Different results for Jitter and Phase-Noise

[netbug]

Dear all,

I have designed a crystal oscillator. I am using a pierce oscillator (source grounded). My goal is to calculate jitter rms. I have started from phase noise, but the results are inconsistent with jitter's results.

So in order to debug this I have some questions:

1 -As I am using a 80 MHz crystal, do I have to use the samevalue for the beat frequency parameter of the PSS analysis or not ? Some application notes claim that should be use half of the freqency.

2 - The biggest problem of all, is when calculating Jitter using PSS+PNOISE the results vary quite a lot due to the upper integration limit. How to set this correctly ?

3 - Point 2 result's are so dramatic, that weak inversion, exhibits better phase noise results than in strong inversion. In terms of jitter is exactly the opposite, strong Inversion has less Jitter than Strong Inversion. I susopect that the integration limits should be wrong, when calculating jitter.

4 - The phase-noise curve that I obtain is quite strange around 80 MHz (Figure attached). If I exclude/include that part in the integration of Jitter, the results vary quite a lot.

Thank you for your attention.

Best regards,

Pedro Cardoso

 

[klystron]

Does this help **broken link removed**

 

[netbug]

Hi,

Thanks for the link. It is very useful to understand the several types of Jitter. But I needed to know how to calculate jitter on Cadence having the phase noise already.

Best regards,
Pedro

 

[skal81]

Regarding your question about simulation, I'll try to answer some of them.
1 - If you are using your crystal at its nominal frequency, then set the beat frequency of the PSS to 80MHz. If you are having your crystal oscillating at another frequency please set that one.
2 - Set the upper limit to half the oscillation frequency. The upper frequency terms are taken into account as folding component of the noise.
4 - You're it's strange. I was expecting a higher peak. If you set your integration limit higher than the oscillation frequency, the main oscillation tone appear in the PN plot (and it's harmonic). So you usually have a quite big peak. But this one is quite small which puzzle me a little... Anyhow I think if you set the integration to half the oscillation frequency this should not happen anymore.

(sorry I'm not sure for question 3)

To calculate the period jitter you can use the following eq.:
J=\frac{2}{2\pi {f}_{osc}} \sqrt{\int S_{\phi }(f)\sin^2\left(\frac{\pi f}{f_{osc}}\right)df}

You can find some information on using Spectre PSS here:
https://www.designers-guide.org/Analysis/PLLnoise+jitter.pdf
And here are some stuff about phase-noise and jitter:
https://www.delroy.com/PLL_dir/tutorial/PLL_tutorial_slides.pdf

 

[erikl]

@skal: You better enclose LaTeX text within TEX tags:

\[J=\frac{2}{2\pi {f}_{osc}} \sqrt{\int S_{\phi }(f)\sin^2\left(\frac{\pi f}{f_{osc}}\right)df}\]