Posted on June 07, 2018 at 05:06Martin, I finally got my motor working thanks to this thread and ideas from you. I am using an 'el cheapo' Chinese motor because I don't want to fry my expensive motors. The ultimate application is for 3 to 5kW 42V specialized UAV motor using FOC for a research project. The test motor is an A2212 14 pole 1000KV. It is made for trapezoidal control but nothing is perfect and should work with FOC (which it does).
On the IHM08M1 I have JP1 and JP2 ON, JP9 OFF, JP3 ON, J5 and J6 in 3SH position.
The control board is set to Nucleo F446RE. I create the project using WB and build under Eclipse.
I had to change a lot of things. The default startup parameters do not work. The FAQs offer solutions to this, once you have found them. The default DIO ports are incorrect and need to be changed to suit the board. Other small changes need to be made too.
The motor runs, but not as well as I expected. Startup still takes longer than using 6-step. Minimum speed is higher than 6-step at which it stops and raises a speed error. There seems to be more vibration also which is speed dependent, ie. there are speed ranges where vibration is felt. I replace the cheap bearings with precision bearings which did not help and my next task is to balance the rotor. At slow speed the speed tracking is unreliable and there is much hunting. Playing with pid terms changes this but I cannot optimize it. Auto-restart seem to do nothing.
Ultimately I would like to see the same performance as commercial 6-step controllers for UAV use, ie. start/stop on demand, braking options, auto restart after a forced stop, etc. in addition to the smooth operation and rapid speed response promised by FOC control.
The Motor Profiler will not work with this combination at all - pity.
This was not the smooth introduction to MCSDK that I was expecting. If I had not invested much time and effort into studying 5000 pages of F4 documentation I may have quit and moved to another platform. It took a long time just to get the motor to run at all.
For sure I can write the code to do the other things I want to evaluate, but first I have to learn the inner workings of the SDK, get the simple app working, then find the time, and this is not my research project, just part of it. Test apps that do more than just run the motor and stop as soon as they are unhappy don't cut it. Developers need test apps that cover more complex functionality, that they can build on to properly evaluate the product for their purpose. We really do not have the time to become expert ST motor control engineers merely to ascertain whether the product is our proper solution.
I will keep working with this and post anything I find out here.
