Harmonic Port and PSS with HB engine

Hi Michell, as a best practice, use the hb Choosing Analyses form (not the pss-hb) for harmonic balance simulation.  (Reference:  Getting the Most Out of Spectre® X-RF 21.1: Maximizing Performance  and Getting the Most Out of Spectre® APS RF 21.1: Maximizing Performance .

You may want to check out:  Spectre Circuit Simulator and Accelerated Parallel Simulator RF Analysis in ADE Explorer User Guide -- Frequency Domain Analyses: Harmonic Balance - Example Using the Harmonic Port with Harmonic Balance.  There's an example you can follow.

Sorry...can't dig into this more now.

best regards,

Tawna

Dear Tawna,

thanks for your help. I still do not understand whether I could use the Harmonic Port option with PSS-HB, I only got from your reply that it would be better to use HB engine directly. So I'll desist.

The circuit I am trying to simulate is maybe too basic or presents some sensitivities I cannot work out on my own and since I have no access to your Support services, I guess my chances are very limited.

I have already tried to explain the circuit and simulator behaviour here but then I was experiencing also other kinds of issues, like the netlist not properly updated when I changed sim type. I think there is a possible issue when one uses "auto calculate" but defines frequencies with variables. I have witnessed that they do not get updated when one changes them and re-runs simulation. So, if I define a clock with fLO and the PSS is "auto-calculating" the fundamental to be this value, it then substitutes the value of the variable at the minute, say 10 GHz. If I now change the variable value to 12 GHz and rerun, the PSS fundamental is still 10 GHz. This stung me at the time of writing the previous thread, and this time I was re-visiting, I caught this behaviour.

The workaround is simple: untick the "auto calculate" and write explicitly your fundamental as a VAR("fLO") statement. Then everything works fine. I haven't checked if one can directly write "fLO" instead (w/o quotes), but I guess that's a possibility too since "fLO"  is also a "freqname"  in the source definition.

Back to my issue.

I do not know why, I've spent a considerable amount of hours trying to make a simple HB + HBAC work for the downconverting ideal mixer I show in the previous post. I use this sim type for much less ideal structures and it works without issues, so I believe this to be some sort of numerical corner I'm pushing the simulator into. However, as said, I wasn't able to find out what (I've experimented with number of harmonics, non overlapping clocks, extra resistance here and there, asymmetry of the load... Nothing helped so far).

PSS-shooting + PAC, on the other hand, works flawlessly. And, to my surprise, even PSS-HB + PAC does!! HB + HBAC does not.

As a visual proxy, here below I'm reporting the output of the baseband side, as it comes from the PSS-HB + PAC, and HB + HBAC:

(Stimulus has been defined as relative, so you have that the "-1" sideband is actually the downconverted signal).

The PSS-HB outcome is in perfect alignment with theory (based on the component parameters I've in the circuit), while HB one is just ZERO.

So, all this very long reply just to tell you that I have my reasons why I've come to use PSS and I can't go back to pure HB, as it would be the recommended way for Harmonic Port use.

Regards,

Michele

I cannot think why PSS in HB mode should behave any differently from HB analysis, since it's the same engine underneath. You should find a way to contact customer support as trying to debug this here will be to no avail. There must be a way of you accessing customer support if you are at a customer with a valid software contract.

Andrew

I cannot think why PSS in HB mode should behave any differently from HB analysis, since it's the same engine underneath. You should find a way to contact customer support as trying to debug this here will be to no avail. There must be a way of you accessing customer support if you are at a customer with a valid software contract.

Andrew