About Internships Projects Contact Resume
Back to Projects

JHU Mars Rover Team / Mechanical Design / FEA

Cycloidal Gearbox Housing

SolidWorks FEA Simulation Torque Analysis DFM CNC Machining

Key Takeaways

CAD model and machined cycloidal gearbox housing

CAD model (left) and machined 7075-T6 aluminum housing (right)

7 kg
Payload capacity
20%
Weight reduction
1.3x
Safety factor

Overview

Part of the Johns Hopkins Mars Rover Team manipulator subteam. I designed the cycloidal gearbox housing for the wrist assembly, sized motors for a 7 kg payload, and optimized the housing for weight while maintaining stiffness. The approach: derive torque requirements analytically, use FEA to validate and push the design lighter, then verify with hand calcs before sending to the machine shop.

Torque Analysis

The wrist has 4 motors, each seeing different torque depending on arm configuration. I derived the torque equations for each joint using statics, then found the worst-case loading at full extension with max payload.

Hand calculations for torque requirements

Torque derivation for motor sizing

BLDC motor and driver board

Selected BLDC motor with driver board

At the first motor, 7 kg payload at r = 0.025 m gives a base torque of 1.7 N·m. With 1.3x safety factor, minimum requirement is 2.2 N·m. I derived equivalent equations for all 4 motors using the same approach.

Motor Torque (Generalized)
τL + τm = Iα
τmotor = M(gr₃ + ⅙a₃²α)
For 7 kg at r = 0.025 m: τ = 1.7 N·m
With 1.3x SF: τreq = 2.2 N·m

This analysis showed we could use smaller BLDCs than originally planned, cutting motor cost by about 50% while still meeting torque requirements with margin.

Housing Weight Optimization

The baseline housing design was conservative. I used FEA to evaluate how much material could be removed while staying under the deflection threshold for gear alignment.

For cycloidal gearboxes, shaft deflection needs to stay under ~8 μm to maintain proper gear mesh. This became the target for the optimization.

ComponentOriginalOptimizedReduction
Housing wall6 mm5 mm17%
Shaft diameter14 mm12.7 mm9%
Flange thickness8 mm6 mm25%
Total mass253 g210 g17%

Combined with additional lightening pockets in non-critical areas, the final design hit the 20% weight reduction target. DFM principles were applied throughout: wall thicknesses stayed above CNC limits, tolerances were specified only where they mattered for gear alignment.

FEA Validation

Ran a static study on the optimized shaft coupling under 70 N lateral load (worst-case from payload at full extension). The goal: confirm deflection stays under 8 μm threshold.

FEA displacement plot showing 6.5 micron max
Hand Calc Check
d = 12.7 mm, L = 30.5 mm, F = 70 N
I = πd⁴/64 = 1.28 × 10⁻⁹ m⁴
δ = FL³/(3EI)
δ = (70)(0.0305)³ / (3 × 71.7×10⁹ × 1.28×10⁻⁹)
δhand = 7.2 μm
FEA Result
δFEA = 6.5 μm

SolidWorks FEA displacement plot (7075-T6 Al, 70 N lateral load)

MetricValueThresholdStatus
FEA deflection6.5 μm< 8 μmPASS
Hand calc7.2 μm< 8 μmPASS
FEA vs hand calc10% diff< 15%PASS

Hand calc assumes a pure cantilever from a rigid wall, which slightly overestimates deflection. FEA captures the additional stiffness from the flange plate and using tetrahedral elements, which have increased stiffness due to triangular nature. The 10% agreement confirms the model is set up correctly and the optimized geometry is good to machine.

Design validated: 6.5 μm deflection is under 8 μm threshold. 20% lighter housing approved for CNC machining.

Bottom Line