digitalo
Junior Member level 1
- Joined
- Aug 6, 2011
- Messages
- 15
- Helped
- 1
- Reputation
- 2
- Reaction score
- 1
- Trophy points
- 1,283
- Activity points
- 1,455
Hi all!
We want to upgrate our flow from purely best-typical-worst view flow to AOCV based. Unfortunately, I cannot find much information on how to generate the tables. I'm using the Cadence ETS suite to simulate them from MC models. The models allow me to specify by how much the models are to be varied, in units of sigma. What is a good value? 1 sigma because the tools expect this and scale themselves? Or 3 sigma to catch "everything" already in the table?
And when I have the tables, do I use them together with best-worst? I assume yes, because they are on-chip variations, where best-worst is between wafers. But wouldn't this then go worse than worst (and better than best) in effect, and therefore end outside the acceptance window of the foundry, leading to overly pessimistic designs?
Regards,
digitalo
We want to upgrate our flow from purely best-typical-worst view flow to AOCV based. Unfortunately, I cannot find much information on how to generate the tables. I'm using the Cadence ETS suite to simulate them from MC models. The models allow me to specify by how much the models are to be varied, in units of sigma. What is a good value? 1 sigma because the tools expect this and scale themselves? Or 3 sigma to catch "everything" already in the table?
And when I have the tables, do I use them together with best-worst? I assume yes, because they are on-chip variations, where best-worst is between wafers. But wouldn't this then go worse than worst (and better than best) in effect, and therefore end outside the acceptance window of the foundry, leading to overly pessimistic designs?
Regards,
digitalo