For our gear train project, we originally had several different ideas. We researched some different types of gear on Google and YouTube, and after initially discussing elliptical gears and various other shapes we initially settled on some triangular planetary gears. It also had features that would allow the gears to stay together even if we lifted the gear-contraption horizontally. This was possible if each gear was composed of basically 3 parts; the inner planetary male gears would have a layer of teeth, a protruding, smooth inner layer, and another layer of teeth. The outer gear and the central gears were both female types with a protruding layer of teeth, a smooth middle layer, and another protruding layer of teeth. The alternative was switching l that layout to have the protruding part on both the central and outside gears, but we eventually realized that that would have rendered it impossible to assemble the design.
After attempting to draw sketches, experimenting with SolidWorks, and looking up the math behind epitrochoids and hypotrochoids, and attempting to model the path with python, we decided that designing triangular planetary gears would be too much work and would be beyond the purpose of this assignment. We briefly discussed the idea of switching back to a slightly more simple shape like an ellipse or a shape of constant diameter but this presented the same problems as the epitrochoid.
As a result, we chose to stick with normal circular planetary gears, and focus instead on its ability to stay connected without some sort of frame.
We used the solidworks toolbox to generate the initial gear layout, and went through several design iterations. Initially our design included four planet gears and one large central gear. However, upon receiving the machinable wax blocks we realized that our center gear was too big to be machined out of the stock we had available to us. We resized the gears to be smaller, which included changing the design to have only three planet gears instead of four so that the tooth ratio would keep the gear train symmetrical.
Because we had male and female gears made of three different layers, we couldn’t make just one mold for each gear. Instead, we had to either do a split-cast mold or simply make half of each gear and combine them afterwards. Our initial plan was to do split-cast, but Dr. Wettergreen advised us against it, so we ended up making CNC molds for half of the planetary gears and half of the central gear. Because the outer gear was too large to CNC, we planned to laser cut that out from acrylic in three layers.
To generate the g code for the CNC machine we used Fusion 360 (an Autodesk product that is free to students). Female molds were created in SolidWorks using the cavity tool and the planet/center gear part files. Those part files were then imported into Fusion 360 and toolpaths were created to cut the part. After a couple of hours of playing with the post-processor settings, we were able to get the g code to open in Flashcut CNC 4. However, upon trying to load the g code onto the actual CNC machine we discovered that the CNC was incorrectly scaling the g code file. We were able to find how to scale the g code in Flashcut, and after several test cuts found and changed the scaling factor to the correct value (0.790 for all axes). Finally (after more toying with the post-processing of the toolpath) we were able to load the g code into Flashcut CNC 4 on the actually CNC machine and cut our parts.
After getting our CNC molds, we used Smooth-Cast 300Q to make our gears.
Because we needed to make 6 halves of the planetary gears (2 halves for each of our 3 planetary gears), we wanted to be able to reuse our mold rather than cutting the same mold in 6 different wax blocks. To do this, we drilled two holes in the bottom of our mold, filled it with Play-Doh, then sprayed the Mold Release inside the mold.
This way, when we poured the 300Q in the mold, we would be able to use a paintbrush to poke the Play-Doh filled holes out from the mold, and the Mold Release would also help it slide out more smoothly. Ultimately, it became clear that only one hole was necessary and the mold-release was barely necessary. In fact, the mold-release may not have been needed at all but we did not think it was prudent to risk not using it. The first few pours of the 300Q did not go perfectly and we encountered some issues achieving smooth surfaces. (Gears 1 and 2) There was also one instance where the Play-Doh application was imperfect and the 300Q leaked out. Ultimately we found the right amount of 300Q to achieve a good, smooth finish (about 28 mL total).
Our next challenge was to find a way to align the two halves of the gear perfectly. Initially we tried using the small nubs left from the Play-Doh holes, even using wooden dowels cast into the gears, but this proved to be a lapse in spatial understanding and did not help. (Gears 3 and 6)
We decided the best way to do it was simply with smooth surfaces. First, we made one half of the gear and removed it. Then, when we were making the second half of the gear, we tried to hold the already-made half over the liquid so that they would set together.
