On 2015-01-30 18:09, [email protected] wrote:
I built a GPS (HP 58503) based NTP server about 10 years ago.  It worked fine 
but eventually died, and I built a replacement.  But I am having no end of 
problems with long offsets.
The new machine is Linux-based with a Z3801 and PPS.  It consistently shows an 
offset as compared to other NTP clocks.
After about a week of trying to figure it out, I got desperate and bought a 
used Datum Tymserv 2100 as a stopgap until I could figure out the problem.  It 
worked great until recently, and now it has a 1-second offset as compared to 
the world.
Both servers share a common antenna (Symmetricom GPS LT) on a HP/Symmetricom 
splitter.  Obviously I make certain that I have good GPS signal and everything 
is locked.
I recently purchased a SECOND Datum 2100 and it had the exact same offset.  So 
I ran a new wire and hooked it to a different antenna of the same model, same 
offset.
So I changed to a different antenna, same offset -1 second as compared with 
known good network servers.
Here is what I have running in this example:
Freiburg = Datum 2100
Augsburg = Linux server with HP Z3801, shared with Freiburg's antenna
Shaun = Datum 2100 on different antenna
When I go into Augsburg, all seems good (PPS currently disabled), but it is off 
by nearly a second in the other direction, +1 second from known-good servers so 
+2 seconds from my other servers.
Clearly I am doing something wrong.  I've been beating my head against a wall 
for quite a while trying to figure it out.  I have another system with a Z3805 
that has an offset of 16000 with a good lock.

See recent posts on timenuts archives about Datum Tymserve 2100 et al early 
leap second, checks, and fixes.

--
Take care. Thanks, Brian Inglis
_______________________________________________
questions mailing list
[email protected]
http://lists.ntp.org/listinfo/questions

Reply via email to