Laser Scanning Microscope from Blu-ray Player #3: Increasing the Resolution

[@havelsand]

Excellent stuff. This microscope series must have far more views. Totally underrated. I'm curious what comes next in your channel.

[@pastarmak]

10 years ago I was working as electronics engineer for a German company - Zwick/Roel - a harness testing machines manufacturers but in their England facilities - Indentec. I had exactly the same idea - using blue ray mechanism for scanning and measuring the indentation, those times people was laughing and the idea was discarded. I'm glad someone has realised it now.

[@andrewphi4958]

About the same time I've seen a cut from a TV show where russian kid and his teacher made a working microscope out of a CD drive with very good resolution and specimen glss embedded into a normal disc to insert. Worked pretty well, down to bacteria size. Unfortunately, I couldn't find any further info.

[@Redcactus5]

You may be able to further increase the resolution by also using the translation voice coils on the laser assembly, in addition to the sample stepper. If you make it scan in a similar way to the way a drive reads (the coils do most of the moving, the stepper only advance when the coils run out of travel), you may be able to improve quality. You could also try adding position sensors to the motor to allow for error compensation. adding an x translation stepper to allow for scanning bigger areas at the same detail level could be neat too. Not sure if any of this would work or even be worth it, but it is an idea. Love this project!

[@SnoopyNug]

What excellent ideas!

[@Thefreakyfreek]

Thats how the dvd reader works

[@Redcactus5]

@@Thefreakyfreekthat’s where I got the idea. These microscopes are incredibly similar to optical drives, so I thought that things could probably transfer over.

[@7LegSpiders]

I'd love to see an encoder on the shaft for closed loop positioning, but a belt or gear on the motor for speed reduction (better position accuracy). Tell me why that wouldn't work to keep the same motor, but reduce the heat generated by splitting phases so small.

[@talkingmetastasis607]

please don't give up that project, you're such an inspiration for me. i dream about a diy confocal microscope every day. cheers bro

[@randolph6013]

Respect, someone with such skills in the fields of precision mechanics, electronics, and IT can be proud of themselves.

[@AJMansfield1]

The stepper driver's microstepping could be the cause of the vertical blur in your images, if it's inducing vibrations at the microstep switching frequency. You may be able to get better results if you add more mass to the stage to dampen this vibration, or actually disable microstepping by setting the driver's configuration pins. Microstepping involves generating intermediate "unit vector" voltages to feed the two stepper coils, and nearly all implementations of this rely on high-frequency PWM to do this, for efficiency sake. Lower switching frequency results in fewer transitions, which means less heat dissipated in the driver. Large stepper motors with a lot of inductance attached to heavy mechanical stages with a lot of inertia will intrinsically damp out frequencies above a threshold, making it a trade-off that's easy to design around. But I suspect the motor and stage you have are a lot lighter than what that driver was designed to control, and some of its micro-stepping PWM frequency is vibrating the stage enough to blur the image vertically.

[@longpham-sj5sv]

With such small motor, he could even make a linear (like in linear power supply) stepper motor driver to get even lower mechanical noise.

[@longpham-sj5sv]

To some extent, the voice coil - focus feedback loop can cancel out the mechanical noise if it's fast enough

[@PaulSpades]

Here's the thing. Microstepping actually wabbles between poles, and there's backlash in the mechanics. The proper way to increase resolution is to gear down and slow down the movement. And average samples between full runs - that should average the mechanical noise, not samples in each positions - which averages laser sample noise but should already be accurate enough. At this scale, I'd try polar coordinate movements, ironically the way discs are already set up to work. But disabling microstepping, finer lead screws(probably not available) and slower speed should give better results. Alternatively, geared down steppers could be available commercially.

[@longpham-sj5sv]

The stepper driver uses PWM and feedback loop to regulate phase current in full step or micro step mode. It's the PWM causing current ripple in each phase, which translates to mechanical noise. I think it could be mitigated a lot by using a linear stepper driver given the motor is fairly small or improving the voice coil - focus feedback loop so that the focus is maintained despite the noise (the PDIC has bandwidth of few 100s MHz while the voice coil has bandwidth of 10s kHz) I've not seen inside of a blu ray stepper motor but i think the linear rail is pretty much tied directly to the motor shaft, combining with the spring loaded "nut", i doubt there's any backlash at all. Adding gears would complicate the mechanical design and increase backlash. To increase the resolution, as someone has stated, the voice coil itself can be used as an actuator.

