How can lenses show different FF/BF on the same body?

[Class A]

EDIT: Sorry, I just noticed that I posted this question to this forum already (-> Why do *lenses* front/back focus?). This one here is slightly different, though.

I tried a couple of forums and still don't know the answer.

How can different lenses have different front focus (FF) or back focus (BF) behaviour on one body?

If a body has a calibration problem (say the AF sensor plane isn't quite in the exact position) then lenses won't focus properly but they all should show the same amount of FF or BF.

Many users report lenses to have different behaviour, say lens A to show FF and lens B to show BF behaviour. How is that possible? The effect cannot be explained by assuming that users complain without reason. Some camera models allow to dial in focus compensation values for different values. I don't think camera manufacturers would put in such a feature just to please users, even if the individual compensations weren't necessary.

Further to my old posting (see above) one probably has to distinguish between smart lenses with a chip (sometimes "rechipping" is said to cure the problem) and others which are purely controlled by the camera, say with a screw. In the latter case, my understanding of AF is that the camera uses a closed feedback loop to achieve maximum sharpness. The lens should then not matter at all.

I can think of two explanations:

1. The chromatic aberration of a lens throws off the prism-based phase comparison.

2. There is no 100% feedback but that the camera always applies a small offset step after having found the correct point. This would be necessary if the optical path lengths between mount and AF unit and mount and sensor respectively where slightly different in each camera and instead of doing an optimal physical alignment, some software correction takes place. If lenses differ in how they actuate this final offset step then they would yield different FF/BF phenomena.

I'm really stumped by the fact that lenses show different FF/BF on a single body. Apparently even copies of the same lens model can differ in their FF/BF behaviour, i.e., replacing a FF lens copy with another can cure the problem.

Are chromatic aberrations of lenses really large enough to support explanation 1?

Could a proper hardware calibration (instead of software compensation) of bodies avoid the problem altogether?

[Colin Southern]

It's an interesting question, and one that I still don't have an answer for.

One thought I did have is that perhaps the AF isn't a fully closed loop - perhaps for the sake of speed the AF mechanism works out how far off the focus is - instructs the lens to move by exactly the amount required to correct this - and leaves it at that. If the amount wasn't quite right, then a small BF / FF might result? But as I say, I don't know - just speculation.

You might like to post the question to Canon's Chuck Westfall for inclusion in his tech tips column?

Cheers,

Colin - pbase.com/cjsouthern

[Class A]

One thought I did have is that perhaps the AF isn't a fully closed loop - perhaps for the sake of speed the AF mechanism works out how far off the focus is - instructs the lens to move by exactly the amount required to correct this - and leaves it at that.

This doesn't seem to be the case with Pentax cameras. Sometimes the lens is moved to a certain position and sometimes (not always) it then moves back (and possibly forth) a very tiny amount. So there appears to be a final check after the lens has executed all turns. I believe users refer to this as "hunting" but it is not the "all the way forward and back" hunting that lenses do when they have no clue where the focus is and need to search the whole range. Some users believe that Pentax AF is slower but more precise than say Canon focus because of this "double check" behaviour, but I have no idea whether there is any truth to that or not. I have difficulty believing that any AF system will just trust the lens mechanics and not do a final check on arrival.

You might like to post the question to Canon's Chuck Westfall for inclusion in his tech tips column?

I have found his (small) tech tips column but no way to add to it?

[Colin Southern]

I'd be interested to know if people get this BF / FF happening in both 1-Shot and SERVO modes, or just one or the other. I would have thought that SERVO mode would definately be a closed loop (although it's not uncommon for the camera to hunt a little) (although I suspect that usually the focus change due to the hunting is probably covered in the depth of field anyway).

Not sure if you know much about Chuck Westfall, but he's a walking/talking encyclopedia of Canon technical knowledge, and a true gentleman - he's posted a link to eMail him at the bottom of http://dirckhalstead.org/issue0812/tech-tips.html. I've exchanged eMails with Chuck several times in the past - he's always replied quickly and comprehensively to my questions, but I do try to limit myself to one or two questions a year because I can only imagine the amount of eMail the poor chap must get from every man and his dog wanting answers. But if you can't find the answers anywhere else - then perhaps that's a good question to approach him with? (he would have access to pretty much anyone within Canon I would presume).

Cheers,

Colin