[ntp:questions] Re: Z3801 Leapsecond bug - Sep 30?

Tom Van Baak tvb at leapsecond.com
Mon Sep 5 17:06:27 UTC 2005


Hi Tim,

The :ptime:leap:date? +2005,+9,+30 error happened
last time too (1998) -- to all HP GPS SmartClocks.
It shows up when the leap second is announced more
than one quarter year ahead of time. You can google
for old Usenet postings about it. As far as I know, no
other GPSDO has the same bug.

The bug goes away on October 1 and you'll get the
correct +2005,+12,+31 response; no interruption in
the 1PPS or 10 MHz outputs.

In this case the bug has nothing to do with 256
weeks or 1024 weeks, etc. I was told it had do
with a misunderstanding between the Smartclock
and Oncore VP firmware engineers.

I'm sure it's a simple bug to fix or to work-around
but HP never did it and they long since sold the
GPS product line to Symmetricom. No one I
know has source code to the old firmware so we
can't make an easy fix ourselves.

But as long as leap seconds are scheduled only
at the end of June and December it's a one-line
fix in any host software to work-around this bug.

See also:
http://www.leapsecond.com/notes/leapsec256.htm
http://www.leapsecond.com/museum/z3801a/eeprom.htm
http://www.leapsecond.com/notes/leap-watch.htm

/tvb
http://www.leapsecond.com/

----- Original Message -----
From: <shoppa at trailing-edge.com>
Newsgroups: comp.protocols.time.ntp
Sent: Friday, September 02, 2005 05:52
Subject: Re: Z3801 Leapsecond bug - Sep 30?

> Martin -
>   You write a lot of stuff, but I still don't see how this is
> applicable here.  The Oncore VP inside the Z3801A says that there is a
> leap second coming (this is true).  The Z3801A assumes that it is
> coming September 30 (which is false, but based on the fact that the
> next opportunity to insert a leap second would indeed be Sep 30th
> 23:59:60).
>
>   If you're trying to say that there is a flaw with the Oncore VP's
> internal GPS week number calculations, I agree this did happen last
> year, but I don't see how it's relevant this time around, because the
> Oncore is correctly saying that a leap second is coming.  The VP
> doesn't have the Bj command that is available in later Oncores.
>
> Tim.
>





More information about the questions mailing list