Re: Monte Carlo and highly matched structures

Correlations can be defined between devices to account for matching.

statistics {
correlate dev=[m1 m2] cc=0.8
}

I have no clue, however, on how to estimate the correlation factor. You should
probably extract a reasonable value from your foundry's matching data, if available,
or from measurements of your own.

You may find the following application note useful for your purpose :
http://sourcelink.cadence.com/docs/files/Application_Notes/2003/spectre_mcmodelingAN.pdf

cheers,

Stéphane

snau-ktv@xxxxxx wrote:
Hi

I'm about to rund Monte Carlo simulations on highly matched structures
(inverters) in th STM 90 nm.
To take corners as well as matching into account I guess I choose the
"process and mismatch" option. (not process only, or mismatch only).

I guess that the simulator does not automatically take into account
that the structures are highly
matched due to a very regular layout?
A netlist from schematics is used.
Does the tool not take the (future) high matching on layout into
account? If not - could I tell it to
do so, in some way?
Thanks.

.