[ntp:questions] Falseticking with PPS

unruh unruh at invalid.ca
Tue Nov 22 21:32:31 UTC 2011


On 2011-11-22, Nickolay Orekhov <nowhere at mail.ru> wrote:
> Hello, all!
>
> I have the following configuration on my client:
> 1 PPS signal delivered from some GPS receiver and 1 NTP server
>
> Here's what I've got:
> QNX Neutrino (localhost) (ttyp0)
>
> ntpq> pe
>      remote           refid      st t when poll reach   delay   offset
>  jitter
>==============================================================================
> xPPS(1)          .PPSE.           0 l    5    8  377    0.000  -356.53
> 1.699
> x10.0.0.197      .GPS.            1 u    4    8  377    0.316  84647.8
> 2.052
>  10.0.0.198      .STEP.          16 u    -    8    0    0.000    0.000
> 0.000
> ntpq>
>
> Here's my config:
>
> tos mindist 0.032
> tinker panic 0 stepout 60
>
> # PPS reference clock
> server 127.127.22.1 maxpoll 3
> fudge 127.127.22.1 refid PPSE
>
> # External servers
> server 10.0.0.197 prefer iburst maxpoll 3
> server 10.0.0.198  iburst maxpoll 3
>
> Ok, I see, they are both falsetickers because of great difference between
> them. But why the difference is so great?
> I think the difference between PPS signal and some other signal can't be
> more than 1 second, isn't it?

PPS will give you accurate nanoseconds, but horrible seconds. Ie you in
general need something else to determine the seconds. 
The more relevant question is why is 198 not working. 

>
> May be I should increase tos mindist to be slightly greater than 1 second?
>
> Thanks, Nick



More information about the questions mailing list