Note: This project started as a team project between myself and Corbin Cowden. Due to illness, this project is large part a solo off-shoot.
Slide Deck: Here
(Link sharing should only work for Rice accounts).
Video:
Inspiration:
The ideation of this project was largely a product of an idea for a final aesthetic. Between Corbin’s knowledge of chainmail and my own experience with wire-wrap jewelry, we wanted to do something that could employ those skills for aesthetic pieces. The natural conclusion was a knight in armor, though a horse’s gallop or the swing of a weapon were the most obvious ways to actually employ mechanical movement.
Moving to a solo project required simplifying this idea significantly, and with not being able to showcase Corbin’s specific interests, I choose to just focus on the movement of a horse.
Example:
Mechanism:
The original scope of this design required incorporating additional movements than were ultimately feasible. In addition to the cyclic leg movement (through movement 22), a secondary set of gears would be used to ‘sway’ or ‘rock’ the body (movement 124). The motion created by adding joints to the legs didn’t have a clear parallel (though 145 comes close, with the rotating gears causing the translational motion of the knees).
Early Formulation
Initially, most work was done in two parts: mechanical pieces on solid works and aesthetic pieces on Illustrator.
On Solidworks, through motion studies using concentric mates and distance mates to hold everything together, the leg motion looked promising. That is, as long as the body is properly distanced from the gears (body-bottom of gear is the length of a straightened leg), the knees move in such a way that the legs never bend ‘backwards’. That is, the legs bend in a consistent way.
A second Solidworks model was made to show the physical spacing and fit of the secondary swaying motion but ultimately was too complex to properly model through Solidworks animation due to the limitation of mates as well as the performance limitations of OEDK computers.
On the illustrator side of things, a horse logo was found on The Noun Project to use as the base for the aesthetic parts. From here, points were designated as the joints. This wasn’t trivial as mechanics (placing hip/knee joints on the same horizontal line) were at a tradeoff with aesthetics/realism (where a horse’s joints are actually located). Joints were marked with 6 mm holes as that was the diameter of the desired dowels.
First Prototypes
As recommended, the first actual prototyping was done by laser-cutting cardboard and getting a sense of how viable the solution is.
First, the base leg motion was cut out with cardboard gears. Though the motion was surprisingly good for cardboard, there were obvious limitations in that both bars holding the gears in place needed to be held down.
A second cardboard test was done to model the more complex system. Though this was done with representational wheels instead of gears, it did prove some valuable results for this design. Notably, the center slot that constrained the horse body to only moving up and down was too short and the inside pair of gears was too large, causing nearly uncontrollable swaying of the body. The gear ratio was changed and plans were made to re-design the center slot to be a pillar spanning the center of the entire device.
Note: This was around the point where I started working solo. This required the more complex design to be disregarded as unfeasible for a 1-person team.
Intermediate Prototypes:
Moving forward, I needed to start moving more rapidly through prototypes laser cut out of wood. The first was to test the viability of the current gears with wood: assessing the chosen pitch angles, if added weight or friction was problematic, etc.
Although this did work quite well, there were some obvious issues. The first was still needing to hold the bars still in order to for the legs to move properly. Second was the more pressing concern of the spacing/stacking of each piece. This seemed to be less of an issue with cardboard since those pieces were more flexible but couldn’t be ignored now, hence the supporting state in the figure.
At the TA’s recommendation, I immediately move to properly raise this prototype to something more vertical.
Two supplemental laser cuts were performed for this, fitting them to the gears and legs I had already made. This was ultimately still quite low-fidelity as hot glue was pretty much the only adhesive used.
- The stage itself, using a series of teeth adapted from a maker-case design similar to the laser-cut box assignment.
- A set of three-dozen wooden ‘spacers’. These are simply rings with a 6 mm inside diameter to hold snugly against the used dowels and a 10 mm outer diameter. These were cut out in a dense grid and were used in all prototyping from here on out, including the final prototype. The main purpose for this was to space pieces out in using a simplest unit of a ‘width’ of the available wood.
This prototype had some issues.
- The first major issue were the gears not staying aligned, generally due to their thinness relative to their size.
- The second main issue was the knee’s contacting the gear as they straightened out as it was the only place were a spacer wasn’t included for safety.
Final Prototyping:
From evaluating the first upright prototype, a list of improvements were made going into what would have to be the final prototype due to remaining time. Each will be addressed separately
- Incorporate actual horse-shaped pieces
This step was actually one of the most challenging of the final prototype. After choosing a final size of the gears (scaled to about 2/3 size and adding 30 degree connection angles to keep length in line with horse body), a maser illustrator file was used to shape and size each piece of the horse onto the gear placements.