However, this meant lining up the two halves of the gear was imprecise and it also meant that one plastic gear was sitting on a layer of liquid, and it wouldn’t mold perfectly level. To fix this, we created a small notch in the wax mold as a sort of marking tab. With this, our gear would have one extra piece of plastic sticking out, but we would be able to line up this small bit with the two halves when we were putting them together. If we were going to put two halves together though, then we needed each half to have a perfectly flat surface. Turns out, if we leave the 300Q to just rest and solidify then the surface is actually a little concave. To solve this for the planetary gears, we borrowed the idea of using a thin piece of plexiglass to rest on the surface of the plastic. The plexiglass would cause the plastic to overflow a little, but then we could peel off the extra plastic and end up with a nice smooth surface. To keep the two halves together, we mixed a small bit of 300Q to spread in between the two halves, held them together, and just waited for the plastic to settle and bind them together.Afterwards, we could easily cut off the bit of plastic on the edge with an Exacto-knife. With this methodology, we made the 3 planetary gears and the central gear. Some of the initial combinations were not perfectly flush (Gears 4 and 5) but we eventually succeeded (Final Gears)
We encountered the opposite problem with the central gear. The faces that had to be bonded were instead those from the bottom of the mold and they were convex. Additionally, the notches on the top of the mold did not help in aligning them, though in retrospect we could have cut notches at the bottom. As it was we aligned them by eye. The convex faces proved problematic because they made it difficult to align the two sides to be flush. Additionally there appeared to be a small gap at the edges. Initially we solved this by sanding down the insides of the inner gear to be flat. This made the gear look nice but it lowered the clearance in the gap so the planetary gears didn’t turn nicely in them. (Gear 7) We reverted to the initial strategy of being extremely careful alignment by hand/eye. In retrospect, this also could perhaps have been solved more elegantly by sanding down the faces and then making the planetary gears notch piece thinner as well, either by sanding, modifying the mold, or pouring less 300Q into the mold when forming the individual pieces. We also tried drilling holes into one half gear to explore the idea of using a central axle. (Gear 8)
For the outer gear, we first prototyped whole gear set using laser cut wood. From this, we found out that there was too much space between the gears and it allowed them to slip between the teeth. We decided that the easiest way to fix this was to shrink the outer gear, as opposed to increasing the size of the central and planetary gears and possibly messing up the teeth ratio. First we cut out of cardboard but experienced some difficulties (first the power was too low and then it was too high. In unrelated news, we worked with Thor to develop a fire-response protocol for the laser cutter).
We learned that with a powerful laser cutter that has been calibrated to wood and with an abundance of wood, there is not any good reason to use cardboard. We switched to wood. After several wooden prototypes, we finally found the right size of the outer gear and laser cut each of the three layers out of acrylic. On the top layer of acrylic, we were supposed to engrave a quote from Galileo, but we accidentally just cut the outline of the words into it. Because the power was much lower than necessary, the letters didn’t cut all the way through so it still looked alright.
For post-processing, we sanded down the faces of each gear to give them all a uniform matte finish and remove blemishes. Later on, this also provided a surface to which paint could bond. We did the same on the central gear. Additionally, the seams on the gears were not flush so they had to be sanded as well. We used 120 grit sandpaper for all the surfaces. To sand the seam on the inner gear we had to put it in a vice and fold an extremely thin piece of sandpaper to run over it which was time-consuming but effective. We also used the drill press to add a hole to the center of the planetary gear through which we epoxyed an axle. Unfortunately the drill press wobbled a bit and our alignment may not have been perfect so the hole is slightly off-center. Doing it again we would add the hole to the negative mold, and even consider adding axles to all the gears, though it makes assembly far more difficult. The epoxy require more sanding but eventually came out well. This all left the assembly in pretty good shape.
At this stage, the outer assembly consisted of the upper gear face, the container ring, and the lower gear face all stacked and held together with three bolts, wing nuts, and washers. The decision to use wing nuts was temporary, but proved effective for prototyping because it made assembly and disassembly incredibly easy. Washers also proved useful in separating the gear layers and decreasing friction on the main gear train. However, the final prototype will have the wing nuts replaced with nuts for a more finished appearance.
We decided that the entire outer gear would look much better with some post-processing so we sanded more finely, with 120 grit sandpaper. We began by using a sanding block on all surfaces and edges of each acrylic outer gear, then going over the sanded surface with 120 grit sandpaper to smooth out the frosted appearance. Because this began to cover up the etched quote, we painted over the letters with black model paint, then sanded down the paint layer. This kept the uniform frost, but let the paint drain into the crevices made by each engraved letter, making them completely black. At this time, we discovered that the laser cutter had almost completely cut through the acrylic layer and paint had begun to leak, so we set aside the upper acrylic gear until the rest of the design was complete. We then moved on to painting each molded gear in different colors. The sanded surfaces accepted the paint nicely but the edges largely didn’t, so we ended up scraping, sanding, and dissolving the paint of off the edges.
We experimented with files, sandpaper, paint thinner, and utility knives/scalpels and ended up using some combination thereof to remove the paint residue. This proved extremely time consuming.
There is some paint left on the gear workings that we expect to come off with use but for now it looks good. We could consider sanding the gear surfaces very finely to smooth out the paint but their current state is technically more planetary so we chose to use it.
