Tube lens tests on D800E full frame

[aidanmoore]

Hi,

Thanks for the data...very interesting.

AS a D800 owner, this is relevant to my setup with a X10 Nikon objective mounted on a Nikkor micro 200 lens.

Can I ask what your opinion is of the performance of, say, the Raynox vs the Nikkor 200mm f4 micro lens used as a tube lens?

Thanks again.....

[ChrisR]

Welcome Aidan
There aren't many rigorous tests of such things.

One thing which matters though, is which Micro Nikkor 200mm. There was one before the current ED version which was not stellar, and at least one member has a current one he's not entirely happy with though Nikon have declared it to be "in spec".

Regarding separations, if you move too far you vignette the image, but some (eg Raynox) lenses show a small improvement if you allow around 50mm. This doesn't apply to camera lenses though there can be reflections bouncing between the glasses which become more noticeable when the gap is small.

[Charles Krebs]

Rodrigo

I find the massive amount of longitudinal CA in the images where Nikon objectives are paired with the Raynox lens intriguing.

... but it would seem like Nikon is offloading some of its CA correction onto the tube lens (I believe Zeiss does that as well).

... Also, despite being "infinity corrected", the objective/tube lens combination works a bit better when they are spaced like a Keplerian telescope, which would mean that the objective should sit ~200mm away from the tube lens (the back focal plane of these objectives is usually inside the lens itself). Has anyone tried doing that?

Nikon, Olympus (current UIS, UIS2), and Mitutoyo do not perform addition color correction in the tube lens. Leica and Zeiss do. Keep in mind that the Nikon 10X BE objective shown in the tests is a very low cost achromat, while the Mitutoyo is Plan Apo. I think that is the primary reason for the high longitudinal chromatic aberration. (This is actually a somewhat less of a problem than it might appear when doing image stacking since you are using primarily the in-focus section of each source image where it is less noticeable).

Nikon recommends 100-200mm from objective to tube lens with the MXA20696. Olympus recommends 50-170mm separation. For some time I used a little over 100mm separation with the Nikon tube lens, but lately use it with much less than 100mm. I have not (but should!) do a comparison test, but nothing jumps out at me as being any different as far as image quality is concerned (on APS-C sensor).

If you mount onto a "regular" telephoto camera lens (as opposed to a microscope tube lens or the Raynox) you need to mount it very close to the front element or you will typically get vignetting.

[rguerra]

Hi Charles,
Thanks for clarifying that. I hadn't realized that the Nikon objective were not APO's. I would have expected the correction to be better, but then again I'm used to monochrome cameras.

Vignetting will definitely be a problem if you place the ITL200 or the MXA20696 200mm away, since their clear aperture is ~30mm, though I imagine the Raynox may be able to do better. The images in the beginning of the thread don't show anything egregious, but I would expect less coma and astigmatism when placed at the "correct" Keplerian spacing. However, since the back pupil of the objective is so small, it's entirely possible that you can get away with placing the lenses closer together without too much trouble.

-R

[Lou Jost]

This is a thread well worth reviving, since all of us who work with infinity objectives face the "tube lens" problem, and the prospect of a $70 Raynox solution is very attractive!

Rik, in your set-up for the tests here, you didn't mention whether you used the Raynox facing forward or reversed. Do you remember which orientation you used? [Edit Dec 30 2016-- I see you did mention at the outset that you are using it in reverse. Sorry, I missed that.] And do you remember the separation you used between the Raynox and objective? I think you usually put an iris between them, right?

[Macro_Cosmos]

Sorry to necro this thread. I figured that rather than creating my own, I might as well voice my grievances here.

I just went through some crude and primitive testing of the Thorlabs ITL200 tube lens I acquired from some bankrupt business. Compared it to a 200mm f/4 ai-s (not micro-nikkor). My 200mm ais is optically superb, went through several copies to acquire one without any issues.

