Professional Documents
Culture Documents
ho
ks
or
Stability and Stability Margin in Freq
W
n
Domain
ai
M
T
T. K. Ghoshal
0K
T1
Jadavpur University
EI
2014
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margin in Nyquist
T
0K
Stability Margin in Bode
T1
Disk Margin
M
N
or
W
Stability and Stability Margin in Freq
n
ai
M
Domain (Units 5.2 & 5.3)
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
or
Background Concepts and Definitions
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
T
Stability Margin in Nyquist
0K
T1
Disk Margin
M
N
p
ho
5.1 Module Introduction
ks
or
5.2 Nyquist Plane Analysis of Stability
W
n
5.3 Concepts of Stability Margins
ai
M
T
5.4 Effects of Pole Zero Cancellation
0K
T1
CT
EI
M
N
p
loop system, such gain/phase/disc margins, regions
ho
ks
of exclusion of poles of close loop system on the
or
complex planes etc.
W
n
Explain and demonstrate numerically the nature of
ai
M
closed loop response near the stability boundary
T
0K
Establish the numerical validation of stability
T1
p
in the loop
ho
ks
Relate stability margins such as gain and phase
or
W
margins and corresponding cross-over frequencies
n
with time/frequency domain response parameters
ai
M
such as bandwidth, damping factor etc.
T
0K
T1
CT
EI
M
N
p
ho
Nyquist plot over the D-zone in Imaginary plane
ks
Nyquist Criterion
or
W
Unit 5.3: Concepts of Stability Margins
n
Concepts of Stability Margins
ai
M
Defining Stability Margin, Stability Margin as ,
T
robustness, 0K
using Nyquist, Bode and Nichols plots to represent
T1
systems,.
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 8
Pedagogic Notes
The theory of stability for (linear) feedback
p
system owes to several engineers of Bell
ho
Labs, USA, who were working to improve
ks
amplifiers for long distance telephony. This
or
Nyquist
include:
W
n
Harry Nyquist (1889 1976)
ai
M
Stability analysis, fax machines and many more
T
Harold Stephen Black (1898 1983)
0K
Bode
Analysis of Negative Feedback
T1
Black
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 9
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margin in Nyquist
T
0K
Stability Margin in Bode
T1
Disk Margin
M
N
or
W
Stability and Stability Margin in Freq
n
ai
M
Domain (Units 5.2 and 5.3)
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected
ks
or
Topics
W
n
Background Concepts and Definitions
ai
M
Nyquist Stability Theorem
T
Interpretations for Open loop stable systems
0K
T1
Disk Margin
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 12
p
ho
ks
or
W
Background Concepts and
n
ai
M
Definitions
T
0K
T1
CT
EI
M
N
Stability of Linear Systems
Stability of a (closed loop or open loop) linear
p
system can be determined from the location of the
ho
poles.
ks
or
Systems which have all the poles in the left half of
W
the complex plane (LHP) can be proven to be
n
ai
asymptotically stable.
M
T
Initial conditions eventually decays exponentially
0K
This can be proven without taking recourse to the
T1
additional advantageCKBS-Jadavpur
for stability anlysis , beyond a 14
Y/N type of answer.
T10KT-December 2014 University, Kolkata
Objective of Nyquist Theorem
Nyquist Theorem allows prediction of the stability of
p
a closed loop linear system
ho
ks
from the frequency domain behaviour (TF) of the
or
corresponding open loop system,
W
n
Over which a system designer may have some choice to
ai
modify
M
T
without solving for the roots of the characteristic
0K
equation of the closed loop system.
T1
stability, it
M
But also
T10KT-December 2014 provides a CKBS-Jadavpur
quantitative measure
University, Kolkata of stability. 15
Open and closed loop Transfer Function
K ( s )G p ( s )
p
Gc ( s ) = = CLTF
ho
1 + K ( s )G p ( s ) H ( s )
ks
G ( s)
or
= (say)
1 + G ( s) H ( s)
W
T ( s ) = K ( s )G p ( s ) H ( s ) = OLTF
n
ai
M
Some definitions first: Note
T
Forward path TF= G(s), Feedback path TF= H(s),
0K
T1
p
G (s) = ; H (s) =
ho
DG ( s ) DH ( s )
ks
or
N C (s) NT (s)
Gc ( s ) = ; T (s) =
W
DC ( s ) DT ( s )
n
ai
Characteristic Equation of Gc ( s ) :
M
DC ( s ) = 0 1 + T ( s ) = 0 1 + G ( s ) H ( s ) = 0
T
0K
T1
above ChE.
