Milled on OX CNC running belts for X/Y. TinyG for controller and auto-leveller.

Milled on OX CNC running belts for X/Y. TinyG for controller and auto-leveller.

Used 0.1mm vcarve bit.

I am impressed at the resolution of the traces. Next time I need to make the pads large enough to accept larger drill holes and some extra room for solder. This at least should help those thinking of using a similar setup to mine.

Of course I used Chilipeppr!

Nicely done!

I build boards like this all the time for work. If you are putting wires through the vias, rather than components, or if it’s going to see any vibration in use, make the pads around the vias massively larger, like 4x4mm copper pours if possible, because the epoxy adhering the pad to the board gets beat up during soldering and is really willing to come loose in usage. Also, it’s a lot easier to solder all the side-to-side vias by stringing a single piece of wire through all of them, soldering them on both sides, then cutting the wire away to produce your conduction paths. Trying to stick individual bits of wire through each hole and solder them is a huge pain.

Hi, may I ask how you get so nice looking PCB. I’m also trying to turn my own PCB’s but I get rather ugly PCB’s because of copper that is curling up on the side of the traces leaving behind a very rough finish.

The CNC I use is a Xcarve with what it supposed to be 0.1 mm Vbit. I draw the schematic is Eagle->Flatcam-> chilipepper for final control.
I use the autoleveling function in chilliperper with a selected cut depth of ~0.05 mm. But this will always results in a slightly touched coppoer but not cut trhough. and when I adjust the depth by 0.01 mm it will completely mess up the finish by cutting to deep.

I’m not sure anymore what the problem could be, so can you give me some hint/tips/tricks on how to get some well done made PCB’s?

Kind regards,


Interesting, a couple of questions. Are you sure that after you’ve ran the probing did you load the probe data into chilipeppr workspace? Also, the depth adjustment from 0.05mm to 0.06 (1mm increment) should not be enough to mess up your finish unless the tip of your 0.1mm vbit has been damaged… I know this from experience. You should check it because it sometimes breaks off during probing which will change the characteristics and profile of that bit.

Finally, there will be “burrs” visible after you do your milling. I used 400-600 grit sand paper to sand these away. This left some additional cleanup where I used soapy water to wash away the bits that were embedded into the trace groves.There may be better ways to do this cleanup but thats what I did.

On our machine, burrs along the cut mean either the bit is dull/low quality or it’s spinning too slowly. Ours spins at 40,000 rpm, so that may not be an option for most people.
I agree on sanding it a bit afterwards.
I’m using a quite different machine/software, but on mine I routinely start by putting the bit down and cutting a line and moving it away and then looking at the line through a handheld jeweler’s loupe to check the depth, and adjusting it up or down a little bit to get the width (10 mils/0.25mm) I want my isolation cut to be. This may be difficult to do in your system. On the work system I have a mechanical adjuster that moves the collet up and down by about 0.05mm per click, and on my home system I can set it up to encode all the z moves (depth) as functions of a variable, and then alter that variable until I’m getting the cut depth I want, while maintaining the probe mapping.

I’m almost sure I load the probe data because after it the gcode in the windows show the text “all new Z”. And there is only one button you can press to load the probe data to the work-space and after many tries I should eventually have used the right button but it still stays the same. So that it for the probing data.

And like you said I also didn’t expect to mess up the finish by adjusting as little as 0.01 but apparently it does for me. But I need to have a better look for the tip, is rather difficult to measure :). But it will be my next point than to check. But for the state it is in now i don’t want to sand it because it is just to awful.

For the speed i can max go up to 27000 rpm, but it just the proportion between feedrate and speed to get a even nicer cut. But if the depth is totally wrong every time (e.g. 0.3-0.2 mm instead of 0.05 mm) it is no use in fine-tuning the feedrate and speed.

So in the end i still have to guess the right solution for my problem.

Something you could try, for improving the cutting depth issue, is shimming the pcb with paper. Each sheet of paper should add 0.05mm. If you put, say, five sheets of paper under it, probed it, started the run, and saw it was cutting way too deep, pull out three sheets and restart the run. If it’s cutting too shallowly, add three sheets. If your physical placement of the board is the same, that should preserve your z probe data (which is usually a measurement of unevenness in the board.)

I would avoid the paper shimming as the probe data will be off unless I’m missing something @John_Bump .

I router mine at 20K rpm and there were most definitely burs visible until I sanded it. You should post a close up pic showing the burs you are seeing… it just may be normal until you touch it up.

Also, I net dollars to donuts your tip got damaged ever so slightly when probing and it cracked off the very tip leaving a larger diameter tip… you can see how this would affect the height (probing data relative to the longer tip before it cracked off) and the width of the tip those screwing up your cut.

I agree wrt the tip chipping off. Carbide is surprisingly delicate, and microscopic chips can have a huge effect on cut quality.