[SOLVED] Can I create a random voltage source for Monte Carlo analysis in Cadence/Spectre/ADE?

Hi there,

I'm trying to model mismatch in a circuit - but instead of using the foundry mismatch model I want to put in data I've extracted from characterizing a test chip we've done.

I'd like to do this by having a bunch of voltage sources with a mean of zero and a sigma of 1V. I can then use VCCS's to add random currents into the nodes I'm concerned about. For those who are interested, one uses Cholesky decomposition of the co-variance matrix to work out the Gm factors.

I've figured out the maths but I'm clueless as to how to make the sigma=1 voltage sources. Can anyone help?

Thanks,

Simon

stc_csr said:

I've figured out the maths but I'm clueless as to how to make the sigma=1 voltage sources.

Click to expand...

Perhaps a verilog-A Gauss distribution might help you?

Hi there,

Thanks for your answer. I'm not sure if that exactly helps - I want to create a set of DC voltage sources such that in Monte Carlo analysis, each will have a random voltage. The random voltages on each should have a gaussian distribution with a sigma of one and a mean of zero, and they should all be uncorrelated.

From these I will create the random currents with appropriate distributions and correlations using ideal transconductors to the target nodes in question.

The link you have provided demonstrates how to create a gaussian pulse - which is not exactly the same thing, but interesting nonetheless. Thanks for that.

Actually I have done the hard work and figured it out myself. It requires typing netlist and "including" it - something I'd hoped to avoid - too bad I guess.

Here's the include file. It seems to work and produces two random voltages - you can add more easily if you want.

Not bad at all! Thanks for reporting back!