[ntp:questions] GPS-PPS, standalone server. NTP

Fida Hasan k.fidahasan at gmail.com
Tue Jun 20 13:18:11 UTC 2017


On Wednesday, June 14, 2017 at 5:37:04 AM UTC+10, David Lord wrote:
> Fida Hasan wrote:
> > On Thursday, June 8, 2017 at 6:37:04 AM UTC+10, David Lord wrote:
> >> :
> >>> Le mardi 6 juin 2017 15:28:27 UTC+2, David Taylor a écrit :
> >>>> On 06/06/2017 13:11,  wrote:
> >>>> []
> >>>>> Some GPS will continue to deliver a PPS signal even if the lock is lost. I'm thinking particularly about the Garmin 18xLVC where it is clearly indicated in the documentation (4.4.1): 'After the initial position fix has been calculated, the PPS signal is generated and continues until the unit is powered down.'
> >>>>>
> >>>>> With the use of that 'kind of' GPS, ntpd will continue to provide time service.
> >>>> As I understand it, NTP will only continue to provide a service if it 
> >>>> has other "time-of-day" sources available.  Should the NMEA output (as 
> >>>> the only time-of-day source) become invalid, NTP would reject it, and 
> >>>> gradually ramp itself up to stratum-16 so as to become invalid as a 
> >>>> server to its clients.
> >>>>
> >>>> [1 - I'm unsure off the top of my head what NTP checks to know whether 
> >>>> NMEA is valid or not.
> >>>> 2 - I wonder what the drift in the GPS 18x LVC is when unlocked?]
> >>>>
> >>>> -- 
> >>>> Cheers,
> >>>> David
> >>>> Web: http://www.satsignal.eu
> >>> At least on my 'Garmin', when a fix is not valid the position is not given but the time message remain available. The GPS internal clock continue to work.
> >>>
> >>> One question is to know how stable and precise can be the internal clock of a 18xLVC GPS model? I don't have yet the answer but if it's comparable with the one in a Raspberry or Odroid chip then I'm an happy man for some hours:)
> >> Hi
> >>
> >> NMEA from my 18xLVC was +/- 300ms so I used fudge stratum so that
> >> it didn't affect time accuracy if PPS wasn't available. Sometimes
> >> there was an inversion layer preventing good GPS reception. The
> >> LVC was swapped out to be replaced by a SURE which was still
> >> reliable when the PC went down in March this year an has not yet
> >> been replaced.
> >>
> >> from my ntp.conf:
> >> server 127.127.20.2 mode 18 prefer
> >> fudge 127.127.20.2 stratum 7 time2 0.407 flag1 0 refid GPSb
> >> server 127.127.22.2 minpoll 4 maxpoll 4
> >> fudge 127.127.22.2 flag2 0 flag3 1 refid PPSb
> >>
> >>
> >> David
> > 
> > Hi David,
> > I was just wondering to know the accuracy you have achieved through driver 20? Did it turned down to 1-miro or around it?
> > 
> > Regards,
> > Fida
> 
> Hi
> 
> with the two gps I've tried, driver 20 is only accurate
> enough for numbering the second, say +/- 400ms for my LVC
> and maybe +/- 100ms for the Sure.
> 
> Combined with driver 22 I had accuracy down to few us,
> mostly better than 1us.
> 
> If you have other internet sources use of driver 20 isn't
> essential.
> 
> 
> David

Hi David,
Combined with 22, in 20 I have also got 1 micro offset (maximum 1.60 micro) in kernel mode. But in user mode it is more than 5 micro or above.
Cheers!



More information about the questions mailing list