Is this acceptable CA for the 70-200 2.8L Non-IS?

I bought a new 70-200 2.8L non-IS from B&H back in late June; it replaced a 70-200 F4L IS. I noticed some strange AF behavior with the new lens, but I figured it must be my technique with the heavier lens. The first chance I really had to put it through a proper, controlled test was more 30 days after purchase (i.e. after B&H's exchange warranty), and it became apparent there was a real AF issue, along with what I would call significant-to-severe CA (purple fringing in front & green fringing behind the focal plane). I sent it to Canon for repairs, and they told me adjustments to the AF mechanism would correct both issues (??). Didn't make sense to me, but they're the experts... Anyhow, I got it back, and the AF was almost more erratic, and the CA was still terrible. Sent it back again, and I just got it back. The AF now seems to be correct, but the CA issue persists...

Below is a shot of a high-contrast scene, with a 100% crop. The shot was obviously intended to highlight (pun intended) the CA issue. Even so, if I were to shoot a backlit portrait, I'm sure this issue would rear its ugly head.

So, my question to all of you who own and/or have extensively used this lens: is this "normal" CA for this lens? It seems unacceptable to me (based on my experience with the F4 IS lens), but I would welcome the perspective of other experienced users. I would be VERY grateful for any and all useful insights. Not looking forward to sending this back a third time (would be thrilled if they'd just exchange for a new one at this point).

Thanks in advance.

First, do you have a "protective" filter on that lens? If so, you might try it without any filter. I've seen filters cause or amplify CA, and even seem to increase focus errors (which also amps up CA).

I don't have the 70-200/2.8 non-IS, so can't say for certain... but I would think that not normal (if there's no filter on the lens).

On the other hand, the lens you were using previously (70-200/4 IS) is one of two Canon 70-200s with fluorite elements, that can make for what appears a sharper image and controls CA very well. The 70-200/2.8 IS Mark II is the other one using fluorite. The other three models (including the oldest 70-200/2.8 non-IS) do not have fluorite elements, though they do have some UD or ED elements.

No filters... Interesting about the fluorite. My F4 was super sharp, and had absolutely no CA, so that may explain it. Thanks for your response.

My 70-200L non-IS is twenty years old, beaten to hell, and still doesn't have anywhere near that kind of CA.

Thanks for chiming in. Confirming my suspicions. I really don't want to send this thing back for a 3rd time (I've actually had it in my possession less time than it's been gone to/from repairs)...

^^ Before you do anything, do some tests in controlled lighting conditions. You have very strong backlighting there, so ideal conditions for CA. Do a proper test, and see how the lens performs under normal conditions. If it's the same, send it back. If it only happens in shooting situations like above, I probably wouldn't sweat it. The lens looks pretty sharp otherwise.

I took that shot precisely for that reason - to push it in that regard.

I agree that I won't shoot in lighting like that very often, but it does happen on occasion (sun or strong strobe as backlight, with strobe or flash as key front light), and it would be nice if it could handle it.

My points of comparison are the F4 IS lens I've mentioned and several 300 F2.8 Mk I lenses I've rented over the years, none of which have given hints at CA issues, much less to this extent. The comment above about fluorite coating may explain that - I haven't researched to verify, or to check if the 300 F2.8 lens has it.

If it's just a limitation of the lens, then that sucks for me not knowing before I bought it (my bad). But if it's worse than a typical copy of this lens, then I want it fixed/replaced, given the price.

I spoke with Canon again today (at length), and it's gonna go back in. The first 2 times it went California, and this time they want it to go to Virginia, so an "engineer" can assess it.

Guess we'll see how that turns out. Pretty bummed by the experience. I know every manufacturer of every type of device is gonna end up with a few bad copies, and someone has to be the poor slob to get it...