In this paper a method is presented to reduce the noise level of a particle velocity sensor, a thermal two-wire sensor sensitive to acoustic particle velocities, which yields a reduction of the noise of 30 dB. The method is based on utilisation of cross- instead of auto-correlation spectra of two of these sensors. The noise of the sensor becomes even considerably lower than the principal resistance noise level (Johnson noise) of the wires. The specific thermal acoustic sensor does not measure sound pressure but the accompanying particle velocity associated with sound. This particle velocity is deduced from a temperature difference between two heated wires. A drawback is the sensor's lower signal-to-noise (S/N-) ratio at higher frequencies. With the described cross correlation method the noise can significantly be decreased. The time averaged cross correlation signal of two uncorrelated noise sources is theoretically zero and this principle is used in the measurement of the cross spectrum of two sensors. The larger the measuring time and therefore the number of data points stored, the smaller the variance of the noise power in the cross spectrum. The theoretical dependence of the variance in the power of the cross spectrum on the number of stored data points is compared to theory, for various situations. Dependent on the number of data points used, noise reductions as high as 30 dB have been attained.
|Number of pages||7|
|Journal||Acustica united with Acta Acustica|
|Publication status||Published - Mar 2004|