This final project was done by both Grant Anderson and Samuel Ramirez.

For our final in ENGI 210, we are to combine our skills in CAD, 3D printing, CNC machining, and molding to create a two mold chess piece. We chose to use this Dragon Chess Set as our basis and in split the file in SolidWorks so we could work with only the queen. We decided on the Queen because we found the piece to be not only cool looking, but also possible to CNC and mold without too many modifications to the original. The King for example can’t be done without removing most of his spikes and it will make his straight back look awkward. The Queen has a nice curvature and only a few back spikes, so she was better suited in that regard.

Once we isolated our Queen,we had to smooth out her back in Fusion 360 so that it would be possible to CNC it in wood. We simply a combination of the smooth tool, selecting an area of polygons on the dragon’s back, and the reduce polygons tool, which allowed us to decrease the overall complexity of the geometry in that area. The pedestal needed the same treatment on the backside, so we did our best to round it out and make it still look like a pedestal. Afterwards, we 3d printed the smoothed out chess piece to see how it would look. It appears to have printed well!With the smoothing completed, we attempted to split the piece in half using Meshmixer, but found that the piece was very uneven if we isolated the back from the front. We wanted to try and make the CNC mold handle the back half of the mold, but cutting the piece straight down was not to go allow that. Thus, we moved the piece into Fusion 360 and cleaved the piece at an angle. After converting them into their own bodies, we added a base for each of the negatives along with peg holes to help align the two sides together when molding. Unfortunately, we discovered too late that Fusion 360 and SolidWorks don’t scale the same way and we had made both our pieces about 10 times as large as they should have been. We weren’t careful about scaling and he had to rescale and proceed work from after we smoothed the dragon only because the scaling wouldn’t work well with bases if we did so.

For the most part, we followed the instructions given to us by our professor, Dr. Wettergreen, to give us our prepped CNC part. The step by step guide was thorough but there were still some challenges. After splitting the piece the big frustration was how to make a base that would line up correctly. Fusion360 has some build in analysis tools but when the piece you are centering doesn’t have a smooth face or easy center there is no thing to measure off of. We decided the best route was to turn the grid on in the highest detail possible then try to eyeball the piece around the origin. This was somewhat successful but later led to allignment problems. Almost every team had alignment problems and if I had to hazard a guess this is the step from which they arose. After the pieces were centered it was relatively easy to extrude a rectangular back and add alignment holes. The piece rose at a weird angle since we had performed an angled cut but this was not overly concerning since the 3d printer could handle these angles.

Now that the pieces were sketched in Fusion it was time to use Fusion’s built in CAM software to prepare the gcode. First we defined a relative stock that was slightly larger then out piece. We didn’t need to make it exact in the software since it did not matter as long as it cut the piece out on the block of wood provided. The Z height did matter since the machine needs to know how thick the wood is. Next we defined our operations. The 3D adaptive clearing function made things very easy. For this operation we used a ⅛ inch ball nose mill since ball mills are better for curved surfaces. We uploaded the tool library for Inventables products and they came with all the recommended presets built in such as spindle speed, coolant, z-plunge, etc. One thing We left on was the full z-retract between cuts. This was probably very inefficient but we did not trust the software not to run into the piece while traversing. This may have been overly cautious and it did end up adding at least 30 minutes to the cut. We also ran a smoothing pass with a 1/16 inch fishtail bit. For the smoothing we ran parallel sweeps in a cross hatch pattern to try and get a smooth surface. Once the operations were done we followed the instructions for adding Easel as a post processing option to Fusion 360 and post processed the adaptive clearing and smoothing as two separate runs on the machine so we could change the bit in between.Once we had our prepped part, we moved over to Easel and the Carvey in order to get our first mold. The second mold was being 3D printed at the same time and came out pretty cleanly. We did forget to take into account the possibility of air bubbles with the front side of the mold, so we would have added a little air tube for the horns and the mouth if we had remembered. As for the CNC mold, we first attempted to carve it using 3/4″ wood and found that our part was too deep for the wood. We attempted to remedy the cuts that were too deep by getting a slab of clay and molding it around the cuts. We dug into the peg holes and smoothed out the extra clay in the wooden mold. We later found out that this little fix would come to hinder us later, due to the clay retaining moisture and preventing the silicone from curing properly. We continued on nonetheless, and sprayed Polycrylic on the wooden mold in order to hopefully prevent the above from happening.

