Chess Piece: King

Introduction:

For our final project, we were tasked with creating a chess piece using a mold and cast technique. In order to create our mold, we first needed to create a positive mold out of a material that can be easily milled.  Once the positive mold is finished, flexible, negative molds can be produced using Oomoo or Moldstar. Using these negative molds, the chess piece can be cast using Smooth-On 300Q. After the part has been cast, it needs to be post processed, as large parting lines are almost inevitable with our mold fidelity. This full process, along with all of our mishaps, lessons, and takeaways, is detailed below.

Selecting the Piece:

As you may have noticed, our piece is a relatively simple chess king. What makes it unique, however, is that is has a second base added on top of the traditional king’s cross. In this way, it is able to be used upside down. The two of us found this to be mildly humorous, and slightly ‘unconventional,’ and so we proceeded using this piece, discovered online.  

File Prep:

We found our chess piece file from Thingiverse, which means that it was in an “.stl” format. However, we needed to cut the file in half, make it a solid part, and reduce the number of faces. We wisely chose the orientation with which we cut the object in half, such that we were able to use one single CNC piece to create two negatives that together would produce a whole piece, which certainly saved us significant amounts of time. We used the program “MeshMixer” to make these adjustments to the file. Once all the proper adjustments were made, we then imported the part into Autodesk Fusion360. Fusion360 required some adjustments to make it a body, as outlined in the tutorial. After aligning the views to be in the proper orientation, we inserted our milling stock size into the settings. Since our full-size part was around 18 inches tall, we had to scale our part down to the proper length.  Once the part was the proper size, we had to change its location so that it would not sit at the very edge of our stock.  

In preparation for inserting milling settings, we needed to insert a bounding box around our part to outline a border. To map the path of the mill, we used the “Adaptive Clearing” option in the 3D drop down menu, and inserted the settings relevant to our setup (coolant disabled, bit size/type, et cetera). This was to outline the path for the rough cut, which uses a flat-end bit.  Next, we copied the adaptive clearing setup and named it “smoothing” to outline our process for finishing the part. The smoothing run used a rounded bit, and used step sizes that were much smaller than the rough cut. Both of these processes produced gcode that was not immediately usable by Easel. Using the process outlined on the Inventables website, we were able to successfully post process the gcode, move it into Easel, and prepare for the cutting process. Below, an image of the roughing process, shown in Fusion360, is given. Note the small rough stepdowns, and the very small fine steps. In addition, note that this particular cut did not remove the material above and below the piece. This issue was fixed in Fusion360 by using ‘selection’ rather than ‘bounding box’ in the relevant setting.

CNC Milling – Wood:

In milling our positive mold out of wood, we used an extremely small step-down size (0.1 inches) in order to put less stress on the mill.  This also allowed for a more accurate rough cut, but it took significantly longer to cut. Also, we struggled to get the file to mill the whole part, as it would leave parts of the edges uncut. To fix this, we used two bounding boxes, the larger to correspond with the rough cut, and the smaller to correspond to the smoothing cut.  This worked, and we were able to mill a mold successfully out of wood. However, the stock used was not thick enough, so the bottom of the mold was almost through the stock. The problem with this was that there was a patch of sap in the middle of the bottom of the mold, so our fingers went right through the sap layer and ruined the mold when we tried to pick it up. After this, we then proceeded to try another piece of wood using slightly different settings (namely, milling out the bounding box). Sadly, this piece was accidentally killed halfway through :(, but rather than continuing, because the quality looked perfectly acceptable, we used a thicker stock made out of wax. Below, images of this process can be seen. Note that the clearing and parallel cuts were performed on two separate days, hence the different bolt colors.

CNC Milling – Wax:

This material proved far more pleasant to work with than did the wood discussed above, given that wax has no grain and is significantly softer than even the pine that we were using. To begin, we first made all the necessary adjustments in Fusion360 (with some mistakes that were quickly remedied with no long term negative effects, and are detailed below in the lessons learned section). We then cut the rough adaptive clearing, which produced a relatively detailed copy. While we personally liked the rough, ‘stepped’ appearance of this piece, we decided to go ahead and cut the parallel cut, in order to smooth it out and appear somewhat more traditional. With a failed first attempt cutting a large gouge, we then fixed the code and completed the cut, and were rewarded with a beautiful (at least we think so) wax mold. In the future, wax is a wonderful material to use as a final piece, but we certainly learned a lot from using pine wood to make several attempted cuts, which was a certainly valuable process. Of course, the wax was not entirely perfect, and some small remnants were touched up and removed using fine sandpaper, overall smoothing out the mold. However, we made a point to only use the paper to clear edges and rough areas, ensuring that machining marks and general shapes were kept during the molding process. Below, images of the rough milling, parallel finishing, and sanding are shown.

