How can AF adjustment be inconsistent across lenses?

Question

How is it that sometimes a lens will work fine on a camera body without adjustment yet others need some AF micro adjustment tweaking. I would imagine that the accuracy of the AF system with regards to front/back focus would be consistent across all lenses and be a body only issue. Yet it seems very common for different lenses to need different levels and directions of adjustment. What is it that could cause one lens to throw off the AF sensor but not another?

For example I have a good quality 100mm f/2.8 macro lens that I haven't noticed any focus issues with (though now I'm going to check again) and a cheap 70-210 f/4-5.6 zoom that needed micro adjustment. I would assume that a much sharper lens with shallower DoF would be more temperamental with AF, but that's not the case here it seems. How would the quality of the lens affect the AF module in the camera though? From what I understand of phase detect AF it should be the same regardless of the alignment and such of the lens, it should be just measuring the angle of light and be able to tell when something is in focus independently for the mechanics of the lens.

I think that this effect might be due to aberrations in the lens making the dynamics of the "circles of confusion" more complected than in a more high quality lens, resulting in poorer sharpness and offset AF.

Answer 1

The camera body is only half the equation - the lens is the other half! The optical elements located in the lens must be properly aligned so that there is the correct distance between them and the sensor in the camera body.

The only physical thing with the body that can require focus correction is a slightly different distance from the lens mounting flange to the sensor when compared to the distance from the lens mounting flange, via the secondary mirror, to the autofocus array. Any other calibration made using the camera's micro adjustment feature is done to allow the camera to correct for inaccuracies in the lens. So it stands to reason that different correction values will be needed for different lenses mounted on the same camera.

With a lens there are several issues that can affect auto focus accuracy and consistency. The accuracy of the auto focus motor to move exactly the amount it is instructed to move by the camera is usually the area that is most troublesome. It is also the area most difficult to correct for because the inaccuracy will vary depending on how far the lens is out of focus when the AF reading is taken.

Most Phase Detection AF systems take an initial reading, calculate how far the lens' focus element needs to move, sends that instruction to the lens, and then takes the picture. The camera doesn't take a second reading to confirm the lens moved the correct amount. How accurately the lens carries out the instruction from the camera regarding exactly how much movement to make is critical to the accuracy of the AF system. The more accurate and consistent the focus motor in the lens is, the more accurate the entire AF system can be calibrated! Newer lenses do provide more information to the camera regarding how far the lens actually moved and newer cameras can take advantage of this information. But even then, normally no second AF reading is made. When AF systems were first implemented the primary concern was focus speed, mainly due to the limited amount of processing power available imposed by the space and power consumption considerations.

Assuming the camera is adjusted so that the length of the optical path to the AF array and the sensor are the same, the camera's firmware may also be more accurate with one particular lens model than another.

Here's a blog post by lens guru Roger Cicala of lensrentals.com that discusses these things in depth with regard to Canon system cameras and lenses. But much of the general concept presented applies equally to other camera makers' AF systems and how they are designed.