1 Chapter 6 Problem Set

Chapter 6
PROBLEMS
1. [E, None, 4.2] Implement the equation X = ((A + B) (C + D + E) + F) G using complemen-
tary CMOS. Size the devices so that the output resistance is the same as that of an inverter
with an NMOS W/L = 2 and PMOS W/L = 6. Which input pattern(s) would give the worst and
best equivalent pull-up or pull-down resistance?
Solution
Rewriting the output expression in the form X = ((A + B) (C + D + E) + F) G = ((AB +
CDE)F) + G allows us to build the pulldown network by inspection (parallel devices imple-
ment an OR, and series devices implement an AND). The pullup network is the dual of the
pulldown network.

A B F
24 24 12
C 24 D 24 E 24

G 12
X
A 8 C 12 G 2
B 8 D 12
E 12

F 4

The plot shows sizes that meet the requirement - in the worst case, the output resistance
of the circuit matches the output resistance of an inverter with NMOS W/L=2 and PMOS
W/L=6.
The worst case pull-up resistance occurs whenever a single path exists from the output
node to Vdd. Examples of vectors for the worst case are ABCDEFG=1111100 and 0101110.
The best case pull-up resistance occurs when ABCDEFG=0000000.
The worst case pull-down resistance occurs whenever a single path exists from the out-
put node to GND. Examples of vectors for the worst case are ABCDEFG=0000001 and
0011110.
The best case pull-down resistance occurs when ABCDEFG=1111111.
2. Implement the following expression in a full static CMOS logic fashion using no more than
10 transistors:

Y = (A ⋅ B) + (A ⋅ C ⋅ E) + (D ⋅ E) + (D ⋅ C ⋅ B)

Solution
2 Chapter 6 Problem Set

The circuit is given in the next figure.

VDD

C B
A

D E

Y
C
A D

B E

3. Consider the circuit of Figure 6.1.

VDD

B W/L=16
W/L=8 W/L=8 Cx1
D C A W/L=16
Y

B A W/L=12
W/L=12
Cx2
C W/L=12

Cx3
D W/L=12

Figure 6.1 CMOS combinational logic gate.

a. What is the logic function implemented by the CMOS transistor network? Size the NMOS
and PMOS devices so that the output resistance is the same as that of an inverter with an
NMOS W/L = 4 and PMOS W/L = 8.
Solution
The logic function is : Y = ( A + B )CD . The transistor sizes are given in the figure
above.
b. What are the input patterns that give the worst case tpHL and tpLH. State clearly what are the
initial input patterns and which input(s) has to make a transition in order to achieve this
maximum propagation delay. Consider the effect of the capacitances at the internal nodes.
Solution
The worst case tpHL happens when the internal node capacitances (Cx2 and Cx3) are
charged before the high to low transition. The initial states that can cause this are:
ABCD=[1010, 1110, 0110]. The final state is one of: ABCD=[1011, 0111].
Digital Integrated Circuits - 2nd Ed 3

The worst case tpLH happens when Cx1 is charged before the low to high transition. The
input pattern that can cause this is: ABCD=[0111] =>[0011].
c. Verify part (b) with SPICE. Assume all transistors have minimum gate length (0.25µm).
Solution
The two cases are shown below.

Figure 6.2 Best and worst tpHL.

Figure 6.3 Best and worst tpLH.

d. If P(A=1)=0.5, P(B=1)=0.2, P(C=1)=0.3 and P(D=1)=1, determine the power dissipation

in the logic gate. Assume VDD=2.5V, Cout=30fF and fclk=250MHz.
Solution
Since D is always 1, the circuit implements the following function Y = ( A + B )C .
P(A+B)=1 = PA=0.PB = 0 = 0.5*(1-0.2) = 0.4,
P(A+B)=0 = 1 - 0.4 = 0.6,
PY = 0 = P(A+B) = 1.PC = 1 = 0.6*0.3 = 0.18
PY = 1 = 1 - 0.18 = 0.82
PY = 0 =>1= 0.18*0.82 = 0.1476
So Pdyn = PY = 0 =>1CoutVDD2fclk = (0.1476)(30.10-15)(2.52)(250.106) = 6.92 µ W.

4. [M, None, 4.2] CMOS Logic

a. Do the following two circuits (Figure 6.4) implement the same logic function? If yes, what
is that logic function? If no, give Boolean expressions for both circuits.
4 Chapter 6 Problem Set

Solution
Yes, they implement the same logic function :
F = (ABCD + E) = (A + B + C + D).E
b. Will these two circuits’ output resistances always be equal to each other?
Solution
No
c. Will these two circuits’ rise and fall times always be equal to each other? Why or why not?
Solution
No. Circuit B appears optimized for the case where the transistor with input E is on the
critical path since it is closer to the output node than in circuit A. Therefore, if input E arrives
later, circuit B will be faster than circuit A since the internal node will already be charged and
only the output capacitance needs to be switched. Even if we assume, all inputs arrive at the
same time, however, the two circuits rise and fall times will not be equal to each other. Con-
sider an input combination where E,A,B,C,D are all low. Circuit A has only one body-
affected device while circuit B has four. Since the associated rise in Vt and fall in output resis-
tance affects only one resistor in circuit A, but four parallel resistors in circuit B, we expect a
difference in the timing waveforms.
VDD
VDD
E 6
A 6 B 6 C 6 D 6

A 6 B 6 C 6 D 6
E 6
F
F
A 4
A 4

B 4 4
B

C 4 C 4

D 4 E 1 D 4 E 1

Figure 6.4 Two static

Circuit A Circuit B CMOS gates.

