Normally, you have a choice with PCB prototypes: fast or cheap. [Stephen Hawes] has been trying fiber lasers to create PCBs. He’s learned a lot which he shares in the video below. Very good-looking singled-sided boards take just a few minutes. Fiber lasers are not cheap but they are within range for well-off hackers and certainly possible for a well-funded hackerspace.
One thing that’s important is to use FR1 phenolic substrate instead of the more common FR4. FR4 uses epoxy which will probably produce some toxic fumes under the laser.
We were surprised at how good the boards looked. Of course, the line definition was amazing, as you’d expect from a laser with details down to 200 microns (a little less than 0.008″), and he thinks it can go even lower. The laser also drills holes and can cut the board outline. A silk screen makes it easy to add a solder mask, and the laser will even cut the mask. [Stephen] also etched “silk screening” into the solder mask and filled it with a different color solder mask to make nice board legends.
Registration is critical and will be extra critical for two-sided boards which is what he’s working on now. We think if you put some scored carrier edges around the board with fiducials, you could make a jig that would hold the board in a very precise position using the holes in the carrier edges.
Vias are another issue. He mentions using rivets, but we’ve also seen people simply solder both sides of a wire through a hole, which isn’t that hard.
For most people, making your own PCBs is fun but not very practical. But there is something about being able to turn around actually good-looking boards multiple times in a day when you are doing heavy development. If you don’t mind fumes, you can laser mark your PCBs.
I see more use in this for creating front panels. As for PCBs, I can get 5 50×50 6 layer boards for less than €2 (excl. shipping) in a week.
Ok, but a week is still a 50,400% increase over 20 minutes.. You never needed something right now?
East over here hacking on his own pacemaker.
Im with Christoph on this one. Even if you have a $3700 fiber laser engraver sitting around. Its too cheap to get far better by farming it out to bother messing around with diy.
Shhh. Nobody tell them it’s cheaper to order your 3d prints too.
Only if youre counting the price of the printer with the print, and even then youll often come out ahead on the print if youre cheap on your printer choice.
I mean sure if youre need a metal print or a multijet print youd probably save going with a service but FDM and mSLA not so much so.
Print Bureaus dont do it for fun, they do it to turn a profit.
This guys running a $3700 laser that the company sent him to test, he didnt even buy it yet. Hes only going to be able to do 2 sided boards unless he gets really creative and like Christoph said ” can get 5 50×50 6 layer boards for less than €2.
Honestly I had a friend ask me to print a whole ton of stuff for him, I said “ah it’s probably cheaper to get JLC or someone to do it” but after pricing it up there was almost nothing in it, and a lot of the time it was more expensive than my base of filament plus electricity, and that’s me buying consumer quantities of both.
If you print a large piece and it fails once or twice it will definitely be cheaper to order it – and you will have better print quality and save time and headache to.
It’s kinda against the hacker ethos though.
And yeah I have a cheap 40 dollar laser attached to my print head for quickly removing resist from copper boards. It can’t drill or cut like the 3700 dollar laser but I can and have made a two sided pcb in under an hour. I like the process and it doesn’t matter to me that I use my time and maybe it costs a bit more.
The drive in oil places can change my oil faster than I can but I still do it myself. The examples are endless.
Not if you have to print multiple iterations to tweak a dimension or break something.
And to think: People build their own CNC machines, lathes and laser cutters. Some people even print their own photos at home. You can get all these things done online and have it shipped to you. Some people even have woodworking tools or do their own painting. Crazy, right? :)
Whilst this is correct, the advantage of this is much the same as having your own 3D printer instead of farming out print jobs – you can iterate the design in hours rather than days / weeks. With that bing said, through hole and double sided / multi layer boards are very difficult to do, and very very difficult to do well. I’m sticking with JLCPCB for now, but this is certainly interesting. I have a 40W diode laser cutter, might be worth investigating if it can vaporise the copper with enough passes.
I have cnc pcb mills at work and at home, and yeah, the iteration is the point to this. We can go from schematic to usable board in about 2 hours for the first one, and subsequent design iterations are about 45 minutes each. We spend a lot of time doing testing for radiated electromagnetic emissions, and that’s a LOT of work in switching out chokes and moving capacitors around. (A good/bad thing about milled boards is that without soldermask you can put parts anywhere if the other end needs to be ground, but they also have a lot of solder-short-to-ground issues.) We regularly go through 6 iterations of a layout before we find one that we decide is worth pursuing, and then we can wait three days for our quick turn board house up the street to crank them out for us.
