The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Increasing the standard deviation of models in cadence
https://designers-guide.org/forum/YaBB.pl?num=1590598119

Message started by dnw on May 27^th, 2020, 9:48am

Title: Increasing the standard deviation of models in cadence
Post by dnw on May 27^th, 2020, 9:48am

Hi Is there a way to increase the sigma of stat models in cadence?
I am looking for a faster way to run 6,9 sigma cases
Thanks

Title: Re: Increasing the standard deviation of models in cadence
Post by Andrew Beckett on Jun 7^th, 2020, 6:43am

What are you aiming to achieve? There is a capability of the Virtuoso Variation Option used with ADE Explorer/Assembler for doing high-yield verification using an approach called Scaled Sigma Sampling which allows you to verify the yield with high sigma targets like six sigma. This approach (giving an over-simplified description) runs a monte carlo simulation with various scale factors of the sigma, and then from the results can then compute the yield. Because it scales up the sigma, it has the benefit of actually simulating points in the failure regions - trying to simulate 6-sigma or beyond in normal monte carlo is very unlikely to simulate anything in the tails because the probability of them happening is very small.

By the way, there is no tool called "cadence", so it helps if you can be precise about which tool/product you're using. Cadence is the name of the company providing EDA tools, not the name of a software suite (and Cadence produces a very wide range of EDA software, so you could be talking about many things).

Andrew.

Title: Re: Increasing the standard deviation of models in cadence
Post by dnw on Jun 7^th, 2020, 9:54am

Hi Andrew,
Sorry for loosely using Cadence name. I am using Virtuoso and running sims using ADEXL.
What I am trying to achieve is look at the tail with less number of simulation. I want to make sure I have enough margins on those corners.
Also to see if the startup is happening properly and if there is any failure case for that.

I think what you mentioned scaled sigma sampling is what I am looking for. I will try to google that also if you have some reference for that it will be really helpful.
Thanks