Lasers Make PCBs The Old Fashioned Way

There are many ways to create printed circuit boards, but one of the more traditional ways involves using boards coated with photoresist and exposing the desired artwork on the board, usually with UV light. Then you develop the board like a photograph and etch it in acid. Where the photoresist stays, you’ll wind up with copper traces. Hackers have used lots of methods to get that artwork ranging from pen plotters to laser printers, but commercially a machine called a photoplotter created the artwork using a light and a piece of film. [JGJMatt] sort of rediscovered this idea by realizing that a cheap laser engraver could directly draw on the photoresist.

The laser dot is about 0.2 mm in diameter, so fine resolution boards are possible. If you have a laser cutter or engraver already, you have just about everything you need. If not, the lower-power laser modules are very affordable and you can mount one on a 3D printer. Most people are interested in using these to cut where higher power is a must, but for exposing photosensitive film, you don’t need much power. The 500 mW module used in the project costs about fifty bucks.

Of course, once you draw on the board with the laser, the rest of the process is like it always has been. Develop the board, etch, and all that. We wish the laser could drill the holes as that’s the part we hate the worst!

We’ve seen powerful lasers just cut boards, of course. You can also forego the photoresist and just let the laser burn off a coating of paint.

46 thoughts on “Lasers Make PCBs The Old Fashioned Way

      1. Been there, on a 3d printer. Other than figuring out how to get the two layers to align by using the 3d printer to use a Dremel to drill the 0.6mm VIAS, using G92 to zero the drag engraver to start the iso routing, I had a 3d printed board holder with 2 locations for 5mm socket head bolts. Next challenge was to arrest bed movement vs Y carriage movement between ISO routing the two layers. My bed was mounted to the carriage via long screws and springs. Typical for an i3 clone. Made brackets to allow bed leveling but arrest any other motion other than up/dn in relation to the Y carriage. Also needed to add a threaded rod diagonally to compensate for XY mechanical skew to get the tram equal.
        End result after 2 months of ripped out hair and a few scrapped boards?

        1 nearly perfect 14x14cm double sided PCB. Side to side via accuracy +/- 0.25mm across the whole board.

        Just send your boards out.. they will be much better looking and you won’t have the non plated through hole soldering problem.

    1. Why always the nay saying. I made a PCB using a similar method, but by spray painting the blank PCB and using a diy scribe attached to my 3D printer. This scraped the paint off and the etch was successful. Even had vias as small as a 10th of an inch. Alignment is actually very easy if you know what you are doing. I will never go back to dry film again. I would be purchasing a laser to replace the scribe soon.

  1. There is now open source / free software to control the type of fiber laser used in one of the linked articles: https://www.bryce.pw/engraver.html

    These lasers are getting cheaper ($2200) though still a lot more expensive than a diode laser. The spot size is about a quarter of the resolution claimed in the article, nevermind the degradation from the theoretical resolution you get with etching. Etching is certainly doable for THT and low-density SMD, but for high resolutions it is hard to do repeatably without a fair bit of equipment.

    1. Depends on etchant. I used to get very fine results with dilluted ferric chloride. 0402 pads could be done repeatedly, 0201 was becoming a hit-or-miss depending on quality of paper used, even toner made a difference as refills were more unpredictable.

      1. Yup. It’s a misconception that 1064nm can’t machine copper. Even a 20W q-switched fiber laser can do it nicely. It does tend to lightly blacken the PCB, but most of that is just copper or copper oxide particles and much can be wiped away. The substrate remains nonconductive (as verified by my megaohmmeter at 500V)

    2. The real issue is that regular dry film photoresist has a hard time going sub 85 ųm track/gap. There are a couple of solutions out there, but the material availability is prohibitive for DIYers. You also need to switch to a copper chloride etchant, dehydrate the photoresist post exposure, control the development cycle accurately, etc. I doubt it’s worth the effort to optimise the process at home. It took us several weeks to get it stable on professional equipment with high-quality film masks in a cleanroom. Once you go sub-100ųm dust, humidity and temperature swings become your enemies.

      And anyway, illuminating PCBs directly with LDI systems is pretty standard in industry, contact lithography became exceedingly rare except for very large volumes or when pushing the edge of what’s possible. Of course the professional systems have more power and scan using an array of galvos.

  2. I have used black paint sprayed onto the copper board and etched with my Ortur laser and the results are stellar. Quick chemical etch with muriatic acid/ Hydogen peroxide bath for just a few minutes. Done!

