You're right Daniel. We probably ought to go over to the new thread that you started. I will play around with the lens when I get a chance later today, or tomorrow.

I'll post on the new thread you started.

Thanks a lot Mark!

I really do appreciate you trying this out for me ... uh us.

Till then,

DK

The way these servos most likely work is that the processor, in auto focus mode, sends pulses to the actuators commanding moves with a second signal telling the actuator which direction to move in. The manual rings mimic the processor in that they too send pulses to the actuators with a direction signal. Thus as long as you never send enough pulses to get the actuator at either end of its range the ring settings should be repeatable. If, however, you turn the ring to the point where the actuator is at he limit and then turn it half a rotation past that the actuator will ignore the command pulses but when you come back will respond immediately to the pulses to go in the other direction. The ring markings would be off by 180° in such a case. It's probably not a big deal to change the firmware to use the pulse counts to keep track of where the ring actually is to prevent sending pulses to the actuator until the ring is where it is supposed to be. This could simulate the action of a mechanical control.

The problem is that there seems to be processor which responds to the frequency of the pulses, and changes the degree of change in focus according to this frequency. The result is that turning the focus ring quickly (higher pulse frequency) causes a larger change in focus across a given pull, than a slow turn (lower pulse freq.) does.

In order to hit marks consistently, one would have to pull focus at the same speed every time. This is the bigger problem with the lens, than the 180 deg. problem.

At least that is how I understand it.

It's quite common in control systems to have a rate sensitive design like the one that might or might not be on the XL2. (I believe XL2 is rate sensitive but others seem to disagree so I'll try to keep an open mind for now.)

The advantage of this system is that it lets the operator make large movements quickly while also allowing for fine adjustment. It also means that the rotational position sensor does not have to be an expensive one because the operator is always in the loop to provide calibration on the fly.

Alternatively, if the system were designed to interpret the input pulses simply as position, there would either be lack of precision in the focussing, or it would take many turns of the focus ring to go from close to far focus settings. Using a more expensive encoder would help, but some mechanical gearing might still be required for fine control.

Anyway, a big thanks to Mark for carrying out these tests and reporting the results on the forum. It is very much appreciated.

Richard

Here are the results of my research on follow focus with the 20x auto lens.

I made a new post, "Canon 20x auto lens follow focus accuracy"
http://www.dvinfo.net/conf/showthrea...threadid=38166

Read your test findings Mark, good job.

Sounds to me like breathing or not, that 16x manual is looking more attractive now...?

Thanks Charles and you are right. After using both lenses I prefer the manual. The 20x has a longer reach, but I'm pretty much sticking with the 16x manual.

I'm a 14x manual guy myself. I like having the iris ring where it should be, and I rarely seem to need more on the long end.

Ya! What the heck was I thinkin', eh?