Mr Singh:
Averaging works, I agree, if there is relatively uniform distribution of the samples around the central tendancy of the data. So your comment was just fine.

The real point of my post was that the momemtum of a rotating mechanical contrivance means that the speed does not tend to change instantaneously. With good sensor placement this helps the averaging process and also the PLL approach. IE, the filter on the PLL need not be set orders of magnitide slower than the speed change characteristics of the rotating device.

Michael Karas