The tool every optical lab needs to have

This video describes how to build a powerful alignment and metrology tool for your optical lab. It provides an accurate reference of infinity (collimated) light and allows to measure very small angles of reflective or semi-reflective (e.g. air-glass) interfaces. Autocollimators are real time-savers in optical labs!
If you would like to have more information on autocollimators and angle measurement, please consult these two posts on our website:
- "A Reference of Optical Infinity" (text version of video) : www.thepulsar.be/article/a-ref...
- "Measuring Small Angles Deviations" : www.thepulsar.be/article/measu...
I hope you will enjoy the video! Do not hesitate to comment!

[@philparsons5051]

I've despaired of youtube of late, it seems to be filling up with clickbait full of mistakes by people trying to be professional creators. Then occasionally I run into the odd gem that impresses me with its depth of knowledge and makes me reconsider my thoughts, and this is one of those very gems that impresses me enough to log into youtube, add a like and sub to the channel and backs it up with detailed coverage on a website away from the advertising generating pit of youtube. Thank you.

[@martonbalogh8255]

Thank you so much man, this video is a life saver!

[@blackout57]

Hi, as usual you gave us a great video. Thanks for sharing the things you like.

[@matthewdom8222]

loved it more please! Best Matt.

[@wikicheedia]

Great video. Any chance you can put the bill of materials on the website links? It flashes up a bit too quick on youtube.

[@yakubumshelia1668]

Excellent 👌

[@maneki9neko]

Very helpful.

[@marcb7107]

Been following your website, esp. the recent ones on Raman -- very nicely done! Two questions: (i) any advice for putting together a system for gas-phase Raman? (ii) I feel like I read somewhere you're working on putting together a laser-diode driver ... anything you can share on that yet?

[@thepulsar]

I have never tried gas-phase Raman up to now. I can test it but I am afraid that I may not have enough sensitivity to do that :-/ I will test anyway Concerning the laser driver, I have some bad news. I just tested the board a few minutes ago and it fried my laser diode in a few seconds. It was working fine with LEDs and I was operating the laser much below its max current but for some reason it did not work. The light output of the laser dropped to zero after less than 20 seconds...

[@ThunderPelocho]

This is an excellent video - well done! I am wondering in what context you acquired these components; perhaps a project at work?

[@thepulsar]

Hi. Thank you for the feedback! To answer your question, I am using some of my spare time to work on metrology instruments as a hobbie activity. I then share the results on my website (and more recently on youtube) to help people reproducing the experiments. For this first video I decided to talk about autocollimators which can be quite useful when aligning or characterizing other projects. If you have a look at my website you will see that I have a few posts on optics with devices such as spectroscope, refractometer, fluorescence sensor for fluids... Next video will probably be about scattering sensor but I am also working on cyclic voltammetry at the moment :-)

[@ThunderPelocho]

Fascinating. I am working in optical sensing myself and I would love to do more experiments on a hobby basis. But as you know optical/mechanical elements, especially from Thor/EO/NP are very expensive for a hobby. The BOM you show is probably upwards of1500 Eu. I am subscriber #2 :-)

[@thepulsar]

That is true. Thorlabs is what I would call almost affordable while EO is a bit more expensive and Newport is simply no-way to go. Clearly, most parts are just impossible to get as amateurs. Lasers, 3 axis translation stages, motion controllers... are just way too expensive. On the other hand, you can get a few lenses and mirrors at a reasonable price. Especially when you do not need a lambda/10. After that, you still have a few more expensive parts (beamsplitters, cameras...) but that are still reasonable as investment because you can use them in almost all your projects. They are a bit like LEGO bricks and, to answer honnestly, I spend most of my time trying to optimize which elements to select such that I can use them in several projects. It is quite rare that I buy a part for only one specific project, usually it is more like 5-10 potential applications. Finally, and when possible, I try to reproduce the electronics by myself because that is a part where you do not get a lot of bang for bucks at Thorlabs. For instance, in this video I am using an homemade LED driver that costs a fraction of the price of the Thorlabs one with almost the same specifications in terms of driving current, modulation depth, noise... But on the downside it costs you time because such a LED driver took me about ~50 hours to complete.

