[time-nuts] Mechanical clock sound pickup circuit
cfharris at erols.com
Sat Dec 12 09:54:18 EST 2015
Back in the day, companies, like Vibrograf, Greiner, Elna, L&R, all
made timing machines that were based on the same principles.
The machines used a crystal, or tuning fork frequency reference, and
divided it down for the various standard (and not so standard) watch
beat rates. The divided reference signal was used to turn a synchronous
motor, which rotated a drum which had a single turn loop of wire wound
around it, in the fashion of a "one turn per drum width" screw thread.
The watch's movement was clamped into a mount that had a piezoelectric
element, usually a rochelle salt crystal, that turned the small
vibrations into an electric representation that was fed to a chain
of amplifiers, filtered to amplify only the audible component until
it was large enough to trigger a tiny little thyratron tube.
The thyratron tube was set so that when the impulse exceeded a
variable threshold, it would fire, and would discharge a capacitor
into a solenoid for an instant, and pull in the armature that caused
a metal "bale", that was as wide as the drum, to strike the wire that
is wrapped around the rotating drum.
Because this wire essentially proceeded along the drum as it rotated,
the intersection of the wire, and the bale formed a scanning element,
not unlike the beam on a CRT oscilloscope.
To record the impulse, a paper tape was fed between the rotating drum,
a typewriter ribbon, and the metal bale. When the metal bale "ticked",
it pushed the ribbon against the paper, and the travel was stopped by
the rotating wire wrapped on the drum, and an ink mark was made.
Ok, why all this complexity?
The idea was to, tick-by-tick, record the difference between the watch
movement's escapement's noise and the smooth flow of time embodied by
the crystal/tuning fork/dividers, etc..
By adjusting the gain of the microphone's amplifier stage, and as a
result, the threshold of the printer, the watchmaker could observe
quiet repeating noises all the way down to the pretty noisy tick of
When you look at these actual traces, your squishy ware's DSP can
easily see the slope of the group of traces, which is the rate of the
watch, and any rhythmic variations of the individual "ticks" recorded
on the paper.
You can see things like irregular spacing of the teeth on the escapement
wheel, and irregular spacing of many of the later wheels and pinions.
By adjusting the gain of the amplifier stages, and the resulting
shift in threshold, you can select out noises of different loudnesses.
And the speed of the thyratron, and charged capacitor allowed multiple
strikes of the bale during each tick.
You can see the noise made by the impulse jewel touching the tuning
fork, and the noise made by the pallet jewels touching the escapement
wheel... Lots of very interesting things that indicate the quality of
the movement, and the state of its lubrication... as well as a nearly
instantaneous indication of the rate of the watch, as it sits in the
various "positions" (dial up, dial down, pendant up, down, right left...)
All of this from the feverish minds of horologists back at the dawn of
the vacuum tube.
I would suggest that any programming you use for your tools do similar
Andrea Baldoni wrote:
> I decided to do some experiments with mechanical clocks, so I worked a little
> on picking up escapement ticking sound, with the idea of processing it and
> obtaining a "clean" digital pulse to feed a counter.
> So far, I have not yet been able to find the best way to obtain a digital pulse,
> but I have already built the preamp for the piezoelectrick pickup, that
> I used to feed the mic input of a PC sound card for spectrum analysis.
> The timing could eventually be done in software because the whole idea of
> measuring watches by picking up their noise almost surely doesn't allow high
> resolution anyway, but I will plan to try hardware solutions as well in the
> future. I hope to be able to measure the jitter of the clock, but it will be
> very hard.
> In the meantime, with the free software Biburo you can download here
> you can regulate your wrist watch.
> Best regards,
> Andrea Baldoni
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts