[ntp:questions] Re: difference between M12 & M12T GPS receiver

Simon Burge simonb at wasabisystems.com
Mon Feb 2 04:41:27 UTC 2004


Hal Murray wrote:

> Being off by a second seems like a common quirk.  I've seen it
> on both a GPSClock 200 and an HP Z3801.
> 
> I don't know where it comes from.  I wouldn't be surprised by
> a bug/glitch in ntp somewhere.

I've noticed this too on a Laipac TF10.  In my clockstats entries are of
the form:

 53037 16292.645 127.127.20.0 $GPGGA,043131.999,3810.2944,S,14418.4202,E,1,07,1.4,26.5,M,,,,0000*20

I don't know if ntpd is only seeing the "043131" and ignoring the ".999"
so that really it should be using the rounded up second?

While I remember, occasionally I see what looks like a single "missed
pps signal":

53036 69259.812 127.127.20.0 9644 0.000233637 0.000000000 0.000270000 0.377961049
53036 69275.022 127.127.20.0 9144 -0.999770778 0.000000000 0.000225000 0.925846448
53036 69292.012 127.127.20.0 9644 0.000264818 0.000000000 0.000255000 0.534537909

and other times a long run of them:

53036 67929.166 127.127.20.0 9044 -7.999051909 0.000000000 0.000255000 0.827809416
53036 67943.873 127.127.20.0 9144 -6.999066889 0.000000000 0.000225000 0.536139550
53036 67960.102 127.127.20.0 9144 -7.999083111 0.000000000 0.000225000 0.910040559
53036 67976.991 127.127.20.0 9144 -7.999091091 0.000000000 0.000255000 0.827300194
53036 67994.947 127.127.20.0 9144 -6.999101000 0.000000000 0.000240000 0.655980721
[ and 20 or so more ]

and then it corrects itself and is happy again.  Could this be an
artifact of a cheap GPS?  I've thought about adding a hack to say "no
tiemstamp received" if the clock has otherwise synced and then gets
stuck in this state, but I'm sure that's not the correct solution :-)

Cheers,
Simon.
--
Simon Burge                            <simonb at wasabisystems.com>
NetBSD Support and Service:         http://www.wasabisystems.com/



More information about the questions mailing list