[@MrImhotep2287]

Could you please put a text for the supplies for this I'm looking to build something like this for checking my surface plate? Thank you

[@Pidrittel]

Really a fantastic video, Chapeau! I've got a few short questions: - What is the chromatic error of the system? e.g. could you use the setup to check the collimation of a beam of different wavelengths? I think about building an autocollimator-setup like this for testing the collimation of a laser beam in an optical setup. For example, if you switch from the lower end to the upper end of the visible spectrum, 400 to 700 nm, would you need to recalibrate the setup? Or would you expect the error to be small because you used an achromat as a collimating lens? I will try to simulate this beforehand, but maybe you did some testing as well? - Do you use 3D-Software to plan those Thorlabs Cage systems ahead? If yes, which software can you recommend? - is the first sequence in this video filmed or rendered footage? If the latter, what software did you use?

[@thepulsar]

Hello, - For chromaticity what I do is I switch the LED depending on the wavelength I need and I recalibrate the autoco; it's quickly done and guarantee a good result! - Today I do everything in SolidWorks prior to assembly but at the time of this video I did not; just a pen and a piece of paper can be enough - First sequence is actual footage :) I used some sort of skateboard made for radial traveling and smoothed the motion in after effects

[@void0alexey]

Great explanation and implementation, thank you. Can you answer if LED laser (the ones from thorlabs also) can be used instead of one you mentioned in BOM?

[@thepulsar]

The problem if you use a coherent source is that you will get an extremely noisy image due to speckles

[@void0alexey]

@@thepulsar need to read about it now)) one more thing, I found article on autocollimator design but they combine multiple lenses. Overall it looks like same for me. Do you think it is applicable here also? www.researchgate.net/publication/289733489_Autocollimator_for_Small_Angle_Measurement_over_Long_Distance

[@movax20h]

Really cool video. How does the temperature and expansion of the metal elements influence the accuracy of such autocollimator?

[@thepulsar]

I haven't checked/simulated but based on the symmetry of the system I would only expect a very slight defocus, but nothing that would be mistaken from tilt/angle change

[@gianlucadezanet6187]

Hi, thank you for your great work! I have a couple of questions about the applications of an autocollimator, I couldn't find any answers around. Basically I have a telescope structure with the secondary mirror supported by some struts, and I should monitor tilts and displacements of it when exposed to a thermal cycle. Is the autocollimator actually suitable for such measurements? I think it is for angles, but could it be good for displacements as well?

[@thepulsar]

if you would like to monitor the motion of the secondary mirror the autocollimator can be used for the tilt. Sometimes it may be required to fix a temporary mirror either at the back or side of the object under test for accessibility purpose (I'll let you look for that). For displacement, I would use different techniques depending on the expected magnitude. There are professional interferometers that can help you to do that but if you can't afford them you may use techniques like sending a beam from a 45° angle and look by how much it moves on a detector at 90° after reflection on your object. Interferometry will give you the most accurate results but may be sensitive to vibrations and stuff like that. Lot's of possibilities actually :) You can also have a look at confocal sensors.

[@gianlucadezanet6187]

@@thepulsar Thank you for your fast reply! For the autocollimator applied to the telescope yes, I was thinking the same: temporary mirror on the back of the secondary mirror, both flat and perpendicular perhaps so that tilts in each direction could be measured. What about the air refraction index? It should be compensated in some way, even though the measurement can be just relative to the original configuration I guess. For the displacements, I had a look at the components of the Michelson Interferometer Kit by Thorlabs (EDU-MINT2), apparently for around 500 £ (recycling the beam splitter from the autocollimator) could work. The confocal sensor seems sooo cool, but prohibitively expensive :D The main issue is maybe the thermal cycle itself and the vibrations coming from the chamber.

[@vh1385]

Thanks for this nice video. I can see how this comes in handy when seeing up a telescope. Say I would like to center two lenses on a common axis, how can this be done with this autocollimator? Can you advise a bit?

