Polarizer for Mitutoyo

[hero]

I meant #1, with the polarizer in infinity space, but as far from the objective as possible.

So, I've been following this thread with some interest, and this raises an interesting if not entirely related question. For an infinity-corrected objective, my understanding is that the exit rays from the objective are parallel, in which case, the space between objective and tube lens is arbitrary ("infinity space" as you put it). But then, why is it that there are recommendations for this spacing for particular implementations; e.g., Mitutoyo objectives and Raynox tube lenses? Is it because the actual focal length of the tube lens is not exactly 200 mm? I'm trying to wrap my head around the optics and this is one of the sticking points in my mind.

For what it's worth, I do understand that the spacing between tube lens and sensor is related to the actual focal length of the lens, and that reversing the lens does not necessarily maintain this spacing (due to telephoto ratio). But I could swear I've seen recommendations for the spacing between objective and tube lens.

[rjlittlefield]

For an infinity-corrected objective, my understanding is that the exit rays from the objective are parallel, in which case, the space between objective and tube lens is arbitrary ("infinity space" as you put it). But then, why is it that there are recommendations for this spacing for particular implementations; e.g., Mitutoyo objectives and Raynox tube lenses? Is it because the actual focal length of the tube lens is not exactly 200 mm? I'm trying to wrap my head around the optics and this is one of the sticking points in my mind.

This is a common confusion. The explanation is that most of the exit rays are not parallel to the optical axis. Instead, each point in the field produces a bundle of rays that are all parallel to each other. The bundle for the center point of the field is parallel to the optical axis, but the bundles for every other point are angled away from it.

Here is a ray-traced diagram for an idealized 10X NA 0.25 objective (20 mm focal length, 5mm radius), separated by 120 mm from a 200 mm tube lens. It shows the bundles for the center and corner points of the field with an APS-C sensor.



With this large separation, the outer rays for the bundle in the corner of the field are striking the tube lens at about 12 mm off axis. If the field were larger, say with a full-frame sensor, the angles would be steeper and the off-axis distance would be even more.

You can see in this diagram why physical separation and position of the entrance pupil for the tube lens are both important. The farther back the entrance pupil is located inside the lens, the farther off axis the corner rays will be when they get there. Traditional telephoto lenses tend to vignette because their entrance pupils are positioned far back. Some designs vignette worse than others, even with the same pupil diameter, because the pupil is located farther back with the troublesome ones.

--Rik

[hero]

Very helpful to see this diagram, thanks! So if I interpret this correctly, for a given infinity corrected objective and tube lens, with the tube lens fixed at infinity focus...

(1) changing the space between objective and tube lens will not change magnification, but if the spacing is too large, the marginal rays could vignette.

(2) reducing the space between objective and tube lens should reduce aberrations caused by the tube lens, because the rays exiting the objective will enter the tube lens closer to the axis.

(3) adding an optical flat between objective and tube lens should not affect magnification or aberration, since it would only displace the ray bundle along the axis, and not their parallelism.

(4) therefore, for a real-world setup, a general rule of thumb is to keep the objective and tube lens spacing small.

Did I get this right?

[rjlittlefield]

Did I get this right?

Alas, if only the world were that simple...

Point (1) looks completely correct.

Point (3) is correct for all practical purposes.

Point (2) is not necessarily correct, because the tube lens may be optimized for off-angle rays to strike the lens at specific off-axis distances. For example, one of my early tests of the Thorlabs ITL200, using only a distant target and an 11 mm aperture to mimic a perfect infinity objective, showed that lateral CA was reduced when the aperture was located a significant distance in front of the tube lens. You can read that report HERE.

Point (4) is therefore not correct either. The optimum spacing unfortunately depends on the particular type of tube lens that you have. It probably also depends on the particular objective that you're using it with, since whatever aberrations are introduced by the tube lens can either add or subtract from aberrations in the objective itself. And finally it depends on the criteria that you're using for "best", because frequently some aberrations will get better but others worse as separation increases.

--Rik

[hero]

Rik, this is why your combination of theoretical understanding and real-world experience is so invaluable, because it's easy to look at the idealized ray diagrams and forget that the geometric optics don't necessarily make these subtleties apparent; conversely, real-world testing can be a bit like looking at black-box output (perhaps a bit more literally "black box" in the case of cameras!), and without the underlying theory, these observed results can be rather opaque to explanation.

But the consequence of this information is that...well, adjusting the spacing (both in front and behind) for a Raynox setup is going to be a pain, isn't it? Someday (when I grow up and become a real photographer) I'd like to do some cross-polarization photography at 10x, but for now I'll be happy to just cobble something usable together!

[rjlittlefield]

... well, adjusting the spacing (both in front and behind) for a Raynox setup is going to be a pain, isn't it?

Going for "the best" is always a pain, in the pocketbook if nothing else.

The secret to happiness, at least for myself, is being able to reach "very good" and then stop and ask again what's the most important thing to be doing, given whatever resources are available.

In the case of tube lenses, I have on hand all the usual suspects to choose from (Raynox, Mitutoyo, Nikon, Thorlabs). But when I want to take some pictures I almost always reach for the Raynox because I know it works very well for all the criteria I care about. For me, those criteria include being easily obtained, easily configured, and giving good image quality corner to corner on all sizes of sensors up to fullframe, so it makes a "known good" solution that I can comfortably recommend to other people without having to agonize about the specifics of their equipment. Someone with other priorities, for example absolute minimum CA on APS-C, will make a different choice.

In your case, perhaps my best suggestion would be to just relax and assume that everything will be fine until shown otherwise. If your Raynox is focused roughly on infinity, and your tubes or whatever are properly flocked, and you're using a decent polarizing filter, then your images will look fine even at 10X. See for example the discussion at http://www.photomacrography.net/forum/v ... hp?t=37664. The images there are only at 2.5X, using a Mitutoyo 5X on 100 mm tube lens, but you can see that there were a lot more interesting aspects than agonizing over whether one setup might be a hair sharper or less CA'd than another.

--Rik