A different article about DBA, translated by Google from German:
http://translate.google.com/translate?prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.poisonnuke.de%2Findex.php%3Faction%3DRaumakustik&sl=de&tl=en&history_state0=This is interesting to me, because he says you need minimum 4 drivers per array to make a planar wave. Using only two at mid-height or two on the floor per array will make cylindrical wave with poor cancellation efficiency and will also create floor ceiling modes. But I can see how the effect could still be positive because while it may not avoid as many modes the wave is roughly planar at ear height because there is a cylindrical wave formed with two horizontally spaced drivers. And side to side wall modes are still avoided.
This picture is interesting.
It shows how the individual spherical waves eventually combine to become planar waves. But this requires some distance. At zero distance, along the front wall, there are only standing waves. At several times the distance of the driver to the wall, and with help from side wall reflections (shown in red,) the spherical waves finally combine to become planar. This is true of front and rear arrays in DBA. I guess the planar formation is about 3x the distance of driver to wall, but planarity increases with distance as the waves naturally flatten with increasing radius.
It's important to note that waves do not have any replationship of affinity to each other. They don't know each other is there. They pass right through each other with no affect on each other. We observe the addition and subtraction at specific times and locations as modes, but there is no real interaction. It is the reflections from the sidewalls and floor ceiling that add more waves to the tank which add to the desired longitudinal wave to flatten out the observed planar wave faster than a single driver would by increasing wave radius alone. The planar wave in this case is not really a single wave, but a collection of spherical waves and wall reflections of spherical waves combining to make the appearance of a single wave.
I think this is why a true cuboid is so important to the planar wave phenomena. Angled ceiling will mess up the reflections to some degree. But I think this will have less affect at lower frequencies with more drivers in the array, since there will be more reflections also.
You want the listening seat and the cancel zone to be within the planar wave zone. There must be some distance between listening seat and the cancel zone where bass respone falls off. So a room needs to have enough length to allow the planar zone to be long enough.
I read in one description (maybe
FoLLgoTT on avsforum?) that the number of drivers in the array affects the high frequency extension of the array because it allows shorter wavelengths to form planar waves coherently. But in light of the need for minimum length of the planar zone, adding more drivers to the array also shortens the distance required to form a planar wave at any freq. So in small rooms, increasing the number of drivers in the array will also create longer "planar zone."