A gear with a single chipped tooth produces an impact vibration every time that tooth passes through the mesh zone. At a gear running at 1,500 RPM with 30 teeth, that impact occurs 750 times per second (at 750 Hz). Vibration analysis finds this repeating impact even when the gear is sealed inside a gearbox and shows no external sign of damage.
What Gear Fault Detection Measures
Vibration analysis applied to gearboxes detects faults in gear teeth and gear shafts through their effect on the vibration spectrum. Each meshing gear pair produces energy at the gear mesh frequency (GMF), calculated as shaft RPM divided by 60, multiplied by the number of teeth on that gear.
Faults in the gear teeth create three distinct effects in the spectrum:
- Sidebands around the GMF and its harmonics, spaced at the shaft rotational rate of the affected gear
- Impulse repetition at the shaft rotational rate of the damaged gear
- Elevated high-frequency noise floor from distributed surface wear
Each of these effects points to a different fault type and severity level. Sidebands indicate a localized or modulating fault on one shaft. A rising noise floor indicates distributed surface degradation. Both can exist simultaneously in a gearbox with multiple fault modes progressing at different rates.
Key Gear Vibration Frequencies
| Frequency | Formula | What It Indicates |
|---|---|---|
| Gear Mesh Frequency (GMF) | RPM/60 × number of teeth (for either gear in mesh) | Normal meshing excitation; elevated GMF amplitude may indicate load sharing problems or tooth form errors |
| GMF sidebands | GMF ± n × shaft frequency | Sidebands spaced at shaft speed indicate a modulating fault (eccentric gear, localized tooth defect) on that shaft |
| Hunting tooth frequency | GMF / LCM(teeth_gear1, teeth_gear2) | Appears when one defect tooth contacts the same defect on the mating gear; produces a very low frequency cyclic event |
| 2x GMF, 3x GMF | Harmonics of GMF | Distribution of energy across harmonics indicates tooth profile errors; concentration in higher harmonics indicates shock loading |
| Residual signal | After removing GMF and shaft harmonics | Reveals distributed or localized tooth defects remaining in the signal after periodic components are subtracted |
The GMF itself is not a fault indicator. Every healthy gearbox produces a GMF peak. What matters is the amplitude relative to baseline, the presence and spacing of sidebands, and how energy distributes across harmonics over time.
Gear Failure Modes and Their Signatures
Wear (Abrasive)
Gradual tooth surface erosion from abrasive particles in the lubricant. The tooth profile degrades uniformly rather than at a single point.
Vibration signature: elevated noise floor across all gear mesh frequencies, increased debris concentration in oil analysis (fine metallic particles), gradual increase in GMF amplitude without sharp sideband structure.
Pitting (Contact Fatigue)
Sub-surface fatigue cracks propagate to the tooth flank surface, creating small pits. This is the most common gear failure mode in high-load applications.
Vibration signature: sidebands around GMF spaced at shaft speed; oil debris showing spherical and irregular wear particles; as pitting progresses to spalling, the signal develops impulsive content with each pit passing through the mesh zone.
Tooth Cracking and Spalling
A crack forms at the tooth root from bending fatigue, propagating toward the tooth flank over time. If undetected, the tooth fractures completely.
Vibration signature: impulsive time domain events repeating at shaft speed as the cracked tooth enters the mesh; GMF sideband growth that accelerates as the crack deepens; high crest factor (ratio of peak to RMS amplitude) indicating the impulsive nature of the events. Crest factor rising above 4 to 6 warrants investigation.
Misalignment
Gear centerlines not parallel (spur gears) or not at the correct angle (helical and bevel gears). Misalignment increases the load on one side of the tooth face.
Vibration signature: elevated axial vibration component, sidebands around GMF, increased bearing loads showing as 1x and 2x shaft speed peaks in the spectrum.
Eccentricity
A gear mounted off-center on its shaft. The effective tooth-to-tooth spacing varies once per revolution.
