Anybody actually milling pcbs with shapeoko?

[kf4mot]

Just wondering what kind of results I can expect.

[zimmer62]

I started to, I didn't have much luck yet with my board design that I was using in the past with the etching process. The traces were too close together and I just couldn't get my cutting bit to cut shallow and thin enough... YET...

I expect with a change in tooling, and tightening up some of my z-axis issues plus a board design with further spacing between traces it should be fine.

[kf4mot]

What's the finest pitch part your working with?
Think it will work with 20 mil pitch parts?

[gazorp]

Sure. I think 10 mill will be no prob with proper tooling and calibration.

[kf4mot]

Thanks for the info.
Now I just got to wait for the next production run.

[zimmer62]

proper tooling and calibration

What do you suppose the proper tooling is here, and what steps would you take for calibration?

[gazorp]

proper tooling and calibration

What do you suppose the proper tooling is here, and what steps would you take for calibration?

Tooling: Use proper pcb endmills at the appropriate feed and speed rates. Good kyocera end mills are available on ebay in many shapes and sizes. Strong clamps required for copper clad.

Calibrate: Triple check all axis movements are precise and repeatable. Use high quality endstops to help with repeatability. Spindle should have less than .0002" runout. Bed should be planar to .002"

[kf4mot]

proper tooling and calibration

What do you suppose the proper tooling is here

I've read good things about these bits here.
http://www.drewtronics.net/

[kosme]

How are you making the copper clad perpendicular to the cutting bit?

[cvoinescu]

Kosme, I suspect that's not terribly important; as long as the spindle is not visibly tilted to one side, it's fine. What gazorp said is much more important: low runout and flat bed.

I was thinking (idly and based on no experience at all, and no ShapeOko yet either -- batch #3 here) that it may be easier to measure how much the bed is off from flat (e.g. by slowly lowering the bit until it makes electrical contact with the copper, at various points) and then have a script alter the Z coordinates of the G-code to compensate. The ShapeOko has enough resolution on Z for that to work, I think. At least for me, figuring out the math and writing a program seem easier than mechanically calibrating the ShapeOko to that precision, and keeping it calibrated. What do you guys think?