I am getting somewhere on my fan design. I got the RPM down to 2000 with adequate flow and pressure. I turned away from simpleFOC because with an uno and the standard shield it was still causing a lot of motor heating in constant torque mode even when the motor was unloaded.
There was a lot of noise too, I am hoping I might be able to solve that using the Lepton, it appeared to be caused primarily by the time it took to run the control loop.
How is the energy efficiency I can expect with a magnetic angle chip encoder and constant torque mode? I’ve become partial to the idea of using an encoder because it provides faster start up and direction reversal, and this matters for me because the fan reverses direction every 10 seconds, and startup time affects net average airflow.
My plan is to re-order leptons and encoder boards (I have to redo the dimensions of the encoder boards so they are identical to the dimensions of the existing encoders, I want things to be compatible). UPS lost the package last time or I would have had em 2 months ago…
Bad plan? I really need the motor to have good energy efficiency. If it gets hot it will melt the plastic. Also the energy efficiency of fans in buildings is measured and rated and considered a significant factor. A bog standard trapezoidal sensorless drive board does a great job except startup is a bit slow and it makes too much motor noise. The fan itsself is super quiet.
I can run a hoverboard motor in free spin with only 0.3A in torque / voltage mode. It’s as efficient as it gets. I’ll post a video later. Actually in my case I use hall sensors. If I use an absolute angle encoder I can get even better results. No current sensing. If you just want to run a fan with a small motor, that’s really a lot easier. The hoverboard motor is like a huge two kilos piece of metal. My driver board however is a lot different (100V / 400A with a G432 mcu) but from algorithm point of view that doesn’t matter.
I have another question that is kind of related, if anyone is out there?
I noticed with my experimentation with the MCF8316 eval board, the system needs to be told the inductance of the motor phases, back-emf voltage and resistance of the motor phases. If the inductance is off by even 15 percent, it makes a noticeable buzzing noise.
in my discussion with the guy from texas instrument, he says that in order to implement FOC, these values are needed. I don’t need FOC, sine wave would be fine.
But how does simpleFOC get the info it needs about the inductance and stuff? I can measure it, but I don’t remember seeing anywhere to input that info. I went through the thing to get constant torque mode working without encountering that…
This is something we are working on, and with a bit of luck it will be part of the next release. In the same way you can set the motor resistance and KV, you will be able to set the inductance.
The algorithms will then take it into account when modelling the motor physics.
At the moment the models don’t include it (which negatively impacts performance, but only at high speeds / power levels).
awesome, if initial exploration with the lepton goes well maybe I can contribute a couple hundred bucks to help with this. I don’t want to pretend I’m mr moneybags, this undertaking is pretty sketch, but it might be sensible. Is 2000 rpm high speed?
For motors, depends on the motor, its KV and pole pairs. For the MCUs, the variation is also wide. For an UNO, that’s probably far more than it can do.
For a fast STM32 MCU or ESP32 it should be achievable.
The uno maxed out at like 200 rpm, I tried it with the shield. I forget exactly but it was extremely slow. The motor is a ~160KV motor with 7 pole pairs running at 24 volts. So the motor is not the bottleneck. If the lepton doesn’t work out I don’t know what I’ll do, but here’s hoping.
