Hi knro,

I can try to 'count' the pulses, or just 'control' the timing so in 1000ms we have 10 pulses with the lenses I have, but I'm afraid that this will heavily depend on the lens we're using. As far as I know, good lenses have very fast and silent focussing mechanisms (USM and other ultrasonic motors), with low amount of backlash and so, but cheaper lenses may have worse specs in this regard, so in the end the conversion between 'timing' and 'steps' won't be the same from one lens to another.

Backlash is another issue with these devices. For example, in my Canon 70/200 f4L I have like 1 step of backlash, so in my opinion it would be better to use steps instead of times.

Well, one is is to perform "relative" focusing, but we need some sort of numerical range for that which doesn't exist. I guess we can simply assume 1000ms = 10 pulses and just use that as basis. So putting 100 would only result in 1 pulse and so on.

We may try it, probably 100ms is enough to do 1 pulse in most of the lenses.

In fact, I think we can try it just without doing any experiment, just by coding something like:

initial_time = get_time()
move_step_focuser
sleep(100ms-(get_time()-initial_time))

If the move_step_focuser really takes less than 100ms, we can guarantee more or less that 1step = 100ms. Does it make sense to you Jasem?.

I didn't fully understand the previous code you wrote here. Probably the timer function didn't work as expected because these devices don't move the focusing ring continuously until you stop them, but they move in a pre-defined way.

The code assumes you can issue a focus command in some direction with some speed that runs for a specific period of time, then you "stop". But the DSLR moves in discrete steps and I guess speed is just "how far" they move when commanded each time. I guess we can incorporate your idea in the driver, but testing would be appreciated

I was reading this thread and was curious to know if this ever was implemented. I am thinking about hooking a second Canon DSLR up piggy backed on my OTA for widefield images and this focusing routine would come in handy for that if it were implemented.