Is there a reason why the gm/id design methodology is not used in practical design?

[dkdlek93]

Is there a reason why the gm/id design methodology is not used in practical design?

 

[arthorios]

What do you mean by not used in practical design?

As far as I know, most people make their design choices in a way that can always be traced back to gm/id, Vov, inversion coefficient, etc. That's because these parameters are very handy when designing analog circuits, and can be easily used to build a simulation-based look-up table for a given process.

 

[dick_freebird]

I don't use it because it "became a thing" well after
I started doing analog design and I've gotten by
well enough without it.

I'd rather just run local optimization loops in a
simulator, where the netlist encompasses the
devices not-under-consideration as well, and
just pick something that looks centered.

Guys I knew who were much more experienced
and better at it than me, at the time (now they're
mostly retired and/or dead) didn't mess with it
either - they would build and solve for first
derivative=0, the h-parameter equations for
the end-to-end op-amp circuit chain and find
the AVOL peak. You really have to like math for
that, and I don't. Back in the day nobody trusted
the simulators, but I trust them more than any
hand calculation - especially if you like
completeness better than idealistic assumptions.