Lu,
I imagine that the torque pulses to the shaft are so rapid as to look like a steady torque, similar to an anti-lock brake system.
The avg value changes during a part of each cycle if a stick input is made, but even then, there is not a large spike being delivered because of the PWM type modulation.
In a hover, rather than 4 large pulses, 1 at each quadrant, you could apply several hundred shorter pulses averaged around the circle to achieve the same torque.
Last edited by Chiplight; 21st May 2005 at 20:11.