I design coreless axial-flux motors (disclosure: my company Turncircles). No iron in the stator means zero cogging, which is why I think they could suit haptics, gimbals and robot joints here. The catch is on the control side, and I wonder if this community could help.
The motors have very low phase inductance: about Phase-to-phase Inductance: 9.5µH at Phase-to-phase Resistance: 141mOhm, 6 pole pairs, 1.91 V/krpm (L2L peak), run at 12VDC.
As far as I understand, that causes two problems:
Current ripple. At the usual 20–30 kHz PWM the ripple is large. Higher PWM frequency, series inductors or a different driver? What have you found works?
Sensorless. Back-EMF observers should be fine at speed, but low-speed and startup are hard. With no iron there is almost no saliency (Ld ≈ Lq), so HFI has little to work with. With an encoder it runs well; without one I’m not happy yet.
Is there anyone willing to test one of our motors with SimpleFOC, sensored and sensorless, and share results in this thread? We could send the motor.
Test motor: ~90mm, 200g, 0.15Nm, 6000RPM max. I look forward to your replies here or DMs.
If you want to size a motor for your own project, the configurator is free to try: cloud . turncircles . com. It also gives phase R and L, so you can check the controller side before building.
In summary it can run on 12V, does 50KHz PWM though I have not tested with a motor with that low inductance, lowest I have tried is around 50uH. Also runs custom firmware not based on SimpleFOC.
Currently I haven’t implemented any kind of sensorless operation in my firmware, so for it to be compatible with my driver the motor needs to have a sin/cos encoder where the encoder is aligned to the electrical revolutions of the motor, doable with 2 linear hall sensors picking up the rotor’s magnetic field. I’m quite busy with other work at the moment so implementing sensorless on my driver is unlikely in the near future, so I can only help you to test the operation with a sensor.
Let me know if this sounds good or this is not what you’re looking for. I’m based in Australia.
I just played with the numbers: correct me if I’m wrong but the motor will draw 85A at 12V and consume 1000W at startup? And because of the low inductance that won’t change much when it revvs up.
I only know coreless axial flux machines as generators for small windmills, where they only feed a 3-phase rectifier.
It’s hard to believe you will tame that beast to make it behave nice in angle mode without burning up something (PSU or FET’s or winding)
What was your design goal? 0.15Nm at 12V and 6kRPM is not unrealistic with regular motors IMHO.
Thank you for the detailed information. This sounds very interesting. It would be interesting for us to see how well the motor can be controlled with a sin/cos encoder. We haven’t experimented with this type of controller/feedback arrangement so far, so it could be a useful test for us as well.
The main concern I have is shipping to Australia. It is not only the shipping time but also the customs process, which is by definition quite painful.
One possibility would be for us to send you only the active parts of the motor rather than a complete motor assembly. Could you let me know whether you would be able to handle mechanical integration on your side? Would you be able to integrate the rotor/stator active parts into a suitable mechanical enclosure including the sin/cos sensor integration and alignment?
Thanks again for your interest. I think this could be a very interesting experiment.
I can handle basic mechanical integration, I have assembled a BLDC motor before for a test (the one in the playing music demo is the one I assembled myself). That process mainly involved screwing the sensor PCB to the frame, using epoxy to glue the stator to the frame and then sliding on the rotor and finally putting in the circlip to hold everything together. I’m not sure how your motor is designed to be assembled but if it involves anything more complex than that I probably don’t have the tools to assemble it myself (things such as press fitting I cannot do).
I can 3D print a jig to attach the motor to for testing. Though I am not sure what kind of mounting holes exist and what kind of shaft it has to take the output from.
For the sin/cos encoder the way it needs to be implemented to be compatible with my driver is 2 linear hall sensors near the rotor magnets, sensing the field from them which ideally should be sinusoidal shape. So usually the motor needs to be designed with this sensor integrated, retrofitting to an assembled motor is probably infeasible, although changing from 3 digital hall sensors 120deg apart to 2 analog hall sensors 90deg apart is probably doable without much redesign of the motor. The DRV5055 is a linear hall sensor I have used before. My firmware automatically aligns the sensor to the motor windings in the calibration sequence.
Customs in Australia doesn’t usually charge fees if the package is low value under 1000AUD.
Some pictures of the motor in both the assembled and disassembled state and a technical drawing would be helpful to clarify and give me a better idea of the feasibility.
If digital halls aren’t too close to the magnets, you can swap them out with analog and use Clarke transform to convert the 120° readings to 90° for use with atan2. Same as how current sensor readings are converted to the 90° DQ coordinate system. See the correction I added here:
Hi Andrew, there is not much flux leakage outside the active area. We should integrate the sensors onto the PCB stator. Let’s meet up and I can present you the active parts as well as the motor assembly. We can make a plan from there. Let me know if and when you could be available kaan.ozgen@turncircles.com
No, I simply used Ohms law.
I have wound/rewound hundreds of BLDC motors in the past and have a well known simulation tool to calculate the outcome (kV, currents, torque, efficiency a.s.o.)