[time-nuts] Re: Phase coherence with 2x GPSDO
Tom Van Baak
tvb at LeapSecond.com
Mon Mar 7 19:15:59 UTC 2022
> I've had the same question but I can't see how it could possibly "know"
> its own frequency error/uncertainty. What would it reference to?
That's why it's a rough internal *estimate*, not an actual measurement
against an external reference. Again, it goes back to how a GPSDO works.
There is the assumption that GPS is "perfect" in the long-term. There is
the assumption that the OCXO is extremely stable short-term. The two are
coupled in a slow feedback loop. Thus it is possible to use the OCXO to
see how bad GPS is, short-term. And it is possible to use GPS to see how
bad the OCXO is, long-term. This is roughly the nature of the TINT and
FEE numbers.
I don't of anyone who actually uses these numbers, or trusts the result.
And the important thing, for this discussion, is that the numbers won't
improve your accuracy. They are more a diagnostic of how the GPSDO is
working. As your application is getting clearer, you do not want a GPSDO
with all its complications and unknowns, not to mention proprietary
firmware. Instead you want a plain GPS timing receiver and do all your
own real-time or post-processing.
> I need to place one of the SDRs on a drone and one on the ground.
Will that work? Almost no quartz-based GPSDO are rated for flight. The
3D acceleration alone will greatly affect the oscillator at ppb levels,
confusing the GPSDO algorithms. Most GPSDO are designed for static use
in stable environments. While they are running you do not touch them,
blow on them, turn them, tilt them, not even by millimeter.
But you should try it to see for yourself. Start a clean phase
comparison between your two GPSDO on the bench. Wait for steady state.
Then pick one up, tilt it, wave it, wiggle it, and keep this up for a
while. See how much your phase coherence suffers and also how quickly
(or slowly) the GPSDO recovers from these external perturbations. This
is another reason to use a plain GPS receiver and not a GPSDO.
/tvb
More information about the Time-nuts_lists.febo.com
mailing list