[time-nuts] OCXO pressure sensitivity before TCXO

[Bob kb8tq]

Hi

Keep in mind that all those designs date “way back” in terms of OCXO design. 
The whole issue of pressure sensitivity really was not well understood / accepted
until the mid 1990’s. Up to that point the whole “is it pressure or something else”
thing was very much a hot topic. 

In terms of an OCXO designed for phase noise (like the 10811), pressure is pretty
much a non-issue. It’s far enough down that it will only impact long term performance
and even then well below the daily aging spec on the part. Humidity is often 
more of an issue on a 10811 than pressure. 

Bob


> On Aug 5, 2019, at 6:30 AM, ew via time-nuts <time-nuts at lists.febo.com> wrote:
> 
> Today most quality OCXO's are in a sealed can but three of the best HP10811, Austron 1150 and OSA 8600/07 are not sealed. Austroon set screw hole has a screw with an O ring. When you try to get the best out of them not part of a loop you may have to consider compensation like with Rb's. Maybe it is not the Xtal but some other part of the circuit. Any Idea or experience. Pressure will most likely be slow. Here in Florida Hurricane  is on us, maybe a time to outline a test.
> Bert Kehren
> In a message dated 8/5/2019 3:34:15 AM Eastern Standard Time, attila at kinali.ch writes:
> 
> On Sat, 3 Aug 2019 14:22:22 +0000 (UTC)
> ew via time-nuts <time-nuts at lists.febo.com> wrote:
> 
>> Are wee missing something. Reading Bill Slade post and old NASA
>> attachments the attached plot to me was an eye opener. Our work 
>> has always been focused on Rb pressure compensation. Are we
>> overlooking something. Has any one done some work on this subject
> 
> No, we are not overlooking anything. Pressure sensitivity of
> OCXOs has been long known and most manufacturers account for it
> in their design. In most cases, the temperature induced instability
> is larger than the pressure sensitive effect, together with the
> pressure sensitivity measurement and characterisation being
> expensive, it does not get mentioned in the datasheets.
> 
> But, if you really want to compensate for environmental effects,
> then you need to have a quite complex model. This makes it rather
> difficult to parametrize, you need measurements over months if not
> years. That makes these parameters also quite difficult to use, as
> they exhibit memory effects (e.g. hysterisis) and change over time.
> 
> If you are interested in this, have a look at the fields of 
> "System identification." Or, if you want to have it a bit simpler
> and more statistical: "Principal Component Analysis"
> 
>             Attila Kinali
> -- 
> <JaberWorky>    The bad part of Zurich is where the degenerates
>                 throw DARK chocolate at you.
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.