One thing that I would question is how can a good speaker NOT be good for all types of music. If "reproduction" is the goal then a speaker that reproduces well must do so on all signals no matter where they come from. My experience is that a good set of speakers sounds good for any sources.

I admit that some problems with speakers are more evident with some program material than others, but the simple fact that a speaker does in fact sound better or worse with different source material is a clear indication that it has problems.

I would ask Wayne what he means by "fairly wide" coverage (please can we not use the incorrect term "dispersion"). To me 90° is the maximum width that can be handled - in a small room - because wider than that will yield too many near field refections. I can actually live with 60 x 40 in a small room, its just that no mid frequency source can do this (not a reasoanble one that is) so this pattern cannot be combined with any LF source without coverage problems at the crossover.

I think Wayne is quite correct in his comments about "sweet spot" as being an obsolete concept. It is easy to see why it is prefered for most speakers by just looking at their off axis response - its abominable. So few people look at or even care about off axis response - they simply measure and sit on axis. To me coverage patterns are all important, mainly because everything else is so easy to do. But just try and get a constant coverage in all directions above 500 Hz. thru to 10 kHz. THAT is not easy. And when you don't HAVE TO sit on axis you can do things with speaker placement that dramatically improves the small room near field reflection problem.

I have been doing some studies of the subjective perception of minimum phase versus non-minimum phase resonances. I would prefer not to give out the results before they happen (due about June), but let me tell you there is a profound difference in the two. A non-minimum phase resonance occurs when the sound path to the listener is longer than the direct one, so it is delayed in time and hence non-minimum phase. Things like cabinet and waveguide edge diffraction and HOM (Higher Order Modes) in waveguides fall into this category. Minimum phase resonances are those like cone break-up, cabinet resonances etc. Historically these two things are considered the same by just looking at the frequency response, independent of the minimum non-minimum phase characteristics.

This study came about as an attempt to define why a new design that I did sounded so good. It was better than expected and so I am trying to explain why. It has led me to whole other thinking about whats subjectively important in loudspeakers. We know now that nonlinar distortion is not - so what is?

At any rate I would love to define how I did the design and what my design criteria is, but that would lead us too far into a commercial area. I initially just did this design for myself - for my own system. But, as I said, it sounded so good that I had to think about thecommercial aspects. That stops me from giving a detailed description.

I like using 90° flares in cornerhorns because the angle of the horn flare matches that of the walls. That's really an ideal solution because the low frequencies are directed by the walls and those higher up by their horns. Everything is matched and the reverberent field is charged uniformly.

Of course, not everyone can use this placement, and when the speaker is placed out away from the walls, then there are reflections to deal with. If the coverage angle includes an incident wall, then early reflections might make narrow coverage a tempting solution. Some pull speakers away from walls, to reduce early reflections. But either way, I still don't like a narrow coverage pattern. Reflections are controlled, provided the horn isn't pointed at a wall or very nearly. But the listening area just becomes too small. The reverberent field is usually non-uniform too, since the highly directional nature of a narrow HF horn is very different than the non-directional LF energies produced by the speaker.

By the way, why do you not like the term "dispersion"? I like staying away from subjective terms as much as you do, because they're just too ambiguous. But I don't see anything unclear about "dispersion", unless you want to be more specific as to what the cause is for the pattern. I did not, since I was looking for an open-ended discussion on high-efficiency speakers in general. Some might like slot devices, some might like constant-dispersion horns. Others will like round horns and some won't like horns at all.

Wayne

These two comments are not true.

"But the listening area just becomes too small. The reverberent field is usually non-uniform too, since the highly directional nature of a narrow HF horn is very different than the non-directional LF energies produced by the speaker"

Even with a 60° coverage the listening area ten feet back is about ten feet wide. And the reverberant field is NEVER uniform all the way down in frequency - thats impossible. But its also not what one wants. I want the power response to rise at lower frequencies in my rooms because I design them to have a lot of low frequency absorption. The increased power response at LF is exactly compensated for by the increased absorption. Further, the ear is not sensitive to early reflections at frequencies below about 500 Hz due to the way it processes LF signals. So one does not need high directionality all the way down in frequency. But it is critical that the coverage change be smooth and that is very doable with careful design.

The word dispersion, if looked up in a physics text, means a wave speed that changes with frequency. SO using it for polar pattern description is coloqial and misused. Hence I don't use it. This is particularly true since the HOM are truely dispersive - in the physics sense, variable wave speed - and hence the use of the term for polar pattern would truely get confusing. The use came about as a loose description for the way a speaker "disperses" the sound, which, as I say, is not a very scientific description.

To me CD always meant "Constant Directivity" which I much prefer.

I hate to be picky, but correct word usage in science is very important.

Non-uniformity of off-axis response is not generally considered to be a plus. Some live with this defect, but I don't think anyone considers it to be a good thing. The whole point of CD horns is to develop a pattern that is constant with frequency.

I do think that gradually narrowing directivity is preferable to having abrupt changes, but I think that uniform directivity is better still.

You must have totally misunderstood my comments and obviuosly have not seen the plots on my web site. I don't know how you got what you claim I said from my post. I reread it and I never said anything about "Non-uniformity of off-axis response" being a "good thing". I said the exact opposite! "I think that uniform directivity is better still" - SO DO I! But to do this across the entire spectrum is simply impossible.

I think that you call a corner a CD horn down to very low frequencies, but to that I do not agree. Since you cannot get the source at the corners apex, there will be some frequency at which this concept fails. And anyways, the entire concept of directivity at low frequencies in a small room is ambiguous. Further a HF CD source placed in a corner is not in the proper plane (ear level) nor pointing in the correct direction (it points upward - or downward). So while this concept seems attrctive - in is not feasible.

What I said was that the ideal for a small room - where a huge system is also not feasible - is to have the directivity narrow down (since at LF a monopole source has a very wide directivity) to the designed coverage angle starting at about 500 Hz. It should be at the design angle at about 800 - 1 kHz and remain constant from that point up. In a small room it would be virtually impossible to get any substantial directivity below 500 Hz. Using a dipole helps, but they have their own set of problems.

I simply do not understand how you got the idea that I was promoting a varioable coverage angle with frequency.

We will have to disagree on this one. I do not see the corner as a prefered solution because I don't believe that it will work as well as you claim. Do you have any measurements to support you belief?

At what frequency do you cross over from LF to HF and how far away are these two sources? Obviuosly you would need more than a 2-way system to do this. Are the design details on you web site?

You claim that you can get a constant directivity across the spectrum with this approach? Even at the crossover points?

Uniform directivity is the holy grail of CD horn designers. It was once only an issue for prosound, where coverage of large areas requires the use of multiple horns, and you want even coverage without overlap. Interactions from overlap cause response anomalies, so the idea is to cover an exact pattern with an even and uniform field, then have an abrupt pattern stop and handoff to another horn that covers another area.

But even in home use, what I've noticed is that speakers with uniform directivity fill the room better. I think they just sound better.

Take for example, a horn with extreme collapsing directivity. This horn has a long necked throat that curves out to a wide mouth. On axis, the sound can be made to have flat response. This is determined by all the factors involved, the radiating diaphragm, it's motor, the crossover, amplifier and any EQ involved. But when you put it all together, the sound should be flat on-axis.

Such a horn has collapsing directivity. As frequency rises, the pattern becomes narrower and narower. If you move off axis, the high frequency response drops rapidly. What this means is that there is a lot more low frequency energy in the room than there is high frequency energy. High frequency energy is focused to a very small point, so not much is required to sound right on-axis. But if you increased high frequency energy enough to make response good 20° off-axis, then the on-axis response would be ear-splitting. Way too much high frequency energy on-axis.

Now take another example. This is a three-way horn loudspeaker system, with each of the three horns having a curved wall flare. What you see then is that the bass horn is nearly omnidirectional but starts become directional as it is run up into the lower midrange. Then as sound is handed off to the midrange horn, the pattern widens up again. As frequency rises, the pattern narrows and begins to beam. Then it is handed off to the tweeter and the pattern widens a second time. But at high frequencies it narrows once again.

On-axis, and outdoors or in a very dead room, this loudspeaker may sound just fine. But if response is good on-axis, then it will be poor off-axis. There will be low bass and some midrange, but upper midrange and treble will not be present. The energy distribution in the room will have wide peaks and valleys.

This is an example of an uneven reverberent field. On-axis sounds very different than off-axis. The sound far way is uneven as a result. Basically, it's just another way of sayng the tonal balance is off. If you listen to a speaker like this, it may sound OK right on axis in the "sweet spot" but if you move out of that zone, it sounds bad. Movement in the room sounds like you're passing through a phase shifter. Balance is poor unless you're right in line with the speakers.

The reflections from the room have poor balance too, so people with speakers like this are usually obsessive about room treatment. That's because any reflections have poor tonal balance, even in places where there are no nulls from interactions. So owners of speakers with poor off-axis performance are usually very fussy about placement and wall treatments. Their speakers sound only sound good anechoic and directly on-axis.

Being in an open space and listening on-axis is always good, but a speaker with good directional characteristics isn't as adversely affected when heard off-axis. When indoors, reflected energies are more balanced and not as unnatural sounding. It really contributes a lot to the overall sound, so I think that's an important part of the design.

If the speaker sounds balanced off-axis, then that means the energy developed is fairly equal throughout the spectrum. The room is filled with sound. Bass and midrange won't be over-represented and the "sweet spot" becomes a large part of the room.

DI = 10 log [180° / arcsin (sin α/2 · sin β/2)]

where α is the wall angle in one plane and β is the wall angle in the other.

If the sound source is placed at the apex of a corner, then sound radiation is confined to eighth-space. Energy is radiated π/2 steradians into the room in the only direction possible, and the maximum radiation pattern is fixed and defined by the wall angles. Because of this directionality, there is 9dB DI increase over omnidirectional radiation.

Beyond that, I don't mean to be rude, but your position seems overly argumentative.

About crossover points, the transition from low to mid in a π cornerhorn is 250Hz and the midrange is a straight-sided horn. Crossover to the tweeter is at 1.6kHz. As you can see, the wavelengths at the crossover points make it pretty easy to position adjacent radiators within a 1/4 wavelength of each other.

Regarding directivity matching, the hardest range to control is the bass, which is bound by the room's walls. The midrange is placed close enough to the corner that the lowest edge of its range is aided by the walls, so where its directivity control begins to fail, the room walls begin to act as flare extensions. At higher frequencies, the horn flare alone sets its directivity. By using this approach, directivity throughout the audio spectrum is maintained within the range of 9 to 11

• Measurements