Casting the Molds:

Luckily, this aspect of the process was relatively straightforward. With our wax mold created, we first cut the sprue out of wooden dowel, and hot glued it into place. We then coated the wax in mold release agent, mixed up some mold cast, and created the first negative. Sadly, it was only after the pour had occurred that we realized we forgot the air channel, but given that we would need a second negative anyways, we were not too perturbed. This mold was removed with no issue, at which point we cut an air channel and glued that into place as well. Once again, the mold was coated in release agent. Sadly, at this point the mold star was nearly gone, and we attempted to make it work by filling gaps in the mold with Popsicle sticks. Sadly, this was not sufficient, and so we did cast one negative mold that had various holes in it. We then switched to Oomoo, with which we were able to readily create a second negative mold that presented no issues whatsoever. It was somewhat difficult to remove, due to the fact that Oomoo had seeped under the air channel, but despite this issue we were able to get it free with minor superficial tearing that did not affect performance. We then produced a third Oomoo mold with the intention of creating a fourth in order to optimize our casting performance, but we ended up finding that we only really needed the first two. Below, the two molds used for our casting are shown, along with the first cast we made. Note the tearing on the air channel of the Oomoo mold.

Casting the Pieces:

After the negative molds were cast, we were able to start casting our pieces.  For the first few trials, we used Smooth-On 300Q, but we observed large parting lines on our finished pieces. Then, we tried to use clamps to press the mold together more, and also used a different material: 325Q. This part did not turn out well because the clamps actually caused the negative molds to deform slightly so that the finished part was more elliptical than cylindrical. Although using clamps did help decrease the size of the parting lines, our finished piece was not satisfactory (as it was not the right shape, and 325Q had a lot of visible internal bubbles in the finished part), so we continued to use 300Q and hand-hold the molds together. Though the hand holding method required more dexterity, focus, and control, and also involved some leakage, we were able to get a few decent-looking parts using this strategy, with only one single part that was badly misaligned.  

Coloring these parts proved to be difficult. We added a good amount of black ink to our 300Q mixture, and yet the part still turned out grey, and the last of the black ink had just been used. In order to get around this problem, we thought about mixing many different inks to make our own black ink, but in order to conserve resources, we deemed our grey part satisfactory for the purposes of this assignment.  Sadly, we were unable to take any ‘action’ shots of these processes, as both of our gloved sets of hands were covered in a variety of sticky chemicals. However, below can be seen the two final pieces in their ‘fresh out of the mold’ glory. Note the bubbles present on both extremes of the crosses; this was present on every single cast we created.

Post-processing:

After having cast multiple acceptable copies of the chess piece, there was still significant remaining post processing for us to complete. As can be seen in the picture above, each piece had a relatively significant parting lines, which formed small fins about one millimeter in length. In addition, each piece also had a significant sprue, which was a fairly sizable piece of plastic that also needed to be removed. Last but not least, each piece also featured a long plastic piece where the air channel (the cavity allowing air to flow out of the bottom of the mold) had been filled with plastic. Using a pair of wire cutters, each of these significant plastic pieces was removed. Once this had been completed, we then proceeded to use exacto knives to slowly shave off the protruding parting lines, which were present around the entire circumference of the piece. Some areas proved a little too tricky to access with the blade, and so were left untouched, as removing them would certainly have posed a safety hazard to our fingers and hands. Once this process was completed, the pieces that survived this process were taken to the belt sander, and the top and bottom faces of the piece were smoothed down, to reduce any visible surface aberrations and ensure that the pieces had perfect faces on which to balance. Last but not least, hand sanding was completed using 120, 220, and 400 grit sandpaper to obtain a relatively smooth finish on the ends and on some faces. Note here that we did not employ sanding on all faces of the piece, as we thought that some level of visible machining lines and the visible facets that remained from the MeshMixer face reduction process were aesthetically pleasing. These can be viewed slightly in some of the images in this write-up, as well as in the physical pieces. Sadly, at this point we had three completed chess pieces, but one of them was dropped in transit, and promptly snapped at the cross. An image of this ongoing process can be seen below, which includes the piece that was broken.

The product!

Below, the final products can be seen! Note the thinned top and bottom surfaces. In addition, we would have recast the black piece to attempt to remove the excessive bubbling, but given that this casting represented the very last of the 300Q AND the last of the black dye, we decided to leave well enough alone.

