This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Low speed operation



I noticed in some recent posts from Chris Clearman that further work on IPD / HFI techniques has been deprioritized because TI have concluded that the vast majority of motors being used don't have sufficient saliency. I saw that one proposed solution is to automatically transition to sensored mode with trapezoidal commutation at low speed. Have you looked at the technique described in "Accuracy and robustness improvement in sensorless PMSM drives at low-speed by direct-axis current injection"? I think in the past you have advised applying a current to one of the coils to align the rotor before startup (and I think that running the Rs estimation basically does this for you). The method in the paper seems to be essentially the same idea except that the alignment current is (a) rotatated with the estimated speed of the rotor and (b) gradually decreased as the estimated speed increases.

  • I've sent this along to some of the other team members to see if they have feedback on the IEEE article.

    Rs Recal injects the D axis current to try to get an initial alignment.
    1. you need to have enough current injected to move any load, which can be difficult
    2. even if you have a good starting angle you still can't track the angle during the initial movement, so in practice there are many scenarios in which the dynamics of the load don't allow for the rotor to actually rotate (brake, backwards movement on a bike on a hill, over-load condition for a drill, etc.) so you lose synchronization.

    ForceAngle rotates the angle used in FOC to attempt to get the motor started, so it is very similar to what you describe. It then transitions to FAST once the FAST frequency estimate is > user.h setting.