The low quality of almost all motor drivers currently available, in my limited experience

Ok, so I have been through about 8 different motors and maybe 12 different motor drivers, more if you count the ESCs, perhaps. The only sensored solution I tried was my own, though.

What I have found is that at the exact same power output of the motor, the energy efficiency, acceleration and noise emission vary greatly. And they all suck at at least one thing. but as I previously posted, and just this morning confirmed, even the best of these drivers, the texas instruments MCF8316A, which is otherwise the best and supposed to be a really good sensorless FOC driver board for fans, has very poor performance.

This morning I got my fan working with a 6 step commutation board for the first time and it gets the energy efficiency I was able to obtain by manually adjusting the voltage on the power supply to find the optimal point.

And yet it does it reliably, with great acceleration. However the noise emission, which is a result of the six step stuff, mainly, is too much.

My point is basically that motor drivers in this class are not very good, with current market availability and systems on offer, even the best of them can be beaten without difficulty, theoretically. The texas instruments driver had a terrible problem with reliability, could not drive some motors at all, and got 260 milliamps at 24 volts under exactly the same conditions that the 6 step board got with only 220 milliamps.

The whole point of FOC is supposed to be torque, energy efficiency and noise??? And that was the most promising chip I could obtain. I also tried one from allegra which was completely useless, it advertised sine wave drive only, not foc. I could not find any other FOC based drivers which I could actually get a dev board for, and given the likelihood of things not working I’m not keen on trying to buy a chip, get a board made with it etc. etc.

So just saying there is actually at least in my little area of exploration, which is sensorless drive for a fan, a ton of room for improvement, which e.g. the next generation SimpleFOC board combined with SimpleFOC could totally beat. I’d have to figure out sensorless commutation, though… With a good lepton 3.0 type board there are a lot of options, though, like in a pinch I could use strategy X which is good except susceptible to, rarely, stalling, and then fall back to back emf for commutation if it stalled, to recover from the stall. Or maybe I could get away with some kind of back emf commutation with sine wave drive, but my understanding is that the non-powering of one phase was part of what’s responsible for the noise, which goes with the back emf position sensing strategy…

I would like to use a motor with hall sensors but I am afraid water will damage the sensors.

Do you know if your motor produces sinusoidal or trapezoidal back EMF? As I understand it, maximum efficiency is when the applied voltage exactly cancels the back EMF. I think 12N14P is usually sinusoidal, but higher efficiency with 6 step may indicate otherwise. I’m not sure if FOC automatically adapts to the shape of the motor waveform, or only ensures that the output wave is in phase with the back EMF.

Something you could try is feeding it an actual trapezoid rather than the square wave output by Trapezoid120 mode. Add a time percentage to ramp the output between the extremes, and adjust it until current is minimum. That ought to reduce the noise while also maintaining high efficiency for trapezoid motors (probably even improving it since the back EMF doesn’t have instantaneous changes either). Could try cosine interpolation too, so the wave looks like this:

1 Like

I’ve used the interrupt-heavy step-dir listener with a motor in torque mode. Everytime the steps stop coming in but the motor still works on emptying the step-counter, I hear it running much smoother. The interrupts have a significant influence regarding noise.
Same applies for sensors with interrupts.

I have checked with oscope, it is sin wave, pretty nice one which is not what the standard advice indicates. it has curved magnets, the sm tpe, not ip type. the ip type is flat magnets, that is supposed to be sine and sm type trapezoid like you say. But the voltage from the scope doesnt lie…

I have previously done just as you say and it is a good idea, it surprisingly had no impact at all. The noise from the 6 step driver I think is because one coil is unpowered at all times