[HuFlungDung]

Hi Dominic,

I don't think your conclusion about servos needing to drive through a reduction is correct, not torquewise anyways.

If you must use a gear reduction, there can be a problem with backlash if the encoders are mounted directly on the screw, while the servo acts through the gear train. What happens is that it becomes difficult to set the PID parameters for the servo because the feedback loop has to deal with this backlash. The motor needs to continually shift back and forth across the backlash amount in order to move the encoder to test the position. The motor will do this, but it is a form of oscillation and makes it quite difficult to find PID settings to give good performance, because the settling time is lengthy.

The best way around this is to simply make sure the encoder is mounted directly on the motor, then it will operate properly. However, the actual position of the mechanism becomes suspect. So for applications that really require the best of both worlds, you can use scale (or encoder) feedback off both the motor and the screw. I don't know much about how that is set up, but I know it is possible to do.

If your machine happens to be constructed in such a way that there is a constant load on the backlash affected axis in one direction, then the backlash can simply be factored into the controller software as a parameter to take into account upon every reversal of that axis. However, the mechanical load will take care of the motor "settling to position" problem. Then you could safely apply the encoder directly to the screw and not have a PID nightmare with the servo running through the gear train.