and Debugging:

 C programming - 35.25

• The testing and debugging was very fast, with only realignment of graphics being
required. This took a couple of hours.

35.8.7 Documentation

35.8.7.1 - Users Manual:

• The documentation included an Executive Summary of what the Program does.

• The Objectives of the program were described.

• The theory for beam design was given for the reference of any program user, who
wanted to verify the theory, and possible use it.

• A manual was given which described key layouts, screen layout, basic sequence
of operations, inputs and outputs.

• Program Specifications were also given.

• A walk through manual was given. This allowed the user to follow an example
which displayed all aspects of the program.
 C programming - 35.26

35.8.7.2 - Programmers Manual:

• Design Strategy was outlined and given.

• A complete program listing was given (with complete comments).

• Complete production of this Documentation took about 6 hours.

35.8.8 Listing of BeamCAD Program.

• Written for turbo ‘C’

35.9 PRACTICE PROBLEMS

1. What are the basic components of a ‘C’ compiler, and what do they do?
 C programming - 35.27

2. You have been asked to design a CAD program which will choose a bolt and a
nut to hold two pieces of sheet metal together. Each piece of sheet metal will have a hole
drilled in it that is the size of the screw. You are required to consider that the two pieces are
experiencing a single force. State your assumptions about the problem, then describe how
you would produce this program with a Top Down design.

3. What are some reasons for using ‘C’ as a programming language?

4. Describe some of the reasons for Using Top-Down Design, and how to do it.
 units - 36.1

36. UNITS AND CONVERSIONS

• Units are essential when describing real things.

• Good engineering practice demands that each number should always be accom-
panied with a unit.

36.1 HOW TO USE UNITS

• This table does not give an exhaustive list of conversion factors, but instead a
minimal (but fairly complete) set is given. From the values below any conversion value
can be derived. If you are not sure about this, ask the instructor to show you how.

• A simple example of unit conversion is given below,

**a simple unit conversion example:

Given,
·
d x = 10m d y = 5ft

Find the distance ‘d’,

keep the units in the equation
2 2
d = dx + dy

2 2
∴d = ( 10m ) + ( 5ft ) From the tables
multiply by 1
2 2
∴d = 100m + 25ft 1ft = 0.3048m

2 0.3048m 2 0.3048m
100m + 25ft  --------------------- ∴1 = ---------------------
2
∴d =
 1ft  1ft
2
2 2 m
∴d = 100m + 25ft ( 0.092903 ) ------2
ft
cancel out units
2 2
∴d = 100m + 25 ( 0.092903 )m

2
∴d = 102.32m = 10.12m
 units - 36.2

36.2 HOW TO USE SI UNITS

1. Beware upper/lower case letter in many cases they can change meanings.

e.g. m = milli, mega

2. Try to move prefixes out of the denominator of the units.

e.g., N/cm or KN/m

3. Use a slash or exponents.

e.g., (kg•m/s2) or (kg•m•s-2)

4. Use a dot in compound units.

e.g., N•m

5. Use spaces to divide digits when there are more than 5 figures, commas are
avoided because their use is equivalent to decimal points in some cultures.

• In some cases units are non-standard. There are two major variations US units are
marked with ‘US’ and Imperial units are marked with ‘IMP’.

36.3 THE TABLE

Major Division

Distance
1 ft. (feet) = 12 in. (inches) = 0.3048 m (meter)
1 mile = 1760 yards = 5280 ft = 1.609km
1 in.(inch) = 2.540 cm
1 yd (yard) = 3 ft.
1 nautical mile = 6080 ft. = 1852 m = 1.150782 mi
1 micron = 10-6 m
1 angstrom = 10-10 m
1 mil = 10-6 m
1 acre = 43,560 ft. = 0.4047 hectares
 units - 36.3

1 furlong = 660 ft
1 lightyear = 9.460528e15 m
1 parsec = 3.085678e16 m

Area
1 acre = 43,559.66 ft2
1 Hectare (ha) = 10,000 m2
1 Hectare (ha) = 10,000 m2
1 Hectare (ha) = 10,000 m2
1 Hectare (ha) = 10,000 m2

Velocity
1 mph = 0.8689762 knot

Angle
1 rev = 2PI radians = 360 degrees = 400 gradians
1 degree = 60 minutes
1 minute = 60 seconds

Volume
1 US gallon = 231 in3
1 CC = 1 cm3
1 IMP gallon = 277.274 in3
1 barrel = 31 IMP gal. = 31.5 US gal.
1 US gal. = 3.785 l = 4 quarts = 8 pints = 16 cups
1 liter (l) = 0.001 m3 = 2.1 pints (pt) = 1.06 quarts (qt) = 0.26 gallons (gal)
1 qt (quart) = 0.9464 l
1 cup (c) = 0.2365882 l = 8 USoz
1 US oz = 8 U.S. drams = 456.0129 drops = 480 US minim = 1.040842
IMP oz
= 2 tablespoons = 6 teaspoons
1 IMP gal. = 1.201 U.S. gal.
1 US pint = 16 US oz
1 IMP pint = 20 IMP oz
1tablespoon = 0.5 oz.
1 bushel = 32 quarts
1 peck = 8 quarts

Force/Mass
1 N (newton) = 1 kg•m/s2 = 100,000 dyne
1 dyne = 2.248*10-6 lb. (pound)
1 kg = 9.81 N (on earth surface) = 2.2046 lb
1lb = 16 oz. (ounce) = 4.448N
 units - 36.4

1 oz. = 28.35 g (gram) = 0.2780N

1 lb = 0.03108 slug
1 kip = 1000 lb.
1 slug = 14.59 kg
1 imperial ton = 2000 lb = 907.2 kg
1 metric tonne = 1000 kg
1 troy oz = 480 grain (gr)
1 g = 5 carat
1 pennyweight = 24 grain
1 stone = 14 lb
1 long ton = 2240 lb
1 short ton = 2000 lb

Pressure
1 Pascal (Pa) = 1 N/m2 = 6.895 kPa
1 atm (metric atmos.) =760 mmHg at 0°C=14.223 lb/in2=1.0132*105 N/m2
1 psi = 2.0355 in. Hg at 32F = 2.0416 in. Hg at 62F
1 microbar = 0.1 N/m2

Scale/Magnitude
atto (a) = 10-18
femto (f) = 10-15
pico (p) = 10-12
nano (n) = 10-9
micro (µ) = 10-6
milli (m) = 10-3
centi (c) 10-2
deci (d) = 10-1
deka (da) = 10
hecto (H) = 102
kilo (K) = 103
mega (M) = 106
giga (G) = 109
tera (T) = 1012
peta (P) = 1015
exa (E) = 1018

Power
1 h.p. (horsepower) = 745.7 W (watts) = 2.545 BTU/hr. = 550 ft.lb./sec.
1 ft•lb/s = 1.356 W
1 J (joule) = 1 N•m = 107 ergs = 0.2389 cal.
1 W = 1 J/s
 units - 36.5

1 ev = 1.60219*10-19 J
1 erg = 10-7 J

Temperature
°F = [(°C*9)/5]+32, °C = Celsius (Centigrade), F = Fahrenheit
K = Kelvin
Rankine (R) = F - 459.666
0.252 calories = 1 BTU (British Thermal Unit)
-273.2 °C = -459.7 °F = 0 K = 0 R = absolute zero
0 °C = 32 °F = 273.3 K = 491.7 R = Water Freezes
100°C = 212°F = 373.3 K = 671.7 R = Water Boils (1 atm. pressure)
1 therm = 100,000 BTU

Mathematical
π radians = 3.1416 radians = 180 degrees = 0.5 cycles
1 Hz = 1 cycle/sec.
1 rpm (revolutions per minute) = 60 RPS (Revolutions per second) = 60Hz
1 fps (foot per second) = 1 ft/sec
1 mph (miles per hour) = 1 mi./hr.
1 cfm (cubic foot per minute) = 1 ft3/min.
e = 2.718

Time
1 Hz (hertz) = 1 s-1
1 year = 365 days = 52 weeks = 12 months
1 leap year = 366 days
1 day = 24 hours
1 fortnight = 14 days
1 hour = 60 min.
1 min = 60 seconds
1 millenium = 1000 years
1 century = 100 years
1 decade = 10 years

Physical Constants
R = 1.987 cal/mole K = ideal gas law constant
K = Boltzmann’s constant = 1.3x10-16 erg/K = 1.3x10-23 J/K
h = Planck’s constant = 6.62x10-27 erg-sec = 6.62x10-34 J.sec
Avagadro’s number = 6.02x1023 atoms/atomic weight
density of water = 1 g/cm3
electron charge = 1.60x10-19 coul.
electron rest mass = 9.11*10**-31 kg
proton rest mass = 1.67*10**-27 kg
 units - 36.6

speed of light (c) = 3.00x1010 cm/sec

speed of sound in dry air 25 C = 331 m/s
gravitational constant = 6.67*10**-11 Nm**2/kg**2
permittivity of free space = 8.85*10**-12 farad/m
permeability of free space = 1.26*10**-6 henry/m
mean radius of earth = 6370 Km
mass of earth = 5.98*10**24 kg

Electromagnetic
magnetic flux = weber (We) = 10**8 maxwell
inductance = henry
magnetic flux density = tesla (T) = 10**4 gauss
magnetic intensity = ampere/m = 0.004*PI oersted
electric flux density = coulomb/m**2
capacitance = farad
permeability = henry/m
electric field strength = V/m
luminous flux = lumen
luminance = candela/m**2
1 flame = 4 foot candles = 43.05564 lux = 43.05564 meter-candles
illumination = lux
resistance = ohm

36.4 ASCII, HEX, BINARY CONVERSION

• The table below will allow conversions between decimal, binary, hexadecimal,
and ASCII values. The values shown only go up to 127. ASCII values above this are not
commonly used in robust applications.
 units - 36.7

