optics for low distortion for 0.5x-1x-objective/macro/te

[jimduk]

I am looking for advice to help understand the best simple/off the shelf optics for 0.5x-1x imaging. Application is measurement (apologies it's not aesthetic photography). Key question is - is there anything special about 1x and where do the trade-offs between macro, telecentric and microscope objectives come in?

Broadly speaking I am trying to get a good low-distortion monochrome image from a planar surface with a mathematical pattern with features in the 10-30 micron range. The imager is 1/3" 3.75 micron, 1280 x 960 pixels. Pattern is exactly (<0.5 degree) parallel to sensor and optics. I would like the DoF to be 0.2-1 mm, and a short working distance (10s of mm) and overall distance from the image plane to the object plane to be compact if possible (e,g, 50-100mm preferable to 200-300 + mm). A lighter, cooler system is preferable to a heavier, warmer one

There seem to be three or four options - microscope objectives, machine vision macro lenses, machine vision telecentric and then maybe mobile phone based optic (reversed or otherwise messed about)

Machine vision telecentric looks like it fits the bill https://www.edmundoptics.eu/p/10x-silve ... ens/15293/ but is expensive and seems like overkill. Advantage - seem to get slightly better DoF

For microscope objectives, getting low mag & the NA for the depth of field seems like the challenge and also the lenses are mostly infinite. For 1mm DoF the calculation is roughly NA = 0.03 (bad approximation of 700nm/(0.03)^2) . This thing from Thorlabs looks good https://www.thorlabs.com/thorproduct.cf ... r=TL1X-SAP but again is very pricy and a bit bulky. It seems like a simple finite conjugate would work but most objectives seem infinite(I have a 2x finite conjugate compact from Edmunds https://www.edmundoptics.co.uk/p/2x-0-1 ... ive/38498/ theoretically I think I'd be OK with this reversed and in NA closer to 0.05 - are these available - I've seen cheap 0.7x finite lens to add to microscopes, but don't really understand if they would work - they have no NA details etc. https://bolioptics.com/0-7x-auxiliary-o ... cope-48mm/)

Finally there are macro lenses from people like lensation - this https://www.lensation.de/product/MC4M-025-194I/ has NA 0.04, good distortion, working distance is a bit longer than desired but probably works

Small M7/M8 lenses seem to work a bit, but hard to get quality specs for them and the optics people I talk to point me towards larger lenses - don't know if this is historical or good advice.

Obviously given the forum I could mount a photography macro lens - benefit is price/availability but downside is size/weight.

I suppose my question is - a) for 0.5-1x mag, 0.2-1mm Dof, low distortion and shorter image plane/object plane distance is there a strong reason to prefer telecentric vs finite microscope vs infinite microscope vs macro
b) What's the simplest thing that might work? is getting a single lens ( or doublet/triplet) and building a system say with f = 20mm and having image plane at 2f, object plane at 2f with lens in the middle a viable approach and where to buy the lens from?

(also happy to engage a specialist if needed)
Jim

[Lou Jost]

I'll just say that if your subject is perfectly flat and perfectly parallel to the sensor plane, then there is no advantage to a telecentric lens.

[enricosavazzi]

Putting your system in perspective, the 1/3" sensor has a size of 4.8 by 3.6 mm. This is much smaller than the cameras normally discussed in this bulletin board. Although macro lenses for system cameras are usually of good quality, they are often designed for much larger sensors with larger pixels. In spite of the relatively low resolution of your camera, its pixels are a bit too small for most macro lenses to resolve satisfactorily.

Finite microscope objectives are also not optimal for working at 0.5x to 1x. With relatively few exceptions they are designed for magnifications 4x and higher. They are also designed to cover image circles larger than your sensor (typically between roughly 15 and 25 mm).

Infinity corrected microscope objectives can be used with a small sensor by using a tube lens of focal length much shorter than the standard focal length (which is usually 200 or 180 mm). A 50 mm focal length might be adequate to produce a sufficiently small image circle. The tube lens needs to be of sufficiently good quality for the 3.75 micron pixels, which is a similar pixel size as current Micro 4/3 cameras. This therefore seems doable.

Machine vision macro lenses specified to provide at least full HD resolution might be an option, but I know too little about them to make any suggestion. Years ago I tested a few zooms with macro capabilities designed for machine vision, but my results were rather poor. These lenses were apparently designed for quite low image resolution. Perhaps things are better now, with full HD videocameras relatively commonplace.

Mobile phone lenses are designed for autofocus by the phone electronics, and difficult to use without these accompanying electronics. These lenses are also generally not designed for low geometric distortion, as well as not designed for the required range of magnification.

[Lou Jost]

There are some tiny fast Cinegon lenses made for small image circles that might be usable.

[jimduk]

Thanks for the replies - really appreciated. The Cinegon lenses look nice, though they seem to be at 0.1 x mag which isn't where we want, but might be able to make work.

