I can forsee a significant problem with most of the tests that I'm
seeing. Most of the tests are shot with the focus point on
something severely canted. Since the AF sensor is not a point (or
even a line) but rather a plane there is a continuum of valid
positions for the lens to be in. Furthermore the AF sensor is not
necessarily completely enclosed by the rectangle overlayed on the
image. Can I suggest a slightly modified test. Have a high
contrast planar object parallel to the focal plane that you focus
on and have your ruler intersect it. The following were taken with
my d30 and are 1:1 crops. The ruler intersects the card at 16.5
inches. In the first I am focusing on the card about a half inch
to the left of the ruler and it appears that the in focus plane is
somewhere inbetween the 16.5 line and the 16.375 line.
on this next picture, I focused on the ruler with the focus point
centered on the 16.5 line. After hunting and failing to find focus
a couple of times it finally gave a focus lock for this. In this
case the 17 line was outside of the AF box etched on the focus
screen.
I repeated these tests a few times and got similar results. When
the AF sensor was over the card it would always be about a 1/16th
of an inch front focused. When the AF sensor was pointed at the
ruler I never got consistent AF although it would sometimes snap
into position as though it new exactly where to go.
All of the above was done in my basement with a 50mm f/1.4 on my
d30 at about 20 inches from the target. The camera was set at
1/160 sec f/1.4 ISO 100 (from which you can calculate an
approximate light intensity if you know how (which I don't off the
top of my head)) in one shot AF mode and the center focus point set.
As I'm going away for the weekend I won't be able to respond to
comments or criticisms so please feel free to trash my name
jim