[time-nuts] NOT digest, I forget the subject not sure why the digest received Richard's reply and not me ? Oh well, never mind.

[Bruce Griffiths]

Stanley Reynolds wrote:
> Guess that would be the case of too good to believe performance unless the lock was intermittent and was not recognized. Guessing that the phase comparison equipment should be fairly equal in precision and accuracy as well, maybe a rapid switch of the measuring by only one device would help the equal part if the sample and recording could be rapid enough. Otherwise some type of rotation to qualify differences in the measuring equipment. Or as I suspect most use better precision and accuracy in the testing equipment than the DUT, once you reach the upper limit of better measurement then use the N cornered hat to cull the bad ones and id the good ones.
>
> Stanley
>
>   
Injection locking can lead to an unbelievably low phase noise, ADEV etc.
The insidious part is that the closer the OCXO resonator frequencies
become the lower the injected signal has to be to achieve injection lock.
ie as the frequencies get closer more and more reverse isolation is
required.
One limiting mechanism is the intrinsic fluctuation in resonator parameters.

Ideally the phase comparisons should be made simultaneously.
Deadtime can confound the results depending on the phase noise processes
that are significant.
If more than one type of phase noise process (white phase, flicker
phase, white FM, flicker FM etc) is present/significant, correcting for
deadtime effects can be problematic.

One method is to use very high reverse isolation (>120dB) buffers
between each OCXO and the phase comparison gear.
If one uses mixers (eg in dual mixer systems), then high isolation
mixers can help.
Minicircuits have a few (2??) level 17 mixers with 80dB LO to RF
isolation (RF to LO isolation should be similar) at 10MHz.

One could use N dual mixer systems together with N high end sound cards
(plus software) and a suitable offset oscillator to make the phase
comparisons.

Bruce