[time-nuts] Watches

Chuck Harris cfharris at erols.com
Sat Dec 1 11:03:19 EST 2007


Hi Didier,

In the days of mechanical watches, every watchmaker had a watch timing machine,
such as a Vibrograf, Griener, L&R, .... .

What the watch timing machine did, was take an accurate crystal oscillator, and
divide it down to the various ticking rates of the mechanical watches of the
day.  The most popular rate was 18,000 beats per hour.  The reference in the
timing machine drove a synchronous motor that spun a drum, and the watch's
noises were amplified to a level that could move a tapper, and the tapper
hit a spiral wire on the rotating drum, and left a mark on a slowly advancing
piece of paper.  This left a record of the watch's rate as compared to the
crystal's rate.

Then the Bulova Accutron came on to the scene, and the watch timing machines
added an inductive loop that could sense the coils that drove the tuning fork,
and a pulse generator that divided the signal heard by the inductive loop into
pulses that the timing machine could record (18000BPH).

Everything else remained essentially the same.

The quartz watch changed the picture slightly.  The only audible sounds
were a very faint 32KHz mechanical vibration, and a 1pps tick of the stepper
motor.

There was however, a rather peppy 32KHz E-M signal.  So, most of the machines
either listened to the mechanical tick, or sniffed for the RF signal of the
quartz oscillator, or both.

The easiest way to go for someone with a time-nuts sort of laboratory is to
use a coil of wire that is resonated at 32KHz, an oscilloscope, and a nice
fractional divide by N synthesizer (ala HP3336A,B,C or HP3586A,B,C, ...)

The synthesizer can either be used to trigger the timebase, or as the horizontal
input, depending on whether you prefer Lissajous patterns or sliding waveforms
for making your adjustments.

You can also use a good reciprocal counter; though, some signal processing will
probably be required for that route.

In any case, the technique used is to set the watch to your reference clock.
Wear it for 1 week, and then note the exact time difference from your reference
clock.  Calculate the number of PPM offset.  Next measure the watch's 32KHz
oscillator using the loop of wire.  If you use a Lissajous, or sliding waveform
to make this measurement, your synthesizer will end up set to the watch's
frequency.  Add or subtract the offset, and reset the synthesizer.  Then
adjust the watch to exactly the new frequency.

Doing the adjustment this way will result in the watch being slightly wrong
on your bench, but the environmental variations that happen as you wear your
watch will be compensated out so that your weekly rate will be nearly perfect.
If it isn't, try again.

Didier Juges wrote:
> I believe watchmakers have a device they use to measure the vibration from
> the stepper motor or the escape mechanism and indicate if the watch gains or
> looses time. I am not sure how accurate that system is, and if something
> equivalent is in use on crystal watches.
> 
> Chuck, can you tell us?
> 
> My son's Bulova was at the repair shop for something like 3 months
> (August-November, I got it back last week) for what I initially thought was
> a dead battery (the watch was erratic). The watchmaker said a capacitor had
> to be replaced (did not ask which, I assumed it was the trimmer capacitor).
> He further said the factory sent the wrong capacitor 3 times, after which he
> decided to replace the entire movement instead, hence the 3 months and $60
> repair (after discount).
> 
> I am not sure if my son's watch is crystal controlled or some other system,
> I know some Bulovas used to use a mechanical tuning fork resonator
> (Accutron?) His watch is only 2 or 3 years old.

Real tuning form Accutrons are collectibles now, and it is not unheard of for
an unscrupulous watchmaker to steal the movement out of one, and replace it with
a cheap quartz movement, all in the name of doing the watch's owner a favor.

The economics are thus:  The Accutron movement can bring several hundred dollars
worth of parts, with the tuning fork coils being the most valuable, and the
quartz replacement costs between $2 and $20 depending on how generous the
watchmaker was feeling.  If this happened, your Accutron will now go tic...tic...tic
instead of hummmmmmmmmmmmmmmmmmmmmmmm...  And it will have been transformed from
a significant antique relic to a worthless piece of junk.

-Chuck Harris



More information about the time-nuts mailing list