Geek Articles

Completing the Teardown of the Sony FE 70-200 f/2.8 GM OSS: Part II

Published February 27, 2017

You probably shouldn’t read this article unless you’ve at least skimmed through Part 1. To summarize, we had a Sony FE 70-200mm f/2.8 GM OSS lens that had been dropped and no longer would focus. We tore it down both to see if we could repair it and to see what-all was up inside there. In the first part, we did the initial disassembly of the major components and a complete disassembly of the back (mount) half of the lens. We found a few surprises and some different ways of doing things. But we didn’t find the problem with the lens.

The Rest of the Teardown

So after a snack break and some more caffeine, we tackled dissecting the front half of the lens. You can see the Southern Fairy Tale ring on top, with the black plastic zoom-link ring right underneath it (you can see a metal zoom key peaking out above it on the side away from Aaron’s hand)., 2017


We start by removing the lock-on ring that connects the front half to the back half of the lens. You can see the ring of 8 screws holding the ring assembly to the front of the lens. There are 4 additional screws in the little notches, two of which are by Aaron’s forefingers., 2017


Taking all of those out let us remove the lock ring from the front half of the lens., 2017


As you might expect, only the plate is fixed to the front of the lens, the locking ring spins freely behind the plate, threading onto the rear half of the lens to attach it., 2017


There’s also a shim under the plate. This would be a spacing shim either for backfocus adjustment or for optical spacing (probably the latter)., 2017


Here’s where we found our first problem with the lens. It didn’t prevent focusing; but it may well have caused an optical issue. Remember, this lens had been dropped. We were told the drop was only about 2 feet, and there was no external evidence of damage (no scuffs, dents, etc.)

When we looked at the attachment plate above more carefully, it became apparent that two of the screw holes on one side of the lens were bent downward slightly., 2017


Two holes on the opposite side were bent upward slightly. This plate is attaching two heavy lens-halves, and it’s fairly thin aluminum (or aluminium, if you have a non-U. S. plate). The picture below shows the thickness of the plate through one of the screw holes; we measured it as 1.2mm. That might be thick enough if it was stainless steel or titanium (I’m no materials engineer) but this is aluminum; we could bend it with our hands. If the screw holes were bent, we had to assume the rim of the plate was also., 2016


Some engineer with way more math than me obviously calculated that this plate was thick enough to hold the mass of the two lens-halves together. But I’m damn sure he didn’t consider the part about force and velocity squared and all, because this plate, the one I can bend in my hands, is not going to resist much force. At all.

While the plate did hold the lens together just fine despite these bends, we’re obviously concerned that the bent plate could affect the lens optically. There are strong elements on either side of that connection and a tilt of a few microns could cause some significant optical disturbance.

Anyway, while we aren’t happy about the plate, it certainly wasn’t preventing the lens from focusing, so we continued on with our disassembly after getting the parts department to order a replacement plate.

With the connecting ring out of the way, we can access the keys to remove the inner ring linking the zoom barrel you turn to the zoom mechanism inside the lens., 2017


Removing those keys lets us take off the connecting ring and then remove the front outer fixed barrel. Like the other outer barrels, it has weather resistant rubber strips; the blue one is a seal, the black fuzzy felt one further back goes under the rotating barrel., 2017


Looking inside the outer barrel, you can see the focus-hold button mechanisms and the flexes that connect them., 2017


Underneath we start to see the inner workings of the zooming groups and front focus motor. You might think ‘look at all the cool stuff in there’. We thought, ‘oh, crap, this is going to take forever’. If you look you can see a position sensor over on the left side, the big GMR unit hanging over the ring USM motor on the right side, and flexes criss-crossing every damn where. More and more, this lens looks like it’s got a lot of Nikon heritage., 2017


Rotating around to the other side, there’s another PCB, presumably doing some position processing since it’s connected to all those position sensors., 2017


The focusing ring was the next obvious removal and with the rubber taken off there was easy access to the screws and collars holding it in place, after which it slips off of the lens., 2017


