Introduction

A chess piece is something most of us are exposed to at quite a young age, perhaps caused by a clever grandparent, middle school teacher, or simply a friend. However, creating one’s own chess pieces is not as common of an occurrence. How would one even go about doing that? Where to start? Well, the OEDK is a great answer to this last question. For our Midterm project in ENGI 210, an engineering design class taught by Dr. Wettergreen, we were tasked with creating a functional chess piece which required the use of several different fabrication techniques. This blog will showcase the entire process of making a chess piece, from asking the question “Where to start?” to delivering a finished piece.

Selecting a Design

Similar to a previous 3D printing assignment, our chess piece journey started on thingiverse.com where a vast array of free chess sets are only a click away. We quickly came upon an animal themed chess set, complete with lion-shaped King, penguin pawn, and horse-shaped knight. We chose to 3D print all three aforementioned pieces using a formlabs SLA printer. 

After having created a physical representation of the pieces that we considered producing, it was easier to spot design features that may cause problems down the road when compared to figuring that out based on an online model. Such difficult features include complex geometry, protruding parts (such as ears, crowns, arms, etc), and detailed decorations. Eventually we chose to go with the horse-shaped knight, which still had a complex geometry but seemed manageable to succeed with.

Creating a two-part mold

While we already had a ready-to-go 3D printed chess piece after SLA printing, this was not an acceptable end-result because we had to make use of casting, molding, and CNC-machining, alongside 3D printing. We were required to cast our final chess piece from a two-part mold. One half of this mold had to be produced with CNC machining, while 3D printing was acceptable for the other half. So, after having selected our design, the next step was to split the 3D model in half. Because of the geometry of the knight’s ears which created an overhang, we chose to split the design in half front to back instead of side to side. This would allow for the back half to be CNC machined, while the front half could easily be 3D printed.

We initially split the part in Windows 3D Builder. However, since we ultimately imported the part into MeshMixer and Fusion360 it would have been easier to have split it in one of those softwares in retrospect. When splitting, we split a few millimeters after the plane with the ears so that the part was also split with enough thickness around the horse’s neck. We then imported the file into MeshMixer, where we decreased the number of triangles in each split half to <25,000 for easier use moving forward. We exported each half as an STL file for further processing. We imported both halves into SolidWorks to add the rectangular casting platforms for the molds, a circular base for the chess piece, lock key pegs for the mold, and a pour hole. Because about ⅔ of our horse encompassed the front-half 3D-printed part, we were able to add the entire pour hole to only one half of the mold. 

 

Below is a picture of our successfully 3D printed half. When looking at our 3D printed half, we realized that it had a cavity underneath its neck. We also realized that during the creation of a mold this cavity would be filled with silicon, which would then cause the part to get stuck in the mold. To avoid this problem, we decided to fill the cavity with playdough. This no frills solution came in handy; without it we would have had to adjust the 3D model, which would have been quite a complicated process. 

CNC-machining.

To produce the other half of the mold, we made use of Computer Numerical Control (CNC) machining. The OEDK is in possession of a prosumer level CNC machine by inventables, called the “carvey”. Before being able to use this machine, our files had to be adequately adjusted. We imported the STL file of the back-half of the horse from SolidWorks into Fusion 360. We then worked to create the g-code to successfully carve the file. We first entered the wood dimensions from which we would cut and then positioned the part within the wood block so that the top of the part was level with the top plane of the wood. We then created the pathways for both a rough cut and a smooth cut. For the rough cut, we selected the adaptive clearing tool path, which ultimately caused difficulties in the carving process. We then selected the bits for the carve. When selecting the drill bits, we measured the deepest distance that the Carvey would need to carve and then selected a bit that was at least that length for the rough cut. We ultimately selected a ⅛ in 1 flute spiral up-cut bit that was 23mm in length for the rough cut. For the smooth cut, we selected a 1/16 in fishtail downcut when creating the g-code. Since creating the g-code, we learned that it is important to use a ball bit for a smooth cut, and so we would change this in the g-code moving forward. 

