Following on from my unscientific test of my 17-40mm f4.0 L lens some months ago, I needed to do some similar work with my 70-200mm f4.0 L lens which I bought in San Francisco. The 17-40 has it's sweet spot at about f5.6, and I've been assuming that the longer lens is the same. Time for a proper look.

Same drill as before: stick a newspaper to the wall, put the camera on a tripod, and using mirror lockup to reduce blur as much as possible, take 3 shots at various apertures. Find the best one of each set of 3, and compare how the changes in aperture affect sharpness. Starting with the lens at 200mm then, I get these 6 100% crops:

Hmmm, interesting. Clearly this lens isn't the same as the 17-40mm. That one is sharpest with the aperture nearly wide open, but this one is clearly softer wide open than it is stopped down. f4.0 is really quite soft compared to the apparent sweet spot around f11. I took quite a lot of shots in this test and I haven't presented them all here. I did look closely at them though, I can tell you that there is hardly any difference between shots taken at f8 through to f16. There is, however, a notable softening between f16 and f18. That's clearly the point where diffraction cuts in, and apertures smaller than f18 soften quite markedly. Let's see if things differ at mid zoom.

I moved things around a bit and ended up with the camera closer, so the image isn't quite the same bit of newspaper, but the procedure was the same as before. What I see from this is that the sweet spot has moved to a slightly larger aperture. f7.1 looks about the best, with f11 being just a bit softer. f8 seems softer than both, which puzzles me. It was the best of 3 f8 shots, all of which were all but identical. Maybe my lens has an issue at f8, or maybe I just didn't take the correct shot. As I said, this isn't scientific or anything. Diffraction comes in at f18, and anything smaller than that is notably softer than the best. That leaves 70mm.

Exactly the same story: f7.1 through to about f11 is sharpest, with modest softening either side of that range, and notably softer at f18 through to smallest apeture. The softening at f8 appears here as well as 131mm - you have to look hard to see it, but it's there.

The conclusion then, is that this lens is at its best around f7.1 to f11, and this is what would normally be expected. It also further demonstrates what a star performer the 17-40mm lens is, being close to its best at wide open f4. This test also shows why this 70-200mm lens is so well regarded. Even though there is a softening across the aperture range, you have to look pretty closely at a particular type of photograph in order to see it. On most real world photographs you'd be hard pushed to spot any difference in the middle of the aperture range, and the softness at the ends of the apeture range would only be notable when looking hard at certain types of images.

One area of little importance to the 17-40mm range, but of great importance to the 70-200mm range is that of how much image sharpness can be maintained while handholding the camera. The above shots were taken with the camera on a tripod, so I wanted to do some experimentation with the camera in my hands.

I read somewhere that the slowest practical shutter speed while handholding is the same as the focal length of the lens. That is, at 200mm, which on a 20D is 320mm equivalent, I shouldn't be using a shutter slower than 1/320sec. I need to take into account my own abilities and I have no idea whether my hand is steadier than average or shakier.

Simple test then. Put the camera into shutter priority mode and take the same shot several times at different shutter speeds. I'm not going to reproduce vast numbers of photos here, since it gets confusing trying to see the wood for the trees. Instead, I'm going to use a table. For each focal length - 200mm, 131mm, and 70mm - I took 3 shots, trying hard to keep the camera steady and the shots identical for comparison. I then looked at the shots, looking to see how many of the 3 were usable.

Starting once again from 200mm then:

Shutter @ 200mm	Usable from 3 tries	Practical?
1/50th	0	No
1/100th	0	No
1/200th	3	Yes
1/400th	2	Yes
1/800th	3	Yes
1/1600th	3	Yes

At 1/50th of a second none of the 3 shots I took were usable. One was less blurred than the other 2, but it was still way beyond sharpening into something useful. Even on that best shot, the text from the newspaper image was barely readable. So, at 200mm, 1/50th sec isn't a practical proposition.

At 1/100th of a second I was still in the same position - none of the 3 shots were usable. They were all about the same, and all were notably better than the 1/50th sec shots. I could make out what the text said on all of them. 1/100th sec still isn't practical though.

Then we get to the usable shots. All 3 I took at 1/200th of a second were usable. They weren't razor sharp, but with a bit of unsharp masking they would come out quite nicely. The same applies to 1/400th, although one of those was too blurry to use. All the shots I took at 1/800th and 1/1600th were pretty much alike, and all were usable.

The conclusion for 200mm, then, is that 1/200th is a slowest practical exposure time I can use. If I really want the shot, or don't have much time to compose and steady myself, going to something like 1/500th should pretty much guarantee a sharp photo.

Zoom out to 131mm:

Shutter @ 131mm	Usable from 3 tries	Practical?
1/50th	1	No
1/100th	1	No
1/200th	3	Yes
1/400th	3	Yes
1/800th	2	Yes
1/1600th	3	Yes

At 131mm, one of the shots I took at 1/50th sec was just about usable. It was pretty grotty by most standards, but in a pinch it might be sharpened into something worth looking at. The other 2 were way off. At 1/100th sec things were borderline. I'd say 1 was usable and the other 2 were very close, but not quite there. At 1/200th, all three were usable. The best was very good - I could make out the individual spots of ink that make up the newspaper images. At faster shutter speeds the quality was pretty consistent, although one of the 1/800th shots was blurred. Clearly attention is still needed even with a fast shutter.

That leaves 70mm:

Shutter @ 70mm	Usable from 3 tries	Practical?
1/50th	2	No
1/100th	3	Yes
1/200th	3	Yes
1/400th	3	Yes
1/800th	3	Yes
1/1600th	3	Yes

At 70mm, 1/50th sec is almost practical. 2 out of my three shots were OK, if not great. 1/50th sec is almost usable, but I'd rather speed up the shutter a stop, because anything faster than that resulted in decent sharpeness.

Conclusion then: the rule of thumb just about works for me without taking into account the 20D 1.6x focal length multipler. At 200mm, if I'm careful and pay attention to breathing, etc., 1/200th sec should result in decent images most of the time. At 131mm, my limit would be somewhere between 1/125th and 1/200th. At 70mm anything faster than 1/100th should be fine, and I might get away with a slightly slower shutter if necessary. These figures represent my limit - to ensure a decent shot under less controlled circumstances, halving the shutter speed again would pretty much ensure a sharp image.