Gear mesh frequency (GMF) is the vibration frequency produced every time a pair of gear teeth come into contact, and on any gearbox spectrum it is normally the loudest peak in the room. Once you know how to read GMF and the sidebands around it, you can tell wear apart from a cracked tooth before the gearbox ever makes a sound anyone in the plant would notice.
GMF is calculated the same way for any gear pair: the number of teeth on a gear multiplied by that gear's shaft rotating speed.
Both gears in a mesh produce the same GMF, since a pinion with fewer teeth simply turns proportionally faster. A pinion with 24 teeth running at 1,492 RPM meshing with a larger gear both generate a GMF of 35,808 CPM (24 x 1,492). In a multi-stage gearbox, each meshing stage has its own GMF, so a spectrum from a two- or three-stage unit will show several distinct mesh frequencies, each with its own harmonics.
Every tooth engagement is a small, repeating impact as load transfers from one tooth to the next, so GMF and its harmonics (2x GMF, 3x GMF) are always present in a healthy gearbox, they are simply a signature of teeth meshing, not proof of a fault. What matters diagnostically is the trend: rising amplitude at GMF or its harmonics over time relative to that machine's own baseline, and the pattern of sidebands that grows up around those peaks.
A sideband is a peak spaced at a fixed frequency offset above and below GMF (or one of its harmonics). Sidebands appear when the meshing process is being modulated, that is, when something makes the tooth-to-tooth contact uneven once per shaft revolution instead of perfectly uniform.
Because the spacing itself identifies which shaft is at fault, sidebands are often more useful than the GMF amplitude alone. A rise in GMF amplitude with no sidebands is frequently just increased load or normal wear-in; a rise in sidebands with GMF amplitude still modest is often the earlier, more actionable warning.
| Condition | Typical spectral pattern |
|---|---|
| Normal wear-in / light load | GMF present at low-to-moderate amplitude, few or no sidebands, harmonics small relative to 1x GMF |
| Distributed wear (many teeth affected) | Rising amplitude at GMF and its harmonics (sometimes harmonics rise more than the fundamental), sideband families broaden around each harmonic |
| Eccentricity or gear misalignment | Strong, fairly symmetric sidebands spaced at the eccentric shaft's rotating speed, clustered tightly around GMF and its harmonics |
| Cracked or chipped single tooth | An impact once per revolution of the cracked gear: a distinct spike in the time waveform at that shaft's rotation period, plus sidebands at that shaft speed around GMF that grow as the crack propagates |
| Backlash or looseness in the mesh | Multiple GMF harmonics with rich sideband content and a noisier noise floor between peaks |
A practical read of a gearbox spectrum follows this order:
GMF analysis rarely stands alone. Overall vibration severity should still be checked against a recognized reference such as ISO 10816-3 vibration severity zones for the machine class, since a gearbox can have a textbook GMF sideband pattern and still fall inside an acceptable overall severity zone for a while longer. It is also worth ruling out contributing causes before condemning the gearset: a shaft that is out of balance can inject 1x running-speed energy that shows up as unexplained sidebands (see dynamic balancing), and coupling or alignment problems upstream of the gearbox can do the same (covered in laser shaft alignment). Ruling these out first keeps you from chasing a gear fault that is actually a balance or alignment problem.
This is exactly the kind of early, hard-to-see change that plants lose to reactive repairs. Fabrico reads machine condition and OEE straight from the line, and computer vision catches wear and behavior changes that vibration sensors alone can miss, so a developing fault auto-routes a work order before it becomes a teardown. It is EU-built with EU data residency and certified to ISO 27001, ISO 20000-1, and ISO 9001. Book a Fabrico demo to see it on your own gearboxes.
No. GMF depends only on tooth count and shaft speed, so it stays fixed for a given speed regardless of load. What changes with load is usually the amplitude at GMF and its harmonics, which is why trending amplitude against a stable baseline matters more than the raw frequency value.
Some low-level sidebands are normal, since no gear pair is manufactured with zero tooth-to-tooth variation and no shaft runs with zero eccentricity. The diagnostic signal is a sideband family that grows over time relative to the machine's own baseline, not the mere presence of sidebands.
Look at the time waveform, not just the spectrum. A cracked or chipped tooth produces one sharp, repeating impact per revolution of the affected gear, which shows up as a single dominant spike in the time waveform at that shaft's rotation period. Distributed wear instead tends to raise sideband amplitude more evenly without that single dominant impact.
Measure the frequency spacing between adjacent sideband peaks in Hz, then compare that spacing to the running speed (in Hz) of each shaft feeding the mesh. The shaft whose speed matches the spacing is the one associated with the developing fault.