[@TankR]

@@longpham-sj5sv Well to be fair, in its normal job it just has to read digital data, and has the advantage of error correction. The circuitry wasnt meant to produce an analog output value from the photodiodes other than for focusing correction which, again in normal useage is an automatic feedback loop. Also we need to keep in mind BR optics are inherently astigmatic. IIRC they also use two side beams (split from the main of course early in the optics path) which either keep a bead on the next and previous track, or just inside them in the smooth area (if the laser wonders of center one side or the other will start or stop flashing as data pips interact with the light). These tracking spots have their own photodiode areas on the sensor, I wonder if they could be utilized to ensure the entire assembly were square to the sample, or at least the stage....... But yeah I totally agree, utilize the positioning coils more. AND SET THE STEPPER DRIVER PROPERLY!!!! Dude needs to stop winging it before they cook the damn thing.... I have a lens block from an old handy cam Im playing with for the zoom functions, and Im just direct driving the coils from an arduino. Just driving one coil, one polarity shift at a time. Easy peasy. The original project from years ago I saw on hack a day.....and the second go by someone else that HaD so professionally pretended was the first time it was seen ever....I believe used linear mag coil drives to move the stage. So as long as the voltage output steps of a DAC kept the area of interest within reach of the lens all they needed to do was county up in binary to move the stage. Linear drives that small sound expensive, eh? Nah, they just used some speakers from an old bluetooth boombox! And some skewers from the woofer cones to the side glass! LoL

[@knickly]

Some researchers are using piezo buzzers for their SPM stages. A really interesting option is slip-stick piezo stages, which allow nanoresolution with long travels (there are some open source printed designs).

[@TankR]

The original project years ago used speakers to translate the slides in XY (it also looked up from below to remove any alignment issues ensuring the laser lens was square to the sample plane). The arduino would drive the voice coils through some basic circuitry and depending on how much voltage was passed through them they would extend or retract the speaker cones. Attach some chop sticks to the cones and the stage, BAM! automated stage ;)

[@johnmorris1162]

If there is a backlash problem between the screws and the carriage nut, you could try adding a second nut with a split washer or some washers with a little give in between the nuts and adjust the nuts to balance least friction against least backlash.

[@DoctorVolt]

I solved that by rotating the lead screw a half turn before the actual scan begins.

[@dh2032]

some sort spring, or at the scale where working at here, a bit sponge or small rubber band, the one's out of the anything will do?

[@medienmond]

Get yourself a TMC2209 stepper driver (StepStick), which has smoother motion and ist more quiet. Also the microstepping is more relieable and the motor does not need as much current. In addition you can set additional parameters via a UART Interface to the driver.

[@TankR]

No, go analog. Grab some speakers from a bluetooth boom box, and set up a DAC circuit so that each increment displaces the voice coil on the woofer just enough to keep the area of interest within reach of the laser, also utilize the focusing coils more for positioning.....like how its supposed to work.... Then some tooth picks or skewers glued between the woofer cones and the slides and you're off to the races. Also, flip the thing over so the laser is looking up though a cover slip.

[@vrmadlab]

an optical probing microscope is crazy for the budget. Well done !

[@DatBoiOrly]

11:48 i think the reason it has those washed out lines is because it isn't truly level so it's scanning it on an angle so it's picking up side information making it washed out

[@SphaeroX]

Great project! Let's start improving and building a PCB with all parts and sell it, perfect for Researchers!

[@Krishell]

Thanks!

[@DoctorVolt]

Thank you so much!

[@morkovija]

this is top tier levels of geekiness

[@markr9640]

Excellent video. I have many years of experience with stepper motors. If you want smooth running, it often helps to add a fly wheel. 'Not a lot of people know this'

[@pvc988]

1. I wonder if it would be possible to recover height map using focus error signal. 2. The problem with horizontal lines might come from leadscrew/nut tolerances. Try preloading it with a spring. 3. Voice coil instead of stepper motor could be more precise. 4. Could this setup be used for microphotolitography?

[@HeadCodeMonkey82]

Incredible work, you have clearly learnt so much to be able to produce such an amazing result

[@SIM31r]

Interesting microscope. It is better than usual optical microscopes. Because it can create 3D images. I think it is possible to modify the program so that the result will not be a 2D picture, but a 3D model that can be rotated in a 3D editor (to show the 3D structure of the material). The most interesting objects for scanning are insects, a fly's head, dragonflies.

