Use of a rotating source for acoustic measurements

Abstract

At the Laboratory of Acoustic Imaging and Sound Control, much research is done in the field of room acoustics. The impulse response is widely regarded as the most meaningful characteristic in room acoustics. For a certain combination of source and receiver position, the impulse response is the sound pressure at the receiver position as a function of time generated by an acoustic pulse emitted at the source position. A major requirement for the source is omnidirectionality: the source must radiate acoustic energy equally in all directions. In 1980, Beentjes [2] has developed an acoustic source consisting of a 32-face source sphere and a separate bass loudspeaker. This source radiates omnidirectionally from 60 to 2800 Hz. Taking in account that the range of interest is from 50 to 5000Hz, a successor is desirable. In the present research, the utility of a rotating loudspeaker for room response measurements is investigated. In theory, reproducing a linear frequency sweep with a rotating source results in decomposition of the directivity of the source. The contributions of the mono-pole, dipole and higher order pole terms to the directivity pattern of the source are separated. A monopole is omnidirectional, thus the response to an omnidirectional source can be calculated. Simulations confirm the theory, the directivity is decomposed. Use of a ‘normal’ linear frequency sweep results in a required measuring time of at least 100 minutes, which is not acceptable. Use of parallel linear frequency sweeps reduces the required time to a couple of minutes. A disadvantage of parallel sweeps is that the position of the turntable at the beginning must be the same as the position at the ending of the measurement. This requires a very constant angular velocity of the turntable. Doppler effects have a large influence on the impulse response. Doppler effects result in a complex directivity, resulting in many higher order pole contributions to the impulse response. In addition, the power spectrum gets the shape of a sinc-function, with a smaller central peak for a higher angular velocity. Experiments confirm the results of the simulation. The impulse response of a rotating loudspeaker under anechoic conditions shows many peaks at the positions corresponding to monopole, dipole and higher order poles. The S/N ratio in room measurements is 65 dB at best and 40 dB at worst, which is a very promising result. Using better equipment, the S/N ratio can become even higher. The rotating source in combination with a frequency sweep is a promising tool for impulse response measurements in rooms. The idea of creating a monopole with one loudspeaker is elegant and the achieved S/N ratio is very good. A negative point is the high demand on the measuring equipment (a very constantly rotating turntable is necessary). A problem that has not been investigated in this research, is the extension to 3D. A rotating speaker has only a monopole directivity component in a horizontal plane perpendicular to the rotation axis. Radiation in the direction of the rotation axis is not influenced by the rotation. Therefore, a loudspeaker rotating on one axis does not radiate isotropically.