PROBLEM # 11.1 Meshack Pavan.

Estimate the noise margins for the inverters used to generate Fig. 11.4.

Solution: -

From the graphs in Fig. 11.4 and the text in p11.3, we have

VOH = 5V
VOL = 0V
VIL = 1.8V
VIH = 2.1V Long channel
Process
Therefore, noise margins,
NMH = VOH - VIH = 5 – 2.1 = 2.9V
NML = VIL - VOL = 1.8 – 0 = 1.8V

For the short channel process, similarly from the graph and text in the book,

VOH = 1V
VOL = 0V
VIL = 400mV
VIH = 500mV

So, noise margins,

NMH = VOH - VIH = 1 – 0.5 = 0.5V = 500mV
NML = VIL - VOL = 0.4 – 0 = 0.4V = 400mV
Assignment #10 Vinay Dindi EE510 Due Date-4/12/04

11.2) Design and simulate the DC characteristics of an inverter with Vsp

approximately equal to VTHn. Estimate the resulting noise margins for the design.

For Vsp~=Vthn, (ßn/ ßp) should be as large as possible.

Let (ßn/ ßp)=90
So Wn=30*Wp assuming KPn=KPp and Ln=Lp
If Wp=10, then Wn=300.
Substituting these values in equation 11.4,
we get Vsp=0.32. (Vtn=Vtp=0.28V from Table 6.3)
From simulations we obtain Vsp~=0.27V.
Problem 11.3 Satish Dulam
Question:
Show that the switching point of three inverters in series is
dominated by the Vsp id the first inverter.

Solution:
Three inverters with different switching points are simulated
as below:
*Stage1 Inverter
.control
destroy all
run
let Icross=-i(vdd)
*plot Icross
plot vout Vin
.endc
.option scale=50n
.dc vin 0 1 1m
vdd vdd 0 DC 1
Vin vin 0 DC 0
M1 vout vin 0 0 NMOS L=1 W=10
M2 vout vin vdd vdd PMOS L=1 W=20
* 50nm BSIM4 models

Stage2 and 3 Inverters with different Switching points:

*Stage2 Inverter
.control
destroy all
run
let Icross=-i(vdd)
*plot Icross
plot vout Vin
.endc
.option scale=50n
.dc vin 0 1 1m
vdd vdd 0 DC 1
Vin vin 0 DC 0
M1 vout vin 0 0 NMOS L=1 W=10
M2 vout vin vdd vdd PMOS L=1 W=200
* 50nm BSIM4 models
*Stage3 Inverter
.control
destroy all
run
let Icross=-i(vdd)
*plot Icross
plot vout Vin
.endc
.option scale=50n
.dc vin 0 1 1m
vdd vdd 0 DC 1
Vin vin 0 DC 0
M1 vout vin 0 0 NMOS L=1 W=10
M2 vout vin vdd vdd PMOS L=1 W=400
* 50nm BSIM4 models
*Three inverters in series
.control
destroy all
run
plot vout vin
.endc
.option scale=50n
.dc vin 0 1 1m

vdd vdd 0 DC 1
Vin vin 0 DC 0
M1 vout1 vin 0 0 NMOS L=1 W=10
M2 vout1 vin vdd vdd PMOS L=1 W=20
M3 vout2 vout1 0 0 NMOS L=1 W=10
M4 vout2 vout1 vdd vdd PMOS L=1 W=200
M5 vout vout2 0 0 NMOS L=1 W=10
M6 vout vout2 vdd vdd PMOS L=1 W=400

* 50nm BSIM4 models

* Don't forget the .options scale=50nm if using an Lmin of 1
* 1<Ldrawn<200 10<Wdrawn<10000 Vdd=1V
* Change to level=54 when using HSPICE

Notice the switching point of the 3stage inverters is around

500mv though the second and third stage switching points are
at 640mv.
Problem 11.4 Rahul Mhatre

Repeat Ex. 11.6 using a PMOS device with a width of 10.

Out
In 10/10 50fF

Figure 11.11 Circuit Schematic for Problem 11.4.

Solution: Replacing the inverter with its digital model, we get the following schematic
VDD VDD VDD

Rp
Cinn =3/2 Coxn Coutp = Coxp

In Out
Cinn =3/2 Coxn Coutn = Coxn
Rn

Figure 11.4.1 Digital Model for the inverter

The total capacitance at the output node is given as

Ctot = Coxp + Coxn + Cload
Ctot = (62.5aF) [WnLn + WpLp] + 50fF
= (62.5aF) [10 + 10] + 50fF
Ctot = 51.25 fF

From Table 10.1,

Rp = 68k/Wp
Rp = 68k/10 = 6.8K

Rn = 34K/Wn
Rn = 34K/10 = 3.4K

TpHL = 0.7 Rn Ctot

TpHL = 0.7 * 3.4K * 51.25fF
TpHL = 122 pS
TpLH = 0.7 Rp Ctot
TpLH = 0.7 * 6.8K * 51.25fF
TpLH = 244pS

The SPICE Simulation netlist is as shown in Figure 11.4.2

*** Problem 11.4 CMOS: Circuit Design, Layout, and Simulation ***
.control
destroy all
run
plot vin vout
.endc

.option scale=50n
.tran 10p 2n UIC

vdd vdd 0 DC 1
Vin vin 0 DC 0 pulse 0 1 500p 0 0 1n 2n

M1 vout vin 0 0 NMOS L=1 W=10

M2 vout vin vdd vdd PMOS L=1 W=10
Cl vout 0 50f

***Models for NMOS and PMOS****

.end
Figure 11.4.2 SPICE netlist for Figure 11.11

The SPICE simulations are as shown in Figure 11.4.3. The TpHL was found to be
100 ps and TpLH was found out to be around 200ps.

