an opamp has 3 fundamental blocks that are ALWAYS there to make it work like it's supposed to
differential amplifier
common source mosfet, high input/output resistance
common drain mosfet, high input resistance, low output resistance
use cap across common drain mosfet to get a dominant pole for stability reasons

large signal operation
looking for superior reproduction of input signal

Vgs of each mosfet tied to corresponding input signal
id1 = 1/2 * uCoxW/L * Vov1 * Vov1
id2 = 1/2 * uCoxW/L * Vov2 * Vov2

play with those, get sqrt(id1) - sqrt(id2) = Vid * sqrt(uCoxW/2L)

square both sides....swap id2 for I-id1, play around, end up with a quadratic
end up with id1 = I/2 +/- sqrt(kn' * WI/L) * Vid/2 * sqrt((1-Vid*vid/4)/(I/(uCoxW/l)))

I/2 is bias point, + branch is id1, - is id2

simplify to id2 = I/2 - I/Vov * Vid/2
gm = 2Id/Vov, if Id = I/2 -> gm = I/Vov

doing some plots, it becomes clear that you have to keep Vid as small as possible to remain in the linear region
can trade gain for better linearity
have to figure out which is more important
want as wide a range as possible on input signal, might not always be a small signal

hw4: online