In the indirect compensation scheme used in the book you use 1 capacitor which is traditionally
the way I have used it as well. I have, however, seen people use two capacitors when doing the
compensation, one to the NMOS cascode and one for PMOS cascode but I have never heard of
a satisfactory answer for why they do this. Is it to make the circuit symmetrical so that the PSRR
from N-side and P-side are they same?
There is no need for two capacitors for small-signals, a single capacitor will work fine and is
preferred for power-supply rejection reasons as discussed in the book. However for larger signals
it is possible for one side to start shutting off or simply have a considerably different small-signal
equivalent behavior. Using two paths helps to ensure that the op-amp is always compensated for
A simple example of when one cap alone for compensation can fail is when you have a step input
and the side of the diff-amp that is connected to the compensation capacitor shuts off (no compensation
while the output changes, and you will get ringing and a recovery time).