[@thepulsar]

Hi Tom. Thanks for your feedback! To my best knowledge, the easiest way to center lenses on a given axis is to first align your autocollimator with a spindle axis where you will mount the lens and then place the lens and watch for the reflected profile. You will need an extra focusing lens in front of your collimator to do that. When the lens is not centred, the reticle will describe a circle whose radius is proportional to the decentring. By narrowing down that radius, you align the lens with the rotation axis. If you check for "autocollimator centering lens" on google image, you will find diagrams that explains the concept. Despite I have often worked with third party companies who uses that technique to align lens, I have never tested it myself so I cannot give you much more advices than that. When I have time I'll try to implement something but that will have to wait a few months - I am currently busy finishing my spectroscopy project :)

[@vh1385]

@@thepulsar Thanks for your reply! I found some pictures and they do look helpful indeed. Might be a bit of a pain though to properly align the focusing lens :-/

[@bdgackle]

If you are willing to share, how did you create this animations? They are both very cool and very efficient at conveying how this thing works.

[@thepulsar]

Sure! I first trace the paths in Illustrator then import them in after effects using the scribble or stroke effect :)

[@rvogel61]

Thanks! Phantastic video! I have one question: If the collimator objective is somewhat tilted it would redefine the optical axis in an unwanted way, wouldn't it? If so, is it possible to use a kinematic holder for the 2" lens to adjust it after roughly setting the correct focus distance, removing the reflective surface mirror and trying to collimate the collimator objective by centering the reflected reticle by the lens with the original reticle image? Does this make any sense or improve the precision of the autocollimator, or is the tilting caused by imperfect mounting in the lens holder negligible in that context?

[@thepulsar]

I would not use a kinematic mount for the lenses here for the following reasons. There are actually three different axis in the setup: (1) the mechanical axis, (2) the lens axis, and, (3) the axis of the projected beam. While it is true that a tilt of the lens will affect the optical axis of the lens compared to the mechanical axis of the system (as well as decentering or tilting of each individual surface of the lens), it is not the major source of departure between the mechanical axis and the projected beam axis. Most of that will actually come from the centering of the reticle which is on the order of 1mm roughly here. Defining an absolute axis is not an easy task which is why autocollimators are most often used as relative measurements comparing a state of a system with another state. Hope this answers your question :-)

[@rvogel61]

@@thepulsar First and foremost, thank you for your prompt answer! I understand that the autocollimator essentially defines a new axis via the projected beam. It might be close to the desired optical axis , but as long as all mirrors, crystals, etc. are aligned along that "unknown" axis, I should be fine. Is that what you meant by relative measurements? Your answer brought one more question to my mind: does the shift of the center of the reticle by, say, 1mm off the optical axis in the X and/or Y planes (translation) have any mentionable effects on the angular alignment process? True, the center point is not in the projected beam axis, but the angle error of an reflected beam against the projected beam is still visible by the shifted return image and can be still be quantified by taking the focal length of the collimator lens and the pixel size of the camera into account. Or am I mistaken here? Also, if the reticle itself is slightly tilted, apart from that not all of the reticle might be focused in the camera picture, does this effect possible measurements of the misalignment of reflections against the axis of the projected beam? As the reticle is in front of a diffusor, does it contribute to the projected beam axis, or is the axis of the projected beam soley derived from the orientation and optical properties of the splitter and collimator lens? Does it make sense at all to try to put the center of the reticle into the axis of the projected beam, it seems arbitrary to me as long as only angular misalignments are measured?

[@thepulsar]

You can (and should) always try to center the reticle to the optical axis of the lens; however getting a truly perfect alignment will be difficult and expensive to achieve (there is no real limits!). Here I chose some kind of tradeoff which makes sense in the context this setup was designed and used. For example, one of the major limitations is that there is slight distortion compared to a F*tan(theta) rule and so the relationship between numbers of pixels and angle is not strictly linear. Nonetheless, it can still measure relatively accurately very small angular motion like the wobble of a translation stage. If you check in the archives of the website, you will find this very example. I hope it will give you an idea of the kind of things you may be able to get with this setup. I also posted a video on how to align a vise on a mill using it :)

[@thepulsar]

www.thepulsar.be/article/precision-motorized-translation-table/ 2mrad that's 2mm at 1meter, just to give an idea :)

[@coolfox999]

Great video thank you ! I'm in the process of building autocolliator myself from old 400mm telephoto lens and some buts I acquire from ebay like led light source and old microscope beam splitter ( the one used on teaching microscopes with two sets of oculars ) with have a bit more complicated prism when 50;50 cube. I have one may be silly question , why you set camera on one side of prism first and after transferred to opposite ? It surely would work in the first place .