hexadecimal

hexadecimal
decimal

decimal
ASCII

ASCII
binary

binary
0 0 00000000 NUL 32 20 00100000 space
1 1 00000001 SOH 33 21 00100001 !
2 2 00000010 STX 34 22 00100010 “
3 3 00000011 ETX 35 23 00100011 #
4 4 00000100 EOT 36 24 00100100 $
5 5 00000101 ENQ 37 25 00100101 %
6 6 00000110 ACK 38 26 00100110 &
7 7 00000111 BEL 39 27 00100111 ‘
8 8 00001000 BS 40 28 00101000 (
9 9 00001001 HT 41 29 00101001 )
10 A 00001010 LF 42 2A 00101010 *
11 B 00001011 VT 43 2B 00101011 +
12 C 00001100 FF 44 2C 00101100 ,
13 D 00001101 CR 45 2D 00101101 -
14 E 00001110 S0 46 2E 00101110 .
15 F 00001111 S1 47 2F 00101111 /
16 10 00010000 DLE 48 30 00110000 0
17 11 00010001 DC1 49 31 00110001 1
18 12 00010010 DC2 50 32 00110010 2
19 13 00010011 DC3 51 33 00110011 3
20 14 00010100 DC4 52 34 00110100 4
21 15 00010101 NAK 53 35 00110101 5
22 16 00010110 SYN 54 36 00110110 6
23 17 00010111 ETB 55 37 00110111 7
24 18 00011000 CAN 56 38 00111000 8
25 19 00011001 EM 57 39 00111001 9
26 1A 00011010 SUB 58 3A 00111010 :
27 1B 00011011 ESC 59 3B 00111011 ;
28 1C 00011100 FS 60 3C 00111100 <
29 1D 00011101 GS 61 3D 00111101 =
30 1E 00011110 RS 62 3E 00111110 >
31 1F 00011111 US 63 3F 00111111 ?
 units - 36.8

hexadecimal

hexadecimal
decimal

decimal
ASCII

ASCII
binary

binary
64 40 01000000 @ 96 60 01100000 ‘
65 41 01000001 A 97 61 01100001 a
66 42 01000010 B 98 62 01100010 b
67 43 01000011 C 99 63 01100011 c
68 44 01000100 D 100 64 01100100 d
69 45 01000101 E 101 65 01100101 e
70 46 01000110 F 102 66 01100110 f
71 47 01000111 G 103 67 01100111 g
72 48 01001000 H 104 68 01101000 h
73 49 01001001 I 105 69 01101001 i
74 4A 01001010 J 106 6A 01101010 j
75 4B 01001011 K 107 6B 01101011 k
76 4C 01001100 L 108 6C 01101100 l
77 4D 01001101 M 109 6D 01101101 m
78 4E 01001110 N 110 6E 01101110 n
79 4F 01001111 O 111 6F 01101111 o
80 50 01010000 P 112 70 01110000 p
81 51 01010001 Q 113 71 01110001 q
82 52 01010010 R 114 72 01110010 r
83 53 01010011 S 115 73 01110011 s
84 54 01010100 T 116 74 01110100 t
85 55 01010101 U 117 75 01110101 u
86 56 01010110 V 118 76 01110110 v
87 57 01010111 W 119 77 01110111 w
88 58 01011000 X 120 78 01111000 x
89 59 01011001 Y 121 79 01111001 y
90 5A 01011010 Z 122 7A 01111010 z
91 5B 01011011 [ 123 7B 01111011 {
92 5C 01011100 yen 124 7C 01111100 |
93 5D 01011101 ] 125 7D 01111101 }
94 5E 01011110 ^ 126 7E 01111110 r arr.
95 5F 01011111 _ 127 7F 01111111 l arr.

36.5 G-CODES

• Note that G and M codes not universal standards, and may vary between
machines. In any case of doubt, the manuals for the machine should be checked.
 units - 36.9

• A basic list of ‘G’ operation codes is given below. These direct motion of the
tool.

G00 - Rapid move (not cutting)

G01 - Linear move
G02 - Clockwise circular motion
G03 - Counterclockwise circular motion
G04 - Dwell
G05 - Pause (for operator intervention)
G08 - Acceleration
G09 - Deceleration
G17 - x-y plane for circular interpolation
G18 - z-x plane for circular interpolation
G19 - y-z plane for circular interpolation
G20 - turning cycle or inch data specification
G21 - thread cutting cycle or metric data specification
G24 - face turning cycle
G25 - wait for input #1 to go low (Prolight Mill)
G26 - wait for input #1 to go high (Prolight Mill)
G28 - return to reference point
G29 - return from reference point
G31 - Stop on input (INROB1 is high) (Prolight Mill)
G33-35 - thread cutting functions (Emco Lathe)
G35 - wait for input #2 to go low (Prolight Mill)
G36 - wait for input #2 to go high (Prolight Mill)
G40 - cutter compensation cancel
G41 - cutter compensation to the left
G42 - cutter compensation to the right
G43 - tool length compensation, positive
G44 - tool length compensation, negative
G50 - Preset position
G70 - set inch based units or finishing cycle
G71 - set metric units or stock removal
G72 - indicate finishing cycle (EMCO Lathe)
G72 - 3D circular interpolation clockwise (Prolight Mill)
G73 - turning cycle contour (EMCO Lathe)
G73 - 3D circular interpolation counter clockwise (Prolight Mill)
G74 - facing cycle contour (Emco Lathe)
G74.1 - disable 360 deg arcs (Prolight Mill)
G75 - pattern repeating (Emco Lathe)
G75.1 - enable 360 degree arcs (Prolight Mill)
G76 - deep hole drilling, cut cycle in z-axis
G77 - cut-in cycle in x-axis
G78 - multiple threading cycle
G80 - fixed cycle cancel
 units - 36.10

