• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Horn Speakers - Is it me or.......

OK, it happened again. I recently listened to some large horn speakers. I have periodically done this over many years at various hifi shows. I came away again with the same thoughts. I just don't get them.

Firstly they were clearly highly directional. Unless I was right bang in the firing line on axis they had no high frequencies.

Secondly, mid range seemed over emphasised with a "cuppy" effect. Exactly like the sound you get if you cup your hands around your mouth.

Lastly they were no more dynamic than any other large speaker.

All the same characteristics I have heard previously.

Is it me? Am I biased? Heard the wrong horns? Some rave about horns but it's lost on me.

What are others experiences?
For my experience :
  • Opening coverage, a constant one, should be adapted to listening distance, for our room/saloon or even dedicated HC it's 90/80° horizontal and can be arround 60° vertically (our rooms are wider than they are high, and our ears are positioned horizontally, which also defines the plane to which we are most sensitive), and the horn should be EQ’d in the DSP in near field (40/60 cm), to take into account all horn behavior in the baffle but not the room interaction. EQ for room, at listening position, should be done only in the modal field (below 250 Hz more or less in a regular sized room).

  • The horn must be CD (Constant Directivity) on the horizontal plane at least, without mid-range narrowing or mid-range beaming to have a 50/50 proportion between direct field and reverberated field that rebounds on walls and goes back in your ears thanks to side walls even if you are in front of the speakers.

  • If you are sit down you can use a modern biradial, biradial loses their vertical directivity sooner than a horn CD on both axes, it's different and almost impossible to indicate who is better, but if you listen both sit and stand up, a horn CD on both axes is advised.
    Note that the AudioHorn bi-rad has a unusual shape in order to address mid-range narrowing

  • Fast opening CD horns (so not the X-Shape for ex) have difficulty to load due to fast opening, regular OS horn are in this case, it's why they are often very big to compensate it but it creates another problem in more that price, the directivity match: at crossover, directivity of horn and woofer should be close or what you describe will happen:
    • the "cuppy" effect you describe can be a woofer that opens too wide comparatively to horn, can happen if the horn is not CD/has accident too
    • At the reverse, some woofers have a reputation of "empty midrange" just because some peoples cut them too high so the woofer is too narrowing compared to the solution producing HF, at crossover region.
  • Be realistic, the compression driver breakup (often after 10 kHz) will push the driver out of plane wave radiation, and temporal behavior will happen. At this moment, the driver more or less stops following the horn profile and it creates accidents. It's not due to the horn, and it's not audible (too high, no "real-life signal" at this frequency, and our ears don’t really catch it). So, being CD is the way, but not ultra high, exciting the driver breakup too much is unnecessary, and moreover, what is present off-axis is no longer present on-axis (the on-axis curve will drop in HF even faster)

    It's like the first law of thermodynamics : "Energy cannot be created or destroyed, only transformed from one form to another".
    A horn is all about energy distribution, sending it only where it’s needed and avoid wasting or losing it in another form.
It's why AudioHorn, cited upper, creates X-Shape corresponding to woofer size (and M-Shape for modular big horn), with a round-over return preconised to remove midrange narrowing and midrange beaming. With an advised crossover range to have a perfect directivity match.

The width of the baffle is important too, a baffle is a 180° horn that ends abruptly, the baffle should not be too large (the driver + wood and some cm but no more) otherwise the woofer wavefront will try to follow the baffle and radiate at 180° when it is already narrowing because of its end of pistonic effect.

The best is a not too wide baffle, no need to round-over for the woofer, but for the horn it is mandatory, it can look strange but a woofer midrange narrowing well placed at crossover will help in most cases, if the baffle is too wide the woofer off axis response begin to "shake" on polar plot as there is: End of pistonic effect => 180° baffle => mid narrowing too far so not well placed in frequency. That looks strange but a good baffle is a not too large one: https://audiohorn.net/mid-range-bea...nge-narrowing-as-a-directivity-control-device there is a paragraph about it in this article.

