how to simulate crystal oscillator phase noise

[chuntian101]

Hi,

I am now designing crystal oscillator, but as you know, crystal's Q is so high that cadence pss simulator will take very long time to converge (sometime maybe not converge at all), but GoldenGate simulator can converge more quick than spectre, however, phase noise result are quite different, so I don't know which one is believeable？ Is there anyone compared these two simulator?

 

[Pictou]

I'm doing the exact same thing right now, and for the noise I use the hbnoise simulation, you can find a tutorial here : (Be sure to use IC615 ISR14 or later to see the new MMSIM12.1 features in the GUI)

https://www.cadence.com/Community/b...e-features-auto-tstab-and-auto-harmonics.aspx

And you are right, PSS doesn't work at all.

 

[chuntian101]

Hi Pictou,

Thanks for you kindly reply, I'm not very confident of crystal core plus buffer phase noise result by using hb+hbnoise, because I do the buffer only phase noise simulation by using pss+pnoise(jitter) and hb+hbnoise, and the latter result is much optimistic than the former, have you compared with these two results? and how are you simulating crystal core plus buffer phase noise? Thanks very much.

 

[dick_freebird]

"Simulating" phase noise will have as good a result as
the models and inputs and numerical methods conspire
to allow.

Often phase noise is determined by something other than
intrinsic transistor noise generation - like, finite edge
rates and finite stage gains, against various noise
aggressors (ground / supply bounce, ambient EMI picked
up on near-linear XTAL input traces, etc.). If you are
not representing the physical design and its losses
and some sort of realistic environment, your piece-
part phase noise in application will bear not much
relation to analysis.