G81-89 - fixed cycles specified by machine tool manufacturers

G81 - drilling cycle (Prolight Mill)
G82 - straight drilling cycle with dwell (Prolight Mill)
G83 - drilling cycle (EMCO Lathe)
G83 - peck drilling cycle (Prolight Mill)
G84 - taping cycle (EMCO Lathe)
G85 - reaming cycle (EMCO Lathe)
G85 - boring cycle (Prolight mill)
G86 - boring with spindle off and dwell cycle (Prolight Mill)
G89 - boring cycle with dwell (Prolight Mill)
G90 - absolute dimension program
G91 - incremental dimensions
G92 - Spindle speed limit
G93 - Coordinate system setting
G94 - Feed rate in ipm (EMCO Lathe)
G95 - Feed rate in ipr (EMCO Lathe)
G96 - Surface cutting speed (EMCO Lathe)
G97 - Rotational speed rpm (EMCO Lathe)
G98 - withdraw the tool to the starting point or feed per minute
G99 - withdraw the tool to a safe plane or feed per revolution
G101 - Spline interpolation (Prolight Mill)

• M-Codes control machine functions and these include,

M00 - program stop

M01 - optional stop using stop button
M02 - end of program
M03 - spindle on CW
M04 - spindle on CCW
M05 - spindle off
M06 - tool change
M07 - flood with coolant
M08 - mist with coolant
M08 - turn on accessory #1 (120VAC outlet) (Prolight Mill)
M09 - coolant off
M09 - turn off accessory #1 (120VAC outlet) (Prolight Mill)
M10 - turn on accessory #2 (120VAC outlet) (Prolight Mill)
M11 - turn off accessory #2 (120VAC outlet) (Prolight Mill) or tool change
M17 - subroutine end
M20 - tailstock back (EMCO Lathe)
M20 - Chain to next program (Prolight Mill)
M21 - tailstock forward (EMCO Lathe)
M22 - Write current position to data file (Prolight Mill)
M25 - open chuck (EMCO Lathe)
M25 - set output #1 off (Prolight Mill)
 units - 36.11

M26 - close chuck (EMCO Lathe)

M26 - set output #1 on (Prolight Mill)
M30 - end of tape (rewind)
M35 - set output #2 off (Prolight Mill)
M36 - set output #2 on (Prolight Mill)
M38 - put stepper motors on low power standby (Prolight Mill)
M47 - restart a program continuously, or a fixed number of times (Prolight Mill)
M71 - puff blowing on (EMCO Lathe)
M72 - puff blowing off (EMCO Lathe)
M96 - compensate for rounded external curves
M97 - compensate for sharp external curves
M98 - subprogram call
M99 - return from subprogram, jump instruction
M101 - move x-axis home (Prolight Mill)
M102 - move y-axis home (Prolight Mill)
M103 - move z-axis home (Prolight Mill)

• Other codes and keywords include,

Annn - an orientation, or second x-axis spline control point

Bnnn - an orientation, or second y-axis spline control point
Cnnn - an orientation, or second z-axis spline control point, or chamfer
Fnnn - a feed value (in ipm or m/s, not ipr), or thread pitch
Innn - x-axis center for circular interpolation, or first x-axis spline control point
Jnnn - y-axis center for circular interpolation, or first y-axis spline control point
Knnn - z-axis center for circular interpolation, or first z-axis spline control point
Lnnn - arc angle, loop counter and program cycle counter
Nnnn - a sequence/line number
Onnn - subprogram block number
Pnnn - subprogram reference number
Rnnn - a clearance plane for tool movement, or arc radius, or taper value
Qnnn - peck depth for pecking cycle
Snnn - cutting speed (rpm), spindle speed
Tnnn - a tool number
Unnn - relative motion in x
Vnnn - relative motion in y
Wnnn - relative motion in z
Xnnn - an x-axis value
Ynnn - a y-axis value
Znnn - a z-axis value
; - starts a comment (proLight Mill), or end of block (EMCO Lathe)
 materials - 37.1

37. ATOMIC MATERIAL DATA

Element Atomic # Atomic weight density g/cm3 Valances Hw

copper 29 63.57 8.96 1/2
iron 26 55.85 7.86 2/3
aluminum 13 26.97 2.67 3
tungsten 74 183.85 19.3 6/8
zinc 30 65.37 7.13 2
silicon 14 28.09 2.33 4
carbon 6 12.01 2.1 2/4
titanium 22 47.9 4.51 3/4

F = 96,500 coulombs

37. MECHANICAL MATERIAL PROPERTIES

Table 1:

yield stress - comp.

thermal expansion
ult. stress - comp.

ult. stress - shear

ten - yield strain
ten - yield stress

ten - ult. stress

Material

Density

poisson
Type

G
E

10-6/deg C
(Mg/m3)

(MPa)

(MPa)
(GPa)

(GPa)

(%)

aluminum typical 2.6-2.8 70-79 26-30 0.33 35-500 100-550 1-45 23

1100-h14

2024-T6

5456-h116

2014-T6 2.8 73 36-41 0.33 410 480 13

6061-T6 2.8 70 26 0.33 270 310 17

7075-T6 2.7 72 26 0.33 480 550 11

 materials - 37.2

Table 1:

yield stress - comp.

thermal expansion
ult. stress - comp.