[@thepulsar]

When the camera is in the first configuration, it allows to self-calibrate the autocollimator so that the lens is at the right position in regards to the reticle. What you observe then is the direct image of the reticle plus a dimmer image of the reticle that actually comes from the lens. It does already work in that configuration but, usually, people do not like to have the direct image of the reticle and prefer to have only the one that passes through the lens. That's why I put the camera in the second configuration. Extra bonus is that it is more compact that way. Also, by removing the alignment lens you get a slighltly better image quality but that's not a large issue in this system but it could be one in very high resolution experiments.

[@coolfox999]

@@thepulsar Thank you for explanations, much appreciated :-)

[@coolfox999]

@@thepulsar Future said about splitter prism I have found kinda unusual design here's link to the schematic ( sorry my drawing skills :-) i.postimg.cc/pLPKqsTF/Prism.jpg love to hear what you have to say about and your view on how this my work in your autocollimator design? This prism have high degree of light cancelling between outputs ( there still ghost image can be seen ) .

[@diatomsaus]

Excellent video! Do you plan on doing one on measuring the resolution and calibrating microscopes using this setup?

[@thepulsar]

In fact I already use this setup everytime I need to align system at infinity or in a 4-f configuration. When I have internship students I always start by explaining them how the autocollimator is working and ask them to align an infinity-corrected microscope setup. Based on experience, it is a very good way to teach student how to align things with precision. I started doing that after seeing a student using a ruler to align a lens/target :-) Could you develop a little bit what you mean by measuring the resolution ?

[@diatomsaus]

Ah, I was more or less referring to the usage of typical resolution charts such as a usaf or star target to measure the system's performance. It's also somewhat hard to get the objective, tube lens and camera in total alignment with such test charts. A bit of tilt throws everything off. I use a thorlabs 95mm rail and one of their filter holders to hold a chrome plated line chart, I would like to measure the resolution of a new lens I acquired for consumer digital cameras and potentially compare that against the Mit 10x objective. I'm still new to this field, really like these precise setups, lots of fun as a hobby (and expensive...).

[@diatomsaus]

flic.kr/p/Hi5s92 flic.kr/p/26WZpZn This is what my older primitive setup looked like. The tube lens I was attempting to use is the ITL200 which I eventually just gave up. The cheap tubes I was using is just horribly imprecise, and as of now, I can't really justify building a complete thorlabs SM2 tube system for I'm guessing just an edge in pictorial performance over the old Nikon 200mm ais lens I am using as a tube lens. As of now, the 2 performs the same. Pic 2 is just a piece of printed text on a card, a very simple method to see aberrations. I have a pretty good test target, just not exactly sure how to use it properly with the stuffs I have. Aligning them has always been a massive headache. I'm thinking that maybe using a collimator will make this process easier? I looked up those parts and what-not, it's really expensive, even though some parts can be sourced on places like ebay. Missed that small camera shown in the video for $80 a while ago, still a bit grumpy haha.

[@thepulsar]

Nice setup and rail! From what I see you are using a high quality Olympus objective. Normally with that you should achieve diffraction limit (or close to) on full field. Most of the time I use just a 200 mm achromatic doublet as tube lens but I also used a 200 mm Nikon AI f/4 lens at some occasions with good results. From what I see you also have some kind of mechanical alignment between your microscope objective and your lens so that should give you some good quality. So I am a bit surprised when you say you suffer from aberrations. The only potential improvement that I see is to align it as a 4-f system to reduce aberrations. You can use an autocollimator to do that (align microscope objective/target first, then put a mirror between target and microscope objective to align the tube lens with the autocollimator and finaly align the camera sensor to have a sharp image). There can be variations on the alignment procedure but the overall idea is there. If you look at the end of my video on aberrations you can see the resolution achieved that way with a custom microscopy objective on a USAF 1951 target (to have an idea of what is "easily" achievable). Of course with your Olympus 10x Plan you will get much better resolution than the ~200 lp/mm shown.

