Self gain approximation

[mirror_pole]

Hello guys,

I have a question concerning the use of approximation for calculations of parameters like ac gain. Im sure many of you are familiar with the self gain approximation gm*ro>>1, where gm represents the transconductance and ro the output resistance of a MOS.

My question is how to define "much greater then 1", where is the limit of this approximation concerning the fact that the self gain is scaling down with newer technology generations. For example, using a 40nm technology the self gain of a NMOS is less then 10, so im i still allowed to assume gm*ro>>1 for my calculations?

For example if i have: s*c,1*gm,1*(1/ro,1)+s*c,2*(1/ro,1)*1/ro,2 and i assume c,1>c2, is it possible to neglect the second term using the self gain approximation, even with a 40nm technology?

 

[Dominik Przyborowski]

My question is how to define "much greater then 1

Arbitrarily. You have to decide what level of accuracy satisfies you. If 10-20% is OK, then 10 is much greater then 1. If such such accuracy is not enough, then you calculate without simplification.

 

[danadakk]

General speaking if using device as an amplifier with feedback then your
forward G determines gain accuracy when fdbk is applied. This and distor-
tion determine your min acceptable G forward. Talking AC here. DC has some
similar considerations.

There are other considerations of course, such as in comparators where G
helps determine overdrive constraints.

All this is calculable with straightforward analysis.


Regards, Dana.