This is at least partially a fly-by-wire-focusing lens so inside the focus ring is a magnetic strip. When you turn the ring a sensor reads the strip and sends a signal to move the rear focusing elements accordingly. The front focusing elements can be controlled mechanically by the ring or through the ring USM., 2017


And now we’re getting down to the inner workings of the front half of the lens. We were pleased to see an optical adjustment collar up here, too., 2017


I’m going to compress about an hours worth of work now, trying to keep this post a reasonable length. But that set of flexes and sensors you saw 5 pictures ago was removed., 2017


After that we spent 4 or 5 minutes colorfully discussing the parentage of the lens’ mechanical designer who saw fit to make the stacking of the ring USM an integral part of the disassembly. In a nice lens, designed with the thought that someone might actually repair it some day, you take out a couple of screws and take off the ring motor assembly. In this lens we (I’m using the Royal “We” here; Aaron does all the hard stuff) have to take the USM apart piece by piece to get further into the lens.

So after carefully marking the alignment, Aaron started counting turns to remove the pressure plate (because it has to go back to exactly the same point)., 2017


After which we unstacked all of the rings and the USM motor was off., 2017


With the motor off, the next ring is a clutch or transfer ring that translates focusing movement to the focusing elements. Well they would, if they worked. We were now able to see the front focusing group didn’t want to move very well since we could turn this ring by hand now. You can see the ring rolls on very robust bearings (red line) and moves the focusing keys connected to the internal group (green line)., 2017


Once we took the keys and one locking pin out, this ring came right off. We were grateful the tensioning springs underneath it came right off with it. Sometimes those spring out like a jack-in-the-box and we have to spend 15 minutes on hands and knees searching for them., 2017


The next ring down comes right off, too. It rotates on a beautifully made bed of 30 ball bearings, which is one of the reasons this focus ring is so nice and smooth. We both did another prayer of thanks, because none of the ball bearings escaped the disassembly table and made a mad dash for freedom. Ball bearings are the escape artists of the lens prison system., 2017


Just in case any of you want to know how we keep the ball bearings from running away during disassembly, they are corralled on a Dycem non-slip pad. Dycem is one of the great accomplishments of humanity, ranking slightly below the invention of fire and the wheel, but well above the printing press in my pantheon of most important advancements., 2017


Next, we took off the front (filter) ring, which like all other joints in this lens has a nice rubber seal. Right below the top ring you can see 3 of the 6 heavy rollers that the front focusing element slides up and down on., 2017


Then we removed the front group, which is two cemented elements. This, of course, could have been done at any time during the teardown, but we were doing most of our work with the lens face down, so it seemed best to leave it until near the end., 2017


Oh, Yeah. About that Focusing Problem

At this point we we had one piece left – the front group barrel assembly you can see in the picture above. This contained some zoom elements in the smaller part of the barrel, and the front focusing group in the larger part. Reaching through the slots with forceps we could tell that the focusing group was still grinding when we moved it up and down, with a lot of resistance. We could also see it was slightly tilted. This usually means a roller or cam was broken or bent.

We removed all 6 rollers, which by the way are some really robust, ball bearing containing, pieces of engineering. They were all working fine and even with them out of place the focusing groups was still scraping when we tried to move it., 2017


Looking through the slots where the rollers had been, we could tell there was some form of spring loaded plate system with the focusing element; three of the rollers above went into the actual focusing element, three into the plate below the spring. But we couldn’t find a way to open things up further., 2017


We decided to put the lens back together, bite the bullet and see if the service center could repair it. But being a suspicious person by nature I did some checking first and found out, in typical Sony fashion, this entire assembly was considered a single part. So if we sent it in they would replace this entire assembly at a repair cost of slightly more than half the price of a new lens. Needless to say, we switched into nothing-to-lose mode and went back to looking for a way into this assembly.

