As a first step in this assignment, we had to choose what chess figure we wanted to cast. We found a set of ghost-themed chess pieces on Thingiverse, and decided to cast the king piece as it was cute and we thought the crown was fun. For the first gate, we 3D printed this piece to assess scale, modifying the scale for the 3D print to be 50 mm tall. Our instructors gave feedback that we should scale the figure up to resolve the crown better, so we scaled up to have an 80 mm tall piece that we reprinted and this new version was approved by the instructors. While this scaling up did help with crown resolutions but caused us to use larger amounts of material throughout the process and will be something we discuss throughout. 

 

 

 

 

 

 

 

Next, we had to prepare the 3D printed positive for creating one half of the mold. We chose to have all complex features of the figure (eyes, crown) on the 3D print half due to the resolution of the 3D printer being finer than that of the CNC machine. We pulled up the file for our print and used MeshMixer software to split the figure in half accordingly into two halves. We then took the half with the crown and eyes and opened it in Solidworks, adding a base under the face of the figure where the cut was made in MeshMixer, and pegs and peg holes into this base. We also added a pour hole, which was the first of many times during this assignment where the asymmetry of the figure created extra challenges for preparing parts in CAD. Due to the lack of a clear center point of the bottom of the figure due to the asymmetrical shape of the legs, an approximate center point had to be chosen for the pour hole. This approximation was sufficient for this stage of the design, so we 3D printed the part and proceeded

 

 

 

 

 

 

In class, we created a negative mold by pouring a silicone mixture into a box we constructed using cardboard and hot glue to securely encase the 3D-printed positive. First, we carefully measured and cut the cardboard to form a box large enough to accommodate the positive model with adequate space around it for the silicone to flow and form a sturdy mold. Using hot glue, we sealed the edges of the cardboard box, ensuring there were no gaps or leaks where the liquid silicone could escape. The 3D-printed positive was then placed inside the box, positioned strategically to allow even coverage by the silicone. After mixing the silicone with the curing agent, we slowly poured it into the box, allowing it to flow around the positive while ensuring all surfaces and intricate details were fully covered. Care was taken to eliminate air bubbles by gently tapping the box or using a thin tool to release trapped air. Once the silicone was poured and evenly distributed, we left it to cure and solidify (24 hours).

 

Next, we had to prepare the CNC machined positive of the other half of the figure. Immediately after opening the other MeshMixer-prepared half of the figure in Solidworks, we identified a problem. For creating the base of the 3D printed positive half, we had used a point visually in the center of the figure where all the lines converged as the point from which a center rectangle for the base was made. However, in the file for the CNC machine half, the point where all the lines converged was visually not visually centered and certainly did not align with the 3D print half, which is a necessary feature for the mold negatives to align perfectly.

In other words, we needed to find a way to make sure the platform for the CNC machined half was centered around the point on the CNC machined figure half that would align almost exactly with the centerpoint of the 3D printed figure half. We spoke with Dr. Wettergreen and the TAs, and came up with a solution eventually. We utilized the assembly feature in Solidworks to align the CNC machine figure half as exactly as we visually could over the figure part of the full 3D print positive file. In a separate file, we rebuilt the platform for the CNC machined positive, which we then imported and aligned exactly over the platform part of the 3D print positive within the assembly. From here, we deleted the 3D print positive from the assembly, and we were left with the CNC ghost half centered almost exactly on the platform in a way that would align almost perfectly with the 3D print half. When designing the pour hole aspect of the platform for the CNC half, we also had to account for the asymmetry previously mentioned when adding the pour hole to the 3D print half. To make sure the pour hole halves aligned, the distance from the platform edges were used to position the pour hole rather than any reference point on the ghost figure half itself. 

After troubleshooting through the preparation of gcode in the VCarve software, an effort that including having to reverse the positions of the peg holes with the pegs in the solidworks file due to an initial orienting error, we were finally able to load the gcode for CNC machining. After the first pass, the CNC machine appeared to be cutting correctly but after a few more, it became apparent something was wrong with how the machine was cutting in the z axis as very little depth was being cut out and most passes did not cut any depth out at all. Additionally, we could tell the piece was being cut at a slight angle as was seen due to uneven cutting depths at different x-y positions that should have been identical, so the surface was not always flat where it should have been. Interestingly, the x-y positions of the cut appeared more or less perfect, but the piece overall looked very strange due to the cut depth problems.

