This was originally a two-part review, with Lynn Olson commenting on Geddes’ book and on horns in general. Lynn has requested that I remove his part of the review and the associated comments. Here is my take on the book:
I recently finished reading Earl Geddes’ Audio Transducers, and thought I would add my two cents. I agree with Lynn on the fact that horn loudspeaker design is sadly lacking in the theoretical rigor that is applied to conventional direct driver speakers. Earl Geddes has brought modern techniques to the subject, both better analytical work and the use of “T-matrices”. I couldn’t verify every equation, but based on my engineering background, his work seems sound.
Geddes is an iconoclast, and introduces some interesting new ideas (at least to me) as well as puts down some “conventional wisdom”. Examples of the former are: the appropriate choice of coordinate system to get separable coordinates, his “Acoustic Lever”, the effect of the horn mouth on reflections, and his analysis of room reflections. Examples of the latter include: the neglect of horn system efficiency (”Loading essentially became a non-issue with the almost unlimited power capability available today.”), and the non-use of time-domain testing (e.g. waterfall plots, etc.). Because of this, Geddes’ book does not replace existing textbooks and theory of speaker design, but serves as a powerful stimulant for others in the field to re-think their analyses, especially regarding horn theory.
In physics, people generally fall into either the “experimental” category or the “theoretical” category. Albert Einstein and Richard Feymann are examples of theoreticians and Michael Faraday and Ernest Lawrence were experimentalists. I fall into the experimentalist category, so was especially looking forward to Geddes’ chapter on measurements. He presents a concept of taking frequency response measurements at various locations and by mathematical manipulation, generating the terms for his T-Matrix models. A similar plan is laid out for determining the coefficients for the nonlinear model. However, I would like to have seen actual tests and results, so that the loop between modeling and reality is closed. Although a good concept, I get the feeling that the software for doing this data reduction has never been done, leaving this as essentially a hand-waving exercise.
So, with the few reservations noted above, I recommend any speaker designer read Earl’s book, especially horn designers!
- John Atwood
Updated May 22, 2007