Class11 PDF
Class11 PDF
CMOS –Inverters
Lecture 2
2
0.15
1.5
Vout (V)
Vout (V)
0.1
0.05
0.5
Gain = -1
0 0
0 0.5 1 1.5 2 2.5 0 0.05 0.1 0.15 0.2
V (V) V (V)
in
in
W 2
VDSsat
I Dsat = k' (VDD − Vt )VDSsat − 2
L
Propagation Delay
VDD VDD
t pHL = t po × = t po ×
VDD − VTn − VDsatn / 2 VDD − 0.733
Transient Response
3. 0V
2. 0V
tpHL tpLH
0V
- 1. 0V
0s 100ps 200ps 300ps 400ps 500ps
V( 2) V( 3)
Ti me
NMOS/PMOS Ratios
NMOS/PMOS Ratios
CL = (1 + β )(Cdn1 + C gn 2 ) + CW
Propagation delay then becomes: (where Reqn and Reqp
are resistances of identically sized transistors):
0.69 Reqp
tp = [(1 + β )(C dn1 + C gn 2 ) + CW ] Reqn +
2 β
Optimal β then becomes :
∂t p
Reqp C =0
β opt = 1 + W ∂β
Reqn C dn1 + C gn 2
-1 1
Propagation Delay
x 10
5
tpLH tpHL
4.5
4
propagation delays
p
require β=2.4.
3.5
3
1 1.5 2 2.5 3 3.5 4 4.5 5
β
Inverter Sizing
•Assume a symmetrical inverter (PMOS and NMOS are
sized to give identical rise and fall times).
Inverter Sizing
Let Req be the equivalent resistance of
the gate. Then the propagation delay is:
t p = 0.69 Req (Cint + Cext ) = 0.69 Req Cint (1 + Cext / Cint ) = t po (1 + Cext / Cint )
Cint = ηC g
η is the proportionality factor and is near to unity in value.
t p = t po (1 + Cext / ηC g ) = t po (1 + f / η )
Input Output
Cg1 CL
N
Cg1 j CL
N
C g , j = C g , j −1C g , j +1
Thus, given Cg,1 and CL, the sizing factor for each
stage to give the same delay through each stage,
and the minimal total delay is:
f = C L / C g ,1 ≡ F
N N
t p = Nt po (1 + N F / η )
F represents the overall effective fan-out of the circuit
and is equal to:
F = C L / C g ,1
t p = Nt po (1 + N F / η )
For a given F (e.g. given a load capacitance to drive
from a minimally-sized gate), what is the number of
stages that obeys the minimal propagation delay
expression derived above?
If N is too large then the first term dominates
(intrinsic delay of the stages dominates). If N is too
small then the effective fan-out of each stage
becomes too large and the second term dominates.
η = Cint / C g ,1
Knowing η, iteratively solve for f using the derived
expression: (1+η / f )
f =e
We know the load capacitance we are trying to drive,
CL, so we also know F = CL/Cg,1.
t p = Nt po (1 + N F / η )
In Out
1 f f2 CL= 8 C1
C1
f = 3 C L / C1 = 3 8 = 2
N f tp
1 64 1 64 65
1 8 64
2 8=641/2 18
1 4 16 64 3 4=641/3 15
1 64 4 2.8=641/4 15.3
2.8 8 22.6
Homework #1
Using the Level-3 CMOS inverter model
given above, determine the tpHL and tpLH
for 6 to 10 different supply voltage
values from 2.5 to 1.1 V.
Do the simulations again using
Ln=Lp=0.25 µm, Wn=2 µm, and Wp=4 µm.
Homework #2
Determine the sizes (f) of the inverters in
the circuit below such that the delay time
between input and output is minimized. (Hint:
First find the ratios between the devices
that minimize the delay. You should find
that: 4C g , 2 4C g ,3 C L
= =
C g ,1 C g , 2 C g ,3
Homework #2 - Circuit
Input
Output
CL = 64Cg,1
1 2 3