“Do or do not. There is no try.”

As we started the final, we still had the preconceived notion of a chess piece in our minds- a relatively cylindrical piece that one could manipulate in their hands. As such, it was inevitable that we started thinking about hilts. More specifically, the hilts from Star Wars. And so, to thingiverse.com we went, scouring the site for an interesting .stl file. Unfortunately, the vast majority of files were made by CaseStudyno8, who incorporated multiple parts in his designs, making them un-CNC able. Luckily, IMakeShips happened to make a model of the LEGO lightsaber hilt, which is what we decided to go with.

That being said, Kyle noticed that the hilt had a fundamental issue with it; its center hole would mean it might interfere with the later steps of the project. To remedy this, he imported the .stl file into OnShape, using it as a base to recreate it. Of course, the main difference was that the OnShape model lacked the aforementioned central hole.

Once the new model was 3D-printed, we brought it to class, where its sizing was approved.

“In my experience, there’s no such thing as luck.”

Moving onto Gate 2, we used OnShape to essentially cut our model in half before adding a baseplate. On this baseplate, we added 2 0.375” circles (in diameter) and 2 0.4” circles (in diameter). The larger circles were extruded outward while the smaller circles were cut inward.

In addition, we used OnShape to generate an offset plane 1 cm above the tallest point of our model. Using the Up To Next feature of extrusion, we were able to create another part that represented the silicon we would be pouring in; the volume of this part is exactly the volume of silicon we would need to pour (about 130 cm³).

“Never tell me the odds!”

In class, we hotglued cardboard around our part (including a baseplate!) to contain the silicon we would be pouring in. Overhearing our classmates, it appeared as though the density of the silicon was greater than        1 g/cm³, thus meaning we would need more than our initial estimate of 130 cm³ of silicon. Going with a very conservative estimate of 140 g of silicon, we mixed 70 g of Part A with 70 g of Part B and poured the solution into the cardboard box.

Meanwhile, we still needed to CNC out the other half. Following the instructions provided, Kyle gave it his best attempt; he imported the .stl file into VCarve and used the provided tool files to create a roughing as well as a finishing pass. Unfortunately, his attempt ended up as the following image.

 

In attempting to figure out why the final result wasn’t as we had hoped, we looked at other people’s files (as they were all saved on the same computer). Interestingly, one of them used the default 1/8” ball nose tool instead of the provided 1/8” ball nose tool. Perhaps this was the cause. Once we replaced the tool on the finishing pass, two things were evident. The piece was now quite clean, but it would take a much longer time to finish. Perhaps the provided tool simply went too fast and couldn’t get a clean finish. Nevertheless, there was still a slight issue on our hands. The toolpaths now created a strange border around the part- was this the result of machining tolerances?

We assumed it was, and realized that if the boundary was changed from model to part, we could avoid this issue altogether. With that simple change, the simulation now looked like the following image.

But still, there was a slight issue. If the toolpath needed to cut the entire top face of the wood, where would the clamps go? In resolving this issue, Kyle peered at past final projects, specifically at Isa’s. He liked her idea of carving slots into the wood itself, and so he attempted to replicate it. Grabbing a drill and a 1/4” bit, he first drilled a series of holes into the face of the wood. After this, he took inspiration from the CNC machines, turning the drill such that the bit could cut into the wood.

“May the Force be with you.”

Unfortunately, the slots couldn’t hold the wood down; every time the bit entered the wood, it visibly shifted; we needed another method to hold the piece of wood in place. Luckily, Iñigo developed an interesting method utilizing other clamp types, specifically one that seemed to move laterally when tightened. Using this arrangement, he was able to start a cut, but it was interrupted due to class starting (also he used a past toolpath, so we needed to scrap this trial). That being said, however, it appeared the end was in sight regarding the CNC machine.

Hence, we came in one night, aiming to finally move on with the project. Despite Iñigo securing the wood in the same way, the CNC machine was simply too powerful; the wood flew out of the supports. Given that both of our strategies of securing the wood failed due to the sheer power of the provided tool file, we knew we had to use the default 1/4” tool. It took three hours to finish, but it was complete.

The next day, Iñigo used cardboard to surround the CNCed piece and poured 140g of silicon into it.

Once the silicon had set, Kyle removed the surrounding cardboard (and made a subsequent mess).

After Iñigo arrived, Kyle used the CAD he made to figure out that we needed approximately 25 cm³ of polyurethane. However, since we wanted to be safe, for our first trial, we mixed 30 cm³ of polyurethane, along with some green paint. As shown from our first attempt, we didn’t rubber band the two silicon halves tightly enough, allowing some of the polyurethane to spill out. Moreover, we figured out that we simply needed more polyurethane, as 30 cm³ did not fill the mold.

That being said, for our next trials, we used 40 cm³ of polyurethane, which was more than enough to completely fill the mold. And yet, we encountered another issue. It turns out the physical size of the two positives were different; blindly rubberbanding them would result in an unequal deformation of the model; the circular cross-sections would stop being circular and the sections would stop lining up.

After three more unique failures, we figured out a method of rubberbanding that resulted in adequate results. Using four rubber bands, we had two horizontal bands that were each wrapped around the silicon twice, and the block was placed on a popsicle stick such that the block was level.

Since we intentionally overfilled our molds, we needed to sand them down. In addition, our method of rubberbanding, while adequate, did not prevent some polyurethane from spilling out. That is, we needed to use an X-Acto Knife (or something equivalent) to clean up the edges of our pieces. Thus, we decided that Kyle would sand the tops and bottoms of the four pieces and that Iñigo would finish the pieces off with a clean cut.

 

 

costs

3 hours of running the Bambu FDM 3D Printer at $.20 per 24 hours (Reddit) – $0.03

PLA filament – $0.89

18” of 2×4 wood at $3.83 for 96” (The Home Depot) – $0.72

5 hours of running the Shapeoko Pro at $0.78 per hour (Carbide Community) – $3.90

1 cardboard box at $43.23 for 100 boxes (Walmart) – $0.43

10 sticks of hot glue at $43.99 for 600 sticks (Innovative Haus) – $0.73

280g of silicon at $49.99 for 6.12 lbs (Amazon) – $5.04

310 g of polyurethane at $132.76 a gallon (Amazon) – $11.30

A set of silicon measuring cups at $11.99 a set (Amazon) – $11.99

10 popsicle sticks at $10.99 for 300 sticks (Michaels) – $0.37

4 rubber bands at $11.39 for 950 bands (Office Depot) – $0.05

1 X-Acto knife at $3.94 per knife (Walmart) – $3.94

26 hours of work for $7.25 an hour (Minimum Wage) – $188.50

Total: $227