Figure 11.4.3 SPICE Simulation for the netlist of Figure 11.1

John Spratt EE 510 Chap 10 HW 4/4/2004

Problem 11.5: Repeat Ex. 11.6 using the long channel process with a 30/10 inverter.

Solution:
*** Top Level Netlist ***
vdd vdd 0 DC 5
Vin vin 0 DC 0 pulse 0 5 1n 0p 0p 1n 2n
C1 Vout 0 50fF

M1 Vout Vin 0 0 NMOS L=1u W=10u

M2 Vout Vin Vdd Vdd PMOS L=1u W=30u

Simulation:

tPHL=69ps

tPLH=67ps
John Spratt EE 510 Chap 10 HW 4/4/2004

Taking into account the output capacitance of the MOS’s:

tPHL=.7*Rp*Ctot=.7*45k/30*(1.75f*(30+10)+50f)=126ps
tPLH=.7*Rn*Ctot=.7*15k/10*(1.75f*(30+10)+50f)=126ps

W/O taking into account the output capacitance of the MOS’s:

tPHL=.7*Rp*Ctot=.7*45k/30*(50f)=52.5ps
tPLH=.7*Rn*Ctot=.7*15k/10*(50f)=52.5ps

W/O appears to be closer to the sim.

Problem 11.6 Steve Bard

Estimate the oscillation frequency of an 11-stage ring oscillator using 30/10 inverters in
the long-channel CMOS process. Compare your hand calculations to the simulation
results.

Solution: The frequency is about 250 MHz when calculated by hand. When simulated on
SPICE, the frequency is higher, reaching 495 MHz. The hand calculations are shown
below, and the SPICE simulation is shown on the next page.

1
f rosc =
n(t PHL + t PLH )

t PHL + t PLH = 0.7(Rn + R p )CTOT

5
CTOT = (C oxP + C oxN )
2

L 15 k
R n = 15 k = = 1 .5 k
W 10
L 45 k
R p = 45 k = = 1 .5 k
W 30

C oxP = 1.75 fF ⋅ W ⋅ L = 52.5 fF

C oxN = 1.75 fF ⋅ W ⋅ L = 17.5 fF

5
CTOT = (52.5 fF + 17.5 fF ) = 175 fF
2

t PHL + t PLH = 0.7(1.5k + 1.5k )(175 fF ) = 368 ps

1
f rosc = ≈ 250MHz
11(368 ps )
From the SPICE simulation, the frequency = 1/T = 1 / 2.02 ns = 495 MHz.

*** Problem 11.6 ***

.control
destroy all
run
plot vout
.endc
.option scale=1u
.tran .1n 6.5n uic

vdd vdd 0 DC 5
R1 vout 0 1MEG

**********D********G*******S*******B****
M1 vout1 vout 0 0 NMOS L=1 W=10
M2 vout1 vout vdd vdd PMOS L=1 W=30
M3 vout2 vout1 0 0 NMOS L=1 W=10
M4 vout2 vout1 vdd vdd PMOS L=1 W=30
M5 vout3 vout2 0 0 NMOS L=1 W=10
M6 vout3 vout2 vdd vdd PMOS L=1 W=30
M7 vout4 vout3 0 0 NMOS L=1 W=10
M8 vout4 vout3 vdd vdd PMOS L=1 W=30
M9 vout5 vout4 0 0 NMOS L=1 W=10
M10 vout5 vout4 vdd vdd PMOS L=1 W=30
M11 vout6 vout5 0 0 NMOS L=1 W=10
M12 vout6 vout5 vdd vdd PMOS L=1 W=30
M13 vout7 vout6 0 0 NMOS L=1 W=10
M14 vout7 vout6 vdd vdd PMOS L=1 W=30
M15 vout8 vout7 0 0 NMOS L=1 W=10
M16 vout8 vout7 vdd vdd PMOS L=1 W=30
M17 vout9 vout8 0 0 NMOS L=1 W=10
M18 vout9 vout8 vdd vdd PMOS L=1 W=30
M19 vout10 vout9 0 0 NMOS L=1 W=10
M20 vout10 vout9 vdd vdd PMOS L=1 W=30
M21 vout vout10 0 0 NMOS L=1 W=10
M22 vout vout10 vdd vdd PMOS L=1 W=30

.MODEL NMOS NMOS LEVEL = 3

.MODEL PMOS PMOS LEVEL = 3
.end
Problem 11.7 Shambhu Roy

Using the long channel process design a buffer with minimum delay (A=2.718) to
insert in between a 30/10 inverter and a 50pF load capacitance. Simulate the
operation of the design.

