I would like to run a Periodic AC analysis (PAC) at frequencies around a carrier not known a priori.
That is, I have an oscillator whose frequency - of course - is to be determined by simulation. The natural way of simulating an autonomous system and get this information is with the Periodic Steady State (PSS) engine. Around the frequency point found by PSS, I would then like to run a periodic small signal analysis.
Is there a way to specify that from ADE L? I mean some 'special syntax' to use in the PAC in order to signal the simulator that he should use the outcome of the PSS, which in turn is dealing with an autonomous system (i.e. oscillator)?
I couldn't find it rapidly, but maybe somebody can give me a hint?
As a plan B, I was thinking in the direction of ADE XL, where you have a device called "pre-simulation script" which maybe could be used to first assess the oscillation frequency and then pass this info to the next, PAC, simulation as a parameter.
I have never even read about this topic so it would be helpful if anybody knowledgeable would direct me :)
That's exactly what PAC does already. If you run PSS with "oscillator" enabled (i.e. for an autonomous circuit), then the PAC will run on the results of the PSS simulation. If you want to do the frequency sweep relative to the fundamental that is found, you tell the PAC to run "relative" rather than "absolute", and then specify the relative harmonic number (typically 1). The frequencies you then specify in the sweep are relative to the multiple of the fundamental (so 1k to 1M would be Fosc*N+1k to Fosc*N+1M where N is the relative harmonic number).
There is discussion on this (for pxf at least) in an appNote in the <MMSIMinstDir>/tools/spectre/examples/SpectreRF_workshop directory, PSRR_Osc_AN.pdf (which is for using pxf to look at power supply rejection for an oscillator).
So you couldn't find any special setting for this, because there isn't a way for PAC not to do this! Even relative sweep mode is chosen automatically for oscillators... (I think).
In reply to Andrew Beckett:
I should probably go home and have some rest :)
I was choosing a 'guess' in the PSS which was so close to the actual frequency, that I could not spot the difference and I thought it was using 'my guess' instead of the proper one. Then, the fact that when using relative you specify your frequencies around that, so zero is a perfectly fine value, totally escaped me.
Thank you very much for this speed-of-light reply, it really made my week-end ;)