A while ago, in the MillOne days, there were pictures of people using the machine to make printed circuit boards. I am guessing that MillOne is a while (or maybe not) before it is available. Can the LongMill be used to do the same thing - route out traces fine enough for PCBs? With that being said, ‘can it be done’ is not the same as ‘I’ve done it successfully’, so maybe my question would be better worded that way.
I think the LongMill being a more robust design over the Mill One, if anything, would be even more capable of cutting PCBs over the Mill One. I’m can’t recall any of our members yet attempting it, but I’ll leave it for anyone else to chime in if I stand corrected.
As far as how, you can refer to an older guide that we put together a while back:
Or refer to Chuck’s or Tom’s videos:
Thanks for the very fast reply. I’m planning on tooling up in the next few months, so I’ll let you know what I end up using.
You’ll have to be very careful. PCBs are notorious for not being truly flat and, given the depth of cut is very shallow, easy to get traces that haven’t been cut out or traces that have been cut through. Your best bet is to look for a combination of software that can probe the surface to account for irregularities and apply it to the Gcode that cuts the traces. Another option I’ve heard seems to work reasonably well is to put together a vacuum table to hold the board. It not only keeps the board in place, but helps to flatten it out.
If you don’t mind doing etching, there’s another technique where you cover the board with resist and use a laser in a CNC machine to burn off where you want to remove copper, then etch. The downside of this is it must be well-ventilated as the burning resist often produces toxic VOCs (volatile organic compounds).
As you experiment with this, do keep us posted.
Michael, I appreciate the feedback. Seems like this (board flatness) would be an issue with any cnc-based PCB routing solution. A vacuum table seems like a good option.
I’m not sure if I want to mess with coding - not that I can’t, I would rather spend my time making boards and not fuss with reverse engineering/debugging code.
I have done plenty of etching over the years, so I don’t mind the laser/chemical route. From what I have read, the resolution can be really nice. The feature I like about doing the entire operation with CNC is the holes can be drilled in the same process, especially nice if you are looking at 100+ holes per board or multiple boards.
One of the video / youtubes I watched addressed an issue with collet wobble/instability mounting a 1/8" collet into a 1/4" collet, so that is another factor I need to look at.
I have also looked into the additive technologies of laying down conductive pastes and what not. The big downside for me with that technology is that you are dependent on the supplier/vendor for all the different pastes and goo. One of my goals is to be self sufficient, other than buying copper clad boards, which can be found almost anywhere.
At the end of the day, I am willing to pay for a solution that works, so it may be a decision between the cost of a dedicated PCB mill versus a general purpose mill that is not as precise. Perhaps if I can source a spindle that has a 1/8" collet, there might be a case for the larger capacity Sienci with a vacuum table. As long as the precision is there, that is what counts. I am fine with switching out a spindle when higher power/larger end mills are required.
I will keep the post updated with my findings and final decision.
If your primary use is going to be tracing circuit boards, you might be better off using a small spindle instead of a handheld router. A 500W or 600W spindle will have more than enough power for that application, plus run considerably quieter. And most of them will support software software control over spindle speed (if you can figure out the wiring).
They use ER11 collets, and you can readily find either the bits you need in metric sizes, or 1/8" or 1/4" collets to fit in the ER11 chuck.
There are gcode senders that do auto leveling - You load your gcode file and indicate your Z zero on your stock, or board, and mount a switch/probe in the router. I made a probe from a piece of aluminum and a momentary contact switch. It connects to the same pins the XYZ zero block connects to, A5 and ground.
Then the sender probes the stock surface at intervals you specify, develops a height map, and adjusts you gcode to account for irregularities, like bumps and dips, in the stock. One I’ve tried that works pretty well is OpenCNCPilot. It’s free and available at https://github.com/martin2250/OpenCNCPilot/releases. Try the next-to-the last version 1.5.8 - 1.5.9 stopped working for me.
BCNC also does auto leveling but I haven’t tried it. UGS is supposedly coming out with a version soon, and cncjs has one that works in the nde.js version but not yet in the desktop version.
You can also use the same process to probe a surface, transfer the coordinates into MeshLab, and create an STL of the surface.
@TGMan If you decide to stick with router rather than a spindle, Elaire in the US makes 1/8" collets for the Makita so that you would not need the adapter to go from the stock 1/4" to 1/8".
Hi all! Apologies, for the bump, but there is such good conversation here, and I figured keeping the past context would be more helpful than starting a new thread.
I’m super close to ordering a Long Mill, and while most of its use would be for wood projects, a significant amount of usage would be for PCBs. I’ve found some pretty great resources in general for milling PCBs (such as this DIY TECH BROS video in addition to the above links), but no confirmed success stories on the long mill.
So i’m curious if anyone in this community has attempted this recently?
Thanks so much in advance!
@nightgen I’ll just welcome you to the group. Ken. That is the limit of my reply, though, since I’ve not tried to mill a PCB. Good luck.