search

xyzzy42 / tg

A program for timing mechanical watches
GNU General Public License v2.0
34 stars 10 forks source link

Support for cylinder/pin lever movements #6

Open LennertVA opened 2 months ago

LennertVA commented 2 months ago

Not a bug or issue really because from what I can tell, the algorithm is tailored to swiss lever movements - but is there any chance of fully supporting pin lever movements and/or cylinder movements? I very regularly repair/maintain such old movements but they are incredibly difficult to get a timing for. The (chinese) Weishi timegrapher I generally use cannot make sense of them. Trying tg occasionally gives some results but not all watches and not all the time.

xyzzy42 commented 2 months ago

I don't have any such movements, or good recordings of their sound, so I don't think it's something I can add without that.

It should possible to support. I'm surprised a pin lever doesn't work. I don't know what the difference in the sound would be that is preventing that. Do the pins drag against the escape teeth more? That might create enough tick to tick variation in sound so as to confuse the auto-correlation.

Cylinder is of course quite different. Basic timing should be possible without any significant change to the underlying algorithm. I don't know how measurement of amplitude on a cylinder escapement would be done. The sounds of the impulse pin hitting the fork and the fork hitting the banking pin are how it's calculated on a lever, and of course that simply doesn't exist.

LennertVA commented 2 months ago

I don't have any such movements, or good recordings of their sound, so I don't think it's something I can add without that.

I have plenty, so I can provide as many recordings of different pin levers as you want. I modded my Weishi so it has a jack-out - so I can use its microphone as input for the laptop, it gets very nice signal.

I did notice that I was still using a version 0.5 so I'll do some retesting with 0.6.1.

Cylinder is of course quite different. Basic timing should be possible without any significant change to the underlying algorithm. I don't know how measurement of amplitude on a cylinder escapement would be done. The sounds of the impulse pin hitting the fork and the fork hitting the banking pin are how it's calculated on a lever, and of course that simply doesn't exist.

Indeed. I'll try to get a clear waveform to look at (but will have to find one that I know runs okay - my current cylinder project runs bad and that was the initial reason for me to look around and ending up finding tg in the first place).

xyzzy42 commented 2 months ago

Try the branch new-stuff. The spectrograms of the sound would be interesting to look at.