Independent project · August 2026–Present

Calculated + packaged in CAD · not yet fabricated or tested

Reaction-Wheel Cube — Single-Vertex Balancing.

A 150 mm cube designed to balance dynamically on one corner using three mutually perpendicular reaction wheels—the same angular-momentum principle used for spacecraft attitude control.

SolidWorks V5 assembly showing the structural cube frame and three perpendicular reaction-wheel modules
01 / V5 structural packagingThree orthogonal axes · swept-volume checked
Envelope150 mmCube side length
Architecture3 axesMutually perpendicular
Recovery target5° / 0.40 sPreliminary requirement
CAD clearance3.16–3.18 mmAdjacent wheel modules

At a glance

Designed as one coupled system.

The challenge is not simply spinning three wheels. Cube inertia, gravitational instability, motor torque, wheel momentum capacity, sensor bandwidth, battery mass, and physical packaging all affect whether the same design can balance.

The current result is a requirements-driven architecture and interference-checked SolidWorks assembly. It establishes a buildable direction; it does not yet demonstrate balancing performance.

My role
Independent analysis, component selection, and mechanical layout
Tools
SolidWorks, rigid-body dynamics, component trade studies
Mass target
1.65 kg target; 1.75 kg hard maximum
Preliminary BOM
$650–$760 before tax and shipping
Current evidence
Calculations, hardware architecture, and CAD clearance checks
Not yet claimed
Fabrication, closed-loop balance, or measured performance

01 / Rigid-body model

Why one corner changes the problem.

Balancing on a vertex places the center of mass above a point contact. Any small angular error creates gravitational torque that drives the cube farther from equilibrium.

01

Center-to-vertex distance

r = (√3 / 2)L
r = (√3 / 2)(0.150) = 0.1299 m

The 150 mm cube geometry fixes the gravitational moment arm before component placement is considered.

02

Effective tipping inertia

Iv = Icm + mr²
Iv ≈ (11 / 12)mL² = 0.0340 kg·m²

For the preliminary 1.65 kg uniform-cube model, the parallel-axis term dominates the inertia about a horizontal axis through the vertex.

03

Gravitational torque

τg = mgr sin θ
τg,5° ≈ 0.183 N·m

At only five degrees of tilt, gravity already consumes most of the preliminary motor's derated torque capability.

04

Instability time scale

λ ≈ √(mgr / Iv)
λ ≈ 7.86 s−1

This fast open-loop divergence motivated a control-loop target of at least 500 Hz.

Model boundary: These values are preliminary analytical estimates using a uniform 1.65 kg cube. Final mass properties will be exported from the completed assembly and used to update the model before controller tuning.

02 / Actuator sizing

Torque starts the correction. Momentum finishes it.

A reaction wheel applies torque to the cube by accelerating in the opposite direction. Motor torque determines the initial correction; wheel angular-momentum capacity limits how much correction remains available before the wheel saturates.

Wheel inertiaIw = ½mwrw²
Reaction torqueτw = Iwαw
Momentum capacityHw = Iwωw

Initial analytical wheel baseline

Geometry
Ø90 × 10 mm
Material
6061-T6 aluminum
Calculated mass
≈ 0.172 kg
Calculated inertia
≈ 1.74 × 10−4 kg·m²
Normal speed limit
5,500 rpm
Momentum capacity
≈ 0.100 N·m·s

The current CAD packaging uses smaller Ø70 mm wheel modules. Final wheel inertia and momentum capacity will be recalculated after the wheel geometry and material are frozen.

03 / Requirement decision

The first target was not feasible. The requirement changed.

Rejected target

10° recovery

At 10°, the preliminary gravitational torque alone is approximately 0.365 N·m—greater than the candidate motor's derated estimate before adding recovery acceleration.

Decision: do not claim 10° recovery.
Selected preliminary target

5° in 0.40 s

The dynamic estimate produced a worst-loaded-axis requirement of approximately 0.210 N·m.

Decision: continue design at the 5° requirement.
Candidate capability

1.16× margin

0.244 / 0.210 = 1.16

The T-Motor MN3508 KV380 remained a preliminary candidate after applying a 20% derating assumption.

Decision: validate on hardware before committing all three axes.

04 / Mechanical integration

From free-body model to constrained volume.

Current transparent SolidWorks packaging model with three wheel modules, battery, controllers, IMU, and mounting structure
Current SolidWorks assembly · transparent envelope shown for packaging context
01

Orthogonal wheel axes

Three perpendicular modules provide torque authority about all three rotational axes while competing for the same central volume.

02

Swept-volume checking

Wheel envelopes were modeled around each rotating assembly and checked with SolidWorks interference detection, not only by measuring static faces.

03

Iterative placement

Axis and component locations were adjusted until adjacent modules retained approximately 3.16–3.18 mm of clearance without unintended interference.

04

Sensor-centered layout

The IMU was placed near the open center of the cube while the battery, motor controllers, and power hardware were distributed around the mounting structure.

05 / Controls architecture

A fast loop needs a complete signal path.

01ICM-42688-PCube attitude and angular rate
02Teensy 4.1State estimation and balance control
03CAN torque commandsThree sensor-based FOC axes
04MN3508 + AS5047PWheel torque and shaft position

The baseline power architecture uses a 6S battery with regulated low-voltage rails. A complementary filter is planned for initial testing; an extended Kalman filter remains an optional later refinement rather than a current claim.

06 / Current status

CAD-ready. Evidence next.

The project has reached a defined architecture and preliminary physical layout. The next phase is to replace assumptions with measured component and system behavior.

  1. 01
    Freeze the wheel

    Finalize Ø70 mm wheel geometry, material, mass, inertia, and safe operating speed; update torque and momentum calculations.

  2. 02
    Validate one axis

    Measure torque response, controller tracking, wheel runout, vibration, and current demand using one module intended for the final build.

  3. 03
    Fabricate and integrate

    Produce the frame and guards, mount the battery and electronics, complete wiring, and measure final mass properties.

  4. 04
    Close the loop

    Commission sensing and FOC, tune edge balancing first, then progress to full single-vertex balancing.

Project takeaway

Requirements should change when the math says they must.

The central design lesson has been traceability: connecting a balance requirement to motor torque, wheel momentum, controller bandwidth, and a CAD layout that can actually be built.