Companion to Gear Lapping: More Than Noise

Gear Mesh Simulator

Every tooth engagement pushes a gear set a little off its ideal motion. That transmission error repeats at the gear-mesh frequency and becomes the whine you hear. Pick a gear type, change tooth counts, speed and tooth errors, then switch between unlapped and lapped to see, and hear, what changes.

Gear pair in mesh

Shown slowed down, with the error motion exaggerated so you can see it. Gold dots and lines mark tooth 1.

Transmission error

UnlappedLapped

Finish

Tooth errors · Unlapped

Plays the simulated mesh at its real frequencies. Switch the finish or the gear type while it plays to hear the change.

Frequencies

Transmission error spectrum

Mesh frequency

Teeth engage at teeth × shaft speed. Errors that repeat on every tooth show up as tall lines at 1×, 2× and 3× this frequency. These lines are the whine, and lapping lowers them most.

Contact overlap

Helical, spiral bevel, hypoid and worm teeth overlap across the face, so several teeth share the load and each engagement is softer. The mesh lines drop as the overlap ratio rises, and drop furthest near whole-number values.

Sidebands

Runout makes the mesh force rise and fall once per revolution, which puts sidebands around the mesh lines at the shaft frequencies. In a planetary set, each planet passing the sensor does the same at the carrier frequency.

Simplified model for teaching, not design. Transmission error is the sum of the mesh harmonics, scaled by the face overlap of each gear type; a fixed pattern of tooth spacing errors on each gear; runout that modulates the mesh once per revolution; and a sliding-friction floor that is highest for hypoid and worm sets. The planetary model sums each planet's mesh, weighted by its distance from a sensor on the fixed ring. The spectrum is a 65,536-point FFT with a Hann window, in dB re 1 µm. The 3D bevel and hypoid teeth use spherical involute profiles (base cone from sin δb = sin δ · cos α) along circular-arc, face-milled tooth traces, which is close to real geometry but not a full machine-setting simulation. Error values and efficiency ranges are illustrative, not measurements. Read the article: Gear Lapping: More Than Noise.