Technical question: why do lenses backfocus/frontfocus?

So we all know that sometimes a lens sample will consistenly backfocus or frontfocus while our other lenses do just fine. The solution is usually to send the lens in for calibration or to replace it.

The question I have, however, is why is it even possible for a lens to backfocus/frontfocus when there's nothing wrong with the camera?

If the position of the mirror and the autofocus sensor(s) is such that the focal distance between the back end of the lens and the autofocus sensor is the same as the distance between the back end of the lens and the main imaging sensor, and the algorithms for handling autofocus (which involve telling the lens to focus in one direction or the other and for how long...basically, directly controlling the focus servo in the lens) are within the camera itself, then how can the lens itself possibly have any effect on whether or not the image is properly focused?

It seems to me, based on those very basic (but sensible from an engineering standpoint) assumptions, that any backfocus/frontfocus issue should be an issue strictly with the camera. And yet, we know that's not the case. So what gives?

I have exactly the same opinion, and I was about to start a thread about this.

Maybe someone can enlighten us: how is phisically possible that a lens can be focused properly on the viewfinder/af sensor but out of focus on the sensor? (And be a lens problem).

My only explanation is that is only a camera problem. Lenses with big aperture magnify the problem. It can go unnoticed with kit lenses, and only exposed when buying a new lense, thus blaming the lense.

This is the way I see it in my simple ways:

The camera issues a focal position setpoint signal to the lens.

The lens control circuit drives the AF motor to the setpoint position....where it stops....no feedback, no trimming action.

There is always some "slop" and fabrication tolerance spread in the lens AF drive/control circuit. Therefore, even for a perfectly accurate AF setpoint by the camera, different lenses are driven to slightly different AF positions. Large errors or AF drifts are manifested as lens back/front focusing.

Lens AF circuit calibration adjustments allow that error to be reduced.

I thought the camera basically tries to find the greatest amount of contrast at a given focus point (AF sensor) to focus, so there shouldn't be a 'set point', unless I'm misunderstanding what you mean by set point. It's possible that there is some discrepancy between the camera's computer and the lens's chip.

This is caused by the fact that focusing is not done in a constant loopback, and is controlled by the camera, not by the lens. Lens and camera are independently calibrated at the factory, within certain tolerance limits, namely within 1/3 of a stop of the DoF for lenses with an aperture larger than F/2.8, and to within 1 stop of DoF for lenses with a smaller aperture.

So, the AF unit measures some contrast differences, and calculates how much the lens needs to be moved, and then sends a signal to the AF motor in the lens, to do so. If they are both out the same way, you may well end up with focusing to the back or front of the object you want to focus on.

You can overcome this, partly at least, by trying to focus 1 time, but once only. It may acquire correct focus in that case. With some lenses, and under some conditions, the camera does this for you: macro lenses come to mind, used for non-macro purposes, and focusing under low contrast conditions.

The reason why you should only refocus once, is because it will very likely overshoot the second time you try to refocus, so ending up with back- or frontfocus again.

Adjustment of front- or backfocus on a lens is a fairly simple thing to do, for Canon anyway, as the printed circuit board in the lens has several discrete, soldered steps for this purpose. So, when calibrating, a new connection is soldered instead of the old one.

Note that because zoom lenses not necessarily stay in focus over the zoom range, it may be very difficult to calibrate to precise focus over the zoom range, as my own experiments with e.g. the 70-200 and 100-400 proved. Although always focusing at the same point at specific focal lengths, with the same object from the same distance and the same lighting conditions, this point was slightly different between focal lengths.

The camera can also be calibrated, by means of adjusting a screw inside the mirror housing with a small Allen key. This is more precise, because it is continuously variable, unlike lens calibration.

However, this doesn't solve the problem of focus change with zooming. It still means refocusing after zooming, and even then it may not be 100% the same as before, due to the in-lens calibration difficulties accompanying this.

HTH, kind regards, Wim

A lot of logic in PetKal's words, probably does not apply with AF Servo which seems to zero in on the ideal focus as long as the button is depressed (an experiment with double stacked TCs showed this behaviour).

I do know there is a AF zero control in the camera mirror box and have seen instructions on how to DIY trim this; not something I would recommend myself and I am used to microelectronics.