[@shadeydave]

Spherical aberation toward the edges. You can compensate with software, or close crop toward the middle where you need a clearer view. The horizontal lines are doubled up vertically likely because of the instability of the stepper motor at that resolution. You can compensate this by using a geared dc motor instead, perhaps..

[@Dangineering]

Whoa, this is the coolest project I have seen in a while, thank you for sharing!

[@amoedoricardo]

Congratulations, great project!!

[@hamid.s.zolfagari]

hi. thanks for the video. you can use gearbox configuration with 1:3 ratio for Example to get higher step per mm and increase he amount of torque you need to rotate the shaft.

[@_droid]

Very nice work! This is something I have wanted to do for decades but never got around to experimenting with it. This information will be helpful!I I see those same blurs when looking at a calibration slide with a reflective light source. I suspect a transmitted light like a traditional microscope won't do that. It's due to the markings being on the wrong side of the glass compared to the light source.

[@АлександрЮжанин-в2ч]

You can try to install optical sensors of stepper motors for a better gradation of the focus points of the location. for example, as in a laser printer or an old mouse with a running ball . But it is generally possible to approach cardinally , by definition, the sine wave of rotation of the magnetic field as in the video head of a VHS cassette recorder . Alternatively, choose a small stepper motor with interchangeable shafts with a smaller thread pitch, but this one will have to be made of wear-resistant materials and a special pen moving along the thread .

[@fox.9879]

12:11 when you under power a servo it jitters it might be a similar issue in that image

[@TheRainHarvester]

😮put a small capacitor on the pwm pin to smooth the ripple?

[@zheyingchin4194]

Try using a second OPU for the y-axis. Then you will have coils for both axis

[@DoctorVolt]

Yes, but it would be difficult to place the specimem on it, because there is not much space and it cannot carry much weight eiter. But it these issues can be solved somehow, this would be a good solution.

[@soulslip]

Doctor Volt sounds like a Spiderman villain name, cool video

[@rougine]

I have to optical focus on glasssubtrates at work. I have to deal with internal reflections and see all patterns more than once. If you look in an angle from above you see the pattern at least twice but faint. Maybe that is what you are getting here too. To test for this you can try to focus on the reflections. If there are none it is something else.

[@nikscha]

I believe a compliant mechanism could solve your stepper issues. I'm thinking some kind of lever mechanism.

[@AlexSouthwell]

Amazing! I loved it. I want one! I should see what old drives I have

[@jimbotron70]

What about using a CRT as electron scanning microscope?

[@marcdohmen4699]

very nice project. I just build the hardware, but I can't get any connection with the website. Why?

[@Scar32]

instead of a higher resolution for each scan you could instead have it so that you could zoom into an area 2X but keep the same resolution so it scans just as fast but zoomed in to a smaller area. then you can save the 960 scans for special occasions, plus I have done multiple calculation projects that takes around 4 hours or more to complete with no progress bar, the trick is to do it overnight since you already waste like 10 hours doing nothing

[@GregCoonrod]

How did you determine when your subject was "in focus"?

[@DoctorVolt]

It is in focus when it reflects a maximum of light. I can this on the LED or the indicator at the bottom of the HTML site.

[@andrewanderson8803]

Why not try a piezeoelectric stepper?

[@across8339]

Just found your channel and what a fascinating project! The motion of a stepper motor is very abrupt and the double horizontal scan lined suggest t you could getting a backlash bounce between the threaded rod and the nut. You could fairly quickly test this by creating a simple pull back force on the carriage (spring?) and a small delay to settle. This will slow the scan but improvements in the vertical resolution will vastly improve the quality, possibly more than your current high resolution setting.

[@Dhirajkumar-ls1ws]

You can also do direct laser lithography

[@emil4390]

Why not to try DC motor with reducing gear box insted of stepper motor

[@nomanshah6761]

Primax venus fuzool hai Primax galaxy ka batayern please q k voltronic based hai

[@Stopinvadingmyhardware]

Reduction gears and belts to smooth them out.

[@DonCarlione973]

Neat little project 👍🏼

[@mantookaibarta3073]

Nice Explanation.... Hope to see more videos like this.... Thank you for sharing your ideas.....