5. [E, None, 4.2] The transistors in the circuits of the preceding problem have been sized to give
an output resistance of 13 kΩ for the worst-case input pattern. This output resistance can vary,
however, if other patterns are applied.
a. What input patterns (A–E) give the lowest output resistance when the output is low? What
is the value of that resistance?
Solution
The lowest output resistance is obtained when all inputs (A, B, C, D and E) are equal to
1. In that case, the output resistance is the parallel of the resistance of a nMOS of width 1,
with a series of four equal nMOS of width 4. Both combinations have the same resistance,
equal to the worst-case output resistance, 13 k Ω . Then the output resistance, in this case, is
half this value, 6.5 k Ω .
b. What input patterns (A–E) give the lowest output resistance when the output is high? What
is the value of that resistance?
Digital Integrated Circuits - 2nd Ed 5

Solution
The lowest output resistance is obtained when all inputs are equal to zero. Each of the
pMOS have the same width, so all of them have the same resistance. The worst case resis-
tance happens when only one of the inputs (A, B, C or D) is equal to 0 while all the rest are
equal to 1. The output resistance in that case is the series of the resistance of two of the pMOS
and it is equal to 13 k Ω . Then, each of the pMOS has an output resistance equal to 6.5 k Ω .
The output resistance is equal to the series of one of these resistance with the parallel of four
of the same resistnaces. Then, the minimum output resistance is 6.5 k Ω + 6.5 k Ω /4 =
8.125 k Ω .
6. [E, None, 4.2] What is the logic function of circuits A and B in Figure 6.5? Which one is a
dual network and which one is not? Is the nondual network still a valid static logic gate?
Explain. List any advantages of one configuration over the other.
VDD VDD

A B A B

B A B A
F F
B B B B

A A A A

Circuit A Circuit B Figure 6.5 Two logic functions.

Solution
Both circuits A and B implement the XOR logic function. Circuit A is a dual network
because the pull up network is dual with the pull down network.
However, circuit B is still a valid static logic gate, because for any combination of the
inputs, there is either a low resistance path from VDD or ground to the output. Circuit B has an
extra advantage. The internal node capacitances are less compared to Circuit A, which make
it faster than Circuit A.
7. [E, None, 4.2] Compute the following for the pseudo-NMOS inverter shown in Figure 6.6:
a. VOL and VOH
Solution
To find VOH, set Vin to 0, because VOL is likely to be below VT0 for the NMOS. If
Vin=0, then M1 is off, so the PMOS pulls the output all the way to the rail. So,
VOH=VDD=2.5V.
To find VOL, set Vin = VOH = 2.5V. The NMOS is all the way on, but so is the PMOS.
To find VOL, we can write a current balancing equation at the output node: IDP+IDN=0. First,
we must determine the region of operation for each device. We can assume that VDS = VOL
for the NMOS is less than VDSAT, so the NMOS is in the linear region. V DS for the PMOS will
be more negative than VDSAT, and VGTp = -2.1, so the PMOS is velocity saturated. The equa-
tion is therefore:

k' p ⋅ W
----- ⋅ V DSAT ⋅ ( VGT – 0.5V DSAT ) ⋅ ( 1 + λVDS ) + k' n ⋅ W
----- ⋅ V o ⋅ ( V GT – 0.5Vo ) ⋅ ( 1 + λV o ) = 0
L L

Plugging in numbers (process parameters such as VDSAT appear in tables in previous

chapters) gives:
6 Chapter 6 Problem Set

– 30 ⋅ 2 ⋅ – 1 ⋅ ( – 1.6 ) ⋅ ( 1 – 0.1 ( V o – 2.5 ) ) + 115 ( 16 ) ⋅ V o ⋅ ( 2.07 – 0.5V o ) ⋅ ( 1 + 0.06V o ) = 0

Solving for Vo gives VOL = 31.6mV.

b. NML and NMH
Solution
Rather than calculating the derivative of the current, we will estimate V IL and V IH from
the simulated VTC. This approach estimates that the noise margin low is about 0.47Vand the
noise margin high is about 1.67V.
c. The power dissipation: (1) for Vin low, and (2) for Vin high
Solution
For Vin low, the NMOS is off, so the power dissipation is 0W. For Vin high,
P=VI=2.5*IDP. We saw in part a) the equation for IDP. Plugging in the value for VOL, we get
P=VI=2.5*120 µ A=300 µ W.
d. For an output load of 1 pF, calculate tpLH, tpHL, and tp. Are the rising and falling delays
equal? Why or why not?
2.5 V
PMOS

M2 W/L = 0.5µm/0.25µm

Vout

M1 W/L = 4µm/0.25µm
Vin

NMOS Figure 6.6 Pseudo-NMOS inverter.

Solution
We cannot use the estimate of resistance from the I-V curve for the HL transition
because the PMOS is still on. Therefore, we will use the average current method for estimat-
ing delay. The average current for the HL transition through the PMOS is 0.5(IVDD=2.5 +
IVDD=1.25). IVDD=2.5 = 0. IVDD=1.25 = -30(2)(-1)(-2.1+0.5) *(1+0.1(1.25)) = 108uA. Thus, Iavg
for the PMOS is 54uA.
For the NMOS, IVDD=2.5 = 115(16)(0.63)(2.07-.63/2)(1+0.06*2.5)=2.4mA and
IVDD=1.25 = 115 (16) (0.63) *(2.07-.63/2)(1+0.06*1.25) = 2.2mA. So, Iavg for the NMOS is
2.3mA. The average current discharging the capacitor is then 2.3mA-54uA = 2.25mA. Then
tpHL = C*delV/Iavg = 556ps.
For tpLH, the NMOS is off, so we can use equivalent resistance to find the transistion
time. From the table of resistances in the text, we can calculate REQ = 31k Ω /(W/Lp) =
15.5k Ω . Then tpLH = 0.69*C*REQ. So tpLH = 10.7ns.
tp = (tpLH + tpHL)/2 = 5.6ns. The rising delay is much longer because the PMOS is very
weak relative to the NMOS.
8. [M, SPICE, 4.2] Consider the circuit of Figure 6.7.
a. What is the output voltage if only one input is high? If all four inputs are high?
Solution
Digital Integrated Circuits - 2nd Ed 7

