Teardowns and Disassembly

A Peek Inside the Canon 24-70 f/2.8 Mk II

Published September 11, 2012

You probably know we got 5 copies of the Canon 24-70 f/2.8 II today and did some resolution testing before they all shipped out. If you read this blog often, you probably know that Aaron and I gave every lens a careful shake, peered inside, and generally looked for any possible excuse to open one up for ‘repair.’ Sadly, all 5 copies were sharp and had no flaws. Even after a second look.

But then we realized Kristin and Tyler had a baby yesterday (congratulations!!!!) so they aren’t here. Drew is running the place by himself and far too busy to check and see if the guys in the repair department are taking something apart they shouldn’t. So as long as the lens got put back together and shipped out on time, who would know?

Seriously, though, I really, really wanted a look inside this one. I’ve written for years about how much copy-to-copy variation there is in the original 24-70 f/2.8 and how it requires optical adjustment almost any time it’s dropped or misused in the slightest.

A group of 24-70s on their way to Canon for optical adjustment


So I wanted to look inside and see if things might be different with the new lens.

The Front Group of the 24-70 Version I

I put this part separately because I’m convinced the front element design of the original 24-70 f/2.8 accounted for a lot of its problems. The front element is a big, heavy piece of glass, but in the original version it is also the element that is adjusted for centering via screws around the edges.


Version I front centering adjustment


It also adjusts for spacing from the other elements via a sliding ramp.


Adjustments for front element spacing


Finally, it has a set of eccentric collars that adjust the tilt of the element.


Collars to adjust front element tilt in the original 24-70


Given all of that adjustability involving the front element, I’ve never found it surprising that the original 24-70 was bad to drop. It often appeared fine after being dropped, but had actually become decentered and image quality suffered. The bottom line of all this is that I hoped the new lens had a different design for the front group. Guess what?

The New Lens has a Different Design for the Front Group

It has a new design for almost everything else, too. But let’s start at the front. As usual, Aaron is doing disassembly honors while I take pictures and make helpful suggestions like, “Don’t tear that flex.”

After peeling off the makeup ring, there’s a sturdy set of 6 screws holding in the front element (versus 3 centering screws in the old version).

Take those out and the front element comes right out. There’s a strong molding ring around the seating areas and long screws used to hold it in place.

There are no adjustments on the front group. It’s permanently sealed in a heavy plastic mold that seats firmly into the lens mount.ย ย If this lens drops you might crack a filter ring, but I can’t imagine decentering the front element.

Around to the Mount End

With the lens mount removed, the circuit board and flex cables show. Now this is the kind of thing that really makes no difference, but I do like to see someone took the time to design all of the cables so they fit neatly, not requiring glue to hold them in place and crisscrossing each other.

With the flexes disconnected and the circuit board removed we see the rear barrel mount. Again, I’m seeing that little bit extra that I’d hoped to see. There are multiple long screws holding everything in place, not the usual 3 or 4. This is all plastic, but it’s thick, heavy plastic, not the thin plastic mounts I worry about breaking or stripping.

Removing 6 screws, the zoom brush on the side (not shown), the zoom keys, the external barrel and rings come off in one single piece containing the switches ย (sorry about the focus) . . .

Leaving all of the optics and the USM motor in a separate piece.

Another one of those really nice touches you’d never know about if you don’t take lenses apart. The distance scale isn’t just a piece of plastic glued to the inside of the barrel. It’s mounted to a metal piece that screws into place, and it’s much larger than it needs to be, which would help prevent light leaks through the distance window.

One more set of screws to remove (notice the screws are paired two images above – I’ve seen few lenses use two screws when one probably would be enough) and the USM assembly slips off in one piece again, leaving the entire optical assembly behind. The entire optical assembly (less the front group we took off first) is in Aaron’s left hand, the USM assembly and second barrel in one piece in his right hand. This is a very cool thing that I’ll discuss more in a minute.

The modularity of this disassembly is nice, obviously, for someone who has to do lens repairs. But the part I really liked seeing is on the central optical core. If you look below you can see the sliding helicoids with screws and nylon collars that hold the lens elements in place. The collars seem larger, heavier, and just tougher than the ones in the older version. Those collars were one of the things we saw wearing out on older copies and causing problems.

