Weird, right? I was able to find some at a small webshop which may or may not still exist by looking for “pertinax”. FR-4 is definitely dominant almost everywhere else and easy to get. Perhaps Snapmaker should start selling some FR-2 PCB material for their new laser.
1 Like
Try looking up Bakelite PCBs, they seem a bit common and easy to find on Amazon. A lot of them list as bakelite phenolic.
2 Likes
You talk about drilling the PCB, isn’t it possible to do the holes with the 2W too?
Btw, I had surprisingly good results in cutting .0025mm copper foil with the 40W laser!
Had some https://amzn.eu/d/06ycy9pS that I needed to cut to some complex shapes and without much hope I placed it copper side up without anything on it as a first test, cut it with 240mm and 100% and got splendid results! Clean cuts and minimal charing on the paper.
Only issue is that small objects tend to blow away when they are freed, but so far I have been lucky and they have cleared the remaining workpieces so it has mainly been a question about finding them in the enclosure afterwards. 
2 Likes
Drilling the PCB must be possible with the CNC router. In the gerber drill file are all the co¨ordinates that are needed.
I tried, but it does not work well - so I got my 10W blue laser out and see how this cuts the PCB carrier material, and it does with ease, but the results are unsatisfying, as the cuts are heavily charred and not very even. So I’ll do what @buckfast_beekeeper suggested: Use the milling toolhead for the drilling (and potentially cutting out the PCB) - and that works well, did this before. In the past I had some success with PCB isolation routing using milling (FlatCAM is the tool that you want to use). The main problem with isolation milling is that you need to get the PCB 100% parallel to the toolhead movement, and that’s a bit of a pain. That’s why I was hoping for the 2W. Hole drilling and board cutting with milling allows for a slight slant in the PCB board, so that’s OK.
Huh, interesting… I tried PCB isolation routing with my 10W, and it does not leave even the slightest mark on the copper…Du you perchance have PCB board at your disposal? Would be curious if the 40W has success and has perhaps even faster results than the 2W…
Well, I don’t have any “real” PCB’s at home right now, but a epoxy based stripboard,
I ran a set of lines on it.
All 100% 40W with air assist with just the auto-focus set to 2mm
From the top,
1000mm/min
750mm/min
500mm/min
420mm/min
300mm/min
240mm/min
180mm/min
100mm/min
Surprisingly I even got a clean cut (measured no connection at all between them even before cleanup) at 1000mm/min!
At about 300mm/min I started to get some “peeling” of the copper where it’s lifting from the board.
The reason I had a stripe-board at my desk is that I’m doing some prototype work right now. Didn’t expect to be able to do this with a blue laser, but I think I have to buy a few boards and test it out!
2 Likes
Btw, just noticed that I managed to cut exactly at the holes on all lines. 
I initially did one test on a different part of the same board where I got the cuts at the “broader” part of the lines with the same results, but then I had 500 as the fastest speed. It looks the same, with clean cuts of the lines.
2 Likes
Holy moly! If I only had known… Thanks for testing!!! (poor Stripboard btw )
That really makes me think if I should get the 40W… If it were not so ridiculously expensive…
It died for a good cause.
Btw, will have some PCB’s delivered Thursday to try out some “real” testing.
And yes, it is expensive, but I must say that I like it more and more.
But, the IR is on the wishlist as well, since it would be nice to be able to cut clear/white acrylic and not have to use the CNC for that.
1 Like
Finally had some time at my hands to try a real PCB circuit. Will do a more detailed post on this later, when I maybe have full success. Until now, I can say:
- A small PCB of 2,5 x 2.5 cm size, holding 5 through-hole components, i.e. low-moderate complexity, lasered with 500 mm/min and 16 passes, 0.2 mm isolation width takes 2:19 hours. Better than I thought, but still slow.
- With above parameters, it was too sooty already (even with hardpaper), and getting the soot out of 0.2 mm slits, is painful to impossible. Isolation is not given.
- I used KiCAD to create the PCB and FlatCAM to generate toolpathes for lasering a reference frame (for work origin repeatibility), isolation routing, hole drilling with different diameters and cutting out the final board.
- GCode for laser needed a bit of modification
- Did a second try with 1000 mm/min and 25 passes (1:52 hrs), which in my previous tests was a very good result, but then (in the other test) with a 0.7 mm gap. With the 0.2 mm gap it left copper, so no isolation either, but this time for the opposite reason. Improved GCode here by including inline power (more modifications necessary), which may be overkill/not really necessary at such speeds.
So some good results, but no full success.
I think I will give it another try with 0.04 mm gap width, 1000 mm/min, 26 passes (meaning ~4 hours job…).
It is a bit frustrating how much fine-tuning this means, and I’m feeling like I’ll in the end not go down that route when creating PCB prototypes, but perhaps better create a reliable way to mill the PCBs.
My verdict would now be: Yes, doing PCB is technically possible with the 2W, but it is not practical.
@Dowser any success with the 40W?
