[time-nuts] Better GPS

[bg at lysator.liu.se]

Hi,

> Hi
>
> I've been digging around at acc.igs.org. They seem to think that their
> clock models are good to below 0.05 ns. They compare that to a 2 ns number
> for the "as broadcast" models. 2 ns is a pretty familiar number if you
> look at a lot of TBolt plots. All of their data is online and available
> for download at some point in time past live. The longer you wait the
> better the data.

You can get some improvement even in 'real-time' by using the ultra-radid
(predicted) data available over Internet.

     http://igscb.jpl.nasa.gov/components/prods.html

> They really like the idea of L1/L2 receivers and fancy post processing.
> They also are after a lot more than just time. I still have not noticed
> any complete L1/L2 receivers for < $100, so I can't do their full solution
> approach. Ionospheric correction seems to be one big thing you loose by
> going single freq.

Yes, but you would only need one dual freq receiver to estimate the Iono.
Or have real time access to one in the vincinity. Another way, would be to
use the models for Iono and Tropo errors that are available in real time.
One example is the SBAS (Waas, Egnos, Msas, etc) signals. They should get
you down below 2 meters. Perhaps there are online models for prediction of
Iono related radio communications that could be useful.

Beeing connected to Internet opens a whole lot of possible improvements
not available for the classic standalone GPS receiver.

> There's a pretty good summary of all that at
> http://acc.igs.org/UsingIGSProductsVer21.pdf
>
> Since the TBolt sort but not quite of does carrier phase you might be able
> to use their data to improve things. I doubt that you would get to 50 ps,
> but even a 2 or 3X improvement would be worthwhile. Even getting it some
> of the time (midnight?) would still be useful. It would be a log lots of
> data and compare thing - not that unusual.
>
> Has anybody else dug into any of this?

I think we should examine the Tbolt measurement capabilities a little
further. One experiment that would eliminate local clock errors would be
to compare two Tbolts sharing both antenna and the same oscillator.

This is one avenue of might be improvements of the already good Tbolt
GPSDO. Other paths as controlling the temperature and optimizing control
parameters are ofcause also interesting.

> Is there a secret stash of cheap L1/L2 fixed location receivers out there
> that I haven't stumbled across?

Check with local surveying outfits, or close by CORS stations they might
have old receivers collecting dust.

--

   Björn