We could have removed the zoom groups from the small end, but that wasn’t going to give us access to the larger focusing area. The only way in appeared to be by removing the second group, which was held in place with  both a spanner ring and also with adjustable collars through the side of the lens., 2017


Oh, and lots and lots of silicone glue. So we (royal “We” again – Aaron did this while I made more coffee) position-marked and removed the collars and chipped out lots and lots of glue., 2017


Finally we got the element out and could actually look down onto the top of the focusing element. (And yes, we realize there’s more glue to clean up before reassembly. Don’t be snarky. You’re tired just from reading about it, so you can understand we were really sick of doing it.), 2017


There’s another spanner ring down over the focusing element, so we took that out and then just about lost our minds. Because there’s another ring of beautiful ball bearings under it. We’ve never seen anything like this before and at this point really had no idea what we were seeing., 2017


Despite multiple cups of coffee and the resultant tremor, Aaron managed to get the focusing assembly out without spilling any ball bearings. (Notice how I use ‘we’ when I know everything’s going fine, and switch to ‘Aaron’ when there are 30 ball bearings that might spill onto the floor.), 2017


So we made our second ball-bearing corral. We kept them separate on principle, but they’re the same ball bearings as the other set. What size you ask? Oh, 2mm, which is just a bit thicker than that connecting plate I’m obviously not happy about., 2017


We looked at the internal focusing assembly and saw there was a pressure spring at the bottom (red arrow), pressing up against the focusing element in a cage-like arrangement. It appeared to be out of position, creating a bit of a tilt of the element that was scraping the inner barrel when the focusing group moved., 2017, 2017


We (yeah, I’m back to taking credit now) reassembled it with the spring properly positioned back under it’s little latches like it should be., 2017


I won’t bore you with reassembly, except to add that we discovered that the spanner ring over the ball-bearings is like that for a good reason. Tightening the spanner compresses this bottom spring and you can ‘dial in’ the proper amount of tension for the focusing element. Too loose and it rattles around a bit. Too tight and it gets a little resistance to sliding up and down. It’s an interesting arrangement.

But if this spring comes out of position with any regularity the repair is going to be a royal PIA. This is just one dropped lens so let’s not run screaming about what may or may not happen again. We may never see this again. Or we might. Only time, and a few more dropped lenses, will tell.

Usually I end these posts by saying we’ve got the lens reassembled and it’s back in stock now. In this case we’ll wait for a new connecting plate, and then optically test and adjust the lens after final assembly, and really, really carefully check it before putting it back in stock. But we did reassemble it with the old connection plate to make sure the autofocus system worked properly.

So What Did We Learn Today?

Well, first and foremost, that lens repair is a miserable way to make a living. And lens repair on this lens will be miserabler than most.

Second, we learned that dropping your lens is bad, so don’t do that.

Third, 1mm thick aluminum plates maybe shouldn’t be used to hold two halves of a heavy lens together. And yes, Fanboy, I know the Sony engineers are smarter than I am. They designed 962 really great parts that I couldn’t have designed. But it doesn’t take massive engineering knowledge to figure out that the thinnest piece of soft metal shouldn’t hold the two biggest pieces together.

And finally, we learned there’s some really elegant and complex engineering in the Sony FE 70-200 f/2.8 GM OSS lens. There’s also some engineering that just seems complex. It may be far more awesome than I realize and I’m just missing the point. Or not.

I’ll repeat what I say all the time: Sony is trying a lot of new things; that’s how you advance. I completely admire the risk taking and efforts to try new things when most manufacturers are just fine-tuning what already is. Some of these new things turn out to be awesome, some don’t. It’s just as important to identify which new things are not better, or not even adequate, as it is to identify when the new thing is a dramatic improvement.


Roger Cicala and Aaron Closz

March, 2017


Author: Roger Cicala

I’m Roger and I am the founder of 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 Geek Articles
  • Roger Cicala

    Rainbow Dash, no that’s it. That part is well designed for repair.

