CNC Jewelry Box

This is a 2.5D design for a CNC router made out of walnut with aluminum inlays and lids. The design plays with the positive and negative space of two intersecting triangles, and has a swoopy line that defines both the inlays and the handles for the lids.

This was an exercise to learn how to use a CNC router and my first time creating CAM toolpaths to make a 3D design. I hope this is useful to others as a project overview with some little tips for making your own project go more easily.

To make this jewelry box I used:

- Rhinoceros to design the object

- Autodesk Inventor to create the CAM toolpaths for the router

- Pier 9's 5 axis DMS router to route the wood

- Pier 9's Omax A-Jet waterjet to cut the aluminum

- Jointer

- Walnut - one 9" x 23" x 2" walnut offcut from MacBeath hardwood (they sell offcuts by the pound)

- Scrap plywood spoil board 9" x ~40"

- 1/16" aluminum 5" x 15"

- Sandpaper and shim-shaped sanding block

- Scotch-Brite wheel and pads for polishing the aluminum

- Cyanoacrylate glue (superglue) for adhering the inlays

- Safecoat Naturals Clear Penetrating Oil and brush to finish the walnut

I always start with a paper sketch. While the design may change a lot once I'm on the computer, that initial idea and the attitude of the object comes to me most easily on paper.

My paper drawing has a number of overlapping triangles, but after iterating a few times in Rhino, I landed on this basic design: just two intersecting triangles. I then created height and thickness. The bottom of the center pocket is the same height as the outer walls of the slimmer triangle. I Boolean-ed out rabbets for the lids and additional shallow pockets for the inlays. I moved the curving line so that it would cross the two pockets I wanted lids for and define a nice shape for little handles.

In thinking about designing the CAM for this piece, I started with my plan for workholding. Since I wanted to route the design out of the stock, I chose to use a spoil board.

Inventor is an amazing tool for doing CAM, but still requires a little strategy. After creating a facing path, I made a 1" 3D Adaptive toolpath to clear out the majority of the material and create a shoulder around the design. Then I set up 1/4" flat end mill passes to get the pockets, and 1/8" passes to finish the rabbets and make the inlays. For the final contouring I chose the 1/2" end mill and added tabs on easy-to-sand faces. That's five tools in total. I watched the simulation over and over to make sure that my choices made sense and that that the resulting part looked like what I wanted. I saved out the .STL of the resulting part to make my workholding map.

After jointing the backside of the walnut, I used a spoil board long enough to leave plenty of clearance around the clamp straps. I attached the spoil board to the stock with brass screws, using a print out of the routing path to put the screws outside of the design. I used screws that wouldn't come into the stock more than 1/4", the lowest height of my main adaptive pass.

I post-processed the G-code into separate files for each tool. I'm glad I did so because later the rabbets did not turn out on the first pass but I updated and redid that section of the work without it being a big deal.

I loaded all of the tools into the DMS and used the 1/8" end mill to set the workspace coordinates at the same corner of the stock as in the Inventor setup.

The facing path took our ugly scrap piece and revealed what a lovely piece of wood it really is!

The first adaptive path removed the bulk of the material, about a third of the project time overall and revealing the main shape of the design. After each section of G-code I swept up the chips and inspected the part.

A few issues that arose:

The finishing passes on the interior pockets snarled coming into the really tight corners, it would have been better to go slower there.

Making the inlay pockets left some toolmarks on the wall of the pocket, luckily these sanded out fine later.

And, as you may have already guessed, there wasn't enough stock for the tabs on the long end to be of much use. After the first contour pass that thin piece holding the design broke!

The waterjet is a magical, but very dirty tool. It comes with Omax Layout, for setting up the toolpaths, and Omax Make for executing them. The software handles the lead ins and outs and handles the kerf so that the kerf is outside the design.

To create the lids and inlays, I took the original design which had sharp points and created fillets based on the tool diameters, so that they would match the real object. Since there is effectively no kerf when using this waterjet I had to add some tolerance to the design so it will fit into the part. To do so I offset the exterior curves of the lids by 0.020".

One important consideration with using thinner stock on the waterjet is that it is flexible and the pressure from the cutter can push the material from the bottom up against the nozzle and clog or even break it. To prevent this I used extra quick release clamps around the edge to make the material extra secure.

Once they were cut, I took the very slight burr off of the parts with the Scotch-Brite wheel, then hand-polished the faces for a brushed look.

I sanded off the places where the router had chewed up the wood and stray tool marks, but left the contouring toolmarks in the bottoms of the pockets. Getting at the inside corners of the pockets was annoying until I found a shim to use as my sanding block.

I glued the inlays into their pockets with regular old superglue. I recommend this Instructible if you want to learn more about inlays: instructables.com/id/Handcut-inlay

I tested both the superglue and the finish on a scrap from the CNC process.

Happy with the way the Safecoat oil looked, I brushed two coats onto the piece, waiting 15 minutes between, then wiping off the excess oil.

That's it! Enjoy!