[time-nuts] Better than average Rb oscillator

[Angus]

Hi Attila,

>May I ask how you do the pressure compensation?
>
>And could you show a phase plot of the measurement?
>Our *DEV measurments hide a lot of detail and interesting
>features in oscillator behaviour. 

I use a c-field circuit controlled by a DAC rather than a pot, and
just feed the output from an analogue pressure sensor into an op amp
in that circuit. The internal pot does need a tweak every so many
years.
Plots attached for half and one day averages.

>> - apart from the ageing of course! 
>
>Well, IMHO the aging that Skip sees is the big deal.
>His ADEV goes straight down to 1e-14. That's on par
>with the best research Rb vapor cell standards I am
>aware of. And there are no aprubt frequency changes
>whatsoever, in his one month measurement. In contrast
>to that, I see a jump in frequency ever few days/weeks.

  I very rarely see abrupt frequency changes with LPROs - although
some with FE rubidiums and *lots* with Temex ones. Some on geriatric
FRKs too. When I first got an LPRO I was interested in looking into
the frequency jumps that rubidiums were supposed to have, but I just
didn't see them when logging. 
  This particular LPRO is somewhere around 20 years old now, and while
the ageing is lower than when this plot was done many years ago, I'm
not sure if it's as stable. I'll need to process some more recent logs
to see if there's anything interesting going on.
  Most of the logs that I've done have been on temperature controlled
ones, and I don't know if that might have helped. I've certainly had
better results overall with temperature control than temperature
compensation.

>It would be interesting to see where the flicker floor
>of Skip's setup is and whether he sees any lamp light
>shift induced jumps (these occur somewhere in the order
>of a few times a year on the most stable Rb standards).

  A frequency plot including temperature and pressure would be
interesting too.

>>    The Hadamard plot shows the performance that can be obtained by
>> just adding a GPS to correct for long term ageing.
>>    These plots are based on 1000s averages, so look a little better
>> than if the 1s data was used.
>
>Well, beyond a day, GPS is more stable than any Rb standard.
>If you have GPS always on, then you need the Rb only for
>short term time keeping and to bridge any GPS outages.
>But, in the sub-1d range, a good OCXO performs as well as
>a Rb standard. So you can get away with a lot lower cost and
>power, if you have GPS available. So, for me, Rb standards
>only become intresting beyond 1d.

  I think it depends....

  The point with the Hadamard plot was that the ageing can be very
predictable, to the point that it appears from the plots to have a
better stability than the GPS receiver used even at a tau of over 1
day. The dips at 1 day, 2 days, etc. also suggest this. The GPS used
was a Motorola M12+T with hardware sawtooth correction.
  The diurnal ripples in the frequency plot - which show little
correlation with the sensor temperature - also appear to indicate that
the current GPS setup is not up to the task. At that point the TC of
the LPRO was about 5E-13/°C. I'll need to get the dual frequency GPS
setup going to measure even this LPRO properly.

  I don't know if the continuously varying air pressure harms the
stability much (I'm guessing it does a bit anyway) but larger c-field
changes usually take longer to settle than I would expect (I really
should get around to testing that).

Angus.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: freq_43.png
Type: image/png
Size: 59719 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210218/b051a1b1/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: freq_86.png
Type: image/png
Size: 57780 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20210218/b051a1b1/attachment-0001.png>