[time-nuts] Re: Estimating phase noise of sampling clock

[Magnus Danielson]

Hi Jim,

On 2025-07-16 00:37, Jim Lux via time-nuts wrote:
> 	
>
>
> If one has a digitized sample (of TBD length) of a "good" sine wave, has someone got a cookbook way to get the phase noise and/or Allan deviation of the sampling process (mostly dominated by the sample clock oscillator, but potentially other noise sources)?
>
> I can do something like break the samples into shorter chunks, fit a sine wave to the sampled data. Use the frequency (or phase) of those measurements as input to a AVAR calculation, for instance.
>
> Typical samples would be at 2.048 MHz for 2-10 seconds (so not particularly useful for AVAR, to be fair)

So, the modern way to do this is to use CORDIC, track the phase output 
for PM, subtract a reference phase-ramp, decimate data step-wise with 
good filters to fit your needs, then FFT to your hearts delight for PM 
spectrum. See IEEE Std 1139 for instance.

Cheers,
Magnus