Hello,

I have some confusion about what the PSS/Pnoise simulation can do.

I am trying to simulate the noise of an comparactor (w/ Autozero function, but that might be inrelavant for this question). I am interested in the output phase noise and the input referred noise voltage.

My understanding is that the normal AC simulation doesn't work in this kind of senario, and I need to use PSS simulation. the simulation bench I built is shown in the picture attached: normal input is applied onto the (noisy) comparator, whose output controls a sampling switch. The switch is used to sample a linear signal, used to present the flow of the time. Thus by measuring the noise on Vt_sampled, we could know the jitter.

The result seems reasonable. But I still am not sure about that.

In my understanding, this way might or might not work. And I need help from you to clearify my confusion.

the reasoning against this method is: the bias condition (Vo, Vin, Vt and Vt_sampled) is taken with PSS, where noise is not applied. Thus when doing Pnoise, the bias (the transit waveform of those signals) stays the same, only AC noise source is applied. As both the trasiant waveform of Vt and Vo(the control of the switch) is not changed, the jitter on Vo is not transferred into noise on Vt_sampled during the pnoise simulation.

the reasoning for this method is: it is known that Pnoise can be used to simulate the osillator phase noise. This means that dispite the fact that bias is determined by PSS, the transfer from voltage to the voltage and phase of every node, and from them to the final output, can be determined by PSS. That is, the relation in figure 2 is possible, even if Vo is not specified in the simulation configurations.

Could you help me? thanks.