Basically the channel is a voltage controlled conductance (gdo) with a high active output impedance when the device is operation in saturation.

Even in saturation the channel has a conductance which is close to gdo, but the impedance at the drain is much higher because the channel is pinched off. The drain conductance contributes a thermal noise current at the drain terminal, so this is why you see references to gdo with a scaling factor (gamma). The active drain conductance is not equal to gdo because the channel charge density is not constant (somewhat triangular distribution) across the channel. For Vds=0, the charge density is ideally constant across the inverted channel and therefore the channel just looks like a thermal resistance. In saturation (pinchoff) the depletion region raises the impedance at the drain, but the thermal noise current comes straight through. So, you can achieve good gain, but the noise is dominated by the channel thermal noise.

Also, to complicate matters, some of the channel thermal noise leaks into the gate via the gate to channel capacitance. Mostly this is taken to be coupling related to Cgs, but really the coupling is distributed along the channel (quasi static model).

I hope this helps, a little, in regards to why the zero bias drain conductance shows up in CMOS LNA discussions.

Thanks Aaron and RFICDude.

Makes perfect sense now. I'll revisit the lna papers and see if this adds to my understanding.

On the question of why I was saying it was hardswitched, for some reason the only way I could visualise this was to think of almost a transistor being used as a switch.

Thanks again,

Thorn