[ntp:questions] Re: NMEA refclock - PPS not working

R Jenkins not at pub.lished
Mon May 15 14:41:45 UTC 2006

>"dave morgan" <morgad at eclipse.co.uk> wrote in message 
>news:0kvg62prkbhj8nciotpqqmgqqosq4hotk5 at 4ax.com...
>On Mon, 15 May 2006 08:34:21 +0100, "R Jenkins" <not at pub.lished> wrote:

>a couple of random thoughts

>a) Have you tried setting ntp.conf to look at the other edge of the PPS 

>b) Section 4c of the PPS kit install instructions (for PPS kit 2.1.5)
>caught me out for a couple of weeks. Have you copied both the updated
>timepps.h and timex.h files so the NTP compile can see them? (I
>initially only copied one)

>http://www.morgad.no-ip.info/index.html    gpg:0x64B5E037
>Distributed Proofreaders: http://www.pgdp.net
>The NTP server pool http://www.pool.ntp.org
>questions mailing list
>questions at lists.ntp.isc.org

Hi Dave,

many thanks for that, I was missing the timex.h file.
I used just the ppskit-lite patch & only found out about the timepps.h link 
by chance.

It must have been seriously messing things up as all the jitters etc. 
immediately reduced to sensible values.

I now get, after around 30 mins running:

ntpq -c peers
     remote           refid      st t when poll reach   delay   offset 
*GPS_NMEA(0)     .GPS.            1 l   17   64  377    0.000   42.046 
-gate.jrw.intra    3 u    -   16  377    0.195  200.062 
+sites.urchin.ea   3 u   41   64  377   22.420  200.693 
+braindead.mfa.g   2 u   54   64  377   88.858  117.665 

ntp_gettime() returns code 0 (OK)
  time c8130f14.e05a1000  Mon, May 15 2006 15:30:44.876, (.876374),
  maximum error 52983 us, estimated error 0 us
ntp_adjtime() returns code 0 (OK)
  modes 0x0 (),
  offset 47.000 us, frequency -0.276 ppm, interval 4 s,
  maximum error 52983 us, estimated error 0 us,
  time constant 2, precision 1.000 us, tolerance 496 ppm,
  pps frequency -0.276 ppm, stability 1.000 ppm, jitter 1.000 us,
  intervals 386, jitter exceeded 12, stability exceeded 2, errors 328.

It looks like the PPS input is seeing the wrong edge of the pulse.
The DCD signal transition is +12 > -12 at the start of the pulse from the 
GPS, which I thought should be correct.

I will try setting the flag2 option which hopefully will be the final fix!

Thanks again,
Robert Jenkins.

More information about the questions mailing list