Asymmetric Horn Patterns
by Tom Danley
I figured since the question was asked and SynAudCon played a large role in the direction I have gone with loudspeakers, I needed to explain it as best I can wearing my Pat or Don hat for a moment. I apologize in advance for any commercialism this might have but it is difficult to talk about this without entering into how I see it and that governed what I have done.
Asymmetric Pattern / Shaded Amplitude Horns
In a SynAudCon class in the 80’s, Don Davis mentioned the old standard method of flying a single horn over the front and aimed at the heads in the last row. To the degree the horn’s directivity was consistent, the shape of the side of the horn’s radiation lobe could be used to make the SPL much more constant with distance than an on axis consideration of the inverse square law would suggest. In fact if one pictures the horn’s equal loudness balloon, one can imagine a shape which if it could be made, would present a constant SPL within some zone from front to back. To be sure, these are point sources and the SPL follows the inverse square law but the pattern defines how much there is at any angle and frequency.
That way, as you get closer to the source, you are increasingly off axis.
The problem is that now days, people want more than the bandwidth of a single horn /driver and often enough, more acoustic power as well, but the balloon shape changes with frequency and the systems are far from constant directivity.
Solutions that have multiple points of radiation often / normally do not have a consistent radiation pattern. Multi-way loudspeakers have a pattern of lobes and nulls in the region where any two drivers are both operating.
The fundamental problem here seemed to me to be that when two sources are less than about ¼ wavelength apart, they add coherently into one but if greater than about ½ wavelength, they radiate independently and produce an interference pattern where the result is governed by the observation point or position.
Unlike simple coherent addition, when you reverse one of two sources, you do not have global cancellation but instead you re-arrange the interference pattern.
While this realization came through working on acoustic levitation sources at Intersonics, it was the thing that drove the Unity and then Synergy horns. The idea there was that if I could get two or more frequency ranges to actually combine coherently within a horn (like subwoofers do when close together), then, the radiation pattern would only be governed by the horn and its size / geometry.
About 12 years ago now I was thinking about how to make a full range speaker that avoided these issues. I had returned from Egypt after measuring the Great Pyramid with my trusty TEF12+ for a movie and a friend teased me about making a speaker with a pyramid and it would “make the sound sharper.” Ha-ha!
A day or two later, now thinking about horns I remembered another SynAudCon where Don was explaining the conical horn. The upside was the near constant directivity but the down side was comparatively poor LF loading compared to an exponential horn.
In a flash, it dawned on me why they had less LF loading. If one thinks of the impedance transformation ability of a horn, one finds it has a “high pass” corner based on the rate of expansion.
For example, a 30Hz exponential horn doubles its area about every two feet and so at 300 Hz it would be 1/10 of that distance. If you examine a conical horn in that perspective, one finds its expansion rate is initially fast but slows as you move toward the mouth so the hf driver mounted at the apex feels less horn loading BUT a little farther ahead the expansion is appropriate for lower frequencies. I went out to the workshop and knocked a horn together with side mounted drivers where it appeared mid drivers “should” be and the first Unity was born. Fast forward through 12 years of head scratching, improvements in understanding and R&D to the modern Synergy horn.
By making the horn mouth the entire front of the enclosure, one gets the lowest pattern control loss frequency possible. By using a simple straight sided horn; one gets much closer to constant directivity. By having all the drivers add coherently, they do not produce radiation lobes and nulls. Some Synergy horns can actually be placed side by side (acoustically array-able) and combine seamlessly with music, some like the SH-50 have a small enough time dispersion that they can even reproduce a square wave (from fair to excellent on an oscilloscope) over more than a decade wide bandwidth covering both crossovers.
These have been powerful tools in commercial sound because they produce a pattern where a large percentage of the total acoustic energy in a single front lobe. In my opinion this is a significant improvement over the multiple lobes from multiples source systems.
In the real world, there are some painful rules. When my world was below 100 Hz (Servodrive subwoofers) or above 20 kHz (acoustic levitation) the two things seemed so far apart it was hard to reconcile one set of rules that could define both. The first light went on when the reason for Hoffmann’s iron law became apparent. Any given “acoustic thing” takes up space in X, Y and Z and so if one goes down an octave the cubic volume it occupies is cubed. Well down the page of unfortunate rules are the ones which control horn directivity. Conceptually it would be great if one could make just “the top half” of the horn pattern. There is no good reason to have half the energy go where the ears aren’t and the big reflectors are! Indeed there is good reason not to, so the concept is really appealing…
In reality, one cannot make a sharp cutoff without an acoustically large size horn and if you construct a large horn with the source amplitude apportioned “as if” it were just the top half, it only radiates like that down to some frequency where it then radiates more like a symmetric horn aimed down.