Also, do you notice the metal studs extending out from the middle barrel above? I hadn’t seen those on any lens like this before (sometimes you see a single one on a rotating ring). They keep the outer ring, with the USM motor, in proper place. But also they actually are stops that stop the barrel when the lens is fully retracted. In other lenses, the collars around the screws in the helicoids stop the lens motion. These metal stops should save a lot of wear and tear on the more delicate collars. Color me impressed. Again, this is just one of those nice touches that nobody will ever know about, but shows some design team was thinking, “How do we make this last a long time?”

Finally, I’ll give you a close-up that shows the adjustment mechanisms in this lens. You can see the two brass eccentric collars that adjust one of the rear elements for centering and tilt. They are far more protected than the front element adjustments on the old version. I can’t imagine a fall or jar knocking them out of place. Of course, the downside is whenever adjustment is needed, it’s going to require a lot of disassembly. But for almost everyone but us, that’s what factory service is for.

One last thing before we end. Look at that solid nylon ring around the screw above and compare it to a new and old ring from the old version below. The new ones certainly seem more robust.

New and heavily used helicoid collar from version I lens


Nothing but time can really tell how well this lens is going to hold up. Nothing but reviews of lots of copies by lots of users is going to show us how copy-to-copy variation will be. But it’s apparent to me that Canon has taken the time to design the lens well and build it sturdily. I totally agree, that for the price, it should be well designed and well built. But experience has taught me that is not always the case for a more expensive lens. I’m glad it is the case here.

Would I get one? Of course – I’m a resolution freak, a gear-head, and I don’t have a 24-70. I’ve just examined the highest resolving zoom lens I’ve ever tested and found it’s also built and designed superbly. I can’t walk away from that combination.

But is it a practical lens for you? Well, if you’re a working pro or a very serious (and well-healed) amateur I’d have to say yes. It is so good that if there were a 40 megapixel camera in your future, you’d want this to put in front of it. Plus it looks like it will hold up over the years better than its predecessor.

But at this price it’s not for everyone. If you prefer wide aperture primes to f/2.8 zooms, don’t shoot standard range a lot, or simply are not rolling in cash, you’d have to consider the price. There are excellent alternatives for a lot less money. Enough less money that you could buy another lens with the difference.

Whether you want to spend the money or not is likely to be a moot point for a while. Finding these is pretty difficult at the moment.


Roger Cicala


September, 2012



Author: Roger Cicala

I’m Roger and I am the founder of Lensrentals.com. Hailed as one of the optic nerds here, I enjoy shooting collimated light through 30X microscope objectives in my spare time. When I do take real pictures I like using something different: a Medium format, or Pentax K1, or a Sony RX1R.

Posted in Teardowns and Disassembly
  • Brian Feinzimer

    Hi Roger-

    A few years since this has been published, how is variation holding up on these ver II lenses versus the ver I. I did read your 2011 post on variation on the 24-70 ver I.

  • Sergiu Mosoia

    Hi Roger, thank you for the test, this is disappointing for me, really.
    Here is why: I had big expectations from this lens, hoping that the optical quality will match the Canon 24-70 II – I frequently use 1/320 @ 70 mm (f/2.8) with this lens on the 5D Mk IV…which is short, very short (on the 5D Mk III I used 1/160 with the same lens, but I also had to correct now for the increase in resolution… and the bigger issue is that my hands are no longer that steady…).
    Again, thank you.

  • Brian anderson

    So it has been four years since the tear down report. Overall how is the canon 24-70 ii holding up?

  • Anonymous

    I would like to to replace the “lens sleeve” on my 24-70 ii as I dropped it from a 2 feet high tri-pod. Looking at the pictures it does not seem complex at all. Is there any steps on the process I need to be extremely cautious on? Do I need to disassemble the rear mount or do I only need to remove the front element? I am pretty well experienced at taking apart electronics, dealing with expensive cars, iphones, ipads, smartphones and such. I would like to do the repair myself since canon quoted me an arm and leg just to replace it.

