is a foam plug the only way to avoid horn HOM ?
or can horns be designed to be free of HOM ?
and if so, what design mistakes causes HOM ?
or can horns be designed to be free of HOM ?
and if so, what design mistakes causes HOM ?
All devices have HOMs, some more than others. Foam reduces all HOMs, and some of the main mode as well, but it will always reduce the HOMs more than it reduces the main mode. So foam is always an net reduction in HOM content.
A "pure" waveguide would be HOM free if feed by an HOM free wave front. The only "pure" waveguide would be a conical one, but no source can create a pure spherical wave to drive it. HOMs, as even Makarski suggests, are a net result of the combination of the driver and the waveguide. Any mismatch between these two will result in the propagation of HOMs.
HOMs are created in a waveguide or horn whenever there is curvature of the boundary. The greater the second derivative of the curvature (slope) the greater the generation of HOMs within the device. The lowest HOM generation would then be a curvature that started out matched to the driver and ended with the desired slope with an intermediate curvature that has a minimum of 2nd derivative. This curve is exactly the OS shape. That's how the OS shape is generated.
A conical horn has a first derivative, but no second derivative. That's why it is "pure" - straight walls, no slope changes.
It is not the change in area of a device that matters, as Webster claims, but the change in shape of the walls. The rate of area change is irrelevant.
A "pure" waveguide would be HOM free if feed by an HOM free wave front. The only "pure" waveguide would be a conical one, but no source can create a pure spherical wave to drive it. HOMs, as even Makarski suggests, are a net result of the combination of the driver and the waveguide. Any mismatch between these two will result in the propagation of HOMs.
HOMs are created in a waveguide or horn whenever there is curvature of the boundary. The greater the second derivative of the curvature (slope) the greater the generation of HOMs within the device. The lowest HOM generation would then be a curvature that started out matched to the driver and ended with the desired slope with an intermediate curvature that has a minimum of 2nd derivative. This curve is exactly the OS shape. That's how the OS shape is generated.
A conical horn has a first derivative, but no second derivative. That's why it is "pure" - straight walls, no slope changes.
It is not the change in area of a device that matters, as Webster claims, but the change in shape of the walls. The rate of area change is irrelevant.
Last edited:
What is most detrimental to HOM production, curvature near the throat, closer to the mouth, or anywhere along the horn/waveguide?
Why? Because I don't believe a thing is so simply because someone says it is?
Remember that book you recommended? I read it. Do you remember what it said about experts?
Still no measurements posted? Seems odd, considering the topic. Dr. Geddes, have you ever measured HOMs in horns/waveguides? Is so, have you posted those measurements on this forum? Or somewhere else accessible? I'd love to see them.
I've tired to find the measurements, but have not. It's a big forum to search thru, maybe a little help is needed. There was some mention of measurements in water, but that's about all I could find. Anyone?
Pano, I think we're kindred spirits here so I think I'm talking to receptive ears. Like I said a few posts back, check out what other branches of engineering say about high-order modes from devices designed to radiate a wave. To me, that's actually more productive than what I see studied in audio, not because everything is exactly the same but because there tends to be more rigor in other fields, at least in my opinion. Also, where there is corroboration across fields, I tend to feel more comfortable that the thing being studied is not some quack thing.
Now, mind you, I'm not saying anything about audibility. It could be that the discontinuities that cause HOM also cause other acoustic anomalies that are more audible. Lots of stuff is like that, in my opinion, like phase. I strongly suspect that absolute phase is irrelevant, and that what we hear from phase shift between sources is actually response abborations, the comb filtering that results from alternating constructive and destructive interference. That's just one example of something I find discussed on audio messageboards that I think is a chicken-and-egg argument. But still, I think that HOM are well known, maybe new to audio but not to other fields. How audible they are - in and of themselves - is another matter entirely.
Here are a few links to earlier posts. Look inside for links to papers in other fields that talk about high-order modes.
