Custom PCB etching with the P1S

[dansan]

If you want to make your own PCB:s quickly at home using your 3d printer, I have primarily found two solutions suggested out there: fitting either a router (to mill the copper away) or a marker (to cover the traces with ink as in classic home PCB etching). I have dismissed those as unrealistic with my P1S, as the Bambu seems more rigid in its operation (less customizable compared to a regular reprap machine). This is what I like about the Bambu of course, but I’d also like to make some PCB:s.

The reason I’d like to use the 3d printer is because using a marker by hand is very rough, hardly realistic for surface mount circuits. Using photosensitive materials require some form of 2d printer which I don’t have, it also adds another step to the process. One option would be to get inspired by how resin printers work, however I don’t think I’d get such a solution going in an afternoon.

I just made a successful demo/test, printing TPU on top of a copper laminated fiberglass board. I used Bambu Studio for slicing. Some workarounds needed in order to print on top of the “foreign” board:

• First, print a pattern on the regular PEI plate, to help center the laminated board. Then tape it to the plate.
• Disable calibration. This makes the printer only check the distance to the middle of the plate, which is where I put the board.
• I printed at 0.2mm layer thickness, one layer in height.
• Understand that the purge line at the edge of the plate (which is printed just before the start of the actual printing) will be made in the air unless your laminated board extends there, so make sure to fetch it.
• PLA didn’t stick to the copper, which is why I used TPU.

I just successfully made some quick test traces and pads with 2.54mm pitch (for headers), 1206 and 5050 sizes SMD, with the 0.4mm hotend. Traces are 1.0, 0.8 and 0.6mm wide. I was able to confirm that the TPU does protect the copper sufficiently when etching (I’m using sodium persulfate).

Only issue is that the edges of the traces are a bit difficult to get sharply defined in certain places. This seems to be due to two reasons:

• The TPU creates a small barrier in the Z axis, somewhat inhibiting the flow of the etching solution. I’ll build something to agitate continuously.
• The TPU is a bit stringy. This can/will also obstruct the etching solution. Will see if I can tune the settings for this. Again, more rigorous agitation may help.

Because of this I left it in the etching solution for considerably longer than I normally would. This doesn’t seem to have hurt the actual traces noticeably, so this isn’t a significant problem. The TPU seems much better at protecting the copper than the ink from a marker, which is a great advantage.

I removed the TPU using my fingernail, no solvents or sanding needed.

So far this method seems promising, though currently it’s necessary to trim some edges with a knife afterwards.

Next I will try 0805 size with the 0.2mm head which I don’t expect to be a problem, I suspect that 0603 will be reliable as well. Just don’t put traces too close together.

In the attached pictures you can see that one 1206 pad is missing, I believe this would not have happened if it had had a trace attached, so I don’t think this is important. Also, the pads in this demo were only large enough for reflow; if designed for manual soldering, the pads would be perhaps twice the size (also alleviating this).

Anyone else tried this/have any experiences?

[JonRaymond]

I haven’t self etched a board in years but it’s neat to see people still toying with new ideas and methods. Thanks for sharing your findings! It will be interesting to see how much of a difference the 0.2 size nozzle will make.

[Destiny]

how about simply making a 1 layer thick masking stencil instead and hit it with some spray paint to create the mask on the copper?

or however thick you need it to be to maintain dimensions when applying.

[dansan]

Yes that’s also an option. Or, put it on top of a photosensitive board. I will consider this if I don’t find a satisfactory solution to the jagged edges, PLA doesn’t have that.

[dansan]

Although the finest/most accurate details would of course be with the original, ie with my first method. If that matters, I don’t know yet.

[dansan]

I have made some larger circuits since this post and thought I’d share my settings.

Although certainly a useful technique, there are limitations to what sort of quality to expect (see attached image). The challenges are avoiding the hotend from lifting up previously printed out areas, and stringing from the TPU.

I’d like to add that leaving it in the etching solution for an extended period of time is recommended as a workaround for the stringing. The TPU protects the copper really well, so that is not an issue, and leaving it in for longer helps etch the areas under any strings.