This one respect it, it's an X-Shape X25 with "official" round over with a 8" (I would prefer a WO24P or WO24TX here ^^, it's a SB23NBAC) with a tiny round over, we can see the roundover transition :
X-Shape-X25-paint-4.jpg


Note: Thanks to Back Electromotive Force, it's possible to "push" distortion generated by the woofer breakup higher and allow a perfect midrange reproduction for a relatively big woofer. In practice, it's usually done with a air coil inductor in series to the woofer that will increase inductance in the breakup region and "fix" the VC movement when the membrane breakup moves and creates unwanted voice coil motion. It works only in active speakers, as the air coil inductor should not be bypassed by a parallel component.

In the case of air coil inductor use, it's calculated to replace DSP EQ at the same time, so the SPL remains exactly the same across the whole bandwidth, but distortion is now much lower in the top part of the woofer range.

18Sound AIC (Active Impedance Control) use another approach for the same goal: A fixed secondary coil is mounted on the pole piece, generating a controlled magnetic field that compensates for the nonlinear inductive effects of the primary voice coil. This reduces distortion caused by the variable inductance of the moving coil. In the value chain, the AIC system intervenes earlier, compensating for nonlinear inductive effects at a stage before they affect the overall performance, thereby reducing distortion in the moving coil at an earlier point in the signal path.

It's not horn-related, but it’s important for horn-usage speakers. In particular for the midrange reproduction.
 
Last edited:
otherwise the woofer wavefront will try to follow the baffle and radiate at 180° when it is already narrowing because of its end of pistonic effect
I can’t catch this. If the driver is beaming (because parts of the membrane emit in different phases, so they sum off-axis to lower magnitude than when in-phase), how would this narrow beam follow a wide baffle?
As example, the standard way to measure drivers and tweeters is on a wide baffle, and we don’t see significant artifacts related to directivity where they are already beaming.
 

I have a couple of questions about the horn in the image you posted.

I assume the horizontal measurements are made at the horn's midpoint height, corresponding to the horizontal "ridge" in the left and right sides of the horn. Is there any significant discrepancy between a measurement made in the plane of that horizontal ridge, and an out-of-plane measurement which does not include the ridge? It seems to me the throat geometry in particular changes quite a bit from the ridge-line to the corner.

I realize this horn differs in detail from the horn in the JBL M2; does it differ in basic concept as well?
 
I can’t catch this. If the driver is beaming (because parts of the membrane emit in different phases, so they sum off-axis to lower magnitude than when in-phase), how would this narrow beam follow a wide baffle?
As example, the standard way to measure drivers and tweeters is on a wide baffle, and we don’t see significant artifacts related to directivity where they are already beaming.
The iso-baffle size is relative to the woofer or driver size, and it’s not infinite. The microphone position, often very close (e.g. 31.5 cm), won’t fully reveal the behavior of the wavefront further out.

When measuring a horn or woofer in its final implementation, the baffle effect becomes noticeable and should not be underestimated.

In some cases, too big baffle according to emissive source and analysed frequency range, you can even observe this effect on-axis if the microphone is far enough. The wavefront edges naturally propagate at 90° from any surface they touch, traveling at the speed of sound at every point along the wavefront. Therefore, it will tend to follow the baffle at certain frequencies. At higher frequencies, it will start to beam. The wavefront "leaves" the surface when the wavelength becomes shorter.

If you "force" the wavefront to follow a hemispherical direction, it will eventually tear apart because it can’t maintain the full energy over an increasingly larger hemisphere. Of course, this depends on the wavelength, but it happens sooner than most might expect.

At some point, the wavefront must leave the surface, and when implementing a system, the goal is to allow it to do so at the right moment, before it tears apart.

The phenomenon of "beaming" is primarily due to the size of the emissive surface relative to the wavelength, rather than phase differences. As the size of the emissive surface becomes larger relative to the wavelength, it can no longer maintain a uniform pistonic motion, and the surface begins to radiate in a more complex manner. This results in the narrowing of the beam, with sound energy focusing more in the forward direction as the frequency increases.

Phase differences across the emissive surface can contribute to this phenomenon, but they are not the main cause. I guess you mean that at higher frequencies, each point on the emissive surface vibrates at a different phase, which affects the distribution of energy. However, it is the relative size of the emissive surface compared to the wavelength, the loss of pistonic behavior, that primarily drives the "beaming" effect.
 
Last edited:
I have a couple of questions about the horn in the image you posted.

