I just watched Benjamin's tuning video https://cloud.vedder.se/index.php/s/jpm9Soiq68HBbdw and learned a lot.
But there's something I have been wondering all along: Why use sensorless mode when the motor has sensors?
In my case, it's a Golden Motor HPM5000 which I imagine was originally intended to be run with trapezoidal communication. As such, it has 3 Hall sensors.
Now I'm running it FOC and VESC Tool's default setup is to change to sensorless after some minimal speed is achieved.
My guess is that the observer can do a better job of predicting rotor position (assuming the setup is optimized). I imagine this may be especially true under transient load / command conditions.
Any insight appreciated!
Run that motor with half or a quarter of the observer gain. Use the latest VESC-Tool for detection. Yes, the Observer does a better job after X RPM and we therefore switch off the Hall. On this motor the Hall sensor setup is sort of funny, using extra magnets for the hall probes. On some of those we had issues with the hall probes. In general Hall probes don't do a good job at higher RPMs.
Also you can run in great trouble if the sensors fail at speed. So for most applications sensorless is the way forward and sensors are used only for startup.
Frank, could you elaborate on this? It calls into question a fundamental motor characteristic I have been confused about.
I was pretty well convinced each Hall sensor was emitting a square wave at 4x the mechanical rotation rate (partly because the motor has 4 poll pairs).
Now I'm thinking it may only be a factor of 3x.Golden Motor 2.jpg
Media Folder:
Excellent! Much appreciated. It is quite different from the image purported to be an HPM5000B posted here (reply #3): https://goldenmotor.com/SMF/index.php?topic=7054.0
(I do understand there were two versions of that motor.)
I will reciprocate with a photo of my own. (It would seem the image must be hosted externally?) Can't seem to make the image visible. It's a .jpg hosted under Google Documents. Hmmm, if you right-click inside the image box, choose "open image in new tab", and wait patiently, it will show the image.
I installed a temporary index-pulse mechanism on the fan side of the motor to produce one pulse per mechanical revolution. I then spun the motor using a hand-held drill.
Channels 1, 2 and 3 are the Hall outputs. Channel 4 is the index pulse. It's clear there are exactly four complete Hall cycles per revolution for each of the 3 phases. I annotated the screenshot in MS Paint to count the number of Hall edges and show the sequence in which they occur. As expected, 24 edges can be seen (21 and 24 are obscured by an onscreen display). The time between the index pulses is measured by the cursors at approximately 200 ms (or a repetition rate of 5 Hz). In other words, 5 pulses per second which equates to 300 mechanical RPM.