Auto focus Micro adjust allows the user to more precisely control the point of autofocus. Autofocus systems in modern DSLRs have focus accuracy errors related to camera body and lens tolerance issues. This may cause a given lens body combination to focus slightly ahead or behind the intended target. Canon and Nikon have introduced a way for the AF of each lens to be calibrated separately and the calibration data to be stored in the camera body which can recognize a particular lens and or lens/teleconverter combination. In the past it was necessary to send a camera and lens to the camera manufacturer service people to tune the lens camera combination and this did not allow for different lens to be calibrated separately. 

When I received the Canon 50D and was satisfied it was operating correctly I set about the task of calibrating all my lens.  Toward this end I did a quick setup on all my lens and recorded the AF microadjustment offset numbers that I determined from the calibration for each lens. After one week of usage I then went back and retested and compared the numbers from the first run to be sure that the system was stable and repeatable. As the offsets were all identical I was reasonably convinced that the AF system was stable and I was getting repeatable results.

The question of what distance to calibrate the focus at is a good one. Logic dictates that calibrating at the distance you shoot at most often makes sense. Indoor distances limit this to 20-40 feet or so usually and limited Depth of Field suggests close distances are more critical.  I performed calibration at 20 feet with my 500mm and x1.4 and I also checked it at 80 feet. When I compared the cal values they were identical!

After several weeks when the weather got colder I decided to test the calibration values again as cold can effect on opto electronic system and as I often shoot outdoors in the winter I decided to check the cal values again and was glad to see that the calibration values of my 500 and 300mm lens with and without a teleconverter had not changed with a temperature of  -2 degrees C.
 
 
 

My method 
1 - Tape a a flat  hi contrast paper target to a wall at a distance of 15-30 feet from the camera. Insure that the target is square to the camera.   Printable AF Target
2 - Set up the camera on a tripod, turn IS off, use automatic flash compensated EV +1 (Canon ETTL mode) with a shutter speed of 1/250 second or so. Use ISO 100 to prevent ambient light from contributing to the exposure which may blur the image do to camera shake. Use manual exposure mode with the  lens set to it's widest aperture.  eg. f2.8 for a 300m f2.8 lens 
3- Select the center AF point and use one shot focus mode. 
4- Set the focus ring to infinity, activate autofocus  and shoot the exact center of the target. Be sure focus was achieved.
5- Go into the AF microadjust menu  C. Fn III and note the value (default is 0). Offset the AF value +5 and hit "set" to enter it,  and shoot the target again.
6- Zoom the image to maximum on the camera LCD display and compare the two images for sharpness. If  the +5 setting was sharper, offset the value to +10 and shoot another image and compare it to the +5 image.
7- Repeat step 4-6 with differing AF values until you determine which value gives the sharpness image. It may be necessary to do this several times until you feel confident the value is the correct one.
8- Repeat the above steps with the x1.4 Tele extender in place as the camera retains a separate value for tele extenders and lens combinations.
 
 

Tips:
1 - Most newer lens should be within +- 10 points or better, any more than this and there may be a problem with the lens or camera.
2 - Repeat the test occasionally to be sure that nothing has changed. If the values are constantly changing then there may an equipment or technique problem!
3 - Record  the value for each lens for future reference. 
4 - Adjustments of +-2 or 3 are seldom noticeable but it depends on the lens.
5 - For zoom lens use the longest focal length or alternately the focal length you use most often