View Full Version : Interesting description of diffraction in horns/waveguides
For those of who interested, Earl Geddes has a post in his thread on waveguides that is a lengthy description of some of his theory on diffraction in waveguides. It's more about his term HOMs (Higher Order Modes). He describes the difference between these modes and resonances that are not HOMs.
What I found enlightening was how he described this difference, typical resonances being minimum-phase due to the fact that reflections travel along the same path as the original wave whereas HOMs caused by diffraction do not. The latter makes them non-minimum-phase.
The page in the thread is http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-523.html
He makes some other interesting points, such as the importance of the mouth termination and what does and does not impact the impedance profile.
dlr
Dave, it would be worth knowing what criteria Earl is applying to his assment of minimum phase vs non min phase.
When I've discussed min phase with him in the past, he applied min phase to sound power observations (ie all sound radiation integrated over a sphere around the source), not sound pressure at one observation point (ie what a mic or our ear would pick up for the first arrival at the listening spot).
When we discuss min phase vs non min phase, we overwhelming do so in the context of one point of observation.
You can have a situation that is min phase pressure at one point, but not in power.
I think we only need to concern ourselves with min phase at the listening location (pressure at one observation point), when comparing min vs non min phase.
Dave
Deward Hastings
11-24-2011, 12:26 PM
It's more about his term HOMs (Higher Order Modes). He describes the difference between these modes and resonances that are not HOMsMany of his comments also apply to cone breakup, which is also not (or is not necessarily) a minimum phase phenomenon.
Dave, it would be worth knowing what criteria Earl is applying to his assment of minimum phase vs non min phase.
When I've discussed min phase with him in the past, he applied min phase to sound power observations (ie all sound radiation integrated over a sphere around the source), not sound pressure at one observation point (ie what a mic or our ear would pick up for the first arrival at the listening spot).
Yes, in past discussions, one in particular when John was involved, he somewhat reluctantly I would say accepted that when speaking of one point that minimum-phase does apply in general. His position had been that drivers are not M-P, but he usually referred to the polar response and whether or not it can be controlled through a crossover.
When we discuss min phase vs non min phase, we overwhelming do so in the context of one point of observation.
In that post he seems to be viewing it closer to how we have, though he is, I think, focusing on waveguides, as in when he said "...all systems (emphasis mine) have group delay, but for the most part they are minimum phase, i.e. they have a group delay that is explicitly tied to their frequency response.".
You can have a situation that is min phase pressure at one point, but not in power.
His focus being on waveguides, he covers this to some degree. I found it all interesting primarily due to the waveguide issues. There are some differences in the fundamental of the radiation between them and piston drivers that he makes more clear, at least as I read it. It's easier for me to quote him completely at key points that stand out, all emphasis mine.
First an internal resonance of the primary wave (non HOM) does not change the the shape of the wavefront at the mouth. It is spherical in all cases, at a resonance or not. This is true to first order, and I will get to that later. This means that its effect will be seen to be the same at all radiated field points and it can in fact be exactly Eq'd by electronic means. This occurs because the wave reflected from the mouth travels back along the same path from whence it came, and as long as the wave velocity is constant (which it is) it is completely synchronous with the first primary wave and as such its effect can only be to modify the amplitude of the radiation. It is indistinguishable from say a diaphragm resonance.
I think we only need to concern ourselves with min phase at the listening location (pressure at one observation point), when comparing min vs non min phase.
Yes and no.
Yes in that in general we can control only the anechoic portion of the response and with piston drivers usually only at one point. The low frequencies are controllable, but are swamped by the room that are, of course, non-M-P.
No when considering that some of the response that contributes to the non-M-P response can be controlled, but this must be done in the acoustic realm, such as foam damping in a waveguide.
Many of his comments also apply to cone breakup, which is also not (or is not necessarily) a minimum phase phenomenon.
Interesting that he briefly commented on this when saying that "(The primary wave) is spherical in all cases, at a resonance or not." and "...the first primary wave...is indistinguishable from say a diaphragm resonance."
This is one point with which I'm not sure I agree, unless he is also restricting that comment to some particular axis or point. If the primary wave is M-P (as seems apparent) in a waveguide with spherical wavefront, I would think that it would have the same response at all points within the the waveguide controlled region of the polar response.
This is not at all the case for a diaphragm resonance, since these almost always have differing polar response influence. But if restricted to comparison at a single point, I can see how they would be the same given such a restriction.