[@henryjulianmurillogutierre3421]

i’m so jealous 😂 i would like to have that level of skills, this project is insane a more detailed series on the reverse engineering of the laser controller and software coding would be much appreciated i’m looking forward on being able to tepalcate this cool project

[@rushbhavaiya1714]

When you start ? , becz I am also looking to start. lets connect !

[@RyLeedepressed]

Have you looked into nema 8 linear slides? You can find 0.9deg step angle versions for around $50

[@DoctorVolt]

I checked several options, but I didn't want to spend that much money.

[@reoproedros]

why there is a different focus point between scans ?

[@PeaceToAllEvenYou]

For some reason, your accent tempts me to address you as Herr Doktor 🙂 I'm really impressed with this! You achieve high resolution on such large samples. As it's proving difficult to move the sled precisely, and you know that the cylindrical lens provides an elliptical focus on the detector array - did you know that the tracking servo also uses the left/right balance of the four diodes in that central detector array? ie, you already state that, given the diode array {A,B} {C,D} you sum A, B, C and D to get your main signal And the focus error signal is the difference between (A+D) and (B+C) I'm guessing that you already know that there's a tracking servo signal in the difference between (A+C) and (B+D) So what I'm excited about trying is this: Try to use the optical tracking servo to find the resonant frequency of the optics... If it's able to resonate, use that as a final-stage scanning option instead of depending on the precise movement of the sled. Examine the signal which could be useful in resolving small details - but maybe hard to use for large samples... in any case, fun to try. Best wishes!

[@DoctorVolt]

You mean that small stepper motor in the OPU that moves the collimator lens? I was always wondering what this is good for.

[@hvanmegen]

​ @DoctorVolt We'll call you "that other guy with the swiss accent" from now on.. (with Andreas Spiess being the OG (original guy) with the swiss accent)

[@my.own.devices]

This project is a major contribution and I give you huge props. I had the same idea 6 or 7 years ago but the interfacing and output part was too daunting for me. You've really done a fantastic job with the ESP application 👍👍 I always wondered if it would be possible to operate it in confocal mode, i.e. have the focus follow the surface contours as it scans. In this way an image much like that from a SEM might be obtained. The potential for low-cost, high resolution student microscopes, leveraging the incredible abilities of these laser modules, is huge.

[@weirdsciencetv4999]

Use a centrifugal fan blower module to blow directly on motor

[@obsurdhotdog8442]

would add vibration to the carriage. bad.

[@tech.curiosity]

Why dont you use a mechanical hard drive head to make the movement? I believe you will have much more precision and faster scaning.

[@nickpelov]

Are you using the electro-magnetic deflector? I think (not 100% sure) there is a el.magnetic coils that can deflect the lens for fine movement. when you pass current through the coils. It's also possible that if you don't pass current the lens will vibrate and cause distorsions.

[@DoctorVolt]

Yes, the lines are being scanned by the coils

[@paravedaparaveda.5308]

piezocrystal for pozitioning

[@khatharrmalkavian3306]

Could you use a couple gears to adjust the effect of the motor? I haven't worked with them so I don't know what causes the overheat. Is it the constant operation or is it that it doesn't like running at that speed?

[@DoctorVolt]

It's because it needs a high torque to drag the sled.

[@thefruit]

Warum verwendest du nicht ein Größeres Zahnrad das zu immer kleineren Zahnrädern führt, dann könntest du vielleicht einen größeren Motor verwenden oder? Ich hoffe du bleibst dran an dem Projekt und schaffst es eine Anleitung für ein günstiges Hochleistungsmikroskop zu erstellen :) das würde vielen ärmeren Ländern helfen, die zb in Krankenhäusern eines bräuchten (aber ich kenn mich nicht aus, vielleicht ist es garnicht für organische Materialien geeignet)

[@hi-tech-guy-1823]

You should of used the Voice coils more The optical Pick Head has a few of these To fine Tune the optical Pickup Lens

[@su2so3]

Have you considered using a flexure instead of threaded rod for the fine axis control?

[@seditt5146]

I wold think some form of Linear motor would be much better than attempting to use the worm gears to convert rotational movement into translational as you will always get some losses that way. Hell, you already have everything needed to make an Atomic force microscope and see atoms by replacing the stage with piezo controlled movements and a sharp pin to detect the electric fields

[@largerification]

Can you increase the dept of focus by applying autofocus on the go?

[@francescocontalbi1222]

Very nice!!! You are a creative engineer

[@yt4krist0f]

Great results!

