3D Printer FYIBUILD VOLUME IS A SUGGESTION #002
Part of Build Volume Is A Suggestion. Channel: James Bruton.
Come for the 3D-printed parts. Stay to find out why the robot now has a saddle.
In an Autodesk Shop Talk conversation, James describes the loop in plain language: the whole assembly starts in Fusion, parts export as STLs into a slicer, other sketches go to CNC, and 3D printing is how mechanisms get into the real world fast enough to fail in public. That is the editorial reason he is here — not a résumé.
What you’ll accidentally learn
- CAD as geometry type, not “best software”
- Iterative prototyping: print the mechanism, wire it, watch it fail, change the CAD
- Printed parts as structure inside a larger electromechanical system
- Actuators, encoders, and why torque at a joint is a gearing problem
- When a printed mount is the thing that melts
- Open-sourcing CAD and code so the next person can repeat the failure faster — see open-source 3D printing
The omni-directional ball-wheeled bike
What happens when “I wonder if I can ride this” occupies the same workshop as omni-wheels.
In I built an Omni-Directional Ball-Wheeled Bike, James prints the large structural chunks with a 1.2 mm nozzle so the infill is fat enough to hold 4040 extrusion and omni-wheel assemblies 120° apart. Finer-nozzle parts nest into those chunks to hold bearings. Drive: ODrive S1s on 8325 motors, position / velocity / torque control, four driven wheels out of six.
The print is not the bike. The print is the interface between motor, bearing, and extrusion. That is parametric CAD earning its keep: change the chunk, reprint the chunk, keep the rest of the machine.
The one-ball bike
Remove one ball. Discover why the control loop was doing so much work.
I built a Balancing ONE-BALL Bike is the sequel nobody requested. New Atlas summarizing his own video: the whole frame and a rider on one sphere, three custom omni-wheels, motors rated up to 2 kW, wheels mounted vertically so the passive rollers do not jam at speed. Each large wheel carries two rows of 18 rollers — 216 rollers — on custom aluminum hubs with TPU tires.
TPU here is a tire compound, not a phone-case flex. If you only know TPU as “the filament that skips,” start from filament choice and then ignore the consumer use-case.
The screw-bike
Mecanum wheels, but you sit on them, and they are facing the wrong way on purpose.
World’s First SCREW-BIKE: off-the-shelf mecanum wheels top out around 254 mm and cost a fortune, so he prints four ~360 mm wheels — 12 little wheels each, 48 little wheels, 96 skate bearings, TPU tires on a 1.2 mm nozzle, hubs pressed in with a hydraulic press because they are that tight. Hackster and PJRC describe the control stack he shows: Teensy 4.1, ODrive, BNO08x IMU, self-balancing like a hoverboard, CAD and code on GitHub (XRobots/ScrewBike).
This is the CAD → printed mechanism → electronics → controls → parking-lot test path in one project. The saddle is not a joke accessory. It is the load the PID has to hold.
openDog
A robot dog is a walking machine that happens to be a content series.
openDog Dog Robot #7 | Electronics Installation is one episode in an open-source quadruped: 3D printing, CNC, Arduino, ODrive brushless ESCs. James’s own description on that video: CAD is available for anyone who wants to build one; “a dog can literally just be for Christmas.” Six ODrive 3.5 boards, 48 V-capable hardware run at 24 V to start, card-slot wiring so the harness is visible.
Robohub’s focus piece — drawing on his own playlists — follows Mini Robot Dog and openDog V2 (back-drivable brushless motors, IMU for balance). The point for this series is the iteration, not a breed standard. Walking gaits are control problems. Printed links are the cheap way to change the kinematics overnight.
The rideable transformer
Sixteen motors. A Teensy. A robot that folds into something you can sit on.
In I built a Real Transformer I can Ride on, James’s own description (quoted by Interesting Engineering): most real-life transformers leave no passenger space because the car is full of robot. He built one you ride on, like a Power Wheels. 16 motors: 2 RC servos, 8 Dynamixel servos, 6 encoder DC motors. On camera he gears Dynamixels ~2:1 (~100 kg·cm to ~200 kg·cm at the joint), slides arms on V-slot extrusion, and admits he ran out of Dynamixels and dropped to cheaper 270° servos geared the same way.
Hackster notes what the video also shows: Teensy 4.1 with acceleration curves, LED panels, humanoid stance a bit taller than James, vehicle mode actually driven, not a walking biped. The printed parts are panels, gears, servo horns, and sliders — cosmetics and mechanisms in the same BOM.
That is 3D printing after you stop treating the printer as the hobby.
What to steal as a method
- Keep the assembly in CAD. Reprint the interface parts. Do not remodel the universe for a bearing change.
- Fat nozzles exist for structural chunks; fine nozzles exist for the parts that hold bearings. Use both.
- Publish CAD and code when the lesson is the mechanism, not the hero photo.
- Rideable is a test. If it only works on the bench, it is not finished.
- Transformation, walking, and balancing are control problems wearing printed costumes.
Sources
- James Bruton on YouTube
- I built an Omni-Directional Ball-Wheeled Bike
- I built a Balancing ONE-BALL Bike
- World’s First SCREW-BIKE
- openDog Dog Robot #7 | Electronics Installation
- I built a Real Transformer I can Ride on
- Autodesk Shop Talk — James Bruton
- New Atlas — one-ball bike
- Hackster — screw-bike
- PJRC — Teensy / ScrewBike
- Robohub — openDog series
- Hackster — rideable transformer
- Interesting Engineering — transformer motor count