Inside the Box — Simulating a Device in Microgravity
The simulation just took a significant step forward.
Until now, the pulsed offset gyro existed as a floating assembly — forces computed, arrows drawn, numbers logged. Useful for understanding the physics, but disconnected from how this device would actually behave in a real environment.
Today it lives inside a rigid housing, sitting on a surface, bouncing off walls.
What Changed
The assembly — three counter-rotating pairs at 120 degrees, pulsed at 3x frequency — is now enclosed in a rigid body housing. The housing has mass, position, and velocity. It interacts with surfaces through penalty-based contact forces and Coulomb friction.
The key insight: the ball mass and frame mass are separated. The balls generate the internal force through their acceleration on the offset race track. The frame mass is what resists that force. They're independent but coupled — exactly how the real device will work. The balls are fixed relative to their rotation axis, mounted to the frame, and the frame is what moves.
The housing sits inside a containment box — six walls with contact physics on every face. In microgravity, there's no preferred direction. Tilt the assembly, spin it up, and the housing pushes against whichever wall the thrust vector points at.
Three Gravity Modes
Press G to cycle:
- Micro-G — zero gravity. The device floats. Only the internal force and surface contact determine motion. This is the target operating environment.
- Moon — 1.62 m/s². The device is heavy enough to stay on the floor but light enough that the internal force matters. Interesting test case.
- Earth — 9.81 m/s². Full gravity. The device sits on the floor and the internal force has to overcome its own weight to do anything laterally.
The comparison is immediate. Same device, same pulse strength, three different environments. You can watch how the force-to-weight ratio changes everything about the dynamics.
Tuneable Parameters
Everything that matters is now adjustable in real-time:
- RPM (T/Y) — starts at zero. Ramp up and watch the force build. The relationship between speed and force is visible, not just computed.
- Offset (O/P) — the race track eccentricity. More offset means higher peak impulse but also more vibration. There's a sweet spot.
- Pulse Strength (W/X) — how hard the speed modulation hits. Zero pulse = constant speed = no bias. Full pulse = maximum asymmetry.
- Pair Separation ([/]) — distance between counter-rotating assemblies in each pair. Affects torque cancellation.
- Assembly Tilt (arrows) — rotate the entire mechanism inside the housing. The force vector follows.
The assembly starts flipped 180 degrees so the thrust bias already points toward the surface. You're looking at the device from the perspective of the surface it's pushing against.
The Thrust Indicator
A 3D indigo arrow shows the predicted thrust direction before anything moves. It's computed from the assembly orientation alone — no RPM needed. Tilt the assembly with the arrow keys and the indicator follows, showing you exactly where the force will push when you spin up.
This matters for the physical prototype. You need to know which way the device will thrust before you power it on. The indicator is that feedback loop.
What This Means for the Physical Build
Every parameter in this simulation maps to a physical design choice:
- Offset → race track geometry (machined into the track)
- RPM → motor speed (controlled by ESP32 firmware)
- Pulse strength → speed modulation depth (firmware parameter)
- Frame mass → housing weight (aluminium, steel, or printed)
- Pair separation → assembly spacing (mechanical design)
The simulation is a design tool now, not just a physics demonstration. Adjust a parameter, watch the effect, decide what to machine.
Try It Yourself
The core physics is now available as an interactive simulator — adjust RPM, offset, and pulse strength in your browser. Set pulse to zero to see the control case.
What's Next
The simulation work continues alongside physical prototyping. The funding page is live — seven phases from rapid prototyping through to independent verification, each delivering real results before the next begins.
Phase 1 is a Bambu Lab X1 Carbon and first printed race track parts. The question is simple: does the ball roll smoothly through an offset channel at speed?
Everything after that depends on the answer.
Simulated with Taichi GPU compute. Built with Claude Code. Published at indigo-nx.com.