The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> OCEAN SCRIPT - MONTE CARLO
https://designers-guide.org/forum/YaBB.pl?num=1122595978

Message started by FastFurious on Jul 28^th, 2005, 5:12pm

Title: OCEAN SCRIPT - MONTE CARLO
Post by FastFurious on Jul 28^th, 2005, 5:12pm

Hi
Does anyone have any brute force ocean script to do device mismatch analysis without using the statistical tool in spectre of Cadence.
I dont have the Statistical tool, and I was wondering is there is a way to use ocean to write a script to do the device mismatch simulation.

cheers

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by Andrew Beckett on Jul 31^st, 2005, 9:20pm

The OCEAN functions for setting up monte carlo also use the statistics license. Essentially OCEAN-controlled simulations use the same licenses as ADE.

Regards,

Andrew.

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by FastFurious on Aug 1^st, 2005, 12:34pm

Thanks Andrew. So no way out but to purchase that statistical tool license then!

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by Geoffrey_Coram on Aug 2^nd, 2005, 9:33am

Maybe no *easy* way out other than purchasing the license ... but if you know enough about statistics and how to generate random numbers, it should be possible. There are some C functions for generating some distributions, you might have to write a wrapper shell script. And you'd have to write the post-processor for the results.

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by sheldon on Aug 5^th, 2005, 5:05am

FastFurious,

Spectre also supports dc mismatch analysis, don't think
it requires a special license. DC mismatch analysis is
equivalent to corner analysis for mismatch. It mismatches
the differential pairs by a fixed amount and calculates the
sensitivity of the output to the mismatch. The ADE -->
Results --> Print --> mismatch summary(?), identifies the
diff pairs in order of their effect on the output. It also
reports the contributions of different factors to the
mismatch, for example, Vt, mobility variations, ... You can
also save the operating point for use with small signal
analysis, the dc operating point includes mismatch, so you
evaluate the effect of mismatch on common-mode
rejection, power supply rejection, etc. The big limitations
are:
1) You need models with mismatch parameters
2) Transient analysis is not supported.

Best Regards,

Sheldon

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by FastFurious on Sep 1^st, 2005, 2:44pm

Hello,
How can I get the mismatch data?
Cheers

sheldon wrote on Aug 5^th, 2005, 5:05am:

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by sheldon on Sep 1^st, 2005, 7:20pm

FastFurious,

Best case the foundry provides you models with dc mismatch supported.
Worst case, you simulate at the transistor level using Monte Carlo
simulation to generate the dc mismatch model :(

Best Regards,

Sheldon

Title: Re: OCEAN SCRIPT - MONTE CARLO
Post by jimi on Sep 6^th, 2005, 10:32pm

Sheldon :

How foundry support models for dcmatch analysis ?
Do they define parameters of mvtwl, mvtwl2, mvt0, mbewl and mbe0 be part of BSIM3v3 models ?
If yes, can I add those parameters by myself from imformation of matching report !?

Thanks!