  • john

    Hi Roger,

    Is there any ADJUSTMENT for the extending barrel tightness? I have 2 copies of 24-70mm II, one is very tight without any wiggle when extended at 70mm and the other one has wiggle about 2-3 mm (especially up-down wiggle).


  • Roger Cicala


    I’m not familiar with repair centers in your area, but definitely the nearest Canon factory service center if at all possible.

  • hi there..i have a canon 24-70 zoom lense,its not working properly,bymistake it was dipped in d water.currently am in oman.plz suggest me where should i give it for repair.

  • Marin

    Hi Roger,
    I’d like to see how Olympus ZD 14-35mm f/2 lens compares to this one. If you could do it on OM-D via an adapter, it would be great.

  • Roger Cicala


    I agree with you, but given the number (and it’s dozens) or dented filter rings we’ve replaced, I can only assume lots of people leave the hood off.


  • OMG!! Roger/Aaron, how do you control your anxiety when are removing those screws of such a high price lens.. I have disassembled my TV/laptop which are pricy too just for the sake of curiosity, but could not be dare enough open 70-200 L ๐Ÿ™‚ Really well done, and well written article.

  • Martin Datzinger

    Thank you very much for the insight! I didn’t know anything about the decentering risk of version I. But I do have to beg the question, who in their right mind would use the lens without the hood in place, which, as it is mounted to the outer barrel, should protect the innards of the lens from any sort of minor bump, or even heavy impact? Now I can absolutely believe that this design had to be traded in for better resolution – and it seems to have paid of in that regard. But now, even more so with the added leverage of the hood in place, there is always stress on the inner tube, every time you put down the camera with the lens attached, let alone if you bump its front. Don’t you think that undoubtedly better inside construction and significantly worse mechanical protection plus higher adjustment effort will even out somehow? You loose the far more favourable flare (and even rain) protection properties of the old design as well, btw.

    Best regards,

  • RP

    Excellent Roger. And regarding having or not having IS what is your opinion?

  • Roger Cicala


    I don’t have enough math to say it always lowers resolution. What I can say is it’s always an additional optical element in the light path, and that optical element changes the entire design of the lens. If you have a great design, you can’t “add IS” as people like to say. You design an entirely new lens with an IS unit. The Tamron 17-50 f/2.8 is a great example. The VC ‘version’ isn’t a version at all. It’s an entirely different lens, and one that doesn’t resolve nearly as well as the original.

    Does that mean an IS unit lens always has lower resolution? No, not at all. The Canon 70-200 f2.8 IS II is sharper than the NON IS version. But it’s an entirely different design with 23 element compared to the non IS 18 elements. The higher price tag isn’t because ‘they added an IS unit’. It’s because they completely redesigned the lens.

    In the case of the 24-70 II, the resolution is incredibly good. Several supposedly reliable sources said there were actually three 24-70 designs that were prototyped, one of which did have IS. I’m making an assumption that the IS lens didn’t make the cut because it didn’t resolve as well as this lens does. I think Canon decided to go with the better resolving lens instead of the IS lens. Even if the prototype rumor is false and this just happened on the drawing board, I’m pretty sure someone made the decision “we can get better resolution in this design without IS”. Plus they had to think about things like “we’ve got to make this more reliable, so no more adjustable front element, and no more heavy internal zoom barrel on small helicoid screws”. One way or another, something had to give and they decided it was the IS unit.

    I’m totally, completely speculating now: I think maybe Canon is designing a set of lenses that are superbly high resolution because they plan on releasing a camera that requires such lenses. If they knew, for example, a 40 Mpix sensor was in their future, I suspect the design team was charged with “Make the highest resolution possible”.


  • Randy

    Roger – Are you saying there is something about IS that reults in lower resolution–or that they couldn’t design around the IS mechanism without losing quality? I guess I don’t see what one has to do with the other.

  • Roger – I appreciate your response. Those were the lenses on my mind ๐Ÿ™‚ I actually have both- The 24-70 from Tamron needed serious AF adjustments and I still feel it is on the soft side that coupled with your finding about the front element has me packing it up to return. The 28-75 is a little slow to focus in low light but otherwise I have been seriously impressed with the lens and after 4 solid years of use it is still sharp and solid as ever.

