Dude…you got it, right!? Get the thickness set and bingo bango? How freaking cool is that. I have a table to build, this joint and James’ table are really getting me in the spirit to do something a little more complicated than I was originally planning.
Thank you for the inspiration (and perseverance)!
Looking good. I will watch the video once my kids are asleep.
Are you going to try more complicated joints? Dovetails would kick it up a notch.
The design of the joint is straightforward enough; the devil is in the details. I made a little finger gauge where each finger is progressively wider by 0.1 mm (and each gap is progressively narrower) to fine tune the finger widths with less trial and error.
Even with this, I have had to make a ton of attempts because of one dumb mistake I’ve been overlooking all along and took me way too long to realize.
Now what I have not yet mastered is precise squareness and flatness. If I try to make a box and all four sides are parallelograms tilting the same way, then it can’t go together without straining the sides. I’ve got a plan for that. As for flatness, I have surfaced the spoil board and I have a spot for a dial indicator on the tool mount to check and if necessary shim the workpiece flat, but then wood is often not straight and I would prefer not to surface the whole workpiece unless I really have to.
That is great 
:
Surfacing the whole board seems like a requirement. If I were using conventional wood working tools I would plane the surface and square up the edges.
Somewhat related, I was considering some joinery on the lowly 2x4:
The outside area is effectively a butt joint and the inside has fingers. The part that makes this fun is the bit is far too short to reach the full depth of the workpiece but if the fingers are shallow enough then it works. And it has the same problem as the box joints above where the inside corners cant be cut sharp, so the mating fingers have to have rounded or truncated outside corners to avoid interference.
Sure, I could no doubt achieve something equally good with a jig and some traditional tool, but why would I finish the joint in 20 seconds when I can milk it for a couple hours of fun under the spindle!
Come on lets see that brain of yours go nuts. What about making some precision templates and using the hand router to get crazy?
Youtube recommended this video to me, it’s a joint of 3 beams, like the corner of a table. Not a box joint, but very cool looking and should be fairly easy to mill it flat!
Using the mpcnc to build a pantorouter is on my to-do list. https://pantorouter.com/
Yes, please!
Hahahahahaha!! I just watched that one last night!
I’ve been wanting one of those since Matias did his build of one.
So an update, this is a fail but in principle I think it could work. I am calling this a bleacher joint because that’s what it looks like.
Each finger has notches cut out of the corners to clear the inside vertical corners of the opposite piece, which won’t be sharp due to the tool diameter. These notches could have been rounded (like the original box joint) but to make the CAM faster I went with a simpler method for clearance.
My skills in Fusion CAM are improving but I’m still a novice. That, plus I have not disabled arcs, made the job take a bit over 3 hours. My phone would barf on such a long video but with OctoPrint’s built-in timelapse I was able to get a video:
After all the CNC work I cut the two halves of the joint apart with a bandsaw.
Unfortunately the bit cuts narrower than the ideal 1/8", for whatever reason, and the parts don’t fit together. I’m pretty sure this would work if the fingers were slightly narrower.
With the box joints I was also fighting tool width, or offset, or whatever you might call it. I need some method to predict more precisely the width of the fingers and gaps and either understand how to fix the tool physically or compensate in software. For some of the box joints I was deliberately understating the tool diameter to get it to cut narrower fingers and wider gaps.
This is a 3D adaptive with 0.5 mm stock-to-leave followed by 3D contour, followed by a 2D pocketing for the holes in the lowest layers. It is effectively doing a finishing pass so I don’t think that’s the source of my error.
Ah well, more learning.
That is so many glue surfaces it is almost wood velcrow!
Hey Jamie, my brother sent me this after showing him your project
he totally cheats at the end with a bent sander. lol. kidding.
I would love to see this attempted.
Yeah those are neat. I’d seen ones like that before. I am going to rebuild my other MPCNC to have a hole/replaceable base and maybe do some joints like this.
@jamiek I’ve gotta show a little respect for this work here, great job. This is pretty cool. I would like to be able to do this. I’ve started a test design and have my tool paths planned out and I’m looking for some improvements. From your videos, it kinda looks like you are using a flat endmill for the entire process, including while doing the contours. Is that correct?
I’m attempting to do the entire process with a ball endmill because it appears as though it will save me a lot of time with the contours.
I was reminded that I had intended to show how the compound joints are generated geometrically.
To start with, most of the finger is just a rectangle with a cylindrical top, and the base where the fingers meet the main body is also a cylinder. The challenging part is the compound curve where the tops of the fingers meet the base, which is not entirely obvious.
This compound curve can be generated by starting with a circle corresponding to the curvature of the base, taking the point where it touches the main body of the finger, and sweeping upward in a circular arc over the top of the finger (which is already a cylinder).
When sweeping the circle in this way, it must stay horizontal. If it were to tilt, for example if you revolve around the cylinder on top of the finger, it would produce a torus and that is not the shape we’re after. It must stay flat, i.e. it looks like this when viewed from the end.
The disc by itself would leave some extra material behind (circled below) so the circles are extended to a hull between two circles, so it doesn’t leave any garbage behind after performing the subtraction.
It is tricky to explain why this compound surface works, partly because I understand almost, but not quite 100% why this works. To test that it works I took a solid block and subtracted two copies of the finger, offset and rotated corresponding to how a box joint is constructed. By taking this negative shape, it would be the ideal opposite mating finger shape. And I confirmed that the mating shape and the finger shape are the same shape, so it mates perfectly with itself.
Now for simplicity it’s much easier not to worry about all the compound curve, and simply cut a channel so you are guaranteed clearance. This channel is deep enough to remove the entire top cylindrical part, which is half the finger width. The width of the channel must be half the finger width too, to remove all the ‘green’ portion in these diagrams, although it can be wider.
I have only attempted these cylindrical box joints but it would also be possible to make rectangles with rounded corners. Conceptually, you can split the finger in half down its length and insert a rectangle, where then the finger width is more than twice the roundover radius. You would widen the space between the fingers by the same amount.
Anyone know how that’s actually done? Vid is 4 years old and I can’t find tool path recommendations etc
It’s seriously making me consider cutting a big slot in my table
For the arbitrary shape joint, the boards need to be mounted vertically. You need access to the end of the board, so that will mean a slot in your table.
In the years since these posts I was thinking maybe dog-bone joints and some half-dowels could make clean joints with less work. I’m not a fan of the dog-bone gaps from an aesthetic perspective, but they should have repeatable, defined shapes that can be plugged quickly in bulk.
