The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Help needed: PNoise analysis error in SprectreRF
https://designers-guide.org/forum/YaBB.pl?num=1159381081

Message started by littlekoala on Sep 27^th, 2006, 11:18am

Title: Help needed: PNoise analysis error in SprectreRF
Post by littlekoala on Sep 27^th, 2006, 11:18am

Hi, there,

I am using the “pnoise” analysis from SpectreRF to simullate oscillator phase noise. After the PSS analysis converged, then the pnoise analysis started, but it failed immediately with a message: Analysis ‘pnoise’ terminated prematurely due to error.

I can not figure out what is going wrong with my simulation. Transient and PSS analyses show that the oscillator circuit should work fine.

Can anybody help me on this problem? Any help and advice will be greatly appreciated.

Littlekoala

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by ACWWong on Sep 27^th, 2006, 11:56am

there isn;t enough information to go on, but whatever the case the best course of action would be to forward you spectre logs to cadence directly.

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by littlekoala on Sep 27^th, 2006, 3:39pm

I included an AGC (auto gain control) in the feedback loop to control the oscillation amplitude. It seems to me that the AGC is the source of the problem. But I still do not know to get the pnoise analysis done.

Did anybody have the same problem before? How to do phase noise analysis with AGC in the circuit?

Thanks.

Littlekoala

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by ACWWong on Sep 28^th, 2006, 8:58am

in principle i don't think the agc should cause the problem, as the pnoise runs based upon the converged pss analysis. The pss solution should be done after the oscillation amplitude has settled (using tstab). Inspect your spectre log files for clues as to why the "pnoise terminated prematurely due to error".

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by littlekoala on Sep 29^th, 2006, 4:59pm

There is an error message from SpectreRF: Assertion failed in file 'shoot.c' at line 6749. I do not know what is going on within the simulator.

Did anyone have the similar problem before?

Littlekoala

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by Ken Kundert on Sep 29^th, 2006, 10:22pm

The message "Analysis ‘pnoise’ terminated prematurely due to error" indicates that the simulator was terminated due to an error that was reported earlier. If you want help, you should include the actual message associated with the error that caused the simulator to terminate.

The message "Assertion failed in file 'shoot.c' at line 6749." indicates that an inconsistency has detected in the simulator. This indicates that there is a bug in the simulator that should be reported to Cadence. However, given how much troubly your are having, you might be using the simulator in an inappropriate way. You should consider describing your situation more fully in the hope that someone can determine where you are going wrong.

-Ken

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by littlekoala on Oct 2^nd, 2006, 8:18pm

Ken:

The oscillator comprises a MEMS oscillator (verilog-A model) with a Q of 10000, a trans-impedance amplifier and an AGC.

The PSS analysis converged in every simulation, but the pnoise analysis failed for most of my simulations with the same error message as I posted before. Sometimes, when I changed tstab to some value, the pnoise analysis completed properly, but these situations are very rare. And I found no relationship between tstab and a successful pnoise analysis.

Thank you.

Littlekoala

Title: Re: Help needed: PNoise analysis error in Sprectre
Post by Geoffrey_Coram on Oct 3^rd, 2006, 6:13am

Are you using method=gear2only in your PSS? (I think you should.)

Have you tried writing out the PSS and reading it back in to initialize a second pss (to make sure it's really converged)?

Have you set any strange options, such as compression=yes? Have you looked at the options for oscillators?

Do you get any warnings from the Verilog-A compiler?

Is there a way to force fixed-point iterations instead of shooting-newton? I didn't see the option, but maybe a Cadence expert can tell us.