Author Topic: Making high quality home etching easy - 6/6mil PCBs exposed with an SLA printer  (Read 11469 times)

0 Members and 3 Guests are viewing this topic.

Offline jz79

  • Newbie
  • Posts: 8
  • Country: lv
Sorry for necroing an old thread, but this was the inspiration to try and ditch the methods involving laser printers, so I thought I'd report on a success, sort of...  ;D
The msla printer I used was a new out of the box Elegoo Saturn 4 Ultra (because it had the high resolution 10" mono lcd, and it has a tilting feature that supposedly makes 3d printing a lot faster, though there are bugs with the whole design), and this is most definitely not an ad, I do NOT recommend it for this PCB exposure thing, because there is no easy way to just pop your image on the screen for certain amount of time, so get around this I you'd need UVTools freeware software, it is used for creating and editing 3d print files, it will open .goo (Elegoo format) files that you can feed from USB into the printer, so I took the example file that came on the USB together with the printer, deleted all layers, imported new layer using custom .png image (had to be resized, because pixel dimensions are 24x19 micron, they aren't square), then edit the layer settings, set exposure time and hit print, aaaand wait for it to go through all the pointless (for this purpose) actions, calibrating some mystical parameters etc, and then override couple errors to get your image to pop up on lcd finally.

I initially tried couple dry films from Ali, the really cheap ones, and they are no good, it seems they are too thick and overall they have defects, like wrinkles in the dry gel that won't adhere properly, and the thickness seems to cause problems exposing narrow features, after exposing and trying to dissolve (develop) the resist, I can actually see crisp top layer of the features seen below (that is just a test image of TSSOP 24 with 0.1mm track weaving in between the pads), but the thin parts in between seem to catch some UV and partially cure also, making the developer a bit more concentrated kind of helped to dissolve some of it, but it also started to lift some other features, basically - a fail.
I tried all sorts of things with the UV intensity settings, exposure time, but nothing really solved the above, the film I bought needed very short exposure, only 10 seconds, going lower started to look like underexposure, developer was kind of starting to attack the exposed resist.

this is what it looked like overetched trying to get the pads to separate


And this is the result with spray on positive resist called Positiv 20, manufactured by Kontakt Chemie, this was first time trying to use wet resist, made a spin fixture out from a pc fan with some double sided sticky tape to hold the pcb, sprayed the resist on, just a thin coat, it doesn't really spray that evenly, spun it for maybe 10-15 seconds, that leveled it reasonably well, baked for 30min at 60C.
Exposure settings - 45sec, etching with plain ferric chloride, slightly warm, maybe 40-45C, etch time was probably 6-8 minutes, face down, constantly moving it

« Last Edit: October 10, 2024, 07:46:34 pm by jz79 »
 
The following users thanked this post: chickenHeadKnob, thm_w, rajhlinux

Offline PlainName

  • Super Contributor
  • ***
  • Posts: 7410
  • Country: va
Nice!

Well done  :-+
 

Offline rajhlinux

  • Contributor
  • Posts: 33
  • Country: us
Last year, I've found some tablet-like OEM displays in the range of 8K resolution.

Edit:

Oh wow, haven't realized the market has catched up selling consumers SLAs with 12K displays. Wow...

Edit:

Wait there are even 14K displays too?...
« Last Edit: November 02, 2024, 02:54:23 am by rajhlinux »
 

Offline rajhlinux

  • Contributor
  • Posts: 33
  • Country: us
Holy Shamoley, thats the most accurate PCB etch I've ever seen in a DIY style.

What are the actual pixel size of the LCD? From my calculations for a 10 inch diagonal screen with a 12K resolution should be around 20 microns.
Your PCB etching is thus accurate down to at least ~25 microns +/-.

You might wanna give a try to Dupont Dry Film which are used for Laser series sensitizing lithography.
 

Offline jz79

  • Newbie
  • Posts: 8
  • Country: lv
lcd pixel size is 19x24um (11520x5120px and 220x122mm screen size), I would have preferred a square pixel LCD, but it seems there is nothing like that in the 10" lcd size range currently, and I think the results show that the squareness of the pixel isn't that important, it is just a small inconvenience having to resize the "square pixel" image out of the CAD package to the 1.25:1 ratio so the final exposure is the correct

