dog snaps's reviews

Canon EF-S 17-55mm f/2.8 IS USM

This lens seemed to have ideal specs. Reviews were almost uniformly good. I thought this would be a great compliment to my 70-200 f/4 IS L, especially for travel and low-light shooting.
Build quality was very good for a non-L lens - at least as good as my Canon 100/2.8 USM Macro. However, AF was very inconsistent on my Rebel XT - maybe 30-40% acceptable at f/4 at 17-24mm, in good light and with contrasty subjects. AF accuracy improved to about 60-70% at 35-55mm under similar conditions. I would consider an AF accuracy rate below 90% to be unacceptable in normal situations, even for the kit lens or the 50/1.8.
Even though I have not had AF problems with the XT, using lenses such as the 70-200/4, the 100/2, the 50/1.4, and the Tamron 28-75/2.8, I thought I might get better results with my Rebel XTi, since it is supposed to have the more precise focusing system of the Canon 30D with lenses f/2.8 or faster. Unfortunately, my results showed no improvement at all.
I was horribly disappointed, especially considering the price of the lens. Fortunately, the folks at B&H Photo were very helpful in arranging a return for refund.
I thought of asking for an exchange until I read a review of this lens at kenrockwell.com. He complained of his lens having significant AF problems, too. Maybe he and I bought the only two bad copies in North America! I don't think I should have to use some manual focus work-around (focus at longest focal length, then zoom out and recompose) on a $1000 AF lens specifically designed for Rebel 300D/350D/400D, and Canon 20D and 30D, AF bodies.
I have several very good Canon and Tamron lenses, but I have never had to return one that cost more than $500. The last I returned was a $200 lens, well over 4 years ago, so I'm not disposed to look for problems, nor am I a "pixel-peeper." I also use BreezeBrowser, Photoshop CS with PTLens, and DxO Optics Pro 4.1, so I'm not using 3rd-rate editing tools. However, I do have expectations from a $1000 lens. They were exceeded with Canon's 70-200 f/4 IS L, but not met with the 17-55 f/2.8. No matter how well-built a lens is, or how well it performs in the lab, it isn't much good if you cannot focus it accurately and quickly.
If this is a calibration problem, then I would wonder about a factory that would ship out thousand dollar EF-S lenses that did not meet specs.
I avoided the impulse to award a very low rating, based upon only one sample, so as not to unfairly skew the average.