  • Rainbow Dash

    I own this lens and have the filter thread damaged.
    As far as I can see the replacement of the front filter ring is quite easy: remove the focus ring rubber and the clear tape underneath. Then I can unscrew and replace the filter ring.
    I do need only a new filter ring and a need strip of that clear tape.
    No optical recalibration needed.
    Am I missing something?

  • EvilTed

    I dropped my 70-200 F2.8, mounted to my A9 from hip-height onto a grass playing field.
    I missed the holster as I swapped between 70-200 and 24-70.

    The lens mount on the 70-200 bent enough that it would not attach to my A7r III very easily.
    On closer inspection, the mount was clearly deformed at two of the mount screws.

    The cost to repair?


    Sorry, but a Canon or Nikon 70-200 F2.8 is a professional grade lens. The Sony is not.
    The lens mount is 1.2mm aluminum as Roger found and this could be bent by hand pressure.

  • Dragon

    I had Cats back in the day and I still have a bunch. The new stuff still holds up very well and it does a lot more work than the old machines. I think their truck engine problems were aggravated by rapid changes in emissions regulations and that division somehow didn’t have its act together and Cat exited that market. The equipment division and the power generation division are doing fine. Often there is a penalty to being too big.

  • hugh crawford

    Perhaps that piece is a sacrificial weak point designed to break rather than transferring force to something harder to fix deep within the lens assembly? A mechanical fuse if you will?
    Reminds me of those idiots that replace the plastic foot that keeps breaking on their flash with a piece of milled aluminum and then tear the top plate off of their camera.

  • hugh crawford

    I grew up on a farm and in the 60s we had a pair of government surplus Cat D6s bought in the 1950s. We would only shut them off to change the oil because they were such a pain to start. Start the pony engine with a rope then use that to start the D6, hate those pony engines. You could rebuild them in the field, in a tent on a tarp of course. Totally unkillable tractors, we kept using them until 2000 or so when we sold them with the farm. On the other hand just Google Caterpillar C13 and C15 to see how far the mighty have fallen. Mercedes Benz in the 1960s and 1970s was the same way, We drove ours for 700,000 miles until a tree fell on it. Now they realize that making things to last forever is bad business. If you want machinery that will just keep running until, well whatever wipes humanity from the face of the earth, the old two cylinder John Deers are what you want. If someone made cameras like that, well back then lots of companies made cameras like that notably Leica, Nikon, Linhoff, and Graflex before that sewing machine company bought them. Zeiss notably epoxied their cameras together to prevent them from being serviced.

  • Ayoh

    Roger you seem to have quite high confidence in the claim of customer that he dropped this lens from a height of only two feet..
    This leads you to a criticism of sony engineers that the the lens is not robustly designed against general rough handling that it might see in practice. Yet is there any evidence that the customer was being honest about how the damage was sustained? Is it less likely that the team of engineers who successfully design this complex mechanical assembly did not think of a seemingly obvious parameter when specifying the plate thickness?
    Who would you give the benefit of doubt in this scenario – a random customer who damaged the lens or a team of specialised engineers responsible for complex design scenarios? You seem to think the engineers made some obvious oversights based on the word of this customer.
    To me as mechanical engineer the internal assembly looks like it is possibly intentionally designed to localise plastic deformation upon overloading to that aluminium plate. If you overload the lens assembly sufficiently (such as due to large load on account of a fall) some material will eventually plastically deform and result in permanent deformation; if the lens assembly is uniformly stiff it will be difficult to localise that deformation to a predictable location (where a deformed part of the assembly could be easily replaced). However, If a section of the assembly in the load path is markedly less stiff, then the deformation will be predictably localised to that location..i.e. in a replaceable component. The fact that a lot of glue is also used to seal lens elements also suggest robustness against possibly decentering.
    Whether this a concentrated failure point design is the best approach for such a telephoto zoom lens I am not sure. But it seems somewhat arrogant of you to imply the engineers do not know what they are doing and did not think of this detail. To me it seems they did think of it when specifying the design.
    If the plate is in fact the aluminium plate is not stiff enough to avoid deformation under general rough handling, replacing it with stainless steel with the same geometry would make it about 3 times stiffer so is an easy design correction.

