[time-nuts] time-nuts Digest, Vol 43, Issue 35

[Tom Van Baak]

"Martyn Smith" <martyn at ptsyst.com> writes:
> Tom has plotted Allan Variance.  I have plotted frequency accuracy.
> 
> I get similar Allan Variance results to Tom.  But if I just measure the
> accuracy, e.g peak to peak frequency variations, rubidium's out- perform
> OXCO's even when both are disciplined to GPS.
> 
> I hope I'm not being stupid here.

Hi Martyn,

Depending on the requirements there is a place for accuracy
and a place for stability, or usually, a need for a certain amount
of both.

ADEV is a good measure because you can make a stable standard
more accurate if you need to, but you cannot make an accurate
standard more stable. Thus, the more fundamental measurement
of an oscillator is stability. With the exception of high-end cesium
standards, accuracy is something you tune for after-the-fact, but
stability is something inherent in the design and manufacturing.

/tvb

p.s. Can you send me the raw data for the two 3600 sample runs
in your paper? I'll send you accuracy and adev plots.