Let's talk tau (τ)

[Ohm_My]

Hello forum,

I've been collecting data from a SRS PRS10 rubidium standard and utilizing the Stable32 program to graph results, etc. I'm having some difficulty in understanding the tau setting with regard to changing it and why.

Several of my data runs are for 2, 6 and 24 hours for the rubidium standard. I'm not using any GPSDO for steering for long-term corrections.

When I import the source data into Stable 32 I'm unsure as to what tau value to apply or if it even needs to be changed from the default. I realize, conceptually, that the tau value is a measurement of time, be it seconds, days or whatever. I'm looking at trying to verify the 1, 10 and 100 second ADEVs for this device according to SRS. I'm using Lady Heather 6.08b for the source data capture. I'm confident I have the multiplier values and a few other settings set correctly.

If I desire to say verify the 100s specification per SRS using a 24 hour capture file how is the tau value needing to be adjusted? The Stable32 program is quite robust and I would like to be reasonably sure of the data results. I did come across a helpful document authored by W. J. Riley, published in May, 2003. Once I digested that bit of material I'm still unsure of the tau setting(s) for the Stable32 application.

I would enjoy hearing from any and all about the tau settings for the app.

Thank you for reading through this.

[chuckb]

Welcome!
I've been using Stable32 for over 15 years. I initially used it to compare two different GPS locked Rbs that I put together. Lately I have used it for voltage stability analysis.

In Stable32, as you import the data file it prompts you for the tau. This value is the sample interval for the data file. If Lady Heather collected data to a file at 1 second intervals, you tau would be 1 second.

If you collect data for two hours the AV results will display a Stabilty plot that covers 1 second to 1 hour. I usually let it collect data overnight.

When you RUN the program and process the data file I select "All Tau" for the plot results. That provides a smoother plot with more data points.
Hope this helps.

[Ohm_My]

Hello Chuck,

Thank you for the quick response. After a more in depth research of the Stable32 document I did finally locate exactly as you said. The program is quite robust and I'm just getting my feet wet in precision timekeeping, rubidium standards and GPSDO.

My initial runs of data appear to be in conflict with respect to the specs stated for the SRS PRS10 rubidium source. I probably just need to tweak some additional settings within Stable32 to get me closer to advertised spec. Hopefully, my freq source is not out of spec yet as I just purchased it new in mid-July.

Thanks again for the knowledge transfer.

Regards.

[5065AGuru]

Hi,
You say your results seem to indicate you are not meeting spec.? Is your counter able to give you enough resolution for the lower Taus? Also is the reference you are testing against good enough that it does not mask your results?
The PRS10 AD spec is 2X10-11th at 1 Sec, 1X10-11th at 10 Sec, and 2X1012th at 100 Sec.
A decent running HP 10811 would be a good enough reference. (Has to have been on for a week or so!)
The counter, or other device (especially for the 1 Sec AD measurement) has to have enough resolution. In the neighborhood of 12 digits/Sec.
There are ways to boost the resolution, DMTD system, Single channel offset mixing, etc.
You can also seek some advise on time-nuts http://lists.febo.com/pipermail/time-nuts_lists.febo.com/


Cheers,

Corby