With three subs you could get a comparable quality, but placement became more critical. One in a corner, one along a side wall and the last one 2/3 the way up to the ceiling along another side wall. This worked pretty well.

In this same study I allowed each of the woofers a totally independent amplitude and phase and let a computer find the "ideal" for each sub. The most interesting thing was that if I took the three woofers and made the amplitude and phase at two of them completely random, I got about the same result as the complex adaptive one. But think of the advantage. The complex adaptive one has to be set for each room, but the random approach works the same in any room.

I'll leave the concept of making a random filter as a task for the reader (Here is a hint: its called a decorrealtion filter).