ult. stress - shear

ten - yield strain
ten - yield stress

ten - ult. stress

Material

Density

poisson
Type

G
E

10-6/deg C
(Mg/m3)

(MPa)

(MPa)
(GPa)

(GPa)

(%)
Brass typical 8.4-8.6 96-110 36-41 0.34 70-550 200-620 4-60 19.1-
21.2

Yellow Brass

cold rolled

annealed

Red Brass

cold rolled

annealed

Bronze typical 8.2-8.8 96-120 36-44 0.34 82-690 200-830 5-60 18-21

Cast Iron typical 7.0-7.4 83-170 32-69 0.2-0.3 120 69-480 0-1 340- 9.9-12
1400

Concrete typical 2.3 17-31 na 0.1-0.2 10-70 7-14

reinforced 2.4 17-31 na 0.1-0.2

lightweight 1.1-1.8 17-31 na 0.1-0.2

Copper typical 8.9 110-120 40-47 0.33- 55-760 230-830 4-50 16.6-
0.36 17.6

annealed

hard drawn

Glass typical 2.4-2.8 48-83 19-35 0.17- 30-1000 0 5-11

0.27

Plate 70

Fibers 7000-
20000

Magnesium typ. alloys 1.76- 41-45 15-17 0.35 80-280 140-340 2-20 26.1-
1.83 28.8

Monel 67%Ni,30% 8.8 170 66 0.32 170- 450- 2-50 14

Cu 1100 1200

Nickel 8.8 210 80 0.31 100-620 310-760 2-50 13

 materials - 37.3

Table 1:

yield stress - comp.

thermal expansion
ult. stress - comp.

ult. stress - shear

ten - yield strain
ten - yield stress

ten - ult. stress

Material

Density

poisson
Type

G
E

10-6/deg C
(Mg/m3)

(MPa)

(MPa)
(GPa)

(GPa)

(%)
Plastics

Nylon .88-1.1 2.1-3.4 na 0.4 40-80 20-100 70-140

Polyethylene .96-1.4 0.7-1.4 na 0.4 7-28 15-300 140-290

Rock 5-9

Granite 2.6-2.9 40-100 na 0.2-0.3 50-

280

Marble 2.6-2.9 40-100 na 0.2-0.3 50-

280

Quartz 2.6-2.9 40-100 na 0.2-0.3 50-

280

Limestone 2.0-2.9 20-70 na 0.2-0.3 20-

200

Sandstone 2.0-2.9 20-70 na 0.2-0.3 20-

200

Rubber typical .96-1.3 0.0007- 0.0002- 0.45- 1-7 7-20 100-800 130-200
0.004 0.001 0.50

Sand,soil,gra 1.2-2.2
vel

Steel 7.85 190-210 75-80 0.27- 10-18

0.30

cold rolled

annealed

high strength 340- 550- 5-25 14

1000 1200

machine 340-700 550-860 5-25

spring 400- 700- 3-15

1600 1900

stainless 280-700 400- 5-40

1000

tool 520 900 8

structural 200-700 340-830 10-40 12

astm-a48
 materials - 37.4

Table 1:

yield stress - comp.

thermal expansion
ult. stress - comp.

ult. stress - shear

ten - yield strain
ten - yield stress

ten - ult. stress

Material

Density

poisson
Type

G
E

10-6/deg C
(Mg/m3)

(MPa)

(MPa)
(GPa)

(GPa)

(%)
astm-a47

astm-a36 250 400 30 12

astm-a5242

astm-a441

astm-a572 340 500 20 12

astm-a514 700 830 15 12

wire 280- 550- 5-40

1000 1400

Stainless aisi 302 17

Steel

Titanium typ. alloys 4.5 100-120 39-44 0.33 760- 900- 10 8.1-11
1000 1200

Tungsten 1.9 340-380 140-160 0.2 1400- 0-4 4.3

4000

Water fresh 1.0

sea 1.02

Wood (dry)

Douglas fir .48-.56 11-13 30-50 50-80 30-50 40-70

Oak .64-.72 11-12 40-60 50-100 30-40 30-50

Southern .56-.64 11-14 40-60 50-100 30-50 40-70

pine

37.1 FORMULA SHEET

• A collection of the essential mechanics of materials formulas are given below

 materials - 37.5

Axial/Normal Stress/Strain P PL
σ = εE = --- δ = εL = -------
A AE

Shear Stress/Strain τ xy
τ = γG γ xy = -------
G

Poisson’s ratio ε lateral

ν = – --------------------------- E = 2G ( 1 + ν )
ε longitudinal

σx σy σz
ε x = ----- – ν ----- – ν -----
E E E

Torsion τ max
Tc
τ max = ------ γmax = ----------
J G
4
TL πr (cylinder)
φ = ------- J = --------
JG 2

π 4 4
σ ten = – σ comp = τ J = --- ( r o – r i ) (hollow tube)
2
γ
ε = --- P = 2πfT
2
 materials - 37.6

3
Beams My bh
σ = – -------- I = -------- (for rectangle)
12
I
4
VQ πr
τ = -------- I = -------- (for circle)
2
Ib

c
L = ρθ ε max = ---
ρ

My EI
σ = – -------- ρ = ------
I M

2
Buckling π EI k=0.5 (both ends fixed)
P < -----------
2 2
- k=0.7 (one end fixed, one pinned)
k L k=1.0 (both ends pinned)
k=2.0 (one end pinned, one free)

