My watch has been out of beat for some time after a slip of the hand while attempting regulation. It has been keeping excellent time on my wrist regardless but I finally decided to attempt to get it in beat the other night for peace of mind. Everything seemed to go very smoothly so I'm interested to know if I've made any incorrect assumptions!
I used a PC running Audacity (an audio sampler), a pre-amp, and a guitar pick-up mic. The movement is a Tudor ETA 2824-2.
Firstly, I'm not trying to time the watch - I tend to regulate on the wrist over a period of a week and the signal through the contact mic is not clean enough to use timing software. Seeing the waveform in Audacity however, it is easy to manually identify the start of each beat. One key assumption I've made is that getting a watch in beat does not require calibration of my soundcard as I simply need to measure the percentage of each beat in relation to the total time for both (which, incidentally, Audacity seemed to get very close to the correct value of 250ms, i.e. 28,800bph). My aim was to use the ms value out of Audacity as a guide but really to concentrate on the percentage difference and get that as small as possible.
I moved the stud carrier and re-recorded until I managed to get the percentage difference down to roughly 0.05% (approx 0.1ms difference) from a starting point of a 0.8% difference (around 2ms). As I never manually wind the watch, I performed these recordings after a full day of wear (assuming that is an average 'maximum' wind). I only tested dial-up. Also, I used just two beats (one tick, one tock) rather than analyzing over many beats. When I saw Audacity was coming back with a consistent 250ms for both beats each time I recorded, I felt that was okay.
Before I start testing different positions, have I missed something here?