HW 6 - Static Dual and Dynamic Domino Logic

P 7.97, 103, 105

Static Dual Logic

61

64

65

70

The worst case occurs by looking at the longest PMOS and NMOS network path, and dividing the number of transistors over the overall range of W/L.

For PMOS:

(W/L)ACD=(15/1)/3=5/1

and for NMOS:

(W/L)BC=(W/L)BD=(4/1)/2=2/1

Thus for VDD=2.5V:

τPLH=1.2RonPC=1.2⋅1.25⋅10−12F40μAV2]⋅5⋅|1.9V|=3.947 ns

and:

1.2⋅1.25⋅10−12F100μAV2]⋅2⋅1.9V=3.94 ns

Thus τP=3.94 ns.

Using tr=tf=3τP=11.8 ns.

73

a

The device implements Y=(A+B)(C+D)(E+F)―.

b

c

When all NMOS are on, then we get an effective 3 transistors in series (worst-case) and two in parallel, giving:

(W/L)N,eff=6/1â‹…23=4/1

d

For the PMOS we get 3 transistors now in parallel, and 2 in series, giving:

(W/L)P,eff=10/1â‹…32=15/1

Domino Logic

97

103

We're given:

Pasted image 20240306092629.png

a

Assuming all Ai=0 from the circuit, then going into the evaluate phase (clock when high), while we have assumed that Ci's are still uncharged (as we just entered the phase). If A0=1 at this moment, then the voltage at node B has charge sharing between C1,C2 (as the A0 related MOS becomes a short, connecting the tops of C1 and C2).

Thus we can use:

VF=C1V1+C2V2C1+C2

Thus for VB:

VB′=C1V1+C2V2C1+C2

Where we know that V1 prior to the switch must be VDD as it's uncharged, and similar for V2=0V. As such:

VB′=2C2VDD3C2=2VDD3

Hence VB is two- thirds of the VDD.

b

Now when A1=1 is changed then we get a similar scenario, where since C3=C2 then our total capacitance is now 4 times as large. Hence we get charge sharing over all capacitors. Here V1=V2=2VDD3 and V3=0V as before so then we get:

VB″=(3C2)2/3⋅VDD4C2=12VDD

so then the voltage gets halved.

c

In the worst case we have our case from (b) (notice if all are on we get that the capacitor just discharges), hence we need to have C1/C2 be such that:

VB″=(C1+C2)

105