Odd DC simulation results using Hspice

Thread Starter


Joined Jul 7, 2010

I am using Hspice to stimate the static power consumption of CMOS gates implemented using a 180nm tech from UMC. I designed a 4-input NOR gate with wn=240 nm, wp=1440 nm and minimum transistor lengths. Simulation after parasitic extraction shown that the design was correct. Later I designed a larger version of the gate (wn=1440nm and wp=6140nm and minimum transistor lengths).

I expected the later gate to be faster and to have a larger static power consumption for every input vector (wider transistors imply a larger subthreshold conduction and larger reverse-bias leakage current).

As expected, the larger gate resulted to be faster, but when all the inputs were set to 0 it had smaller static power consumption than the original (smaller) NOR gate.

Can anybody explain this behavior?

Thanks in advance.


Joined Aug 29, 2009
Can you just simulate a portion of the gate (a subset of transistors?) - with the aim of reducing to the smallest generic circuit quanta that may then allow you to compare your two sizing options in a simpler way - you could then vary parameters to gain a better understanding of how the static loss occuring.

Ciao, Tim