Posted by Wayne Parham [ 209.136.24.187 ] on April 18, 2007 at 00:18:33:
In Reply to: New speaker designs on the way posted by Duke on April 17, 2007 at 20:47:16:
Sounds like a "swarm" of good audio! I'm anxious to see your work.
I think most would agree that multiple subwoofers are a good thing. So selling them in a group is a good idea. Might be cool to bundle the subs with the CARA room acoustics modeling program. With CARA, the installation can be optimized and the best positions chosen.
For historical perspective, let me clarify my personal position. A few years ago, I discovered the multi-sub approach as described by Todd Welti. His studies concluded that the best placements were four corners, four wall midpoints or two wall midpoints. At that time, I began to suggest putting woofers in those locations. Four cornerhorns could be used, as an example.
After talking with Earl Geddes at the 2005 GPAF, I decided to experiment with his proposal of random placement. It intuitively made sense, and would take no more effort. If Geddes configuration proved true, then instead of placing all four speakers in corners, I would suggest the Geddes arrangement instead. If cornerhorns were used as mains, then their woofers would suffice as the corner position, and two subwoofers would be placed at pseudo-random positions, one above mid-height and another random but not in a corner. If it worked well, I planned to embrace this.
What I found, when modeling with CARA, was that this arrangement did not always work well. Some random arrangements were OK, but others were pretty bad, to tell the truth. The energy distribution in the room was jagged and non-uniform. Sometimes the min/max variance was reduced, other times not, but in most cases, the sound distribution within the room developed an unusual pattern that I would not want.
The normal behavior of nodes in a symmetrical arrangement is they'll be lumped in certain symmetrical places, like in all four corners or in a figure-8 pattern, something like that. The center of the room forms a "bubble", and other areas also form bubbles. These bubbles represent areas of good uniformity and between them mark dead zones.
Asymmetrical arrangements don't make clearly defined "bubble zones" like that. Instead, they make jagged regions that may be good at one frequency but are bad at another. A particular spot might be good, but just a short distance away is bad. It is impossible to know where is good and where is bad unless you model the room or take measurements. The other obvious problem with asymmetrical arrangements is localization, and of summing with distant midwoofers.
I'm beginning to think it may be best to use both approaches. It seems to me the room actually helps to randomize the energy distribution because it is almost never symmetrical. Rooms have has entry ways, irregular features and furniture, and those break up symmetry. It may be that most situations lend themselves to a placement that is neither Welti or Geddes, but something in between. Certainly midwoofers, those run relatively high in frequency should be close grouped with mains and probably symmetrical for best balance. Subs run deeper can be placed further away, and probably don't need or maybe shouldn't have any particular symmetry.
Irrespective of woofer placement, another important thing is damping. If the walls are lossy, that helps to reduce room modes by damping them. So the best thing is to have furniture and walls that absorb bass frequencies. This does more than layout. But to find the best layout, I think it is important that people make informed choices, and not be told that there is a "magic bullet" solution. If a person is to make a blind decision, I personally would rather go with a symmetrical arrangement, because at least then you can know where the nodes will be.
Either way, symmetrical or asymmetrical, one thing is sure. Using a group of subwoofers helps to reduce room modes by smoothing them out. One can find the best placement for them, if they are so inclined. But even if they're not optimized, a group is better than one.
[ AudioKinesis Forum ] [ Help ]