[time-nuts] Re: Phase coherence with 2X GPSDO

[ed breya]

I actually have this kind of setup, with two "identical" GPSDO boards I 
salvaged from some telecom gear long ago. I've yet to be able to 
identify them by make or model. All I know is they use Oncore UT or GT 
RX units, and seem to work, according to the LED indicators and the 10 
MHz outputs.

I built them into one box with a common supply, and common RF in from a 
Wilkinson splitter, fed from an active antenna. When I first got it all 
fired up and working, I was disappointed that the outputs drift in 
relative phase. Sometimes they happen to remain close for long times, 
and  sometimes they obviously slide quickly enough to watch the shift 
over say a half hour or so, observed on a scope. The direction can go 
either way at any time. After I thought about it some, I realized that 
they can't be perfectly aligned, even if there were no sawtooth errors 
from each one's independent situation and activity.

I think the basic limit besides the GPS processing and noise, is the 
discreteness of the frequency control. They use 20 bit DACs for tuning, 
which seems pretty good, but the overall effect of two separate systems, 
each trying to perfect its own deal according to the GPS process, to 
compensate for temperature and their individual characteristics, just 
leads to a certain amount of relative drift. I think the times when they 
seem very stable are when everything happens to be just right, but 
mostly that's not the case. Presuming they actually are working right, 
over the long term, they should both agree on the average frequency, but 
not moment by moment.

So, this is all with two same-type units in the same environment 
conditions, more or less, and getting the same GPS signal, more or less. 
Imagine different units in different locations and conditions.

Ed