[@diatomsaus]

Thanks! The objective I'm using is a Mitutoyo 10x mplan apo. Fragile little thing, mitutoyo objectives are made to break... good job haha. I guess olympus and nikon objectives are far superior in terms of longevity. Read too many horror stories about amateurs breaking their mittys from just a tiny bump or so. My aberration issues, from what I can tell, stems from imprecision. I am using sets of cheap M42 tubes and a wobbly adapter with the itl200, the entire system is unstable. Using such cheap tubes also bring in other issues, such as thread precision causing misalignment of the optical axis, which could contribute to aberrations. Upon further research, there is actually a sweet spot in the "infinity space" from the tube lens. Calculations show the objective should be about 65-70mm from the end of the tube lens, and thorlabs quoted 70-140 for optimal performance. The mitutoyo MT-01 tube lens on the other hand quotes 0-165 or something, so it should be okay. I'm using a D810, which is more demanding than typical industrial cameras. Focused to infinity, the itl200 is great in the centre and abysmal in the corners. The 200ais is a hair worse, but the quality retains almost throughout the frame. This is shot wide open. I assume putting an objective in front would be equivalent to closing the iris down, I may be wrong. If that's the case though, IQ would be great on the ais. Read some articles on your site, very nice and informative :) Thanks for the great work! I've decided to bite the bullet. Going to purchase the Thorlabs' SM2NFM2 (nikon F to SM2 tubes) and a bunch of SM2 accessories for the itl200. Gotta do it justice! I'll share the results when the setup is built! I still stand by my statement, that the 200 ais would be almost as good as the itl200, and a lot better overall if corners are a concern. But anyway, even if that's the case, I'm left with high quality tubes, not the cheap junky m42 tubes, which have horrible internal reflection issues, hence manual flocking is also needed.

[@Lebedis66]

Amazing video. Can I ask why did you switch to 200mm lens rathern that microscope objectives you had in the past? I guess in theory you get a slighly better NA, but surely abberation correction is not as great?

[@thepulsar]

You can actually use any focal; the longer the focal length the more sensitive the instrument. But on the other side long focal also means short measurement range. Here I wanted to achieve high sensitivity :)

[@alfonsoianniello9139]

Hi, thank you for great video. Following your project you think it is possible to put the camera instead of the light source place? to reduce the total length of the autocollimator? Thanks

[@thepulsar]

Sure :) The alignment procedure stays the same also

[@alfonsoianniello9139]

Thanks! The cube splitter orientation also remain the same or not? Do you thinks is possible to implement a system to have a greater magnification on the camera?

[@thepulsar]

The cube orientation itself can remain the same ; nothing else should change even for the alignment. Having a magnification is possible but it depends on where you want the magnification to happen. If it is inside the autocollimator itself (so image of the reticle) you will need to add a magnification system inside the autocollimator which is not trivial. Better use a larger slit/reticle then. If you want the magnification to happen in relation with another system (e.g. you are aligning a microscope objective at infinity), you can change the autocollimator focal length. The magnification ratio will be given by the ratio between their focal lengths then. 200 mm like here is good to align microscope objective but if you project the reticle the other way around it will be very small with short focal length objectives. The solution is then to replace the 200 mm lens by a 100 mm or 75 mm one.

[@mr.chindo8570]

Thank you

[@jimmyzhanneutron]

Hi, thanks for the great video on autocollimator! Can you explain what are the differences (functionally speaking) between this configuration where the collimator lens comes after the beam splitter, versus the one on your website ("measuring small angles deviation") where the collimator lens is before the beam splitter?

[@thepulsar]

It's a tradeoff between cost, size and optical image quality. For short focal length and angle measurements, the version on the website is the best. For long focal lengths and focus adjustments, this version is more suited.

[@jimmyzhanneutron]

@@thepulsar I made the autocollimator based on your design here! However, when it's done, I'm seeing two reticles (you mentioned one is the ghost). The bright one moves when I tilt the mirror slightly, and the dimmer one does not move when I tilt the mirror. Why does the dimmer (ghost reticle) does not move when I tilt the mirror? Let me provide more info: the dimmer reticle will go in and out of focus when I tilt the mirror significantly, but it doesn't move! The bright one behaves expected, it moves even with minor tilt of the mirror.

[@thepulsar]

@jimmyzhanneutron I noticed the same behavior in the past but I would have to do more experiments to pinpoint the origin of the issue. I'm pretty confident that it is a secondary reflection on the target and it's not linked to the cube or lens internal reflection. My theory is that this secondary reflection acts as a "cat's eye retroreflector". I also think you can get rid of it using a 0.7 ND between the cube and the lens but I didn't validated it yet.