 2
W  V
– ----------------- ⋅ ( 1 + λ ⋅ V DS )
I = k′ ⋅ ----- ⋅ V ⋅V min
D L  GT min 2
 

Consider a case when one input is high: A = VDD and B = C = D = 0 V. Assume that
Vout is small enough that Vmin = VDSAT for the PMOS device, and Vmin = VDS = Vout for the
NMOS devices. Solve for Vout by setting the drain currents in the PMOS and NMOS equal to
each other, |IDP| = |IDN|, where the drain currents are functions of Vout, VDD, and the device
parameters.
Vout = 102 mV, and ID = 35.7 µA.
Now verify that the assumptions for Vmin are correct. For the PMOS: VDS = -2.34 V,
VDSAT = -1 V, VGT = -2.1 V, therefore Vmin = VDSAT. For the NMOS: VDS = 102mV, VDSAT =
630mV, VGT = 2.07 V, therefore Vmin = VDS.
Consider the case when all inputs are high: A = B = C = D = VDD. For these hand calcu-
lations, this is numerically equivalent to a circuit with a single NMOS device with W/L =
4*1.5 and its gate tied to VDD. Now, the analysis used above for the case when one device is
on can be reused, replacing W/L of the NMOS with 6, and using the same assumptions for
Vmin. Vout = 25 mV, and ID = 35.9 µA. The assumptions for Vmin are correct.
b. What is the average static power consumption if, at any time, each input turns on with an
(independent) probability of 0.5? 0.1?
Solution
Notice in part a) that the drain current in the PMOS is 35.7 µA with one NMOS on and
35.9 µA with four NMOS devices on. The current in the PMOS can be approximated as 35.8
µA when any number of NMOS devices are on and 0 µA when all four are off. The probabil-
ity that all four NMOS devices are off is (1−ρ)4 where ρ is the probability an input is high.
Therefore,

4 4
P AVG = P OFF ⋅ ( 1 – ρ ) + P ON ⋅ 1 – ( 1 – ρ )

where POFF = 0 W, and PON = 89.5 µW. PAVG = 83.9 µW when ρ = 0.5 and PAVG = 30.7
µW when ρ = 0.5.
c. Compare your analytically obtained results to a SPICE simulation.
Solution
From SPICE: Vout = 98.7 mV, and ID = 38.2 µA with one NMOS device on and Vout =
23.5 mV, and ID = 38.3 µA with all NMOS devices on.

VDD

(W/L) = 0.6

F
A B C D
(W/L) = 1.5

Figure 6.7 Pseudo-NMOS gate.

9. [M, None, 4.2] Implement F = ABC + ACD (and F) in DCVSL. Assume A, B, C, D, and their
complements are available as inputs. Use the minimum number of transistors.
8 Chapter 6 Problem Set

Solution
VDD

F F

A A A A C

B C B C C D

C D

10. [E, Layout, 4.2] A complex logic gate is shown in Figure 6.8.
a. Write the Boolean equations for outputs F and G. What function does this circuit
implement?
Solution
G=A(XOR)B
F=A(XNOR)B
b. What logic family does this circuit belong to?
Solution
It belongs to the DCVSL logic family.
c. Assuming W/L = 0.5u/0.25u for all nmos transistors and W/L = 2u/0.25u for the pmos tran-
sistors, produce a layout of the gate using Magic. Your layout should conform to the fol-
lowing datapath style: (1) Inputs should enter the layout from the left in polysilicon; (2)
The outputs should exit the layout at the right in polysilicon (since the outputs would prob-
ably be driving transistor gate inputs of the next cell to the right); (3) Power and ground
lines should run vertically in metal 1.

VDD

F G

A A A A

B B

Figure 6.8 Two-input complex logic gate.

d. Extract and netlist the layout. Load both outputs (F,G) with a 30fF capacitance and simu-
late the circuit. Does the gate function properly? If not, explain why and resize the transis-
tors so that it does. Change the sizes (and areas and perimeters) in the HSPICE netlist.
Solution
The gate doesn’t function properly, because the PMOS devices are strong and the
NMOS pull down network can not switch the output nodes .
Digital Integrated Circuits - 2nd Ed 9

If you decrease the PMOS sizes to W=0.5um, then the logic gate will function prop-
erly.
11. Design and simulate a circuit that generates an optimal differential signal as shown in Figure
6.9. Make sure the rise and fall times are equal.
A Y Y
Y
A 0 0 1
Y
1 1 0

Y
50% of VDD
Y

Figure 6.9 Differential Buffer.

Solution
The circuit is shown below.

A A A A

Y Y

A A A A

If the inverters are sized for equal rise and fall times then you can achieve equal rise
and fall times on the differential outputs, as long as the other FETs are sized symmetrically.
12. What is the function of the circuit in Figure 6.10?

A B

Y Y

Figure 6.10 Gate.

Solution
The circuit implements an S-R latch. Set is A and Reset is B. The invalid state is when
both A and B are 0.
13. Implement the function S = ABC + ABC + ABC + ABC, which gives the sum of two inputs
with a carry bit, using NMOS pass transistor logic. Design a DCVSL gate which implements
the same function. Assume A, B, C, and their complements are available as inputs.
Solution
10 Chapter 6 Problem Set

The two cases are shown in the figure below.