I don't think dry films are worth the trouble to be honest, always tricky without a clean room and right equipment to get good repeatable results, I find the spray-spin-bake easy and quick enough process, and I think I'd rather buy a stack of presensitized boards from quality manufacturer like Bungard (I'm in Europe) to have something on hand rather than dealing with the dry film

biggest "issue" for me currently is the closed nature of the firmware on the printer and the lack of support, there's a feature on the printer to show a test image to check if the screen is working correctly, it would have been perfect if this feature had a dialogue to browse and select which image you want to display on the lcd (time duration setting there is already present), mentioned it as an idea to the manufacturer, so they can potentially get additional customers to buy their product if their printer could be used for PCB exposures, but zero response
and it seems it is running most likely LinuxCNC under the hood, since the printer will accept G code, there is even specific G code to display an image, but it doesn't seem to work, the screen lights up blank, like it can't load the image I specified in the G code command, that would greatly simplify using this particular printer for this task, but even this simple request the support staff completely ignored - to explain how to display images on the lcd using G code...
even worse is that the lcd driver is some custom board soldered right on the "motherboard" of the controller, I expected it would have hdmi interface like the mono LCDs controllers on ali do, and I thought in worst case scenario I could stick a raspberry inside there to run the thing, but sadly no such luck due to lcd driver interface, I did however find a script and set of tools to unpack the original firmware, but hadn't have the time yet to dig into it more to figure how to add the functionality I want

so, if anyone is thinking about buying a printer for this - avoid the ones that don't accept standard G code, and talk to support first just to get a feel if they are willing to help if any problems arise, because Elegoo, while rich on features on paper, are absolute garbage when it comes to solving issues with their products, I understand that 400eur resin printer may not have built in margin to fund extensive support, but not helping a user to tell them how to display an image on the screen using G code is no support at all, doesn't cost them anything, and it isn't like they can add to the price if they start to include pcb exposure as a feature when competition can do the same with just a Gcode command
« Last Edit: November 01, 2024, 09:46:21 am by jz79 »
 
The following users thanked this post: rajhlinux

Offline rajhlinux

  • Contributor
  • Posts: 33
  • Country: us
19x24 microns = 456 square microns/pixel, that is not enough for my standards of accuracy, need something more in the range of <= 200 microns.  ;)

Actually I think the area of the pixel is not useful in determining the accuracy of precision imaging, but rather the diagonal size of the pixel, for 19x24 microns and using Pythagorean theorem, the overall size light emitted from a pixel of the size of 19x24 microns should be ~29 microns (Diagonal Size). Which is great and can be used for many high precision PCB sensitizing.

I agree, I've read about dry film usage in the PCB lithography process and it is really cumbersome to work with, everything needs to be clean and perfect and less forgiving. But I do know many of the state-of-the-art PCB fabs uses dry film.

However when precision comes into mind, I think the spray/liquid photoresist on a spinner is the winner, hands down. Its the process used for making Integrated circuits on silicon, various methods are used. I agree it's easier to work with than film based and adds higher accuracy.

Have you considered simply buying the OEM 12K or 14K LCD displays from Ali and simply design your own lithography box? I was planning to do this years ago when 8K LCDs were the highest resolution. However I simply decided to stick with something that is much more simpler to work with, using professional grade printers, particularly Epson Pro grade printers (they cost a pretty penny, buy them used after testing their accuracy of nozzle print checks). I've done many testing and completely shocked at the level of detail it can produce. I believe they are accurate down to 15 microns -/+. The interesting part about these printers is that their printhead nozzles are produced in a semiconductor fabrication process of lithography to produce ultra precise and small nozzles. Canon professional grade printers are good too and somewhat cheaper, produces excellent results, but their printheads are thermal based and doesn't give much freedom of using various types of inks compared to Epsons.

My main ideal PCB lithography system would be a custom made ultra-precision, ultra-high-resolution scanning laser system, concept is similar to laser jet printers. As accurate down to sub microns. However building such device requires a PCB, and many high-precision machinists machinery and skills therefore need to start somewhere to produce fundamental PCBs.
« Last Edit: November 02, 2024, 03:55:02 am by rajhlinux »
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 11036
  • Country: us
  • $
lol that dry film shit, try this method

1) clean sink good
2) fill sink with water
3) open dry film under water
4) align with PCB
5) pull it out slowly at an angle
6) put into hot roller machine a bunch of times
7) do again for other side

it sounds bootleg as hell but it works. i was losing my mind trying to do it in the air

the key is to use a big sink. trying to get tupperwares and shit is beyond ridiculous
« Last Edit: November 02, 2024, 03:51:26 am by coppercone2 »
 
The following users thanked this post: rajhlinux

Offline rajhlinux

  • Contributor
  • Posts: 33
  • Country: us
Yes I've seen people doing this and swears by it.
 

