John Pettitt said the following on 10/03/2007 01:53 PM:

> I'm not sure the whole GPS diversity argument holds water.    There 
> really is no good alternative that provides the accuracy that GPS offers 
> (CDMA is GPS and the clock solution don't offer the same accuracy do to 
> propagation delays).      I'm open to somebody convincing me that there 
> is a good alternative to GPS but I've not yet heard a compelling 
> argument.

For what it's worth, I'm currently running three radio refclocks on
separate servers, and comparing their performance across the local
network on another server that gets its PPS and clock from a cesium
standard.  You can see the current stats at
http://www.febo.com/time-freq/ntp/stats/index.html

All three systems are stable to about +/- 5 microseconds, though they
all have some spikes that are significantly larger (out to around 500
microseconds for short intervals); I think this is actually network
related rather than an actual time jitter -- remember that I'm not using
internal statistics, but instead offset as seen across the wire.

The WWVB system is the worst of the bunch, mainly because it adjusts
sync only in 100 microsecond steps, and if it loses lock entirely (which
happens infrequently), you see a short excursion of about -30
miliseconds.  But even it can usually be relied upon for better then 500
microsecond accuracy.  (Note -- if you look at the stats, it's currently
running at an offset of about -350 microseconds.  Almost all of that is
a systematic offset that I haven't yet compensated for.)

The LORAN C is by some measures the best of the bunch, but it suffers
from the disadvantage of requiring manual PPS acquisition; if the LORAN
station it's tracking goes insane, which sometimes happens, it requires
human intervention to recover.

John
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to