PMSM FOC control using hall only

[frank78]

Hello,

I need to adapt an exisiting PMSM controler to a new motor.

The existing design is a FOC using a 4096 step encoder, based on an example project from the NXP website. The PCB has a Kinetis K53.

For the new project, I need to keep the hardware and only adapt the software. The new motor unfortunately does not have an encoder, but only hall sensors. Since BackEMF cannot be measured, the only position information I have are the hall sensors.

My questions:

Is it possible to use FOC for the new motor, even without precise position information? (60 deg steps)

Does anyone have experience using a position estimator from hall sensors ?

Thanks in advance.

Frank

[Daniel_Popa]

Hi frank78‌,

You can compute the rotor position quite accurate using the hall sensors as an input to a speed estimator. Then you integrate the speed and obtain the position needed for PMSM Clarke/Park transformations.

Best regards,

Daniel

[frank78]

Hi Daniel,

Thanks for your reply. I think this is what I will try.

Has anyone already used FOC on an estimated position ?

My fear is that it won't be stable since the position is not the real one, only the linear integration between 2 hall positions. This in return causes position jumps at each hall interrupt, when I update the estimated position with the real one.

Maybe I should integrate twice to also take into account the acceleration ?

Best regards,

Frank

[philip_drake]

Hello Frank,

If the PMSM hardware is going to stay the same, then the contol feedback is a question I have. What is provided back to the controller for phase current, and DC bus voltage?  Are there shunt resistors in each of the phase lines?  If so then a sensorless FOC control is possible. If not then a BLDC 6-step control is possible with the hall sensors. 

For the person designing a motor controller like this I recommend you start with Kinetis Motor Suite. The GUI is a free download.  KMS is able to be run on a number of platforms within the Kinetis KV series of MCUs including FRDM-KV31, FRDM-KV11 and TWR-KV31 and TWR-KV11 Go to www.nxp.com/kms to check it out. 

To answer your questions FOC control can be done sensorless or with a precision encoder.  A motor startup can be enhanced using the hall sensors but this requires special software projects that currently being developed now. 

If you would like to explore how to implement the FOC on Kinetis K53 look up DRM148 . DRM144 shows you how to do BLDC sensorless control with Kinetis.  Application note AN5169 - Three-Phase BLDC Sensorless Motor Control Using the MKV4x In Quadcopter Application is a good read as well.

Best Regards,

Philip

[frank78]

Hello Philip,

thanks for your answer.

The board has indeed current feedback in each phase (shunts to GND) and a measure of DC bus voltage. However, there is no measure of Back EMF since the original design used an incremental encoder.

Do you suggest I determine the positon from hall sensors as 60° steps and use this in Park + Inverse Park ?

Thanks in advance for your feedback.

Frank

[philip_drake · Accepted solution]

With the current feedback you have you can use the Kinetis PMSM sensorless FOC control. Is the Kinetis MCU staying with K53.

You also have the option to use a simple 6-step Trapezoidal BLDC control with the hall sensors.

Philip

Sent from my iPhone

[frank78]

Hello Philip,

the project is a proof of concept for our customer, to reduce cost & time I have to use an existing product. If the proof of concept is successful, he will order a new design, where we can change the MCU. So far I stick with K53.

Thanks for your help.

Best regards,

Frank