I am running the circuit below and get the phase noise using pss/pnoise as well as hb/hbnoise. It is a frequency divider(divide-by-two) in which I used sample-and-hold circuits instead of latches to keep it "analog". pss/pnoise gives expected results whereas hb/hbnoise shows zero phase noise at the divided outputs ckI and ckQ. Is the switch not supported by hbnoise or could there be another problem? The log shows the following warning. The frequency in the warning is always the last point I specify for hbnoise.
Warning from spectre at freq = 10 MHz during HBNOISE analysis `hbnoise'. WARNING (SPECTRE-16518): Arithmetic exception in analysis `hbnoise' .
Schematic snapshot and netlist attached.
Spectre Version 188.8.131.527.isr6 64bit -- 4 Jul 2018
Somehow the netlist isn't getting attached. It is at https://drive.google.com/file/d/1zfNfs2q9zO5wT3-pGWHjGT9QkP39nOBO/view?usp=sharing
If it works with pss/pnoise but doesn't with hb/hbnoise and also produces a warning about an arithmetic exception in this case, I would say that it's likely a bug in the hbnoise analysis. I suggest that you examine if the problem also exists in the most recent Spectre release and report it to Cadence Support if it does.
I checked the netlist with the latest Spectre version, and it also still produces the warning. I think the warning is benign, but I filed CCR 2180892 to get this addressed. I found some earlier similar requests to R&D but thought a specific request for this issue would make sense.
Note that it's nothing to do with the switch (relay) components as it still happens if they are commented out. Nor is it anything to do with the use of the noise file source.
Thanks for taking the effort to check and for filing the ticket.
BTW, I don't see an issue if I remove the divider(e.g. the circuit below). There is no exception reported and the results are as expected. I have also used hbnoise with a number of other circuits without any problems. (This is why I unjustifiably assumed that the switch was the issue. It was the most exotic component in my previous circuit :)
Actually this seems to be dependent upon E0 - the first of the two vcvs in the circuit. If I remove it, or reduce the gain to 1, or remove the min and max options (of course, this alters the behaviour of the circuit) the arithmetic exception warning goes away.
Hopefully we'll clarify this if R&D can look at it.