N
p
ho
DC ( s ) = 0 1 + T ( s ) = 0 1 + G ( s ) H ( s ) = 0
ks
N G N H DG DH + N G N H
1 + G ( s) H ( s) = 1 + =
or
DG DH DG DH
W
n
N G (s) N H (s)
G (s) H (s) =
ai
M
DG ( s ) DH ( s )
T
0K
A zero of 1+G(s)H(s) is a pole of GC(s)
T1
same.
M
N
p
ho
KG ( s )
1 + KG ( s ) H ( s )
ks
KG ( s ) H ( s )
1 + KG ( s ) H ( s )
or
W
n
ai
Zeros: (1,2,3,4) Zeros: (1,2,3,4)
M
Zeros: a, b, c, d
T
0K Poles: (a,b,c,d)
T1
Poles: (5,6,7,8) Poles 5,6,7,8
Unknowns
CT
EI
p
loop poles on the right half plane from the Nyquist
ho
ks
theorem ,
or
W
If none of the closed-loop poles are on the right half
n
plane, the closed loop system is stable.
ai
M
If there is any closed-loop pole on the right half plane,
T
the system goes unstable.
0K
T1
p
ho
ks
or
W
n
ai
M
( s z1 )( s z 2 )
T
F (s) =
0K
( s p1 )( s p2 )
T1
p
ho
ks
or
W
n
ai
( s z1 )( s z 2 )
F (s) =
M
( s p1 )( s p2 )
T
0K
T1
CT
EI
p
Q s
ho
by a vector
ks
originating on the
or
point a and
W
a terminating on the
n
(current location of
ai
M
the) point s.
T
0K
( s z1 )( s z 2 )
F (s) =
T1
( s p1 )( s p2 )
CT
p
Q s
a) can be
ho
ks
represented by a
or
vector opposing
W
a the vector (s-a)
n
and has a
ai
M
magnitude equal
T
0K
( s z1 )( s z 2 ) to the reciprocal of
F (s) =
(s-a)
T1
( s p1 )( s p2 )
CT
A zero inside a
p
Q s
ho
closed contour
ks
or
contributes 360
W
a
deg. angular
n
ai
M
change.
T
0K
( s z1 )( s z 2 )
F (s) =
T1
( s p1 )( s p2 )
CT
A zero outside a
p
Q s
ho
closed contour
ks
or
contributes o
W
deg. angular
n
ai
M
a
change.
T
0K
( s z1 )( s z 2 )
F (s) =
T1
( s p1 )( s p2 )
CT
p
contributes 360 deg.
ho
angular change.
ks
Pole/zero outside
or
contour has 0 deg.
W
angular change.
n
ai
For clockwise traversing
M
around s-contour, a
T
0K zero of F(s) inside
yields rotation in CW,
T1
p
ho
ks
or
W
n
ai
M
N = P-Z Z = P-N
T
0K
T1
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
ai
M
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
M
Interpretations for Open loop stable systems
T
Stability Margin in Nyquist
0K
T1
Disk Margin
M
N
p
is proper and has no unstable pole-zero
ho
ks
cancellations.
or
W
Let N be the number of times that the Nyquist plot for
n
ai
Go(s) encircles the -1 point in the clockwise direction.
M
Also let P denote the number of poles of Go(s) that lie
T
0K
to the right of the imaginary axis.
T1
CT
Then, Z = N + P
EI
M
p
ho
transfer function.
ks
A simplified version of the criterion exists for
or
W
open-loop stable systems.
n
A negative value of N indicates that the -1 point is
ai
M
encircled in the opposite direction (counter-
T
clockwise). This situation implies that each
0K
countercurrent encirclement can stabilize one
T1
ai
M
systems
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margin in Nyquist
T
0K
Stability Margin in Bode
T1
Disk Margin
M
N
p
ho
P = 0 and thus Z = N.
ks
Consequently, the closed-loop system is unstable if the
or
W
Nyquist plot for GOL(s) encircles the critical point, one
n
or more times.
ai
M
Which in other words is that for closed loop stability,
T
the Nyquist plot should not encircle the critical point
0K
T1
at all.
CT
p
I
Indentation <
ho
0
ks
IV Real
R<
or
R
III
W
n
ai
M
Another simplifying assumption is that there are no
T
0K
poles on the imaginary axis either.
