# Cadence Monte Carlo Mismatch simulation - Correlation

Status
Not open for further replies.

#### avt

##### Member level 5
Hi,

could someone explain to me how to determine (or guess) the correlation coeffcients that one can define under the menue option "Simulation/Define Correlations" in the Monte Carlo toolbox under ADE/Spectre for Mismatch Suimulation.

I would assume that the distribution of the mismatch error according to something like sigma=x/sqrt(W*L) (like described in the Pelgrom paper , IEEE JSSC, vol SC-24 pp 1433-1439 October 89) is already delivered bei the MC-simulation library - so I assume that the correlation coefficent determines how good I realized the matching between dedicated transistors in my design - correct ??? - but how the heck can I get then something like correlation coefficient ? (and why can they also be negative ?)

#### avt

##### Member level 5
yes, quite insightful - but this means : I have to measure devices before I'm realy capable to define correlation coefficients on my own, right ? (no other way ...) (no foundry support ? )

Status
Not open for further replies.