With the coming of very cheap blue laser diodes, PCB fabrication has become increasingly interesting. Instead of making a photoresist, placing it over a piece of pre-sensitized copper clad board, and putting the whole assemblage under a blacklight, it’s possible to put a photomask on a board with a tiny bit of very blue light. All you need is a CNC machine. A 3D printer can be a very precise CNC machine, and when you combine these two ideas together, you can make printed circuit boards with an Ultimaker.
[Geggo] had the idea of attaching a blue laser diode to his Ultimaker to burn a few traces into presensitized copper board. With a 3D printed adapter, he was able to mount the diode and associated electronics right on the extruder body. With a small ring to tighten up the aperture, [geggo] was able to put a 50 micrometer wide dot of light on a piece of copper. The laser is powered directly from the PWM fan output on the printer controller board, allowing this entire mish-mash of cheap electronics to be controlled via G-code.
A few experiments were necessary to determine the correct speeds and power settings, with the best results being 1000 mm per minute at 40 mA. The finished board looks fantastic, and a few minutes after [geggo] was done etching a board, he started using his 3D printer as a printer. It’s a result that is so good, so easy to accomplish, and requires so little effort it makes us wonder why we don’t see more of this.
58 thoughts on “Etching PCBs With A 3D Printer”
50 micron is crazy narrow for home etching, let alone anything with a bit of voltage on it.
Anyone know of a program that will raster a Gerber (or Eagle) file and output Gcode?
Gerber was originally used for photo plotters with aperture wheels determining trace widths.
There is also a ULP program for EagleCAD, but I like this group’s work on a generic solution:
ULP program for Eagle (CNC island routing exporter):
I’ve tried this PCB2Gcode converter. Does a decent job. You will need to do fine tuning and ensure to keep spacking between routes a bit loose (Eagle also has an app that can warn you if they get too close). Keep also in mind, that the holes won’t be drilled, so you either need to rig up the 3D printer mechanics to allow for this, or take it to a router. I routed PCBs with a v-bit, then had them drilled. Works great, but will take a lot of fine tuning, especially bed leveling. I am very impressed with the narrow widths – difficult, if not impossible to do with a decent router and v-mill bit.
I used to use linegrinder (windows/freeware/very basic) to turn my gerber files (from KiCAD) into G-code, Now I use Flatcam (Python based with GUI, annoying at the beginning but does actually an exellent job and allows tweaking/adding/repairing traces, importing multiple files etc).
I use FlatCAM: http://flatcam.org/
That is very impressive, seeing an idea discussed on HAD become a reality so quickly is very satisfying.
Not very useful without a drill to make vias. Only limited to single sided boards. I’ll stick with OSHPark or China.
Obviously useful for quick hacks, where you don’t want to wait for shipping…
Not true. You can still drill the holes later and with proper alignment strategy and planning make very good doublesided boards. Plated through holes and vias are still problematic, but the only good solution for these has been board houses anyway. With this and some colored UV cure epoxy you can make your own solder mask and silkscreen layers as well.
1- He’ll have three revs of his board tested before you get your first back.
2- Why wouldn’t he be able to do double sided?
3- Do you think he made this project to replace overseas PCB fab? This is for rapid prototyping not runs of professional boards.
You wrote a disparaging post about a cool project on a site dedicated to showing off cool projects.
I mean come on! It’s got blue lasers and 3D printers in it! Throw in an Arduino and we would have reached maximum coolness on HaD.
Most 3d printers are arduino mega2560 based. Bam!
Absolutely. I’ve done all version of making/buying PCBs and this one blows me away – the routes are mega fine. I can’t do that with my cnc router (10mil routes and spacing is about it). Drilling the holes – well a router really helps there. I envision a 3D printer with dual extruders converted to replace one extruder with a laser and the other extruder by a drill. (am building my dual extruder Cobblebot – I call it Hobblebot) and it’s very tempting to do something like this.
+1 for the smackdown
To reach Maximum Coolness it would also need a 555
It’s meant to replace the old UV light through translucent paper method, or toner transfer, or whatever other light-based etching method. Not a bloody factory in China. Which can’t offer the sort of production time this thing does.
When you make your own PCBs, which is what this is for, you drill your own holes. It’s not hard.
Even if it were limited to single-sided, and I don’t see why it can’t do double, that’s still plenty enough for many things. The guy’s done well, and it just needs a new head on a 3D printer. If you have a 3D printer, you might well benefit from being able to do your own PCBs. So it’s a good idea.
Your comment is not a hack.
How about retrofitting one of those Silhouette paper cutters? They’re much cheaper than a 3D printer and it’s really easy to fit custom cutters and pens into the carriage.
Any XY style plotter with sufficient accuracy and repeat-ability should work.
If you can find one. They were expensive back in the day. My school had one. I think mostly they sold them with very expensive CAD equipment to companies doing draughting, etc. I dunno if it’s the sort of thing anyone would throw out, at least in any sort of repairable condition. And there were so few sold to start with.
Hm, the Atari and Commodore 8-bits had plotters though, little ones, I think paper-roll based. Wonder if you could adapt one of those? Mess about with the roller to drive a flat bed.
