PetaPixel

Sensor Stack Thickness: When Does It Matter?

Single glass piece from the sensor stack of a Canon (left) and Micro 4/3 (right) camera. Image credit: Aaron Closz.

Single glass piece from the sensor stack of a Canon (left) and Micro 4/3 (right) camera. Image credit: Aaron Closz.

The first post I made on sensor-stack thickness wallowed deeply in PhotoGeekery. This one is meant to be of practical use so I’ll try to leave the Geek stuff out. We’ll start with the simple facts.

  1. There are several pieces of glass right in front of the sensor of every digital camera.
  2. The thickness of this layer varies from less than 1mm to slightly more than 4mm depending upon the camera.
  3. The thickness of the stack can affect the optics of a lens mounted to that camera.

There is some confusion on when this stuff matters so I’m going to attempt to accomplish two things with this post. First, we’ll do a general summary of when it might matter. Second, we’ll start a database of information that’s not readily available so those who are interested can come back to this page and find out if a certain camera-lens combination might have a problem.

rogertoy13

When It Might Matter

Testing

This is pretty straightforward.

  1. When testing a lens on an optical bench (right now this is limited to me, apparently) it may be necessary to have glass of the appropriate thickness between the rear element of the lens and the MTF sensor. If it’s not done, the measured MTF curves will be falsely low.
  2. When testing a lens using Imatest, DxO, or other computerized target analysis, as long as the lens is mounted to the camera it is designed for the proper filter stack is already in place and the measurements are accurate.
  3. When testing a lens using Imatest, DxO, or other computerized target analysis on a camera the lens is NOT designed for (testing a Leica M mount lens on a Sony A7r body, for example) the test shows accurately how the lens performs on that body but we can’t make any general conclusions about the lens. It might be better on it’s native body.

I’ll expand on number 3 just a bit. If a tester tests a Leica lens and a Canon-mount lens by mounting both to an A7r, he has created a valid test showing how those lenses perform on that camera. That is good, practical data for any A7r owner to have. But it’s completely false to use those test results and say the Leica lens is better than the Canon-mount lens. The results might be totally different on another camera, or if each was shot on the camera it was made for.

The original Metabones Speedbooster (left) compared to the two new Blackmagic Speedboosters.

Using Lenses on an Adapter

The practical importance comes when we want to use a lens designed for one camera on a different camera. (I’m assuming the adapter contains no optics itself.) Several factors come into play here.

  1. The difference in sensor-stack thickness between the camera the lens was designed for and the camera actually being used.
  2. The maximum aperture of the lens. Wide-aperture lenses are going to be more sensitive than narrow aperture lenses.
  3. How telecentric the lens is. (More specifically, how far forward the exit pupil of the lens is.) A lens with the exit pupil far away from the sensor is not affected by the thickness of the sensor stack very much. A lens with the exit pupil very close to the sensor is affected a lot.

The exit pupil is an optical phenomenon – the exit pupil is not the physical location of the rear aperture or the rear element. It can be measured, but those measurements aren’t readily available. In general more telephoto lenses have very forward exit pupils and aren’t affected by sensor stack thickness very much. Wide-angle lenses may have very close exit pupil distances. Reverse-telephoto design wide-angle lenses (SLR lenses basically) have the exit pupil more forward than Rangefinder wide-angle lenses, generally.

So in theory, a 135mm f/4 SLR lens isn’t going to care much about the sensor stack thickness. A 24mm f/1.4 rangefinder lens can be hugely affected. Here’s a theoretical example that Brian Caldwell published on Photo.net almost 7 years ago. This is calculated at 40 line pairs/mm and a 50mm exit pupil distance is fairly close, but you can see even a 0.5 mm difference in filter stack size could have an effect.

stacked1

Sensor Stack Database

The bottom line is if we want to predict what lens-camera combinations will have problems, we need to know something about sensor-stack thickness and exit pupil distances. Since it’s really hard to find that kind of information I’ve started a database here. Right now it’s pretty limited but we’ll continue to expand it as we get more information.

The sensor optical measurements are made by Brian Caldwell and measure the optical equivalent as if the sensor stack was made of glass with a 1.52 refractile index. The actual physical measurements may be a little different depending upon what types of glass were actually used.