Offline jz79

  • Newbie
  • Posts: 8
  • Country: lv
don't get me wrong - I'm not saying dry films are bad in general, just that the ones I tried were bad

Bungard sells their dry film also, but smallest roll is 80EUR (0.3x25m roll, so that is a lot of film), and the issue with clean conditions to apply it still remains, I'm not going to bother using sink full of water, that is just too much hassle IMO when spray-spin-bake works just fine

quick google search showed that Dupont to be available from around 20EUR

and I didn't go for 12k 10"mono lcds from Ali because they end up costing around 200EUR and there basically is no warranty whatsoever, and a printer is 300~400EUR
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 11036
  • Country: us
  • $
how do you spin a board of substantial size? I only ever saw people make spin tools for little PCB

you can do a 12 inch x 12 inch PCB in a fairly normal kitchen sink
 

Offline jz79

  • Newbie
  • Posts: 8
  • Country: lv
double sided sticky tape on a 2$ atx power supply fan inside a cardboard box... powered from lab power supply, since need to limit the rpm, no problem with 200x120mm boards, that is max what I can fit on the LCD anyway, anything larger I would most likely outsource, but I very rarely if ever need a board longer than 200mm

that was the big surprise for me that even a ghetto spin fixture like that ends up having better than expected results compared to dry films I tried, I didn't even clean the copper particularly well, the board was one of those cheap paper ones, and more than 40 years old, some scotchbrite to get rid of most oxidation, acetone to degrease, then spray, it had some unevenness even after spinning, but dried flat enough, if you ever spray painted anything and had seen water droplets from condensation in the paint, that is what the sprayed and spun coat looked like, but it turned out in the end just fine
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28243
  • Country: nl
    • NCT Developments
Holy Shamoley, thats the most accurate PCB etch I've ever seen in a DIY style.
It looks good indeed but are there no shorts between the pads? There can be barely visible shorts between the pads if the etching wasn't completed. I have done etching with ferric chloride for decades but never went under 0.2mm trace width / seperation. Not because of the lithographic process, but due to the etching process. BTW, I always used pre-clad boards from reputable brands.
« Last Edit: November 02, 2024, 09:39:17 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline cunningfellow

  • Contributor
  • Posts: 22
  • Country: au
It looks good indeed but are there no shorts between the pads? There can be barely visible shorts between the pads if the etching wasn't completed. I have done etching with ferric chloride for decades but never went under 0.2mm trace width / seperation. Not because of the lithographic process, but due to the etching process. BTW, I always used pre-clad boards from reputable brands.

Spray etching can help with that.
 

Offline chickenHeadKnob

  • Super Contributor
  • ***
  • Posts: 1061
  • Country: ca
double sided sticky tape on a 2$ atx power supply fan inside a cardboard box... powered from lab power supply, since need to limit the rpm, no problem with 200x120mm boards, that is max what I can fit on the LCD anyway, anything larger I would most likely outsource, but I very rarely if ever need a board longer than 200mm

that was the big surprise for me that even a ghetto spin fixture like that ends up having better than expected results compared to dry films I tried, I didn't even clean the copper particularly well, the board was one of those cheap paper ones, and more than 40 years old, some scotchbrite to get rid of most oxidation, acetone to degrease, then spray, it had some unevenness even after spinning, but dried flat enough, if you ever spray painted anything and had seen water droplets from condensation in the paint, that is what the sprayed and spun coat looked like, but it turned out in the end just fine
Was your PCB 1 ounce (34 micron) or 0.5 ounce (17 micron) copper weight? The half ounce is better to get  the really thin tracks without undercut.
 nice work by the way.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28243
  • Country: nl
    • NCT Developments
It looks good indeed but are there no shorts between the pads? There can be barely visible shorts between the pads if the etching wasn't completed. I have done etching with ferric chloride for decades but never went under 0.2mm trace width / seperation. Not because of the lithographic process, but due to the etching process. BTW, I always used pre-clad boards from reputable brands.
Spray etching can help with that.
Been there, done that. Spray etching is actually is worse because getting the spray pattern even is hard to do and this method etches extremely fast. If I would get an etching machine again, I would go for one which uses FeCl and foam.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Teal

  • Newbie
  • Posts: 6
  • Country: 00
I've been trying to get into this too but couldn't find any software that would work for me. So I wrote my own. The script converts gerber /svg / png files and patches Anycubic printer files with new image data.
Maybe someone will find it useful: https://github.com/BleakyTex/Anycubic-PW0-Converter

I haven't tried to make PCBs with it yet but the display of my Anycubic Photon Mono 4 resolves 0.05mm traces and gaps.

2453071-0

2453075-1

2453079-2
« Last Edit: Yesterday at 10:55:02 am by Teal »
 
The following users thanked this post: PlainName


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf