Professional Documents
Culture Documents
40
60
80
100
120
of Electrical and Computer Engineering The University of Texas at Austin EE 382M.7 VLSI I Fall 2011
September 12, 2011
80
1 / 35
Topics
mm 40 60 80 100 120
DC Response Logic Levels and Noise Margins Transient Response Delay Estimation
60 40
80
1 / 35
Transistor Behavior
mm 40 60 80 100 120
If the supply voltage of a chip increases, the maximum transistor current will If the width of a transistor increases, its gate capacitance will If the length of a transistor increases, its gate capacitance will If the supply voltage of a chip increases, the gate capacitance of each transistor will
80 60
4 5 6
2 / 35
Transistor Behavior
Behavior in dierent situations (increase, decrease, or not mm 40 60 80 100 change).
1 2
120
If the width of a transistor increases, the current will increase If the length of a transistor increases, the current will decrease 40 If the supply voltage of a chip increases, the maximum transistor current will increase If the width of a transistor increases, its gate capacitance will 60 increase If the length of a transistor increases, its gate capacitance will increase If the supply voltage of a chip increases, the gate capacitance 80 of each transistor will not change
3 / 35
In between, Vout depends on transistor size and current By KCL, current must be such that Idsn = |Idsp | 60 We could solve equations, but graphical solution gives more insight
80
4 / 35
Transistor Operation
Current through transistor depends on the region of operation mm to identify for what60in and Vout are nMOS100 pMOS 120 40 80 Need V and in Cuto, Linear or Saturation nMOS Operation
40 Cuto Vgsn < Vtn Vin < Vtn 60
Linear Vgsn > Vtn Vin > Vtn Vdsn < Vgsn Vtn Vout < Vin Vtn
Saturated Vgsn > Vtn Vin > Vtn Vdsn > Vgsn Vtn Vout > Vin Vtn
5 / 35
pMOS Operation
mm 40 60 80 100 120
Linear Vgsp < Vtp Vin < VDD + Vtp Vdsp > Vgsp Vtp Vout > Vin Vtp
Saturated Vgsp < Vtp Vin < VDD + Vtp Vdsp < Vgsp Vtp Vout < Vin Vtp
80
6 / 35
I-V Characteristics
Make pMOS wider than nMOS such that n = p = mmox W C L
40 60 80 100 120
40
60
80
7 / 35
40
60
80
8 / 35
60
80
9 / 35
DC Transfer Curve
mm 40 60 80 100 120
40
60
80
10 / 35
Operating Regions
mm 40 60 80 100 120
Region A B C D E
80
11 / 35
Beta Ratio
mm 40 60 80 100 120
Noise Margins
How much noise can a gate input see before it does not recognize mm 40 60 80 100 120 the input?
40
60
80
13 / 35
Logic Levels
To maximize noise margins Select logic levels at unity gain point 80 DC transfer of mm 40 60 100 characteristic
120
40
60
80
14 / 35
Transient Response
mm 60 80 DC analysis40 gives the Vout if Vin is constant 100 120
Transient analysis tells us Vout as Vin changes Input is usually considered to be a step or ramp (from 0 to VDD or vice-versa)
40
60
80
15 / 35
40
60
Idsn80 = (t)
0
2 (VDD
V )2
Vout (t) 2
VDD Vt
Delay Denitions
tpdr : rising propagation delay
V mm From input to rising output crossing80 DD /2 40 60 100 120
tcdr : rising contamination delay tcdf : falling contamination delay tcd : average contamination delay
tpd = (tcdr + tcdf )/2
ECE Department, University of Texas at Austin Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 17 / 35
80
60
80
18 / 35
Delay Estimation
mm 40 60 80 100 120
The step response usually looks like a rst order RC response 40 with a decaying exponential Use RC delay models to estimate delay
60 C = total capacitance on output node Use eective resistance R So that tpd = RC Depends on average current as gate switches 80
19 / 35
Example: Sizing 3-Input NAND Gate for Equal Rise and Fall Times
mm 40 60 80 100 120
40
60
Determine the transistor widths to achieve eective rise and fall resistances (times) equal to that of a unit inverter R 80 Annotate the 3-input NAND gate with gate and diusion capacitances
ECE Department, University of Texas at Austin Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 20 / 35
40
60
Determine the transistor widths to achieve eective rise and fall resistances (times) equal to that of a unit inverter R
80
Annotate the 3-input NAND gate with gate and diusion capacitances
ECE Department, University of Texas at Austin Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 21 / 35
22 / 35
23 / 35
24 / 35
Elmore Delay
Finding the delay of ladder networks ON transistors look like resistors
mm 40 60 80 100 120
Ritosource Ci
80
NOTE: Ci includes all the o-path capacitance on nodes that are connected to node i
ECE Department, University of Texas at Austin Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 25 / 35
40
60
80
tpdr = (6 + 4h)RC
Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 26 / 35
60
tpdf = (2C)
80
R + (6 + 4h)C 2 = (7 + 4h)RC
R R + 2 2
27 / 35
40
60
80
28 / 35
40
60
80
R R R R 9C + 5C + + 7C + 3RC = 12.33RC 3 3 3 3 o-path Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 ECE Department, University of Texas at Austin Lecture 5. CMOS Delay =
29 / 35
80
30 / 35
80
100
120
80
100
120
Delay Components
mm 40 60 80 100 120
Delay has two parts Parasitic Delay 6 or 7 RC 40 Independent of Load Eort Delay 60 4h RC Proportional to load capacitance
80
33 / 35
Contamination Delay
Minimum (Contamination) Delay
mm 40 60 80 100 120
Best-case (contamination) delay can be substantially less than propagation delay Example, If both inputs fall simultaneously Important for hold time (will see later in the course) 40
60
80
tcdr = (3 + 2h)RC
Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 34 / 35
Diusion Capacitance
We assumed contacted diusion on every source/drain Good minimizes diusion area mm layout 40 60 80 100 Example, NAND3 layout shares one diusion contact
Reduces output capacitance by 2C Merged uncontacted diusion might help too 40 120
60
80 These general observations can be used for initial estimates of area and performance using tools to extract parasitics will provide more accurate results for a particular technology
ECE Department, University of Texas at Austin Lecture 5. CMOS Gates: DC and Transient Behavior J. A. Abraham, September 12, 2011 35 / 35