I used to think that the problem in trying to blend a subwoofer with a planar was primarily a transient response issue (see my post above, dating back about 15 months). Now I believe that it is primarily a room interaction issue.
By way of introduction, consider the case of a single subwoofer located somewhere along the wall in front of us. There will be one path length from the subwoofer to the listener's ears, and then there will be another path length from the subwoofer to the wall behind the listener, thence reflecting back to the listener's ears. At the frequency where the difference in these two path lengths is equal to one-half wavelength, the energy reflecting off the back wall arrives 180 degrees out-of-phase with the direct sound and the result is a cancellation notch. At the frequency where the path length difference is equal to one wavelength, the reflected energy arrives in-phase and the result is a reinforcement peak. There will be other room-interaction-induced peaks and dips from other reflection paths, and from bass standing-wave modes if the listening position is near a node or an antinode. We can re-arrange these peaks and dips somewhat by moving the subwoofer or the listening position or both, but we cannot eliminate them by placement alone.
Equalization can smoothe the response at one listening position, but may well make it worse at other locations within the room. Whether or not equalization is the right choice depends on your listening style. I will be focusing on an acoustic rather than an electronic solution.
Now note that the problem is not that there are too many of these room-induced peaks and dips; rather, the problem is that there are too few! If we had a large enough room that the peaks and dips were more plentiful and therefore averaged out over short intervals (the ear averages out the sound across roughly 1/3 octave intervals called "critical bands), we'd have subjectively much smoother bass. It is the wide spacing of room-induced peaks and dips that makes them audible and objectionable in the bass region.
Research published in the Journal of the Audio Engineering Society by Todd Welti et al indicates that spaced mulitiple low frequency sources in a room will give smoother in-room bass than a single low frequency source. Briefly, the individual peak-and-dip patterns from each distinct low frequency source is unique enough that the average of several of them is considerably smoother than any one of them could be. We might say that having spaced multiple low-frequency sources "de-correlates" the low frequency sound field, and that is desirable.
Here's a link to a Harmon paper written by Todd Welti. It's not a technical as his AES papers, and is geared towards home theater applications, but the low frequency acoustic priciples apply to two-channel reproduction as well. In case you don't feel like reading the whole thing, Welti concludes by advocating four subwoofers arranged in one of two symmetrical patterns - either in all four corners, or in all four midwall locations:
http://www.harman.com/wp/pdf/multsubs.pdf
How does this apply to dipole speakers? Well, a single dipole can be thought of as two monopole speakers back-to-back separated by a baffle, with the phase of the rearward-facing monopole reversed. So we might expect a dipole speaker's in-room bass smoothness to be comparable to what we might get from four monopoles. In fact, James M. Kates wrote a paper entitled "Dipole Loudspeaker Response in Listening Rooms", which clearly shows the superior in-room bass smoothness of a dipole compared to a monopole. Many dipole owners can attest to the superior bass pitch definition of dipoles, which can be attributed to their much improved in-room smoothness.
So when we try to match up a single subwoofer with two dipole speakers, we are trying to match up systems that have significantly different in-room characteristics. And in most cases the mismatch sticks out like a sore thumb - the low bass simply does not have the same characteristics as the rest of the spectrum. Those who have tried two subwoofers with dipole speakers seem to have a much higher rate of success, as we would expect if the foregoing discussion has merit.
The approach that I advocate is to use four small subwoofers and scatter them asymmetrically around the room. The reason for the asymmetrical scattering is to generate greater dissimilarity in the respective in-room peak-and-dip patterns from the four low frequency sources. I got this idea from Earl Geddes, and use it in a commercial system with his permission. Earl did a computer simulation study comparing four asymmetrically-scattered subs against four symmetrically-placed ones, using one of Welti's recommended configurations:
http://www.gedlee.com/downloads/sub%20study%20.pdf
One final characteristic that the subs should have for this asymmetrical scattering is a steep-slope low-pass filter, so that none of the subs will betray their location by audibly contributing up in the midbass region. Also, good dipoles like Maggies and Quads already have superbly articulate midbass and the last thing they need is a subwoofer mucking that up.
While I do build a packaged multi-sub system (specifically intended to blend well with Quads and Maggies), these basic concepts can be incorporated into a multi-sub system that anyone can assemble themselves from subwoofers that are readily available on the market. The point of this post is to indroduce a new way of thinking about the problem of integrating subwoofers with planars, focusing on room interaction instead of subwoofer transient response.
Duke
