My initial estimate of the time > constant, based on the perfect performance, of 1 tau = 3 hours, was way > off. The real tau is more like 1 day. So I'll wait another day before > making the next adjustment. > > Do you have plots that cover longer time periods? I've done lots of short runs while tweaking this thing...

If time constants is allowed to be limited by frequency drift (regardless of its source) then you can use a long term drift estimate to control the time constant of the control loop. Essentially being an adaptive filter. It is still a gross oversimplification, but may still be a handy one. Cheers, Magnus

If the motional parameters of the crystal were known then the required capacitance change can be estimated before dissassembly.

If you've got 10,000 cell sites, each with an Rb timebase, MTBF figures can provide a pretty accurate estimate of how often one will fail and need to be replaced, at least during the normal lifetime (between the infant mortality and wearout stages). Coupled with MTTR statistics, one can figure out how many spares should be stocked.

materialId=slides&confId=4896 I found particularly interesting the ones by Maximiliano Sioli, where he explained the two mistakes found in the OPERA data acquisition chain and how, after correcting for their best estimate of their effects, the time of flight is compatible with a speed of c. I saw the webcast of the event.

Some of them correct for Rb frequency error to a first approximation -- for example, I believe the Symmetricom units with BesTime (e.g., TS-2500) monitor the Rb frequency as compared to the GPSDO during lock, then correct the output frequency during holdover (when the Rb is in use) by adjusting a DDS according to an estimate of the Rb frequency error --

It is somewhat related to what Dave Mills calls "the Allan intercept" except this you can actually measure and not just estimate. You run several (long!) test-series with different timeconstants in your PLL, and you record the resultant EFC and phase offset as a function of time.

I'm experimenting with using a temperature sensor to estimate local oscillator frequency changes. My goal is to have a decent holdover clock for a NTP server with not so great GPS antenna placement. I've been sampling temperature every minute, measured by a DS18B20.

. >* * * >The input voltage noise @1Hz can't be used directly to estimate the >PN noise at 1Hz offset. My message referred readers to previous list messages for a more detailed explanation, and recommended a particular search term.

I just checked Google Earth and the elevation of your office is 5260', only about 24' off of your GPS estimate if that is your location.