This required a lot of the Pythagorean theorem and a small amount of trig to ensure proper spacing and equal leg lengths
. - Double the thickness of the gears
This step was largely trivial but does represent a major misstep in the final part. To me the immediate path was to just cut out each gear twice. During later assembly, I realize a huge missed oppurtunity which was to only have the center axle run through one layer and the ‘ankle’ joint run through the other layer. This would have removed significant amounts of post processing as the dowels would already have been hidden and would improve the final appearance. - Improve the fidelity of adhesive
This step was the simplest and really just required more planning ahead. Since each dowel was generally only adhered to one piece of a leg or a gear, those pairings could be wood glued early on. - Improve spacing to reduce gear wiggling
Spacing was the most radical improvement. After cutting out the final pieces, I made a mock up of the final model in order to understand how many “wood-widths” were required at each joint/dowel. I used these as a basis for initial lengths, leaving a few millimeters on each side to leave room for sanding.
Once these pieces were made, I cut down to final lengths using my existing parts as a jig, stacking gears and washers to the required height for each dowel to draw on the length.10 mm caps were added to the end of each dowel, ultimately I kept these un-stained as they provided a more toy-like appearance to me, and a fake sense that a second type of wood was used.
- Add Crank
The crank was kept simple, with a few wooden rings of increasing width to provide a decent sort of grip, like a doorknob. To match the exposed dowels on the gears, I glued the largest ring flush to the dowel, which came out looking the best out of these exposed endpoints.
- Add vinyl decal
My vinyl detailing was inspired by a few things. Immediately, I didn’t want to do anything like facial features as having vinyl at just once point would look a bit tacked on. Hooves were considered but since I needed to have a wooden cap to prevent the leg piece from sliding, they would be obscured. Talking with Matthew and Anas and looking at the really nice detailing on their Pokémon-based design, I realized that I was a single fire decal away from a classic Pokémon: Rapidash (also similar to a nightmare or hellsteed among other mythical creatures). This also had the benefit of being a major part of the design.
The fire effect was created by isolating the tail and head from my main body piece and repeatedly off-setting the path. This process was non-trivial as very complex/curved paths had some odd behavior when shrunken but its not interesting to discuss. It was just surprisingly hard.
From here, the selection of available vinyl was actually a major concern. The immediate goal of red – orange – yellow was unavailable as there was a single sheet of red vinyl with very little continuous area.The next path was a blue-based flame. There wasn’t a lot of ideal white vinyl so I chose purple – blue – teal. This looked slightly odd at first but it grew on me. Perhaps it has a butane flame. Maybe its just a really punk horse. Either way I’m pleased with it for the pattern if nothing else.
- Add sign
This step was admitted rushed. The engraving came out well but, in sanding, I found it hard to keep the sign in a perfect rectangle. For a moment I thought this would look nice almost like a fence. However, the added fragility didn’t turn out great. Ultimately I simply glued it against the front of the frame/stage. Finally, I am just now realizing that ‘122’ was erroneously printed.
- Detailing/Staining.
Not a ton to say here, staining just about everything other than the dowel caps to a nice dark brown as I’d seen it done well before. I think my inexperience is apparent but its not too distracting, especially with how small the frame is. Pictured is a test piece for the final color at its purest/best.
Cost Analysis
(Note cost is considering only the materials used in the final iteration, though the time required to get to that point is included. All tooling costs considered a shared resource, not included.)
Plywood: 1x 2’x4′ sheet $9.15 = 9.15
Painters Tape: 1x Roll $7.09 = $7.09
Wood Glue: 1x Bottle $2.59 = $2.59
Superglue: 1x Bottle $4.28 = $4.28
Vinyl Roll: 3x Rolls (due to color requirement) $0.84 = $2.52
Dark Stain: (some amount of) 1x Can $5.98 = $5.98
Labor: (20 hr before the group split) + 25 hr after group split = 45 hr * $20/hr = $900
Total = $931.61
Once again, the overwhelming amount of cost is (tooling and then) labor.
Overview
Ultimately, despite the required simplification of the design and huge amount of invested time that ultimately didn’t end up in the final product itself, this process was thoroughly enjoyable. The word count for this post is already excessive but I think this project was a fitting check point for and expression of the skills learned in this class thus far.