[@adinsched7190]

Hi, thank you for great video. What is the estimated resolution and/or measurement error for such installation? Did you check it?

[@thepulsar]

The actual performances are difficult to evaluate due to the very small angles involved without relying on expensive equipement. On the other hand, it is possible to evaluate the theoretical performances based on a ray tracing analysis in an optical design software. The measurement sensitivity is directly given by the focal length and the pixel size and is about 0.0015°/px here. With a typical (based on personal experience) correlation model to detect the center of the cross, it is possible to get 1/4 of a pixel in displacement so we can say that the sensitivity to small angle deviation is about 1 arcsec. This does not take environmental effects into account however (air, vibrations...). On the accuracy range, you first have the accuracy on the focal length which is

[@adinsched7190]

Exhaustively. Thanks a lot.

[@amirimiev3803]

Thanks for your video. Very interestingly. Ready engineering decisions with such accuracy rather expensive. I consider that your project can be ideal balance of the price and quality for young researchers and students. I intend to collect the same system as at you, only on a granite precision basis and with more rigid metal framework. The device will be is in our laboratory with steady climate on an anti-vibration table. How do you consider whether it is worth replacing F=200 achromat with another, with longer focus (for example F=300 or F =400) for the sake of increase in accuracy of measurements?

[@yerensaygi]

Maybe I ask a silly question but can we use 1" achromatic doublet lens instead of 2"?? Very very good project by the way

[@thepulsar]

yes no problem to use a 1" doublet :) depending on the resolution you try to achieve, you can also use a smaller focal length like 100 mm or 150 mm

[@berndvonderlinde1035]

Great Video. Can you leave list of the components and where to get them.

[@maneki9neko]

All these components come from Thor Labs. The parts list is at 5:40 in the video. All the part numbers are Thor Labs part numbers. If you build one of these you might post here, how it goes for you. I am contemplating the same.

[@joelevi9823]

Thanks for this great video..Is there a cheap alternative to thor? I mean only the laser cost 216$

[@thepulsar]

It is indeed possible to build a cheaper version but that would require major reengineering. I would not sacrifice on lens quality though

[@TrekDreamer11]

I have watched this vid a number of times now and crawled your website. With all but minor parts in my possession my progress continues to hampered by supporting software. On your website i notice use of matlab using With some supplied code. Is there a commercial or open source software that you use.

[@thepulsar]

you can have a look to OCTAVE ; it is a free version of Matlab :)

[@GenaBazarko]

Hello ! I have an autocollimator in which there is no objective lens (focal length 1000 mm, diameter 55 - 60 mm). Can you know where to buy such a lens?

[@thepulsar]

Have a look at Edmund optics they might have something in stock in their planoconvex sets

[@GenaBazarko]

@@thepulsar Thank you .

[@user-qe5vi8df1h]

Hi, thank you for great video. Following your project you think it is possible to create a collimator instead of an autocollimator? Maybe replacing Crosshair Reticles with a very small pinhole or a Usaf resolution chart ? I'm trying to build a collimator that allows me to have an image projected at infinity. Thanks

[@thepulsar]

Yes/no. While it's fairly easy to replace the reticle by any other target (pinhole, slit, slanted edge, ronchi, dot pattern, usaf etc.) the question is more on the quality you expect. If you aim at quantifying contrast for mtf evaluation you will need very high imaging quality which is not easy to achieve. In the current design form, the BS is going to introduce aberrations to the image. However, you can align your target with a second lens using the autocollimator.

[@thepulsar]

It's also important to budget the allowed depth of focus for the autocollimator. If I remember correctly, it's around 0.2 mm for this design (f=200 mm). You may need to increase the focal length while maintaining the f# depending on the lens under test and the precision you want to achieve

[@user-qe5vi8df1h]

@@thepulsar I meant from your setup removing the beam splitter and replacing the target to create a collimator instead an autocollimator. But how to calibrate it without BS? Also what did you mean : "However, you can align your target with a second lens using the autocollimator"?

