Addition of source with know statistical data

Create a file with a ".scs" suffix, and set up the distribution you want (see "spectre -h montecarlo" for more details.

Then reference this file with Setup->Model Libraries in ADE. In the ADE variables, define variable MYDC, and give it a value (this will be your mean value).

// statistics block for my
// design variables

statistics {

   process {
      vary MYDC dist=unif N=0.3
      }

   mismatch {
      vary MYDC dist=gauss std=0.05
   }

}

On an instance of the vsource, you can use the design variable "MYDC". If you want mismatch to be modelled, you'll need to ensure that the vsource is within a subckt - so create an additional level of hierarchy around the vsource in order to do this.

That's it.

Regards, Andrew.

Hi Andrew,

thanks for your fast reply... However that fails.


I have defined a variable in ADE named bgv = 1.2 V
I have defined a new cell including a vsource component whose DC value is the variable "bgv"

I have created a "bgvoltage.scs" file:

statistics{

mismatch{
vary bgv dist=gauss std=15e-3
}

}

Two things:

1. When you reference the "bgvoltage.scs" on the model libraries form, do [u]not[/u] specify a [i]section name[/i]. If you do this, it will expect the statistics block to be within a [i]library[/i] statement, which you don't have.
2. If you want mismatch to work, the voltage source to be within a subckt, as I said before. Essentially mismatch in spectre works by having a parameterized component within a subckt - the mismatch parameters are then made different for each subckt.

The specific problem you're describing here is (almost certainly) down to the first of the two above; I just wanted to mention the second as I don't want you to solve the first only to find that there is no mismatch!

Regards,

Andrew.

Great,

indeed it was caused by the first problem you described !!
Thanks for you help !!

Best regards