Originally Posted by ZilchLab
I'll refer everyone again to the early OS papers, which used an elliptical OS, and yes, there was an on-axis suckout in that, as well.
I agree with your analysis with respect to JBL. Geddes was working with them back then, and there was an agreement that JBL would cover the cost of patenting in exchange for a license to the technology. That process revealed that OS was already in prior art, so the deal ended.
I'd surmise that during the course of that, JBL engineering sussed it out thoroughly with actual prototypes. They have since made EOS products, but only with dome tweeters -- LSR series being the prime example, but there are others in the consumer product lines. An on-axis suckout is a deal-breaker in SR, of course.
It's not as if, being in the public domain, OS technology hasn't been available to the entire industry to exploit, and it's likely a good guess other big hitters have looked at it, as well. I'm not suggesting that it doesn't or can't work, merely that only Geddes himself has found means to convince others it does, and is also big whoop, thus far....
The only thing I'd add to that is that JBL and other pro speaker manu's have very different design considerations than Geddes or any of us. OS has certain design considerations such as the suckout and the lack of horn loading.
Although I haven't tested it, I hypothesize that the closer to true elliptical (as opposed to super-elliptical or squared) and the more axes of symmetry present (ie circular being the greatest), the deeper and most concentrated the notch will be. The greater the asymmetry and less "round" the horn is, the wider and shallower the notch will be.
I don't have the research, data (or even scientific explanation as I'm a layperson in this field) to prove my hypothesize, but this is what I think is occuring in layman's terms:
1. As a horn's mouth terminates, some diffraction effect occurs (I'm not sure how to explain this).
2. The diffraction effect, occurs at a specific time in the wave propagation depending on the path distance from the source.
3. If a point at the mouth termination is the same distance from the source as some other point on the mouth termination, the same diffraction effect will occur.
4. These two identical diffraction effects spaced apart will causing cancellation at a frequency based on the spacing of those effects.
This could be completely crackpot, but it is a guess to what is happening.
Here is how I see it relating to different shapes:
Any point on a circular horn mouth would have an infinite number of points on the mouth with a diffraction effect at the same time and spacing causing a specific null frequency to occur.
If you eliminate the roundness of a circle, you basically get a square. For a square, each point on the square will have 1 opposing point with the same path distance and spacing. Those two points will have a matching set of diffraction points in the opposite axis for a total of 4 points at a given distance. The distances for each set of equidistant mouth point are also quite spread out. The nulls will occur over a spread of frequencies and will not be very noticeable.
If you take away one symmetry of the square you get a rectangle where each point will have an opposing equidistant point but won't have a matching pair in the opposite axis like a square. The cancellations would be spread out over greater frequencies and would have less strength showing as a broadband notch.
Moving back to a more circular arrangement from the rectangle you get an ellipse. An ellipse would act like a rectangle, but with a tighter spacing of frequencies effected. The cancellations would be somewhere between the null on the circular horn and the broadband notch of the rectangle.
The closer the ellipse is to being circular the narrower and stronger the notch becomes. Also, the closer the ellipse is to being rectangular (aka a wide super-ellipse) the wider and weaker the notch becomes.
I think it also holds up that this only occurs on-axis since the horn path lengths to the mouth quickly lose their equidistant nature once the horn is rotated.
Anybody think that this holds water? Don't be afraid to shoot it full of holes. I'm just thinking aloud.
If this is legit (or is close to the target), I think it might push us to using a super-ellipse. I also think it explains why JBL uses the funky looking horns in their PT series and many manufacturers offset the throat. Could it be eliminated by using a horn that is so large as to have this phenomenon occur out of the passband (like the JBL mega-PT)?