Find a good app note from Maxim about clock jitter and phase noise conversion. It has detailed derivation. In summary, carrier spectrum is like below:
Then we can use below to calculate RMS clock jitter:
In spreadsheet, we use linear approximation where a is slope rate for that linear region.
Spreadsheet uses the following measured clock phase noise as an example.
So RMS clock jitter due to 10Hz-100KHz band noise is given in cell F15. Peak-to-Peak jitter is 7.44 * RMS jitter assuming noise is Gaussian dist.
I also like to see how much contribution floor noise (noise in band >100KHz) can make to jitter/total-noise. Turns out quite a lot. But normally osc output is filtered so this band noise can be dropped.
Great answer, thanks for sharing.