Dedecahedral speaker array

12/04: Peter Kassakian has more recently worked on controlling the radiation pattern.

At CNMAT in Berkeley I was fortunate to work on this project during the summers of 2000 and 2001. CNMAT, the Center for New Music and Audio Technologies, mostly makes avant garde music with the help of new technologies developed there and elsewhere. Being an undergraduate student still working in lower-division, there were not many options open to me, but they gave me this one. I worked on fabrication design of what became to be known as the dodecaphone. At first we were considering having the whole box manufactured with plastic, but luckily we ran across Thomas Harder, a carpenter who could work with many materials besides wood.
The speaker array contains twelve almost identical faces. Each face is equipped with a mid-range driver (about 20W), which has its own dedicated channel, so that the radiation pattern of the speaker can be precisely controlled using carefully calculated filters. A simplified version of this idea would be two speakers next to each other producing signals exactly 180 degrees out of phase with each other. A plane of zero amplitude is created in the middle where the waves cancel each other out.

Alternatively, the vbap~ object for Max/MSP allows a signal to be panned over the surface of the speaker without resulting in any loudness changes. The "width" of the apparent source can also be adjusted. This method is simpler but still presents many possibilities of different sounds. For example, the differing kinds of reverberation produced by parts of a room can be exploited by beaming the sound in different directions.

Here is a model of one of the faces. A simplified version of the design would be 12 of these glued together. We chose the dodecahedron shape because it is one of the platonic solids (tetrahedron, cube, dodecahedron, icosahedron), so all of the angles between the faces on the inside are identical (if irrational). This property is useful for controlling the radiation pattern. In order to try to avoid thermal breakdown of the coils in the drivers, we mounted aluminum plates on the faces. We also wanted to separate the drivers from each other to help isolate the channels, so we attached cones to the backs of each face. Here, a small piece of the cone can be seen on the right. Mr. Harder constructed them out of aluminum-foam sandwiches.
This is what the inside of the cones looked like. The black foam was given to us by Meyer Sound, which also sold us the drivers, gave us hints on construction, and helped us measure the speaker after it was completed.
The picture below on the left illustrates again the assembly. On the right, the final product can be seen on a stand from Universal Support, where it is being eyed suspiciously by Professor David Wessel.

The speaker has been used for several concerts as well as experiments on sound cancellation and radiation pattern programming.

Back to the top