Making fine pitch PCB prototypes with fiber laser

How To

on 2021-01-07 40 Comments

There are quite a lot of inexpensive and fairly reliable PCB manufacturers that can produce double-sided PCB’s with solder-mask and silkscreen in a day or two, but what if you need it now? I will not cover expensive prototyping machinery nor messy printer/etching process instead let’s use a 20W Fiber laser. Because… who does not like lasers, right?

Goal

The ultimate goal is to have a soldered fine pitch (0.1mm/4mill track and clearance) component as fast as possible (30min) along with a few passive components.
The material of choice is common FR-4 with 0.018 or 0.035mm copper thickness. It would be perfect to have double-sided routing, but via copper plating process complexity and time restriction imposes limits to single-sided design without drilling (though with some drawbacks it can be done).

Design preparation

Gerber files

For this experiment, I used one of the fresh upcoming camera design. Skipping Gerber export step as it varies from software to software. For this test prepare 3 Gerber files:

TOP tracks layer
TOP Solder-mask layer
Contour layer

FlatCam import

FlatCAM lets you take your designs to a CNC router. You can open Gerber, Excellon or G-code, edit it or create from scatch, and output G-Code. Isolation routing is one of many tasks that FlatCAM is perfect for. It’s is open source, written in Python and runs smoothly on most platforms.
http://flatcam.org/

FlatCam is also an ideal tool to prepare DXF/DWG vector files for further processing. Feel free to explore this powerful and open-source tool.

File / Open / Open Gerber… and load all prepared files

Double click on each Gerber layer and generate Isolation Routing layers
Double click on each Geometry layer and Generate CNCJob object

Then export each CNC job layer to an SVG file. Here we go, Gerber files were converted to more usable DWG files that can be used by EZCAD software.

EZCAD import

LIT Laser systems represents the latest in smart fiber laser technology. Our approach has drasticly reduce the price of laser systems. This unprecedented value and versatility has enabled students, technicians, engineers, inventors, makers, and hobbyists a chance to own laser systems that have been out of reach for many.
https://www.litlaser.com/ezcad

EZCAD comes with the vast majority of laser marking machines and is considered a de-facto tool. Unfortunately has some imperfections (actually tons of them). One of the major ones – outdated and buggy vector file importer. So after opening the top layer we will see a broken image (which is easy but tedious to fix)

Make copper traces

First, let’s fix the vector file imported in EZCAD. Here is the workflow that worked:

Select all geometry
Click Edit / UnGroup
Select each jump line and delete. I know this is tedious work, but could not find any quick workarounds
Click Edit / Group

This is another nasty EZCAD bug – if you would apply hatching now, geometry will be totally messed up.

Quick workaround – Click Modify / Curve Edit / Auto Connect (0.010mm) and geometry hatching looks better (might need grouping / ungrouping step)

For track isolation use hatching and laser parameters like in the screenshot above. Not ideal, but should be good to start experimenting.

It took about 20 cycles to cut through 0.035mm copper foil. Some of these passes were with increased power to 90% to speed up the material removal rate. A good hint when to decrease laser power is to observe copper color – with decreased layer thickness it will heat up more and change color. 50% of power will not greatly damage the substrate, while higher power settings make resin burn.

Prepare soldermask layer

Use PCB UV Curable Solder Repairing Paint (also good choice Dynamask 5000 dry film)

Spread the paint evenly on the PCB (use a thick transparent plastic sheet and rubber roller to spread evenly)
Make sure there are no bubbles
Introduce to 365nm UV light to cure. This should take a few minutes, for exact timing read material specifications, and experiment.
The protective sheet should peel off without damaging the freshly cured solder mask

More or less evenly spread and cured solder mask. This forms a pretty resistant layer that can withstand soldering in high temperatures.

Open solder mask

Rinse and repeat importing solder-mask layer into EZCAD software.

Laser settings in screenshot above. Depending on your layer thickness several passes will be required.

Cut PCB contour

The final step is to cut PCB out of stock material. A laser will not warp and skew FR-4 material so this is a very nice way to have PCB made to size. It will burn edges though, but light touch with 400 grit sandpaper will fix edges.