The Silhouette Cameo is already cheap: $269 on Amazon. And their Portrait and Curio models are even cheaper.
Their parent company makes commercial vinyl cutters for large signs, and those do cost thousands of dollars. But I’ve never needed a circuit board that big!
YES. Oh my gosh, yes. They have a solenoid on the head that lifts and drops the stock blade, which would be a cinch to retrofit into an on/off control for the laser. Maybe even just put a SPDT switch so it can go back to pen/blade mode… My Silhouette has pretty much been gathering dust lately, and I’m starting to get into photoresist etching for decorative brass. Off to buy a laser diode, I guess! Thanks for the idea, Chuck.
Custom pens and cutters yeah. A chunk of stiff PCB rather than a flexible piece of paper, less yeah.
Actually they use a fairly stiff carrier sheet under the paper. More flexible than a PCB, yes definitely, but it’s not like a printer where it has to go around any corners. If my first-gen Sil is any indication, a normal 1-2mm board should be fine.
Tried this with a cricut, which is similar (and more of a crappy company, don’t get one, I got mine just as they started their legal FU to everyone campaign.) I still have software which works, though I donated the machine to the local makerspace.
It doesn’t produce *good* results doing any sort of cutting/drawing (sharpies). Though with a laser it might work fairly well.
It’s not a bad idea, although for the money, might be better shelling out the extra for a 3D printer, then at least you have one. If you already have a paper (/ vinyl) cutter, it’s OK, assuming the wife doesn’t mind you chopping up her expensive crafting machine. I wonder if it could cut paper or vinyl with a strong enough laser? Then it’d be an upgrade!
I wonder if a Silhouette, or equivalent QVC-style card cutter, could be re-purposed into a 3D printer? You’d have to add a Z, obviously. Wonder if it’s accurate enough? Wonder if they do cheap Chinese knockoffs? I’d guess, not yet.
both links to g3gg0.de return empty html document, other wordpress posts on his site work fine, did you kill his server or something?
also https for g3gg0.de returns bad cert
These links don’t work… His homepage works, but not the detailed posts.
work fine for me and also some other people i asked.
if it doesnt work, try on wayback machine:
you have some stupid wordpress plugin installed that detects high load and kills specific blog posts, both links returned empty html body elements for at least couple of hours
you also use bad certificate
I really want to see someone cut out the middleman and put a powerful green laser on to remove the copper directly. It’d have to be crazy powerful with very precise timing, but otherwise a similar setup, plus eye protection.
Why shouldn’t this one have eye protection?
Look at this devices – probably not cheap – but doable: http://www.lpkf.de/produkte/rapid-pcb-prototyping/laserstrukturierung/laser-strukturieren-leiterplatten-prototypen-kleinserien.htm
Having talked to people that tried this on a laser cutter. The problem is, when provided enough power to burn away the copper, burning away the board itself only takes a fraction more power.
Normally a laser cutter uses a CO2 laser. This wavelength is reflected very well on copper. So it is not good to remove it. You need green or even UV light and very short pulses. But of course a frequency doubled Nd:YAG pulse laser is not exactly cheap.
interesting, so with good tuning, you could burn through holes as well…
Why no eye protection?
I’m too afraid I’ll get the wrong eye protection, for the wrong wavelength. Or.. hey.. I’m cheap. Maybe it will be advertised as being for the right wavelength but actually not protect my eyes worth a crap.
I think that if/when I try this I’ll just build a woden box to place over the printer. I’ll get everything queued up and cover it before I hit the start button. At least with this I won’t have to worry about a nozzle getting plugged up or anything like that while I can’t see it.
Maybe I’ll even add a switch that the box sits on. The laser does not come on if the switch isn’t pressed. Yah.. I think I’ll do that for sure. Two switches even!
Reminds me of going retro. A modern version of a Gerber photo-plotter. If you could use the laser through some optics with an LCD aperture (or LDP mirror) you would have a Gerber and create the set of apertures needed. Never need to translate Gerber to something else again :-) So, the mirror from a tiny projector, a few quartz lenses and a blue laser. Hmmm. Like 3D printing, probably not long before a print head is be very affordable.
Wonder if the Watercolorbot (Super-Awesome Sylvia’s creation) could be fitted with a cutting head, like a Dremel flex shaft tool & a burr, to remove copper from a prototype PCB?
Totally depends on what kind of circuit you want to make. If you are aiming at something with less than 0.5mm trace/space width, probably not. There is a good reason why you need a good spindle with proper bearings (minimal runout), a drive mechanism with zero backlash, and high position repeatability under load. And not to forget you need to have the right milling tools, the frame should eliminate resonances and so on, and so forth…
It looks easy when you look from far away, but to get a decent board for todays smd-components pin pitch and size, you have to have some precision in your mill.
Linked site is not responding (blank page). 01:07AM UTC 04 Mar. 2016. Could be the HaD effect. Could be just that WordPress remains the crap that it is.
Can we put a small DC motor with a collet and a micro servo in there too? no big holes bit big enough for through holes or pilots? Just for drills, not traces.