The reason we are on a 1/3" sensor with 3.75 micron is our first prototype was with a cheap 1MP USB microscope (kind that costs £25 on Amazon) which didn't work too badly and had a similar imager. There was distortion at the edges but in general not too bad. They have a longish 100mm + body with the lens going up and down in the middle (fat plastic one piece - see https://youtu.be/Bjn8QxYUd_A?t=210) so we felt we could make that or something similar work.

If we needed to go up a bigger pixel size we could - but we only need approx 1k x 1k pixels, and there don't seem to be too many options at that size for modern, available sensors( e.g. could go 4.8 micron). We could do 1" sensor with 2k * 2k and bin the pixels to get 9.6 micron, but bigger sensors are also more expensive and produce more heat, but that might be a trade-off worth making.

Will look at the infinite corrected objective with a smaller tube lens option - thanks. Also some of the machine vision cameras are certified for 5-10MP so will keep looking at them.

Dumb question - what's wrong with turning a 0.5 x finite microscope objective the wrong way round? I'm not an optics person, but presumably these systems are reversible, though I suppose the object and image NAs may not be the same.

Thanks again for the advice - still learning

[enricosavazzi]

[...]
Dumb question - what's wrong with turning a 0.5 x finite microscope objective the wrong way round? I'm not an optics person, but presumably these systems are reversible, though I suppose the object and image NAs may not be the same.

Nothing wrong in a literal sense, but most finite objectives are designed for a tube length of 160 mm. To make it work reversed at the optimal design distances, the subject would have to be placed about 150 mm from the reversed objective, and the sensor just a few mm away. NA would also change.

[rjlittlefield]

jimduk,welcome aboard!

Telecentric optics are recommended for measuring applications because they allow measurements to be made accurately even when the subject-to-lens distance may vary slightly. The optical DOF of a telecentric lens is identical to that of an ordinary lens with the same NA, but the telecentric lens is usable (can meet accuracy requirements) over a wider range of focus. If you can precisely control the subject-to-lens distance, then you don't need telecentric optics.

Your requirements are unusual: small subject, small sensor, and short working distances.

One approach that comes to mind is to place two infinity objectives back-to-back, with an infinity section in the middle. For example, placing two 4X NA 0.10 objectives back-to-back will give you a 1:1 system with the same field and sensor sizes that the objectives would normally have, so roughly 6-10 mm diagonal.

In that case, used at 1:1, the system would also have zero geometric distortion because it would be completely symmetric. Used with any other conjugates, you would probably get field curvature and other distortions.

Pairing other combinations of objectives such as 4X and 2X to give 1/2X might work OK.

Mounting will be non-trivial, but basically it's just a matter of providing an intermediate plate into which both objectives can screw.

--Rik

[jimduk]

Thanks

One approach that comes to mind is to place two infinity objectives back-to-back, with an infinity section in the middle. For example, placing two 4X NA 0.10 objectives back-to-back will give you a 1:1 system with the same field and sensor sizes that the objectives would normally have, so roughly 6-10 mm diagonal.

This is really interesting - so potentially I use two of these https://bolioptics.com/4x-infinity-corr ... ce-17-3mm/ (or similar) back to back and I've got a system at a fairly short length, good working distance for a reasonable price. Mounting shouldnt be an issue. The NA here is 0.1 which gives 70 microns depth of field theoretically, still a bit tight but may work. Seems like there are also longer range objectives with smaller NA ( mitutoyu 2x @ £800, but more expensive (don't understand optics pricing) or this https://www.thorlabs.com/thorproduct.cf ... r=TL1X-SAP but spending £3k on two 1x lenses seems a bit painful, though the NA is good

Thanks again

[enricosavazzi]

[...]
Seems like there are also longer range objectives with smaller NA ( mitutoyu 2x @ £800, but more expensive (don't understand optics pricing) or this https://www.thorlabs.com/thorproduct.cf ... r=TL1X-SAP but spending £3k on two 1x lenses seems a bit painful, though the NA is good

The Mitutoyo M Plan Apos are expensive, and good when you need a long working distance and a high image quality. In your case, image quality is a requirement but not a long working distance, as far as I can see. Infinity objectives with a shorter working distance are likely going to cost you less.

Another interesting property of paired infinity objectives is that the second objective, at least in theory, cancels out the geometric distortion introduced by the first objective. This is of course true only if the two objectives are identical and the combined magnification remains 1x.

The principle of pairing two identical infinity objectives of the same type and magnification back-to-back yields a combined 1x magnification regardless of the magnification of the objectives. However, the image circle on the subject side of these objectives does have a relationship to objective magnification. You should weigh up your requirements and constraints when choosing a magnification for a pair of infinity corrected objectives.

In practice, a 2x Mitutoyo M Plan Apo has an image circle on the subject side of at least 11 mm (these objectives are typically used on Mitutoyo microscopes with 10x eyepieces with a 22 mm field stop, which limits the image circle visible to an observer), while a 5x objective has an image circle on the subject side of at least 4.4 mm. So a pair of 5x objectives would seem to be a better match for your sensor size than a pair of 2x in terms of utilized circle of view. Paired 5x would also give you a higher NA but a lower DoF. With your sensor, however, the higher NA would likely give you no advantage in captured image quality, and the loss of DoF might be objectionable for your purposes.

