Monte Carlo simulation


#1

Hello guys,
I would like to simulate capacitors 3% mismatch using Monte Carlo. What I’ve done is:
Process Corners: mc, 3s, montecarloruns= 500, LOT=1, DEV=1 DEVLIST: cpp (c1,c2) [I tried both]
Nothing seems to work correctly. Transient response is not different from the tm simulation, and Histogram does not make sense to me. Is there anything I’m missing? Anything else can be done?

Thanks


#2

Hello Mohamed,

I will see if I can get Risto to respond. I have run Monte Carlo simulations, but generally not through Electric. What you’re doing sounds generally correct but a lot of the hard work of setting up the MC parameters is handled by the simcorners script. If we can’t figure the problem out with a few back-and-forth postings, then one of us can access your account (with your permission) and look at the simulation netlists.

Regards,
Tim


#3

Hi Mohamed,
Yes, in PDK EFXH035B MC on cpp works.

DEVLIST is an obsolete parameter (it’s a no-op, there is no selective control over what models participate). You likely copied a corner-control instance from ef_examples, which is a bug: We need to cleanup old example schematics that still show that parameter. If you place an AnalysisProcessCorners directly from the tech_EFXH035B library it won’t show that parameter.

If you don’t mind, further debug of your issue should be carried out via the helpdesk, not in the forums. In the Design/Verify desktop use the Help tab at lower right. Thanks.


#4

Thanks Sto,
You are right. I copied the corner-control instance from ef_examples.
Now I used the one I have in tech_EFXH035B library (without DEVLIST), and I think it gives me the same results.
The other point is, if there is no usable DEVLIST, where can I put the mismatched components?
Thanks