STM32 ZeST sensorless BLDC FOC

Just saw that article in the STM newsletter, wondering if simpleFOC could make any use of it?

1 Like

If it’s not part of a closed source binary blob lol

1 Like

To me it seems, the HighSpeedObserver (HSO) algorithm would run on many STM32 MCUs.
The motor tuning of sFOC could rely on values derived from that algorithm, like

  • realtime motor resistance
  • rotor speed
  • rotor position
  • no need for back-EMF reading at startup (openloop velocity will do)

Especially the last point is interesting for my motor/generator setup for wind-turbines

I think ST Micro is a constant source of interesting information for us, and this is no exception… thanks for pointing it out!

People seem quite hard at work on sensorless control in the # sensorless discord channel, and some are certainly using STM32.

In terms of their motor control SDK, we can’t use it or its code directly, even if license terms would allow it. But their examples are of course an important source of information on how to use and configure their chips, which is certainly part of their intention in publishing it.

And their algorithms, which I assume are based on academic research in the end, are certainly also interesting to look at, but if you want to implement them for SimpleFOC it will be necessary to do them quite differently in the end, to make them cross-platform, support floating point math, etc…

1 Like

Does anyone use the latest STM Cube?
Would be interesting to know if our beloved B-G431 ESC board would be compatible to run HSO. The newsletter only mentioned G4 MCU, so there’s a fair chance.

For the life of me I don’t understand why they chose to take the blob approach. A well written article on the fundamentals of the ideas and algorithm is an important step forward. A blob I can never use, I cannot depend on it for reliability. I can’t even be lazy and just try to mix and match their bits and pieces, that’s more work anyway in the end. IMO it’s not good engineering. If they only make $ on the chips anyway, they should fork or contribute to an open source modular system like simplefoc. Actually, come to think of it maybe we should ask them for some funding, perhaps to bake a sensorless drive with the next gen flagship board. That’s a way better way to sell chips than just handing out blobs like this. And even if it’s not binary, the approach they take gives C code that is so convoluted imo it might as well be binary.

A bit more information in this video and this presentation

HSO sounds a lot like HFI. I might be wrong, but that’s like something a management meeting would come up with just to make it sound different and slap a marketing moniker on a well known concept. I’m willing to bet $100 this HSO is just a proprietary HFI implementation for the ST hardware.


We suspect ZeST is some kind of HFI, but they are very secretive about it.
HSO is an observer and requires 3 phase currents/voltages.

It’s expected for ST to protect their IP, especially when revere engineering this is so trivial. There are plenty of very smart and dedicated people out there with a lot of time and $$$ in their eyes, and the only thing they need is an oscilloscope. All the best to them, let the smartest win.