Multi pole servo motor control at slow speeds

I was doing some tests the other day on slow speed control (1rpm). I thought that sinusoidal commutation and a high resolution encoder (4K CPT) would be "the thing" for controlling a multi-pole flat motor well (with detent torque - naming the EC90 flat from maxon). This would enable the controller to have better control during the commutation between coil switching. During the commutation the natural attraction of the rotor to the poles pulls and pushes the rotor into position. This is something that I thought sinusoidal commutation would help with. I was pleasantly surprised that using another controller that had just block commutation but a 53.5Khz current loop proved far superior in comparison to the 10KHz control on the sinusoidal drive!
<- - Make a Comment - ->
If the motor has detent torque then you are going to fight to get smooth performance every step of the way and it probably has a high harmonic content BEMF. Motor designers skew the stack to smooth out this detent torque and to provide a sinus BEMF waveform. Non sinus BEMF waveforms and detent torque are all disturbances into the control loops of the motor. Comping them out going to be a battle.
<- - Comment made by: Bill Kraz - ->
a 53.5Khz current loop of the amp won't cause a real bandwidth with the same value. It's dependent on the motor value's like resistance and inductance. In practice you'll reach a bandwidth in the 1kHz area. I think your new controller might have a better algorithm and possibly is better in coping with the motor cogging (at cogging points it's not only the cogging torque, but also the varying inductance that distrurbs smooth control).
A higher res encoder won't help you any further i think.
<- - Comment made by: Dirk Godz - ->
The "slottless" motor directly addresses the root problem by removing the back-iron in the stator which creates the detent torque in the first place. This, combined with a good current loop (20 kHz or more) , high resolution (4K CPR is fine, more is always better) feedback, sinusoidal commutation, multi-pole motor, skewed magnets ... all are tools to eliminating velocity ripple at low speeds.
<- - Comment made by: Dave Bedro - ->
Smooth performance at low speeds - everything is relative - but if you want the best performance: I recommend good current sensors (DC Offset) with a high resolution / accuracy position feedback. We see 1024 line sin/cos encoders giving 1 part in 4 million feedback resolution and 30 arc min accuracy as being capable to very accurate low speeds. However, everything component works in a system and its only good as the weakest link - you may need PWM resolution as well, depending on L/R.
<- - Comment made by: Sicr - ->
There are many drive and control companies providing closed loop vector control for stepper motors these days. Perhaps you can approach your problem with that type of solution. Stepper motors exhibit all the stuff previously mentioned and more...high pole count, harmonic distortion, high detent torque... We have tried a variety of them and have been very pleased with both velocity control and positioning performance. I think Bill already mentioned that velocity is relative to the user but we have had great success and we are a high precision stage builder.
<- - Comment made by: Mark Cacha - ->
Torque detent, trapezoidal commutation...are all disturbances at the torque level. How well a velocity loop can compensate/regulate with the presence of those torque disturbances depends on: current loop bandwidth (the inner loop), velocity measurement (and at low speed signal-to-noise ratio) and velocity control algorithm and update rate. So I can see a faster current loop helping (even with the simpler, torque ripple inducing 6-step commutation), but maybe the 2 controllers also had different velocity control parameters. From control theory we know that the inner loop bandwidth dictates what can be achieved with the outer loop.
<- - Comment made by: Jan bost - ->
You may like:
serviceServo Motor & Drives