CyberGene With the automatic software calibration how would you know what coefficient you should use to transform measured speed to MIDI and whether one will suit the entire keyboard?
The calibration I am talking about in this context is the maximum value that the ADC can read. As such, it's extremely easy to calibrate: just ask the user to play every key (making sure the hammer reaches the stop bar) and read the maximum ADC value as, well, "maximum ADC value". Once done the first time, the software could even (if desired) adjust it continuously for the future.
Obviously this is not the only calibration needed even though a similar procedure can apply for some other needed calibrations which in your case are purely mechanical (e.g. what ADC value to consider for escapement, which I believe in your case is vertical adjustment of the sensor)
That said, the other objections you raise are valid for yet other settings, e.g. how to make a MIDI velocity out of a set of measured ADC values. This has been discussed at length in a earlier thread, but this is not what we are discussing here.
Here we are purely discussing the fact that @stem_piano ADC maximum achieved ADC value changes per-sensor and that he is using potentiometers to make them equal. My argument is that if the ADC has enough resolution, one does not need to make them equal and can simply rescale the read ADC value according to the maximum read value. Of course this does not apply if one utilizes a too low resolution ADC.