So let’s see if I can finish a project! Here will follow a series of updates (hopefully) on the ongoing progress of building a device to perform the photosensitive-exposure step in building my own PCB Manufactorum.
The plan is:
LED source of UV - low power, easy, safe.
Capacity for a 100x200 PCB
Able to expose double-sided PCBs
Adjustable timer with auto-off
Progress so far? Still in the design stage, but parts have been ordered…
More details on my blog for now, and the source files for the circuits (KiCAD) and hardware (SolidWorks) are on GitHub.
It’s pretty good. It’s a bit quirky, and has some manual steps that Eagle does automatically but it’s perfectly workable. The one thing I cannot work out is creating custom components. The schematic outline is fine, but the PCB pad part is bonkers. They’ve switched to some weird github hosted system, but it’s very poorly documented.
Happy to give a demo on Thursdays, or I will be at the HS this Sunday if someone wants a bit of a deeper dive into this.
Currently working on (in my head) a way of mounting the two windows in such a way that they are sprung. This would mean closing the lid will compress the mask onto the PCB nicely. Not sure how yet…
Got loads done at Sunday’s hack session. The box is pretty much complete, although I will probably still do some cosmetic cleanup on it - round the corners a bit, paint it, that sort of thing.
I also made a start on the Veroboard for the LED panels:
I always do it before soldering. Otherwise the drill bit snags on any adjacent soldered pins, or worse, pulls the track up off the board. I find using a drill-bit (or one of those little tools specifically for this) easier than using a knife, and much clearer to see where the breaks are. Knife is still handy for those “between the pins” breaks, when space is at a premium.
While trying to work out a neat and non-insanity-producing way of putting 170 LEDs onto Veroboard, I realised I could stick them all into breadboard, and use that to bootstrap a home-made PCB for the LED panels! Here’s the bootstrap device:
I haven’t posted any updates for a little while, but have done lots! Made some significant changes to the schematics (and therefore the boards) after testing on breadboard (pictures to follow), and also about 80% complete on code.
Manufacturing
Biggest problem right now is not the box, but the stuff that comes afterwards. I’ve been having problems getting the exposure time right, and even copper-etch. I believe this is due to the way the inkjet printer is printing on the film - it leaves a visible grid of tiny dots, which are being etched.
Here you can see the result of exposure. At first glance, looks pretty detailed (ignore the finger smudges), but the slight amount of green resist remaining in the middle (by the two parallel tracks) is enough to cause problems etching.
Here is the same board etched. Oddly, the board is over etched at the area where I expected it to be under-etched.
Here’s another example. This one is showing massive exposure problems, simply due to light leaking through the ink on the acetate film. I have also tried using paper, and then oil to make it more transparent, but this is messy and absurdly inconsistent.
Conclusion? Inkjet might be a bust, unless you have a particular printer that prints very solid blacks. It’s not just a cheap-printer problem, either - the Canon ip7200 I have is pretty decent, it just doesn’t print on film that well.
I now have acquired a laser printer and tracing paper, so trying that. I also have some new chemicals, so will be trying Sodium Metasilicate Pentahydrate instead of Sodium Hydroxide.
Yep, the render is from KiCad, although I had to make a couple of parts in Wings3D that are missing from the core library.
Main problem with the inkjet seems to be that the whole process relies on the ink wicking slightly in the paper, so on acetate that doesn’t happen. You get this sort of spotty effect where it’s all black, but there are ‘peaks’ of blackness at the centre of each spot. Haven’t tried the inkjet on the tracing paper yet.
Okay, so the breadboard UV panel didn’t really work out. I think having all the UV coming from a really narrow strip in the middle is what caused only the central strip of the PCB to be properly exposed. I built a panel of UVs on Veroboard, so the light was much more spread out, and here’s the result…
Overall, very pleased with it! This was with a 2 minute exposure time, and it seems that some areas of the board were still not quite exposed enough, so those areas were still being protected from etching. Holding the board up to the light really helps to see the issues.
Main problems were some tracks etched away, a couple of places where nothing was etched away, and then a non-etch-related issue, which was isolated pads or those only attached to a thin track were ripped off when drilling.
I fixed the tracks by using a small drill-bit in my drill-stand, setting the depth-stop to just below the copper depth, and milling the copper where needed.
Here’s a top tip to reward those bothering to read down this far: put a flag of tape onto the drill bit. As it spins, it creates enough wind to keep the board clear of debris! No more huffing and puffing at your boards.
I have adjusted my design rules for future layouts, making most pads ovals and the tracks a bit wider.
