Practical pros/con of laser VS milling of PCB's?

[Mephitus]

I was wondering what kind of experience any of you have had with CNC laser cutting PCB's as well as milling?
If you can either comment or explain what kinds of practical limitations or benefits that each method provides over the other?
I am mostly curious specifically for DIY machines and not professional prototyping kit.

[mikeselectricstuff]

There is no such thing as a DIY PCB laser cutter. LPKF make a very expensive one - well into 5 figures I believe.

[onesixright]

I have none, so no experience. However i did read a bit about it. I have basically seen two versions, and a few serious issues.

• A solution where you paint the clad dark (black), the laser etches the traces. After tat you can go in a conventional way of etching. This basically is a 'easier' of printing your circuit on the board. But still there are the chemicals...
  
• A more power-full CO2 laster solution.

• Now what i understand is that the reflection is a serious issue. Because of the shininess, it basically reflects the light.
• Then there is the depth thats hard to control.
• And last but not least here is toxic fumes that needs to be dealt with.

Maybe go with CNC milling?  (that was my conclusion...)

[mikeselectricstuff]

Good old fashioned chemical photoetching is by far the easiest & cheapest way.

[KL27x]

It gets easier and cheaper. Unless you're doing your own plated vias, I think it will be rare you can beat toner transfer for time, effort, results, and cost of materials. Unless you are doing 4-6 mil traces on huge boards. Just watch Ben Heck's instructional video and do not do anything he does.

[PA4TIM]

I just took an old CDROMburner apart for the laser. But I'm not sure if it usable. I have a HP plotter and want to mount the laser instead of a pen. My idea was to spray-paint a piece of pcb and then remove the paint with the laser.  I used the plotter before to draw the pcb layout direct on the pcb. That worked very well with one type of permanent  marker and after they stopped selling that I could not find a good substitute.

[Sbampato12]

I have none, so no experience. However i did read a bit about it. I have basically seen two versions, and a few serious issues.

• A solution where you paint the clad dark (black), the laser etches the traces. After tat you can go in a conventional way of etching. This basically is a 'easier' of printing your circuit on the board. But still there are the chemicals...
  
• A more power-full CO2 laster solution.

• Now what i understand is that the reflection is a serious issue. Because of the shininess, it basically reflects the light.
• Then there is the depth thats hard to control.
• And last but not least here is toxic fumes that needs to be dealt with.

Maybe go with CNC milling?  (that was my conclusion...)

Copper is like mirror to CO2. Low power machines/tubes use copper mirrors sometimes.

But I've used this method once, I've paid a laser company here to 'etch' a black paint layer on a pcb. Then etched the board. They used a 'marker' machine, with 'fiber laser'. But the results isn't that different than the usual toner or photo process.

[Sbampato12]

Some years ago I've found a guy (German) on internet that build a exposure machine.
He put a photo-painted pre-sensitized pcb on the machine, and the laser do a exposure where the traces are. He didn't need to print the art on transparency.

Looks great, used a rotating mirror in front of a blu-ray laser or something like that. Small machine.

[Mephitus]

I see now I was missunderstanding what processes where at play.
It would seem that a UV laser of industrial power would be needed to cut copper at any kind of realistic ability. And that some very high end products are only newly capable of doing what I had originally thought was an otherwise simple concept.

[nbritton]

I have never milled a PCB, but I have owned a CNC milling machine. I can tell you that tooling for a mill adds up to a lot, end mills are not cheap. I broke most of my end mills through incorrect milling technique or programming errors, the are unforgiving when it comes to mistakes. As an amateur I found it took a lot of time and money to keep my mill fully operational.

[lukier]

I've tried paint + laser method, and after multiple attempts (laser power, focus, type of paint, feedrate) it can be done in principle.

In my case I've used my CNC 3020T with a 1W 445nm laser, high temperature black matt spray paint and modified FlatCAM slightly to generate isolation milling G-codes, but with IO on/off for the laser.

The problem is that this is awfully slow. 1W laser is not super-powerful and isolation milling takes a long time (but probably faster than actual milling with a conical bit - there feedrates have to be based on the tool and material).

Second issue is that even evaporated paint can leave transparent coating on the copper that is impenetrable to the etchant. Usually this requires some extra solvent (isopropanol), maybe ultrasonic bath and a lot of care to clean out places in between the tracks without destroying the remaining paint.

Another problem is quality, isolation milling doesn't produce nicest looking PCBs by its very nature, also I've tried multiple passes and that can leave some tiny bits of copper here and there.

