[time-nuts] Dual GPSDO - any advantage?

Florian Teply usenet at teply.info
Tue Apr 10 07:00:50 EDT 2018


Am Tue, 10 Apr 2018 17:43:25 +1200
schrieb donald collie <donaldbcollie at gmail.com>:

> I`ve just bought two GPSDO`s from China @ US$9-50 each [This is not an
> error!] They are stated as being new, and use a Trimball dual oven
> OCXO. I plan to run these in parallel [2 antennas, 2 feedlines, and 2
> GPSDO`s] It`s been said that a man with two watches is never happy -
> unless, of course, they agree with each other ;-). Being identical
> the outputs should be coherent, unless one becomes faulty [one
> advantage of having two - any discrepancy and you know something must
> be wrong]], but is there some crafty way I can squeeze a little more
> "accuracy" out of two than if I only used one?
> TIA for your
> comments............................................Donald Brett
> Collie ZL4GX

well, my experience in that regard is very limited, not to say
nonexisting, thus take the following with a grain of salt.

Assuming a system with two antenna locations where one device would
be fed both antennas, one could potentially employ some extra
correction based on extra knowledge from these two antennas. Even
though the antenna location usually is known with limited accuracy
(error on the order of 1m), the relative location of two antennas can
be known to a much greater accurracy (order of centimetres or better).
If the system can take that extra information into account, one could
maybe gain some extra accuracy for the solution of time/location.
For example, if the antennas are mounted 1m apart, and everything else
being equal, it is a priori clear, that the calculated position for
both antennas should also be 1m apart. As the signal received by both
antennas should be correlated, while their respective noise
contribution is not, one should gain some SNR. If on top of that time
resolution of the receiver system is very high, one could even take the
phase shift between signals received from both signals into account
similar to the way beam steering is done with phased antenna arrays,
just on receive side, again improving SNR. 
So far for an integrated system, where all signals arew processed in
one place. Now, as I read it you'll rather have two entirely separate
receiver systems. Here, depending on how it is implemented in the
receivers, one could potentially  still gain something extra over a
single unit, but I'd bet this can only be done in postprocessing.
After all, receiver A has no access to the signals as received by
receiver B and vice versa, therefore the spatial relationship can not
be used while deriving the timing solution. Maybe the independent
position solutions could be used to correct the timing solutions after
the fact.

Best regards,
Florian


More information about the time-nuts mailing list