In addition, the following is an image of all of the components that were produced along the way, and helped us reach the end goal.

Lessons Learned:

Software

Throughout this process, there were many lessons learned, and we will do our best to recount all of them here. First and foremost, we both developed significant program specific expertise and learned many lessons in all of the programs learned, including MeshMixer, Fusion360, and Easel. First and foremost, we learned that written notes are extremely beneficial during the learning process. Without the notes that Jacob took during the first working class day, the many steps taken during the process would have been quickly forgotten, and being able to reference this document in the beginning of the process was certainly instrumental in aiding our learning process. In addition, during the software component of this project, we were able to employ a lesson learned from a previous homework assignment: that, sometimes, the fastest way to learn is by playing around. While instruction may have been acceptable, we both learned a ton by sitting alone with the programs, and trying all of the different buttons to see what they did. Playing around with other files in MeshMixer, we found that we were easily able to mix or remove parts of STLs, which could prove beneficial in the future if we want to create our own unique items to 3D print by combining components from thingiverse. In Fusion360, we were able to hover over all of the different settings in the parallel cut settings tab, and find different settings that allowed us to increase the fidelity of our CNCed wax block (such as the stepover setting, which we made very small.)

One very significant Fusion 360 lesson was that, if both a roughing and parallel cut have been created, changing the overall setup and stock dimensions will only change one of the cuts. We found this out the hard way when we went to cut our parallel cut on the wax block, and ended up gouging a huge slot through the side of the mold. Luckily, the soft wax gave way easily (and we realized that after switching to wax from wood, we could have easily increased our step down) and we changed our parallel cut to reflect our new stock size.

CNCing

Here, the primary lessons were learned through direct and indirect experience. First, from watching others have issues with losing zero, our first lesson was that we should ensure that our CNC files remain within the physical range of the machining device. It is clear that the machine will try to do everything we tell it to (see the inch long gouge in the side of our wax mold for evidence) but this does not mean it is able. While I have not personally confirmed this, it is my suspicion that the Carvey was unable to achieve some of the stepover and step down settings that it was told, (eg, remove 0.12 width at a step down of 0.25 inches) and, when it tried to do so, was unable to physically move the appropriate distance, and had some sort of slippage that changed zero. From observing this, we decided to reduce all of our various step values, and we in fact had absolutely no “loss of zero” errors.

Another lesson learned, is that there is a significant trade off between time and accuracy. While not necessarily an issue, many of our cuts did indeed take a long time, the final cut taking 3 hours with a 30 minute parallel smoothing cut. That said, our results are quite pleasing, and our produced wax mold met all criteria. In the future, however, we may seek out the maximum speed for test cuts, and only reduce speed on the final cut, overall saving time and effort.

Casting

There were several lessons learned here as well, most of which had to do with the overall process of casting. First and foremost, this process would certainly have been easier with registration keys. While we were in fact able to produce very solid casts without them, we did have several failed casts that had significant alignment issues (sometimes comically so). Reticence from using these keys was mostly due to unfamiliarity with CADing within the Fusion360 program, but in the future, especially if the casting process is not a one-off thing, these keys would certainly be an integral component of our molds. In the same vein, a channel to allow the air to exit the bottom of the mold would also have been nice to have as part of the CNC mold. While our hot glued dowel rod served admirably, it was not aligned perfectly (and had molding material get stuck underneath it, as can be seen in two of our negatives). Having one machined directly would eliminate this issue. In addition, placement of this channel would have been changed, as a branched channel that also attaches to the cross portion of the king would help eliminate the air bubbles that formed there in every single cast we created. As it is, it is nigh on impossible for this air to escape.

post processing

Here, there is nothing we would have done differently in terms of our post processing. However, having a tapered sprue would certainly have made removing this component more straightforward. In addition, the fact that we used a bit with a ball end made it so that the bit was unable to remove the tiny bit of material around the piece in the wax mold, guaranteeing that we would have a non-negligible parting line in our final positive casts. In the future, investigating ways to avoid this would certainly be a positive step.

Conclusions:

In conclusion, we both certainly learned a massive amount about a subject that neither of us had significant experience in. Working as partners, we were able to learn twice as fast by working together, seeking out different answers at the same time, and otherwise bouncing ideas off of one another. We certainly both feel much more confident in Fusion360, using the Carvey, and using the various casting materials and techniques that we have learned in the course. Overall, this was a wonderful synthesis of all of the subjects covered in the latter half of this course, and we both look forward to carrying these lessons forward with us, both on minor projects and perhaps on larger ones, as the case may be!