[Feature] Custom periodic tick marks on the time strip graph

[xyzzy42]

Problems with a wheel in the train can show themselves as periodic changes in amplitude or rate, with the period equal to the period of the wheel. E.g., one minute for the center wheel.

The time strip chart current has marks for 10 seconds and 1 minute. Allow custom tick marks, perhaps as marks on one side, at a custom interval. This can be set to the period of one of the wheels.

To be more accurate, the tick marks should not be based on elapsed time, but on ticks. The time would be more or less, depending on how slow or fast the watch is running at the moment. For example, a 15 tooth escapement in a 18,000 bph movement takes 3 seconds to revolve. But the the marks would be every 7.5 tick-tock pairs, not every 3 seconds. But just counting ticks would lead to long term errors for longer periods like the center wheel or the mainspring barrel, as tg will miss a few ticks due to problems with audio noise or quality. So it would be necessary to try to account for that.

Instead of just a single user supplied period, the user could inter the wheel train's information. The teeth on each wheel and the number of leaves its pinion. Then the period of each wheel could be automatically determined and multiple wheels could be shown at once. This might make it easier to identify when an error corresponds to the period of a wheel.

Instead of just a mark, a continuous sine wave could be drawn. The sine wave would show the position of the wheel in its revolution.

[petergeelhoed]

Is this what you are looking for? https://www.watchuseek.com/threads/can-you-explain-this-timegraph.5332495/page-2#post-53910120 My code sits here https://github.com/petergeelhoed/tico

[xyzzy42]

Yes, something like that.

I was thinking maybe there was some algorithm to do this automatically. Consider if we a time series of rate (x sec/day fast/slow) data. Suppose we convert this from time domain to frequency domain. Would we see peaks in the spectrum at the period of the escape wheel, the 3rd wheel, etc.? Based on how much rate of the watch has periodic cycles.

Would techniques from harmonic analysis be better?

One also has the non-periodic and non-linear effect of the main spring winding down. The time series isn't stationary. Should one try to remove that? E.g., look at lagged differences.