questions in simulating jitter performance of crystal osc +squaring buffer

specialest

Hello,

These days I was working on crystal osc +squaring buffer. I care about the jitter performance mostly. I have searched many literature and forum, but still have some questions here.
the mmsim is 10.1, the ADE is 5.1.41.

when simulating the jitter performance.
1, since the most noise contribution is from the squaring buffer, (the output of crystal osc is a nearly pure sine wave) . I simulate the buffer separately with a vsin from analogLib (to mimic the output of crystal). So I choose pss +pnoise (jitter mode, PM). and set the threshold value to vdd/2. and calculate the jitter from Jee curve.
jitter= sqrt(integ(Jee^2, 10k, 1M))
it's about 100fs. and the most contribution (90%) is Rf from the first stage.

2, I simulate crystal +buffer together. now it's a autonomous circuit, I can't choose pnoise jitter mode(PM), so I choose pnoise source mode.
and calculate the jitter from phase noise curve.
jitter=sqrt(2*integ(10^(phase noise/10) 10k, 1M))/(2*pi*fosc)
it's about 30fs, and no Rf contribution in noise summary.

I can't explain which is correct. For the application, I certainly care about the threshold crossing jitter. but I can't simulate that for crystal +buffer together.

crystal+buffer

[img][/img]

since I need to know which simulation give the real-world jitter performance, then I can do the trade-off in deciding the parameters for each component.

fuyibin

in two case, the impedance is different. Rf noise current contribution should be converted to noise voltage. And the noise voltage amplitude will affect the jitter. the xtal impedance is parallel with Rf. So in these two case, noise current from Rf is same but impedance is different. That is why the jitter is different

lwjee

You Should select (jitter mode, PM) with XTAL.

specialest

回复 3# lwjee


   I can't choose jitter mode PM in pnoise, since it's an autonomous circuit. If I choose jitter mode, it will be FM automatically. No option to change that. Maybe a problem with mmsim 10.1 version.

lwjee

回复 4# specialest


   Or time domain noise, PM