After the files had been prepared, we moved on to the actual CNC machining. First, we took our selected wood piece and fastened it to the carvey. We then started the rough cut with the ⅛ in bit, with the rough cut being one of two cuts to be completed. This step was intended to take about 80 minutes, during which we watched the Carvey with great anticipation. Because of the adaptive clearing tool path that we selected, the Carvey did not carve in smooth movements but moved up and down quite frequently. About ⅔ of the way through the cut, we realized that it looked like the Carvey was unfortunately cutting the file with a few millimeters of misalignment. We paused the cut and ensured that the piece of wood was clamped down securely. It was secure, and so we determined that it was likely that the Carvey shifted during the cutting process. This may have been due to the length of our cut and the fact that the bit was moving up and down so often. After conversations with Douglas and Dr. Wettergreen, we decided to complete the carve and then proceed with the fine cut. After switching the bit to the 1/16 in ball bit, we started the fine cut, which took only 5 minutes. The smooth cut appeared to be re-aligned in the correct position, “cutting air” in the areas in which the rough cut had eliminated existing material due to the misalignment. 

We concluded that the misalignment might be due to the way in which the g-code for carving was automatically optimized with adaptive clearing. Not only did this considerably lengthen the time of carving, but it appeared as if it decreased the precision of cutting too. However, due to time constraints and the Carvey malfunctioning, we were unable to give this a try. One way or another, we accepted the results and appreciated the charm of imperfection. Moving forward, we could use the pocket clearing tool path for the g-code to reduce the time and also hopefully prevent misalignment errors. Another issue that did not impact the quality of our piece was that the Carvey attempted to cut the rectangular mold base out of the wood after completing the cutting of the piece during the rough cut. In the future, we could also modify the g-code to prevent this from happening. 

Pouring the molds

The completion of two positive halves using different techniques marked a major milestone, but it wasn’t the end of the process. To create a functioning mold, we had to transform our positive halves into negative halves so that we would actually have something to pour into. To achieve this, we used a high fidelity type of casting with high-grade silicone. We prepared our positive halves for casting by encasing them in a so-called “box of indeterminate size”. This box is easily created by glueing together oversized pieces of cardboard, starting at one edge and working your way around until the box is closed. For the 3D printed half, this went very smoothly. For the CNC’d half, we had to cut our pieces a bit more precisely, because they had to fit into the extruded cavity. We used hot glue to hold the walls together and create a leak-proof box. We then continued to mix two parts of silicone, which we mixed together and poured into the boxes. 

Two small leaks occurred, but we were able to apply more hot glue to close them. After a day’s wait, we returned to demold our silicone. It seemed like the newly created negative molds had come out very well, and we were excited to start the final step and actually cast our chess pieces. 

Casting and Post-processing

For casting our chess pieces, we had to bind the two halves of the mold together using elastic bands. To approximate the volume of our chess piece, we first poured in water. This way we determined that we would need about 30 mL of liquid plastic for each piece. After thoroughly drying the mold, we combined two parts of liquid plastic which we slowly poured into the mold. It only took half an hour for the plastic to dry completely so we were able to demold it soon. Our first piece came out pretty well. As expected, the rough texture of the CNC’d part was also reflected in the casted piece. Furthermore, the ears did not seem to have filled up with material. After several more iterations, we determined that the ears were not going to work. Changing the pouring speed, swirling around the mold, and pouring the plastic directly into the ears all seemed to have no effect. This makes us think that there is a limit to how small a hole can be for liquid plastic to enter. We decided to keep one of our casted pieces in its original form without any post processing just to show our process. But for the two pieces we decided to hand in, we used a dremel with a piece of sandpaper to smoothen out the roughness. The coloration of our pieces was achieved by adding a bit of dye to the plastic, which we carefully swirled to create a tie-dye effect. 

Conclusion

So we did it! We created a chess piece from scratch (except for the design itself). Not only did we create a chess piece, but we did it in a way that allows for mass-production, because the two-part mold is reusable until it wears. Looking back, we gained a lot of experience with different fabrication techniques and how they can work together to create something like what we made. This makes us realize how many aspects are involved in designing and then producing something. We were also prompted to think creatively about achieving the best results. For example, when we had to add putty to the neckline of our chess piece to avoid problems later on, or when we decided to sand down our final product to create a smoother finish. All in all, we learned a lot and we can now say that we have made our own chess pieces 🙂

Cost-analysis. 

• SLA resin – at $150/L and a print of about 75mL = $11.25
• Wood –  at a bottom price of $450 per m^3 of pinewood, for a piece of 14cm by 9cm by 5cm = $0.5
• Liquid silicone – at $96/L and estimating a use of 150mL per half mold = $28.8
• Liquid plastic – EasyFlow liquid plastic goes for about $48/L and estimating that we used about 30mL per piece x 5 iterations = $7.2
• Dye – at around $7 per jar and a minuscule amount we used this cost is negligible. 
• Labor – at a minimum wage of $7.25/hour and about 15hours spent on the process by both of us = $217.5 

Total = $265 (making this the most expensive project we have done so far)