And here’s a photo of the result. Right: first try, left 2nd. You can see that my first cutout was a bit too aggressive Btw. the first job for my 200W CNC. The Snapmaker 2 flute 3.175 flat end mill seems not to be the best bit to cut PCB…
Hmpf, getting more success and more frustration…
Did today 1000 mm/min, 26 passes, 0.4 mm gap (3:45 hrs). In most places, it is close to a success, but I am getting soot again. And in a few places something went utterly wrong and I burnt half the PCB thickness! Most likely I messed up something in the GCode while editing it, or it is a problem with Inline power…
What is really frustrating me is that the test squares with a wide range of parameters yielded excellent results, and now a single pass seems to change it from not enough to too much… I am missing something…
The good thing: the single flute flat end mill from Snapmaker does a nice and clean board cut. Here’s all three tries in a row:
Holding them against the light shows the differences:
And here’s my test squares held against the light:
You can see several squares that are perfect! Here are the parameters used for each square - left to right:
Top: 1500 mm/min, 30, 40, 45, 50 passes - gap - 2000 mm/min 40 and 50 passes - gap
Bottom: 1000 mm/min, 20, 25, 30, 35 passes, 500 mm/min, 10, 12, 15, 23 passes
Weekend over, not sure when I’ll have time for this again…
Btw.: Other experiences and tipps highly welcome!
2 Likes
Asking the obvious, want to help and do some brainstorming:
- Inline laser power is more helpful at greyscale engravings, isn’t it? I guess it would be more repeatable if it is turned off.
- Is your bed, x, y and toolhead parallel like it should?
- Is it possible that the copper layer thickness varies too much (maybe from plate to plate)?
I would guess it should be possible to get save repeatable results so it could be some little out of focus error?
What does the math say, how many copper (-thickness) could be vaporized at 1000mm/sec as example?
35um copper thickness means a removal of less than 1um per pass. Oh man the bed isn’t that flat 
.
I guess slightly defocusing could do the trick to not cut through but hard to tell.
Maybe the setting with the least cutting of the hard paper would be the best for the whole process?
2 Likes
The same for the PCB. That is the greatest problem for milling without floating tool.
2 Likes
Thanks @xchrisd & @buckfast_beekeeper for your input! I guess I can say something to that:
- Flatness of bed, PCB or parallelity of the toolhead path, variable thickness of the PCB: I experimented during my tests with intentional defocusing - the thought was: Defocussing would mean broader laser spot, would mean to achive 0.x mm of gap instead of x/0.02mm parallel passes I’d only need x/0.0y mm (y > 2) passes. I can say that a significant effect was only visible by 2-3 mm of defocussing. My print bed is certainly bumpy (the wasteboard not so much I presume), and not parallel, and the PCB may be a bit warped and imperfect, but we are talking of fractions of a mm - this is not relevant.
- Copper thickness varying between PCB plates: My test squares and my circuit tries are happening on exactly the same PCB board, just different corners of it - if copper thickness varies, it does so across this single board. Not sure how likely this is…
- Copper vapourization: In my test squares I vapourized 35 µm of copper in 25 passes @1000 mm/min, so a bit more than 1 µm per pass. What did not happen in the test squares was excessive soot, even with more passes. And that’s where inline power comes into play: My understanding of inline power is a) that I can specify laser power with each move by stating G01 Xxxx Yyyy Szzz - Szzz giving power for the move. That is what helps with grayscale, and very much so. But inline power also means b) that the laser power is adjusted to the real speed of the laser. During direction changes and when starting or stopping, the laser will be much slower than 1000 mm/min while accelerating or deccelerating. Inline power adjusts the power of the laser accordingly, so that the power depositied per distance stays roughly constant.
And effect b) of inline power is what I need: The isolation paths are more complicated than a simple square, and I have sharp turns of the laser. Inline power would be crucial here - and the nearly burnt through parts of my PCB tell me that inline power did not happen - and rightly so: I looked at my GCode and I made a mistake. Inline was not active!
This however gives me hope: The aginst-the-light-photo of try #3 does not look to bad after all, except for a few places, and they seem to coincide with the starting/stopping points and places with sharp turns. So I’ll correct my GCode, will perhaps add one more pass (making it 27) because it looks that in palces copper was left, and make sure inline power is on. Let’s see, this may finally yield good results…
When I read around for Laser inline power, documentation was surprisingly bad/scarce! My best source was actually @brent113’s pull request on Github. If you know more or better about inline power, I’d be glad to get references for a proper documentation - @Zoe @Jade do you perchance have something on the specific implementation of inline power with Snapmaker 2.0?
Thanks again for the input: Me thinking about it gave me these ideas!
im curious. i know this was a test of direct copper removal, but what if i used the laser to draw the circuit that i want etched instead. will that suffice?
what if
we painted a pcb with black paint
we then used the laser to burn the paint off all areas we want copper gone
then etch the board
the z-offset issues should be greatly eased because we are just vaporizing paint
That is what many people do, and you can do this with the blue laser with ease. There are many tutorials for that on the net. My goal is to avoid classical etching and the involved chemistry.
And I still do not believe I have a Z-Offset problem.
We expect to release the SM2.0 G-code command set this Friday, which will include some details on the M3/M4/M5 commands.
That’s great! I’ve in the meantime been doing a bit more reading around, and in the Marlin documentation states that for the adoption of laser power to acceleration/decceleration the Marlin config should contain LASER_POWER_TRAP.
/**
* Scale the laser's power in proportion to the movement rate.
*
* - Sets the entry power proportional to the entry speed over the nominal speed.
* - Ramps the power up every N steps to approximate the speed trapezoid.
* - Due to the limited power resolution this is only approximate.
*/
//#define LASER_POWER_TRAP
@Zoe do you know if this is enabled with Snapmaker 2.0? It seems to need a relatively new Marlin version - which Marlin version is currently the base of the Snapmaker-firmware?
The Marlin code used in 2.0 then did not have this option.
Additionally, the 2.0 machine is based on the 2019 2.0-beta version, which is a very early version.