This is an early test of a method to cut box joints on CNC with the material laying flat.
Of course there are a million methods and jigs for box joints on table saws or other tools, but on CNC it seems especially nice if one can cut out the whole thing in one operation while the material (say plywood) is laying flat.
Since a traditional box joint has square slots between the fingers, it can’t be cut laying flat on a CNC machine because of the the square inside corners. By rounding the slots between the fingers, it’s machinable but then the mating parts of the other fingers need to also be rounded to fit properly. Both of these are fairly simple.
Where it gets a little tricky is where the root of the fingers meet the root of the corresponding space. Carving out a surface that leaves no gaps and can be cut by a flat endmill is not entirely obvious.
Next I’d like to try it in real wood. I think I might need to design in a tiny bit of clearance because it’s a little difficult to get it together, even in foam. I’ve uploaded a test shape to thingiverse: https://www.thingiverse.com/thing:3996745
I’ve had a bit of difficulty getting clearances as tight as I would like. The foam is forgiving in that I can mash it together and hide clearance issues.
Here is my latest attempt as of last night. Getting very close.
I think I need to adjust the clearance another 0.05 and I need a better solution for the inside corner. As it is now the z height of the top surface has to be perfect and flat while milling. I need a tiny beveled butt joint for the inside corner to accommodate imperfections and still look nice. I’ll update more when I get it completed.
That is definitely close enough to use though. Just sand and fill. I hope you’re planning on leaving the ends a bit long. It is tough to make them longer if you cut them short .
I dont know why the gap is so big. It according to my model it should close if the fingers are pressed together all the way. I am wondering where the error is, and at the same time I am ready to shrug and move on.
Maybe put a different spin on it. I actually like seeing that detail in the joint. Like you said if not just add a corner bead to cover it.
I do understand wanting to get it perfect though. So maybe you are not perfectly perpendicular, are you cutting them with the same gcode or rotating them 180 per pattern?
Single gcode cuts out both parts. Multiple CAM operations spliced together into single gcode file.
I had another idea for an approach I’ll post here shortly, which also covers imperfection in the upper surface Z. I’m definitely not above “cheating” to get a perfect seam. But I will cheat on the inside hidden surfaces rather than with a visible bead on the outside.
A little update, my previous idea didn’t really work. I thought I could have the parts collide a bit and sand down to a perfect fit while hiding non-flatness and non-squareness and it somewhat works on a test piece but for a full box it doesnt.
I now realize that a test corner needs only to be square in one direction but for a whole box each corner has to be square in 3 ways. I had never bothered to properly square my machine and it was off significantly so I added some hard stops to at least get decently square.
I have a new plan for cutting a bevel for the inside corner and aligning everything. We’ll see. I suspect part of the difficulty is the “precision” in the 3d carving vs. the 2d toolpaths which I have to work around, although I’m not sure. It is probably my fault.
I really want to get this perfect so I can buy $50 wood and make $300 boxes.
I have been thinking about how I would use the CNC for my knife handle work. I would want to mill into a handle 6 inches high.
Why not have a cut out in your table, lower and clamp the boards through the cutout, then just mill the box joints with the plane of the wood. You could run several in a single pass. You could also add spacers to your legs.
Is not nearly as sexy and impressive as your flat carved approach.
Yes, this would work, and it would be the straightforward way of doing things. You can do mortise and tenon joints and dovetails or more complex shapes. I’ve seen cool “anytail” shapes on youtube that are a generalization if dovetail joints.
The justification for this method is not having to mount the workpiece on end, and being able to mill an entire piece of furniture say with a single setup. But the motivation for me personally is more around being novel and unusual.
Nowadays there are special bits to cut flat dovetails. The old school way to do it is at the end of the CNC table you have a jig that holds pieces vertically like you said. Can’t really do that with the MPCNC so you’d have to have a hole cut in the table, it would be awkward to make it work.
A lowrider however would work great for this application.
Maybe hog out most of the material with the router and then switch to a chisel/shaper with a 4th axis to control the tool orientation. Then z up, z down, z up, z down to carve “handmade” dovetails.
Possible with enough effort, but no way will I ever be able to find the motivation for that.
.