  • VTC Media

    So just what happens, customer wise, when they return a lens bent like a flamingo leg?

  • Matthias Welwarsky

    Well, Sony are not alone to not consider heavy mechanical shock in lens design. I had my Tamron 24-70/2.8 break right through in the middle after taking a fall of around 2.5m inside the photo bag, mounted to the camera. And I think they’re right. As a lens maker you can rightfully assume that photographic equipment is well cared for and protected from mechanical harm. And you’ll admit that the little aluminium ring is a good indicator for a repair mechanic about what has happened to a lens. I’ll even venture saying that little piece is the “canary bird” of this lens. Had it been any sturdier, the lens would have been broken just as good with much less indication as to what has happened.

  • Jase1125

    Always have to be the village….. well you know… in the thread.

  • disqust101

    “I think that one key difference is that people who buy serious construction equipment tend to be much more knowledgeable about what they’re getting than photographers”

    Hahahaha. Just snorted coffee thru my nose. Most construction types have no idea and abuse their equipment until if fails. I know, I ran a cstr biz for 20+ years. The good news is good equipment is built with that sort of ignorance in mind – so companies (like CAT) tend to overdesign things. In the good old days, Terex (owned by GM at the time) used to produce the lamest junk you’d ever see. It would “look” like a CAT, but run like and last like it was a Yugo.

  • Prefer Anonymous

    the deformed aluminum piece may have been a ‘sacrifice’ design, to prevent other parts from impact failure.

    it may have been a specification error, as aluminum alloys comes in varying types as well, some softer, some a tad more rigid, even brittle.

    all that silicone?, seems like a curing process was used as most silicone takes a good day or so to become fixed, maybe some cute new light curing stuff that polymerizes in a few seconds?

    as for the little springy thing that came out of place? shame shame shame on sony!

    know any good shops that do mirror refurbishing? ive got a few cat’s that have seen too much moisture over the years.

  • Some are nearly this bad, but I can’t think of any worse off of the top of my head. There are certainly some lenses that we just won’t work on because it’s not worth the time involved. We’ll still carry it, but not try to repair it in house.

  • Terry

    Thanks for the detail story and sharing! Always happy to read and learn from you guys!

  • Pete No Surname

    “And lens repair on this lens will be miserabler than most.”

    Hi Roger, are there many that are worserer?

    And do you, once finding a lens/e that is worserer, banish it/them from your hallowed turf?….

  • Pete No Surname

    Would it help you if we turned our screens to portrait mode?…..

  • Konstantinos

    It is funny reading all comments below that people is expecting for a lens being made mainly from glass and ligthweigth materials to be working normally after a drop of 2-3 feet… When lenses are heavy (Zeiss, new Sigma 85mm Art) everyone is complaining about the size, weigth and bla, bla, bla… when manufacturers use ligther metals (eg: Aluminum) people complains about it bending under pressure. Hum…
    Again, lenses are not meant to be dropped. If they do, things will brake. I am a Nikon user but I believe Sony´s engineers weren`t using a “crash test” approach when designing this lens.

  • Munchma Quchi
  • I think that’s about it — it will probably be different on the Mk 2.

  • LAB 2.35:1

    Roger – with this sort of of finding (the aluminum ring)… what do you think are the chances that Sony’s team might consider a parts change and swap out the ring for something more robust in the future production runs? Given your close relationship with their engineers, I would suspect that they take your feedback more seriously?

    This reminds me of the infamous issue with the plastic lens mount on the first A7xx bodies, which were obviously replaced with a proper metal ones on the Mk2.

  • I once dropped a Sigma 150-500 straight down on its rear end (my Pentax K-5 was attached, so it took the brunt of the brutal crash into the gravel). Say from 2 feet height, and the only thing that happened was a scratched LCD protector, and the mount lost contact with the OS in the lens, that’s all! And all covered by my insurance — nice!

