[ntp:questions] Re: NTP server with Rubidium PPS
David L. Mills
mills at udel.edu
Tue Oct 11 20:35:35 UTC 2005
Thank you for using plain ASCII, which is much easier for me to read.
I suggest using the kernel interface for the Rubidium to discipline the
frequency only and leave the time discipline to the external servers.
You will need to craft minor changes to the ntp_loopfilter.c source to
do that. The changes should be obvious from the comments. You do not
need or want the PPSAPI interface.
Mauro Fiacco wrote:
> Hi all,
> I have a strange problem which I would like to share (and may find a
> We have a Rubidium source which we use as PPS for our NTP server (running
> freeBSD 5.4, ntpd 4.2.0 - standard ports)
> My NTP server is also configured to use a number of high stratum NTP
> servers as timeservers.
> I have been experimenting with this settings for quite long time and I
> come to the conclusion that if the offset of the PPS (which is
> unsynchronised) is greater than 128ms respect to the offset of the
> timeservers, won't allow ntpd to step it's time and reduce the offset.
> Am I correct?
> I have tried to use the tinker configuration word in my conf file
> (tinker step 0.512) without success.
> The only way I seem to be able to obtain synchronisation _and_ PPS is to
> turn on a off my PPS source until the offset is below the critical
> 128ms... Which takes too much time!!!
> Any ideas on how best use the PPS signal?
> Thank you in advance.
More information about the questions