[ntp:questions] NTP and PPS (and gpsd etc..)

David Anullo david_anullo at itacosystems.com
Tue Apr 24 14:31:20 UTC 2007


Hi,

I have a problem (Houston..). :-) 
I must connect a Navman Jupiter 21 receiver to a linux-embedded machine. 
I must also use PPS signal for accurate time to syncronize the internal 
machine clock.
After a lot of various tests i have big trouble upon the utilization of ntp.
i have used:

1 gpsd (version: 2.32, 2.35beta) with ntpd (version: 4.1.1, 4.2.4, 
4.2.5_24) (ntp.conf using shared memory 127.127.28.0/1)
2 only ntp (ntp.conf with NMEA/PPS 127.127.20.0/127.127.22.0)
3 a old Navman Jupiter 12 receiver (with Refclock Jupiter 127.127.31.0)

In the test 1, the ntpq -p show a offset upon SHM1-0 that without 
warning jump to +/- 300/400 msec (ntpd sometimes show the "no new value 
found" in shared memory)
In the test 2 i lost the PPS signal
In the test 3 i have a relevant excursion of peerstat offset (4 msec..)

In all test i have make various variations that i don't report to you.. 
this is only for example.

My question is: what i must to do with all this "instruments" to have a 
stable time source for my machine? How i can use this strange PPS?

How i can make to obtain a syncronization time-source from this reicever 
(or from a receiver)?

Help me!

-- 
David Anullo

Itaco S.r.l.
Via Zoe Fontana, 220 Ed. B1
00131 ROMA - ITALY
Tel. +390641294922 - Fax +39064191740
www.itacosystems.com




More information about the questions mailing list