[@critic1236]

Great work. Thanks for sharing.

[@marianodiaz461]

This is a very interesting project, and I would like to make it myself , did you made any PCB to interface the laser ?

[@DoctorVolt]

No, I built the electronics on perfboard.

[@marianodiaz461]

@@DoctorVolt is there any chance that you can take a photo of both sizes of the perforated board assembly? I’m not to good following schematics in a perforated board

[@DoctorVolt]

Ok, I will upload the photos to Github when I find the time.

[@gibbyrockerhunter]

Totally guessing here. But if the blurry artifacts are not from software, reflection or refraction from stray energy pollution...id bet the the issue would be from the physical and digital X an Y axis are not perfectly square. Obviously there are a few factors so no clue for certain. Great build. My electron microscope from 8 years ago looks like a mad scientists joke by today’s standards. Lol

[@MarkxTube]

Hey great work. Have you ever heard of the OpenFlexure project? This could give you the fine control you need without the overheating problem. Next you need a vertical axis and some fluorescence dye ;)

[@DoctorVolt]

Would be interesting to know how accurate it works. As far as I can see, it functions with stepper motors and 3D-printed gears. So I guess that this highly depends on the quality of the 3D printer.

[@MarkxTube]

@@DoctorVolt True, however your drive is quite direct from motor to shaft. If you come up with something that increases travel distance but reduces movement and works against a slight resistance you will get a much smoother ride. I think that was the principal behind the openflexure. But it has been a long while since I had a look at it.

[@SergeyPoptsov]

Is it possible to continuously accumulate data from the led sensor to differentially get images of the areas that had just included in the spot and leaving it? You could take images every 120nm to obtain a lower resolution than spot size of 480 nm.

[@DoctorVolt]

Maybe...

[@chrisarmstrong2721]

Would you make this for a privet buyer? If so, roughly how much???

[@qweqwerty7537]

Sir, very nice work. Where do you find a motor with a ball screw pair? Can you give a link on your motor? Please answer.

[@qweqwerty7537]

Stepper motor with lead screw, sorry, i mean this device.

[@DoctorVolt]

I bought it on Ebay, but don't have the link any more. Just enter "stepper motor with lead screw" into the search field and you'll probably find something.

[@Sulzart]

Woah, found your video as a suggestion from YT. Nice project. I'm looking for a way to scan/photograph open ICs. More as an art project than an exact reproduction. Have you scanned IC yet ?

[@patrickday4206]

Nice incredible work

[@Ben-ht5yf]

Great results. Maybe a dedicated adc or a analog sweep driver for the scan coil instead of the pwm would make the optics more stable? Or is it a mechanical limitation?

[@DoctorVolt]

Only the X direction is scanned by voice coil. This is ok. The problem is the Y direction which is driven by stepper motor.

[@Ben-ht5yf]

@@DoctorVolt some preload on the carriage may help, maybe? Impressive results nonetheless

[@DoctorVolt]

@@Ben-ht5yf Yes, I will consider it if I do a follow-on for this.

[@BobSimpkin]

Fascinating, thank you for showing us.

[@RobSchofield]

What about trying a NEMA-8 size motor?

[@DoctorVolt]

NEMA-8 seems to be powerful enough, however it requires a redesign of the base part.

[@talkingmetastasis607]

that's very interesting, I also want to build one

[@keithcarpenter5254]

How about a simple kit to adapt 3d printers into microscopes.......massive market already there..

[@meyemediauk]

i transmitted data via laser when i was a child , but the reciever had to be in a fixed position, the room had wooden floorboards , if i walked arround it would cause a vibration and lose connection , also made a laser speaker, transmitted a song to the output cool stuff

[@DoctorVolt]

Young Sheldon :)

[@Zalo29]

Boy, I've been thinking on your project and came across the lightscribe technology They managed to use the laser to etch the backside of a CD It would be interesting to know how they managed to keep the laser focused It should be useful for you

[@DoctorVolt]

I guess that they use the focus error signal to do that. But if I re-focus every scanpoint, scanning would take much, much longer than it takes anyway

[@Gabuhmi]

Super Projekt, schöne Umsetzung ! Ich möchte echt nicht zu nahe treten, aber ich würde vielleicht überlegen entweder KI für die Audios zu benutzen oder einen Sprecher outsourcen. Gibt günstige Freelancer. Video/Schnitt technisch finde ich es Klasse !

