Worm Gear Noise and Vibration — What the Sound Reveals and How to Engineer It Out

Progressive noise increase in a worm gear drive over months almost always indicates one of three processes: (1) Abrasive wear — running-in particles from the initial operation period were not removed at the 50-100 hour oil change (which many facilities skip), and have been abrading the tooth flanks progressively, increasing profile deviation and mesh noise. (2) Lubrication degradation — the original oil has accumulated metal particles and oxidation products that increase mesh friction and noise. (3) Bearing wear — rolling element bearings in the worm shaft or wheel shaft are developing fatigue spalling. To distinguish: if the noise increase is a smooth, gradual increase proportional to load and speed, (1) or (2) is likely. If the noise has developed a periodic knocking or clicking character, (3) is likely. Drain and replace the oil first — if the noise does not reduce after the oil change and 2 hours of operation, proceed to bearing inspection.

Yes, with appropriate caution. Modern smartphones contain MEMS accelerometers and microphones that are adequate for detecting frequency content in the 20-2,000 Hz range — which covers all gear mesh frequencies for typical industrial drives. Free vibration analyser and FFT (Fast Fourier Transform) apps are available for both iOS and Android. The measurement is most useful for identifying periodic frequencies: a sharp peak in the FFT spectrum at a known frequency (calculated mesh frequency, bearing defect frequency, or shaft rotation frequency) is a reliable indicator even with smartphone measurement quality. The limitations: absolute amplitude measurement is unreliable (smartphone placement and coupling affect the reading); very low-frequency content (below 20 Hz) is not captured; and the measurement requires the smartphone to be in contact with the housing or mounting structure, not held in air.

Load-proportional noise in a worm gear drive has two primary causes. The first is simply that higher load produces higher mesh contact force, which generates higher-amplitude acoustic output at the mesh frequency — this is normal behaviour and does not indicate a problem unless the absolute noise level is unacceptable. The second cause, which indicates a specification problem: inadequate contact pattern (less than 70% face width coverage) concentrates the mesh load on a small tooth area. Under light load, the contact force is low enough that even the small contact area generates acceptable noise. Under full load, the same small contact area is heavily stressed, producing high-amplitude force spikes at each tooth engagement — which radiate as load-proportional mesh frequency noise. To distinguish normal load-proportional noise from contact-pattern-driven noise, compare the noise increase rate: if doubling the load doubles the noise amplitude (6 dB increase), this is normal force-amplitude scaling. If noise increases more than proportionally with load, inadequate contact pattern is the likely cause.

60 dB(A) at 1 metre from the gear housing is achievable for a worm gear drive at low-to-moderate load and speed. Achievability depends primarily on three parameters: (1) Module size — smaller module produces lower mesh frequency and lower acoustic output at the same load ratio; (2) Precision class — DIN 7 thread-ground gear set with documented >=70% contact pattern is typically 8-14 dB(A) quieter than DIN 9 at equal load; (3) Enclosed housing — an oil-bath housing with no acoustic transmission paths to the machine structure provides 6-10 dB(A) additional noise isolation compared to an exposed gear set. For very sensitive acoustic environments (medical offices, concert halls, recording studios), specify DIN 6 or DIN 7 gear set with PA66 nylon wheel if torque allows, PAO lubricant, resilient anti-vibration mounts, and acoustic foam lining on the housing interior.

Airborne noise is acoustic pressure waves propagating directly from the gear housing through air to the listener. Structure-borne noise is vibration energy travelling through the machine structure — mounting bolts, frame members, panels — and radiating as acoustic energy from a larger surface area further from the gear. The distinction matters because the remediation is different. Airborne noise is reduced by acoustic enclosures around the gear or by reducing the gear noise source. Structure-borne noise is reduced by breaking the vibration transmission path between the gear housing and the radiating structure — using resilient anti-vibration mounts, flexible couplings, or acoustic damping pads. In practice, most worm gear noise complaints in industrial machines are dominated by structure-borne noise — the gear housing couples to the machine frame through rigid bolts, and the entire machine panel becomes a large-area radiator at the mesh frequency.

A noise that is prominent at only one specific operating speed but not others is characteristic of structural resonance. At the specific speed, the mesh frequency (f_mesh = n_worm x z1 / 60) coincides with a natural frequency of the housing, mounting structure, or machine panel. At that frequency, the structure amplifies the gear mesh force vibration and radiates it loudly. Solutions in order of implementation ease: (1) Change the operating speed slightly (even 3-5%) to detune the mesh frequency from the structural resonance — if a variable speed drive is used, this is a controller parameter change; (2) Add mass or stiffening to the resonating structure to shift its natural frequency away from the mesh frequency; (3) Add damping (constrained layer damping material) to the resonating panel to reduce its response at resonance; (4) Change to a different gear ratio to produce a different mesh frequency at the same operating speed.

Yes, and it is usually not a sign of a problem. Cold mineral gear oil has much higher viscosity than at operating temperature — ISO VG 460 mineral oil at 5 degrees C may be 6-8x more viscous than at 40 degrees C. This high-viscosity cold oil creates increased viscous drag as the worm thread churns through it, producing low-frequency churning noise. As the housing warms and oil viscosity drops to its design operating range, the noise level reduces. If the startup noise is a churning or gurgling character and resolves within 10-20 minutes of running, this is normal cold-start behaviour. If the startup noise is a metallic knock or grind that does not resolve with warmup, this is a different problem — stop and investigate. To eliminate cold-start noise: switch from mineral to PAO synthetic oil, which has a much higher viscosity index (VI >150) and maintains more consistent viscosity across the startup-to-operating temperature range.

Korea Ever-Power does not provide acoustic test data for gear sets as standalone components — acoustic output depends on the complete machine including housing, mounting structure, coupling, and operating conditions, not on the gear set alone. For EU Machinery Directive noise emission documentation (required under Annex I, Section 1.7.4), the responsible person is the machine manufacturer, not the gear component supplier. Korea Ever-Power can support the machine manufacturer’s noise emission assessment by providing: the gear precision class (DIN class number) and contact pattern coverage percentage — both relevant to predicting mesh noise contribution; the recommended lubricant specification — relevant to lubrication noise contribution; and any application-specific noise test data from previous installations of the same gear set specification, where available from our application engineering records. Request this information at order placement for inclusion in the machine technical file.