Next step for me is to try SMD. I expect 1206 and 0805 to work well, perhaps even 0603.

Nozzle: I use the 0.4mm nozzle because it seems Bambu advises against using the 0.2mm nozzle with TPU as it might clog. I will probably try this in the future.

Line width: Keep it wide, while considering your detail level. This is because each additional pass against an existing extrusion risks tearing it up. I use 0.6mm, and adjust my traces/layout accordingly (ie, my traces are 0.6mm wide). The 0.4mm nozzle is probably suitable up to a line width of 0.8mm, if you don’t require the finer detail level.

Elephant foot: disable, otherwise it distorts the proportions as we’re only printing the initial layer.

Resolution (precision): you could lower this value, if you are unhappy with the detail level as seen in the slicer preview.

Layer height: Keep this thick, and prolong the etching time to work around the inhibited flow along the edges that this may cause. The reason to keep this thick is because it’s really difficult to hold the laminate really flat. I tend towards 0.3mm.

Number of layers: 1. Adding additional passes on top risks tearing it all up.

Speed: I started at 10, and haven’t tuned this yet.

Wall generator: If the slicer refuses to draw your finer detail, try switching to arachne.

Nozzle temperature: 240 (so far this seems to stick to the copper better than 220)

Heatbed temperature: cold (unheated), otherwise it seems even more difficult to get the TPU to stick.

Regarding your cad file: avoid short strings for the hotend to print, if at all possible. Every start/stop involves a greater risk of tearing it up and having the hotend drag it along, disturbing the rest of the print. For example: remove all text, only include the actual tracks.

I saw a piece of laminate (avoid cutting, as this will likely make it slightly bent) and inspect it carefully with a flashlight for flatness. I use thin double sided tape to stick it to the printing plate. As before, disable bed leveling as we don’t need to confuse the printer with the areas outside the laminate.

It would be really great if it was possible to get the printer to perform bed leveling within the confines of the laminate only, and with a tight spacing. This would help tremendously with adhesion, which seems to be the biggest issue with this technique overall. I might look into hacking the gcode for this in the future, if that’s possible.

[dansan]

That’s 2.54mm hole pitch in the attached picture.

[TAF_International]

have you worked with or considered (and if so why not?) conductive PLA at all?

[NeverDie]

MICRO 3D Printing with a 0.1 mm nozzle

[NeverDie]

I thought you were wasting your time with a 0.4mm nozzle, but maybe not if printing with
a 0.1mm nozzle is truly possible.

[dansan]

Such materials have huge resistance, so they are not suitable for electrical circuits.

[dansan]

I don’t understand why I’d be wasting my time?

I actually consider moving to up 0.8mm rather than down in size.

The 0.6mm wide traces I have used are quite narrow for an etched circuit, I would not want to reduce this further.

There’s also the layer thickness, where smaller nozzles can’t print as thick. As you remember from my recent post I conclude that the major issue is adhesion since there is only one point used for the bed leveling. For this reason alone it makes sense to move to something larger than 0.4mm, not down in size.

[NeverDie]

Sorry, what I meant was that you can etch PCBs with better results on even a $100-$200 hobby CNC machine, so I didn’t see what the point of it was.

[dansan]

No worries. There are many ways to skin a cat. The method detailed here is an improvement over using a marker by hand. You mean using a router? That would produce a lot of dust in my home, be very noisy, and frankly I don’t like the look of pcb:s made in that way. It wouldn’t suit my particular situation at all.

[debonr]

Having etched and machined many PCBs, with art-work created by many different methods, I think what you are doing is not at all bad. You possibly could get a track between pads - you’ve always got the chance of trimming it with a scalpel. The major snag I see is the lack of the ‘hole’ in the centre of pad used as a guide for drilling. For my small PCB’s, which were embedded in potting compound, so appearance did not matter, I made them ‘copper rich’. Saved etch time, better for heat transfer, etc. Instead of drawing the tracks, I ‘drew the gaps between’. If you are printing in tpu, then there would be a larger surface area and less chance of peeling. Although it does not look as if it is causing a problem, sharp corners could be avoided. Sweeping curves were popular in the early days of hand laid track-work.