### ASA 125th Meeting Ottawa 1993 May

## 2pEA12. Optimized constant beam width arrays.

**Jefferson A. Harrell
**

**
**
*Jet Propulsion Labs., 4800 Oak Grove Dr., Pasadena, CA 91109
*

*
*
**Elmer L. Hixson
**

**
**
*Univ. of Texas, Austin, TX 78712
*

*
*
It has been previously shown by the authors [JAES 34, 221 (April 1990); J.
Acoust. Soc. Am. 91, 2326 (A) (1992)] that the main beamwidth of a loudspeaker
array can be held constant over an octave by the superposition of two arrays by
using suitable filter functions. The beam patterns are identical at the end
(design) frequencies, however, deviations occur at frequencies between the
extremes. The deviations are representative of unwanted acoustic energy if the
deviation is positive when referenced to the desired pattern. While the array
has some error at these design frequencies (for example, undesired side and
backlobes), the total error energy integrated over the octave may be greater
than that of other arrays that use element weights that allow more error at the
design frequencies. Presented is an algorithm to minimize the deviations
between design frequencies while minimizing the error energies, which is based
on a Wiener adaptation. The results are compared to hand-generated guesses. The
impact of linear and nonlinear cost functions is considered.