[ntp:questions] xGPS_NMEA / xPPS

Dave Hart hart at ntp.org
Fri Feb 24 19:38:48 UTC 2012


On Fri, Feb 24, 2012 at 05:45, Alby VA <albyva at empire.org> wrote:
>
>  I've watched as my NTP session has gone between normal and falsetick.
> Anybody want to toss out an idea of why this is happening and the
> solution to make it stop?
>
>     remote           refid      st t when poll reach   delay
> offset  jitter
> ==============================================================================
> -ntp.alaska.edu  .GPS.            1 u   69  128    7  127.532
> 0.581  19.471
> +utcnist2.colora .ACTS.           1 u   69  128    7   61.424
> 15.184  16.376
> *time-a.nist.gov .ACTS.           1 u   70  128    7   16.060
> 26.839  25.833
> +tick.usask.ca   .GPS.            1 u   75  128    7  106.332
> 20.798 24.112
> -cronos.cenam.mx .GPS.            1 u   68  128    7   81.165
> 1.985  11.542
> xPPS(0)          .PPS.            0 l   13 16 377 0.000
> 28.105   3.330
> xGPS_NMEA(0)     .GPSb.           0 l   12 16 377 0.000
> -11.920   5.476
>
>
>     remote           refid      st t when poll reach   delay
> offset  jitter
> ==============================================================================
> +ntp.alaska.edu  .GPS.            1 u   75  128    7  127.532
> 0.581  19.471
> -utcnist2.colora .ACTS.           1 u   75  128    7   61.424
> 15.184  16.376
> -time-a.nist.gov .ACTS.           1 u   76  128    7   16.060
> 26.839  25.833
> -tick.usask.ca   .GPS.            1 u   81  128    7  106.332
> 20.798 24.112
> +cronos.cenam.mx .GPS.            1 u   74  128    7   81.165
> 1.985  11.542
> oPPS(0)          .PPS.            0 l    3 16 377 0.000
> 28.630   3.022
> *GPS_NMEA(0)     .GPSb.           0 l    2 16 377 0.000
> -25.386 12.305
>
> ** Ignore the low reach. I just restarted ntpd.

See http://www.eecis.udel.edu/~mills/ntp/html/warp.html ("How NTP
Works") and the pages it links to.  Pay particular attention to the
clock filter, clock cluster, and clock selection pages.  The
falseticker 'x' label falls out of the clock cluster algorithm.

I would be tempted to try switching from separate NMEA and PPS
refclock drivers to using the NMEA driver alone with its PPS support
enabled via fudge flag1 1.  That will result in one less input to the
cluster algorithm and may resolve your issue.

Note that "ignore the low reach" is bad advice -- ntpd won't.  Low
reach means there are entries in the clock filter register with the
maximum dispersion value, which will affect the NTP algorithms by
design.

Cheers,
Dave Hart


More information about the questions mailing list