My First Attempt at Milling a PCB on a Mini CNC Machine

TLDR - so far good enough for single sided boards. Need to refine process for double sided.

For over a decade, I’ve been interested in milling printed circuit boards (PCBs) using a CNC machine. Chemical etching can be messy, and ordering from manufacturers can be a hassle, so I was excited to finally try it out at South London Makerspace with the help of Andy and Kyle.

To start, Andy had already purchased a set of “PCB” tools with end mills ranging from 0.8mm to 3mm on a 1/8 collet, as well as a set of 3.175mm V bits.

I designed my PCB in KiCad, adjusting the pad sizes as large as possible, the clearance to 0.5mm and the minimum track width to 1mm and then exported the file as an SVG.

In InkScape, I created a negative of the PCB by filling the white space with black and deleting all other layers.

Then, I used VCarve to generate the GCode for the CNC, selecting the Pocket ToolPath function and adjusting the cut depth to 0.04mm. For tools I selected an 0.8mm end mill and a v-bit. To offset the path and leave more trace/pad, I added a pocket allowance of 0.2mm for the v-bit path.

Kyle suggested we run the calculation, but then ignore the end mill clearance, only running the v-bit path, and this worked well and significantly reduced job time.

For the holes I adjusted the cut depth to 1.6mm and used the 0.8mm end mill.

To fix the blank PCB to acrylic, I used masking tape and super glue. Masking tape is applied to both the board the acrylic, and you then glue them together. The test cuts went well, but I found that the v-bit was either cutting too deep for double-sided boards or left nasty burrs when scratching the copper. Essentially the depth of the trace cutting was determined by my Z axis touch off. However, the holes drilled well and were good enough.

Overall, it was a successful first attempt, and I plan to refine the process by using 0.4mm end mills for cleaner cuts with less material loss. I’ll keep you updated on my progress!!

11 Likes

Nice work! Thanks for trying this out and sharing!

1 Like

Oh wow i am definitely excited for updates. This sounds like an amazing option to have.

2 Likes

In my latest attempt with 0.4mm end mills, I was able to achieve a beautiful etch. However, I encountered frequent breakages of the bits, costing nearly £4 each, which was frustrating. Upon reflection, I realized that the spindle speed was too slow. To address this issue, I have ordered a variety of bits from Ali Express and plan to conduct further testing in a few weeks.

In the meantime, I decided to try a new approach by purchasing v bits specifically designed for PCB manufacturing. I utilized a 10 degree 1mm bit for engraving and employed the profile toolpath with the outside vector, offsetting it by a few tenths of a millimeter. This method proved to be very successful, and I used an 0.8mm end mill for the holes and a 2mm end mill for the mounting holes (although I made a classic mistake by confusing radius and diameter when cutting out the board). The end result was my first fully functional board, which was more than adequate and would work well for double-sided boards. However, the settings for the 0.8mm still need work, as the cutting process is currently painfully slow. Despite this, I am thrilled with the excellent progress I have made so far.

The board? A very simple breakout for a buck converter, exactly the kind of basic utility pcb that this process will be invaluable for.

![miniCNC-3|375x500]

Another broken bit…

The etch with a 0.4mm end mill before it all started to go south…

Screenshot_2023-04-18_10-19-11

The v bit that worked well for me.

![miniCNC-8|375x500]

A successful cut (design errors aside)

Soldered board

Top of the board.

6 Likes

The ground trace between in and out is over designed. Without it, if one soldered the board back to front it would still function, with it, it melts.

Actually looking at it, I think I have soldered it back to front! - I’m on a train so can’t check. If so I’ll have to cut the trace. Glad I spotted that.

As for the design? I’ll revise it.

Also happy to share the files for these experiments on GitHub if people are interested?

Could have sworn I continuity tested it thoroughly with a meter, still it was 6:15am. Luckily didn’t run any current through it.

Just did some boards today and the tip kept snapping on the V bit. I ended up using it snapped, which was good enough, if irritating

2 Likes