[ntp:questions] Re: Differences in behaviour of ntpd 4.1.1 and 4.2.0 on Windows XP
neil.burn at marconiselenia.com
Fri Nov 21 12:40:04 UTC 2003
I am not actually using the PPS output of the acutime to drive ntp.
It has a serial output which provides the time to ntp. That said, the
PPS output width is configurable and I have tried 10ms and 10us and
there is no discernable difference in the offsets that 4.2.0 achieves
with either PPS width.
As I type this I think I have spotted the problem. There are a few
differences in the refclock_palisade.c file between versions 4.1.1 and
4.2.0. Specifically there has been a change to use nsecs instead of
usecs. The problem appears to be caused by 3 missing zeros...
I made the following change to "refclock_palisade.c":
497 <! pp->nsec = (long) (getdbl((u_char *) &mb(3)) * 1000000);
497 !> pp->nsec = (long) (getdbl((u_char *) &mb(3)) * 1000000000);
I've rebuilt 4.2.0 with the 3 missing zeros added and it all seems to
Thanks David/Terje for pointing me in the right direction.
(Do I need to send an email to bugs at mail.ntp.org to report this?)
Terje Mathisen <terje.mathisen at hda.hydro.com> wrote in message news:<bpjbc4$ssn$1 at osl016lin.hda.hydro.com>...
> David J Taylor wrote:
> >>I am a little confused over the differences in behaviour of ntpd 4.1.1
> >>and 4.2.0 on Windows XP.
> >>I have an acutime 2000 connected to a PC running ntp. I am
> >>comparing the offset between the acutime pps output and the pc time as
> >>shown by an oscilloscope with the offset as reported by "ntpq -c pe".
> >> With 4.2.0 the offset as reported by ntpq tends to zero as expected
> >>but the oscilloscope shows an offset of anything up to 700
> >>milliseconds (this offset value varies each time ntp is restarted).
> >>With 4.1.1 the offset as reported by ntpq agrees with the oscilloscope
> >>to within the millisecond.
> > Is there any question of the edge polarity of the two versions being
> > different? I.e. is the pulse about 700ms wide and NTP is triggering off
> > the wring edge? I don't run a PPS source, just throwing in the
> > suggestion....
> Nice suggestion, the default polarity of PPS have been changed at least
> once during the ntpd development process.
> If this have happened here, then a switch to the opposite edge would
> indeed fix the problem.
More information about the questions