servo failure on matrixpilot
Paul and Matt,
In general I agree with your comments about autopilots stressing the servos. Some additional thoughts:
1. The main problem that I have seen is that the small, extra motion of the servos creates a "scrubbing" motion of the potentiometer wiper arm, that gradually wears away the resistive material on servos that are less well made.
2. The result of the loss of resistive material is a deadband for the potentiometer.
3. The combination of the feedback of the autopilot and the feedback in the servo compensates for the deadband for a while. But at some point the deadband is greater than what can be compensated for, and then the control for that axis falls apart.
What I now do is to check the servos each time I power up for a flight. I move each servo a little bit in manual mode, and look for smooth operation. If there is any indication of twitching, I replace the servo. I have had to replace a servo on three occasions for this reason in all of the time that I have been flying with an autopilot.
I have tried your dead zone idea with my simulator and it is interesting. -3° +3° of dead zone on the servo command and the servo stops jittering when the plane flies straight amid high turbulence. The performance is not screwed up too much. The lateral error is less than 10 meters.
Yes, solutions are the same as with any LRS system:
ferrite coils on servo wire (near servo, if possible) - few winds of servo wire
twist servo wires
good antenna (with balun) - also helps a lot
picking right servos also helps (some tests on RCG - look for "900MHz VTX servo interference", but data is old. I usually choose from big bag of servos)
caps on servo H-bridge input (sometimes hard to do - not much space inside, but usually even small tantalum+ceramic cap helps)