Season 4 Episode 10: “Finale – The Rook” : Caz and Andre team up to mass produce rooks on a scale never before seen by ENGI 210. Will they be able to overcome their biggest challenge yet?

For real, this is the last project of ENGI 210, and it is a long one. Andre and I (henceforth referred to as “We”) are given the challenge of molding and casting a chess piece from a CNC positive. The toolchain is vast and looks a little something like:

Solidworks > Fusion 360 > Easel > Carvey CNC >      Hand Postprocessing > Rebound Molding > Smooth-On Casting > Hand Finishing

I took lead over the first half of this toolchain so I’ll be detailing it in this blog. Head on over to Andre’s blog to see the details of the molding and casting process.

Initial CAD Creation

We spent a solid amount of time looking through Thingiverse and other sites for suitable rook stl files to CNC and mold. Unfortunately, most were either too complex or too simple, and all had a glaring flaw: the overhang. Since the Carvey is a 3-axis mill, it can only cut from the top down, not from the side. That means that any wood must be completely supported from the bottom with no overhangs. The top of a rook includes crenelations that jut out from the top, making the standard piece impossible to CNC. I took to Solidworks to create my own alternate rook with a simple + on top rather than the standard o. This is able to be cut very well with the Carvey. I cut the rook in half and exported it as an stl file to send to Fusion360.

G-Code Creation

Since our rook was low-poly enough, we could send it straight to Fusion 360 with no issues. We did the standard steps for creating a CAM model, including converting to B-Rep, creating a bounding box, and setting an Adaptive Clear and Parallel Cut. Slightly more difficult steps include setting the stock size by measuring our wooden stock piece, aligning the chess piece in the stock to not interfere with any clamps, and setting feed speeds based on Carvey recommendations and wild guesses. The hardest part of this initial setup was getting the tool settings correct. There is no preset tool, so we had to measure the bits of the Carvey and input those into a custom tool setting in Fusion 360. Once we had a proper setup, we exported the G-code and opened up in Easel.

Attempt 1: Catastrophic Failure

First attempt on the Carvey seemed to be working fine throughout the first plunge depth. We saw a vague chess-like structure emerge, and about halfway through, we paused, opened up the Carvey, and vacuumed out all the wood dust. When we resumed, all hell broke loose. The Carvey somehow became miszeroed, and it started chewing its way straight through our piece. We stopped it in a panic then pulled the piece out, admiring how ruined it had become in only a few short seconds.

 Attempt 2: Just a Little Failure

On the next trial, we decided to try and not open the Carvey to protect against unknown breaking of the Carvey. We also lowered the plunge depth a bit to try and mitigate slipping of the bit against its calibration. Unfortunately, we did not lower this value enough. The Carvey again became misaligned halfway through our cut, even though we didn’t pause it at all. The timing belt could have slipped or had some other mechanical error, but the end result was that our cut just did not work. As a bonus, I attempted a finish cut on this piece, and the finish cut deep into the wood on the first pass. Stopped that one right away and proceeded to the next trial.

Attempt 3: Pretty Great

For the next Adaptive Clearing Cut, we lowered the plunge depth even further, to a measly 0.125″. This means that the cut takes far longer and does many more small passes, but there is less chance of the Carvey outright breaking, which is always good. This cut turned out surprisingly well. The edges were clean, the shape was correct, and there were absolutely no giant holes from the Carvey deciding to tear its way through our piece randomly. We decided to pull this piece out, save it in case of further catastrophe, and move on.

Attempt 4: Fine Cut Try

For our last CNC cut, we tried an adaptive cut with two separate clearing passes, one parallel cut for the flat edges, and one ramp cut for the steep areas. The adaptive clearing went off without a hitch, and the finishing cuts went … ok. They cut a little deeper than expected, which is strange since the Carvey zeros from the bottom tip of the bit. The finish also wasn’t quite as smooth as expected, maybe due to us using a 1/8″ ball end mill instead of a smaller one and having passes slightly too far apart. In the end, we decided to make molds of both Attempt 3 and Attempt 4, since both had similar surface and shape quality.

Mold Preparations

Although we had half rooks cut into our piece of wood, we didn’t have any of the other stuff that makes a good mold, like pour holes, air holes, or registration keys. We added these into our cut wood pieces manually using scrap pieces. We cut a dowel in half, then glued the semicircle into our piece to create a pour hole. We took an acrylic strip and used it to connect the bottom of the piece to the top near the pour hole. This became the air release. We also used dowels to attempt registration keys, although we made a critical error in spatial visualization which led to these keys being only partially useful. Find out this error on part 2 in Andre’s blog (Cliffhanger!!)