Arlo, it's not just the first cycle, the controller issues a 6-step sequence until the RPMs have risen enough so it can generate the proper sinewave current shape. The transfer is done progressively.

Bas, about my question of yesterday, on the way to calculate the commutation frequency, I think I was wrong. After some reflection while I was sleeping, I came to the conclusion that the frequency is much higher than what I initially thought, because of the "electronic gearing". But I haven't been able to figure out the formula yet.

I have this nice program that shows you the winding scheme for any combination of number of slots and poles, and it also gives the number of "cogging steps" per rev of the motor. I believe that the cogging steps are the same thing as the number of cycles necessary to make the motor rotate one revolution. (But is that number for one phase only, or the total for the 3 phases?)

So it might be that the correct formula for the commutation frequency is:

# cogging steps X motor RPM , divided by 60

A search on the web didn't produce any conclusive answer on this.