Mechanism Design Guide: How to Select the Right Topology

Why mechanism selection is the hardest decision in mechanical design

Every mechanical product begins with the same problem: an input motion has to be transformed into a specific output behavior, subject to constraints on size, force, speed, accuracy, life, and cost. The choice of topology — which family of mechanism, and which specific architecture within it — locks in 80% of those tradeoffs before the first dimension is ever drawn.

Get it right and dimensional synthesis is a finishing exercise. Get it wrong and no amount of optimization will recover the lost performance — you will be fighting your own kinematics for the life of the program.

The five questions that drive topology selection

1. 1. Input → Output transformation. What is the input motion (rotary, linear, oscillating)? What does the output need to look like (continuous rotary, reciprocating linear, indexed, programmable curve)?
2. 2. Degrees of freedom (DOF). Is one driven input enough, or does the task require two coordinated inputs (differentials, parallel manipulators)?
3. 3. Motion profile. Is constant velocity acceptable, or do you need dwells, asymmetric strokes, sinusoidal motion, or a custom displacement law?
4. 4. Force, torque, and back-driving. What load is at the output, and must the mechanism hold position when unpowered (worm gear, toggle)?
5. 5. Precision and life. Is the priority repeatability and zero backlash (harmonic, cycloidal), or low cost and serviceability (four-bar, slider-crank)?

Mechanism selection matrix

Use the matrix below as a first-pass filter. Each row is a topology in the Kinexaai library; columns are the selection criteria most engineers apply first.

Linkages vs cams vs gears — when to use which

Linkages

Linkages dominate when the required output is a planar path, a function generator, or a specific motion law that needs to be tunable by geometry. They are cheap, robust, and tolerant of dirt and lubrication failure. The penalty is that the motion profile is coupled to the geometry — you cannot change it independently of link lengths.

Cams

Cams are the right answer whenever the output motion is fundamentally arbitrary — a programmable displacement law with dwells, rises, returns, and prescribed velocity and acceleration limits. They cost more to manufacture and inspect than linkages and wear faster, but they decouple motion from geometry: change the profile, get a new behavior.

Gears

Gears win when the task is power transmission at a defined ratio with continuous, smooth rotation. They tolerate the highest torques, deliver the best efficiency, and scale cleanly through planetary, harmonic, and cycloidal architectures when ratio or compactness demands it. They are not the right tool for path generation.

Intermittent & spatial mechanisms

Geneva drives, Maltese crosses, and ratchet-and-pawl systems exist for one reason: an input that rotates continuously, and an output that must move in discrete indexed steps. Spatial linkages (Sarrus, Bennett) are reserved for cases where planar motion cannot satisfy the workspace — deployable structures, parallel manipulators, foldable machines.

A worked example: indexing station for an assembly line

Suppose the input is a continuously rotating motor and the output is an eight-station rotary table that must dwell long enough at each station for a robot to place a part, then index quickly to the next position. Run the five questions:

• Input → output: continuous rotary → indexed rotary.
• DOF: one driven input is enough.
• Motion profile: long dwell, fast index, smooth acceleration to avoid disturbing the parts.
• Torque: moderate; the table inertia is the dominant load.
• Precision: repeatability matters more than absolute accuracy.

The matrix narrows the answer to two candidates: a Geneva drive for low cost and a robust dwell, or a cam-driven indexer if the motion profile needs tuning beyond what eight-slot Geneva geometry allows. Both are intermittent-family mechanisms — the gear and linkage rows in the matrix are eliminated automatically because neither can produce a true dwell from continuous rotation.

From selection to production-ready synthesis

Selecting the topology is half the job. The other half is dimensional synthesis: choosing link lengths, cam profiles, gear ratios, and tolerances that satisfy the functional constraints while staying inside manufacturing, packaging, and cost budgets. This is where most mechanism design programs stall — the search space is high-dimensional, the constraints are coupled, and the only person on the team who can navigate it confidently is usually the most senior mechanical engineer.

Kinexaai automates both steps. Describe the inputs, outputs, and constraints; the platform evaluates every topology in the library, runs dimensional synthesis on the viable candidates, and returns a production-ready design with the dimensions, motion profile, and tradeoff rationale spelled out.