B B C C

A
S

A
Pass Gate Implementation
VDD

S S

A A A A

B B B B B B B B

C C

DCVSL Implementation

14. Describe the logic function computed by the circuit in Figure 6.11. Note that all transistors
(except for the middle inverters) are NMOS. Size and simulate the circuit so that it achieves a
100 ps delay (50-50) using 0.25µm devices, while driving a 100 fF load on both differential
outputs. (VDD = 2.5V). Assume A, B and their complements are available as inputs.
Y Y

A M1 A M1 M1 A M1 A

B M2 B M2 M2 B M2 B

A M3 A M3 M3 A M3 A

B M4 B M4 M4 B M4 B

Figure 6.11 Cascoded Logic Styles.

Digital Integrated Circuits - 2nd Ed 11

For the drain and source perimeters and areas you can use the following approxima-
tions: AS=AD=W*0.625u and PS=PD=W+1.25u.
15.
Solution
The circuit implements an XOR. The sizes of the transistors are M1: 28u/0.25u, M2:
28u/0.25u, M3: 10u/0.25u, M4: 10u/0.25u. MPinv: 4u/0.25, MNinv: 0.375u/10u
16. [M, None. 4.2] Figure 6.12 contains a pass-gate logic network.
a. Determine the truth table for the circuit. What logic function does it implement?
Solution
The truth table is shown below

AB Out

00 1

01 0

10 0

11 1

The circuit implements an XNOR.

b. Assuming 0 and 2.5 V inputs, size the PMOS transistor to achieve a VOL = 0.3 V.
Solution
The PMOS device will be velocity saturated and the NMOS passgate will be in the lin-
ear region. IDN+IDP=0, so

W- ⋅ V
k' p ⋅ ---- W- ⋅ V ⋅ ( V
L DSAT ⋅ ( V GT – 0.5V DSAT ) ⋅ ( 1 + λVDS ) + k' n ⋅ ----
L o GT – 0.5V o ) ⋅ ( 1 + λV o ) = 0

We know that Vo=0.3V, so we can plug in numbers and solve for W/L for the PMOS is
7. Let the PMOS be 1.75/0.25.
c. If the PMOS were removed, would the circuit still function correctly? Does the PMOS
transistor serve any useful purpose?

1.5/.25 Out

B Figure 6.12 Pass-gate network.

Solution
No. If the PMOS were removed, the output node could remain low when AB=00
because it would be floating. The PMOS device pulls the output node high when it would oth-
erwise be in a high impedence state.
17. [M, None, 4.2] This problem considers the effects of process scaling on pass-gate logic.
12 Chapter 6 Problem Set

a. If a process has a tbuf of 0.4 ns, Req of 8 kΩ, and C of 12 fF, what is the optimal number of
stages between buffers in a pass-gate chain?
Solution
m opt = 1.7 t p ⁄ ( R eq ⋅ C ) = 3.47 ≈ 3 gates between buffers.
b. Suppose that, if the dimension of this process are shrunk by a factor S, Req scales as 1/S2, C
scales as 1/S, and tbuf scales as 1/S2. What is the expression for the optimal number of buff-
ers as a function of S? What is this value if S = 2?
Solution tp ⁄ S
2
S ⋅ tp
m opt = 1.7 --------------------------------
- = 1.7 ---------------- = 4.9 ≈ 5 gates between buffers.
2
R eq ⁄ S ⋅ C ⁄ S R eq ⋅ C
18. [C, None, 4.2] Consider the circuit of Figure 6.13. Let Cx = 50 fF, Mr has W/L = 0.375/0.375,
Mn has W/Leff = 0.375/0.25. Assume the output inverter doesn’t switch until its input equals
VDD/2.
a. How long will it take Mn to pull down node x from 2.5 V to 1.25 V if In is at 0 V and B is
at 2.5V?
Solution
To determine the time required for these transitions, we will find the average currents
in the FETs Mr and Mn. The equivalent resistance method will not suffice since it does not
account for both devices being on.
For Mr, IVDD=2.5 = 0 since VDS = 0. For the other case, the PMOS device is velocity sat-
urated, so:
IVDD=1.25 = (-30)(1)(-1)(-2.1+0.5)(1+0.1*1.25) = -54uA. The average current in the
PMOS is -27uA.
Mn is in the velocity saturation region for both endpoints of the transition. The two cur-
rents are therefore:
IVDD=2.5 = (115)(1.5)(0.63)(2.07-0.63/2)(1+0.06*2.5) = 219uA.
IVDD=1.25 = (115)(1.5)(0.63)(2.07-0.63/2)(1+0.06*2.5) = 205uA.
And the average current in the NMOS is 212uA.
The total current DISCHARGING the capacitor is 211uA - 27uA = 185uA.
The time for the transition is then

t = C ⋅ ∆V- = 50fF
--------------- ⋅ 1.25V- = 338ps .
-------------------------------
I avg 185µA

b. How long will it take Mn to pull up node x from 0 V to 1.25 V if VIn is 2.5 V and VB is
2.5 V?
Solution
For the LH transition, the PMOS “keeper” is off. The NMOS Mn is the only FET that
is on for this transition. We present both methods for finding the pull-up time.
Equivalent Resistance: We need to perform a different sweep for this measurement
than the regular ID vs VDS sweep. In this case, V DS is changing because the source node of the
FET is rising. Since the source voltage is changing, VGS also is reducing as node x rises. This
effectively “turns down” the current the NMOS can sustain. Performing the appropriate
sweep and measuring REQ gives REQ = (11.3k Ω + 34.7k Ω ) / 2 = 23k Ω . Thus,
t = 0.69*C*REQ = 0.69*50fF*23k Ω = 794ps.
Average Current: When x = 0, the pass transistor has a VGS = 2.5 and a VDS = 2.5, so
it is velocity saturated.
Ix=0 = (115)(1.5)(0.63)(2.07-0.63/2)(1+0.06*2.5) = 219uA.
Digital Integrated Circuits - 2nd Ed 13

When x = 1.25, the pass transistor has VDS = 1.25 and VGS = 1.25. It is still velocity
saturated, but notice that VGS has decreased. Thus,
Ix=1.25 = (115)(1.5)(0.63)(1.25-0.43-0.63/2)(1+0.06*1.25) = 59uA.
The average current is then Iavg = 139uA.