  • I’m not understanding your “go through the shim” comment. That’s an 0.01mm flexible shim. Are you trying to say it would tear if there was stress or should be a shock absorber? Neither would be the case.
    As to the screw bending there was no sign in the outer set of screws, but the bending of the inner set made us concerned that the entire plate could be warped slightly, especially after we found could bend the metal with finger pressure.
    But we don’t know for certain, of course.

  • Mansoor Guuwa

    Can someone please explain exactly how the damage to the aluminium plate was caused.

    “When we looked at the attachment plate above more carefully, it became apparent that two of the screw holes on one side of the lens were bent downward slightly.”

    Actually the bending you showed in the photo was upward ie towards the screw heads. I know it’s mentioned that on the the holes opposite the bending is in the opposite direction, I am aware that the lens has been reassembled but i would’ve like to see the other holes to understand the forces involved.
    The aluminium plate was screwed flash against the lens with a shim between the two surface. In the assembled state the forces that bent the plate would have had to go through the shim. Also any forces, due to the front part of the lens being knocked against something, would be taken up by to 4 outer screws and not the inner 8 screws.

  • It will definitely be better the second time. But we still haven’t looked into how long the optical adjustment would take. I think it’s reasonable to think we could do the disassmbly/reassembly in 3-4 hours pretty easily. But the optical adjustment could be 30 minutes, could be another 4 hours. So too early to tell.

    But if you’re speaking how much would we charge in $$ – I’m pretty comfortable too much to be worth anyone’s while.

  • Laurent Fournier

    I am curious…how much would it take for you guys to do that same repair again? surely you learned quite a bit from the experience and the second time would be less painfull, right?

  • Marc P.

    I am no lens specialist, but a 1.2mm ring made out of aluminium seems to be a design error – and/or Sony have had missed the material stress into here, in terms of a 2ft drop. Stainless steel would have been also being cheaper perhaps then alu, i’d guess. It’ll be better for Sony to design this part out of steel then. Again, thanks a lot for this way interesting Lens Tear-down, Roger.

    For instance, i’ve dropped last december my (way old) EOS 20D, whileas taking pictures. The attached Lens was thankfully a cheap one, and got a stuck, defective zoom mechanism, but still took pictures, as a fixed “prime” now being stuck at around 35mm setting. Whileas the Canon Body only had a small scratch at the left corner, barely noticeable, it wasn’t a drop from 2 ft, but when i was standing and taking pictures at the lake here, i’ve slipped onto the mud, and felt backwards, whileas having the DSLR touching the ground with the bottom and not that heavy, the DSLR and lens came horizontal to the ground with my arm, and the 20D still shoots very well. But something into the Lens then prevented the Zoom mechanism from functioning, and when i slightly shake the setup, something is rattling inside the Lens…this is the 1st ever lens i’ve broken since 1987 for real, whileas being physical from apperance intact.

  • JJ

    These “old” A-Mount lenses are my point. They should be very robust, these are the designs everybody is asking for? And there are plenty of Minolta lenses out there wich are still working fine for a good price.
    But if someone needs pixel level sharpnes on a 50MP sports body build like a truck … I bet we still have to wait a few years more, the industry is not quite there.

  • AStarbucks

    Didn’t what you say just made it the only sensible thing to break in the middle?

    Given how heavy such telephoto lenses are, you prefer the body’s mount joint to break from the camera body?

    The whole point is that in the field, you break the telephoto lens (so stuff it in your bag), but you can still carry on shooting with other (shorter) lenses instead of having a huge hole in your camera body. Lens Rental has already confirmed that this seems to be the case with the larger telephoto lenses.

  • Just noticed in his bio that Roger likes shooting the Pentax K-1. Considering Roger’s engineering mind, to me that says a lot. I’ve been wanting to get my hands on a K-1 for a long while now… We’ll see how it goes…

  • Ralph Hightower

    It seems that lens repair was simpler in the manual focus days. But just image what lens from the 70’s & 80’s could have been with the optic materials and computer power of today.

Follow on Feedly