[@thepulsar]

@user-qe5vi8df1h you cannot align without a beamsplitter. One solution (among others) is to use the autocollimator to align a second system (the collimator) that has the same fnumber

[@user-qe5vi8df1h]

@@thepulsar Thanks! I'm trying to build a system that allows me to verify the quality of a camera lenses. A setup like this: on one side a collimator that projects the image of a resolution test target or pinhole at infinity and on the other side behind the camera lens to evaluate a ½" CMOS Camera with a 160mm tube and microscope lens mounted. I hope I explained it well :)

[@KyleDouglass]

Thanks so much for making this video! I've built such an autocollimator (AC) for our lab and have been playing with it for about a week now. Regarding the "ghost" image of the reticule that appears after the camera is put in its final location: do you know from which surface of the beam splitter it's coming? I noticed that blocking the light between the final AC lens and a mirror that reflects light back into the AC will eliminate both the real and ghost image. Furthermore, only the real image will move when the mirror is tilted, whereas the ghost image remains fixed. The location of the ghost and real image also become displaced axially when another lens is placed after the AC lens but before the mirror. I find this behavior a bit inconsistent with all of the hypotheses I could think of regarding which surface the ghost image is coming from. Any ideas would be greatly appreciated!

[@thepulsar]

To be checked (just a quick idea thrown here) is a double reflection involving the cmos sensor. One solution to test is to insert a ND filter at the beamsplitter, that should reduce the ghost :) tell me if you try it and get some results!

[@KyleDouglass]

@@thepulsar Thanks for the reply! I think I figured it out yesterday evening. If all one has is a mirror at the end of the AC, then the ghost image is from light that actually passes twice through the AC. After passing through once, it reflects off the reticule and then passes again to the mirror and finally to the camera. Why doesn't the ghost image move? If you draw a ray trace diagram of the AC with a tilted mirror, but trace the reflected rays back to the reticule and make them reflect off of it, you will find that these rays actually return to their original point of origin on the reticule. (Assuming I haven't made a mistake 😜.) So in this way, aligning the main image with the ghost actually ensures that the mirror's surface is perpendicular to the optics axis because only the cross from the first reflection moves. Pretty cool! If you have another lens between the AC and the mirror, then the secondary images in this case are coming from spurious reflections. For example, if you use a plano-convex lens with the flat surface facing the AC, you can make it perpendicular to the optics axis by aligning the cross from light reflected off its surface to the cross coming from light reflected off the mirror. The Thorlabs camera seems to have enough dynamic range to see both. Thanks again for the video! I now know that an autocollimator is way more interesting than I first thought. I'm trying now to find a way to extend it to be able to align lenses in all three dimensions.

[@KyleDouglass]

In case someone finds this comment, I should add that, upon further testing, I found that you can only ensure that the mirror is perpendicular to the reticle when the reticle center is exactly on axis, which likely won't be the case given the tolerances of the Thorlabs cage system. As was mentioned in another comment, this makes the autocollimator better suited for relative measurements.

[@pavelpenshin2871]

Hello, Could you please elaborate on why would you not advise a color camera but a monochrome?

[@thepulsar]

That's because of the Bayer pattern and the subsequent "debayerisation" process that is manufacturer dependent but which will affect always resolution. The process can also introduce non-linear response of the position detection due to the fact that we are only using 1/4th of the pixels with the red light

[@thepulsar]

That however only apply to Bayer type but there is no real advantage of using a tri-cmos camera for monochrome applications

[@sourabhkumawat4863]

Thanks for saving my 10000$. Can you please share details of the raw materials you used.... I mean like where one can buy them

[@thepulsar]

Thorlabs dot com for the supply and the bill of materials is in the video :)

[@Simon28298]

Where are you in Belgium?

[@fernandofarrera.1764]

Can I use this to collimate my camera lenses?

[@thepulsar]

That's exactly how I use it myself :)

[@mannycalavera121]

Do you have another channel, you sound really familiar

[@thepulsar]

No other channel, that must be someone else :)

[@corbinlinder2631]

Very cool build, unfortunately it's nearly $2000 USD. Do you think an autocollimator can be built for under $500?

[@thepulsar]

With some major redesign and custom parts you can probably go below 500USD indeed

[@beingbanana]

enjoying those lab snacks ?

[@thepulsar]

"a threat to humans" ;) that makes you wonder what they actually put inside

[@leicaindia2321]

hii ThePulsarBE How to built autocollimator with the help of auto level

[@thepulsar]

Hi Pushkar, could you develop a little bit ? I am not sure to understand what you are referring to with auto level. :-/

[@petrucciMsk]

My gut feeling tells me one can pickup a couple of used autocollimators at a price of the components used in this setup, so this video should rather be named "how autocollimator works" or something like that :) But thanks anyway, subscribed.

