First successful 5-iteration H-Tree

I’m still pushing the wood around when cutting, but it didn’t break this time, so that’s progress. Going back to a higher number of shallow passes for the next test to see if that helps with the small bits of the workpiece moving. I also need to find some balance between “stuck down enough not to move” and "can get it back off the spoilboard without breaking it. This time, I got “stuck down enough”, but had to break out the heat gun and a razor to get it back off the spoilboard (3M industrial double sided tape). May try the blue tape and CA glue trick next.

Looking good! Are you ramping into the cut?

Also… What is it?

Edit 1: I’ve been asked to make a tongue drum and I’m wondering if a truncated non-symmetrical pattern might work. Experiment time!

Edit B: A good quality plywood would probably be good for strength on the narrow parts. I’d be concerned about splitting along the grain with a hardwood.

1 Like

That’s not really designed to be easily removable. Definitely try the blue tape & CA glue (I like Starbond).

If you really prefer double-sided tape this is the one I use for CNC router work.

It’s a calibration piece / torture test more than anything else. I’ve been playing around with algorithmic pattern generation (see pygdk on github) and am torturing my machines to sort out autocalibration for the gcode generator. This particular run was straight plunged (yeah, I know) and cut in a single 3/4" pass (yeah, I know). Here’s a video with some callouts of things I did wrong.

Solid wood does fail along the grain exactly as you would expect. Here’s an early failed run in some very dry reclaimed mystery wood.

If I have time this evening, I’ll go back to playing with ramps, DoC, F&S, etc. and see if I can dial it in.

Regarding the tape, I’ll have to give the yellow double-sided stuff a try – I picked up a roll a while back (when ordering accessories randomly before I had the machine), but haven’t opened it. If that doesn’t make for happy mounting, I’ll pick up some blue tape and give that a go.

Thanks y’all!

Edit: A tongue drum looks like an interesting project – might have to play with doing one myself. I’m sure there are ratios and formulas and such for getting the tongues calibrated to the right notes, which is exactly the sort of thing I like playing with.


So if I’m reading right you are algorithmically generating the g-code too? I may sketch up a similar pattern to yours and see what Fusion360 comes up with in terms of cut path.

I think there are likely guidelines for size and shape, but my research says that tuning is needed without fail due to the variable density and grain structure in a natural product (wood). I’d love to see what you come up with!

Have you tried dental floss? I know it works on car emblems…

That’s a good idea!

In my experience on cars you still have to break out the heat gun or park it in the sun for a couple of hours.

Dental floss is indeed a great idea. Sounds a lot safer than a razor. Thanks for the tip, Les!

Yes, Nick, I’m skipping over the visual design portion and generating my gcode algorithmically using a python module I’m writing. It currently supports basic moves, some DRO-style operations, Logo/turtle-style moves, and most recently, L-systems for defining recursive patterns like this one.

I’m building up tool tables (compatible with CAMotics) as well as feeds-and-speeds tables which use manufacturers recommendations when available and general assumptions (like there is a known best constant surface speed for any given tool/workpiece material combination) when not. End goal is to be able to pick a tool, define the workpiece and desired features, and everything else is magic. I’ll probably wind up including a configurable fudge factor on the calculated feeds and speeds as you said because there are differences from specimen to specimen for even the same type of wood.

That sounds awesome! I’m afraid my programming days are behind me (for the moment, at least) but I’d gladly be a guinea pig if you need anything.