I'm repairing a Canon Ixus 870IS. I have replaced the lens compartment, and i experienced that when zoomed in, picture is out of focus. The camera is reporting that focus is OK. In wide angle zoom, the image is sharp, focus is OK.
This raises an interesting question that might be related to the manual focus troubles we have been seeing with some specific cameras.
I'm starting to think that cameras do not focus using any information calibrated for distance. Instead, they use a feedback loop to adjust the lens mechanism until they see a sharp image as measured by average rate of change of pixel to pixel intensity variation in the focus zone (I think). They then use the position of the lens focus mechanism and the zoom setting to calculate (or lookup) the subject distance for EXIF info purposes.
Cameras with MF capability reverses the above mentioned process when in MF mode, using the requested subject distance setting and current zoom position to calculate the needed Mecha setting.
Under the above assumptions, if something is misaligned in the lens mechanism, it may still be possible to focus but the reported subject distance will not be correct. Furthermore, if CHDK tries to set the subject distance on one of these cameras, the calculated mecha position needed will be wrong and the camera will not be focused where it was requested (in fact test have shown where it not focusing anywhere at all).
So for CoIT's issue, it may not be a calibration problem at all. Instead, the focus mechanism is somehow misaligned and not able to achieve focus lock at long subject distances. If so, then the mechanism must be fixed to solve this problem - any recalibration is not going to help.
All of the above is just my current working theory. I've been using a test script comparing mecha position to reported SD to play with validating this.