Vibration signature: strong 1x shaft speed sideband around GMF; amplitude modulation of the gear mesh signal visible in the time waveform as a cyclic variation in amplitude at shaft speed.
Advanced Analysis Techniques
Cepstrum Analysis
The cepstrum is the Fourier transform of the log spectrum. It converts evenly spaced sideband families into single peaks, making them much easier to identify than in the raw spectrum.
A peak in the cepstrum at a quefrency (the inverse of frequency, in seconds) corresponding to a shaft period indicates sideband families modulated at that shaft's rotational rate. This directly identifies which shaft's gear is faulted, even when multiple gear pairs share overlapping frequency ranges in the spectrum.
Time Synchronous Averaging (TSA)
TSA averages the vibration time waveform over many shaft revolutions, phase-locked to a tachometer signal on the shaft of interest. Asynchronous content (noise, other gear pairs, electrical interference) averages to zero. Only gear-synchronous events remain.
The result is a clean representation of one gear's contribution to the vibration signal. TSA enables detection of very small tooth-level defects that would be invisible in the raw waveform, because the technique isolates each tooth's impact from the background noise.
Kurtosis
Kurtosis measures the impulsiveness of the time domain signal. For a signal with a Gaussian (normal) amplitude distribution, kurtosis equals 3. As impulsive events appear in the signal from developing tooth defects, the distribution becomes non-Gaussian and kurtosis rises.
A healthy gearbox typically shows kurtosis near 3. Values rising into the 4 to 6 range indicate early-stage impulsive content. Kurtosis is sensitive to early damage but saturates at later stages as the entire signal becomes heavily impulsive, which is why it works best as an early warning metric rather than a severity measure.
Oil Analysis for Gearboxes
Vibration analysis and oil analysis are complementary for gearbox condition monitoring. Each detects different aspects of the same degradation process.
Oil analysis detects wear metal concentration (iron and chromium from steel gears; copper from bronze worm wheels) and particle morphology. Fatigue-type particles have characteristic shapes: smooth spheres indicate rolling contact fatigue; flat platelets indicate normal wear; angular particles indicate abrasion from contamination or surface damage.
A gearbox showing elevated iron in oil analysis but a clean vibration spectrum may have distributed surface wear that has not yet concentrated into a detectable frequency signature. The converse also occurs: a gearbox with a developing sideband pattern may have normal wear metal levels because the crack has not yet reached a stage where it generates significant debris.
Running both methods together covers both scenarios. Oil analysis catches distributed wear early. Vibration analysis catches localized tooth damage and structural changes. The combination reduces the chance of a gear fault reaching a critical stage undetected.
Measurement Points and Setup
Accelerometers mount on the gearbox housing as close as possible to each gear mesh or bearing position. The signal attenuates as it travels through the housing structure, so proximity to the source matters for detecting early-stage faults.
For a two-stage gearbox with three shafts, use a minimum of one measurement point per shaft, positioned radially toward the gear mesh. If the housing geometry allows, add an axial measurement point to detect misalignment.
Two setup requirements are non-negotiable for accurate gear fault detection:
- Steady-state conditions: take measurements at constant load and speed. Variable speed operation smears gear mesh frequencies across a frequency range rather than producing a sharp peak, making sideband detection unreliable without speed tracking (order analysis).
- Known gear geometry: you need the gear ratios and tooth counts for every gear pair before analysis. Without these, you cannot calculate the expected GMF for each mesh or interpret sideband spacing. Get these from the gearbox nameplate or manufacturer documentation before starting.
Baseline measurements taken when the gearbox is known to be in good condition provide the reference against which all future measurements are compared. Without a baseline, you are comparing only against a generic severity guideline rather than the specific machine's normal operating signature.
Continuous Gear Mesh Monitoring
Tractian's vibration monitoring tracks gear mesh frequencies and sideband patterns continuously on gearbox-driven assets, trending changes from the established baseline.
See Mechanical Monitoring