[time-nuts] Accuracy/drift of Garmin GPS 16 HVS 1 PPS output under invalid fix conditions...

[Achim Gratz]

Steve Olney writes:
> I think the level of precision you guys are aiming for is massive
> over-kill for my application.   Holding +/- 1 us over 15 minutes would
> be a doddle by many means as has been mentioned by you guys.  I just
> wanted to know whether the basic GPS mouse PPS would stay accurate
> enough to do the job for my application.

The trouble is, you can't tell -- at least if all you have is a single
GPS mouse.  Until you try and compare to some other source of time you
won't know what the firmware is doing.

I have a bunch of NavSpark mini and for some of them the antenna
location is bad enough that I see them getting in trouble at certain
satellite constellations / wheather conditions regularly.  These are not
timing modules, so they need to see 4 "good" sats to get 3D fix and
provide stable PPS.  When they drop below that threshold, you'll see the
position solution drifting and the PPS accuracy degrades almost
immediately.  The shift associated with the (wrong) position is almost
immediate, if the bad reception continues the PPS is starting to drift
further almost linearly.  A typical drift rate of the PPS when losing
lock is 100µs/hour (keep in mind I did shield the modules from air
drafts).  One or two times the GPS simply did not aquire a 3D fix again,
but the PPS stopped drifting at about 5ms offset as long as they see at
least one or two sats.

Now, any other GPS or even the same one with a different firmware
version might do a completely different thing.  I wouldn't know about
that either if I didn't have seven such systems scattered around my
network, so I can clearly see the glitches on one of them as seen from
the others.

> Radio astronomy is a hobby - i.e., a source of fun - although I do
> strive to maintain a level of scientific rigour to the best of my
> abilities.  One of the key drivers of the fun is doing things with
> simple systems.   I get extra pleasure in using a $20 DBTV dongle as
> my pulsar receiver to achieve the detection of a pulsar glitch (AFAIK
> the first time an amateur has ever done so on any pulsar).

OK, that explains nicely why 1ms is your target as the USB frame jitter
already introduces that much uncertainty.  Now, why do you need that
trigger contraption you were talking about rather than timestamping the
USB frames from the SDR and keeping the computer clock in sync with NTP
(you'll get to use the PPS from the GPS for that, otherwise a below-ms
timing accuracy won't happen anyway)?  Are you injecting a signal into
the antenna as a sync?


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

SD adaptations for KORG EX-800 and Poly-800MkII V0.9:
http://Synth.Stromeko.net/Downloads.html#KorgSDada