Laser parameters are in the screenshot. To cut through 1.6mm FR-4, wobble parameter is enabled (click Advanced... button)

Same parameters can be used to drill holes.

Will it solder?

Moment of truth – 0402 capacitors soldered with hot air and after Ultrasonic cleaning look charming.

Of course bare copper will oxidize very quickly so proper care should be taken. Other than that usual soldering job.

And for your entertainment and understanding macro video with each step

Final thoughts

Well with all experiments, preparations and writing an article it took half a day. If issues with vector file import are fixed, it should be possible to have simple in 30min.

While making PCBs with 1064nm fiber laser is quick and a clean process it has some drawbacks.

FR-4 material burns and smells, also there is a danger of having burnt resin between tracks. Good ventilation is a must!
Isolating tracks cam be uneven. Some areas can be burnt through, some will have remaining copper.
High energy pulsed laser would work much better, but the price of such a machine is beyond the roof.
Stencil from stainless sheet metal can be also cut with laser. Main concern is not to overheat stainles steel. Another issue is uneven edges. This is not an issue for 0402 component pads, but smaller pads will have issues.
During import into EZCAD, missed cleaning some geometry so some pads are incorrect. But for proof of concept and laser parameter tuning this is irrelevant.
A good UV light source is a must. While the surface of the etched PCB can harden pretty fast, the deeper layer can remain untouched. Soldering with hot air or cleaning can open these defects spreading uncured paint.

Comments ( 40 )