The angled effect we observed was easily attributed to the file being manually made to appear flat in vcarve using the orient tool, as the imported file initially was at a weird angle (not lying on any axis). While the cuts looked flat visually while prepping in vcarve so we thought the orientation was alright, it is easy to imagine it was not actually perfectly flat which could lead to the observed angling problem. However, the problem with the dimensioning in the z axis being radically off while being preserved in the x-y axis was much harder to identify. After much discussion with an OEDK worker and amongst ourselves, we realized it could maybe be a problem with unit conversions across different softwares. We had prepared our file using millimeters in solidworks, and then converted all given parameters to millimeters in vcarve accordingly. However, when the gcode file was uploaded into carbide motion, all the units were converted to inches. While this explanation could not account for the preservation of x-y scaling, we decided to try addressing this inconsistency as we and also the OEDK worker could not come up with any other explanation. In solidworks, we first fixed the angling of the piece by using the assembly tool to align the base of the cnc positive to a identically dimensioned rectangle 0.001 inch thick which had been constructed on the top plane, therefore aligning the whole piece to the top plane. We then adjusted the units in the piece to inches, and re-exported it. We prepared the file in vcarve again, and confirmed that the print should work with the OEDK worker who had graciously helped us out throughout the morning – although admittedly neither of us were sure whether this new format would work either as we both had also thought the first cut would work and were not sure what the underlying problem was.

With the file ready we imported it to carbide and turned on the CNC machine. We first set up our block for cutting by making indentations on the side so that we could use fasteners to hold it firmly in place throughout the cut. We then calibrated the machine by taping the calibrating to the bottom left corner and manually moving the drill bit into its designated hole. After calibration we were able to cut. This time it cut much better and at the proper depth. We first did a roughing pass with a ¼” drill bit which worked great. After that was completed we switched to a 1/16” ball point drill bit to finish the piece. It was working very well until it had to finish the edges. Because we had walls that were taller than the bit the chuck hit the wall at one point and so we paused the cut and sanded down our walls without moving it from its fastened position until it would no longer interfere with the chuck. After this tedious sanding, we unpaused the cut and it was able to finish as desired.

After cleaning up our mess at the CNC machine we went and made a box of indeterminate size around our wooden CNC’d positive. To do so we used hot glue and duct tape similar to fasten cardboard walls around the mold, similar to how we did to make the first negative from the 3D print positive. Once the walls were set we mixed 200 mL of part A and part B to create the silicone mixture which we then slowly poured over the positive. We checked for leaks and left it for a day to cure. The next day, we removed it and now had two negative halves that align via the pegs we added to the base. At this stage we were ready to pour the final polyurethane casting.

With our silicone molds prepared, we went on to the final step required: casting polyurethane into the molds. For this, we mixed part A and B of the polyurethane solution together in silicone cups that we used to pour into the mold. More specifically, we first used heavy duty rubber bands to hold our silicone molds in place, to align the negatives, and to reduce leakage. After this was done, we added part A and part B to silicon cups in equal parts, separately, until we were ready to mix them. We then quickly mixed the two into a third cup before pouring it into our pour hole. 

With our first object, we assumed the volume of the mold, which proved to be the wrong decision because we needed much more polyurethane to fill our mold. Our first pour was about 60 mL, but when this proved to obviously not be enough, we added 40 more mL to reach our ghosts’ volume of ~98 mL. So, for the next ghost, we resolved to mix the 100 mL altogether at the beginning. However, we were using small silicone cups that went up to 100mL; therefore, the top part of the solution was not being mixed properly. Because we did not make the connection at the time, we tried for a few more times, but we gave up after our resulting ghosts came out with everything cured but the head.

 

 

 

 

 

 

The next time we came in, we used a bigger silicone cup so we could make sure to mix the solution thoroughly. After this change, the rest of our chess piece came out nicely, with everything cured as it was supposed to. Throughout the casting process, we encountered various issues related to our design that confused us for a while until we were able to figure out the issue. The first issue was the amount of volume we needed. Since most other groups were using 100 mL cups because their object required relatively less volume, making it easier to pour when the cup was full. In addition to this, our pour hole was small making it hard to pour the solution without spilling. When we simply used a bigger mixing cup, these problems were solved.

At the end, our first piece was white, while the other 3 were white, pink, and blue. Because the two sides of the mold weren’t perfectly aligned, we post processed with sandpaper to smooth out the ghost and remove misaligned edges. We were then left with a cute ghost family of 4, each with its own quirks and a story behind it 🙂

 

 

 

 

 

Cost Model

Cost Type	Cost	Price	Source	Quantity	Total
Materials	Silicone mold making material (A&B) – (2 x 19oz)	$20	Amazon	(~900 mL)	$17
	Casting material – Polyurethane (A&B) – (2×30 oz)	$41	Amazon	800 mL	$19
	4×6 wood	$24/10 feet	Home Depot	6 inches	$1.20
	Sandpaper assortment	$13/25	Amazon	3 sheets	$1.56
	Heavy Duty Rubber Bands	$6.50/220 bands	Uline	4	Negligible
Labor	3D printer operator	$18.85/hr	ZipRecruiter	1 hour
	Solidworks prototyping engineer	$34	ZipRecruiter	4 hours
	CNC Operator	$22/hour	ZipRecruiter	1 hour
Overhead	Facility Cost (Machine Time) – CNC	$4.85/hour	Rockler.com	4 hours
	Facility Cost (Machine Time) – 3d print	$0.10/hour	slashplan.com	2 hour
Total					$234.60