I DO like the idea of programmable servos vise the matchbox, but I don't think varying TX pulses explains why every servo shouldn't be able to be mechanically set to a prescribed "zero" position. A TX "center" position with no sub trim offset should give a pulse width of 1.5 mSec going to the servo. I think that was Russian's point was this : theoretically, if you put two servos on a y-harness, they should both be able to be centered exactly the same mechanically (90 degrees to their arms, or whatever measure you like, but BOTH servos should position the same in this way without doing anything special -- correct me if I'm wrong, russian).

It seems to me, that the mechanical centering difference between servos issue is really the inability to mate the pots to a known output gear position. This is an issue on every servo, even analog servos. Once you put all the gears inside the package, and then locate the potentiometer to the output shaft which indicates actual spline position, I think it's tough to actually align the pot so it's dead nuts one spline to another when compared to a different servo. That's why it's just a lot easier to re-program the center position, which can be done much more precisely than gear or spline teeth can be positioned. Pots have manufacturing tolerances, and I just don't think it's possible to electronically trim them so they will match mechanically in this sort of system every time from one servo to the next.