[@kredwol2103]

Why not just add a 10 to 1 gear on the motor you have?

[@CRAIG86]

i had that out dated samsung blu ray player, worked great to wach youtube intell it sttarted to refuse to connect to wifi

[@7LegSpiders]

Do house flies have toenails?

[@valentinoboi96]

Amazing video🎉

[@RyLeedepressed]

Impressive work

[@ifrite3]

how do you sound german and indian at the same time

[@cyantosolar]

Great project! Inspired to try something like this based on your work. Want to use an optical setup that has both red and blue lasers in one optical path. It seems like Sony BDPBX370 has optical head like this. Will your electronics control this player? or would you suggest another model that is readily available? Thanks!

[@DoctorVolt]

My electronics will control any OPU. I was using a BDP10G from a Samsung player. With other heads it's up to you to figure out the pinout.

[@cyantosolar]

How did you figure out the pinouts? And how would one go about figuring out pinouts for a different model? Thank you.@@DoctorVolt

[@DoctorVolt]

Have you seen part #1? There I show how I did it.

[@ricoreyes6044]

Very interesting project. I've been wondering what to do with the drives I have in my junk bin, and I think this is an awesome idea to work with. I wonder if adding a little counter-tension (maybe a rubber band) would pull the backlash from the drive. Also, is there any chance there's vibration or external influences causing the positioning noise?

[@DoctorVolt]

I'm quite sure that it must be caused by vibrations. Waiting some microseconds between moving and capturing the value helps there, but also taking the picture takes longer, then.

[@ricoreyes6044]

@@DoctorVolt is the sensor output stable and clean if you travel continuously instead of discrete steps?

[@valeriydergachev]

cool i want this tool :( good work my friend!! its real cool !!

[@lockdown9402]

It does not work this way.... you cant get 150nm resolution. The blue ray spot is not a spot and niether is it 150nm wide. It is a diffraction pattern and looks like concentric bright and dark fringes...... Yes for a CD where the tracks are a set distance apart you can get that resolution because you know here the dark and the reflective white spots are or are expected to be. So every consequtive track is around 150nm or less apart but every adajacent bit on the same track have to be spaced out further because that is the information you dont have and you are trying to find out. However when you are imaging something that you have no idea what it actually looks like and does not have a set periodicity you cant make that assumption. The best angular resolution you can get is still around 200nm ( theoritically). As a reference I would like to point out the book by James Pawley "Handbook of biological confocal microscopy". Apart from the resolution claims, this is a very nice device and clever use of electronics. I and amazed that you were able to find the relevant pins from the ribbon connector and these things rarely have a datasheet. Great job.

[@peterrichards1989]

Can you scan the surface of a CD and show how it looks?

[@DoctorVolt]

I tried, but I could only see dust and scratches on the surface of the coating, but not the pits and lands.

[@BrunoTorrente]

@@DoctorVolt I was interested, because an open source OPA platform could emerge from your work, of course today readers have become commodities, but they are stagnant, several concept tests could occur or even replicate cases such as the use of femto lasers (similar to the silica project) for data recording. There are preservation projects that need such technologies but are limited by closed hardware/software.

[@eVITORIOe]

maybe you need to apply lens correction

[@noanyobiseniss7462]

This earned you a Big ars SUB and wish I could upvote multiple times! :)

[@gerd3136]

Wow, Hammer 👍🏼😎!

[@MrAlex3461]

Good, great! Great great! Great Scott that's good great

[@trinocerous]

You're awesome! Thanks for sharing!

[@RonLaws]

So, would this be called a Scanning Photon Microscope 🤔

[@001Nero001]

Dont use microstepping. use mechanical reduction

[@faevara]

If it gets hot during operation, could you operate it in a freezer? Edit: My logic is that if anything is expanding due to the heat it would cause an adjustment to the measurements and scans.

[@DoctorVolt]

I can also cool it down with coolant spray or dry ice

[@chrisgj198]

For some mechanical ideas, you could read the article about making a ruling engine, in The Amateur Scientist column of Scientific American magazine, April 1975. This machine moves a carriage with accuracy about 0.000001 inch or 25.4nm, which should be good enough.

[@janthran]

you could also use an array of 4 different lasers to achieve higher resolution on the same area

[@hvanmegen]

Hey, another other guy with that swiss accent!