Is Sigma problems created by Canon?

Echo Johnson wrote in post #14912173
No

Expanding on what John from PA and gjl711 and said:

With regards to focus, imagine 0 as perfect - which, as gjl711 noted, is quite difficult to acheive. Camera and lens manufacturers can't trash every lens and camera that isn't perfect, so they build within tolerances, let's say -20 to +20.
If your camera body is -8 and your lens is +8, you'll have, essentially, a perfect combo. Even if it's -9 and +12, it will still be pretty good. But when you start getting high-value same-sign numbers, like +17 and +19 or -20 and -16, you'll find noticeable front- or back-focusing. Just remember, each copy of every lens and every body can and will be different.

So your 28-70 might be a -10 lens, which is within tolerance. Your 20D might be a +8 body, which is also within tolerance. Together, they focus pretty damn well. OTOH, your 600D might be -10. Suddenly, you start noticing that the focus isn't quite where it should be.

Note: all numbers are arbritrary and for illustration purposes only.

This nails it spot on.

No, not that one. A previous one that was better reviewed at the time. Sigma EX 28-70mm f2.8 DF aspherical. I believe it would not be compatible with the t3i...

Anyway, not big deal. I just wanted to try that baby again. My 15-85mm answer most of my needs for now.

I think you hit the nail on the head. Lenses need to be calibrated for each camera. Modern cameras have an item in the menu (sometimes hidden until Customer Services or a forum member tells you) that allow you to test and calibrate your lenses. Usually, they go back certain number of lenses.

gjl711 wrote in post #14911984
The main reason for front and back focus is that the light path used for the AF (non-live view) is not the same one that will be used to expose the actual image. Light is diverted by a mirror to a separate sensor where it is evaluated and the AF amount calculated. If there is any variance between these light paths then the AF will not be perfect.

As to why things can change, no piece of hardware can be perfectly manufactured. There are tolerances a +/- amount that is considered within specs. A perfect lens and body will both be +/- 0 but this is near impossible to achieve. If the lens is off a little to the + side and so is the body, then thing are real close but if the lens is a little to the + side and the body to the - side the differences are additive and it can be farther off the either alone. Those with micro focus adjust and FoCal, an automated MFA adjust tool, clearly see this as nearly all lenses require some adjustment to be spot on.

kendon wrote in post #14911876
afaik sigma does reverse engineering, while tamron pays for licensing.

If this is true, I had the wrong information. I do know that Tamron designed the 18-250mm f3.5-6.3 supertelephoto lens. I also know that Pentax licensed the design from Tamron to offer a similar Pentax. They are exactly the same. It could be the same that Tamron made in a Canon mount.

The lenses themselves are designed by each company, Canon, Tamaron, Sigma and such. It's the EF/EF-S interface that needs to be reverse engineered by third party manufacturers. THe EF/EF-S interface is canon specific and Canon controlled and they are under no obligation to inform other manufacturers of lenses when this interface changes.

What you are mentioning is the licensing of a lens design to the manufacturer. THat's a different thing.

Sigma problems are caused by sigma being sigma. =p Really, Sigma has turned around in the past few years and is now on a roll, releasing great lens after great lens. But it didn't used to be that way. They used to release crappy unreliable lens after crappy unreliable lens. =p

Think Hyundai. If you have ever seen one of their cars in the 80s...yeah. Generally awful, and amazing when you still see one that is running. But now? They rock.

Not that dirty filthy ROM again

i have a sigma 28-105 2.8-4 that i used on my original rebel & xt with zero problems. bought two newer bodies ( xsi w/ xs for backup) and neither body would work with this lens, so i put it on the back corner of my desk figuring i could always use manually. i also have a 40d that it worked fine on, but it came with ( ick ) 28-135 canon, so why bother to keep it in the bag as it didn't work on the other 2 bodies...

recently i put my hamds on a t3 for the higher iso, as i shoot a lot of concerts. and surprise, surprise. it works flawlessly on my t3. tried it on a mates t3 and worked perfect on his too.

go figure. i almost wrote this sigma off...

well, according to the above quoted wikipedia page and this one (last paragraph) tamron also gets the protocol by reverse engineering. i stand corrected, so far at least, as all these are still speculations. we will probably never know, as i don't think the involved parties would disclose such information.

btw, i was not implying that either reverse engineering or licensing would have any particular impact on af quality. if sigma has more problems with that (which needs to be proven better than by the amount of internet ranting about it) then it probably has some other reasons, tolerances being one of them. what echo johnson says explains it pretty good.

Sigma like any manufacturer has problems on a few select lenses. The 30mm f.4 for instance seems to have a lot of QC issues. So many returns on it that Adorama actually met with Sigma about it. Meanwhile, I have seen very few problems with the 70-200 f2.8 EX DG OS HSM, the 10-20 f3.5-5.6, the 17-70 f2.8-4 OS, or the 17-50 f2.8 OS.
Meanwhile, I've seen a lot of complaints about the Canon 17-85 and the Canon 70-300.
Bottom line is that I don't think Sigma has any more or less problems than anyone else. They are just different. They have different specs, different strengths and different weaknesses. But you can often get a comparable lens for half the Canon price and if you get a bad one you can usually send it back to get a better one. Considering the warranties of 3-5 years compared to Canon's 12 month, you really can't go wrong by choosing Sigma unless the specific lens is weak in the areas that you need to be strong.

RTPVid wrote in post #14912100
I've read that several times here, but no one offers any source of the information. Is this real, or is this POTN Legend?

As far as I know Tamron has a licenced EF protocol.
Got this information from a trusted and most likely good informed person I know.
Probably I should add that I am an image engineer with some contacts in optics industry.

This old problem with the Canon EF lens protocol on the older lenses is based on a firmware change Canon made. For electronics specialists there is a solution - see this longEF protocol thread in a German forum.

Reengineered flash devices have sometimes the same problem - for example these Sigma EF-500DG work on EOS 5D and 350D, but not 1100D as far as I remember.

I wouldn't be surprised if Canon was actually actively trying to sabotage third party lens manufacturers, by messing up with their AF algorithm. In fact, canon used to be much more blatant about that. I bought recently a sime-pro film camera Canon EOS-5, and was shocked to find out that with MF lenses (from Nikon etc., with a non-chipped adaptor) it meters with a huge error (a few stops), which was done very clearly on purpose. Secondly, with a non-chipped adaptor and MF lens, after trying to make a shot the mirror would get stuck in an open position, until you remove the battery. At that time, the camera was advertised to "only work properly with genuine Canon lenses".

Canon can only guarantee that their cameras will work with Canon lenses because Canon has no control whatsoever of other lenses made by other companies. If they made any claim other than that, then they would be responsible to make sure that anything they do to their cameras worked for every non-Canon lens as well. Instead, Canon makes what changes they want to their cameras when they feel like it and they make sure their own lenses work. If not, they may address it with a firmware update. But if someone else's lens doesn't work, that's not their problem. They aren't going to go out of their way to sabotage however because this would be blatantly obvious and would anger a lot of people who maybe don't want to buy all that extra-expensive Canon glass at which point they may jump ship to Nikon or Sony, etc. In short, competition in the market place keeps them from doing such things. But they aren't going to go out of their way to ensure backward compatibility to 3rd party lenses when they design new cameras and algorithms.