[time-nuts] Happy Leap Hour Day !

[Tom Van Baak]

To address some of the recent DST postings...

Clocks with hard-coded USA DST rules will be off by an
hour for the three weeks from 11-Mar to 1-Apr-2007.
Given how long it's been since the DST rules changed you
can see why someone in a cheap or weak moment would
design a clock with hardcoded rules. Still, a bad design.

WWVB radio-controlled "atomic clocks", on the other hand,
do not have hardcoded rules; instead they switch to DST
based on command from Boulder. This is why many of us
saw our RC clocks/watches do the right thing this morning.
This is a good design.

But a  big problem remains. The WWVB subcode has only
the ability to give 1 UTC day of advanced warning of a DST
change. So if your clock happened to have poor reception
yesterday it still doesn't know of the DST change and will
remain in error by an hour until it ever gets good reception,
which could be day(s) later.

The problem is compounded by the fact that most RC clocks
only enable reception late in the evening (e.g., starting at
11 PM), that the DST switch occurs at 2 AM local time, and
that most of the USA is 5 to 8 hours left of Greenwich.

These three factors make the window for DST notification
much smaller than one day. And the result is that every time
a DST change occurs there are tens of thousands of RC
clocks that get it wrong (by not getting it at all). It's all a little
embarrassing since these clocks are often advertised to be
accurate to a millionth of a second, etc.

Further embarrassing is that NTP, the great internet clock is
so academically pure, and that GPS, the great clock in the
sky, is so globally available, that neither dare taint themselves
with the geographical and political mess of timezones or DST.

A solution would be to carve out a few more DST bits in the
WWVB subcode. So instead of giving a few hours of notice
an RC clock would see, for example, a 7- or 15-day binary
countdown to the DST event. That way, poor reception the
night before DST, or even a couple of nights before, would
not make the clocks fail at 2 AM Sunday.

Don't hold your breath waiting for a fix; but at least you better
understand the problem now. Actually, the solution may be
that more and more people are using cell phones instead of
clocks/watches/computers to get accurate local time...

/tvb
http://www.LeapSecond.com


Links:

NIST Radio Station WWVB
http://tf.nist.gov/stations/wwvb.htm

WWVB Time Code Format
http://tf.nist.gov/stations/wwvbtimecode.htm

WWVB Radio Controlled Clocks 
http://tf.nist.gov/stations/radioclocks.htm

http://tf.nist.gov/general/pdf/1976.pdf
http://tf.nist.gov/general/pdf/1383.pdf

Decoding WWVB from a Sony atomic...
http://www.leapsecond.com/pages/sony-wwvb/

WWVB Subcode Test Generator - wwvb2
http://www.leapsecond.com/notes/wwvb2.htm