mismatch only and process only

[yxo]

Hello!
I wonder, what the difference is between mismatch only and process only in Analysis Variation in Monte - Carlo Analysis. Thank you.

 

[erikl]

Mismatch only considers only the mismatch of devices, which are located close to each other (adjacent), and which should match as closely as possible. The algorithm used usually is the PELGROM equation, and the mismatch is calculated depending on the W & L sizes, and a (or a few) parameter(s) dedicated to the process and provided by the fab.

Process only only considers the variations of the process, not only of all the chips on one single wafer, but also the variations on different wafers, and even on different lots. The parameter spreading also is provided by the fab and dedicated to the individual process. The spread ususally lies between ±1σ and 3σ (sigma) mean deviations. Actually, these comprise all possible variations between the standard process corners (ss sf tt fs ff).

Of course it is also possible (and desirable) to use both mismatch & process variations in Monte Carlo Analysis. From a circuit design point of view, however, it is important to see the mismatch only variation, in order to decide upon (e.g.) necessary enlargement of devices to achieve better matching.

From process only variation MC simulations you can get a yield estimation for your design depending on the process spreading. Considering both variations will give you an overall yield estimation for your design.

 

[yxo]

thank you. I got it