    Thanks for all you do!


  • Jim Thomson

    Since you took the lens apart did you determine if it had the sensor required for the impoved AF?

    I’m glad I shoot a crop body so I’m not tempted to purchase this lens.


  • Robert

    Cool, can’t wait to see the AF report!

    In regard of your explanation for the reason why it doesn’t have IS, I think it’s possible but it could be weigth also as Canon clearly said when they announced it that they were responding to the demand of pros who wanted less weight. Probably would have weighted more than the current version if they would have added IS and keep the high resolution.

    They could have at least updated the 24-105mm with better optics and a slight price increase (~1600-1800)! And but that in a combo with 5d3. Now they’re going to lose the amateur going for the new tamron since it beats both the old 24-70 and the 24-105 (probably including myself if it holds up the test of time)…

  • KyleSTL

    Wow. Failure on my part. I feel like a total idiot. I liked the old format better. Any chance you’ll go back to the old spacing and font for the blog, Roger? More well-defined headers between blog posts would have saved me some embarrassment.

  • Roger Cicala


    I did get the shots for and AF test, but it will be days before I can run them through Imatest and process them.

    As to IS, I don’t know anything for certain, but there’s a very strong rumor that Canon had 3 prototypes of 24-70s, one of which had IS. I think it’s very possible that the IS design had lower resolution and Canon decided to go with resolution over IS.

  • Robert

    Thanks a lot Roger, truly insightful!
    Now only a small AF test (hope you did this before shipping them all!) to see if this lens takes advantage of the new closed loop AF algorithm that you’ve discovered (most probably is) and were done with this almost perfect lens.

    Why oh why didn’t Canon includes IS at this price point! It would truly have been perfect walkaround low-light standard zoom! I guess they need to make some more profit in a few years with a new lens release!

  • Roger Cicala


    I think the obvious, and a very good alternative, has got to be the Tamron 24-70 VC. I’d also consider the ancient Tamron 28-75 f/2.8, particularly if you have a crop camera. It’s still a good lens and hard to beat for the money (about $600). I shot with those for years on D30 and D40 and was very happy with it. On a full-frame, though, the corners are bit soft.

    In the past, this would be where I would plug “or a few nice consumer grade primes” because in the past lenses like the 50mm f/1.4 and 35mm f/2 would be as good as the zoom at that focal length. But the new primes, at nearly $1,000 are going to add up quickly. The older primes aren’t quite as good, but the 35 f/2 and 50 f/1.4 are certainly cost effective alternatives. I’m not as thrilled about the older 24mm f/2.8 or 28mm f/1.8.


  • Roger! I love you ๐Ÿ™‚ And your writing. Now please suggest some”. . excellent alternatives for a lot less money.”

    My credit card would appreciate that ๐Ÿ™‚

  • Kai


    I’d say that the title of Lord Superior Lens-Geek is more appropriate for you (and Aaron). I absolutely love that you professionals share these strip-downs and resolution test results that us mere mortals can only fantasize about.
    I’m looking forward to the AF accuracy testing report!
    My only ‘complaint’ is that of Ian above – this is quickly proving to be one set of very expensive posts for me.

  • Ian

    Roger, these are very expensive posts (for me and many others) you are putting up on your blog, but thank you for your fine work. My wife and AMEX bill will probably be less appreciative.

  • KyleSTL, the results are in the chart mid-way down this page:

  • puppy_kicker

    KyleSTL, must you be “that guy”? The VERY NEXT link, which goes to an article Roger posted within an hour of this one, is called “Canon 24-70 f/2.8 II Resolution Tests.”

  • When I see you guys disassembling lenses (expensive lenses), it give me a pain in the heart … lol
    Nice work!
    (sorry for bad english… using GTrad)

  • You are a brave man. Or… well let’s leave it at brave. Seriously cool article. I want one, what’s inside does matter.

  • KyleSTL

    [quote]Iโ€™ve just examined the highest resolving zoom lens Iโ€™ve ever tested and found itโ€™s also built and designed superbly.[/quote]
    Sharper than the 70-200mm IS II? Care to share the test results?

Follow on Feedly