Hi!

I'm working with the Specter circuit simulator.

I'm trying to get transfer function for the circuit on the first pic. At first step I make test bench for STB analysis and see some poles and zeros. Next I want to replace every element of the initial circuit with it small-signal model to get it's small-signal representation and I'm started from the PMOS.

This PMOS has a current biasing function so I replace it with resistor and capacitor as at second pic. Parameters I get from the OP analysis. Resistor have rout value and capacitor is sum of the cdg, cds and cdb. After simulation I get very different result - poles and zeroes are well shifted from it's original positions, gain and phase margin are increase. I think its predictable because rout is huge resistance so system gets more stable.  Replace rout by ron also don't give desireble result in terms of gain, but what I should to do?

In saturation region small-signal resistance of the MOSFET described by it rout OP parameter. Small-signal model of the MOSFET if pretty straingforward in this case, and circuit simulator do the same things as me - replace MOSFETS with it small-signal models. So why I get such unpleasure results? I don't want to get 100% accuracy, but what I see makes me sad.