C ⋅ ∆V 50fF ⋅ 1.25V
t = ---------------- = -------------------------------- = 450ps .
I avg 139µA

Clearly, the two solutions are not very close together. The actual simulated transition
time is about 644ps. The Iavg approximation underestimates the solution because the true
average current in this case is not close to the average of the endpoints. In a typical inverter
(PMOS pullup and NMOS pulldown), VGS doesn’t change over the transition, so the current
is reasonably linear with VDS. For that case, the average current is close to the average of the
endpoints. In this problem, the pinch-off of VGS-VT in the pass transistor means the average is
closer to the smaller value. Numerical calculation of the average current from an HSPICE sim
gives Iavg = 93uA which would give a transition time of t = 672ps, which is much closer to the
actual value.
c. What is the minimum value of VB necessary to pull down Vx to 1.25 V when VIn = 0 V?
Solution
In order for Mn to pull node x low, the current in Mr must equal or exceed the current
that charges up the capacitor at every point in the transition. The maximum current in Mr
occurs when x = 1.25 V, and it is (from part a) IMr = -54uA. We can write a current equation
for Mn at this point in the transition and solve for VB:
Note that Mn is velocity saturated at this point: 54 = 115(1.5)(0.63)(VB-0.43-
0.63/2)(1+0.06*1.25).
Solving gives VB = 1.207V.
VDD

VDD
Mr
B
M2
In x Out
Mn

M1

Figure 6.13 Level restorer.

14 Chapter 6 Problem Set

19. Pass Transistor Logic

pass transistor network VDD
VDD = 2.5V
B (W/L)2 = 1.5um/0.25um
M2 (W/L)1 = 0.5um/0.25um
x Out
(W/L)ni = 0.5um/0.25um
A Mn1

M1
kn’ = 115uA/V2, kp’ = -30uA/V2
B
VtN = 0.43V, VtP = -0.4V
Mn2

Figure 6.14 Level restoring circuit.

Consider the circuit of Figure 6.14. Assume the inverter switches ideally at VDD/2, neglect
body effect, channel length modulation and all parasitic capacitance throughout this problem.
a. What is the logic function performed by this circuit?
Solution
The circuit is a NAND gate.
b. Explain why this circuit has non-zero static dissipation.
Solution
When A=B= VDD, the voltage at node x is VX=VDD-VtN. This causes static power dissipa-
tion at the inverter the pass transistor network is driving.
c. Using only just 1 transistor, design a fix so that there will not be any static power dissipa-
tion. Explain how you chose the size of the transistor.
Solution
The modified circuit is shown in the next figure.
VDD

VDD
Mr
B
M2

Mn1 x
A Out

B M1

Mn2

The size of Mr should be chosen so that when one of the inputs A or B equals 0, either Mn1
or Mn2, would be able to pull node X to VDD/2 or less.
d. Implement the same circuit using transmission gates.
Solution
Digital Integrated Circuits - 2nd Ed 15

The circuit is shown below.

Out
A

e. Replace the pass-transistor network in Figure 6.14 with a pass transistor network that
computes the following function: x = ABC at the node x. Assume you have the true and
complementary versions of the three inputs A,B and C.
Solution
One possible implementation is shown.

B C

A
C ABC

C
B

20. [M, None, 4.3] Sketch the waveforms at x, y, and z for the given inputs (Figure 6.15). You
may approximate the time scale, but be sure to compute the voltage levels. Assume that
VT = 0.5 V when body effect is a factor.
21. [E, None, 4.3] Consider the circuit of Figure 6.16.
a. Give the logic function of x and y in terms of A, B, and C. Sketch the waveforms at x and y
for the given inputs. Do x and y evaluate to the values you expected from their logic func-
tions? Explain.
Solution
x = AB and y = ABC
The circuit does not correctly implement the desired logic function. This stems from
the fact that x is pre-charged high, and thus node y is discharged as soon as the evaluation
phase starts. Although x is eventually discharged by the first stage, y cannot be charged high
again since it is a dynamic node with no low-impedance path to Vdd (during evaluate). Com-
mon solutions to this problem areto either place an inverter between the two stages (thus
allowing only 0-to-1 transitions on the inputs to each stage during evaluate) as in Domino
logic or employing np-CMOS. The latter is presented in (b).
b. Redesign the gates using np-CMOS to eliminate any race conditions. Sketch the wave-
forms at x and y for your new circuit.
Solution
16 Chapter 6 Problem Set

2.5 V VDD
φ
0V φ Mp
2.5 V
x
A
0V CL = 100 fF
A Ma
2.5 V
y
B Ca = 10 fF
B Mb
0V
z
Cb = 10 fF
φ Mn

Figure 6.15 Dynamic CMOS.

