A rotating shaft which is used as a power transmission element usually runs with torsional vibration as well as transverse vibration. The vibration couples with other mechanical element such as a gear and is the major cause of noise and vibration problem. Therefore it is necessary to analyze the vibration of multi element vibrational system for the reduction of vibration. For this purpose, a vibration measurement is to be applicable to every part of the system without any extra attachments with excess mass and certain volume.
A method to measure the torsional vibration of a running shaft is investigated in this research. A sticker having fine striped pattern was produced by photo-reduction process. As the shaft runs, black and white stripe runs across a photo sensor and gives a pulse train whose frequency is proportional to the instantaneous rotation speed. Since the frequency corresponds to the surface velocity, fluctuation of the frequency presents the vibration velocity of the torsio
nal vibration. The pulse train is transformed to the vibration velocity signal through a frequency to voltage converter.
For the reliable measurement, two optical sensors was introduced at the opposite position about the center axis to eliminate transverse vibration. The mean of two signals give accurate torsional vibration. In turn the difference between two signal gives transverse vibration. Finally, to observe complete behavior of the shaft vibration four optical sensors were installed, a pair horizontally opposed for torsional and vertical vibration, another pair vertically opposed for torsional and horizontal vibration.
There is discontinuity of the signal due to the ends of sticker which produce a spike in the waveform of vibration. Therefore the FFT analysis is performed as such that the spike is not in the analysis window.
The measurement was applied to a proctical gear reduction to verify the method with an acceleration measurement. Basically it proved to be efficient to observe the vibration of the shaft. However, some noiselike components were found to disturb the real signal. It is not due to the basic principle, however, the cause of the observation must related to accuracy of the striped pattern.