[time-nuts] Crystal filters in test equipment

[Bernd Neubig]

>> Learned Gentlemen,
>> Both the HP 106 and 107 have a post oscillator crystal filter.  There 
>> is also a 10 MHz crystal filter used in my Tracor 527E FDM.
>> So the question I have is there anything to be gained by adding 10 
>> MHz crystal filters to the 10811 and similar OCXO's?  They are very 
>> inexpensive to purchase.
>> Regards,
>> Perrier

Some of the commercial ultra-low phase noise OCXO today are using crystals filters at (or just in front of) the output buffer stage. These filters are usually heavily overdriving the crystal into the non-linear region, therefore the long-term reliability of such a circuit is suspect.
The purpose is to improve the far-off noise floor, which works fine, but sometimes the close-in noise gets worse, as Rick said before.
Cheap monolithic filters (duals) are much more sensitive to high drive levels, and thus are not recommended at all.
Another effect of a crystal filter in the output is, that its output impedance outside the passband is far off from 50 Ohm. So it fools the power splitter at the input of cross-correlation based phase noise test equipment, and may display lower unrealistic phase noise "dBc/Hz" values at the noise floor, sometimes lower than the thermal noise floor of a 50 Ohm system.

Regards
Bernd