[time-nuts] TDEV measurements on a telco sync unit

[Christophe Huygens]

Hi all,

I am playing around which an HP 55400 system, using an HP Z3801A as PRS, 
conditioning a rubidium holdover card. The system has 8 2048kHz outputs 
(the 1/5/10MHz output cards are supposed to exist, but never seen one in 
20 years). Using the rubidium (FE-5680A, checked inside) and HP 
"Smartclock" this system should generate a "better" clock with superior 
phase stability which is what telco folks care about (I think this means 
that frequency stability is ok too but not 100% sure).

What is interesting (at least for me) is that this system reports a lot 
of statistics on it's PRS input and one can apply these measurements to 
other inputs even if they are not used in the smart calculations, these 
are just measured against this generated supposedly perfect clock.

Indeed when I measure TDEV of the Z3801A versus a Thunderbolt using the 
5370 TI (the best I can do) this seems close to the measurements on the 
PRS using the above (reports at 1 4 16 256 and 1024s TDEV). Does this 
mean I should do the effort of diving in and extracting this perfect 
clock and turn it into my reference 1 PPS? Or am I better of investing 
in a 10811-60159. Don t want to spend more than a few 100.

I did measure the 2048kHz out (in frequency) and these are not so good 
but of course these are the result of another PLL locked to this clock.

br / tnx

Xtof on4iy