Acoustic Envelope Analysis of a Running 3D-Printed PLA Gear Pair
Anoushkaa Upadhyay
13/11/27
3D-printed gears made from polylactic acid (PLA) are widely used in classroom kits and other low-cost mechanisms, but they wear during repeated use, and there is no convenient way to tell that a gear is changing without stopping and dismantling the system. This study asks a deliberately narrow question: whether a single-channel sound level recording, of the kind a phone-based sound meter produces, contains time-localized events that can be detected reproducibly. One 42-minute run of a battery-driven two-gear PLA assembly was logged as relative sound level against time at 9.51 Hz and analyzed in Python. Because that sampling rate restricts the analysis to frequencies below 4.76 Hz, and because the gear pair carries 100 and 40 teeth, the gear-meshing frequency lies outside the measured band for any motor speed above 7.1 rev/min. The measurement is therefore treated throughout as an envelope measurement, describing how loudness varies over time rather than the acoustics of the meshing teeth. The record divides into a loud opening segment and a stationary remainder 7.0 dB quieter; only the stationary segment is analyzed. Within it, four events exceed the detection threshold, at 1,454, 2,019, 2,127 and 2,180 s. The same four events are returned by every analysis band between 0.5 and 4.5 Hz and every threshold between 6 and 9 dB above the segment mean, and none of 500 phase-randomized surrogates with the same power spectrum produced any detection at all. The study does not establish that these events reflect gear damage.