[time-nuts] Method for comparing oscillators

[John Green]

Part of using this method is that, at some point, the DUT must be set to the
same frequency as the reference. (As close as I can get it anyway.) So,
accuracy and stability are addressed separately. For TCXOs, set to frequency
and watch the drift over time. I have an environmental chamber so I can do
freq vs temp testing. This is actually better done with a good frequency
counter using the Tbolt as reference. The problem with OCXOs is that when
you adjust frequency, you have to wait a long time for the thing to settle
in. That is why I will set up and go about my daily duties, only checking in
periodically. I mainly want to look at 2 things. Short term   stability, by
this I mean from turn on to maybe an hour. When used in communication
equipment, how fast an oscillator gets to within tolerance as well as
retrace are important. Next is long term. Hours to weeks and even months. I
would like to be able to do years, but at some point I have to be realistic.
Yes, Allan deviation plots are interesting but I am more concerned with
ordinary spectral purity and stability over time and perhaps temperature
that will have practical effects on either test equipment measurements or
communication system performance. This is interesting conversation, by the
way. Thanks