[Ground-station] $89 GPSDO works pretty well

[Phil Karn]

Hi all, I don't know if this is appropriate for this group but I suspect
at least some of you might be interested.

I recently bought this 10 MHz GPS disciplined oscillator off ebay for
$89. It's apparently one of a series of designs by some guy in China.

https://www.ebay.com/itm/10MHz-Sinwave-PLL-GPSDO-GPS-DISCIPLINED-OSCILLATOR/122957393972

Not expecting much for $89, I took it over to Kerry Banke's place this
afternoon for a test. We used his HP interval/frequency counter and his
free-running (but carefully tweaked) Rb oscillator as a reference. (I
think he doesn't always keep it automatically locked to GPS to avoid any
unexpected artifacts from the correction algorithm during measurements).
I also used his rooftop GPS antenna.

As expected, my unit started off frequency and asymptotically approached
10 MHz as the oven warmed up. After 5-10 minutes (I didn't time it) the
red alarm LED went out. I took to mean that the oven was up to
temperature and the GPS was locked. (Disconnecting the GPS antenna
turned the alarm LED back on.)

We watched over 45 minutes as the unit continued to asymptotically
approach 10 MHz. At that time it began to bobble +/- 1 millihertz around
10 MHz (1 millihertz is 1 cycle per 16.7 minutes!) We then compared my
unit and his Rb oscillator against his own Trimble GPS receiver, which
gave comparable stability figures of 100e-12. (I suspect that's an Allan
variance, but if so we're not sure of the averaging interval.)

I switched from Kerry's outdoor GPS antenna to the provided antenna to
see if it was any worse inside his garage but we didn't see any change.
However, by that time the unit had probably tamed the crystal oscillator
pretty well so I wouldn't expect much immediate change even if the
indoor GPS signal was much noisier.

Strangely enough we were unable to make the red alarm LED come on by
wrapping the GPS antenna in aluminum foil or putting it in a metal box.
It lit pretty quickly when no antenna was connected, so I suspected
maybe it was seeing a loss of LNA supply current. But it didn't alarm on
Kerry's antenna, and he says it's capacitively coupled and has its own
power supply.

I also took a look at the 1pps output. It's a 1Hz 3.3V square wave,
which raised the question of which edge is on the second. We took a few
minutes to find a working HF receiver (Kerry actually had one!) and tune
in WWV. The answer: the rising edge.

Anyway, this seems like a pretty good unit, especially for $89. And I
know we're gonna need some good stable time and frequency references in
this project (and many others -- I have ideas for beam forming and
frequency hopping on HF).

I can now continue my project of evaluating local ATSC 8VSB pilot
carriers as poor-man's frequency references. Some hams seem to be
assuming that they're highly accurate (e.g., see the AMSAT Reno
proceedings) but that's not what I see. Here in San Diego, channel 8
(KFMB) is pretty close but channel 10 (KGTV) is about 4-5 Hz high,
consistent with past observations. I suspect this will be typical
elsewhere (in places that have ATSC).

73, Phil