Solution:
For a 30/10 long channel process using table 10.2

Cin = 3/2(1.75 * 30 + 1.75 * 10) fF = 105 fF

Cout = 70fF

Delay without buffer = 0.7 (1.5k + 1.5k) 50pF = 105 nS

For minimum delay A=e

Number of inverters in buffer N = ln(50pF/105fF) = 6.16

So choosing N=6

1
50 pF 6
A= ( ) =2.79
105 fF
So the 6 Inverters to be used in the buffer are
84/28, 233/78, 651/217, 1817/606, 5071/1690 and 14150/4716

Delay with buffer = 0.7*6*(1.5k+1.5k)(70fF+2.79*105fF)=4.6nS

*** Solution 11.7_long with buffer CMOS: Circuit Design, Layout, and Simulation ***

.control
destroy all
run
plot vin,vout

.endc

.option scale=1u
.tran 10p 8n UIC

vdd vdd 0 DC 5
Vin vin 0 DC 0 pulse 0 5 1n 0 0 2n 4n

M01 v1 vin 0 0 NMOS L=1 W=10

M02 v1 vin vdd vdd PMOS L=1 W=30

M11 v2 v1 0 0 NMOS L=1 W=28

M12 v2 v1 vdd vdd PMOS L=1 W=84

M21 v3 v2 0 0 NMOS L=1 W=78

M22 v3 v2 vdd vdd PMOS L=1 W=233

M31 v4 v3 0 0 NMOS L=1 W=217

M32 v4 v3 vdd vdd PMOS L=1 W=651

M41 v5 v4 0 0 NMOS L=1 W=606

M42 v5 v4 vdd vdd PMOS L=1 W=1817

M51 v6 v5 0 0 NMOS L=1 W=1690

M52 v6 v5 vdd vdd PMOS L=1 W=5071

M61 vout v6 0 0 NMOS L=1 W=4716

M62 vout v6 vdd vdd PMOS L=1 W=14150
Cl vout 0 50p
Problem 11.8 Harish Reddy Singidi

Repeat problem 11.7 using an area factor, A of 8.

Solution: -
A=8
Cin1 = 1.5 (Coxn + Coxp)
Cin1 = 1.5 ((1.75)(30) + (1.75)(10))
Cin1 = 105fF
Cout1 = Coxn1 + Coxp1 = (1.75)(10)(1.75)(30) = 70fF
Rn1 = 15K/10 = 1.5K
Rp1 = 15K/10 = 1.5K
N = ln((Cload/Cin1))/ln(A)
N= 2.9 Stages = 3 Stages

(TPHL + TPLH) = 0.7* N(Rn1+Rp1)(Cout1+A*Cin1)

(TPHL + TPLH) = 0.7* 3*(1.5K+1.5K)(70fF+8*105fF)
(TPHL + TPLH) = 5.73ns
Assignment #10 Vinay Kumar Dindi

11.9) Derive an equation for the switching point voltage, similar to the derivation of Eq(11.4), for
the NMOS inverter seen in Fig11.24a.

To find the switching point voltage, let Vin=Vout=Vsp

At the conditions mentioned above, both M1 and M2 are in saturation.
So Ids1 (sat)=Ids2 (sat)
(ßn2/2)(Vsp-Vthn2)2= (ßn/2)(Vdd-Vsp-Vthn1)
Solving for Vsp gives
Vsp= [Vthn1√ (ßn1/ ßn2) + (Vdd-Vthn2)] / [1 + √ (ßn1/ ßn2)]
If ßn1=ßn2 and Vthn2 =Vthn1, then
Vs= Vdd/2.
Problem 11.10 Rupa Balan

Repeat problem 11.9 for the inverter in fig 11.24c. Note that the PMOS transistor is
operating in the triode region when the input/output are at Vsp.

Given M2 is in the triode region.

For M1 to be in saturation, Vds> Vgs- Vthn

Vd>Vg-Vthn
Vthn>0 ( Vd=Vg=Vsp )
Hence M1 is in saturation.
Current through M1, IDN = (βn/2) (Vsp - Vthn)2

Current through M2, IDP =βp [((Vdd – Vthp) (Vdd –Vsp)) – ((Vdd - Vsp)2/2)]

Equating the currents, (βn/2) (Vsp - Vthn)2= βp [((Vdd – Vthp) (Vdd –Vsp)) – ((Vdd - Vsp)2/2)]

Simplifying we get,
βp/βn = (Vsp - Vthn)2 / [(Vdd – Vthp)2 – (Vsp - Vthp)2

Cross-multiplying we get an equation of the form

aVsp2+bVsp+c =0
where

a= βn + βp
b= -2 (βn Vthn+ βp Vthp)
c= βn Vthn2- βp Vdd2+2 βp Vdd Vthp

This is a quadratic equation which can be solved to get Vsp.