What was interesting to me was the discussion of the details with regard to horns and waveguides.
dlr
Deward Hastings
11-24-2011, 04:22 PM
This is one point with which I'm not sure I agree, unless he is also restricting that comment to some particular axis or point. If the primary wave is M-P (as seems apparent) in a waveguide with spherical wavefront, I would think that it would have the same response at all points within the the waveguide controlled region of the polar response.
This is not at all the case for a diaphragm resonance, since these almost always have differing polar response influence.The radiation from a cone in breakup is clearly not spherical . . . it it were there could be no cancellation (diffraction) nulls, which clearly are present. I would think that the same argument would apply to a horn . . . if something (anything) before the mouth or at the mouth produces lobing outside the mouth then the wavefront is not spherical and the presumption of minimum phase is questionable at best.
The radiation from a cone in breakup is clearly not spherical . . . it it were there could be no cancellation (diffraction) nulls, which clearly are present. I would think that the same argument would apply to a horn . . . if something (anything) before the mouth or at the mouth produces lobing outside the mouth then the wavefront is not spherical and the presumption of minimum phase is questionable at best.
No one was saying that anything from a cone is spherical. His statement was that the shape of the primary wavefront (first order effect) at the mouth of a compression driver waveguide is spherical and that "...it can in fact be exactly Eq'd by electronic means. This occurs because the wave reflected from the mouth travels back along the same path from whence it came". He was only saying that a primary wave resonance in a waveguide is indistinguishable from a cone resonance. You'd have to ask him to explain his thoughts on that, he did not elaborate.
My intent was to try to drum up some interest in something other than the latest version of a 2-way system. ;)
dlr
It was actually me who even gave some written proof that baffle diffraction is minimum phase, no matter what the shape of the baffle is. And yes, he "reluctantly" accepted that. The strange thing is, it was something so obvious mathematically and simple, I didn't understand what I had to even argue with him. There was no talk about polar response or anything. He initially had said diffraction is more offensive because it is not minimum phase. I said it is very much so, gave proof to it.
Yes, I remember. A few of us had been arguing the point from an empirical standpoint, having for years been able to accurately generate HBT phase form the SPL that matched measured phase including diffraction effects and driver resonances far into "breakup". I could match phase in a 10" Vifa up to 20K.
So, this concept as you can see, has nothing to with polar, power response.
He should really find some other term to what he is describing, not to confuse people.
I think he considers M-P such that the "effect will be seen to be the same at all radiated field points and it can in fact be exactly Eq'd by electronic means." On any given axis this is usually the case, but is seldom, if ever the case on a polar basis, hence the sound power position. At least that's they way I'm interpreting his position.
dlr
It was actually me who even gave some written proof that baffle diffraction is minimum phase, no matter what the shape of the baffle is. And yes, he "reluctantly" accepted that.
Yes, you showed it analystically (it was a beautiful proof BTW) and I showed it through posting of direct experiemntal results.
Funny how credit gets tossed around on the web. ;-)
Dave
No when considering that some of the response that contributes to the non-M-P response can be controlled, but this must be done in the acoustic realm, such as foam damping in a waveguide.
Disagree. Why should one care if a reflection or reverberation, once heard, was originally launched as MP or non-min phase?
Disagree. Why should one care if a reflection or reverberation, once heard, was originally launched as MP or non-min phase?
It's not in the after-the-fact context as we hear it, it's whether or not it can be corrected in a crossover as part of the design. That's his focus having to do with the OS waveguide. What he calls Higher Order Modes are non-M-P and cannot be corrected with a crossover whereas any primary wave or resonance that is M-P can be. His use of acoustic foam in the OS waveguide damps both the primary wave and the HOMs, but with EQ afterwards his claims is that the HOMs are essentially relatively lower in level afterwards. As he points out, proving such is not simple, but his claim is that his studies support this. In this regard I'm not aware of any proof for or against.
I'm trying to describe this very much in brief, but I think it's correct. I do find the logic in his arguments to be compelling as it relates to a waveguide. He has done peer-reviewed research and has his thesis on the OS waveguide that I certainly am not in a position to refute. He says that were it not so, the math would not work. I'm not aware of anyone who has shown otherwise, though certain posters in that thread argue without any solid basis, just their opinions. I do not find them compelling at all.
dlr
I haven't looked at diyaudio thread, but if this is so, it is misuse of terminology and concepts. Above has nothing to do with minium phase concept at all. Even a perfect cone speaker driver on infinite baffle will need to be called "non minimum phase" because it will have different off axis fr shape than on axis, equalize off axis on axis will be off. What has this got to do with minimum phase...
I agree with you when viewed on a strict basis. I was one who argued on the term minimum-phase, but that was as most of us do, it's in the context of some particular point in space. We also know that at any other point in space, we can predict the result at that point for any crossover applied. His position is almost always in relation to the polar response. I'm not saying that I agree with every position he takes.
Maybe I should qualify this. If I'm being accurate, his take is that for typical horns and even an OS waveguide without foam damping, even the direct anechoic response will have both M-P and non-M-P components, the latter being the HOMs. The latter cannot, of course, be controlled with a crossover. It can only be treated acoustically. This is part of the first arrival because it's a component of the output from the waveguide and is present on all axes if not treated acoustically. But he also emphasizes the power response as not being completely controllable with a crossover. From that perspective he's correct. I don't think of it in terms of M-P however.
Many here too often ignore the polar response with the assumption that it's primarily the direct and room response and that anechoic polar response is of little importance as long as the power response in-room is addressed. They are both important to me, hence my emphasis on diffraction control that most only address in a cursory way. Were it considered more important, there would likely be more emphasis on addressing it, although it may be that the aesthetics (felt) or construction (roundovers) gets in the way. In any case, compromise (of the polar response) is generally the rule.
The polar response is one reason why I was so impressed with John K's newest system the Note.
dlr
Just to clarify I am not saying polar response is not important, or being able to effect all the pollar response with electronic equilization doesn't matter.
My point is, using the term and concept of minimum phase'ness in these is just misuse of it, doesn't belog there. Unless you believe that ANY diffraction, regardless of the amplitude of the echoed signal, when summed with the original signal makes the sum minimum phase; which is wrong.
Putting the polar response aside, if I understand his position, it is that the HOMs (his term) are non-M-P, very low in level though audible and not correctable via EQ. I suspect that he considers it non-M-P because the primary wave can be EQ'd with precision while the HOMs cannot and IIRC (I'd have to re-read some of the thread) the FR and phase cannot be re-created one from the other. The difficulty that he admits is that measuring the HOMs separately and distinguishable from the rest is difficult and not currently doable, again IIRC. His basis is his theory on the OS waveguide.
We know rather easily through empirical methods that diffraction from a baffle is M-P. We can measure it. If, however, a waveguide is some special condition, it may be that the HOMs as a separate signal and are not correctable, thus they are non-M-P. We can isolate or even eliminate baffle diffraction and measure the change. Not so to this point with diffraction in a waveguide as far as I know.
dlr
daryl
11-25-2011, 02:03 PM
Really hanging on Earl's every word and dissecting his meanings.
Hmmm.
Remember when he was here before was unable to comprehend room-gain?
I would say however that I have always liked what he is pursuing with his wave-guide.
That is achieving a constant directivity waveguide that does not need multiple diffractions to maintain it's pattern as CD wave-guides normally do.
Excellent!
daryl
11-25-2011, 02:26 PM
A reflection, diffraction or any other action is exactly that, period.
Makes no difference whether on the face of a cabinet, off a wall or within a wave-guide.
They are ALWAYS minimum-phase!
If it happens in nature it is minimum-phase period!
The only reason you might say a particular reflection is not minimum-phase is that in the case where the amplitude of reflection(s) is great enough to cause null(s) there is no possible value to multiply zero by to achieve the original signal magnitude and no phase can be derived.
Signal timing is strictly determined by phase and if phase response is aligned to zero reflections will be neutralized.
Obviously nulls cannot be realligned to zero and in practice you would not want anything close to a null because of the noise/error problems it would create.
It's not that systems with nulls are not minimum-phase but just that there is no possible correction.
johnk...
11-25-2011, 04:14 PM
As far as I know or remember the one thing about Earl's discussion is that he has made illusions tot he non MP character of HOMs but never showed any direct evidence to support his position. Measurements of wave guides like his typically show MP behavior.
Earl used to argue that drivers were not MP because the driver's transfer function was 3-D and dependent on where it was measured. From his point of view at the time, only when the output was single valued, like an electronic filter, could the device be MP. I argued the standard fair, MP implies a relationship between phase and amplitude and that the response on and off axis was still MP, but different MP. Much of this came about in the discussion of eqing diffraction and dipoles. That argument has little to do with MP or not MP. It is simply a matter of the fact that a single correction is not a 3-D correction, mp or not. Also, with regards to dipoles I repeatedly made the statement that for a true dipole the response to be EQ'ed is that of an LP filtered dipole where the polar response was truly constant directivity (figure 8) and that in that frequency range a single eq function was adequate. Other member of the discussion constantly chose to ignore the part about LP filtering the dipole response to support their argument that dipole diffraction could not be eq'ed; The typical DIYA discussion of "I don't care what you say, I going to distort it to support my position."
