And Another Loudspeaker Array Study

By Pat Brown

This is the latest of a series of measured vs predicted comparisons that Pat Brown conducted on various loudspeaker array types.

Loudspeaker arrays are very difficult to measure. They are large, heavy, and difficult to rotate about their acoustic center. This is why there is much interest in predicting array performance using software. A large number of variables exist, and there is a diversity of opinion on what matters and what does not. The fog clears when comparisons are conducted of measured vs. predicted response. This is the latest of a series of such comparisons that I have conducted on various array types over the past three years.

One-octave-centered polars are provided below of measured vs. predicted (1/6-oct is used internally in the prediction). The measured data is from the CLF data file, and the predicted data is from CATT-Acoustic. The data was magnitude-only, and box interaction/diffraction effects were not considered.

The verdict? Measured vs. predicted correlation can be quite high – more than adequate for design work. Since predicted responses are by nature approximations they cannot account for some of the subtleties of element interaction. As found in the previous comparisons, a simple approach using accurate geometry and magnitude data can produce very good correlation with measured data, and save a lot of time and money by freeing us from measuring the array. pb

Make/Model: Danley Sound Labs SH46

Array Weight: 156 kg (343 lbs)

Measurement Distance: 7 m (23 ft)

Angular Resolution: 5-degrees measured, 2-deg predicted

Data Frequency Resolution: 1.46 Hz

Data File Frequency Resolution: 1/3-oct avg

Measurement Platform: Monkey Forest

Data Format: Transfer Function (mag and phase)

Number of Measurements: 667 (full-symmetry)

Test Lab: Electro-acoustic Testing Company, Inc.