I've noticed with my new 40D that the AI-servo mode is not as sensitive/capable in low-light (and perhaps low-contrast as well) situations as my 30D.
With my 30D, if it can lock focus in one-shot or AI-focus mode, it will focus and hold in AI-servo.
With my 40D, the ability for it to lock focus in one-shot or AI-focus modes noticeably exceeds its ability to hold focus in AI-servo mode. In other words, AI-servo mode will hunt looking for focus on targets that AI-focus and one-shot modes are able to lock onto.
Furthermore, I've noticed the 30D's ability to hold focus in AI-servo mode on static low-light targets actually exceeds the 40D's ability to do the same -- significantly so, in fact.
On my 30D, I used to use it almost exclusively in AI-servo mode (with the custom function set so the "splat" button would activate autofocus). The 30D's ability to hold focus on a static target in AI-servo mode was identical to its ability to lock focus in AI-focus and one-shot modes. But with the 40D, I've found I can't use AI-servo all the time, because it doesn't seem to be able to hold focus on low-light static targets -- instead, I have to use AI-focus or one-shot mode in order to be able to focus on the same targets as the 30D's AI-servo.
Is this consistent with your experiences with your 40D? Because if not, it means my camera's focus system is likely defective (or at least needs adjustment).
One more thing: my 40D's peripheral focus points do not appear to be as capable of focusing in low light as the 30D's center focus point. The question is whether or not that's as it should be...
Most of my tests are with the 24-105L lens, by the way, so the diagonal focus sensors in the center are presumably not being used in that case...
I haven't yet done a formal test, but the above appears to be consistently the case for me thus far...
