Now, I know what you think, I also agree that most people who say there's something wrong with their lenses (like the 28-70/2.8 myth) actually are poor testers. I said MOST, to all you good testers about to complain.
But I had AF trouble with both the 20D and 50/1.4. I know you have a 5D, have you tried that lens on a 20D? The lens reportedly works flawlessly on a 5D and up (that from someone who had AF trouble on the 20D as well).
I'm not saying that the 50/1.4 has good and bad batches ('cause I don't know for sure) and also not saying that it is a bad lens. My hypothesis is that the lens' rather lower quality micro USM reacts unexpectedly with the 20D's AF, or the 20D's AF behaves unexpectedly in some specific conditions, that the 50/1.4 just happens to create more often.
For instance, these pics I took with the 135/2 + 1.4xTC:
http://forums.dpreview.com/forums/read.asp?forum=1019&message=17639986
Notice that I wasn't testing for AF when I took the pics. But I just noticed two of them were very OOF (the first and the last, which are similar but different, and taken at different times with other shots in-between).
There's also people who say they don't have this trouble, but then again there's different circumstances involved. It wasn't always that I got OOF pics with the 50/1.4, but many times in what should be rather easy conditions of AF'ing.
So my opinion on the subject is that while I really don't think there are as many "bad copies" as so many people may lead others to believe, there should be at least a little bit, since nothing is perfect.
My 50mm f/1.4 and 85mm f/1.8 work perfectly, and have since day
one. As does my 16-35L. Since the 16-35L was the lens I chose over
the 17-40L last summer, I see no reason to buy a 17-40L now. Since
I prefer manual focus, an f/4 maximum aperture is not my prefered
lens.
But thanks for your kind offer.
--
Peter White