Now, mind you, I'm not saying anything about audibility. It could be that the discontinuities that cause HOM also cause other acoustic anomalies that are more audible. Lots of stuff is like that, in my opinion, like phase. I strongly suspect that absolute phase is irrelevant, and that what we hear from phase shift between sources is actually response abborations, the comb filtering that results from alternating constructive and destructive interference. That's just one example of something I find discussed on audio messageboards that I think is a chicken-and-egg argument. But still, I think that HOM are well known, maybe new to audio but not to other fields. How audible they are - in and of themselves - is another matter entirely.
Here are a few links to earlier posts. Look inside for links to papers in other fields that talk about high-order modes.
- The dirty little secret of horns. (post #332)
- Are Most Horns Fundamentally Flawed? (post #227)
Thanks Wayne, I'll start digging.
I just worry that this might be like cable skin effect. Applicable to RF, but negligible at AF. That won't stop a lot of people talking about how important is is for audio, tho. 🙂
I just worry that this might be like cable skin effect. Applicable to RF, but negligible at AF. That won't stop a lot of people talking about how important is is for audio, tho. 🙂
And to think I was having all that fun yesterday afternoon listening to a 6khz sine wave. 😀
I always look at it from the RF point of view of standing waves et al. Scimitar antennas as wideband as they are function wonderfully over a very wide bandwidth should shed a little light on the subject of how to avoid the HOM's pitfalls.
I always look at it from the RF point of view of standing waves et al. Scimitar antennas as wideband as they are function wonderfully over a very wide bandwidth should shed a little light on the subject of how to avoid the HOM's pitfalls.
I agree Pano, could be. Definitely could be.
One thing though, even if the HOM themselves are inaudible, I think there's little doubt that many of the same things that cause HOM also tend to cause response ripple. A discontinuity in a horn almost always results in a weird response curve. It can be seen in the impedance curve as well. So it very well could be that these are all interrelated. That is my suspicion, actually, sort of like how phase difference between adjacent drivers causes comb-filtering. Same thing with diffraction and HOM, could be what we are most sensitive to is the response aberrations they cause. Chicken and eggs.
It seems like the longer the path that an HOM has to travel the more delayed it gets behind the main mode. Based on my research this would suggest that the further down the device the more objectionable it is likely to be.
That's the 64k$ question. No one knows for sure and don't believe anyone who tells you otherwise.
We know this. If you do things that reduce HOMs you tend to also reduce internal reflections. This "combination" of things is clearly a benefit as you can see that by just looking at the modern genre of waveguides and horns. They all are getting away from sharp discontinuities etc. Why? Because they sound so much better. Is this because the HOMs have been reduced or because the internal reflections have been reduced? It is not easy to tell since they are highly coupled and sorting out which is which is a very complex task. One that I am certainly not equipped to do, although I do know how it could be done.
We also know that the kind of sonic aberration that is created by an HOM - a delayed high passed signal - has a characteristic that it becomes more audible the louder the playback level. This is shown by Geddes and Lee in their 2008 AES paper. Internal Reflections should not be level dependent because they are a lot like a band pass feedback circuit and are linear. (This has never been tested however, it is just assumed to be true. If it is not true then ALL resonances would be more audible at increased SPL - which is a very interesting hypothesis in and of itself.)
Hence, we know that HOMs exist, we know that improving devices such that the HOMs and the internal reflections are reduced improves perception, but we don't know what % is HOMs and what % is internal reflections.
Here is my position - it doesn't matter. If the right design reduces both then that's what should be done. Prescribing what effect to what perception is just an academic exercise that I, for one, don't have the time for.
We know this. If you do things that reduce HOMs you tend to also reduce internal reflections. This "combination" of things is clearly a benefit as you can see that by just looking at the modern genre of waveguides and horns. They all are getting away from sharp discontinuities etc. Why? Because they sound so much better. Is this because the HOMs have been reduced or because the internal reflections have been reduced? It is not easy to tell since they are highly coupled and sorting out which is which is a very complex task. One that I am certainly not equipped to do, although I do know how it could be done.
We also know that the kind of sonic aberration that is created by an HOM - a delayed high passed signal - has a characteristic that it becomes more audible the louder the playback level. This is shown by Geddes and Lee in their 2008 AES paper. Internal Reflections should not be level dependent because they are a lot like a band pass feedback circuit and are linear. (This has never been tested however, it is just assumed to be true. If it is not true then ALL resonances would be more audible at increased SPL - which is a very interesting hypothesis in and of itself.)