Also, the ugly words “pattern flip” need to be muttered, the difficulty making an asymmetric radiation pattern without pattern flip is another dimension of unpleasant rules one must obey to succeed. You will notice most of our horns are symmetric or only modestly asymmetric.
By far, the symmetric case is simplest and as Don Keele’s rule of thumb for the pattern loss frequency the physical size and wall angle determine the pattern loss frequency in either plane.
The Shaded Amplitude Horn
So what is the shaded amplitude horn?
Start with the single horn over the front, aimed at the heads in the back row.
Now, make that horn a full bandwidth source without lobes and nulls at xover and large enough to have pattern control down fairly low (in a perfect world to DC I suppose). Make the vertical pattern a High Q or narrow angle source, it being aimed at the back row.
Then with vanes or “other means”, one leaks out energy from the bottom of the pattern to make up for having a high Q (which defines the top of the beam). Since all of the energy has the same effective origin, it remains a point source.
The GH-60 was the first of the boxes to apply the shading and to be honest I have learned some more about doing this. The Birdhouse speaker used at Turner Field was an application-specific speaker which had a very short time fuse. It was pretty close to what was needed right off the bat and ended with only a dB or two variation over a large area.
In the Jericho horn, I used a layered combiner to add the outputs from three coax compression drivers into a radiation bubble of the bottom half of a horn. At high frequencies, that old rule about being less than ¼ wavelength comes into play. At 20 kHz, the wavelength is about 5/8-inch while the drivers are about 5 inches in diameter. Hence, the best one can do is producing the wave front shape that would have been produced if it had been a single driver, further back, something you cannot do with a Y throat or manifold. That single part, the combiner took months to figure out, a great example of commenting “I think we can do that” based on a fleeting mental image. For a while I was sure Mike (Hedden) was going to be mad at me but he never gave me any grief (thanks Mike). In that speaker, the low frequency horn has an exit above and below the mid bass/mid /high horn. That was due to the large acoustic power the LF section had to produce, I just didn’t see a way to radiate enough LF energy from a single horn then.
I can’t go much farther right now without spilling some secret Infocomm beans but let’s say I have not stopped working on this, but hopefully what I have explained might help explain what’s going on.
Executive summary: make a full range single relatively high Q point source, leak out energy on the bottom side to suit the needed contour, wave the magic wand and in 6-12 months “voila!”
Why do it this way?
The answer is blowing in the wind – well sort of.
When one uses a line array, they have a system that works by producing an interference pattern and this is under the assumption that when the different modes are dense enough, you can’t hear them.
To be sure, most radiate as separate sources still and if one were to do an ETC of a uniformly driven line source, one sees the energy begins at the closest arrival and ends at the farthest radiation and so impulsive events are stretched out.
When there is a cross wind, this is when the inference pattern becomes plainly audible as the wind modulates its position even just a little. With a wide band point source like the Synergy horns, the radiation balloon is not an interference pattern made of countless lobes and nulls but a nearly homogeneous pressure over a wide bandwidth.
As a result even in a strong crosswind, there is no “swishy swishy” (comb filtering) and the source only has one time arrival over a broad bandwidth so it sounds like the music.
Also audible, unlike the self interference-dependent approach the other unspoken flaws of line arrays (frequency response changes with physical distance and left to right position) are also avoided. Only air absorption of HF is a factor.
Mike dragged a Jericho and a small generator out into a hot Georgia parking lot and made a video I think captures that continuous sound field effect (given camcorder mics). It turned out later the actual distance to the far point is 700 feet. If you have headphones handy use them.
http://www.facebook.com/video/video.php?v=1717077460850
So for you younger guys, seriously, that’s what a bunch of SynAudCon / TEF classes in the 80’s and 90’s did for me. Think about your next 20-30 years and what you have learned and is festering away in there..
Like someone at NASA said of an engineer I knew, but also true of elections, the correct Vector is more important than great Magnitude.
Best Regards,
Tom Danley
Chief Engineer – Danley Sound Labs