An Experimental Investigation of the Impact of Random Spacing Errors on the Dynamic Transmission Error of Spur Gear Pairs

Abstract

Noise and vibration performance of a gear system is critical in any engineering industry. Excessive vibrational amplitudes originated by the excitations at the gear meshes propagate to the transmission housing to cause noticeable noise, while also increasing gear tooth stresses to degrade durability. As such, gear designers must generate designs that are nominally quiet with low-vibration amplitudes. This implies a gear pair fabricated exactly to the specifications of its blue print will be acceptable for its vibration behavior. Achieving this, however, is not sufficient. As the manufacturing of gears require them to be subject to bands of tolerances afforded by the manufacturing processes employed, the designers must be concerned about variations to the performance of their presumably quite baseline designs within these tolerance bands. This research aims at demonstrating how one type of manufacturing error, random tooth spacing errors, alter the vibratory behavior of a spur gear pair. Two pairs of spur gears are tested for their dynamic transmission error performance. One gear pair with no tooth spacing errors form the baseline. The second gear pair contain an intentionally induced random sequence of spacing errors. The forced vibration responses of both gear pairs are compared within wide ranges of speed and torque. This comparison shows that there is a clear and significant impact of random spacing errors on spur gear dynamics, measurable through examination of their respective transmission error signatures. In the off-resonance regions of speed, vibration amplitudes of the random error pair are higher than the no-error baseline spur gear pair. Meanwhile, at or near resonance peaks, the presence of random spacing errors tends to lower the peak amplitudes slightly as compared to the no-error baseline spur gear pair. The presence of random spacing errors introduces substantial harmonic content that are non-mesh harmonics. This results in a broadband frequency spectrum in addition to an otherwise well-defined frequency spectrum with gear-mesh order components, pointing to an additional concern of noise quality.