Hi all (first interaction here),
I have a general question regarding MonteCarlo simulations. Should there be a difference in the result of a MonteCarlo simulation if I use the 'm=100' multiplier on a transistor cell or a vector notation <1:100>? I've played around with a differential pair, looking at the difference in drain current but couldn't reach any specific conclusion. The results seemed to converge to the same level, but not exactly, with something like a 2~3% error (1000 iterations).
I'm aware that using the 'm' multiplier is generally better for simulation speed, as the instance won't be multiplied 'm' times on the netlist. But during an MC simulation using 'm' also means that all those transistors will vary on the same direction (and value). Meanwhile having different instances with vector notation <> will make all the transistors go to different directions.
Intuitively, I'd say both approaches should converge to the same result. But maybe there is difference in the amount of MC iterations one or the other needs to reach the same confidence level? Any preferred approach or thoughts?
Thank you very much!