Of course this has been put under user control on a lens by lens basis on the 1D III, interesting that Canon have done that - presumably they are happy to trade more helpline loading from users that have screwed up their camera in the hope of reducing returns and warranty budget for AF issue.

Also, and this is really anecdotal, there is some kind of lens adjustment for AF precision. It is not clear if this is some sort of electronic twiddle control or mechanical shimming or both. The descriptions are not clear.

The problem is the AF is an open loop system (Beyond me why that is): There is no feedback: The camera takes a look, determines focus is of by amount 'X', so gives the lens the command to move 'X'. *It doesn't look again*.
Good explanation inn post #3here

bacchanal wrote in post #3302350
I thought the camera basically tries to find the greatest amount of contrast at a given focus point (AF sensor) to focus, so there shouldn't be a 'set point', unless I'm misunderstanding what you mean by set point. It's possible that there is some discrepancy between the camera's computer and the lens's chip.

(1)Someone please correct me if I am wrong, but I understand Canon SLRs use phase detection as opposed to contrast measurement in their camera's AF subsystem.
(2)In effect that is a "setpoint", or generally a "control/governing signal" issued by one controller (camera AF) to another/subordinate controller (lens AF drive).

From the link I posted: "the activated AF sensor "looks" at the image projected by the lens from two different directions (each line of pixels in the array looks from the opposite direction of the other) and identifies the phase difference of the light from each direction."
So 1) yes. 2) Yes. (If I understand what you mean correctly)

Okay, so I'm feeling a bit confused and enlightened at the same time. What exactly does 'phase difference' mean pertaining to an AF system?

OK let's see it other way. I had front focus problems even using manual focus. That can't be a lens problems, can it?

erpetao wrote in post #3302524
OK let's see it other way. I had front focus problems even using manual focus. That can't be a lens problems, can it?

No. But you've moved the goalposts. The original question was:

kcbrown wrote in post #3301858
The question I have, however, is why is it even possible for a lens to backfocus/frontfocus when there's nothing wrong with the camera?

bacchanal wrote in post #3302519
Okay, so I'm feeling a bit confused and enlightened at the same time. What exactly does 'phase difference' mean pertaining to an AF system?

To answer myself:

Wikipedia wrote:
Phase detection is achieved by dividing the incoming light into pairs of images and comparing them. SIR TTL passive phase detection (secondary image registration, through the lens) is often utilized in film and digital SLR cameras. The system utilizes a beam splitter (implemented as a small semi-transparent area of the main reflex mirror, coupled with a small secondary mirror) to direct light to an AF sensor at the bottom of the camera. Two optical prisms capture the light rays coming from the opposite sides of the lens and divert it to the AF sensor, creating a simple range finder with a base identical to the lens' diameter. The two images are then analysed for similar light intensity patterns (peaks and valleys) and the phase difference is calculated in order to find if the object is in front focus or back focus position. This instantly gives the exact direction of focusing and amount of focus ring's movement. Although AF sensor is typically a one-dimensional photosensitive strip (only a few pixels high and a few dozen wide), some modern cameras (Canon EOS-1D, Nikon D2X) feature Area SIR sensors that are rectangular so as to provide two-dimensional intensity patterns in 2D. Cross-type (CT) focus points have a pair of sensors oriented at 90° to one another, although one sensor typically requires a larger aperture to operate than the other.

PetKal wrote in post #3302431
(1)Someone please correct me if I am wrong, but I understand Canon SLRs use phase detection as opposed to contrast measurement in their camera's AF subsystem.
(2)In effect that is a "setpoint", or generally a "control/governing signal" issued by one controller (camera AF) to another/subordinate controller (lens AF drive).

Yes correct, that is why it works in less than the usual P&S focus time of an ice age. I guess it is faster because phase detection is coherent and so more noise proof requiring less averaging to get a usable result.

erpetao wrote in post #3302524
OK let's see it other way. I had front focus problems even using manual focus. That can't be a lens problems, can it?

That is a different issue. Assuming the camera mount, mirror and focus screen are all in good alignment it is difficult to see how that can happen, it seems more likely the viewfinder and sensor optical paths are not well matched.

Can you explain what tests you have done to come to your diagnosis?