[ntp:questions] Maximum time2 fudge value for NMEA refclock?

David J Taylor david-taylor at blueyonder.co.uk.invalid
Tue Jul 20 09:19:18 UTC 2010

"Rhys" <user at mailinator.org> wrote in message 
news:MPG.26b01082c6206af6989680 at news.optusnet.com.au...
> I'm amazed it quickly converges, using separate NMEA and PPS on mine,
> with the correct offset, and correct drift file, its an hour or 2 before
> it finishes going 500ppm/step/500ppm/step.

With a standard NTP system, either with a LAN sync-source, or a 
single-connector NMEA/PPS source on either Windows or FreeBSD I've not 
seen that sort of behaviour.  That's with a GPS 18 LVC on Windows and 
FreeBSD, and with a GPS 18x LVC on Windows.  I don't recall explicitly 
testing the 18x on FreeBSD.  There can be an hour or three before the 
ultimate accuracy is reached.

Perhaps the separate NMEA and PPS are confusing it?  Is it clock hopping? 
I think you need someone more expert than me to comment.

> Running around 600ms offset now, I charted it against PPS, and it looks
> roughly like 560-640 ms, slowly going back and forwards across the day.
> Using combined NMEA and PPS it just ignores the NMEA until the PPS is
> working, but thats not as self-sufficient as i'd like.

I'm not sure what you mean by that.  Within a couple of cycles (20-40 
seconds) my NMEA shows as synced with a "*" (ntpq -p) and the PPS as 
synced with a "o".  Admittedly I don't restart these systems at all 

> Thanks for your notes on Freebsd stuff, helped a lot setting it up.

Delighted it helped.  If you found an error on something which could have 
been explained better, just shout.


More information about the questions mailing list