(F=female M=size of threads or male)
To adapt the ITL200 to my D810, I used several tubes:
1. Set of Nikon extension tubes. These aren't the Chinese cheapos, these are the made in Japan, rather expensive ai-s tubes.
2. F>T2
3. Custom T2 adapter made for the itl200. It has threads that allow positioning of the tube lens internally
4. Russian T2>M42 coupling tube (F-F)
5. Various M42 tubes
6. M42>M26
7. Mit 10x apo
8. Resolution test target

The problems I have encountered are: (refer to numbers assigned above)
1. The made in Japan supposedly official tubes exhibit some wobble
2. This cheap Chinese adapter is abysmal. It wobbles A LOT.
Due to 1 and 2, the optical axis is thrown off (optical axis of the entire system, meaning the tube lens, the objective, and the surface of D810's CMOS, I can't think of a better word, there's a technical term but I just can't pull it out of my drowsy brain).

3. This tube adapter is relatively well made, the tube lens does wobble slightly inside however. I adjusted the tube lens for infinity and tried to get the best results possible. The distance from ITL200 to the surface of the sensor should be 148mm.

8. Getting the test target completely parallel with the objective was also an annoyance.

The entire system was just a failure. There's just too many factors that are out of control, and precision is a detriment to using the ITL200 on a camera. In addition to that, the ITL200 was NOT designed for consumer cameras, which explains why the corners are literally unusable. This tube would be better off on a crop body.

If all that is not enough, there's another problem. The optimal distance from the back of the objective to the ITL200 is anywhere from 70 to 210mm. This is a massive range. I was wondering why, so I consulted a specialist. His explanation was simple:
1. It's mass produced, hence cheap
2. The big range is playing hit and miss, which means the ITL200 isn't a dedicated tube lens, it's rather a "try to fit it all" solution. An easy example is a zoom lens, 28-300mm. It can do a bit of everything, but it's essentially good at absolutely nothing (or rather being a Swiss army knife it its only redeemable quality).
3. Precision matters. Adapting this weird tube to any camera system will bring numerous complications.

The only solution to all these problems I can think of, will simply be to invest in Thorlabs' SM1 tube system. I made 4 adapting designs, and one of them is based purely on the Thorlabs SM1 tube system. How much does it cost? Well, about $400. Using a cage system will bring more stability, which will add another $200. Too much for a uni lad who lives on one meal per day.

My other designs are using purely M42 tubes, hybrid of M42 and SM1, hybrid of M42 and F (which is the one I went with due to being cheap). And yeah, I get what I paid for.

Regarding to pictorial quality, which is what most people are ultimately interested in, it's well, inferior to the 200mm ais (my setup involving F-mount tubes and M42 tubes):
1. It has more LoCA than the 200mm ais
2. Resolution is a tad lower
3. They are equally sharp though
4. 200mm ais adapting is intuitive. Tube to infinity, M52>M26 done. ITL200 requires work, it is also longer
5. Horrible defocus elements. Gross.
6. Corners are unusable. Expected since this was designed for a microscope setup, not consumer cameras.

Summing up, unless one is ready to spend the money, time, and effort to invest in a precise tube system, don't bother with the ITL200. It might not even be that good compared to a 200mm ais lens. I have no way to verify it currently.

This begs another question, I'm genuinely curious. Rik, how did you set yours up? My results were slightly worse than what can be seen in the comparison.

As for the ITL200 and its associated tube, I might list it in the exchange section, or maybe wait till I'm bothered and cashed up enough to import thorlabs parts for a better tube setup.

This is what the setup looked like:
ITL200 Test
Test target not shown, I started with a piece of paper and took this shot.

[Macro_Cosmos]

Update: I just drafted up another adaption method which avoids the problematic tubes whilst being somewhat reasonable in terms of cost. Maybe I'll give it another go.

[Lou Jost]

I'm curious, why not stick with the simple and economical camera lenses/Raynox/LSA as tube lenses? It is easy and quality is higher than with the dedicated tube lenses.

[rjlittlefield]

If all that is not enough, there's another problem. The optimal distance from the back of the objective to the ITL200 is anywhere from 70 to 210mm. This is a massive range. I was wondering why, so I consulted a specialist. His explanation was simple:
1. It's mass produced, hence cheap
2. The big range is playing hit and miss, which means the ITL200 isn't a dedicated tube lens, it's rather a "try to fit it all" solution. An easy example is a zoom lens, 28-300mm. It can do a bit of everything, but it's essentially good at absolutely nothing (or rather being a Swiss army knife it its only redeemable quality).
3. Precision matters. Adapting this weird tube to any camera system will bring numerous complications.

