Originally posted by stevebrot I wonder what tells it where to stop? Part of the efficiency of PDAF is provided by knowing current focus position with very high precision and using the calculated number of steps (similar to a numeric controller) to to projected the projected stop point. Canon does this so well (or so they say), they don't even recheck position against the sensor.
If it knows the position and speed, it can continue to run the motor for a specific period of time so it arrives at the desired position. This also requires the stopping being precise.
This still means it is "relatively" stepless, I would think, due to the precision of the calculations being done. But all of this is also primarily compared to the screw-drive focusing, I would think, where there minimum movement possible is significantly larger for that mechanism. I don't know the details of the older SDM workings, but presumably they don't rely on the above described position and speed data.