[time-nuts] PRS10 MDEV

Christoph Kopetzky ck at cksd.de
Tue Oct 16 06:52:56 EDT 2018


Tom,

thank you for your explanations.
Such information helps the newer time-nuts as me to get easier in touch 
to this fundamentals.
I did not know the reasons for these quasi perturbations which also I 
could see on my own reference oscillators. Now I understand where they 
come from...

-- 

with best Regards

Chris
Am 15.10.18 um 07:31 schrieb Tom Van Baak:
> When I see an ADEV or MDEV plot that looks unusual I like to check the phase or frequency plots for clues. Thanks for your .TIM file.
> 
> Attached are:
> 
> Burford-6h-adev.png -- TimeLab 'a' command. This is a straight line with slope -1; what you expect for a self-test. Note the ADEV at tau 1 second is right about 7e-11, 70 p/s, which is typical noise for the TICC counter.
> 
> Burford-6h-mdev.png -- TimeLab 'm' command. As you observed this has, a plateau. Sort of unexpected given you weren't using a GPSDO.
> 
> Burford-6h-phase.png -- TimeLab 'p' command. Phase difference plot. Notice all those bumps.
> 
> Burford-6h-phase-100s.png -- TimeLab 'g100' command. Phase difference plot with 100 second averaging makes the variations stand out much clearer. Can you guess what that is?
> 
> So the smoking gun is that you have quasi-periodic perturbations in phase on the order of about 100 ps. The shape looks thermal to me. Do you have a thermostatically controlled heater in the room or house? I suspect this is causing your MDEV plateau.
> 
> If so, you've accidentally done a nice experiment -- which is to establish your measurement noise floor and to examine the environmental conditions of your work bench. Ah, the annoying joys of working with precise time!
> 
> Again, assuming it is due to thermal effects, you can now find out if it's your TICC or your cables or your PRS10. Since TimeLab will display plots in real-time a heat gun or cold spray may help you isolate the culprit quickly.
> 
> /tvb
> 
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.
> 



More information about the time-nuts mailing list