Originally posted by BarneyCG Wouldn't that be due to the way that the DOF tends to extend further behind the desired focus point than it does in front of it so when focusing from infinity there is a little more leway ? In which case you'll see exactly the same problems no matter which manufacturer you pick.
In an ideal world, the autofocus would focus perfectly on your subject. But in the real world, inaccuracy will cause the autofocus to stop just in front or just behind your chosen focus point. In a well designed lens/camera system this inaccuracy should certainly be less than the depth of field (at maximum aperture).
I'm not sure that I fully understand your point, but I agree that the non-symmetric nature of depth of field is important. In my opinion, a front focusing system is preferential to a back focusing one because in a front focusing situation the larger depth of field behind will more likely bring your subject into the zone of acceptable sharpness.
In the case of my 21mm, the depth of field at maximum aperture appears to be shallower than the K10D autofocus accuracy. I also wonder if that is why they made this lens a f3.2 rather than f2.8, 2.4 etc. They perhaps knew that their autofocus accuracy was right at the limit at this fstop and short focal length and that we might not notice !! I can't say that I've seen a problem on my 14mm/f2.8, but this is not the sort of lens I use for close shots at maximum aperture.
I still don't understand why Benjamin is having trouble at f9, f14. There should be bags of depth of field at these apertures. Look at the sums (e.g 1m subject at f9 results in a depth of field of 0.95m (0.71m to 1.66m). Also, if the scene is out of focus at these apertures then surely you would notice it in the viewfinder before stopping down?
Andy