Once both molds were created, we poured a 1:1 ratio of the silicon mixes into our molds. Unfortunately, we forgot that we needed a block base for each of the molds so that they can easily be handled and manipulated later on. Thankfully, we remedied this quickly using a technique shown to us at the beginning of the semester to create a box of indeterminate size around each mold using cardboard. These boxes would serve to keep the silicon from leaking out of the mold and would create a silicon block that the mold would sit on. We taped the cardboard all around its edges to try and seal them and proceeded to pour more of our silicon mix on top of our earlier mix. These molded together, so no harm was done.After the cure time of about 7 hours, we went in to check on the two molds. We found that the mold based on the 3d print was completed, but that the mold sitting on the CNC part was still rather goopy, so we gave the silicon a full 24 hours before checking on it again. Unfortunately, we found that the mold was still goopy despite the extra time. Regardless, we still separated the silicones from the molds, and found that the clay mold was sticky, due to the clay retaining the moisture and preventing it from curing properly. Our next step was to allow the semi-cured silicone to air dry on its own before continuing.

Once we got the silicone as good as it could be it was time to make some dragon pieces. It took about 1.5 fluid oz. of easy flow to fill the mold. To start off, we rubber banded the two pieces of silicone together and poured the easy flow in. About 20 minutes later we demolded and beheld a monstrosity. The nose of the dragon had not come out since an air bubble had been unable to escape and the two halves were at least a quarter of an inch misaligned. We tried one more and got the same result. After talking to Nick Lester we realized that the nose needed an airhole for the air to escape from. We took one of the clay working tools and stabbed a hole in the mold that worked just fine. We also tried using sharpie lines to figure out how to align our silicone. This was a little bit imperfect but it got us closer than we had been. At this point, Dr. Wettergreen told us that we were not rubber banding the mold correctly. Instead of directly applying rubber bands to the silicone and deforming it we should be putting the silicone between two pieces of wood and rubber banding the wood. This helped minimize the deformities.We had attempted some alignments earlier, but found that the pegs weren’t helping alignment whatsoever. Our first pieces were completely off, with their backs about 4mm lower than they should have been. After our first pieces, we cut off the pegs and then marked off using sharpie how our molds were attached together by drawing lines across the molds to indicate alignment. We then measured the amount of misalignment on each of the failed chess pieces and adjusted the alignment distance each time and drawing a new sharpie line with each test. We eventually got two pieces as closely aligned as possible, a lavender and pink set of dragons, and used some diagonal cutters to chip away at the funnel of silicone and prepare for post processing.

Once we had some properly aligned dragons it was time to post process. This mostly looked like sanding out minor imperfections left over from alignment. The easyflow sands were well so we started with 120 grit sandpaper then moved to cleaning up with 220 and 440. This made a mostly smooth finish but uncovered some small air bubbles in the easyflow. This may have been due to over vigorous mixing but the world may never know.

Cost Analysis

Materials

• Easy Flow ($123/2gallons) * 1.5 oz.*0.0117188 gal/oz = $1.08
• Platsil 73-25 Silicone Rubber $250/16 lb * (0.3125 lb) = $13.02
• 3d printing filament = $1.48 according to 3dPrinterOS
• Scrap piece of 2”x4”x12” wood = $0.35
• Rubber bands = $0.40
• Labor
• 21 man hours * $7.25 = $152.25

Total: $168.23

 

Note: The cost to make additional pieces is approximately $4 dollars assuming labor and materials.