Poul, Thanks for the informative reply. I guess the short answer is that we should use a very clean long-term GPS trend record to correct the PRS10 rather than the 1PPS from the GPS unit, which has horrible short-term stability.
I believe that Tom VB mentioned once that a daily update to the PRS10 EFC should be sufficient. I can see his point. It would be bad to add all that short-term noise to the rubidium's PLL. I will discuss with the other engineers here the possibility of using the phase info from the counter to bump the EFC control on the Rb source. It seems like a fairly easy thing to do. I imagine a simple proportional frequency control loop (human or computerized) would be sufficient for our needs since the integral error would be much smaller than our milliseond-level time accuracy requirement. I'll also look into setting up the SR620 counter to average the phase over perhaps 1000 samples to do the first-level noise reduction. Then we'll get 86 useful numbers per day instead of 86000. Finally, we will be getting a maser again in a couple months and we will probably want to stabilize it this way (with a 30 day adjustment period, perhaps) after the VLBI run, unless the VLBI folks have a better idea. I'll have to ask them. _______________________________________________ time-nuts mailing list time-nuts@febo.com https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts