Interesting topic.... I was thinking of the same thing when looking for an original project last year that would lead to some new lessons learned but I cheaped out. I was thinking about a DDS ENG-1 waveguide surrounded by 6 or 8 of the Extremis 7inch mid-bass drivers, crossed around 1.5kHz (which IMHO is as low as you should cross that particular waveguide). Anyways, as I see it, here are the key issues:
1) For me personally, a big attraction regarding line arrays is the limited driver excursion and lower levels of distortion created by distributing the music power load across multiple drivers. This principle would still hold true in your circle array as it would in any configuration of multiple drivers grouped together in some fashion to share the drive responsbilities
2) Another, albeit not clearly proven, princicple is the cylinder like wave launch of multiple drivers in a vertical line, thus giving the line array some unique imaging and spacial properties that appeal to some listerners and not to others. Talk to Rick Craig about his experience here and also note the fact that he seems to put as much if not more so efforts into his point source designs as his line arrays, though both are extremely impressive IMHO. Your circle array would not benefit from this potential.
3) Focused field array - I believe there is a paten on this from Legacy audio but I am not sure, anyways, the 64,000 dollar question here is that we seem to understand the affect of placing two drivers along a vertical access and as an alternative what happens if they are place together along a horizontal access, but not a lot of information on using four drivers to apply both at once. Even less information about more than four in various symetical geometries around a shared focal point. I say "information" meaning just what I find around the forums as I am not a truly dedicated audio engineer going through the tombs of research available to those with such passion and discipline.
4) In relation to the point above, you have lobing and combing artifacts that may or may not be desirable depending on where they occur and how the may relate to driver response, crossover points (when using asymetrical slopes), and desired voicing. The key here is "center to center" driver spacing and the realtion between this distance and the wavelength of the upper frequency limit you are calling on those multiple drivers to reproduce. In my opinion, once you go around in a circle, the c-t-c is not the drivers beside each other but the drivers across the focal point from one another. This is a large distance and now forces you to break all the rules regarding this relationship (best explanation in the world AFAIK is Dr. Jim Griffins white paper on line arrays). There is also enough anecdotal evidence to conclude that these artifacts may or may not be as noticeable as the engineers think and that people respond differently to them, and that both points are heavily driven by usage, i.e. - do you like to sit in one place, do you want everyone on the couch to get a decent image, do you want everyone mingling about the room in either standing or sitting positions to at least here a balanced frequency response regardless of image specificity. All of these preferences make power reponse, frequency response, lobing and combing artifacts more or less important to each speaker designer (again all IMHO, and I am just a hack that is addicted to reading four of five well known audio forums on very regular basis)
My own conclusion, if you were to set a circle array system up like a near field studio monitor, where each driver is pretty darn close to be of equal distance to your ears, then you may achieve a great result due to lack of distortion and power compression in a group of share frequecy spectrum drivers, especially if they were mid-bass drivers and you had a very capable tweeter in the middle, preferrably a pro compression driver or horn loaded tweeter of some sort. For general use it is wide open and I don't think there is much confidence in anything other than trial and error, knowing that the odds are a bit stacked against you as the power response of such a system is bound to be all over the place on both horizontal and vertical listening axis.
Regards,
Greggo