K d/2 r
3.0 P P

d/2
2.5

2.0

1.5
r/d
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
 materials - 37.7

K
r
2.5

P D
d
P
2.0

D/d = 2.0
1.5
D/d = 1.5
D/d = 1.25
D/d = 1.1

1.0
r/d
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
 materials - 37.8

K
T
r T

D d
2.5

2.0

1.5
D/d = 2.0
D/d = 1.33
D/d = 1.2
D/d = 1.1
1.0
r/d
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
 bibliography - 38.1

38. BIBLIOGRAPHY

38.1 TEXTBOOKS

38.1.1 Slotine and Li

Citation: Slotine, J. E., Li, W., "Applied Nonlinear Control", Prentice Hall, 1991.

Topics/Chapters:

Introduction
Phase plane analysis
Fundamentals of Lyapunov theory
Advanced stability theory
Describing function analysis
Feedback linearization
Sliding control
Adaptive control
Control of Multi-input physical systems

Comments: Directed to mechanical undergraduate students.

38.1.2 VandeVegte

Citation: VandeVegte, John, "Feedback Control Systems", Prentice Hall, Second

Edition, 1990.

Topics/Chapters:

Linearized dynamic models

Transfer function models of physical systems
Modeling of feedback systems and controllers
Transient performance and the s-plane
The perfoarmance and dynamic compensation of feedback systems
The root-locus method
 bibliography - 38.2

Frequency response analysis

Frequency response design
Digital control systems
Digital control system analysis and design
State space analysis
Introduction to state space design
Multivariable systems in the frequency domain
Nonlinear control systems
App: Vectors matrices and determinants
App: Computer aids for analysis and design

Comments: Directed to mechanical undergraduate students. Chapters cover a

range of topics systems including motors, pneumatics, fluids, thermo, etc. Chapters 1-8
use classical design techniques. Chapters 9-10 present z-transform methods for controller
analysis, some design issues are presented.
 input output equations - 39.1

39. TOPICS IN DEVELOPMENT

Topics:

Objectives:

39.1 UPDATED DC MOTOR MODEL

ea = KE ω

Te = KT ia

2
T e – T L – C d ω = JDω
2
T e = JDω + C d ω + T L

v a = r a i a + l a Di a + e a

va = ia ( ra + la D ) + e a

va = ia ( ra + la D ) + K E ω
Te
v a = ------ ( r a + l a D ) + K E ω
KT
2
( JDω + C d ω + T L )
v a = ------------------------------------------------ ( r a + l a D ) + K E ω
KT
2
( JDω + C d ω + T L ) ( r a + l a D ) + K E K T ω = K T v a
2 2
Jr a Dω + C d r a ω + T L r a + Jla DDω + l a DC d ω + l a DT L + K E K T ω = K T v a
·· · 2 ·
ω ( Jla ) + ω ( Jr a + l a ωC d ) + ω ( K E K T ) + ω ( C d r a ) = v a ( K T ) + T L ( – r a ) + T L ( – l a )
 input output equations - 39.2

Measure the motor resistance,

r a = measured

When stalled a voltage is applied and the torque is measured,

0 ( Jl a ) + 0 ( Jr a + l a ωC d ) + 0 ( K E K T ) + 0 ( C d r a ) = v a ( K T ) + T L ( – r a ) + 0 ( – l a )

T L ( ra ) = v a ( K T )
TL ( ra )
K T = ---------------
-
va
At a steady state speed,
2
0 ( Jl a ) + 0 ( Jr a + l a ωC d ) + ω ( K E K T ) + ω ( C d r a ) = v a ( K T ) + T L ( – r a ) + 0 ( – l a )
2
ω ( KE KT ) + ω ( Cd ra ) = v a ( K T ) + T L ( –r a )
2
ω ( C d ra ) + ω ( KE KT ) + ( TL ( ra ) – va ( KT ) ) = 0

At lower steady state speeds (little drag),

0 ( Jl a ) + 0 ( Jr a + l a ωC d ) + ω ( K E K T ) + 0 = v a ( K T ) + T L ( – r a ) + 0 ( – l a )

ω ( KE KT ) = va ( KT ) + TL ( –ra )
ra
ω ( K E ) + T L  ------ = v a
KT
The static friction limit,

0 ( Jl a ) + 0 ( Jr a + l a ωC d ) + 0 ( K E K T ) + 0 ( C d r a ) = v a ( K T ) + T L ( – r a ) + 0 ( –l a )
KT
T L = v a  ------
ra
 input output equations - 39.3

·· · 2 ·
ω + ω A + ωB + ω C = v a D + T L E + T L F
 input output equations - 39.4

39.2 ANOTHER DC MOTOR MODEL

TL

-
ia Te
vs + 1 -
-------------------- 1
-------------------- ωr
La D + ra Ka JD + Bm
+
-

Ka

where,

K a = motor speed constant

L a = motor armature inductance
r a = motor armature resistance
J = inertia of rotor
B m = damping of rotor
v s = voltage applied to armature
i a = armature current
T e = effective motor torque
T L = load torque
ω r = motor speed
 input output equations - 39.5

