Monticarlo Analyses in Cadence

Dear Suta,

Attached below the picture of my recent simulation with 500 Sample,

Now suppose I want an actual voltage of 1.65 with +-50 mV is accepted,

As you mentioned before that Sigma is a result of simulation and requirement, if I would follow your procedure for 3 sigma....... 50 mV/3 = 16.66 mV.... thus if I want to go for 99.7 % of Yield my VCO max must be no more 1.65 V+ 16.66 mV =1.666
Similarly, VCO min = 1.65-16.6 mV = 1.6334 V. Since I didn't cross this limit in my MC result it means I have satisfied the sigma and the yield condition.

Please correct me if I am wrongly following your procedure.

The other question related to your discussion, you told I should look where Sigma is saturating, what this mean and how can I monitor it please



Thank you very much once again

No, I think if you want 3 sigma yield, your VCOmax should be 1.65+/- 50mV because the 50mV is equal to 3 sigma, while 1 sigma is 16.6mV. Either way your simulations show you are well below these limits.

When I said sigma saturation, I meant that you run, say 100 runs and get some sigma, then run 200 runs and get somewhat different sigma, then run 300 runs, 500 runs, etc and as you increase the number of runs the sigma changes less and less until it doesn't practically change anymore i.e. saturates and this will be kind of the number of MC runs that are giving good results. But this is just for experimental purposes. I think 200 runs already captures most of the statistical variations in the circuit. The rest is just some small adjustments.

Dear Suta,

I understood it now completely from you about Sigma concept and how to see if my samples is enough or not.
I will continue working based on this procedure, if I face some problems I will post it again

I just want to share this to any reader who see this post concerning the difference between the Latin and Random Sampling

https://www.lumina.com/blog/latin-hypercube-vs.-monte-carlo-sampling

https://www.uio.no/studier/emner/matnat/math/STK4400/v05/undervisningsmateriale/Sampling methods.pdf

Thank you very much for your kind help