Delays of Alarms of Self-Setting Clocks

New: 2002 December 31

                                                
From: David Dunham [dunham@erols.com]
Sent: Wednesday, December 25, 2002 5:22 PM
To: Michael Schmid, IAP/TU Wien
Cc: scotty399@shaw.ca; jan.manek@worldonline.cz; ALucas0217@aol.com;
pdmaley@yahoo.com; rickf@stic.net; stevepr@netstevepr.com;
hhcuno@t-online.de; H.-J.B@t-online.de; beisker@gsf.de;
rob515@swbell.net
Subject: Re: Problem with time-signal controlled alarm clocks

Michael,
      Thanks for your message, and warning about the time
delays for alarms with radio-controlled clocks.  We'll
need to check this out; if these delays are common, we'll
need to evaluate it for each model of these clocks, and
will need to do that for the American ones that use WWVB.
      David

At 09:59 PM 12/25/2002 +0100, you wrote:

Dear Dr. Dunham,

in your Occultation Timing Primer at
   http://iota.jhuapl.edu/timng920.htm
you have suggested recording the alarm of a radio time-signal
controlled clock on tape or the audio track of a camcorder
for timing.

This may be an easy solution with some alarm clocks, but not
with all of them.
I have tried with my alarm clock controlled by the German
DCF77 time signal and comparison with a home-built time signal
receiver accurate to approx. 0.01 s. At room temperature,
the LCD display of the alarm clock was less than 0.1 s behind
the true time, whereas the alarm was consistently 0.25 +/-0.01 s
after the time. Also all later beeps started at 0.25 s after
a full second. I guess that the reason for this "late" alarm
is avoiding radio interference of the alarm getting into
the clock synchronisation process (the alarm uses a fairly loud
inductive transducer, which probably creates a lot of
electromagnetic interference, including overtones near the
77.5 kHz time signal). The DCF77 time signal uses 0.1
and 0.2 second pulses starting with the full second
(the pulse length encodes time and date information). Thus,
a short beep at 0.25 s after the full second will not interfere
with the radio receiver any more, even during the periods
of clock synchronization.

Since my alarm clock is a fairly common one and there are
probably not too many types of integrated circuits for
time signal receivers around, I fear that many time-signal
controlled alarm clocks will suffer from the same problem.
Of course, as soon as a user understands the problem, he/she
can easily take it into account, but otherwise a 0.25 s
delay might go unnoticed.

Best New Year's wishes from Vienna,

Michael Schmid