RE: Advice on Digital Servos for drive by wire throttle
I'd be hesitant to use hobby parts in an automotive application.
As far I as can tell, RC parts are designed to consumer grade requirements with maybe some attention paid to vibration and water/dust proofing, but not much. 100 C is going to have a noticeable effect on failure rate.
In the hobby application, the servos are typically in use for only 10 to 20 minutes at a time. People who fly a lot (say four flights every day for competition practice) will replace servos each season as a PM item.
My personal experience with about 60 servos over the last five years has resulted in four servo failures, and a couple that occasionally misbehaved in a way that discouraged me from using them in mission critical applications. Most of the failures were premium servos in demanding applications (mostly helicopter tail rotors with fast gyros).
As for your specific questions, I'll answer what I know.
The absolute best resolution you could expect would be 10 bits. Servos typically have a deadband specified, which is the amount of input pulse width error required for the servo to take corrective action (i.e. move). A very low deadband might be 1 us. The input pulse width range is 1 ms, giving around 10 bit resolution. Deadbands higher than this are much more common. I would expect something less than 10 us in most cases.
There is no error integration. I've never measured digital servos, but the normal analogue servos will typically need about 5 - 10° deflection to develop rated torque. There is no standard for the amount of position error required for torque measurement. Attempts have been made to find this out from the manufacturers, but a definitive answer has not been forthcoming.
The 9451 is representative of the state of the art.
Some servos come with Al heatsink cases.