Where can you get them for that price?
“We are getting really close”
Still only single or double layer without proper vias. A protoboard or deadbug style circuit is still the only way to quickly make a prototype.
2 layer with vias is doable at home: https://hackaday.com/2021/06/11/open-source-method-makes-possible-two-layer-pcbs-with-through-plating-at-home/
The quality is lower than a professionally made PCB but you can have your PCBs the same day and are good enough also for SMD components. On top of that some people like to tinker with this process so we’re still making PCBs at home. I do.
It could be a great way to make a small extension board for a smd component to make it breadboard friendly. Then 2 layers should be enough in most cases.
I find that a lot of stuff can be done in 1 layer, or 1 layer + solid ground plane. And getting the PCB right now does wonders to “keeping in the flow”.. way too often I have had PCBs arrive and then lack the motivation to assemble them.
Indeed. A full GND plane for the 2nd layer is a good strategy. It avoids machining the other side (and it’s alignment issues), and putting GND on the bottom frees up a lot of area on the top layer. And having a solid GND plane should be part of any PCB in these modern times.
As C mentions, Via’s are still cumbersome. PCB rivets still seem to be the best way for DIY via’s but I never used them.
An Idea that may work for via’s:
1. Stick a wire though the hole.
2. Bend the wires in an L shape.
3. Solder one end of the L to the PCB.
4. Cut off the excess wire, either with a precision side cutter, or a knife.
4. Repeat for the other “via’s”.
5. Turn the PCB around, clamp it on some heat resistant surface and solder the other side.
I think the critical point with this method would be the quality of the clamping. The PCB will bend a bit. Lots of clamping stategies can be tried. From a sponge with a few layers of paper over it, or just putting kapton tape on the back.
It may even work if you solder the GND plane first. With a bit of luck (and thin wire) there simply is not enough heat transfer through the wire to melt the solder on the GND plane.
Putting the GND plane on a wet sponge also helps.
As the PCB gets more complicated, you can draw “wire bridges” by enabling a few extra layers in your PCB program, and use wire instead of the copper PCB layers for your prototype. But overall, this is getting into the area where ording PCB’s is the much better option.
DIY PCB’s are mostly useful for the “simple stuff”, or when quick turnaround is a big benefit. For example designing PCB antenna’s, where you can do multiple iterations on a day to optimize the design.
Suggestion I saw on the video comments: “Sew” one wire back and forth through every single via on the board like a messy rat’s nest. Solder the tops and bottoms. Then cut off the excess wire. My problem with that is that I use a hot plate for soldering and that won’t leave the bottom a flat surface.
That was me, because I’ve done this a zillion times. If you cut the wires flush, it is possible you might be able to put it on a hot plate and let the bottom side solder melt and do your work that way. Or, what I do: solder the rest of the board and do your vias last.
I’ve also used the nc drill file on my cnc mill to drill an aluminum plate that has lots of holes in it, put that on the hot plate, and put the board on top, so the lumpy bits are all in the holes, but that’s a LOT of work and we only used that a very few times for an extremely specific, high-price/high-value board.
Another option is to buy thinner board instead of the standard 1.65 mm stuff. You make two single sided boards – front and back. That way you can solder components on both sides at the same time in your reflow oven without using different solders and temperature profiles. Then you simply glue them together and solder or rivet the vias.
How does getting an USD 1200 gadget for free qualify as “not sponsored”.
I’m ok with companies sponsoring youtubers by giving them free gadgets for review, and I also appreciate the disclosure of the youtubers saying so, but please do not call it “not sponsored”.
And after that it gets a bit complicated. Many of the youtubers are honest, and if a product does not live up to it’s expectations, they say so in their review. Other less hones youtubers praise all products in order to get more stuff for free. And the “real world” is somewhere in between. Youtubers have to generate enough interesting content to stay relevant, while not pointing out too many of the products flaws in order to get more stuff. So where is the line? You have to judge both the motives and the knowledge of the reviewer to set the scale. Stephen Hawes’ main business is his SMT PnP machine. I hope it goes well with those machines, and growing his youtube channel with other related content (another common strategy) works well for him.
You underestimated by a bunch,
Xtool f1 Ultra isnt a $1200 gadget, its a $3700 gadget
Thanks for the note! You’re totally right, I should have used different wording. xTool didn’t pay me for the video, and had no editorial review. I was just trying (poorly) to indicate that the opinions were my own. I’ve left a pinned comment on the video explaining this, and will be much more explicit in how I say these things in the future!
