[time-nuts] Z3801A Problem

Tom Van Baak tvb at LeapSecond.com
Tue Nov 6 19:19:37 UTC 2012


> What can the firmware do if it decides that it doesn't like a satellite?  The 
> GPS receiver does the math and sends over the PPS.  How would the firmware 
> adjust that calculation to not use a particular satellite?

Hal,

Just let the firmware do what it does. The detection and adjustments can be done off-board. Not unlike what all the pros do with post-processing.

Look at the Oncore @@Ha or @@Hn message -- you get the timing solution contribution for each SV. What you see in the hardware 1PPS is simply a real-time, uneducated, equally weighted mean of this calculation. It works pretty well. However, if you can apply an educated manual weighting (perhaps based on a long- or short-term history of SVN, Az/El, signal levels, std dev from mean, etc.) to each SV in the solution, it seems to me it is possible to calculate a virtual 1PPS that is more accurate than the real 1PPS.

When you think about it, a static elevation mask, ignoring the role of azimuth, ignoring your antenna 2D tree profile, ignoring S/N ratios, and treating all SV in the sky as perfecty equal  is simple, but dumb.

Someone on the list could make a fun project to explore these algorithms to improve the performance of Oncore, TBolt, or u-blox timing receivers. I've looked into this for the Oncore and TBolt but haven't posted anything. Contact me off-line if interested.

/tvb





More information about the time-nuts mailing list