Etching PCBs with a SLA 3d printer

[TMaxElectronics]

I recently started making my own PCBs again because I got tired of waiting for two weeks and paying 35€ shipping, just to test out an idea. But I was thinking about ways of improving the process and looked into all of the more widely known "fast" ways of making PCBs. They basically all involve direct masking/exposing of the OCB material and avoiding the use of transparency film and range all the way from milling the PCB with a small cnc machine to a modified laser printer with a blue photodiode that then exposes the photosensitive film on the base material. But they all involved either equipment I didn't have, or some more or less extensive modification.

But then I though of my little SLA (LCD masked UV resin) 3d printer that was sitting in the corner collecting dust and started looking into the possibility of using the <10um pixel pitch LCD for direct PCB exposure. The chitubox file format is quite simple and only packs the image data with run length compression, and there was another java project that decoded and encoded the format already out there, so I quickly threw together a piece of software that converts the gerbers directly to the printer's format:



Since the printer handles all of the exposure timing for you all you need to do is to tell it for how long to expose the PCB and which type of photopolymer is applied to it, copy the file to the printer and hit print with the PCB on it instead of the resin tank.
Well... at least thats the idea

So far I have gotten mixed results, and managed no better than 0.2mm spacing and 0.15mm trace thickness,though those were both not to a quality that I'd call acceptable and it is really only useable down to 0.3/0.3mm. The issue seems to be the distance between the LCD and glass that the PCB sits on, as there is a clearly visible blurriness when just looking at it with a piece of paper.  I think .3/.3 is still certainly usable (especially since it reduced the time from gerber export to finished PCB to only around 25min) but it's not perfect.

I have actually ordered a replacement screen for the printer, so it might be possible to remove the glass in front of the LCD screen and get better resolution. It might even be easy enough to make a pcb toaster (or waffle iron? xD ) that has two of the displays and can expose a double layer PCB in one go... but that would be a project for later 

What do you guys think? Is this something worth pursuing or am I wasting my time?

[Twospoons]

Whats the light source? Is it a point source or an area light?  A point source should produce a reasonably sharp image.
The best light source would be a laser, as these do behave like a true point source.

[klugesmith]

Sounds interesting.  Not long ago, a board with 0.3 mm line and space rule was fine resolution!
Rules document from one industrial fabricator today says the cost steps up when lines or spaces are smaller than 0.1 mm (on a 1/2 ounce inner layer).  At work we have done large boards full of 0.081 mm striplines for years.  What's the rule for inexpensive quick-turn shops?

Are there any optics between your LCD array and the normal resin interface? Like GRIN lens strips used with line sensor in inexpensive scanners? Then the image would be sharpest at an image plane well above the surface on which resin tank rests.

Some readers may have heard about the sudden closure of Fry's Electronics store chain in western USA.  20 years ago that company was notorious for restocking defective or damaged products that had been returned by customers.  I once saw a photosensitized PC board in a bag on the correct hook, without its lightproof wrapper.  Don't remember if somebody's board design had been developed on it.

[johnf]

People are using cheap laser engravers to do this
Silicon chip Magazine ran an article at the end of last year on how to do it
you spray the board with paint let it dry then run the engraver over and brush away the paint ash

[TMaxElectronics]

seems that I missed the first few responses... sorry about that

Whats the light source?

A 2x2 array of UV LEDs, so not exactly small aperture size to begin with. A laser would in deed be better but be limited in output power and provide a too small spot size.

Are there any optics between your LCD array and the normal resin interface?

Not really. Its just a big reflector, some clear plastic screen in front of that with the LCD on it and the resin tank on top of that, so I suspect they just accept the resolution loss with the LCD not being in direct contact.
Here's an image of the optics (very low quality and out of focus... the printer is in the darkest corner of my shop ): https://i.imgur.com/KolVda3.jpg

I actually remember going to frys in las vegas when we visited probably 10 years ago. I think I quite liked it because they actually had some individual electronic components on the shelves which I didn't really know from out local shops here in germany, as they have a counter where you have to ask for the stuff you need.

People are using cheap laser engravers to do this

I know, but they have issues with filled areas (at least when I last looked into them) and are relatively slow. There is a project from some guys at my local hackerspace to modify a laser printer to expose the PCBs directly. But since my intention was to find a cheap way of doing this with devices that people are fairly likely to have (and without modifying those) I didn't pursue that further either.

I did some experimentation with using the LCD without the glass in front of it (which is a fairly hard modification) and it got MUCH better results:


There was some dust on the screen so the 0.1mm spacing didn't quite work out but I must say that I am really impressed with the results, as 0.15/0.15mm (and even 0.1/0.1mm) seems absolutely doable if I clean the screen properly. The issue is though that the amount of modification required means that the printer can't really be used for printing anymore (at least not easily) so not really a good second use for the printer.