Unfortunately, I don't have high power CO2 laser cutter. I've read some success stories there, if one disables air assist and vaporizes paint back to front (assuming the exhaust is on the top). Also, as someone mentioned, copper is a mirror for LWIR, so I would be worried about destroying the focusing lens.

Anyway, for now I'm back to doing toner transfer. It has issues (mostly registration and scale problems, CNC drilled holes don't often match the printed pattern very well on bigger boards - poor laser printer I suppose).

Here some photos of one of the first "successful" PCBs done with the laser + paint method:

[lukier]

Some years ago I've found a guy (German) on internet that build a exposure machine.
He put a photo-painted pre-sensitized pcb on the machine, and the laser do a exposure where the traces are. He didn't need to print the art on transparency.

Looks great, used a rotating mirror in front of a blu-ray laser or something like that. Small machine.

UV exposure might make more sense. UV laser optical power shouldn't be an issue and the machine could in principle move much faster in a raster manner (probably much faster than my CNC can).

Rotating mirror might work, but it would distort the focus and spot size at the extremities. That's why fast marking lasers use very expensive F-theta lens to correct that while passing optical power through.

I've read somewhere that laser printer mirror and lens assembly is of not much use, as it is designed for IR I believe and the optics doesn't pass much power through in the 445nm band (and UV as well I suppose).  Lasers from printers are rather low power.

[Sbampato12]

I don't use the CO2 + paint mainly because that, I don't know how the optics or safety will work with the reflected beans (if it is relevant). And I use the toner method, because I think it is easier.

If we had the sensitized boards here (in my country), I think I would use the photo method. But to spread a thin layer of photo-paint, and have to wait one day (without the drying cabinet) is not easy or fast.

[jwm_]

something I have been experimenting with recently is i cover a pcb in a adhesive vinyl sheet, then use a laser to cut the vinyl, which i then peel off easily. I had mixed results with paint, I may have just not found the right paint.

I think i may have the poorest laser cutter out there though, which limits my results. I will post a tear down of this ebay gem sometime. But it was only about a hundred bucks and does cut paper/vinyl and burn wood just fine.

[IconicPCB]

I use CNC milling method to make prototype boards for my clients.
I make single double and multilayer boards ( palladium activation process).

I have seen LPKF laser in action. and in fact have samples of its output.
A neat machine. It does not cut copper it only marks the surface and uses an acoustic effect to pop the copper surface off; a really interesting method.
Wish i could afford it but a couple a hundred thousand dollars don't come easy.

So I stick with milled boards. I get excellent results. My mill is home brewed far more capable than stepper motor driven commercial offerings.

[Kjelt]

I wonder when we will finally see a mature inverse process: printing conductive ink circuits on bare board.
There has been tons of research, I saw printed circuits on paper and thin plastick carriers but never something that could stand a reflow process. 

[tautech]

I use CNC milling method to make prototype boards for my clients.
I make single double and multilayer boards ( palladium activation process).

I have seen LPKF laser in action. and in fact have samples of its output.
A neat machine. It does not cut copper it only marks the surface and uses an acoustic effect to pop the copper surface off; a really interesting method.
Wish i could afford it but a couple a hundred thousand dollars don't come easy.

So I stick with milled boards. I get excellent results. My mill is home brewed far more capable than stepper motor driven commercial offerings.

Pics please

[Sbampato12]

I use CNC milling method to make prototype boards for my clients.
I make single double and multilayer boards ( palladium activation process).

I have seen LPKF laser in action. and in fact have samples of its output.
A neat machine. It does not cut copper it only marks the surface and uses an acoustic effect to pop the copper surface off; a really interesting method.
Wish i could afford it but a couple a hundred thousand dollars don't come easy.

So I stick with milled boards. I get excellent results. My mill is home brewed far more capable than stepper motor driven commercial offerings.

Pics please

Agree...

[The Magic Rabbit]

You can't laser off any material that is highly conductive to heat unless you use a high power laser. Even lasers that cut through 5mm acrylic struggle with standard FR4 board. It's just not practical.

Sent from my XT1039 using Tapatalk

[Mephitus]

This thread has been very informative, thanks for the responses. As a long term goal, I would like to build a precision CNC mill for both prototyping as well as for other home projects for the family. I have some boards I want to make and would like to use those presensitized pcb. I have heard good things. Anyone else have experience with these that can comment?

[IconicPCB]

Tautech,

Some pics...

Machie can process a panel 600 x500mm.
Spindle runs up to 100KRP with a low and well defined runout of the cutting tool.

