Hi everyone,

*** I've just received my gear back from Canon Service, see the post starting on page 3 for the details ****

I've been around the forums long enough to hear all sorts of stories of woe about focus problems. I've had a 70-200 f/4 IS for nearly a year now and have, in the back of my mind suspected it has some kind of problems. However, I've always assumed it's more likely that it's just me, and not the lens (or my 20D). I went out shooting birds the other day. I had the 70-200, the 20d, I was shooting handheld and using the center focus point only, without doing any focus-recompose (I know, who shoots birds hand-held with a 200mm lens.. but bear with me..). In 300 shots had almost no keepers, and a huge percentage of really out of focus shots. Usually, I get better results. I decided to do some tests today and call canon support. I learned about the focus point tool in canon's Zoom Browser software, and tried some tests there. Here are the samples:

1) An example of a bad bird photo. I was only about 30 feet away from this guy, and I just don't think I could miss by that much. 100% crop, no other changes. Note the nice focus on the little ring in the water in the foreground.

I have only had a quick look at the images but it looks like the camera is front focussing. Send the lenses and body into Canon for calibration. Usually if at least one lens is within warranty then the calibration will be free.

yep, front focus. Have it calibrated, it'll be great then.

Quick question, are the focus problems with this lens only?

yes, if you try another lens you shoudl see better focusing.

Agreed: forward focus problem. You can see how much sharper the bubbles are in the first photo. If this is the only lens you have which exhibits this problem then just send the lens away with a copy of these shots so the techs can work it out.

Andrew, a test I do to determine the focus point is to photograph your keyboard looking down at about a 30 degree angle. Focus on a particular key (I focus on the little button cursor thingy on my Thinkpad laptop). You can quickly tell if it's front or back focusing. Although yours looks like it's front focusing. BTW, my 70-200 f/4 IS is scary sharp!

Yes, it is bad. Are your other lenses fine or do they present some kind of focusing issue too ?
I believe this will most likely end up with you sending the camera and 70-200 to Canon, but you may also have to send in the other lenses, so that they do not get your camera to misfocus them afterwards.

Hi all,

Thanks for the replies. I tried the keyboard test that Richard Nye mentioned, and I wasn't able to reproduce the problem. I think that the problem is more likely to occur if I'm standing 20+ feet away from the subject (at least this is what I found when I noticed these troubles in the real world). I tried standing both around 6 feet away and around 15 feet away from the keyboard and every shot I took was acceptably focused. I didn't spend too much time on this test however.

A few others suggested that I try another lens. I duplicated my original test with my 28-135 and I wasn't able to reproduce the problem, at least not exactly. All of the shots (about 30) I took with the 28-135 seemed to be quite good, except for one which was way off. Here's the screen cap from ZoomBrowser:

4) Strange behavior with the 28-135, which otherwise seemed fine. To me, it looks like the camera might have picked a different focus point(s) then the center point highlighted in red. (I was in center only mode, as before)

I was worried about that when seeing your previous tests, but now it seems even more likely: and what if the AF system was not properly aligned in the camera ? I am not sure whetehr it is plausible, but I think it would explain the results.

Note that the focus points are far larger than the boxes shown in the viewfinder. They probably overlap with the boxes in your posted shots. The AF points tend to usually pick up the object of greatest contrast. Use a proper chart, etc. to test it.

I agree with simon_says.
Trying to hit auto focus on the post between the boxes is not a good test. The AF system is smart but not that smart.

Second, If you have the time send it all to Canon. I sent a 20D in and just because I didn't want any questions after it came back I sent my glass with it.
It all came back spot on. Then the 40D came out and a lucky member here got a great deal on a freshly calibrated 20D.

The AF systems are good and getting better. But not yet 100%.

Focus points are not confined to the little box, they do extend beyond. I would put the camera into AI Servo, then place the focus point on the box and hold the shutter half way, then move the focus point left or right and see when the focus falls off the box and into the background subject. In some of your pictures it could still be touching one of the two boxes. On my 10D it would definitely touch, however the 20D has a better focus system and may, or may not, extend as far beyond the squares. That is the center point on the 20D and could be a cross type sensor that may not be as effective with an f4 lens (usually f2.8 or faster). One of the left or right points may be more accurate on a vertical subject being that those focus points are vertical bars rather than horizontal like the upper points. Once you are sure where the fall-off point is then position your subjects with enough distance apart to test.

To be sure about the focus, there is a fold up test target image out on the net that you can download and print. It folds up so that the target is a flat plane with enough distance between it and the included distance scale that it will not allow the focus point to accidentally lock on the scale. I'll try a search and post a link when found. This will let you know the distance of focus ahead of and behind the target your lens has. There must be some sort of chart to check against for a given aperture, but generally 1/3 the distance in front of the subject; 2/3 behind.

Much of this info may already be clear to you, but I'm repeating it here in case other readers are not familiar with it.

It is quite possible to have a 70-200 that focuses reasonably well close up, but the further away you go the more the focus misses... My non-IS f4 did that, (although in my case it was backfocus at long distances) and because bench top tests close up looked fine, it was a beast to diagnose. It's hard to set up decent targets 50-100 feet away, and test unambiguously, because of the issue of the rather larger than you think focus sensors covering more objects at longer distances.
In the end I got fed up with the uncertainty and sent it in to Canon for a checkover just before the warranty expired and they fixed it up. If you aren't sure, my advice would be to send it in before the warranty runs out and get them to check it over.

A few posters mentioned that the AF point box and the AF sensor point are not equal. I did have that in mind when I was testing with my other lens (the 28-135). I moved in a little closer to the target so that the frame was filled the same at 135 as it was at 200 and with the same tests I found that I could focus on the fence just fine. I also purposely picked areas of contrast (both vertical and horizontal) to help lock on. So I think I created a reasonable test scenario.

In any case, it's now up to the folks at Canon, as the camera and the 70-200 are on their way to Irvine. I'll follow up with whatever comes of it.

Thanks for the suggestions and help,
Andrew

It looks as if the camera's AF point is over the fence, yet the boxes are in focus. A fun test would be to move the boxes out of the way and see if the camera still frontfocuses the same amount. Then move the boxes forward or backward and see if the distance the camera frontfocuses changes.

Since the boxes are in focus, it's likely the camera focused on them inadvertently instead of the fence.

Either way, Canon should take care of it for you over in Irvine.

I was happy to see my 20d and 70-200 make it back to me on Friday. My experience with Canon service (this was my first time) was quite good and in the future I would not wait as long as I did before getting things checked out.

It turns out that the Canon service folks found issues with both my lens and 20d:

For the lens they identified the symptom as "rear focus" but stated that "the exact cause could not be identified" and that "electrical adjustments were carried out".

For the 20d they found that "the adjustment of the AF assembly was incorrect" and again "electrical adjustments were carried out".

My initial impressions after the fixes are really good. I find that the focus is much more consistent, even in low light. I'll try my target tests again and see what happens (it's been a rainy weekend) though I expect quite an improvement given what I've seen so far.

I tried to look carefully at how much of the actual focus point might go outside the center focus box, and my best guess so far is that the focus point extends out past the center box by roughly half the size of the box. Does that fit with what other folks believe on the 20d (with an f/4 max aperture lens)?

Thanks,
Andrew