When you do AF calibration tests, you need to be somewhat scientific about your approach to make sure any missed focus truly is caused by the AF system ... and nothing else. This means you can't just use your typical photography shots to diagnose the camea. If you're seeing issues with your shots being soft, there are too many things that can cause this that are something other than an error in the auto-focus system. The whole point of isolating all the variables by doing a controlled test is to make sure that the AF problem is actually the camera's focus calibration and nothing else is contributing to the problem. That means using a test-chart in a controlled test is the best way to identify (and hopefully correct) the problem.
Here's my test chart. The tool has a flat high-contrast card that serves as the focus target. In single AF point focus mode, I put the focus point on that area of the chart. But to the right is a rule that is leaning on a 45º slope (it's built into the tool). The idea is that the camera locks focus on the high-contrast target and you take a shot. But then you inspect the ruler on the right to see if the "0" point is technically the sharpest and you look up & down the scale to see if it's actually sharper elsewhere on the scale. The parts below the mid-point are closer and the parts above are farther. This tells you if the camera is front-focusing or back-focusing.
The target should be parallel to the camera's image sensor (don't allow the target to be at an angle or you'll get bad results).
Also, lenses do not necessarily have "flat" fields. Meaning the point of perfect focus at the center may NOT necessarily be the perfect focus point at the corners. Lens makers try to produce lenses with fairly flat fields -- but they aren't perfect. So make sure the target is centered.
The camera should be on a tripod. The focus target can also be mounted on a tripod but you could just place it on any flat surface. The key thing is that you need to be sure that the distance between camera and target cannot possibly change after the camera has locked focus and if you hand-hold the camera then you can't be sure (you could have moved and now you're making camera adjustments for something that wasn't caused by the camera's AF system and that could make things even worse.)
Also... you want LOTS of samples before you draw any conclusions. It's always possible for a camera than generally nails the focus to just occasionally miss. It's also possible for a camera that might be consistently missing focus to just occasionally get lucky. So you want to make sure you shoot a LOT of sample shots before forming any conclusions.
And finally... the focus system is mechanical. There are gears. These gears can develop slop (gear backlash). For example... suppose your subject is 10' away but your lens if focused for 5' away. The camera lens has to focus "out" to lock focus. But now suppose the lens was focused for 15' away... now the lens has to move the focus "in" to lock focus. Due to gear backlash (slop) it's possible that the camera might nail the focus in one direction... but miss in the other direction. What you WANT is consistent focus regardless of whether it has to focus "out" vs. focus "in" to achieve focus.
I set up the camera & target for the test, make sure the target is centered and parallel to the image sensor on the camera, etc.
Then attach the lens I plan to test and deliberately de-focus the lens randomly to some position which is too close and let the camera focus and take a shot. Then de-focus it too close again... and let the camera focus and take another shot. Repeat this exercise about 10 times so now you have a good 10 samples or so... and for ALL of them the camera had to focus "out" to achieve correct focus.
Now repeat... but this time run the focus too far out so that the camera has to focus "in" to achieve focus. Again, repeat this test another 10 times or so.
(Make sure you note which images had to be focused "out" vs. "in").
Now import all the images to your computer and evaluate the focus.
What you want to see is consistent focus regardless of whether focus had to move in or out. If the focus is different depending on the direction of focus then it means your lens likely has excessive gear backlash and needs to be serviced. This is NOT the sort of focus issue that the in-camera auto-focus micr-adjustment can correct. AFMA can only correct missed focus if the missed focus is consistent.
Next ... assuming the focus is consistent (it either generally always nails the focus... or if it misses focus it should at least always miss by roughly the same amount and in the same direction.) If you see this happening then you can apply some AFMA correction using the in-camera menu settings (and don't forget to re-test.)
When you evaluate all your images, you might just occasionally find one that missed focus. That's ok if it's one. As long as the overwhelming majority of the images are doing the right thing then you're ok. If, however, focus is all over the place... then you have other issues -- and there's a strong possibility that it's the lens and not the camera.