Registration ( top to bottom ) is based on tooling pins inserted into delrin bungs.
Servo positioning includes position and velocity feedback to minimise following error during coordinated moves.
Control loop implemented in LinuxCNC and motion control PCI based hardware.

Linear Positional resolution ( via Renishaw linear scale ) 0.005mm.
Motor shat rotational resolution 2048 PPR.
Machine age 8 years.

Apologies for lousy images. Have an old phone.



To the Magic Rabbit

You are correct in stating no cutting with CO2 laser.
The trick is use a UV laser ( typically IR passed through a non linear frequency multiplying crystal mediu and then re focused as UV onto copper.
Again No cutting.
Rather use UV to achieve two goals... micro ablation followed by general area treatment to cause adhesive beneath copper sheet to boil off and detonate removing micro chunks of copper.

 

[lukier]

IconicPCB:

very nice machine and the spindle is just perfect for PCB processing (100kRPM!). With such spindle and rigid machine you can probably mill much faster than my crappy-wobly CNC 3020T with 200W DC motor.

However, your setup is quite far from "hobbyist" level I would say. The specs of your machine are probably better than some low end benchtop LPKF milling machines.

RF guys must like you, they love controlled impedance milled PCBs in Rogers material

[IconicPCB]

We have done some work with a local microwave company manufacturing 3D filetr structures in a multilayer Rogers material.
A real challenge to press and ensure good adhesion.
The client did not want the surface of the copper to be mechanically abraded. Adhesion of Rogers prepreg proved to be the challenge which tested our limits.
Apparently the copper surface needs to be treated with some ugly chemistry which we did not have.

Machine frame is indeed very rigid designed and built during soviet era . Actually I used to work for a machine sales organisation run by a German engineer. Enough said.

[Mephitus]

IconicPCB, that is one hell of an impressive setup.     Would you happen to have any documentation on its construction to share with the class?

[KuchateK]

Here is to link of machine exposing with laser mentioned above.




Another one uses converted printer.


Hardware for it is cheap, but printing a transparency and sticking everything under a lamp is quicker and the rest stays the same.

[IconicPCB]

Mephitus,

Sadly all of original drawings were pencil and paper, nothing in electronic format.

On the other hand Controls are implemented in LinuxCNC. Originally I used BLDC drivers designed and made in Australia. Drivers had a step direction interface ( for all intents and purposes they looked like stepper motor interface).
Set up was driven almost open loop that is to say LinuxCNC had no feed back signals. The loop was closed inside the drivers themselves.

When one of the drivers failed I replaced it only to find the driver hardware had remained the same but the firmware was optimised for a particular application and no longer behaved as need ( in my application).

This gave m impetus to do further work on the controls. A the new set up is based on analog drivers with tacho velocity feedback while positional feedback is applied to LinuxCNC through a Mesa 5I20 FPGA card runing hostmot2 firmware ( alla stadard setup under LinuxCNC).

Bearings and screws are preloaded  NSK parts.  Alfred Jaeger 33mm body diameter spindle completes the picture.

Spindle sits inside a purpose designed floating head assembly which allows for localised deviations in laminate flatness .

The work platen is tooled up with soft tooling. Delrin bungs are screwed into aluminium platen. The bungs are then drilled by the machine to accommodate 1/8" tooling pins. The tooling pin arrangement matches Gerber processing set up so that both data and product panel can be flipped over and retain precision inherent in the machine resolution.

Delrin bungs will wear out eventually or move with temperature and time. They are unscrewed and replaced with new bungs.

The work surface of the aliminium platen was also machined by the spindle itself to ensure a uniform surface at a uniform distance to the spindle collet.

The tooling pins on the milling machine match tooling pins on a laser direct imaging set up.
Laser machine is used to directly expose liquid photo image-able solder mask and component overlay.
Same Gerber data processing is employed to generatesoldermask andlegend information for both top and bottom sidesand same control arrangement is implemented. However laser machine is driven by linear motors and can achieve 15000 mm /min operation at 3000 mm/s/s acceleration.
9000mm/s produces good results at higher speeds and or acceleration ringing is observabe.
Finished boards are electrically tested on a PROBOT flying probe tester and if needed prototype boards are assembled on a Mechatronika M10V pick and place machine.

[The Magic Rabbit]

I want one of those laser circuit flatbed systems now. Ooooooh if only it were hot-swappable with a 3D print head, I'd be in heaven!

Sent from my XT1039 using Tapatalk

[IconicPCB]

It is.

All you need is the signal from the extruder to switch the diode laser driver on.

Get a sixty dollar laser housing.diode combo build a current driver to suit the laser diode .. and PTOINNNNGGG!