T1
p
ho
assumptions, we need
ks
to plot the polar plot
or
corresponding to the
W
positive values of
n
=
ai
frequencies for zero to
M
infinity only.
T
0K
Stability is ensured if
T1
=0
the critical point is
CT
=0 to = .
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 42
Simplified Nyquist
May be extended for open loop poles on imaginary
p
axis.
ho
ks
Also may be extended with open loop unstable
or
W
systems.
n
ai
With extreme caution
M
T
0K
T1
CT
EI
M
N
ai
M
T
0K
T1
CT
EI
M
N
Stability margins on Nyquist.
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
p
ability to quantify the degree of stability by two
ho
quantities:
ks
or
Gain Margin (GM) and
W
Phase margin (PM)
n
ai
The Gain margin is the change quantifies the
M
change (alone) in the open-loop gain (in terms of
T
0K
ratio or equivalently in dB) to make the closed-loop
T1
system unstable ,
CT
p
alone required to make the closed-loop system
ho
ks
unstable.
or
W
This is computed at unity gain (Gain Cross Over)
n
frequency by 190 degree + phase at GCF.
ai
M
Example-1: at PCF, the gain is 0.5,
T
0K
the gain margin is ratio 1/0.5= 2 times = 6dB
T1
p
ho
Why PM?
ks
Delay Margin
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
systems
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 50
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margin in Nyquist
T
0K
T1
Disk Margin
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
The frequency at which
or
the phase equals 180
W
degrees is called the
n
GM
phase crossover
ai
frequency
M
GM
T
The frequency at which
0K the magnitude equals 1 is
T1
called the gain crossover
frequency M
CT
EI
M
p
ho
cuts the negative
ks
or
real axis at the
W
critical point.
n
ai
It also intersects
M
the frequency
T
0K
response plot
T1
where the
CT
GM is to be a
M
N
ratio =a here.
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 54
Stability Margins-3
The stability margins are in effect measures of
p
stability robustness to plant parameter
ho
perturbations.
ks
or
These PM and GM quantify how much change in
W
system phase and system gain respectively can be
n
ai
tolerated before the closed loop system becomes
M
unstable.
T
0K
Phase shift can occur because of neglected dynamics
T1
digital controller.
EI
system.
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 55
p
ho
ks
or
W
n
Stability Margins in Bode
ai
M
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margins in Nyquist
T
0K
Stability Margins in Bode
T1
CT
Disk Margin
M
N
p
ho
to analyse stability and
ks
assessing stability
or
margins if the Open
W
loop TF is stable.
n
ai
In such cases the
M
phase and gain
T
-180 0K
margins may be simply
T1
plot provided
EI
available and
both the GCF and the
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 58
Example
p
40 K
ho
G (s) =
( s + 2)( s + 4)( s + 5)
ks
or
W
n
ai
M
T
0K
T1
CT
p
the right hand side pole location of the closed loop TF is
ho
important
ks
or
The characteristic equation having at least one root on RHS of s-
W
plane makes the system unstable.
n
ai
For analysis of relative stability of closed loop system Gain
M
Margin and Phase Margin are useful
T
0K
The gain margin indicates how far the gain is to increased so that
T1
phase is -180
N
p
ho
with respect to frequency
ks
Gain margin is the amount of gain in decibels (dB) that can be added
or
to the loop before the closed-loop system becomes unstable.
W
Phase cross over frequency is the frequency at which the phase is 180
n
At the PCF the dB is to be recorded from the magnitude plot to get the Gain
ai
margin. A positive GM indicate stability
M
Phase margin (PM) is defined as the angle in degrees which is to be
subtracted at the frequency where the magnitude plot intercept 0 dB
T
line so that the phase is 180
0K
T1
Gain cross over frequency is the frequency at which the magnitude is 0 dB
At GCF the phase is to be recorded and the difference from 180 has to be
CT
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
Fig taken
EI
from B.C.
M
Kuo
N
p
C (s)
= T (s)
ho
R( s)
ks
n2
or
= 2
s + 2 n s + n2
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
Typically, the addition of any poles may change the correlation between step
ho
transient re-sponse and frequency response . However, the derived results for
ks
second order systems may be applicable to higher order systems in the presence
or
of a dominant second order system poles
W
For an LTI higher order systems with a dominant second order pole, the following
relationships generally exists
n
ai
The value of Mr is indicative for the relative stability. A satisfactory performance is
attained for 1 < Mr < 1:4, which corresponds to a damping ration of 0; 4 < < 0:7.
