shing wrote:
I ran the focusing test (downloaded from the Internet) with all my lens on the 10D. I found all the shots of my Sigma 70-200mm EX f2.8 were front focus (about 2.5cm). My other lens were no problem. Here are all lens used:
Have you stolen my 10D? 
Seriously, I'm having the exact same problem, my 10D front focuses by pretty much the same amount with the Sigma 70-200/2.8, but focus is spot-on with my other lenses (all Canon).
I emailed Sigma a week and a half ago and received this response:
From Sigma International Division:
1. The lens itself cannot change or adjust the Auto Focus point of the lens, since AF is controlled by AF line sensor and its algorithm in the camera. Although, AF accuracy of the modern SLR cameras has been improved very much, it is still not accurate as Manual Focus. That is a reason most of the professional photographers do not use AF of the camera.
2. AF accuracy will vary depend on the type of the lens. Generally, faster aperture, tele-zoom lens, such as 70-200mm F2.8 and wide angle zoom lens would be not accurate like standard lens.
To prove this analysis, please test Canon 70-200mm F2.8 or Canon 17-35mm F2.8. You will see the similar AF Shift phenomenon.
3. Since EOS-5 has 5 horizontal sensors, it does not affect by the test chart positioned in 45degree to the lens axis. On the other hand, EOS10D has 5 horizontal sensors, plus one sensor at the upper part of the finder and one sensor at the lower part of the finder, which cause the disturbance of the AF sensing with peripheral subjects. That is the reason your EOS5 has better test results compared to EOS 10D.
4. Different AF point is caused by focusing direction ( i.e. from infinity to the subject or near distance to the subject) is inevitable, but acceptable, because the difference of two focus points is due to the depth of the focus. Generally, AF of SLR cameras will not stop the center of the depth of focus, but stop the threshold of the depth of focus, especially with small low contrast subjects, such as the test charts.
As the conclusion, Sigma 70-200mm F2.8EX has no problem with AF Mechanism of the lens.
I haven't had the opportunity to test my 10D with any fast Canon lenses so I can't compare the AF performance with them. Anyone in SW Finland got one so we could do a quick test? Rajala (a local camera store) didn't have any when I was there.
According to item 3 in their response, I think they were assuming I was using all focus points and not just one. If I had been using all focus points, I would understand the end result might be off, but I was only using the center focus point. Oh well.
I think their item 4 is also due to some misunderstanding, as in my original message to Sigma I stated that I get the same front focus irregardless to whether I start AF from infinity or from the minimum focus distance.
BTW, I also inquired as to why my EOS 5 turns the focus much faster (and reliably) than the 10D, and the answer was:
All EOS cameras including EOS-5 have +/- 0.05mm depth of field. However, Canon has received a lot of complains regarding the "near focus and far focus" of AF with D30 as well as D60, because it is easy to check AF accuracy by PC but not easy by film.
Therefore, Canon had to change the depth of focus of EOS10D from +/-0.05 to +/- 0.035mm, which has more accurate AF than +/-0.05 but requires slower AF speed compared to EOS-5.
I guess I'll have to email Sigma again.