I assume the horizontal measurements are made at the horn's midpoint height, corresponding to the horizontal "ridge" in the left and right sides of the horn. Is there any significant discrepancy between a measurement made in the plane of that horizontal ridge, and an out-of-plane measurement which does not include the ridge? It seems to me the throat geometry in particular changes quite a bit from the ridge-line to the corner.

I realize this horn differs in detail from the horn in the JBL M2; does it differ in basic concept as well?
I'm not sure I fully understand the first question, but a polar measurement is typically taken at the APEX point and most often at 1 meter (then gated to avoid register reflections). You can find more details here: https://audiohorn.net/guide/apex-polar-map-measurement/

Kolbrek describe it well to, it's an important point of polar measurement, he also describe well midrange beaming and narrowing in his book :
https://hornspeakersystems.info/

There are strict conditions and specific procedures that need to be followed to obtain a relevant polar plot, so be careful with scaling.

For distortion 31.5cm is good as for a SPL level (so the level at 1m) you just need to send 10dB more (HifiCompass does this).

Regarding the second question and the M2, aside from the appearance, the design is completely different in reality. The X-Shape has a 90/85° coverage, "very CD" (constant directivity) and no mid-range narrowing or mid-range beaming if proper round-over is used.

While the CD directivity is pushed higher in the 1.4" version (the X40). The JBL M2 horn is designed for easy molding, while the X-Shape is not. The throat is very different and more complex on X-Shape side, it's only possible with 3D printing.
 
The wavefront edges naturally propagate at 90° from any surface they touch, traveling at the speed of sound at every point along the wavefront. Therefore, it will tend to follow the baffle at certain frequencies. At higher frequencies, it will start to beam. The wavefront "leaves" the surface when the wavelength becomes shorter.

If you "force" the wavefront to follow a hemispherical direction, it will eventually tear apart because it can’t maintain the full energy over an increasingly larger hemisphere. Of course, this depends on the wavelength, but it happens sooner than most might expect.

At some point, the wavefront must leave the surface, and when implementing a system, the goal is to allow it to do so at the right moment, before it tears apart.
I'll still try to understand what’s written, and most importantly, how to use it in practice. For now, it doesn’t seem to help much in explaining how a wave front that has detached from the surface starts to follow the surface again…
The phenomenon of "beaming" is primarily due to the size of the emissive surface relative to the wavelength, rather than phase differences.
Any changes in the frequency response (FR) are associated with phase changes… Narrow radiation is due to the fact that off-axis, signals of the same frequency travel different paths, so initially they don’t sum efficiently, and as the frequency increases, they begin to cancel each other out—right? Accordingly, the causes of this in drivers are their size and (for any non-ideal driver) non-pistonic behavior, which increases the number of radiating zones with phase discrepancies.


As the size of the emissive surface becomes smaller relative to the wavelength, it can no longer maintain a uniform pistonic motion, and the surface begins to radiate in a more complex manner.
On the contrary, a large driver breaks up at a high frequency.
This results in the narrowing of the beam, with sound energy focusing more in the forward direction as the frequency increases.

Phase differences across the emissive surface can contribute to this phenomenon, but they are not the main cause.
The explanation of the cause is missing here. I’ve already presented my version above.
I guess you mean that at higher frequencies, each point on the emissive surface vibrates at a different phase, which affects the distribution of energy. However, it is the relative size of the emissive surface compared to the wavelength, the loss of pistonic behavior, that primarily drives the "beaming" effect.
Once again, the explanation is missing. What else is harmful about the loss of pistonic behavior, besides the emergence of zones radiating in different phases (for each frequency)?

So far, I still don’t understand why a wide baffle is supposed to broaden directivity where the driver starts to beam.

All tweeters are mounted on very wide (relative to their size) baffles, and yet they still beam at higher frequencies according to their physical size.
 
I explained it badly and reverse it lol, sorry, the larger the driver, the sooner it leaves pistonic movement so the sooner it starts to narrow in directivity. This allows for a lower crossover point from a directivity point of view, and it's especially useful to use a bigger horn in that case.