I'm guessing that your specialist did not take a lot of time to think about this problem.

The range is large simply because it's the length of an "infinity section". Within that section, when the tube lens is at the correct distance from the sensor, the image of the focused subject consists of beams of light, one beam per point on the subject, where all the rays within each beam are parallel ("focused at infinity"). Because of this very special focus arrangement, the only effect of changing distance between objective and tube lens is to alter how far off axis each beam strikes the tube lens. It is a very different and much more tolerant situation than any case where the light is either converging or diverging. All of the tube lens manufacturers quote similarly broad ranges, for the same reason. Mitutoyo even goes so far as to specify that the separation can be from zero to whatever causes vignetting.

Rik, how did you set yours up?

Let's see, June 2014, comparative test of lenses needing different extensions... That was undoubtedly using an Olympus bellows with an Olympus bayonet-to-M42 plus other screw adapters on the front, and in the rear a modified mount that locks into the bellows and presents a bayonet for mounting the camera. In general I avoid bayonet extension tubes because they inevitably get floppy beyond a certain amount of weight, whatever is needed to overcome the springs in the receptacles. In theory that can happen at the front of the bellows also, but those springs are pretty strong in the unit that I have, so separation has never been a problem.

--Rik

[Macro_Cosmos]

The range is large simply because it's the length of an "infinity section". Within that section, when the tube lens is at the correct distance from the sensor, the image of the focused subject consists of beams of light, one beam per point on the subject, where all the rays within each beam are parallel ("focused at infinity"). Because of this very special focus arrangement, the only effect of changing distance between objective and tube lens is to alter how far off axis each beam strikes the tube lens. It is a very different and much more tolerant situation than any case where the light is either converging or diverging. All of the tube lens manufacturers quote similarly broad ranges, for the same reason. Mitutoyo even goes so far as to specify that the separation can be from zero to whatever causes vignetting.

Thanks for your detailed explanation! Since the separation varies, does a so called "sweet spot" exist? I'm thinking that varying the objective-tube distance won't change performance in chromatic aberrations. I did a little bit of reading, and this distance is called "infinity space" by many manufactures. It allows other components to be inserted in the system, such as a beam splitter. I looked up some formulas and did a little bit of crunching:

Focal length of the Mit 10x is 20mm, NA=.28
Exit pupil d=2*20*.28=11.2mm
24mm is width of an FX sensor, itl200 has an f of 200mm and an entrance pupil of probably 20mm. Thorlabs' website says not available.
200(20-11.2)/24=73.333...
This leaves with such a figure, in the 70-170 range but quite a tight one I must say. I hope my calculations are correct here.

Let's see, June 2014, comparative test of lenses needing different extensions... That was undoubtedly using an Olympus bellows with an Olympus bayonet-to-M42 plus other screw adapters on the front, and in the rear a modified mount that locks into the bellows and presents a bayonet for mounting the camera. In general I avoid bayonet extension tubes because they inevitably get floppy beyond a certain amount of weight, whatever is needed to overcome the springs in the receptacles. In theory that can happen at the front of the bellows also, but those springs are pretty strong in the unit that I have, so separation has never been a problem.

--Rik

I agree here, a high quality bellows is preferable to flimsy tubes. The M42 tubes I have are cheap, and there's no internal fine threading to eliminate ghosting and the such. It would be crucial to flock them in my opinion. Since I sold off all my bellows, I'm stuck to tubes. Can't find a good PB-6 as of now. I'm thinking of something like this: http://www.reallyrightstuff.com/Lens-Su ... ck-Release
There's many good Chinese ones available for a fraction of the price. Hejnar makes them too.

Really disappointed that my ITL200 setup doesn't perform as well as the 200ais. Reflecting on your "parallel lines" statement, at least in theory, the ITL200 should perform better than the 200ais.

[enricosavazzi]

Some potentially useful information, paraphrasing from Seward 2010, Optical design of microscopes (SPIE Press), page 10:

The tube lens typically can be moved along its axis by up to 10% of its focal length without introducing significant aberration.

I interpret this as the tube lens moving relative to the eyepieces/sensor plane, since we know that the distance between tube lens and objective is not critical.