2.5V
x
0V

2.0V
y
0V

1.5V
z 0V

The modified circuit using np-CMOS is shown below together with the waveforms at x
and y. The desired logic function is now correctly implemented
22. [M, None, 4.3] Suppose we wish to implement the two logic functions given by
F = A + B + C and G = A + B + C + D. Assume both true and complementary signals are
available.
a. Implement these functions in dynamic CMOS as cascaded φ stages so as to minimize the
total transistor count.
Solution
Dynamic gates with NMOS pull-down networks cannot be directly cascaded. This
solution uses a domino logic approach.

clk clk
G
F
A B C D

clk clk

b. Design an np-CMOS implementation of the same logic functions.

Solution
Digital Integrated Circuits - 2nd Ed 17

VDD
2.5 V
φ φ φ
0V y
2.5 V x
A C
A
2.5 V
B B φ

C φ
0V

(a)
Figure 6.16 Cascaded dynamic gates.
2.5 V x
0V

2.5 V y
VDD
0V
φ φ
(b) np-CMOS x
A
2.5 V x
0V B C
y
φ φ
2.5 V y
0V

The circuit is shown below

clk clk

A B C D
G
clk F clk

23. Consider a conventional 4-stage Domino logic circuit as shown in Figure 6.17 in which all
precharge and evaluate devices are clocked using a common clock φ. For this entire problem,
assume that the pulldown network is simply a single NMOS device, so that each Domino
stage consists of a dynamic inverter followed by a static inverter. Assume that the precharge
18 Chapter 6 Problem Set

time, evaluate time, and propagation delay of the static inverter are all T/2. Assume that the
transitions are ideal (zero rise/fall times).

φ φ φ φ

IN Pulldown Pulldown Pulldown Pulldown

Network Out1 Network Out2 Network Out3 Network Out4

φ φ φ φ

Figure 6.17 Conventional DOMINO Dynamic Logic.

a. Complete the timing diagram for signals Out1, Out2, Out3 and Out4, when the IN signal
goes high before the rising edge of the clock φ. Assume that the clock period is 10 T time
units.
Solution
The timing diagram is shown below.

0 T 2T 3T 4T 5T 6T 7T 8T 9T 10T

CLK

IN

OUT1

OUT2

OUT3

OUT4

b. Suppose that there are no evaluate switches at the 3 latter stages. Assume that the clock φ
is initially in the precharge state (φ=0 with all nodes settled to the correct precharge
states), and the block enters the evaluate period (φ=1). Is there a problem during the evalu-
ate period, or is there a benefit? Explain.
Solution
There is no problem during the evaluate stage. The precharged nodes remain charged
until a signal propogates through the logic, activating the pull-down network and discharging
the node. In fact, this topology improves the circuit’s robustness in terms of charge sharing
affecting the output for any generic pull-down network, and reduces the body effect in the
pull-down network.
c. Assume that the clock φ is initially in the evaluate state (φ=1), and the block enters the pre-
charge state (φ = 0). Is there a problem, or is there any benefit, if the last three evaluate
switches are removed? Explain.
Solution
There is a problem during the precharge stage. If all precharged nodes are discharged
during the evaluate stage, when the precharge FETs simultaneously turn on, the pull-down
Digital Integrated Circuits - 2nd Ed 19

networks will initially remain on, creating a short circuit. This continues in each gate until the
previous gate charges, disabling its pull-down network.
24. [C, Spice, 4.3] Figure 6.18 shows a dynamic CMOS circuit in Domino logic. In determining
source and drain areas and perimeters, you may use the following approximations: AD = AS =
W × 0.625µm and PD = PS = W + 1.25µm. Assume 0.1 ns rise/fall times for all inputs,
including the clock. Furthermore, you may assume that all the inputs and their complements
are available, and that all inputs change during the precharge phase of the clock cycle.
a. What Boolean functions are implemented at outputs F and G? If A and B are interpreted as
two-bit binary words, A = A1A0 and B = B1B0, then what interpretation can be applied to
output G?
Solution
F = A0B0 + A1B1, G = F(A0B0 + A1B1)
If A and B are interpreted as two-bit binary words, output G is high if A = B: a com-
parator
b. Which gate (1 or 2) has the highest potential for harmful charge sharing and why? What
sequence of inputs (spanning two clock cycles) results in the worst-case charge-sharing
scenario? Using SPICE, determine the extent to which charge sharing affects the circuit
for this worst case.

2.5 V 2.5 V 2.5 V 2.5 V

φ 2/0.25 1/0.25 φ 2/0.25 1/0.25

G
F

1/0.25 A0 0.5/0.25 1/0.25 0.5/0.25 CL

B0 1/0.25 B0 A1 2/0.25 A1

φ 2/0.25 B1 2/0.25 B1

φ 2/0.25

GATE 1 GATE 2
Figure 6.18 DOMINO logic circuit.

Solution
Gate 2 has the higher potential for harmful charge sharing because the capacitance that
contributes to charge sharing is larger than in gate 1.
The sequence of inputs resulting in the worst-case charge sharing is A 0 = B 0 and
A 1 = B 1 for the first cycle. Then A 0 = B 0 and A 1 ≠ B 1 for the second cycle such that
A1/A1 transistor that is on during the second cycle is the same as in the first cycle. For exam-
ple, A0 = B0 = A1 = B1 = VDD in cycle 1 and A0 = B0 = A1 = VDD, B1 = 0 V in cycle 2. This
20 Chapter 6 Problem Set

will cause the charge at the output of gate 2 to be shared with the total parasitic capacitance at
the drains of the A1, A1, and B1 transistors.