The breakup is a physical phenomenon related to the membrane happening way after the end of pistonic movement : membrane resonance causes several effects such as nonlinear distortion, peaks and dips in the frequency response due to modal behavior, disruption of the intended directivity pattern of course (because out-of-plane wave radiation in the case of compression drivers), and energy storage with delayed release (ringing), which compromises temporal accuracy. All of this are easy to measure.

A large baffle is just a 180° horn with a mid-range narrowing at he extremity, no more. You can easily measure this with polar plots in real life or simulate it using FEA or BEM methods.

As for when the wavefront “releases” from the surface, well, when CD horn designers simulate horns, they look for a smooth and controlled polar response, especially for CD horns, free from accidents even in the final speaker box (either in real life and/or in FEA), so this behavior is built into the design.

We agree on phase, of course, but you should also consider the shape of the wavefront. It's related to the same phenomenon, but it's not about cancellation. When we leave pistonic movement, the shape of the wavefront—i.e., the energy distribution—changes. This new shape makes it easier for the wavefront to beam (or leave the surface) due to its edges. Additionally, the shorter the wavelength, the easier the wavefront leaves any surface. That's why understanding how the wavefront edges want to propagate at 90° from surfaces is important.
 
Last edited:
A large baffle is just a 180° horn with a mid-range narrowing at he extremity, no more. You can easily measure this with polar plots in real life or simulate it using FEA or BEM methods.
Probably a waveguide, not a horn, although that's something to think about :)

That's the point — if I could model it, I wouldn't be asking you questions, I’d be modeling it myself. But since the idea was yours, I thought maybe you had modeled it or seen some models and could share them. You have to admit, it would be easier to explain the concept that way, rather than suggesting someone else test it just to find out how it works :)
We agree on phase, of course, but you should also consider the shape of the wavefront. It's related to the same phenomenon, but it's not about cancellation. When we leave pistonic movement, the shape of the wavefront—i.e., the energy distribution—changes. This new shape makes it easier for the wavefront to beam (or leave the surface) due to its edges. Additionally, the shorter the wavelength, the easier the wavefront leaves any surface. That's why understanding how the wavefront edges want to propagate at 90° from surfaces is important.

I think I’ve understood your idea. But (without modeling)) it still seems to me that it doesn’t work that way.

By the way... That JBL device from the audiohorn's article you referred, according to JBL themselves, doesn’t narrow the dispersion — it actually widens it (Acoustic Aperture Technology, see JBL brochure).
1744234025762.png
 
Yes, you right, 180° WG is more appropriated that "horn".

The 2023 JBL brochure said "Featuring patent pending Dual Dissimilar Arraying and Acoustic Aperture Technology, the JBL 200 series provides remarkably uniform coverage and smooth,accurate sound reproduction for every member of your audience."

And JBL website : "The two 15" woofers 2275H work with the "Acoustic Aperture" technology. This combines an acoustically shaped part with the woofer. The transition from bass to treble (midrange), which is critical for all 2-way systems, is thereby optimized."

The article don't said it narrow the woofer, it said that it guide it (horizontally), take control on the horizontal plane and do what he have to do to match the coverage of the upper horn, JBL take care about Directivity Match.

My jobs in fact is to do horn/waveguide/acoustics lens modeling and FEA simulations, I simulate wavefront in FEA every days ^^, I have simulated this kind of lens btw, the goal is the Directivity Match, there is several way to do it and some are simpler and others more complex, It's also about whether it's worth it both technically and in terms of production costs. But it's not the subject here ^^.

At some point, it becomes difficult to describe what we do; we essentially manipulate an initial energy/wavefront to achieve a specific goal in the case that concerns us here. Sometimes, acoustic impedance (useful for phase plugs, for example), vibro-acoustics (for enclosures), and turbulence + velocity analysis (for ports, for example) are all part of the game.
 
Huh? Could you elaborate?

Rob :)


A radial horn is revolved in one plane. Some of the radial horns from JBL were radial but some were not. But I think all of the so called "biradial" horns were radial. Biradial would imply that it's revolved in both planes, something none of them were!

Here's an example of horn that is truly biradial:
snap dual radial horn2.png


We were considering releasing a true biradial horn and planned to call it dual-radial to avoid any confusion.
 
