I see a lot of discussion around cogging torque in BLDCs. If you must have very low cogging torque for your application, you’ll eventually need to get to the point of ensuring the motor itself is designed for low cogging due to reluctance torque generated by the magnets and poles of the stator. In industrial AC servo drives, this is accomplished by skewing the magnet poles and designing them for a broader, more de-centralized magnetic field. The tradeoff is a lower motor constant and probably lower peak efficiencies. I’ve had really good results using SimpleFOC to drive surplus AC servos i find on eBay from time to time. With practically zero cogging torque, I’m able to control position down to the limit of the encoder - easily down to sub-degree repeatability. As a side note, pole pair checks usually fail with these motors due to the skewed magnets, but FOC works beautifully.
I’ve attached some pictures of how the rotor and stator of these motors are designed - this one was rated for 500W and several N-m of torque.
Very interesting! Do you think that a sufficiently advanced scheme to model or measure the torque ripple and then compensate in software might be workable though? Unfortunately building a good quality motor is quite a project. If we can instead pack the skill and knowledge into software then that may be easier to roll out to more people and lead to greater utility for us as a group.
Yes this should be as simple as that.
I’ve fetched some docs from research gate while back where they explained just that. There were different patterns, offsets and shape of magnets.
Alternatively a multiple magnet ring would work too. Much weaker and suuuuuper expensive to make.
I am noticing one could just put two motors in a line, on the same shaft, with the rotors slightly rotated relative to each other, this might end up cheaper than trying to make your own motor. Gimbal motors have hollow shafts so you could do that.
Oh, yeah, it’s a low resistance/high kv type intended for spinning propellers fast. But the point is that split ring motors do work well and don’t add significantly to the cost, so hopefully we’ll be seeing a greater variety of them on the market in the future. I’m surprised it’s not the standard construction for gimbal motors, which are intended for precise direct drive positioning.
I agree with you on the last statement. I believe this should be a standard construction for gimbals and ptz type or anything that requires low speed and high accuracy and decent torque.
On the cost really depends. Samarium Cobalt multipole radial magnets are expensive to manufacture according to few companies that i talked to. We talk here about 15USD per piece at MOQ of 500.
NdFeB are bit cheaper. Alnico would be chapest i believe. And also it seems that OnceTop motors use both alnico for multipole magnet as well as diametrical magnet for sensor.
I have a couple of Nema230motors from Anaheim Automation. 4poles, but if i rotate the shaft zero cogging. I suspected that they used this kind of offset.
They run 0.08% speed deviation during open loop commutation.