Thanks to Shane Elen at Beyondvisible.com for all of the Nikon physical measurements. Hopefully we’ll have some more contributors soon. For right now, please remember most the physical thickness measurements are single measurements (if there are two confirming measurements, I’ve made the number bold). Until we get several repeated measurements for each sensor-stack, take them as a suggestion, not an absolute fact.

Screen Shot 2014-06-10 at 9.19.50 AM

Exit Pupil Database

The location of the exit pupil, along with the aperture of the lens, predicts to a large degree how much effect a difference in sensor stack might have. Brian was kind enough to prepare another theoretic graph of how much difference the exit pupil location makes.

stacked3

We’ve been able to get exit pupil distances for a few lenses, which I’ve listed in the database below. The data in this table is largely from Brian Caldwell and Joseph Wisniewski, with a few from manufacturers publications where I could find it.

Screen Shot 2014-06-10 at 9.21.13 AMScreen Shot 2014-06-10 at 9.21.38 AM

Before you strain your eyes looking, the data many of you want, exit pupil distance for M-mount lenses, isn’t listed yet because I don’t have it. You can assume the wide-angle M-mount lenses will have shorter exit pupil distances than any of the SLR lenses, but this is the best I can do for right now.

Conclusion

I know at this point, most of you want to ask about this camera or that lens. I’m sorry; this is all of the data I can find. We obviously need a lot more. Some of you are skilled enough to make these measurements, or may know sources I haven’t been able to find. So if you have access to any of this data, please forward it and I’ll add it to the tables we’ve started. I’ve written some manufacturers to see if they’re willing to share numbers with us (but I’m not holding my breath).

Of course, Brian and I are going to continue to make more measurements and will add them as we do. But it isn’t a project we can do in a day or two. For physical measurements to be really accurate we have to destroy the sensor, so I’ll be doing that as cameras fail and aren’t repairable. (If any of you have a dead camera you want to donate to the measurement cause, send it to me and I’ll send you back the pieces :-) )

Over time I hope this database becomes quite large. We’ll leave it here so it remains available to anyone interested.


About the author: Roger Cicala is the founder of LensRentals. This article was originally published here.


 
Get the hottest photo stories delivered to your inbox.
Get a daily digest of the latest headlines:
  • Ken Elliott

    Thanks, Roger. As a side note, I noticed certain Nikon lenses were exceptional performers when used with a Sony NEX-EA50 video camera (APS-C sensor) while others did not. As an example, I expected the Nikon 24mm f/1.4 G to outperform the 14-24mm f/2.8 G, as it does on the Nikon body. But test footage and real world use shows the 14-24mm zoom @ 24mm to outperform the 24mm prime. I had not been able to understand why this was the case, until I read your article. The 24mm prime has an exit pupil of 75mm, while the zoom at 24mm has a 96.5mm exit pupil. Your list seems to match my experience – lenses with 70-100mm exit pupils perform well on the Sony, but shorter ones suffer from lower performance that I see on a Nikon body.

  • http://www.fotto.be Otto Van De Steene

    very interesting. yet… one a few questions pop up in my mind.

    1) what about lens manufacturers that make lenses in different mounts. I especially think about the sigma’s, samyangs, and tamrons in this world. I can’t imagine that they will be optically different as some of them can even interchange the mount as after service.

    2) I imagine that a sony a lens with sony convertor on a e-mount camera is the exception?

  • Thomas

    You can tell this is a high precision optics lab because they have that big hammer nearby. :)

  • Mike

    I wonder how old lenses perform on digital cameras, considering they were developed for film, without any filter stack.
    The Helios 44 series, for example, is very popular with EOS bodies (in part due to the very simple adapter required), but are we getting maybe a lower IQ because of these filters?

  • Roger Cicala

    I think for the majority of Nikon and Canon cameras, the stack seems to be about 2mm, which probably makes it fairly easy to use the same lens on both, just changing the mount.

  • OtterMatt

    Every good geek knows the three questions of engineering: What is it, What does it do, and How do I break it?

  • hdc77494

    I think the people who might most affected by this are video shooters using older manual primes on Canon dslrs. I’m really curious if you can run into the same problem with a Red Dragon, the Canon C series, or other video cameras.
    I have some Zeiss Contex manual primes I’ve used on my t2i, but I haven’t tried them on my 5D3. If sensor stack is a serious issue, why haven’t we heard about it before, and how are so many successfully using old primes every day?