[time-nuts] Counter OCXO behaviour

Fabio Eboli FabioEb at quipo.it
Mon Jan 14 23:08:58 UTC 2013


I answer here to all, thank Ed David and Charles for your thoughts.
David, LOL, you posted the pic of the exact counter in question.
Not a similar unit, I mean it's exactly *that* 1992 that is being
measured ;)

The pictures on eevblog show the early measurements I made with
the counter. One of the Rb wasmeasured to about 10MHz+60mHz.
After a while (I think a pair of monthes) the reading was slightly
over 10MHz+70mHz and remained there for more than 6 monthes, until
I trimmed the counter few days ago.

The trimming is very touchy, and the simple opening of the screw on
the rear is traumatic: the frequency rises for a pair of minutes and
then it takes a while to return to the previous level.

Talking about this OCXO, how much is it good?
I mean for who has eperience with many OCXO is
this a good unit, or it's an average one?
Is this the kind of stability and reaction that can
be expected by an OCXO useful for GPSDO (less the
EFC control of course)?

Measuring this OCXO, just now I'm starting to feel the precision
of the FE5680, or 5682. It's so easy to set the Rb to follow
GPS closely for days...

I have some pics of the measurements I made, they are a little messy,
I hope can be fun and interesting to see:

This was before touching anything, top right the phase between
the GPS and OCXO:
http://www.flickr.com/photos/14336723@N08/8382126024/
it was slow by about 7x10^-9. This was the first trim attempt:
http://www.flickr.com/photos/14336723@N08/8382125734/
and this the last trims, the dips in phase are evident, they
happen when I open the calibration screw on the rear:
http://www.flickr.com/photos/14336723@N08/8381043177/
and at last these are logged after last trim action:
http://www.flickr.com/photos/14336723@N08/8381044033/

I tried to extract more data from last log:
http://www.flickr.com/photos/14336723@N08/8382126108/
Here I made the first and second derivative in time of the phase,
data are heavily averaged, so are not very accurate. The
10MHz period error (vs. GPS) changed from -0.8x10^-10 soon after
the trim, to +1x10^-10 after two days (top right).
The OCXO period was rising from almost 2x10^-15 soon after the
trim to 1.3x10^-15 after two days (bottom left).
Seem the stabilization will take a while...

Fabio.




More information about the time-nuts mailing list