A gear pair is a common means for transmitting rotational energy from shaft to shaft, while maintaining or amplifying speed and torque. The ongoing search for effective ways to utilize the vibration signature of meshing gear teeth to estimate the existence and location of faults spawned numerous approaches. Amongst these are the calculation of statistical condition indicators of a preprocessed time history, wavelet transform of the signal, frequency content analysis. As previously suggested by Randall[1], McFadden[2] and others, a fault alters the frequency/phase modulation of the gear meshing signal. In the frequency or order domain, the gear meshing frequency is the carrier signal, and modulations (both amplitude and frequency/phase) are represented by side bands to that carrier signal. In our previous study[3], the effectiveness of summing the modulation energy for the detection of the existence of a fault was demonstrated. In this study, an attempt to increase the detail of frequency content observation has been made. In addition to the discussed total modulation energy, SNM, or Spectral Normalized Moment is introduced as a further property of the modulation. In the presented study it is shown that in combination with the modulation energy, SNM may allow further understanding of the type of fault dealt with. The primary objective of this study is to show that additional information regarding the condition of the gear is obtainable from a further refined analysis of the modulation side bands, SNM being the relatively crude first step on this path.
