I played around some with masking sats that were in poor signal areas, but the problem I wound up with is once you exclude a sat the Tbolt firmware no longer tracks its position so you don't know when its safe to start using it again. To determine that properly you would need to model the sat orbits, etc. It opens up a rather messy can of worms...
Mark, Have you considered using the individual sat solutions (in the 8F A7 message), removing one or more of the sat's in software, and then re-calculating the mean clock bias? This way you don't stop tracking any SV; you just selectively use or not use the SV bias data of your choice. This might work well with your new external software disciplining approach. Another approach that might be fun is playing with weighting the SV clock bias contributions rather than the all or nothing model. You may find that variable weighting by Az or El or signal level or SVN or time-of-day, etc improves the timing quality a bit. If nothing else you could collect a lot of per-SV clock bias data and then crunch a week's worth of data to look for correlations. Given that you could then "replay" the timing solution and see what effect different algorithms would have had on the timing output. /tvb _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
