Now, there is obviously a few problems with this approach. First of all, how does one get the whole thing started? Remember, a motor does not generate any back-EMF when it is just sitting still. So just about any sensorless algorithm is going to have to have some kind of starting algorithm. Just like a car when one steps inside, turns the key, gets the engine turning over and over until finally it kicks in, it is something like that. One has to do the same thing with the brushless DC motor and just start supplying some sequences to the motor, like with the stepper motor. Yeah, the designer said never drive it that way but in this case one has to do something like that to get the motor started. It runs rough, the rotor is just going to follow the commutation sequence, but eventually it will catch on and one will start getting some back-EMF zero crossings. But there is another problem with this and that is the whole premise of this is based upon the assumption that it is running at constant velocity. What happens if it is not running at constant velocity? What if it is accelerating or decelerating? Since the designer is always trying to predict a future event based upon past information then one is going to always miss the true commutation boundary if the velocity is changing. Now, implement some observers. There is a whole section or a whole different study on observer technology and one can actually track the change in the commutation zero crossings, monitor that, and then actually compensate for the acceleration. As a reminder, it takes a while for it to learn the fact that the speed has changed. So, even a good observer is still trying to predict a future event based upon information which has occurred in the past. Finally, as the motor is going slower and slower, what one will see is the back-EMF amplitude getting smaller and smaller. So this technique which again is based upon measuring something in the back-EMF signal typically does not work well at very low speeds. So for these reasons and probably a few more, this does not represent a good way to do sensorless control of brushless DC motors.