M
A large Mr indicates a high overshot and slow damping.
T
If the system is subject to noise signals whose frequency are near to the resonance
0K
frequency !r, the noise will be amplified in the output causing a serious problem.
T1
The magnitude of the resonance frequency !r indicates the speed of the transient
response. Large !r indicates faster time response [smaller rise and settling times]
CT
The resonant peak frequency!r and the damped natural frequency !d of unit step
response are very close to each other for lightly damped systems.
EI
M
N
p
= T (s)
ho
R( s)
ks
n2
= 2
or
s + 2 n s + n2
W
n
ai
M
T
0K
T1
CT
EI
M
p
n = = BW = n (1 2 2 ) + 4 4 4 2 + 2
ho
Ts Tp 1 2
ks
or
The first set of relations show the inter dependence
W
of natural frequency, peak time and settling time
n
ai
(low damping).
M
The second equation provides approximate 3-db
T
0K
Bandwidth.
T1
NF=90 degree BW
CT
BW/GCF=1.72
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 69
PUOS vs zeta
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
M
N
p
BW = n (1 2 2 ) + 4 4 4 2 + 2
ho
ks
4
BW = (1 2 2 ) + 4 4 4 2 + 2
or
Ts
n
BW = (1 2 2 ) + 4 4 4 2 + 2
ai
M
Tp 1 2
T
0K
BW
T1
1
from value at dc (0 rad/sec), or M =
EI
.
M
2
N
p
ho
ks
or
W
n
Nichols Charts
ai
M
T
0K
T1
CT
p
of open-loop frequency response
ho
ks
or
W
Phase margin of open-loop frequency response
n
ai
Can be written in terms of damping ratio as following
M
T
0K
2
M = tan 1
T1
2 2 + 1 + 4 4
CT
EI
M
N
p
ho
BW = 3.7
ks
or
W
n
ai
M
T
0K
T1
CT
PM=35
EI
M
N
2 2 + 1 + 4 4
p
= 0.32
ho
Solve for PM = 35
ks
or
W
4
Ts = (1 2 2 ) + 4 4 4 2 + 2
n
ai
BW
M
= 5.5
T
0K
Tp = (1 2 2 ) + 4 4 4 2 + 2
T1
BW 1 2
CT
= 1.43
EI
M
N
ai
M
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margins in Nyquist
T
0K
Stability Margins in Bode
T1
Disk Margin
N
ai
M
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margins in Nyquist
T
0K
Stability Margins in Bode
T1
Disk Margin
M
N
p
ho
ks
or
W
n
ai
M
T
0K
T1
CT
EI
Presented in J-N-P
M
N
Popularized Gille et al
T10KT-December 2014 CKBS-Jadavpur University, Kolkata 84
disk margin-3
p
ho
ks
or
W
n
ai
M
T
0K
T1
p
ho
Y ( s) c T
Adj ( sI A)b N ( s )
ks
1
G ( s) = = c ( sI A) b =
T
=
or
U ( s) det( sI A) D( s)
W
n
Obtaining the TF is straight Forward.
ai
M
Note that both the numerator N(s) and the denominator
T
D(s) are polynomials in s, corresponding to the numerator
0K
and the denominator characteristic polynomials.
T1
ai
M
T
0K
T1
CT
EI
M
N
Contents
Pedagogic Perspective
p
ho
Example Course Material of Selected Topics
ks
Background Concepts and Definitions
or
W
Nyquist Stability Theorem
n
ai
Interpretations for Open loop stable systems
M
Stability Margin in Nyquist
T
0K
Stability Margin in Bode
T1
Disk Margin
M
N
p
represented by
ho
Point in state Space
x3
a point in this
ks
x2 state space.
or
W
n
ai
x1
M
State Space: The n-dimensional space
T
0K
whose coordinate axes are the individual
T1
p
; m<n
ho
ak
dt k
dt j
y = C(x) + D(u)
ks
k =0 j =0
or
x = f (x, u) or x = f (x) + g(u ) The terms ak, bj etc or for linear
W
system, elements of the A, B, C,D
y = h(x, u) matrices are called parameters.
n
ai
M
For LTI systems the parameters are constants.
T
For linear time varying systems, some of the parameters
0K
are functions of time
T1
or
W
for
n
ai
M
Your kind AttentionT
0K
T1
CT
EI
M
N