What about using the AtomStack A20/K1 Pro 1064nm?
It is clearly cheaper (https://atomstack.com/products/atomstack-a20-pro-1064nm) and has a bigger (400 * 400 mm) engraving area
Wouldn’t using a cheaper engraver clearly reduce costs if you make enough PCBs?
There’s honest and there’s “honest”.
I’m very dubious about the “free but not sponsored and I have full editorial control” reviews, the temptation to play down the bad point is just too great.
I pointed this out on one review. The reviewer got quite annoyed, stated he definitely wasn’t a shill, and then told me I was wrong by basically quoting the product puff piece back to me. This was for a LED light available in two versions and I pointed out the one he “reviewed” wasn’t the best for the job. Dude had no idea what I was on about, but of course I was wrong.
I agress. Dave at EEVBlog didn’t understand this either. Magically, things he paid for were crap, and things he got for free were great. Founny how are barin works. All you need to do is be upfront about if it to the audience. Dave refused to do that.
I think this is going to be game changing for little breakout boards for sensors and other chips I just want to play with or just boards holding a bunch of other little breakout boards and displays and buttons and pots so I don’t have to either use a breadboard for prototyping or manually wire things together in a once-off prototype. Or see if your dimensions and wiring are correct before sending the board off. Who would wait a week or more if they could have that in 20 minutes and move on with the project? By far most boards I get made by JLCPCB are little breakout boards that could be a single layer. Followed by “Does this layout feel right? Does it fit in the enclosure I’m planning?”.
At the other end: You can get things printed or machined or laser cut or injection moulded or whatever for cheap online yet many of us do these things at home anyway. I think this fills a similar niche. We’re hackers, makers, DIY types, etc, right?
It’s easy enought to wire up proto boards for stuff like that. Just get a supply of smd “adapters” so that you can have the right pin count and pitch when you need it. Other than that, a spool of “wire wrap” wire and a roll of solder and you’re in business without the headache of laying out and fabricating a board.
I have a two litre container of those smd adapter boards. I still never have the right one. Not everything is even a rectangular chip with legs anyway.
So what about putting solder into a 3D print-like head and having the machine squirt just a bit of molten solder into all the vias on one side, then same on the other side?
Might have to pre-heat the via surround a bit to get it to take. or dwell a bit at first touch.
Probably want to use extra high-temp solder.
Try it manually with one hole drilled through a double-sided pcb. Obviously isolate the surrounding copper so you’re not soldering together two giant copper heat sinks. I suspect you’ll quickly find that the solder does not want to go into the copper-free hole.
This is what I found. You cannot get solder to bridge through a standard thickness pcb. As soon as it heats up, it balls up and retracts. Try bridging over the edge of a board from one side to the other. Very frustrating. Down in a high aspect ratio hole, is even worse.
Probably better to use conductive ink or epoxy.
He also said something about different pcbs from different times seen to have more copper. Hate to inform you but you can buy pcb material with different amounts of copper. But then if you had designed things back before CAD/CAM, you lid know these things because you had to, too do your job and speed out the material for the job. Buying FR4 because that’s the norm is really a silly comment.
On the price against the short order houses, I think many of you need to actually read what he said. First time is not free then add in the cost of silk screen materials washing and cleaning the screen and equipment after. It’s much easier to get a four layer pcb with a ground plane and power plane with traces on top and bottom. Then make a board that should be four layers (maybe) and have power lines running next to signal etc.
But hey if it floats your boat go for it.
I could see making a quick 1-up pcb for a project but not at the price, maybe fiber lasers will be in the sub $1K someday.
LPKF has a solution to the double sided via issue using a conductive polymer that is squeegeed into the via holes with a vacuum bed and then baked. A fixture with 3 tooling pins takes care of the registration.
I could see making a quick 1-up pcb for a project but not at the price, maybe fiber lasers will be in the sub $1K someday.
LPKF has a solution to the double sided via issue using a conductive polymer that is squeegeed into the via holes with a vacuum bed and then baked. A fixture with 3 tooling pins takes care of the registration.
Search youtube for “alphacrow hldi” – a russian diy dryfilm pcb laser exposing device. Many many succesful implementations. You can gut an old inkjet and build one for cheap.
I can’t see much toxic effect from the FR4 epoxy, however the evaporated copper should worry one more. Keep that in mind when playing with fiber lasers in PCB
etching.