[time-nuts] Physical Clocks - adjusting

Chuck Harris cfharris at erols.com
Thu Apr 28 08:56:46 EDT 2005



Mike S wrote:
> What got my interest in precision time started was adjusting clocks - especially wristwatches. I've tried to keep
> track of watch time vs. WWV and tweak adjusters to make them better, but it's a very long process, and I've not been
> particularly successful, probably due to lack of patience and diligence.
> 
> So, with that interest, I thought to put together an electronic calibrator. My thought is to use a microphone or
> piezo disc somehow coupled to a watch to pick up the mechanical vibrations, amplify (filter?), and feed that into a
> TI counter. 3e-7 gets 1/sec/month, as much as can be expected with a normal wristwatch. That should be easily done,
> and allow adjustment in a matter of minutes. TVB's precision wristwatch isn't practical for my purposes. :-)
> 
> A commercial unit is shown here: http://www.bmumford.com/microset.html
> 
> They have a "simple watch sensor" ( http://www.bmumford.com/mset/access/watch.jpg ), which I guess to be a piezo
> sensor, fixed at one end with a brass post attached to the other, which in turn protrudes through the top of the
> case. Mechanical vibration in a watch in contact with the brass rod would generate the signal.
> 
> Anyone with any experience with such a calibrator? My biggest concern is that there will be too much noise in the
> signal to get an accurate TI measurement.

Hi Mike,

Watchmakers have used timing machines that work as you have described
for about 75 years now.  The machines that they use have crystal, or
tuning fork, driven oscillators that are divided down to a rate near 60Hz.
The 60Hz signal is used to drive a synchronous motor that spins a drum
that has a spiral of wire that makes one complete turn across the length
of the drum.  The motor also, through a gear train, advances a piece of
paper over the top of that drum.  You can think of the spiral of wire as
being a mechanism with which to create a scan across the width of the paper,
much like the horizontal sweep of an oscilloscope. (In fact, a watch timing
machine is simply a specialized version of what was once known as an
oscillograph.)

The watch is attached to a microphone of sorts that only hears physical
vibration.  All of the parts of this microphone are massive, so they don't
readily couple to sounds traveling through the air.  The sounds detected
by the microphone are amplified and filtered and become a driver signal
for a thyratron trigger tube that causes a solonoid driven bail to strike
the paper, through a ribbon, into the rotating drum and its spiral of wire.
A mark is made on the paper where the bail and the spiral of wire intersect.

The bail makes a mark on the paper which represents the position where
a loud sound inside of the watch occurred relative to the reference oscillator
that drives the timing machine.  A trace is made that will be straight if
the watch is running at the same rate as the drum, etc., or will be at an
angle if the rate is high, or low relative to the drum, etc.

The watchmaker can tell the rate of the watch, and see the operation quality
of the watch, in seconds.  This makes setting up a watch to a desired rate
very quick and easy.

After the watchmaker has set the watch, and given it back to the customer,
he tells the customer not to set the watch, but rather to wear it in a normal
way, and come back in 1 week.  At that time, the watchmaker makes note of how
much the watch is off of "real" time, and puts the watch back on the timing
machine, and resets the watch's rate, but this time, subtracting out the
"personal rate" shift that occurred over the week of the watch's being worn
by its owner.  The customer is now happy as a clam, because his watch always
shows the correct time within seconds a day.

-Chuck (an amateur watchmaker)



More information about the time-nuts mailing list