The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Trouble with PSS and a frequency divider
https://designers-guide.org/forum/YaBB.pl?num=1130430091

Message started by danmc on Oct 27^th, 2005, 9:21am

Title: Trouble with PSS and a frequency divider
Post by danmc on Oct 27^th, 2005, 9:21am

I'm trying to simulate a frequency divider (built with static CMOS logic) using PSS.

If I run a transient sim, everything looks good. All the waveforms are what they should be. When I run PSS however, I'm unable to get it to converge. I selected my clock source as the Fundamental Tone and set it as a Large signal. For the beat frequency I picked the divider output frequency (1/24th of the clock source). For tstab, I picked a time which is long enough for the divider to fully go through a couple of output cycles so I've hopefully flushed out any wierd initial states.

In all 20 pss iterations, I see the same large value for the convergence norm:
Conv norm = 72e+03

Also, once I get past this problem I'll want to look at the noise at the output rising edges. Is there a way to automate the selection of the time for the timedomain noise? It would see like if I could figure out how to make a spectreRF compatible S-R flip-flop, it would be simpler to set the (noiseless) flip-flop with the divider output and reset it with something derived from my clock source. Then I wouldn't need the timedomain option but could go from the phase noise output directly. Comments on this?

Thanks
-Dan

Title: Re: Trouble with PSS and a frequency divider
Post by Ken Kundert on Oct 27^th, 2005, 2:32pm

There should be one fundamental, which is the output frequency of the divider (1/24th of the clock source frequency).

-Ken

Title: Re: Trouble with PSS and a frequency divider
Post by danmc on May 6^th, 2006, 9:13pm

seems there was one extra flip-flop internally dividing by 2 one more time so while the output frequency was 1/24 of the input, the fundamental frequency of the circuit as a whole was 1/48. duh! Its no wonder the convergence norm wasnt' getting smaller :-X

so, any good way to automate finding the rising edge time to feed into the timedomain pnoise sim?