A comprehensive procedure in predicting faults in gear transmission sy
stems under normal operating conditions is presented. Experimental dat
a were obtained from a spiral bevel gear fatigue test rig at NASA Lewi
s Research Center. Time-synchronous-averaged vibration data were recor
ded throughout the test as the fault progressed from a small single pi
t to severe pitting over several teeth, and finally tooth fracture. A
numerical procedure based on the Wigner-Ville distribution was used to
examine the time-averaged vibration data. Results from the Wigner-Vil
le procedure are compared to results from a variety of signal analysis
techniques that include time-domain analysis methods and frequency an
alysis methods. Using photographs of the gear tooth at various stages
of damage, the limitations and accuracy of the various techniques are
compared and discussed, Conclusions are drawn from the comparison of t
he different approaches as well as the applicability of the Wigner-Vil
le method in predicting gear faults.