That got me thinking though: what if I was to build sort of a LCD waffle iron with screens top and bottom and some alignment pins on the side that could expose 2 layer PCBs in one go. Just throw the board in, copy the gerber files onto a sd card and hit go. The printer screens are fairly cheap and have some documentation available from what I have seen. I'll keep you posted if I actually build that

[Twospoons]

The doublesided exposure idea is interesting - neatly avoids all the registration issues.  Except for the initial setup, of course!
I see there are resin printer LCD screens of various sizes available on Aliexpress, so should be easy enough to get something suitable.
For your optics, and since speed is less critical with PCBs (single exposure), you might look at using a single UV led with a Fresnel condenser lens. Done right that should give you reasonably even exposure, and reasonably parallel light for better definition.

[TMaxElectronics]

My idea was to try and see how well normal LCD backlight material works with UV. It is after all intended to give a nice and uniform brightness across a large area and would be a lot thinner than geometrical optics.
Not sure if I can get enough light with that though, since the LCD does absorb a ton of it. There are some monochrome ones available (which are used in slightly more modern sla printers) that absorb much less light, but they are also about 3x as expensive.

I have also been experimenting with that UV curable solder mask available on aliexpress and have gotten some pretty decent results, with the only issue being the layer thickness. I need to put some sort of plastic film over the top of it so it doesn't stick to the screen and that seems to mess up the coat. So far I've been using some 120T (threads/mm) screen printing mesh to apply the layer, which is too fine I think and doesn't let enough of the paint through.

But it might just be possible to make some 0.1/0.1mm PCB with solder mask, silkscreen and plated through holes at home which would be really cool

[Twospoons]

You can buy dry-film photoresist on Aliexpress too. Should be a lot less messy than screenprinting, and far more consistent.

[TMaxElectronics]

The photoresist for etching is pre-applied to the PCBs, I am applying the solder mask as UV curable paint and that is effectively 100% consistent with the screen printing mesh
There is some dry film solder mask out there too but that is hard to get from what I can make out.

[klugesmith]

>> make some PCB with solder mask, silkscreen and plated through holes at home which would be really cool

Plated through holes sounds like the hardest part. Got any references for DIY method?  Hole wall activation chemistry?  Extra photoresist cycle to avoid electroplating background areas to be etched away later ?

I found a hackaday article whose title is misleading, because it's about using metal eyelets instead of plating.
https://hackaday.com/2015/02/25/diy-through-hole-plating-like-a-boss/

[TMaxElectronics]

I'd have to search for it again but a while ago I found somebody using graphite based anti ESD varnish and a vacuum cleaner for the hole activation and then electroplated that with copper. According to his results he managed to plate down to 0.4mm holes 100% reliably. Much better than the formaldehyde based electroless copper plating mixures.

Protecting them from etching is another issue. I want to avoid needing a seperate dry film resist so I might give some other metal coatings a try and just etch inverted. I think some etchants can be stopped with a tin plating, which is really cheap and easy.

[klugesmith]

I saw a graphite paint and vacuum cleaner project online today, maybe the same one you are thinking of.   Assortment of hole sizes, and the 0.3 mm holes failed but bigger ones are OK.

I remember the tin resist process from industrial board shops, decades ago. Could not easily find a reference online today.
1. drill holes in copper-clad board stock.
2. Process photoresist to be _open_ where we want copper in finished board.   
3. Do the hole plating thing, which also adds top and bottom copper where we want it.
4. Electroplate tin or tin/lead to cover the visible copper.
5. Remove photoresist. 
6. Remove background copper with an etchant that is resisted by the tin on traces, pads, and in holes.

We know that's not a universal method.  SMOBC (Solder mask over bare copper) boards were common more than 30 years ago.

[hardboiled99]

The chitubox file format is quite simple and only packs the image data with run length compression, and there was another java project that decoded and encoded the format already out there, so I quickly threw together a piece of software that converts the gerbers directly to the printer's format

I recently saw a couple of Youtube videos on this (etching PCBs using SLA printer), and wanted to try it out myself. Having some file conversion challenges, and was wondering if you ever pursued this further?

[TMaxElectronics]

I did actually. Also just happens that I made a post about it on the EEVBlog forum aswell

For file conversion I wrote a little program that spits out the files for chitu 3d printers directly from gerbers, with adjustable layer alignment and all. But I haven't finished it yet, at the moment it is very hacky and hardcoded for my printer. I'll work on it some more soon though. You can check the github repo for any updates I make in the future: https://www.eevblog.com/forum/eda/making-high-quality-home-etching-easy-66mil-pcbs-exposed-with-an-sla-printer/new/#new