Convention Paper
Determination of the Quality Factors of Loudspeakers by a Statistical Method
AES Convention 90 · Paper 3039 · February 1991
Abstract
Methods of measuring loudspeaker quality factors are available. The most widely used is that of THIELE and SMALL and is now a standard on which many authors and manufacturers rely. Unfortunately, a loudspeaker does not always behave as an ideal parallel R-L-C circuit as assumed by the aforementioned method, and the results of the measurement of f1, f2, [Z1] and [Z2] can not be guaranteed to be free of errors, this having an effect on the precision of the quality factor calculation. To overcome this problem, a statistical method based on the linearization of the impedance curve, which uses several measurement points and linear regression has been developed. This method improves the accuracy of the results, because statistical errors tend to be self-cancelling, and because a set of measurements distributed over a frequency interval is used, thus diminishing the influence of local disturbances on the impedance curve. The statistical method is also well-suited for use with computer controlled automatic measurement systems. The use of a computer is nevertheless not a requirement, because many pocket calculators execute statistical calculations and can be programmed according to the mathematics involved.