The text does not elaborate on what constitutes acceptable aberration, or what types of aberration are introduced by doing as described. It also says nothing of whether high-NA objectives are more sensitive to this displacement.

[Lou Jost]

I had tested various distances between coupled lenses in the past, and both astigmatism and lateral color aberrations changed slightly with spacing. They changed in opposite directions (one got worse while the other got better) with increasing distance. I suspect that there is no universal best distance; it will depend on the aberrations of each particular lens combo.

[rjlittlefield]

Since the separation varies, does a so called "sweet spot" exist?

Yes, but as Lou Jost points out, various aberrations can change in different directions, so the location of any sweet spot can be pretty vague and depend on your specific criteria.

I'm thinking that varying the objective-tube distance won't change performance in chromatic aberrations.

Well, for off-axis points changing the objective-tube distance changes which radial zone of the tube lens gets used. The farther out they get, the less well corrected the lens is likely to be, for the same reasons that an ordinary lens usually degrades if you open it wide up. The big difference is that in a tube lens setup, the variability of aberrations off-axis is greatly reduced because the lens remains stopped to typically f/18 or f/20 for all beams, no matter where they hit the lens. But I would not like to bet that any property remains exactly the same.

I did a little bit of reading, and this distance is called "infinity space" by many manufactures. It allows other components to be inserted in the system, such as a beam splitter.

More specifically it allows planar components to be inserted. This is the big reason why modern designs use the infinity design. With finite objectives, added components always had to be precisely non-planar, if they were to have any chance of avoiding added aberrations. With the infinity design, the added components still have to be precise, but precise and planar is much simpler to achieve.

200(20-11.2)/24=73.333... This leaves with such a figure, in the 70-170 range but quite a tight one I must say. I hope my calculations are correct here.

Probably should use 28 mm diagonal, not 24 mm frame width, so 200*(20-11.2)/28=62.9 and it's a lost cause. Fortunately that 70 mm minimum distance is a matter of specification, not reality. In fact the lens works fine with much shorter distances too. Remember the Mitutoyo spec.

a high quality bellows is preferable to flimsy tubes.

On the other hand rigid tubes beat bellows in a lot of respects. Yes, flocking is important. See Raynox DCR-150 tube assembly with flocking for a setup with both.

Reflecting on your "parallel lines" statement, at least in theory, the ITL200 should perform better than the 200ais.

I'm missing something in that line of thought. The tube lens is just looking at a virtual object located at infinity. I assume the 200ais is able to take a good landscape shot. The only thing much different from that when used as a tube lens is the aperture location. If any telephoto does not vignette when used as a tube lens, and it gives a decent landscape image when opened just enough to avoid vignetting as tube, then I'd expect it to make a good tube lens also. The big place where ordinary telephotos fall down as tube lenses is that they tend to vignette, particularly when used with full frame sensors.

Some potentially useful information, paraphrasing from Seward 2010, Optical design of microscopes (SPIE Press), page 10:

The tube lens typically can be moved along its axis by up to 10% of its focal length without introducing significant aberration.

I interpret this as the tube lens moving relative to the eyepieces/sensor plane, since we know that the distance between tube lens and objective is not critical.

The text does not elaborate on what constitutes acceptable aberration, or what types of aberration are introduced by doing as described. It also says nothing of whether high-NA objectives are more sensitive to this displacement.

That section of Seward's book is not very satisfying. Its treatment of telecentricity is misleading at best, since the aperture position that it requires is not one found in any microscope I know of.

But anyway, another way of looking at the problem is that an infinity objective + tube lens is pretty similar to a finite objective by itself. High NA objectives will definitely be more sensitive than low NA, for the usual reason that changing the focus distance in the back will change focus distance in the front, leading to spherical aberration in the objective. But even so, the standard graph for tube length tolerance, at http://www.science-info.net/docs/etc/Tube-Length-na.gif , shows that 10% variation in a 160 mm tube is acceptable up to about NA 0.5 .

I suspect that there is no universal best distance; it will depend on the aberrations of each particular lens combo.

I share that suspicion, but I'll let somebody else demonstrate it experimentally. It seems like quite a lot of work for not much gain.

--Rik