2021-01-11jipe
Hi, Really good job! I just wonder how powerful is your laser? Is that a CO2 laser? At least more than 50, or even 100W ?
- 2021-01-11saulius
  Thanks! Actually it's 20W fiber laser (1064nm). Powerful CO2 laser will not do better job at removing copper without substantial FR4 material damage.
2021-01-11TheDarkTiger
Awesome! A clear step by step process, that's very nice. Quick question, you say "Same parameters can be used to drill holes." ; did you try it? I think you need to remove coper beforehand where you "drill"?
- 2021-01-11saulius
  No need to remove copper. Drilling 0.2mm vias with laser looks not so shabby. Will be updating instructions with more details in a few days, stay tuned.
2021-01-11TheDarkTiger
Impressive! If only there was a single format for pcb and an other for laser cut, one could create a program to automatically configure everything without always tweaking the parameters until it fits. Thank you for sharing your discoveries!
- 2021-01-11saulius
  That would be perfect! My goal is to fine-tune parameters so it would work as a baseline. Gerber processing works fine already (though more automation would be nice to have) and EZCAM already does not require geometry cleanup.
2021-01-11Matt Hope
You could finish the end product in a liquid tin bath, it's easier to solder and protect the copper from oxidation. Also, for the via, I don't think there is a quick solution for home-made PCB. I've used in the past via rivets, it does a great job, but it is time consuming and the via need to be quite big (0.6mm drill, 1.5mm pad). I saw some people having success with carbon powder and solder paste, but it required a lot of trial and error to make it work.
- 2021-01-11saulius
  Hi Matt, thank you for reminding about rivets! This could be helpful in some cases. And regarding oxidation - laser does good job cleaning copper from oxides so if PCB is soldered within a few hours (or days if conditions are dry) there is no need to coat fresh made board.
2021-01-11B Thomas
Are you aware of any open source alternatives to EZCAD which will work with one of the inexpensive fiber lasers ?
- 2021-01-11saulius
  Hi Thomas, not that I am aware of. There is an alternative (also commercial) called Lightburn.
2021-01-11andreas
Thank you for all the details! I've seen videos of people successfully structuring FR4-PCBs, but up to now unfortunately most times without details on the used parameters. That is exactly the purpose I bought my 30W fiber laser for last year. Unfortunately I did not have any usable success up to now. I managed to remove copper, but always burned (really burned...) the FR4. With reduced power I was not able to remove the copper. I even had trouble to engrave into the backside without copper cladding without burning the FR4. Sometimes it even ignited a flame. I looking forward to read more. Maybe I can adopt something. Thanks a lot so far!
- 2021-01-11saulius
  For Copper/FR4 transition - try reducing power just slightly above threshold where copper is being vaporized. Also keep low pulse per area count (reduce laser freq, increase speed).
2021-01-12cejasmasmas
Hi, when you export to svg, you can open the svg file in inkscape, and save as .plt (HP Graphics Language Plot File), this way ezcad2 opens the file correctly without anymore processing, you have to install and additional plugin in but after that it just works
- 2021-01-12saulius
  Thank you! Was trying to avoid extra software packages and Inkscape has bugs of it's own. Anyway found better solution - export G-Code from FlatCam and open in EZCAD. Works like a charm!
2021-01-12s-ol
Really nice! Are you aware of PCB via rivets? I don't have access to a fiber laser, but with a co2 laser and a CNC mill I was able to make double-sided boards using them. They are basically just little copper rivets that you insert on one side and mushroom over on the other. I recommend also soldering them, purely mechanical contact tends to turn intermittent over time IME. http://fabacademy.org/2020/labs/opendot/students/sol-bekic/projects/laser-board/
- 2021-01-12saulius
  Thank you for reminding about rivets... but they are huge and require tedious work to install.
2021-01-13Felipe
A engraving laser module can do the job too? 40W Laser Module Kit Laser Cutting Engraving Module Blue Light Module Set Laser Engraving Machine Cutting Tool Metal Engravable https://a.aliexpress.com/_mr6HgqH
- 2021-01-13saulius
  Felipe it will not work. You need at least Fber or YAG laser.
2021-01-13cejasmasmas
Forgot about G-Code export, thats even better, I bought a 50w fiber laser for this purpose, tested a few copper clad boards, got carried away with the power settings and totally destroyed the boards, managed to create a few with fair results, In the end I spray painted them and removed the paint as with a Co2 laser, the good thing is that the fiber laser doesn't leave the transparent mask the Co2 leaves, so after removing the paint you can go directly to the acid, getting really nice reliable results, the one thing I didn't test was to actually cut them, it's good to know it can be done, thank you
- 2021-01-13saulius
  I did many unsuccessful testes earlier. There are many parameters to tinker with. As you probably noted trickiest thing is to remove copper with minimal damage to FR4 material - try removing copper in multiple passes with minimal power settings. Soldermask will fix minor FR4 damages.
2021-01-13joposter
You have got me looking for machines. The result looks very usable! may i ask what precision you could achive? did you try some stress tests to find minimum trace width and pitch?
- 2021-01-13saulius
  Technically laser spot is 0.02mm (on papers). Have not tested what actually can be achieved, though. 0.1 is looking pretty good. 0.07mm should be also doable. Everything smaller should be not quite reliable.
2021-01-15Shingo
Within an hour after reading this, I ordered CO2 laser cutter. My idea is to use Kapton adhesive tape/film instead of PCB UV Curable Solder Repairing Paint, Dynam"a"sk (you have typo) and UV cure machine. Mine will be delivered in March so if you're interested in, please try it!
- 2021-01-15saulius