25. [M, Spice, 4.3] In this problem you will consider methods for eliminating charge sharing in
the circuit of Figure 6.18. You will then determine the performance of the resulting circuit.
a. In problem 24 you determined which gate (1 or 2) suffers the most from charge sharing.
Add a single 2/0.25 PMOS precharge transistor (with its gate driven by the clock φ and its
source connected to VDD) to one of the nodes in that gate to maximally reduce the charge-
sharing effect. What effect (if any) will this addition have on the gate delay? Use SPICE to
demonstrate that the additional transistor has eliminated charge sharing for the previously
determined worst-case sequence of inputs.
Solution
The additional precharge transistor should charge the node that is shared by the A1 and
A1 transistor drains and the F transistor source. Assuming the gate delay is dominated by the
precharge stage, this will reduce the gate delay by briefly aiding the precharging of gate 2.
SPICE output with additional precharge transistor.
Digital Integrated Circuits - 2nd Ed 21

b. For the new circuit (including additional precharge transistor), find the sequence of inputs
(spanning two clock cycles) that results in the worst-case delay through the circuit.
Remember that precharging is another factor that limits the maximum clocking frequency
of the circuit, so your input sequence should address the worst-case precharging delay.
Solution
The worst-case delay results from A = B for two consecutive cycles. This results in the
maximum charging and discharging of the internal nodes
c. Using SPICE on the new circuit and applying the sequence of inputs found in part (b), find
the maximum clock frequency for correct operation of the circuit. Remember that the pre-
charge cycle must be long enough to allow all precharged nodes to reach ~90% of their
final values before evaluation begins. Also, recall that the inputs (A, B and their comple-
ments) should not begin changing until the clock signal has reached 0 V (precharge
phase), and they should reach their final values before the circuit enters the evaluation
phase.
Solution
The maximum clock frequency is ~4.4 GHz.

26. [C, None, 4.2–3] For this problem, refer to the layout of Figure 6.19.
a. Draw the schematic corresponding to the layout. Include transistor sizes.
Solution

0.5/0.5
Out
B 1.5/0.25
A 1/0.25
C 1.5/0.25
22 Chapter 6 Problem Set

b. What logic function does the circuit implement? To which logic family does the circuit
belong?
Solution
The circuit implements Out = A+BC. It is in the pseudo NMOS family.
c. Does the circuit have any advantages over fully complementary CMOS?
Solution
The circuit uses less area than a fully complementary CMOS implementation.
d. Calculate the worst-case VOL and VOH.
Solution
VOH = VDD = 2.5V. To find VOL, assume that we can combine MB and MC into one
NMOS with W/L = 0.75/0.25. Then the worst case VOL occurs when A=0 and the combined
BC NMOS is on. Assume that VOL is less than VDSATn. Then the NMOS device is in the linear
region. The PMOS device will be velocity saturated. Equating the currents at the output gives:

W W
k' p ⋅ ----- ⋅ V DSAT ⋅ ( V GT – 0.5V DSAT ) ⋅ ( 1 + λV DS ) + k' n ⋅ ----- ⋅ V o ⋅ ( V GT – 0.5V o ) ⋅ ( 1 + λV o ) = 0
L L

The only unknown in this 3rd order polynomial is Vo. Solving for Vo gives VOL=
51.2mV
e. Write the expresions for the area and perimeter of the drain and source for all of the FETs
in terms of λ . Assume that the capacitance of shared diffuusions divides evenly between
the sharing devices. Copy the layout into Magic, extract and simulate to find the worst-
case tpHL time. For what input transition(s) does this occur? Name all of the parasitic
capacitances that you would need to know to calculate this delay by habd (you do not need
to perform the calculation).

Figure 6.19 Layout

of complex gate.

Solution
Call the PMOS device P, and name the other devices by their input signal.
ADP = ASP = 19 λ 2. PDP = PSP = 15 λ .
ASA = 40 λ 2. PSA = 18 λ .
Digital Integrated Circuits - 2nd Ed 23

ADA = (3x8 + 3x12) λ 2 / 2 = 30 λ 2. PDA = 16 λ /2 = 8 λ .

ADB= ADA. PDB = PDA.
ASB = 36 λ 2/2 = 18 λ 2. PSB = 6 λ /2 = 3 λ .
ADC = ASB. PDC = PSC.
ASC = 60 λ 2. PSC = 22 λ .
We can narrow the number of transitions to look at for determining the worst case tpHL.
The worst case capacitance occurs when the internal node between MB and MC is charged up
to VDD. Then the worst case delay will occur when either MA or the MB, MC pair discharges
this capacitance. If the series devices are doing the discharging, we need to consider the case
where MB is initially on and where MB is initially off.
The simulation shows that the worst-case transition occurs over three cycles: ABC =
010 to 000 to 011 produces the worst-case tpHL. This is worse than when MA discharges the
node (ABC = 010 to 110) or when MB is initially on (ABC = 010 to 011).
We could calculate tpHL using either the equivalent resistance method or the average
current method. In either case, CL would include the following parasitic capacitances:
CGDPMOS + CDBPMOS + CGDA(no Miller effect b/c input not changing) + CDBA + CGDB +
CDBB + CGSB + CGDC + CDBC.
27. [E, None, 4.4] Derive the truth table, state transition graph, and output transition probabilities
for a three-input XOR gate with independent, identically distributed, uniform white-noise
inputs.
Solution
The truth table of a three-input XOR gate is:

A B C Y

0 0 0 0

0 0 1 1

0 1 0 1

0 1 1 0

1 0 0 1

1 0 1 0

1 1 0 0

1 1 1 1

Table 1: Truth table

As the inputs are independent, identically distribute, uniform white noise, each of the
possible combinations of three input values, has a probability equal to 1/8. From the table, the
probabiliy of having the output equal to 0 is p 0 = 0.5 . In the same way
28. [C, None, 4.4] Figure 6.20 shows a two-input multiplexer. For this problem, assume indepen-
dent, identically-distributed uniform white noise inputs.
a. Does this schematic contain reconvergent fan-out? Explain your answer.
Solution
24 Chapter 6 Problem Set

This schematic has reconvergent fan-out because both inputs of the or gate depend on
the value of S.
b. Find the exact signal (P1) and transition (P0 → 1) formulas for nodes X, Y, and Z for: (1) a
static, fully complementary CMOS implementation, and (2) a dynamic CMOS implemen-
tation.

