Markus wrote:

> a 0.5 mHz (not Hz) hourly drift won't affect decoding much.

> The difficulty is only that the actual center frequency is
> not easily predictable,

You're right.  It's impractical to search for.  The only way
is to prefix the message with a spell of carrier long enough
to measure F.   But strictly speaking that carrier should
be included in the overall Eb/N0 budget - it will raise the
average Eb and knock about 3dB off the performance.

Here I've been testing a Trimble 57963 GPS with PPS and 10MHz
DOCXO and it works quite well.  Plenty available on ebay.
Not quite plug and play because it uses an undocumented RS232
protocol which I've pretty much figured out now.  Will have
to modify gpsd so that it can work with it.    It could become
the timer/oscillator for a proper LF receiver.

Three MC1496 are staring at me from the junk box.

--
Paul Nicholson
--