[ntp:questions] Re: GMT: UT1 vs UTC
Tom Van Baak
tvb at leapsecond.com
Mon Nov 3 07:21:48 UTC 2003
> However, you raise an interesting idea. The on-time epoch is the
> beginning of the minute that the timecode is actually sent. From the
> leap warning you know the epoch that a leap must be made and in the
> previous minute the DUT1 sign. From this you can predict the DUT1 after
> the leap and thus the leap correction. Clever. I'm still worried about
Correct. But it's even less fragile than that.
1) By convention you know that the leap second
will occur at the end of the last minute of a month.
2) You know what month it will be because the
leap second pending bit goes on and stays on
weeks ahead of time.
3) You also know months (or even years) ahead
of time what the DUT1 sign is.
So 1+2+3 allows you to know well in advance
when and what type of leap second it will be.
Thus there's no requirement to be following the
timecode during the exact minute before the leap
second. Which is a good thing because it is likely
you won't have good reception anyway.
By the way, this is a much better situation than
the DST bits. WWVB-based atomic time radio
controlled clocks that don't happen to get good
reception the night before a DST change will not
make the change when they are supposed to.
> the three-bit DUT1 magnitude in WWV/WWVH.
Is not bit 50 the DUT1 sign for WWV/WWVH? See:
> Notwithstanding the above, I was told by NIST staff that the NIST
> timecode generators do not implement a negative step. I have no
> information about other stations.
That's correct. Which is why it's important to state
that the WWVB subcode spec itself is fine; it's various
pieces of hardware (including the Ft Collins TCG) and
end-user software than may not be prepared for a
negative leap second should one ever be scheduled.
More information about the questions