[time-nuts] Mechanical clock sound pickup circuit

[Andrea Baldoni]

Hello.
One answer for all.

> Peter
> The watch or clock produces many sounds during its operation and the issue
> is selecting the required one.  As you wish to investigate the timing

I noticed it, unfortunately it seems that often the different sounds are very
similar in level. I'm thinking about to make a multichannel system to take
maybe advantage of the phase.

> Chuck
> 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.

I see that the software solutions out there mimic the capabilites of the
original printing machines (you gave a very nice description of!).
This is extremely interesting in the horological sense; my goal instead is to
obtain a single signal out of the many, the more representative of the actual
balance movement as possible, to measure its own jitter.

> Bob
> You *may* find that moving the passband of the mic up above 4 KHz helps things a
> bit.

I found that the energy goes up to 20kHz in my setup (further, I cannot
measure). I thought that being a very short "click", it's the
sound equivalent of a step function, filtered by the elastic response of the
contact surfaces, bearings, case, piezo pickup, etc... Surely not all watches
are the same (in particular those made of plastic), I should investigate
further, or the assumption of step function is not correct from the start.

Actually the preamp has a highpass filter at 10kHz and the ambient noise has
been completely eliminated.

> Dave
> Someone is in the process of writing open-source watch timing software.
> You may want to look into it.

Thank you for the link, it seems good to do experimenting and troubleshooting
of the audio capture part.

> Alexander
> to get a reliable digital signal from a noisy analog signal is the most
> reliable way to use an analog PLL with a linear multiplier type phase detector
> ...
> low pass filter or integrator, the price of the method is that it also
> eliminates the phase-noise of the the input signal, .

In effect I'm exactly interested in the phase noise of the source, adding
the least possible noise in the capturing process. The varying
sound levels coming from the watch from tick to tick and the different sounds
coming because of different mechanical parts interactions make this a
challenge. The various timing machines instead focus to those different
sounds to infer the mechanical condition of the movement.

I'm thinking to process the signal through an antilog amplifier so
peaks will be sharper.

> Azelio
> Maybe this can be useful to make the pick-up:
> http://www.meas-spec.com/downloads/LDT_Series.pdf

Initially I thought it could have been possible to infer clock ticks by
the fact there is a moving mass inside it, but they are (fortunately) so
perfectly balanced, that they don't vibrate on the scale of the Hz/tens of Hz
those flexible piezo films are aimed to pickup.
The only thing you can pickup is the "noise" coming from the contact surfaces,
on a totally different frequency scale they are not sensible to.

Best regards,
Andrea Baldoni