Anyway, I have a lot of respect for Earl and his contributions by I thinks sometime he is a little close minded and it is difficult to get him to look at things from a different perspective.
Pete Schumacher ®
11-25-2011, 04:55 PM
I'd like to know how a waveguide is any different than any other baffle. Other than the shape, the physics is the same, with a waveguide doing constructive diffraction, uniformly across the band.
All I know is, the normal edge diffraction present on a flat baffle with small radius roundovers, is completely absent with the 8" waveguide I've been using with the RS28 domes. The guide is smoothly terminated on the baffle, and that baffle has a sharp edge, but there's not a hint of diffraction ripple normally associated with flat, sharply terminated baffles, between 1KHz and 20KHz.
A reflection, diffraction or any other action is exactly that, period.
Makes no difference whether on the face of a cabinet, off a wall or within a wave-guide.
They are ALWAYS minimum-phase!
If it happens in nature it is minimum-phase period!
The part that is still a question in my mind is beyond the first order diffraction effects. I'm not necessarily in agreement with everything he says, but within the restriction of a waveguide that will have multiple order diffraction effects, I don't see anything that proves one way or the other. He does say that without the HOMs as he describes them, the math would not work. On that aspect I have to accept it as I can't refute and it I'm not aware of anyone who has. The question devolves then to whether or not they can be non-MP as he says they are. This is what I found interesting to consider. At first glance they should all be.
Higher order diffraction on a flat baffle is probably vanishingly small in most cases given the angles, distances and distribution. Within the confines of a waveguide, however, it seems logical that they may be a lot higher and might be significant enough to be audible. Whether or not they would be considered MP is, I think, something that can't be assumed.
His point about non-MP is IIRC more an extension of the HOMs he describes from his theoretical work as having excess group delay, his term, rather than the typical group delay one would see from a MP signal. This does seem to describe the same result that one gets from baffle diffraction that has its own group delay that includes excess delay due to the extra distance from source to baffle edge. This latter we know to be MP, although what we see is primarily the first order diffraction effects. The rest is swamped.
I just find the debate interesting. Prior to reading the threads at diyAudio I had no interest at all in waveguides. They are more interesting than what dome or cone tweeter can be installed in a PE waveguide. I've played with them a bit, but the result is generally nowhere close to the efforts that have gone into the commercial compression driver systems.
dlr
I'd like to know how a waveguide is any different than any other baffle. Other than the shape, the physics is the same, with a waveguide doing constructive diffraction, uniformly across the band.
As was pointed out once when I forgot to keep it in mind, a flat baffle is a waveguide of sorts with it's coverage being 180 degrees and the control range region being dictated by the baffle dimensions. The difference is the gain. For a flat baffle it's a plain 6db over no baffle for a theoretical point source.
All I know is, the normal edge diffraction present on a flat baffle with small radius roundovers, is completely absent with the 8" waveguide I've been using with the RS28 domes.
The waveguide control is dictated by the diameter of the mouth. An 8" waveguide would then control down to about 1700Hz. There is probably some kind of transition region to the area of no control, so it's not surprising that you would see no baffle edge effects. It will be almost all baffle step if anything. The polar response is probably a bit more erratic.
dlr
It's not in the after-the-fact context as we hear it, it's whether or not it can be corrected in a crossover as part of the design. That's his focus having to do with the OS waveguide. What he calls Higher Order Modes are non-M-P and cannot be corrected with a crossover whereas any primary wave or resonance that is M-P can be. His use of acoustic foam in the OS waveguide damps both the primary wave and the HOMs, but with EQ afterwards his claims is that the HOMs are essentially relatively lower in level afterwards. As he points out, proving such is not simple, but his claim is that his studies support this. In this regard I'm not aware of any proof for or against.
I'm trying to describe this very much in brief, but I think it's correct. I do find the logic in his arguments to be compelling as it relates to a waveguide. He has done peer-reviewed research and has his thesis on the OS waveguide that I certainly am not in a position to refute. He says that were it not so, the math would not work. I'm not aware of anyone who has shown otherwise, though certain posters in that thread argue without any solid basis, just their opinions. I do not find them compelling at all.
dlr
Dave, I think you missed my point. John explained it again, which was that Earl felt that a system radiating in 3D space wasn't min phase even if it was observed to be min phase at each observation point.
I believe that if one feels MP is important, then this is necessary only for first arrival. Is Earl saying that the first arrival of a horn in non MP withjout the foam to control HOMs? Or is he saying that it is min phase in a single point but not integrated over 3D (ie power) which I think doesn't matter given the room contribution and multiple surface reflections.
It is trivial to determine if a horn is MP in the direct arrival, with or without the HOM-busting foam. Just measure its frequency response at a point in space, calculate excess phase and Min Phase. If excess phase is constant slope over frequency (ie flat excess delay), then it's Min Phase. If it's not, then its not MP.
Why all the mystery surrounding this? One measurement will tell the tale. Anyone have a horn available that they can measure frequency response on?
Dave
A reflection, diffraction or any other action is exactly that, period.
Makes no difference whether on the face of a cabinet, off a wall or within a wave-guide.
They are ALWAYS minimum-phase!
If it happens in nature it is minimum-phase period!
Take a room, assume its height and width is same for simplification. Assume front and back wall are anechoic, again just to make the example simple, and assume the side walls, floor and ceiling are very well reflective. Now assume a point source in the middle and say a few feet away you have the mic and equal distance from the side walls and ceiling and floor. If you consider the mirror sources from the side walls, ceiling and floor, the mic receives the direct sound, and 4 same time arriving reflections from the mirror sources. The sum of reflections amplitude will be higher than the direct signal, unless the path length difference between direct and reflections are too much, which most often won't be. In this simple example the system, which can be only defined by the mic at that position is not minimum phase. So it can happen in nature.
But all that said, I don't see how a linear system concept such minimum phaseness with a very well definition of its own, can be used for polar response being equilazable or not. It has nothing to do with it, just misuse of terms and concepts, as I said early on. It causes nothing but confusion, as evidenced here.
Pete Schumacher ®
11-25-2011, 06:33 PM
The waveguide control is dictated by the diameter of the mouth. An 8" waveguide would then control down to about 1700Hz. There is probably some kind of transition region to the area of no control, so it's not surprising that you would see no baffle edge effects. It will be almost all baffle step if anything. The polar response is probably a bit more erratic.
dlr
You'd think so, but that is not the case. The curves are almost as smooth in the guide as they are on the very large baffle. In fact, with just a small offset angle, the responses are even smoother than the on-axis response, and show remarkable consistency as you move well off axis.
It's not shown here, but I recently ran a set of plots, and at 60deg, only the top octave is showing a roll off.
http://i12.photobucket.com/albums/a203/pete_schumacher/shallowellipticalnothroatwithmediumroundover0-30-45.png
Dave, I think you missed my point.
It wouldn't be the first time. :)
John explained it again, which was that Earl felt that a system radiating in 3D space wasn't min phase even if it was observed to be min phase at each observation point.
That has been his position, yes. On that, it's no different than when we see the difference in diffraction effects on a flat baffle that are MP at any measurement point, but that cannot be EQ'ed to a "uniform response", if you will. There is a bit of semantics involved.
I believe that if one feels MP is important, then this is necessary only for first arrival. Is Earl saying that the first arrival of a horn in non MP without the foam to control HOMs?
He seems to make it more nuanced. The "primary" wave is MP, the "HOMs" are not. The problem with this is that, to paraphrase, measuring them so that they can be distinguished is difficult. In fact, I don't think there's a way to do that. I believe he's basing all of his descriptions from his mathematical theory on OS waveguides. On that he says that the wave guide with lowest possible amount of HOMs is the OS waveguide. I think that I have described this correctly. The HOMs still exist with the foam, they are attenuated with it, more so that the primary wave after EQ.
Or is he saying that it is min phase in a single point but not integrated over 3D (ie power) which I think doesn't matter given the room contribution and multiple surface reflections.
This is, I think, his fundamental position. I would say that there are two power responses that have to be considered. There is the power response as radiated by the driver, waveguide here, integrated over a sphere enclosing the driver. This is the anechoic power response. The room power response is simply the driver power response modified by the room conditions. The two are separate. In a perfectly anechoic environment, the power response would be moot vis-a-vis perception, there would only be first arrival. But the power radiated by the driver would be the same as it would be in a non-anechoic environment.
It is trivial to determine if a horn is MP in the direct arrival, with or without the HOM-busting foam. Just measure its frequency response at a point in space, calculate excess phase and Min Phase. If excess phase is constant slope over frequency (ie flat excess delay), then it's Min Phase. If it's not, then its not MP.
Why all the mystery surrounding this? One measurement will tell the tale. Anyone have a horn available that they can measure frequency response on?
Dave
I think that it has to do with his term HOMs. They are low in level, possibly to the point that a measurement might not make them distinguishable. But he believes that they are sufficiently audible. This, though, is based on his studies. The problem is that, to my knowledge, there are not any other studies in support. Much is based on his OS waveguide theory.
All of this assumes that I'm properly interpreting and presenting what he's said.
dlr
You'd think so, but that is not the case. The curves are almost as smooth in the guide as they are on the very large baffle. In fact, with just a small offset angle, the responses are even smoother than the on-axis response, and show remarkable consistency as you move well off axis.
It's not shown here, but I recently ran a set of plots, and at 60deg, only the top octave is showing a roll off.
http://i12.photobucket.com/albums/a203/pete_schumacher/shallowellipticalnothroatwithmediumroundover0-30-45.png
I disagree. I'd say that it's (very roughly) controlled down to 2K or so. What is also part of a design is the coverage angle as well as the frequency. Remember that the PE "horn" is likely not a very precisely designed one. The mouth termination has an effect at the low end as does the throat higher up. One unknown fact is the effect of the "source" being a dome, whereas in a typical horn or waveguide the source is a compression driver. The wavefront from a dome in a PE waveguide is quite different altogether.
In my experiments with the PE horn, the throat was very critical. One day I'll probably put up a page of these experiments. I've just let it all sit unused for a couple of years.
dlr
Pete Schumacher ®
11-25-2011, 08:43 PM
I disagree. I'd say that it's (very roughly) controlled down to 2K or so. What is also part of a design is the coverage angle as well as the frequency. Remember that the PE "horn" is likely not a very precisely designed one. The mouth termination has an effect at the low end as does the throat higher up. One unknown fact is the effect of the "source" being a dome, whereas in a typical horn or waveguide the source is a compression driver. The wavefront from a dome in a PE waveguide is quite different altogether.
In my experiments with the PE horn, the throat was very critical. One day I'll probably put up a page of these experiments. I've just let it all sit unused for a couple of years.
dlr
This is not the PE 8" waveguide. It's not a conical guide by any means, but roughly an 8" radius profile, 2" deep. There is no "flat" section to the guide, but a continuous curve. There's also no "throat" section to speak of.
http://www.vaporsound.com/wp-content/uploads/2011/11/Aurora2.jpg
It wouldn't be the first time. :)
He seems to make it more nuanced. The "primary" wave is MP, the "HOMs" are not. The problem with this is that, to paraphrase, measuring them so that they can be distinguished is difficult.
Hey Dave,
I don't think it matters. The direct signal (ie the total combined first arrival signal at the listening spot), is min phase or its not.
Why does it matter if the horn first arrival is min phase or not? Think about this a bit more deeply. TP designs are min phase. Typical multiway loudspeakers are not. I have yet to see compelling evidence that TP systems have any inherent sonic advantage. I originally got into speakers a long time ago because I was fascinated with the question of group delay audibility. My 4th year EE thesis was on this and I went to do my masters inspired on this topic. I've studied it to death since the 80s. I've yet to see compelling evidence it matters.
So, the argument seems really weak on two counts, if this is indeed the argument (I don't know firsthand, my comments are based on your interpretation of Earl's argument):
1. If the technique provides MP integrated over a sphere, I don't see the benefit when a non MP room makes all but the direct arrival non MP
2. If Earl is saying the technique makes the direct arrival of the tweeter MP when a regular horn is not, is this really a benefit? Consider two things
a. the horn is still being combined with a woofer that results in a non MP system response, just like any other multiway loudspeaker
b. There is no compelling evidence that MP results in better or more accurate sound over non MP
Earl asks us to believe that typical speaker distortion is not audible. If your interpretation is correct, he's asking for quite the leap of faith to believe that MP correction of a typical non MP horn is audible, especially when combined with the woofer, the system is still non MP!
Dave
This is not the PE 8" waveguide. It's not a conical guide by any means, but roughly an 8" radius profile, 2" deep. There is no "flat" section to the guide, but a continuous curve. There's also no "throat" section to speak of.
http://www.vaporsound.com/wp-content/uploads/2011/11/Aurora2.jpg
There is a throat, it's the junction of the "horn" with the dome perimeter. The profile certainly makes it difficult to say what the effective mouth diameter is. Since it's not a typical horn such as the PE one, I don't know how you'd classify it's diameter. It's a bit like trying to say how wide a baffle is that is 6" with a 3" roundover. Do you call it a 6" baffle with roundovers or a 12" baffle? The measured results are neither that of a 6" baffle of course, nor a flat 12" baffle. Compare a circular roundover bit cut to some other profile cut and the response will be different, though the baffle "width" will be the same.
No doubt the long transition at the mouth is a benefit for the smoothness in response. The "mouth" also appears to be a good match for the particular dome used. But it's not a classic horn which was what I was referring to. That one's a lot like the DXT.
dlr
Pete Schumacher ®
11-25-2011, 10:08 PM
There is a throat, it's the junction of the "horn" with the dome perimeter. The profile certainly makes it difficult to say what the effective mouth diameter is. Since it's not a typical horn such as the PE one, I don't know how you'd classify it's diameter. It's a bit like trying to say how wide a baffle is that is 6" with a 3" roundover. Do you call it a 6" baffle with roundovers or a 12" baffle? The measured results are neither that of a 6" baffle of course, nor a flat 12" baffle. Compare a circular roundover bit cut to some other profile cut and the response will be different, though the baffle "width" will be the same.
No doubt the long transition at the mouth is a benefit for the smoothness in response. The "mouth" also appears to be a good match for the particular dome used. But it's not a classic horn which was what I was referring to. That one's a lot like the DXT.
dlr
It's actually more classically horn shaped than the conical waveguides used today, such as Earl's.
It is like having an 8" radius roundover, starting at the tweeter flange/dome interface at roughly 45deg, but flared forward instead of back. What surprised me when gathering data on it was the uniformity of the off axis, along with the smoothness of response. And now, I've tried the aluminum dome RS28 in there, and it behaves identically to the silk dome. Perhaps this particular shape compliments the off axis character of a dome tweeter particularly well.
When we attempted similar measurements with more typical conical CD guides, there were all kinds of issues with on axis anomalies (non min phase?).
Like this
http://techtalk.parts-express.com/attachment.php?attachmentid=16213&stc=1&d=1307074548
And this
http://techtalk.parts-express.com/attachment.php?attachmentid=16250&stc=1&d=1307288724
And this
http://techtalk.parts-express.com/attachment.php?attachmentid=16251&stc=1&d=1307289804
Here's the thread where these plots came from.
http://techtalk.parts-express.com/showthread.php?t=223719
It's actually more classically horn shaped than the conical waveguides used today, such as Earl's.
It is like having an 8" radius roundover, starting at the tweeter flange/dome interface at roughly 45deg, but flared forward instead of back. What surprised me when gathering data on it was the uniformity of the off axis, along with the smoothness of response. And now, I've tried the aluminum dome RS28 in there, and it behaves identically to the silk dome. Perhaps this particular shape compliments the off axis character of a dome tweeter particularly well.
When we attempted similar measurements with more typical conical CD guides, there were all kinds of issues with on axis anomalies (non min phase?).
I doubt that any of it is non-MP. What we're typical able to measure would, if I read Geddes right, include the HOMs, but they are of such a small amplitude that we can distinguish them within the measurements that we can make. I'm not sure what to make of that, however.
The anomalies above about 10K in the PE horn (my only personal experience outside of the DXT and MDT-37) are strictly related to the dome/mouth junction and some small distance up. I was able to nearly eliminate this with some select damping, though it's hard to duplicate reliably the way I did it.
It's also not surprising that anomalies are primarily an on-axis issue, since the on-axis is the axis of 360 degree symmetry and FR changes at 10K and above require rather small changes in distance. It's an identical situation with the surround of a driver, just differing scale. I found the same thing with the PE horn and the tweeter I used. Change the tweeter dome interface at the mouth and the anomalies will change. Just change the dome geometry and you'll likely see a change in response, even if both diaphragms are flat on-axis on an infinite baffle. The horn, any horn, is highly subject to the wave front shape of the source. This is why, I'm sure, they use compression drivers to try to achieve close to a flat wave front at the throat entrance.
dlr
Pete Schumacher ®
11-25-2011, 10:59 PM
I doubt that any of it is non-MP. What we're typical able to measure would, if I read Geddes right, include the HOMs, but they are of such a small amplitude that we can distinguish them within the measurements that we can make. I'm not sure what to make of that, however.
The anomalies above about 10K in the PE horn (my only personal experience outside of the DXT and MDT-37) are strictly related to the dome/mouth junction and some small distance up. I was able to nearly eliminate this with some select damping, though it's hard to duplicate reliably the way I did it.
It's also not surprising that anomalies are primarily an on-axis issue, since the on-axis is the axis of 360 degree symmetry and FR changes at 10K and above require rather small changes in distance. It's an identical situation with the surround of a driver, just differing scale. I found the same thing with the PE horn and the tweeter I used. Change the tweeter dome interface at the mouth and the anomalies will change. Just change the dome geometry and you'll likely see a change in response, even if both diaphragms are flat on-axis on an infinite baffle. The horn, any horn, is highly subject to the wave front shape of the source. This is why, I'm sure, they use compression drivers to try to achieve close to a flat wave front at the mouth entrance.
dlr
The A and F version domes are different shape, even though they are the same diameter, and we all know that fabric behaves differently than aluminum with surface modes.
It is definitely an interesting topic. And I'm looking forward to trying out different domes in this particular "guide" and determine if it is accommodating to more than just the RS28.
daryl
11-26-2011, 04:43 AM
Feyz you're right reflections are not always minimum phase.
I thought the null was the big deal when reflections were equal and opposite the main signal but it is when reflections EXCEED the main signal that excess phase suddenly appears out of nowhere.
Below in the impulse chart the main pulse is the green one while two reflections (red and blue) are delayed by 10mS.
The red reflection has a magnitude 2db less than the main signal while the blue reflections magnitude is 2db greater.
The magnitude, phase and group delay charts show the frequency domain transfer characteristic of the interference between the main signal and reflections.
The black / gray traces are for the green pulse (main) with the red reflection (-2db).
The teal / light teal traces are for the green pulse (main) with the blue reflection (+2db).
Once the reflections magnitude exceeds that of the main signal it's phase (and group delay) near the nulls heads the opposite of what it was doing when the reflections magnitude was less than that of the main.
The magnitude vs. frequency is identical for both systems except that the system with the +2db reflection is a constant 2db higher than the system with the -2db reflection.
Since both have the same magnitude vs. frequency both will have the same minimum phase which is shown in the gray trace for the system with the -2db reflection.
The system with the +2db reflection has the same magnitude response (with a 2db offset) as the other but also has an additional 360 degrees excess phase every 100hz making it definately NOT minimum phase and could only be corrected with a DSP system.
http://i1189.photobucket.com/albums/z427/darylpatterson/MP.gif
Here it is again with the reflections only +/- 0.1db, shows how identical the signals are except at the nulls.
http://i1189.photobucket.com/albums/z427/darylpatterson/MP2.gif
johnk...
11-26-2011, 08:24 AM
I think it would be of benefit to look at the section on High Modes in ducts in Morse and Ingard, chapter 9, section 2. The thing is that HOMs travel through a duct at speeds greater than c in free space. Thus one might conclude that since the wave speed of different modes (phase velocity) is different that the net response may not be MP(?). I don't know that this would be a correct conclusion.
TN Allen
11-26-2011, 10:34 AM
It would be great to see some waveguide designs based upon this discussion. I've set up to turn the Zaph Audio waveguide and another very similar to Mr. Schumacher's waveguide, and I can send a few of these out for others to test.
I'm willing to do other configurations if anyone can send me sectional views or solid model geometry from which I can develop a tool path to cut more complex waveguides on a CNC machine, the idea being to share these with anyone who can evaluate these and produce data to post.
Perhaps a uniform mounting system and dimensions would make it possible to set the waveguides in either cabinets or test baffles?
I think it would be of benefit to look at the section on High Modes in ducts in Morse and Ingard, chapter 9, section 2. The thing is that HOMs travel through a duct at speeds greater than c in free space. Thus one might conclude that since the wave speed of different modes (phase velocity) is different that the net response may not be MP(?). I don't know that this would be a correct conclusion.
I haven't followed Earl's HOM theory but reading Morse and Ingard ch 9 section 2, is Earl saying the throat acts like a duct at higher frequencies causing sound waves to bounce back and forth off the walls? If this is the case, it must apply to only very long throats or very high frequencies. In that case it's understandable that its non MP at very high frequencies as M&I show that the phase velocity increases with frequency for the wavefronts that bounce back and forth down the duct (ie throat). But given the dimensions and frequencies involved, wouldn't this only be true at very high frequencies?
Based on the equations shown in M&I, the ideal foam would have absorption ratio low in the axial direction but high tangentially in Earl's application.
But again, I wonder at what frequency this matters and why is it considered audible, outside the effect on frequency response?
But again, I wonder at what frequency this matters and why is it considered audible, outside the effect on frequency response?
If I understand correctly, Geddes bases this entirely on some studies of audibility that he and his wife conducted.
dlr
johnk...
11-26-2011, 04:43 PM
Well the thing is that the audible differences, IMO, are those associated with response aberrations resulting form HOMs. Not necessarily MP or non MP. Also, in the discussion of Earl's wave guides the discussion ultimately came down to that the improvement in "horn sound" came from the damping, not the OS form of the guide. Obviously since the HOMs are of much lower amplitude and travel in tangential directions the damping is more effective on them than on the direct sound. So, the result would be, with regard to horns and wave guides, just stuff it.
When you think about this it isn't all that different than Wilson's use of open celled foam for grills some of his speakers (Watt Puppy comes to mind).
Powered by vBulletin® Version 4.2.0 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.