Hence, we know that HOMs exist, we know that improving devices such that the HOMs and the internal reflections are reduced improves perception, but we don't know what % is HOMs and what % is internal reflections.
Here is my position - it doesn't matter. If the right design reduces both then that's what should be done. Prescribing what effect to what perception is just an academic exercise that I, for one, don't have the time for.
No, they are a byproduct of poor horn and phase plug designs which can be responsible for harsh, fatiguing sound, see post #328.
All the usual things that cause harsh, fatiguing sound, (ragged frequency response, distortion, air non-linearity at high SPL, diffraction effects) can be measured fairly easily, but HOMs can not.
To quote Earl Geddes:
"HOM stands for Higher Order Mode. In a waveguide device there is a "main mode" which is where the wave moves down the axis of the device and is everywhere normal to the walls. But other modes are possible, indeed, required, which are not normal to the walls but in fact bounce off of the walls as they propagate. These HOMs would be dispersive in that they move down the device slower than the main mode.
The effects of the waveguide itself will dwarf any effects of the HOMs, making the HOMs very hard to find."
To be without a HOM would be good, but a good HOM is hard to find 😉.
Art
Hi Art
To make matters even more complicated consider this. In all likelihood when a wave front is reflected from a discontinuous part of a waveguide, (at the driver junction, a diffraction slot or the mouth) it is most likely that the reflected wave has HOMs. It would be very unusual for the reflection to be perfectly uniform so some HOMs must be created by this reflection. Now we have HOMs move back up the device towards the diaphragm. They bounce off of the diaphragm and then back out the device delayed even further and most likely generating even more HOMs on the way. It could well be that a poor design generates HOMs exponentially with the amount of discontinuity.
Try and sort all that out.
To make matters even more complicated consider this. In all likelihood when a wave front is reflected from a discontinuous part of a waveguide, (at the driver junction, a diffraction slot or the mouth) it is most likely that the reflected wave has HOMs. It would be very unusual for the reflection to be perfectly uniform so some HOMs must be created by this reflection. Now we have HOMs move back up the device towards the diaphragm. They bounce off of the diaphragm and then back out the device delayed even further and most likely generating even more HOMs on the way. It could well be that a poor design generates HOMs exponentially with the amount of discontinuity.
Try and sort all that out.
that new klipsch elliptical tractrix looks nice (no corners, no flat walls).
Klipsch loudspeaker corner
Klipsch loudspeaker corner
So it would seem that an effort to reduce them in the driver and phase plug would be the most beneficial focus, as there is no shortage of smoothly curved horns/waveguides to be found.
I have believed that this is the next incremental step in better sound for compression drivers and waveguides, yes. That's why I have several patents on how to do this.
There is something unclear in this synopsis. A conical horn has no second derivitive. An OS has a significant second derivative between the throat and conical section, in comparison. You say that HOMs are generated in a waveguide when there is a curvature of the boundary, but that a OS generates less HOMs.
How do you define a HOM free wavefront? Is a "perfectly pure" plane wave not a HOM free wavefront for a conical horn?
Last edited:
Dr. Geddes,
You write that the conical waveguide combined with the source that produces spherical waves is theoretically ideal. Does that mean that in theory the combination of a dome speaker with a conical waveguide is an optimal one? How does it compare against OS+ planar source (ribbon, AMT) or OS + compression driver?
You write that the conical waveguide combined with the source that produces spherical waves is theoretically ideal. Does that mean that in theory the combination of a dome speaker with a conical waveguide is an optimal one? How does it compare against OS+ planar source (ribbon, AMT) or OS + compression driver?
mostly the crossover is to blame ... and almost inclined to say always ... unless ofcourse the design is fundamentally flawed
but when using thin plastic horns and very cheap drivers, there probably are no excuses
which reminds me ...
almost half the world know how horns can flesh your ears 😱
which may also explain why so many people can relate to how bad HOM sounds 🙄
- Status
- Not open for further replies.