Saturday, June 28, 2014
Digital CNC Joinery I
Think about it, try to drill a hole in a piece of wood and expand that hole to a 1x1" square area by drilling more holes. No matter what you do, the four corners of that square will always have a radius in the corners (half the diameter of the drill), and that's just the problem with combining CNC, joinery and traditional woodworking in general. That also means you can't do many traditional wood joints with a CNC machine and that also explains why there are so few carpenters out there that utilize CNC machines in this way, for cabinetry for example.
There are a few companies that came up with some very creative and practical solutions to deal with that problem already. Thermwood for example created http://www.ecabinetsystems.com/ a few years ago with an ingenious way to make it happen.
There is also a good article over at http://makezine.com/2012/04/13/cnc-panel-joinery-notebook/, that seems to be based on the work of Jochen Gros’s 50 Digital Wood Joints project.
Turns out that all that research was done at the HFG (Hochschule fuer Gestaltung/Highshool for Design in Offenbach, Germany), and published by J.Gro's. Being a German myself, I was very intrigued to try that out, not so much to see if they were right, but more to see if the Momus could do what they've been doing in Offenbach back then... If you Google for that topic in general, it still seems all that is shrouded in a big cloud of mystery.
There's also already patents and lawsuits around that, which deal with how to join one piece of wood to another using CNC machines.
Long story short, I had to try.
Actually, I tried it twice - a long time ago when I wasn't really know what I was doing yet, mainly because I didn't understand flex and cutting forces and didn't know the difference between climb cutting and conventional cutting, nor the idea of a glue gap concept. It was a total failure back then. I mean, it worked, but the gaps between the two pieces were all over the place, take a look.
Lately, having the machine under better control and knowing a bit about these things, I tried it again and a after a few tries (15 in total), it started to work. :-)
To test what the HFG published, I chose the relatively simple half blind puzzle joint. This joint is ideal for cutting on a CNC, because there are no sharp turns. Other properties of this type of joint include a lot of area for the glue to join the fibers and easy assembly, it just slides in when done properly - Lego style...
After making 15 pieces of firewood, I learned that it's really important to tighten your machine up once in a while, to avoid any play and backlash. Use a sharp bit. Use a downcut bit, upcut will work too, but there will be fringes and fiber blowout. Divide your toolpaths into a roughing and finishing operation.
Machine the female area about 8 thousands of an inch bigger than the male (glue gap) This might sound silly in wood, but when doing something like this, there's a big difference in precision. I found that at 5 thousands, the parts would not go together well (having to apply force, clamps required, things would crack/break). At about 9 to 10 thousands, the joinery would wiggle a little, not providing for reliable assembly. The sweet spot seemed to be at about 7-8 thousands, the joint would slip into place with no "wiggle" room.
I felt like I just transitioned from 2d to 3d in terms of making things on the CNC, imagine what else could be accomplished with this simple puzzle joint.