[Lou Jost]

The idea of back-to-back stacked lenses (one reversed on the other) is a very good one. I've done that with ordinary lenses, with good results. The Mitutoyos, though, seem very much like overkill, and this is one of the few cases where long working distance is a disadvantage. You mentioned that the photos will be monochorme. Does that mean the light source is also monochrome, or just your sensor? If the light source is monochrome, you would eliminate the possibility of chromatic aberrations, so a cheap achromat migfht give you results just as sharp as an expensive apochromat like the Mitutoyos, or the super-apochromat Thorlabs objectives. Most of the high price of those objectives is due to their color correction using exotic glasses and complex lens formulas. You don't need any of that. In fact the lens complexity may be a disadvantage since you want compactness.

[jimduk]

The sensor is monochrome, the target is monochrome and if needed we can control the illumination with a strobe, so it sounds like that is something we can & should do. We can also lose the edge ( keep centre 80% ) if needed. We want low distortion/ planar projection (probably below 0.1%), so that's why plan achromat is good but we don't need the achromat bit.

For cheaper objectives this https://bolioptics.com/2x-infinity-corr ... ce-18-3mm/

seemed possible at $70 USD, as Enrico pointed out 4-5x might be better ( image circle), but this has a lowish NA ( and also if illuminating we need a bit of working distance if not going coaxial). Can't tell how you tell the quality of these, but at this price I think it's just a case of buy them (or similar) and try it out. The machine vision macro lenses which may work are in the EUR 500-1000, so this would be very cost effective.
Thanks for all the help

[Lou Jost]

Of course the incident light has to be monochrome too, even if the target is black and white. Best way to do that is lasers with milk de-specklers; but it is almost certainly good enough for your application to just use a green-emitting LED (though you may still need de-speklers).

Milk despecklers:
https://www.photomacrography.net/forum/ ... 962d29af10

[Macro_Cosmos]

jimduk,welcome aboard!

Telecentric optics are recommended for measuring applications because they allow measurements to be made accurately even when the subject-to-lens distance may vary slightly. The optical DOF of a telecentric lens is identical to that of an ordinary lens with the same NA, but the telecentric lens is usable (can meet accuracy requirements) over a wider range of focus. If you can precisely control the subject-to-lens distance, then you don't need telecentric optics.

Your requirements are unusual: small subject, small sensor, and short working distances.

One approach that comes to mind is to place two infinity objectives back-to-back, with an infinity section in the middle. For example, placing two 4X NA 0.10 objectives back-to-back will give you a 1:1 system with the same field and sensor sizes that the objectives would normally have, so roughly 6-10 mm diagonal.

In that case, used at 1:1, the system would also have zero geometric distortion because it would be completely symmetric. Used with any other conjugates, you would probably get field curvature and other distortions.

Pairing other combinations of objectives such as 4X and 2X to give 1/2X might work OK.

Mounting will be non-trivial, but basically it's just a matter of providing an intermediate plate into which both objectives can screw.

--Rik

Interesting, never tried that before.

I decided to pull out 2 crummy RMS 10x achromats I have and came up with a very neat solution.


Two fakromats, two Thorlabs RMS-SM1 adapters, one Thorlabs SM1 tube.


Put the adaptors on. Back to back, into the tubing. Both Fakromats are inside the tube.


Complete with a helicoid. M26-M52, into M52 focusing helicoid.


Now adapt the long and naughty...endoscope looking thing to the F-mount.


It's very likely far more than 1x, but surprisingly FX is covered. I think focus is off by a lot but case in point, it works. Very short working distance as well. Zero visible geometric distortion.


It wouldn't work with larger objective lenses but one can always get the SM2 tubes anyway. I've only seen 2 or 3 objective lenses that are larger than 50mm in diameter.

[Lou Jost]

Interesting test! I wonder why it's not 1x though. Maybe it has to do with them not being infinity-corrected optics? In that case it seems the distance between them might be important. Do you have a pair of identical infinity-corrected objectives you can try? I tried this once with matching camera lenses, each focused at infinity, and it worked well.

[Macro_Cosmos]

Interesting test! I wonder why it's not 1x though. Maybe it has to do with them not being infinity-corrected optics? In that case it seems the distance between them might be important. Do you have a pair of identical infinity-corrected objectives you can try? I tried this once with matching camera lenses, each focused at infinity, and it worked well.

Yeah good point, those were some inexpensive no-name objectives. Unfortunately I don't have a pair of identical infinity-corrected lenses.

These objectives require 160mm tubes. Just a shot in the dark, but maybe I have to space them out by 160mm?

A couple 4x infinity corrected objectives could be around $50, not willing to spend that money just for an experiment like this. Hope someone with 2 copies of the same objective lens tries the coupling method. Maybe the PN105A can be dethroned... though I think it won't happen.

Admin edit RJL, for style