When we first saw the assignment for this final, Joey and I had a hard time choosing an object to CNC. Most of the CAD models on Thingiverse (https://www.thingiverse.com/thing:1447610) are made to be 3D printed and thus most are either too intricate or have angles that a three-axis CNC cannot cut regardless of piece orientation. Joey ended up choosing a chess piece while I selected a model of Jabba the Hutt. We got the crash course in Fusion 360 from your YouTube videos and spent a while learning the software. We ended up scaling down both of our models and changing their angle in order to ensure that when we cut them in half all of the detail would show through on the CNC. We eventually realized that Joey’s chess piece would not work for this project because it was symmetrical so we continued with Jabba the Hutt.
When scaling Jabba down we realized that the height of the highest point of each half would need to be equal (or less than) the length of bit protruding from the router collet. If the height was greater, the router itself would hit the model as it cut and destroy it. We then made tool paths for the model using a 1/8th inch ball-nosed router bit and exported the tool paths to gcode. We decided to cut our model out of wood after you suggested that the wax might have issues with melting. Joey cut 2x4s down to rough size and we measured them. We made the stock in Fusion the same size as our 2×4 pieces. This allowed us to both cut deeper without hitting the router collet and also ensure that we would be able to build a box around the part to cast the Oomoo without issues. The first time we tried to cut out our two halves we spent two hours struggling with the CNC machine, reading the manual, and watching YouTube videos. We didn’t get anywhere. After talking to you, we realized that there were a few simple lines of our gcode that needed to be deleted in order for our file to cut. When we came back in, we had to ensure that the piece of wood was lined up in the CNC the same way that we measured for it when calculating the stock necessary. We then set the zero and began cutting. We were able to cut both pieces with no issues at all. Both the front and the back needed frequent vacuuming to get rid of excessive chip buildup. We set the first pass of the CNC to roughly one centimeter and thus had a relatively low feed rate of 24.
Front of Jabba
Side view of Jabba’s front
Jabba back
Cutting shot
Once the pieces had been cut I built up a box around each 2×4 section using cardboard and masking tape. I made sure the boxes wouldn’t leak before spraying each with mold release.
While we let the mold release set for the required five minutes we mixed up a batch of Oomoo. We poured a sizable amount of Oomoo into each mold in order to ensure that the molds would not deform when we placed rubber bands on them later. We came back the next day and removed the cardboard and tape before gently pulling the molds away from the wood. They came off easily, especially because we did not have any large protrusions from our model.
We then prepared our models for casting. We did not include any keys in our design because we thought we would be able to line the two mold halves up easily enough since both of our 2x4s were of roughly the same size. We measured in from two sides that were roughly the same distance from model to edge on both halves and made sure we always kept an equal space when lining up the molds. We cut our pour hole into the bottom of the model which was overall the right decision but left us with some bubbles as I will describe later.
I oriented, sprayed mold release, and rubber-banded the mold while Joey got the casting agent ready. We decided to use the SmoothOn 300Q because of its fast cure time. We started off using green dye. For the first batch we got a little excited and we didn’t mix the two parts together for long enough because we were worried about exceeding the 30 second window before the cast cured. Due to this, when we removed the cast it was still wet to the touch and not all of the green die had been mixed so it had a non-uniform color.
We quickly fell into rhythm with Joey mixing up the SmoothOn while I readied the mold and timed him. We realize that the best way to cast was to allocate 10 seconds to combining part A and B, 10 seconds to stirring, and 10 seconds to pouring and shaking the mold. The next few casts had huge bubbles towards the bottom of the figure (the top of the mold). They would not yield to my gentle tapping. As we progressed I realized that the large flat surface we were pouring into trapped bubbles in the viscous liquid. To fix this Joey stopped the pour right under the top of the pour spout while I angled the mold to roughly 45 degrees in each direction while giving it vigorous pats. This got just about all of the bubbles out. I had a few errors with alignment before perfecting my mold placement and, once we were satisfied with a green version, we tried two red casts (that turned out more pink than red).
Trials from left to right with visible bubbles on the first few casts
To clean up our casts I took a pen knife to the line between the mold halves since some material squeezed out of the sides of each mold. We then used a saw to cut off the pour spout area and used a piece of course sandpaper, laid flat on the reference surface of the workbench, to sand the spout area flush. This was the main benefit of locating the pour spout at the bottom of the model, we would not lose any detail from the model itself. We washed the shavings and our finger prints off before taking final photos.
Once we figured out the code-based roadblocks to our CNC-ing I enjoyed this project. It was interesting to chain so many complex tasks together to make such a small final cast. While I was happy that our Oomoo mold worked the first time, I was sad to see all of our work on the CNC used for just one step along the way to a final cast. I think that the final casts turned out really well. They even look like Jabba from a distance. The main thing is that the molds were able to copy the exact geometry of the CNC cuts and the final casts retained all of that geometry. That was the mark of success to me.