A X
S C

Z
C
Y
B C
Figure 6.20 Two-input multiplexer
Solution
Assuming a fully complementary CMOS implementation:
X is the output of an AND gate with independent, identically-distributed uniform white
noise inputs. As only when both inputs are equal to 1 the output is 1, P 1 = 0.25 . On the
other hand P 0 → 1 = P 0 P 1 = 0.25 ( 1 – 0.25 ) = 0.1875 .
Y is also the output of an AND gate with independent, identically distributed uniform
white noise inputs. The analysis is the same as with X.
If we represent the truth table of the schematic we will see that P 1 = 0.5 . Then
P 0 → 1 = P 0 P 1 = 0.5 ( 1 – 0.5 ) = 0.25 .
Assuming a dynamic CMOS implementation:
In the same way as before, for X, P 1 = 0.25 . In order to obtain the transition prob-
ability, an n-tree dynamic gate will be assumed. In this case: P 0 → 1 = P 0 = 0.75 .
The analisys for Y is equal to the analysis for X.
For Z, using the truth table of the schematic we obtain, again , P 1 = 0.5 . For the
transition probability, it will be assumed that a np-CMOS structure is used.. Then, Z is the
output of a p-tree dynamic gate. Then: P 0 → 1 = P 1 = 0.5 .
29. [M, None, 4.4] Compute the switching power consumed by the multiplexer of Figure 6.20,
assuming that all significant capacitances have been lumped into the three capacitors shown
in the figure, where C = 0.3 pF. Assume that VDD = 2.5 V and independent, identically-distrib-
uted uniform white noise inputs, with events occuring at a frequency of 100 MHz. Perform
this calculation for the following:
a. A static, fully-complementary CMOS implementation
Solution
Switching power is:
2 2
P SW = α ⋅ f ⋅ C ⋅ V DD = ( α X0 → 1 + α Y0 → 1 + α Z0 → 1 ) ⋅ f ⋅ C ⋅ V DD

We calculated in Problem 27 the probabilities of a 0->1 transistion for each node:

P 0 → 1 for X and Y is 0.1875 and P0 → 1 for Z is 0.25.
Thus, PSW = (2*0.1875+0.25)*100MHz*0.3pF*2.52 = 117.2uW.
b. A dynamic CMOS implementation
Solution
Digital Integrated Circuits - 2nd Ed 25

In Problem 27 for a dynamic np-CMOS gate, we calculated the probabilities: P 0 → 1

for X and Y is 0.75 and P 0 → 1 for Z is 0.5. Thus, PSW = (2*0.75+0.5)*100MHz*0.3pF*2.52 =
375uW.
30. For the circuit shown Figure 6.21 ignore DIBL and S=100mV/decade.
a. What is the logic function implemented by this circuit? Assume that all devices (M1-M6)
are 0.5µm/0.25µm.
Solution
A(B+C)
b. Let the drain current for each device (NMOS and PMOS) be 1µA for NMOS at VGS= VT
and PMOS at VSG= VT. What input vectors cause the worst case leakage power for each
output value? Explain (state all the vectors, but do not evaluate the leakage).
Solution
When the output is high, the worst-case leakage occurs when two transistors leak in
parallel: ABC = 100.When the output is low, the worst-case leakage also occurs when two
transistors leak in parallel: ABC = 110 or ABC = 101.
c. Suppose the circuit is active for a fraction of time d and idle for (1-d). When the circuit is
active, the inputs arrive at 100 MHz and are uniformly distributed (Pr(A =1) = 0.5,
Pr(B=1) = 0.5, Pr(C=1) = 0.5) and independent. When the circuit is in the idle mode, the
inputs are fixed to one you chose in part (b). What is the duty cycle d for which the active
power is equal to the leakage power?
VDD = 2.5V VDD = 2.5V

B M4 M6
A

C M5 Out

A M3 CL = 50fF

B M1 C M2

Figure 6.21 CMOS

logic gate.

Solution
d*Pactive = (1-d) Pleakage. Pactive = α 0->1*f*CL*VDD2 = (3/8 * 5/8)*(100*106)*(50*10-
15
)*(2.52) = 7.3 µ W. – VT
---------
– 0.43
-------------
S 0.1
Pleakage (ABC = 100) = VDD*2*IleakM1 = 5* I o 10 = 5* 1uA10 = 251pW.
Plugging the power numbers into the activity equation and solving for d gives d =
3.4*10-8.
26 Chapter 6 Problem Set

DESIGN PROJECT

Design, lay out, and simulate a CMOS four-input XOR gate in the stan-
dard 0.25 micron CMOS process. You can choose any logic circuit
style, and you are free to choose how many stages of logic to use: you
could use one large logic gate or a combination of smaller logic gates.
The supply voltage is set at 2.5 V! Your circuit must drive an external
20 fF load in addition to whatever internal parasitics are present in your
circuit.
The primary design objective is to minimize the propagation delay
of the worst-case transition for your circuit. The secondary objective is
to minimize the area of the layout. At the very worst, your design must
have a propagation delay of no more than 0.5 ns and occupy an area of
no more than 500 square microns, but the faster and smaller your circuit,
the better. Be aware that, when using dynamic logic, the precharge time
should be made part of the delay.
The design will be graded on the magnitude of A × tp2, the product
of the area of your design and the square of the delay for the worst-case
transition.