Motor parameters can be measured by applying a voltage and holding (stalling) the shaft.
The stalled motor torque, and the current are measured. When stalled, the load and effec-
tive torques are equal, and the feedback loop is zero, allowing the simplified system
block diagram. This allows the calculation of the motor speed coefficient directly.

vs 1 ia TL
--------------------- Ka
La D + ra

T
K a = -----L-
ia
The resistance of the windings can be found by direct measurement with a DMM, but can
also be verified by using the steady state current and voltages.
i 1 - 1 1
---a- = ------------------ = --------------------------- = ----
vs La s + ra L a ( j0 ) + r a ra
v
r a = ----s
ia
The inductance of the windings can be found using the time constant - phase shift
????????????????
i 1 -
---a- = ------------------
vs La s + ra XXXXXXXXXXXXXXX
 input output equations - 39.6

If the motor is unloaded (TL = 0) the following differential equation can be derived. If the
system is allowed to reach steady state, the derivatives of motor speed will become
zero.

ia Te
vs + 1 -
--------------------
1
-------------------- ωr
La D + ra Ka JD + Bm
-

Ka

2 2
D ω r ( L a J ) + Dω r ( L a B m + Jr a ) + ω r ( r a Bm + K a ) = Vs K a
2
( 0 ) ( L a J ) + ( 0 ) ( L a B m + Jr a ) + ω r ( r a B m + K a ) = V s K a
2
ω r ( ra Bm + K a ) = Vs Ka
2
Vs K a K a
Bm = ------------ – ------
ra ωr r a

The rotor inertia can be found using .....................

2 2
D ω r ( L a J ) + Dω r ( L a B m + Jr a ) + ω r ( r a B m + K a ) = V s K a

• Another approach to finding the motor parameters is system identfication.

 input output equations - 39.7

TL

-
ia Te
vs + 1 -
--------------------
1
-------------------- ωr
La D + ra Ka JD + Bm
+
-

Ka

1
i a = ( v s – K a ω r ) ---------------------
La D + ra

Di a ( L a ) + i a ( r a ) = v s – K a ω r

–K a –ra
Di a = v s  ----- + ω r  --------- + i a  --------
1
La La La

1
ω r = ( K a i a – T L ) --------------------
JD + B m

Dω r ( J ) + ω r ( B m ) = K a i a – T L

K –B
Dω r = i a  -----a- + T L  ------ + ω r  ---------m-
–1
 J  J  J 

–r –K 1-
-------a- --------a- ---- 0
d- i a
----
La La ia La vs
= +
dt ω Ka –B m ω r –1 T
r
------ ---------- 0 ------ L
J J J
 input output equations - 39.8

The matrix can be parameterized,

–ra –Ka 1-
-------- --------- ---- 0
d- i a La La ia La vs i vs
---- = + = A B a + E 0
dt ω Ka –B m ω r –1 T C D ωr 0 F TL
r
------ ---------- 0 ------ L
J J J

1
E = ----- La = E
La
–ra
A = -------- r a = – AL a = – AE
La
–Ka
B = --------- K a = – BL a = – BE
La
Ka Ka – BE
C = ------ J = ------ = ----------
J C C
–Bm – BED
D = ---------- B m = ---------------
J C

39.3 BLOCK DIAGRAMS AND UNITS

- each block in a block diagram has an input and output. Making sure that the units
are consistent is useful for detecting mistakes and ensuring consitency.
 input output equations - 39.9

39.4 SIGNAL FLOW GRAPHS

- these invert block diagrams.

·· 1- · 1-
x --- x --- x
D D

39.5 ZERO ORDER HOLD

- t.

1 – Ts
1–e
G h ( s ) = ------------------
s

t
0 T

39.6 TORSIONAL DAMPERS

-
 input output equations - 39.10

39.7 MISC

- controller goals
stabilize - to make the system stable
regulate - to control the system about a central design value
follow - to track an input/feedback signal
disturbance rejection - to minimize the effects of noise

- state diagrams can be used with multiple input/multiple output (MIMO) systems

- state based controller designs,

state feedback - in an ideal controller the full state variables can be mea-
sured
observer-estimator - determines the system state using limited data.

- Linear Time Invariant (LTI) systems are linear, having constant coefficients. For
example, state coefficient matrices are all constant values

- Linear systems allow superposition

- leaf springs are non-linear and have a cubic relationship to displacement

- if a system input (setpoint) is zero, or not used, then the control system is called a
regulator.

- stepper motor motion plans should not be approximate, the stepper motor should
follow the path, although trapezoidal profiles are more common.

39.8 Nyquist Plot

• The method is used to determine system stability using the characteristic equa-
tion 1+Gc(s)G(s)H(s)

• If over the range of frequencies the system enters the unstable range of the
Nyquists contour shown below.
 input output equations - 39.11

• The basic method for producing polar plots is,

1. Write the characteristic equation for a feedback system Gc(s)G(s)H(s)

2. Use phasors to determine a gain and angle.
3. For a large range of frequencies, plot the gain and angle on polar axes.
4. Look for ’encirclements’ of the point at a gain of -1 and -180°.

• If the number of clockwise encirclements is equal to the number of poles in the

right hand side of the plane, the system is stable.

