[time-nuts] OCXO Question

[Bob kb8tq]

Hi

I don’t really see anything in your plots that looks like it correlates with temperature.
You may have processed it out with some of your fitting and regression stuff. 

There *is* something going on with he red line. What it is … very unclear. Best guess
would be a bum solder joint somewhere. 

Pretty much all of these OCXO’s (regardless of seller) come out of a pretty brutal
scrap process. It is “normal” to get more bad then good when you buy a group of 
them. Generally, the closer you look, the more you throw out. That’s been true for
quite a few years now. 

Bob

> On Apr 25, 2019, at 11:43 AM, Bob Quenelle <bobqhome at live.com> wrote:
> 
> In testing a new OCXO, I found it had more drift than a previous one.  I'm 
> happy with the better one, not so much for the poorer one. I'd like get some 
> opinions on whether the performance of both seems reasonable.
> 
> I bought both from a US seller. They’re UCT 108663-01, a double oven OCXO. I 
> didn't specifically want DOCXO, just had mixed experience with China-based 
> sellers and decided to try a US seller. Both have over 100 hours of run 
> time.
> 
> To compare the 2 OCXO's I let both OCXO's and an LPRO-101 run for about 12 
> hours to stabilize. Then I turned off DAC updates so the control voltage was 
> constant and collected phase of each OCXO relative to the LPRO-101 every 10 
> seconds. I subtracted successive samples and scaled to get mHz. Next I 
> subtracted the linear regression line so I could see deviations from 
> nominal. I also used exponential averaging with factor 0.1 to clean up the 
> noise. I'll attach the graph and see if it makes it through. In summary, 
> over the 6 hour run the better OCXO shows a peak to peak deviation of 0.2 
> mHz and the poorer one showed 0.9 mHz. This is deviation/wander from 
> nominal, not overall drift. The graph also shows room temperature.
> 
> I've run a test with both OCXO's in the same box on different days and seen 
> similar results so it's not the voltage reference or DAC. I show this test 
> because both OCXO's ran at the same time.
> 
> <OCXO_1.JPG>_______________________________________________
> 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.