      1. Nope. You need a resolution of 300 dpi or better for a decent PCB. Also photoresist wants 400 nm light with substantial power. And of course, the image must be perfectly focused and distortion-free. Now guess why advanced hobbyists still make contact copies with an old-fashioned UV lightbox…

          1. This will only work with very crude PCB designs, scattered light in the LCD’s front glass will blur the shadow. Useless for tiny stuff like MSOP-10 with 0.5 mm pitch. With the classic contact printing method there is no such scattering, because the ink or emulsion side of the mask is in direct contact with the photoresist.

      1. Have you tried making a pcb inductor?
        Perhaps you could try stacking bifilar coils in an alternating stack of pancakes? So each layer also tries to counter it’s own self capacitance. I’ve always wanted to try that with copper wire, but I’ve never had the funding or patience. It seems like a 0.1mm track could in theory make that a lot easier.

        1. Vladimir Trondin’s build using a polygon mirror LSU from a Samsung laser printer – easy to build as you only need a single moving axis and no backlash issues to deal with as the axis always moves
          in a single direction.

          https://youtu.be/zA2SxA9U4ys

          Andry Black published an opensource Raspberry PI + STM32 based high speed
          laser PCB exposer similar to Alphacrow’s HLDI which uses the guts of an inkjet printer including
          the high res optical encoder tape which inkjets use.

          https://youtu.be/Fx7F5YykWQc

          https://easyeda.com/andryblack
          https://github.com/andryblack/pcb-printer-host
          https://github.com/andryblack/pcb-printer-firmware

          1. Also a nice solution. I know a few guys that built this 7 years ago and this inspired me. They had issues getting the focus equal over the full with as the lens Was not made for Blue Light. Anyway, for small Boards this is no issue.

            I’d like to look into reusing LCDs and UV sources from SLA Printers.

  3. I might try multi-layer PCB process with a similar technique. I’m still fighting layer alignment better than +/- 0.25mm and cannot through plate holes. Problem is, hard to beat OSH or PCB-way.

    1. That is the conundrum these days. Professionally made boards are cheap, but you may have to wait 2 or 3 weeks. I have actually never had a need for a printed circuit board “now”, and if I need it now, I perfboard, breadboard or “dead-bug” or something like that, and use it until the nicely made boards come in. I have made single layer boards for fun, and have used them in permanent projects, but have not really ever found it necessary, as my time to make one from scratch is often worth more than just buying nice ones and waiting while I do other things.

  4. I tried the same with K40 laser cutter, but it didn’t work out.

    Even at the lowest power level at which it would lase it was too powerful and would burn the resist instead of exposing it. I guess being IR means exposing wouldn’t really work in any case. Normal developer did nothing but fortunately the burned resist dissolved in isopropyl alcohol faster than the unexposed resist.

    But fine resolution was not possible because the burned edges were kind of flaky. Here is a PCB with 0.5 mm pitch IC and 0603 passives: http://jpa.kapsi.fi/stuff/pix/k40_pcb1.jpg
    And for comparison here is a PCB made with transparency film and traditional exposure, but same etching process: http://jpa.kapsi.fi/stuff/pix/valotus_pcb.jpg

  5. Yes, this approach could work.
    Incidentally I found that “cooking” the resist works almost as well, though not as reliable if you keep the board at a controlled (low) temperature it can work to prevent blurring and resulting low resolution.
    Active cooling with a Peltier can be effective here and ensure repeatability.
    For something like this it doesn’t matter what the wavelength is, the side emission from a 445 or 450nm laser is more than enough and only needs <30mW so a PL450 will do at a pinch if focused well to ensure a minute spot.
    These aren't cheap but filtering is comparatively simple, and an OD5.5+ (red pass) goggles set with additional measures like running it in a light tight box and use of a CMOS camera + remote display will ensure a margin of safety.
    I found that the sort of diodes used in PS3 or HD-DVD modules DO work but only if you use parts from the sled itself, such as the waveshape plate mounted in low melting point alloy to clean up the beam.

  6. Why not:

    Print artwork on laser printer compatible transparency sheet
    Position over sensitized PCB
    Expose with DIY UV LED array, or a cheap UV (finger)nail lamp

    I’ve got one of the $70 (years ago) 1W laser engravers typically of the NEJE “brand” (with many clones) which uses CD-ROM laser transport mechanisms for laser movement (which is why they were so cheap). I’d like to avoid the need for photosensitive resist entirely and have considered burning off a thin, flat black lacquer coating on a PCB instead. Anyone tried that?

  7. Have a somewhat damaged but working LS drive (busted sled) here, so a minimal modification should do it.
    Incidentally, the biggest problem with these is bypassing the eject signal, so the resulting image can be etched multiple times.
    Yes you can use a LS CD-R as a negative, though again there are faster ways to do this.

Leave a Reply to Bryce+SchroederCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.