  Thank you for noticing a typo Shingo! Fixed. Kapton tape cut with CO2 laser might actually work. Though thinking about many issues like bunt tape edge and lesser CO2 laser resolution (making an assumption that it has mechanical beam drive instead of galvoscanner).
2021-02-01electronrancher
Great tutorial and nice boards! I am so far unable to recreate with my 50W but I will try more with a new PCB substrate soon. Would you email me your supplier of copper clad? I would like to try the same substrate you use. BTW: The reason for the unusual behavior of flatcam's SVG is due to the "lens" setting it applies. Not sure what it's supposed to be doing, but all my tools seem to have a hard time with flatcam data as it has a lens effect applied to every line. I fix it using CorelDraw: Import SVG, ungroup all. Then select all lines and Open the Lens docker through Window->Docker->Effects->Lens. Select "No Lens Effect" and you will see that the Lens property is removed from each path in the Object Manager. Now the data will be easy for any tool to use with no miscalculated lines or weird shapes. I usually export as AI, and import into EZCad - no manual editing needed with the cleaned data. I expect there are other programs that can strip the Lens effect, so if you don't use Corel you may try inkscape, etc. Hope it helps!
- 2021-02-01saulius
  Hi, actually used scarp FR4 material so can't tell manufacturer. Probably these were left overs from LPKF material. As someone suggested - copper thickness is an important factor. Regarding SVG import - also found better solution. Use G-CODE export and it will work perfectly. No need to use one more non free program!
2021-12-12nsb
It's been quite a few months since the last comment, but I wanted to add my feedback here. This video was one of the reasons I decided to pick up a fiber laser. I had been milling PCBs on my benchtop mill, but that's slow and tedious...also, who doesn't want a frickin' laser to play with? So far it's going pretty well. I picked up a 30W Raycus unit, and am running Ezcad 2.14.11 for my testing. I found that midrange power settings and multiple passes worked quite well - for me, 60% power and 600mm/s with 5 or 6 passes gave me a very clean isolation run. At that power level the laser will burn the copper but will leave the fiberglass substrate untouched, so you wind up with some very clean PCBs with no charring at all. I did numerous test runs at various power and speed levels and found that higher power was less even in its material removal, so that by the time the middle of large areas of copper was gone, there was still material close to the traces. This is probably due to heat dissipation, but the end result was that to remove the remaining material required another pass, but at higher power, another pass meant charring the exposed substrate. It is probably worth trying a pass or two at high power and then finishing off at lower power. I think saulius mentioned that as well. Then I moved on to drilling holes for components and that's where things are currently falling apart. Haven't found any settings that don't result in serious charring yet. It's possible that I'm not being patient enough and waiting for heat to dissipate, but at the moment I'm just producing a lot of carbon. My workflow currently is: - PCBs in Kicad (v5.1.10) - Plot gerbers and drill files - Import into flatcam (v8.994 beta at the moment) - Single 0.01mm isolation routing on PCB traces - Edge cuts at 0.01mm - Drill file - Export all as SVG - Import into inkscape - Ungroup everything - Delete toolpaths - Group everything - Export as svg again - Import SVG into EZcad - Ungroup file - Apply hatch * Burn! I'll follow up here when I solve the through hole problem.
- 2021-12-12saulius
  Thank you for sharing your settings! Agree that keeping power settings lower helps with charring. While 1064nm fiber laser makes nasty drills (not perpendicular and all burnt), technically, it is not as crucial as charring is local and will not change the impedance.
2021-12-12nsb
Eesh - apparently formatting didn't carry over.
- 2021-12-12saulius
  Unfortunately, WP can't handle nice comment formatting yet.
2022-01-27Bryce
Some new developments in this area: There is open source software for controlling the engravers: www.bryce.pw/engraver.html LightBurn and Meerk40t are working on adding support for these fiber lasers.
- 2022-01-27saulius
  Thank you for sharing! Will try once I have more spare time.
2022-07-26rlevy
Great work! I don't know if you're still responding to comments on here, but I'm looking at picking up a laser for this use case. I'm torn between a 50w fiber and a 30w mopa. Do you feel that the increased frequency range of the mopa laser (1-4000 khz) is better suited for this application? It also allows for the on time duration of the laser to be adjusted so that might help with heat absorption issues?
- 2022-07-26saulius
  Hi Rlevy, have not tested but UV laser should be even better. Usually power rating is smaller, but PCB should burn less.
2023-01-12Zai1208
What are the specs I should look for when buying a laser cutter?
- 2023-01-12saulius
  First of all - 1064nm Fiber laser is far from perfect to make PCB's. Main issue is burnt FR-4 substrate. If this is acceptable (you can deal with it in the next step) then almost all 20W+ pulsed fiber lasers will be suitable.
2024-08-20MG
I’m curious if anyone is mastering two-sided layer alignment? It’s generally a nightmare, not in so much that you can’t get a board square to the laser head, but that laser calibrations are imperfect. To get a .25mm via to line up after a board flip has taken days of magic calibration both in ezCAD for the lens and then applying asymmetric transformations in the DXF. What a mess!
- 2024-08-21saulius
  You are right, you can do mechanical matching with 2 precision pins a drilled holes, but it will not solve all of the issues. Can't think of easy solution with EZCAD.
2024-09-05Nils
What lens did you use on your fiber laser machine to engrave the copper PCB? I tried using a 175 x 175 mm lens on a 30 W MOPA fiber laser, without success.
- 2024-09-05saulius
  Hi Nils, it was 110mm lens on 20W Q-switch fiber laser (not a MOPA).