The article don't said it narrow the woofer, it said that it guide it (horizontally), take control on the horizontal plane and do what he have to do to match the coverage of the upper horn, JBL take care about Directivity Match.
The article says exactly that: "The principle here is to take control on the horizontal part of the wavefront, guide it slightly, and simultaneously induce a midrange narrowing to achieve the desired directivity that matches high-frequency horn directivity."

And in this configuration, the JBL woofer operates up to 1700 Hz, so the purpose of this device is to extend the directivity above 800 Hz. (Тo doubt to ensure smooth directivity transition).
1744271178944.png


My jobs in fact is to do horn/waveguide/acoustics lens modeling and FEA simulations, I simulate wavefront in FEA every days ^^, I have simulated this kind of lens btw, the goal is the Directivity Match, there is several way to do it and some are simpler and others more complex, It's also about whether it's worth it both technically and in terms of production costs. But it's not the subject here ^^.

At some point, it becomes difficult to describe what we do; we essentially manipulate an initial energy/wavefront to achieve a specific goal in the case that concerns us here. Sometimes, acoustic impedance (useful for phase plugs, for example), vibro-acoustics (for enclosures), and turbulence + velocity analysis (for ports, for example) are all part of the game.
It's a pity you didn't share your simulations. A single image would be worth a thousand words here. :) The verbal explanations have rather confused than clarified things.

(Although, if your simulations show the opposite of JBL's data, they could have made things even more confusing)
 
There is no generic explanation or even generic simulation data for that, because it completely depends on what you want to achieve. I was targeting 750hz with 15" not 1700hz.

The 200 series is designed for 100 or 105 degrees of horizontal coverage. At 1700 Hz, of course, the woofer is way too narrowing and you have to widen it, in this case yes.

So they simply do what’s needed to stay close to that at the crossover point, it's just Directivity Match. But if your target is 90 or 85 degrees at, let’s say, 750 Hz, the design will be a bit different, and you will need to narrow, not widen. Sometimes you want to narrow it, sometimes you want to remove some accident, widen it...

My comment about the speaker box width was related to the X-Shape, which are 90 degrees, and in the case of the X25 I showed, cross at 1250 Hz with an 8", with a 90° goal — different.

But I see your point, it should be "manage" and not "induce" in the article, to avoid confusion and address all cases, as all cases are different it's difficult to said something that covers "all", the response is often just "it depends" lol.
 
A radial horn is revolved in one plane. Some of the radial horns from JBL were radial but some were not. But I think all of the so called "biradial" horns were radial. Biradial would imply that it's revolved in both planes, something none of them were!

Here's an example of horn that is truly biradial:
View attachment 443306

We were considering releasing a true biradial horn and planned to call it dual-radial to avoid any confusion.
I see many such bi-radial horns with radius on both planes in their current and older models, exemplary


 
These just popped up in my Instagram feed and I immediately thought of this thread:
IMG_1545.jpeg


They even sounded like someone playing music through a funnel on my iPhone. Any of these big, megabuck horn speakers sound like crap to me.

I find a huge difference between those types of speakers, Avantgarde et al, and horns / waveguides used by JBL and Klipsch and the like:
IMG_0640.jpeg


Martin
 
A radial horn is revolved in one plane. Some of the radial horns from JBL were radial but some were not. But I think all of the so called "biradial" horns were radial. Biradial would imply that it's revolved in both planes, something none of them were!

Here's an example of horn that is truly biradial:
View attachment 443306

We were considering releasing a true biradial horn and planned to call it dual-radial to avoid any confusion.

OK so I thought you were talking about symmetrical coverage. Are you planning on only symmetrical with the same coverage or are you going to go asymmetrical? If you go asymmetrical wouldn't use use different flare rates? Are you limited to symmetrical only?

I attached a profile of both planes of a 2344. Looks to me like both are curved/revolved. If this doesn't fit both revolved can you explain why?

What am I missing?

Rob :)
 

Attachments

  • Improvements in Monitor Loudspeaker Systems_Page_02.jpg
    Improvements in Monitor Loudspeaker Systems_Page_02.jpg
    53.4 KB · Views: 54
I couldn't find any waterfall or polar charts on these, but could anyone speculate on their directivity pattern?

And also predict how much its directivity might differ from that of the TH-4001 horn?
 
Back
Top Bottom