How to auto-calibrate a PMSM motor and encoder

[RubelCA]

Hi,

I found a nice thread where it was explained how a PMSM (brushless DC motor) is calibrated for torque ripple and other manufacturing and magnetic tolerances:

http://www.eng-tips.com/viewthread.cfm?qid=134381

Is there anything else or new/different/easier that can be added to this methods?

My next question is how to calibrate a quadrature encoder when connected to a PMSM motor for shaft alignment problem when installing and/or manufacturing tolerance. I am developing a servo motor drive for a 100 pole PMSM motor. The problems I am facing is as following:

1. The speed servo loop has ripples. I am using a 20000 step/rev quadrature encoder. I believe this partly happening because of the motor torque ripple, magnetic and manufacturing tolerance and mostly due to encoder alignment error. Now, how to calibrate an encoder when the open loop rotation the motor itself is not smooth due to detention torque or cogging and other nonlinear torque?

2. I am looking for a method to calibrate the encoder and the motor when the motor is running freely without any load while commissioning the servo motor.

Expert opinion will be highly appreciated. Thanks and regards in advance.

RubelCA

 

[blacksea]

Most critical for your design will be , back-EMF quality (harmonics,imbalances) and amplifier quality (offsets, imbalances). Of course sine commutation and core-less type of PMSM will be preferred. As a rule not need calibration of digital encoder.

 

[cswilson]

The only "calibration" of the encoder that I see is to figure out (or set) the orientation of the once-per-rev index pulse on the motor. Once you do that, you will just be using the measured position relative to that pulse to calculate whatever corrections or compensations that you come up with as a function of the motor angle.

 

[RubelCA]

Thanks, guys! Suggested by your keywords, found some good journal papers explaining the processes in details. Although math intensive, I believe, I'll be able to grasp it because of the rusty control engineering math background.