[time-nuts] GPS Filter

[Tom Van Baak]

> I think it's simple, at least in the nice/common cases.  If the antenna 
> geometry has a point that everything swivels around, consider that to the the 
> location of the antenna.  I think that covers the typical alt-az mount:
>  http://en.wikipedia.org/wiki/Altazimuth_mount
> The point is where those two axes intersect.  Now just fudge the coax delay 
> to correct for the time/distance from the real antenna location to that 
> point.  That's the before location coax delay (in there with the ionospheric 
> delay) rather than the post GPS antenna-to-box delay.
> 
> Of course, it gets a bit more complicated than that if you want to track 
> several satellites in real time.  That probably takes an antenna per 
> satelite.  But again, VLBI  geeks have been doing that sort of math for ages.

I've always wondered if someone would do this. Place a bunch
of mini az-el mounts in a 3x3 or 4x4 grid 10 cm apart and let
each 9 or 16 antennas pick a unique SV to listen to. After all,
the receiver knows exactly where each SV is and so can point
right to them. Seems it might also be a way to reduce multi-path
as well since each antenna can then be much, much narrower.
I don't know about the RF side. But phase delays and the offset
within the grid can be handled in software. Or let each antenna
have it's own receiver, sharing a common clock.

It would be very fun to watch this antenna in time-lapse. For a
wonderful demonstration of the GPS constellation, imagine on
a misty evening emitting a laser beam out of each antenna.

This was partly inspired by photos PHK posted many years
ago of an array of M12 receivers.

It also gave rise to the idea of creating an az-el sundial; not a
conventional one where the shadow moved across the face
clockwise each day, but one where a microcontroller and light
sensor array drove the mount so the sundial always had exactly
no shadow at all (another thread from several years ago).

/tvb