[time-nuts] Re: GPSDO - GPS1300-10-1000 by RFX Ltd. UK

[Bob kb8tq]

Hi



> On Mar 1, 2022, at 9:28 AM, Lux, Jim <jim at luxfamily.com> wrote:
> 
> On 3/1/22 12:54 AM, John Moran, Scawby Design wrote:
>> David - thanks for the reply, but these seem designed for SDR and I wanted 1pps. Nice and cheap though.
>> 
>> Paul - thanks too; it seems that you are saying that the performance of all GPSDOs are the same, but that wasn't the impression I had got from listening to discussion here. Fine when the thing is locked (except for sawtooth stuff and these hanging bridge things, and digital vs analogue (British) control loops, etc.) but what about hold-up performance when it's not.
> 
> 
> Holdover performance is going to be determined by the oscillator's stability. The time error is essentially = tau * ADEV(tau)  - so if you have a ADEV(1000sec) of 1E-9, after 1000 seconds, the uncertainty of your 1pps would be 1E-6 seconds after 1000 seconds of holdover.

There are a variety of tricks that can get played for holdover. You can correct for frequency error
when locked to some finite level ( maybe 1x10^-11 .. who knows) you might then “learn” the offset
and pump that into an “out of lock” correction process. 

Oscillator aging does get into this. Again you can learn aging (maybe). You also could guess aging 
(it generally goes in one direction for a group of oscillators). Again, you pump that into some sort
of correction process. 

> 
> A more subtle effect is "what happens when the GNSS signal returns" - Does your 1 pps abrubptly snap back to GNSS epoch within a few ns? Or does it gradually drift back to sync. Downstream uses might care.

Indeed, and on an OEM module off eBay, who knows what the original gear cared about or did
not care about. The same thing applies to frequency correction and how much they care about
(how much of) that. 

There are some applications out there that are not telecom targeted and they do indeed get a
GPSDO that works very differently than the typical telecom unit. 

Bob

> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.