With a combination of laser photoresist, etch, then coming back for drills and a solderpaste mask from plastic – if you can get something to stick – this would be a nice package
Ironically we had this exact idea on 4HV, some folks have success with Bungard PCBs.
The best bet might be one of the Osram PL450’s as they are single mode and lower power than your typical >500mW Bluray burner but the longer wavelength means it should be less temperamental if you use a heated bed.
A bit of careful masking and this could work as is or just using an old rotary scanner from a broken laser printer with a geared motor for the X axis if you wanted to fit it into an old flatbed scanner which could then also with care do double duty as a PCB quality checker and copier/UV flusher once developed to get the resist off for later soldering operations.
Removing the resist was never a problem with some “cleaning gasoline” hydrocarbon solvent. There are also weaker 405nm diodes from Br Players (50 to 100mW). But I don’t see why somebody wants a weak laser. Just expose faster.
Although I would also like the raster exposure method with a polygonal mirror like a laser-printer. That remembers me of an old rotating mirror assembly sitting in the corner and some – unfortunately very weak (50mW) BR diodes sitting in the corner.
You dont want a more powerful laser , it doesnt simply expose faster. Most diode laser have horrible modes, and horrible beam abberations. Add to that , very poor optics (which most diode lasers are equipped with) , and your beam profile is toast. More power only exaggerates the issue ,causes light spillage over the edge of the trace, causing a poor exposure. You actually want as LITTLE power as you can get away with.
I have a Synrad galvo marking system that i have used to raster the image onto coated clad, both paint , and dry film resist. However, it actually ablates the material. Which works very nicely, and I dont have to develop the resist.
I also had a related idea to alternately deposit (presumably chemically) copper and photosensitive spray onto the pre-developed and etched board to do home multi-layer PCBs.
For something like this depositing an inverse pattern of UV cured resin though expensive would be a good way to get a flat layer, then seal completely and add self adhesive copper foil etc.
This is incidentally how the very early multi-layer boards circa 1963 were made, at the height of the Cuban Missile Crisis.
This has been being done by amateurs for quite some time. J Tech Photonics sells laser upgrade kits for 3D printers and other devices. See http://jtechphotonics.com/?product_cat=laser-and-driver-combo-kits and http://jtechphotonics.com/?product=laser-etch-bw-image-engraving-sw-license Last board I made with the was Saturday and have been doing work with it all week.
Reminds me of the Diyouware.com machine in some ways, but he was a bit more into it and actually designed the all-in-one machine from the ground up.
So is he using positive or negative photoresist?
*A positive resist is a type of photoresist in which the portion of the photoresist that is exposed to light becomes soluble to the photoresist developer. The portion of the photoresist that is unexposed remains insoluble to the photoresist developer.
*A negative resist is a type of photoresist in which the portion of the photoresist that is exposed to light becomes insoluble to the photoresist developer. The unexposed portion of the photoresist is dissolved by the photoresist developer
Positive apparently, can’t see any scanline artifacts in the copper areas.
Talking of scanlines, it said “rasterising” in the article. Is this an actual step? Turning it into a bitmap image?
Cos, why? PCB designs start off as vectors, and a 3D printer or a PCB mill works on vectors, why have a raster stage? Does that mean the 3D printing software has to re-discover all the vectors in a bitmap image? Cos that’s a bit, like, insane.
Unless I missed something here – the results are far less than I get with Toner Transfer.
I like the idea of laser / UV sensitive but that is because I expected much better results. People always say UV is better than Toner Transfer. I get 5mill traces with toner transfer.
The other thing is that it should be far easier to align (register) double sided boards. I am still trying to work out how to do that accurately with Toner Transfer.
What turned me off doing this is the safety issues for the LASER. I would have to make a full enclose for the diodes that were available when I last looked at this. The other thing is that there is no load for moving the LASER (no contact with the work material) so it should be fairly cheep to make a CNC for this – no load – cheap motors – no z axis – no ball screws – small bearing surface areas – cheap!
I hope that someone continues to improve this – it would be great.
I so much want someone to get this down pat! I personally don’t have any experience with the chemical / exposure / development process.
It uses the worst of both worlds: it scans the thing, so then why not do engraving? and then it uses optical so why not simply print on a transparency and expose which is a 2 minute job?
1. i press a button in eagle that will export tool vectors
2. i put the SD card and the PCB in our 3D printer and press go
3. i go and drink a coffee
4. then i go on with the standard procedure of etching
and this at $2 extra cost plus 2 days of enjoying doing that hack.
i don’t get why this is the worst of both worlds.
but i guess you are on your way to publishing a much better article of how to get it done in a much cooler way ;)
Don’t get it. Guy in original article is linking to 5mw laser. Is it enough??? Surly it can’t be! My 100mw 405nm laser will not even scratch that copper….
Am I missing something?
Am I missing something? Guy in the original article is linking to 5mw laser. Surely that’s not enough! My 100mw 405nm laser will not even scratch this copper….
Please be kind and respectful to help make the comments section excellent. (Comment Policy)