• The gain and phase margins can be determined from the polar plot as shown
below,
 input output equations - 39.12

φM 1
G M = --------
OC
-1
C O

ωc
A

39.9 NICHOLS CHART

- a plot of the gain vs. the angle of a system transfer function after a phasor trans-
form.

• The basic method is,

1. Write an expression for the closed loop gain Gc(s)G(s).

2. Apply a phasor transform to find the gain and phase angle.
3. Gain and phase values are calculated for a large range of frequencies and plotted
on the following paper.
4. Measure the gain and phase margins from the plot.
 input output equations - 39.13

0dB
25
0.5dB
-355° -5°

20

-350° -10°
15

3dB
10
-1dB

5 -330° -30°
dB 12dB

-300°
-5
-60°
-9dB
-10
-270°
-90°
-15
-240° -120°
-210° -180° -150° -18dB
-20
-280 -260 -240 -220 -200 -180 -160 -140 -120 -100 -80
phase (deg)

• The gain and phase margins are measured from the plot as shown below,
 input output equations - 39.14

180°

φM

0dB

GM

Where,
φ M = phase margin
G M = gain margin

• These correlate to values on the Bode plot. To find the approximate gain/phase
margins with a Bode plot, locate the point where the gain is 180° to find the gain margin,
and find the phase margin where the gain is 0dB.

• In unstable systems the function will circle the center point of the plot.

39.10 BESSEL POLYNOMIALS

- put in root locus chapter?

- The following table of functions provides polynomial roots normalized to a

motion that is complete in 1second [How, 2001].
 input output equations - 39.15

polynomial roots
order

1 -4.6200
2 -4.0530±2.3400j
3 -5.0093,-3.9668±3.7845j
4 -4.0156±5.0723j,-5.5281±1.6553j
5 -6.4480,-4.1104±6.3142j,-5.9268±3.0813j
6 -4.2169±7.5300j,-6.2613±4.4018j,-7.1205±1.4540j
7 -8.0271,-4.3361±8.7519j, -6.5714±5.6786j,-7.6824±2.8081j
8 -4.4554±9.9715j,-6.8554±6.9278j,-8.1682±4.1057j,-8.7693±1.3616j
9 -9.6585,-4.5696±11.1838j,-7.1145±8.1557j,-8.5692±5.3655j,-9.4013±2.6655j
10 -4.6835±12.4033,-7.3609±9.3777j,-8.9898±6.6057j,-9.9657±3.9342j,-
10.4278±1.3071j

- the roots are used to construct the charateristic equation of the system.

- To increase, or decrease the reposnse time the root values are multiplied by the
desired settling time.

39.11 ITAE

- The Integral of Time Absolute Error (ITAE) integral equation becomes large for
errors that persist for a long time.

J ITAE = ∫ t e ( t ) dt
0

- is used to select poles to minimize the integral value when selecting a characteris-
tics equation.
 input output equations - 39.16

39.12 ROOT LOCUS

- A design point on the root locus curve can be selected for a specific damping
coefficient.

cos θ = ζ
θ

39.13 LYAPUNOV’S LINEARIZATION METHOD

• This method addresses the local stability of a non-linear system when linearized.

• In essense it validated the use of linear control systems on non-linear systems.

•
 input output equations - 39.17

The proof begins with a taylor series expansion of a function that has a continuous dif-
ferential. The second term in the expression is the higher order terms. There is an
assumption that the system starts at equillibrium, so there is no initial value. The
result is a linear approximation.

∂f  ∂f  ∂f
----- x = f ( 0 ) +   A =  
d
x + f ( t ) =
dt  ∂ x x = 0 h.o.t.  ∂ x x = 0 x = Ax ∂x x = 0

∂f ∂f
x =   x +  
d-
---- u + f h.o.t. ( t ) = Ax + Bu
dt ∂ x x = 0, u = 0 ∂ u x = 0, u = 0

∂f ∂f
A =   B =  
∂ x x = 0, u = 0 ∂ u x = 0, u = 0
A linearized model of the control law can be written and the linearized model rewritten.

∂u
u ≈   x = Gx
∂x x = 0

d-
---- x = Ax + BGx = ( A + BG )x
dt

• Lyapunov’s stability theorem focuses on the stablilty (Eigenvalues) of matrix A

at a given point.

- The system is asymptotically stable at the equillibrium point if all of the

Eigenvalues of A are in the left hand complex plane.
- The system is unstable as the equillibrium point if the matrix A is unsta-
ble, with points in the right hand side of the complex plane.
- If all of the Eigenvalues are in the left hand plane, except for one or more
on the complex axis, then the system is marginally stable, and the linear-
ization may be stable or unstable.

39.14 XXXXX

- ea.
 input output equations - 39.18

39.15 XXXXX

- ea.

39.16 XXXXX

- ea.

39.17 XXXXX

- ea.

39.18 XXXXX

- ea.

39.19 XXXXX

- ea.

39.20 XXXXX

- ea.

39.21 SUMMARY

39.22 PRACTICE PROBLEMS

1.
 input output equations - 39.19

39.23 PRACTICE PROBLEM SOLUTIONS

1.

39.24 ASSGINMENT PROBLEMS

1.

39.25 REFERENCES

39.26 BIBLIOGRAPHY

How, J, "16.31 Feedback Control Course Notes", MIT Opencourseware website.