[@andrewyork3869]

I was looking at that too, may see what can be made, namely the skeleton of sorts that he is using as a frame. (Granted I could be thinking of something else, how ever I believe those rods to be pretty pricey.) A fully enclosed system makes more sense to me.

[@andrewyork3869]

Actually depending on what you substitute and what you substitute with, this can be pretty compelling if you a willing to build a housing.

[@hunkaryilmaz6311]

Hi, thank you for great video. Following your project you think it is possible to create an autocollimator for test the optical quality of Dsrl camera lens? Like an optical bench! Maybe replacing Crosshair Reticles with a very small pinhole for generate an airy disk or airy pattern? or a Resolution chart ? To be more precise something as described in this article www.reduser.net/forum/showthread.php?158004-Let-s-make-STAR-LENS-TESTER-to-get-best-lens-copy- easily- (Airy-disk-PSF-point-source)

[@thepulsar]

You can use the principle but you will need some magnification in front of the camera because the PSF is likely to be small (e.g. a few µm) and you don't want to be limited by the pixel size of the camera. However, if you want to assert the quality of a photographic lens, I would recommend to use a different method, the slanted edge, for several good reasons. (1) it is much more straighforward to implement as you just need a properly illuminated "L" shape with good contrast. (2) It is closer to how you will actually use the lens (instead of a double pass system with a mirror). (4) At contrario to a pinhole measurement, you are also influenced by off-axis points which is, once again, what you will measure with your optics unless you aim it at stars. (5) Slanted edges are much more energy friendly when compared to pinholes which quickly get you intro trouble because of camera noise. (6) Last but not least, it is recognized as ISO standard for MTF evaluation. So I would not say it is "forbidden" to measure optical quality using a pinhole/camera, but it makes much more sens to use the slanted edge unless you are targetting very specific applications. On the other hand, autocollimator setups are very handy for a lot of different tasks (aligning optics, measuring angles, finding focal length...). I hope this answer your question :) If you want to have a look at the slanted edges, a quick look on google images will provide you with plenty of information

[@hunkaryilmaz6311]

Hi, thank you for the reply! For hobby I like to repair DSRL Camera lenses, my big problem has always been the optical alignment, specially the adjustment and checking of the optical axis of the repaired lenses. I'm trying to build an instrument similar to the one used by Nikon for optical centering, that I think is like a PSF based instrument. In the meantime I'm trying to find also other solution. If you have some advice or help for my project I would be very happy.

[@thepulsar]

Thumb's up! That is a great project :) do not hesitate to share when you have something running. Also, and based on what you are doing, an autocollimator is precisely the type of technology that you need. Because a tilted lens is equal to a lens + a wedge, if you send collimated light through a tilted lens in rotation around its mechanical axis, you will see your '+' pattern discribing a circle around the rotation axis. The smaller the circle, the better the lens alignment is in regards to the mechanics. That is precisely the technique used for ultra-high precision lens centering such as described in this video: kzbin.info/www/bejne/mnXQl2yfa6qWi7s . To build the setup, either you could split the illumination and place the camera at the focal position of the lens group to see the '+' mark (easiest solution), or you can use the autocollimator with a mirror at the focal position of the lens group. I haven't tested this myself but I know the principle works (maybe you will need to find a few tricks to have it running smoothly). When you list all the potential applications of autocollimators, you realize that is a very powerful tool to have in a lab :)

[@thepulsar]

Sorry, I meant decentered lens and not tilted lens. Tilted lens will generate on-axis coma.

[@hunkaryilmaz6311]

I'm tring to make an optical alignement instrument like this www.midwestcamera.com/images/backfocus_jig.JPG , Nikon manifactured. In order from bottom to top we find: condensed bright light - 30 µm Pinhole - in the middle a plate with a Nikon mount bayonet for the lens to test/adjust - on the top a monochrome ccd camera with a high magnification lens. But what kind of lens was used? telecentric? macro? magnification? I did not understand it!

[@user-ux5zw1kk9e]

The sensor is insanely small for a autocollimator, 1/2.3" senor only measure about 1 degree devation, which is too small for practical use!