Here is the latest on shelving. I made some cubby-style cable organizers for all our HDMI, USB, etc cables.
One change I had to make from my previous joint systems is increasing the "slit" for sliding two edges into each other. I had used .107", but for this project I had to use a bigger slit, .114". I think as the slit gets deeper, it needs more tolerance. The smaller slit would actually skew the joint since the two sides of the MDF are different (smooth and rough).
I think we need two more shelves to organize all of our cables, with extra cubbies to grow into, but I'm out of MDF.
Each shelf uses about two sheets of 2'x1' (about $2 worth of material).
My design process involves:
1. Sketching it on the whiteboard
2. Scripting the geometry in combscript.
3. Exporting, using a script I wrote to adjust the stroke-width and image size and then send to the laser cutter using Bret's setup. (Script is attached. Use it: `coffee print.coffee myFile.svg`.)
4. Cutting, punching out the cuts, assembling.
5. Go back to step 1. Every project so far takes at least 2 iterations.
I'd like to think that steps 1-3 can be combined into a single step, done at full scale on a "dynamic drafting table".
I like combscript's geometric semantics--a hierarchy of rectangular regions, each positioned relative to its parent, with each measurement a number or math expression, with proper units. I find the iteration semantics, the region_grid construct, a little weak. There is a way to reuse geometry, however reuse cannot be parameterized. The syntax is about 2x more verbose than it needs to be. The booleans are broken enough that I stopped trying to use them.
It would be nice if step 4 could be made faster. Some ideas from easy to hard:
1. Let's get a vacuum for sucking up the small bits.
2. I don't leave the laser unattended, but it would be nice to be able to. What if we installed a webcam with a live feed so we could keep an eye on the cutter while in the vicinity, but not in the room?
3. I use the laser software's duplicate in a grid feature to laser multiples of an object. But it's not smart enough to know when it can reuse cuts. Reusing cuts would almost halve the time it takes to make these repeated rectangular shapes.
4. We could do camera vision on the material to automatically place cuts.