Good question. I suppose they haven't been mentioned because this thread is aimed at more traditional deep expo horns rather than the shallow wave guide types. But I don't see why driver to WG matching would not also be important. Perhaps someone will let us know. @marco_gea
The Geddes style is an axisymmetrical shallow conical horn or waveguide with an OS flare rate adaptation at its throat. One might suppose that the CD internal flare should match the expansion rate of the conical WG, but that OS throat could change things. I simply don't know.
The Geddes style is an axisymmetrical shallow conical horn or waveguide with an OS flare rate adaptation at its throat. One might suppose that the CD internal flare should match the expansion rate of the conical WG, but that OS throat could change things. I simply don't know.
Noteworthy post IronmanIV, thank you.
In addition, where horn expansion creates diffractions which take their own path, would there have been such a thing as a perfect entry? Is it open to interpretation?
This appears to be an assumption, and I see a lot of it. Correct me if I'm wrong, but wouldn't the instantaneous conditions be more relevant as one moves down the horn's axis, rather than the expansion seen over that distance? It brings up questions like.. how does the wavefront know what's ahead?
In addition, where horn expansion creates diffractions which take their own path, would there have been such a thing as a perfect entry? Is it open to interpretation?
Earl Geddes OS throat was approximated (appropriated..) by Charles Hughes for Peavey:
https://peaveycommercialaudio.com/w...atic-Throat-Waveguide-by-Charles-E-Hughes.pdf
Generally, drivers with shallow depth and their phase plug extending near the throat are used with this type of waveguide, and the throat arc can be adjusted to match any entrance angle within the usual range.
That said, using longer, small angle expansions like the old Western Electric, Altec, JBL, TAD etc. with flare constants in the 150-200Hz can increase the low end output generally lacking with a largely conical flare, without a negative impact on dispersion below the throat diameter wavelength.
A "perfect entry" simply avoids a rough diffraction causing transition.
The big question separate from the driver exit angle is whether it's phase plug ends up producing more of a (hemi-) spherical or plane wavefront, and whether the horn was designed using an assumption of either.
One can't even make an assumption based only on the phase plug design, as different diaphragm materials have different breakup modes which vary the effective "time of flight" through the inner and outer plug channels.
Art
Last edited:
This has been analyzed to some depth in one of my "papers" - http://www.at-horns.eu/release/R-OSSE Waveguide rev7.pdf#page=21 (page 21).
Today it's no problem to take a BEM solver at look at all of this systematically. One "issue" is that in the end it doesn't always behave as one would think - see e.g. http://www.at-horns.eu/exar-story.html
Today it's no problem to take a BEM solver at look at all of this systematically. One "issue" is that in the end it doesn't always behave as one would think - see e.g. http://www.at-horns.eu/exar-story.html
I'm glad I was able to assemble something coherent or discussion provoking. I think Geddes has said the "instantaneous conditions" are the most important or the only important thing when determining diffraction or HOM issues and that all other loading and gain predictions can be made by using the lumped impedance models of the throat size and horn mouth size.
I am very curious about the BEM solver approach. To bring up Geddes again, he has said that no further work should be done trying to build complex math models for more complicated to model horns, for example biradial, since the math really is complex and BEM/FEA is the way this will be done. For the diyer, this presents a problem though since I'm not sure the BEM tools are ready for us plunkers and we still need good guidance on using easy models, or analytical approaches like the Hughes paper., which yes, does look a bit derivative of Geddes, but I think much easier to use.
I think the BEM approach is really what we're going to need to embrace though, and it requires a different kind of discussion. It's going to be hard, I think, for mabat to generalize his findings in the ways we are used to asking for simple design heuristics. We'll need to look at the graphical plots of results, follow the links to the report. When he says "doesn't always behave as one would think", I think we should extend that to say, we should not even try to predict (think) the results based on our old heuristics since the BEM simulation is so effective.
Thanks so much for the considered replies, this was very helpful to me to understand the current consensus in this group, which I think is the deepest dive forum on horns anywhere. It is such a treasure to be able follow these discussions as much as 16 years back in time. I think it may be time to try to acquire and understand the BEM tools although I have regrets about not becoming more capable in Hornresp.
Cheers,
Jamie
I think the BEM approach is really what we're going to need to embrace though, and it requires a different kind of discussion. It's going to be hard, I think, for mabat to generalize his findings in the ways we are used to asking for simple design heuristics. We'll need to look at the graphical plots of results, follow the links to the report. When he says "doesn't always behave as one would think", I think we should extend that to say, we should not even try to predict (think) the results based on our old heuristics since the BEM simulation is so effective.
Thanks so much for the considered replies, this was very helpful to me to understand the current consensus in this group, which I think is the deepest dive forum on horns anywhere. It is such a treasure to be able follow these discussions as much as 16 years back in time. I think it may be time to try to acquire and understand the BEM tools although I have regrets about not becoming more capable in Hornresp.
Cheers,
Jamie
Mabat, I'm very confused by your paper. Your Figure 12 shows the throat impedance of two different waveguides, one with a throat exit angle of the driver at 0 degrees and one with an angle of 15 degrees. The impedance curves are quiet different for the two throat angles, which is unexpected for me.
And, Figure 16 shows the throat impedance plots for two different waveguides, one with a larger mouth and one with a small mouth. A large difference in mouth size. But, the impedance curves are almost exactly the same. This is unexpected to me also, and I see that the response graph, Fig. 14 does show a difference in response, as I would expect.
If Fig. 14 shows a response difference near the cutoff region, an acoustic impedance dominated region, doesn't the impedance plot need to show a difference also?
And, Figure 16 shows the throat impedance plots for two different waveguides, one with a larger mouth and one with a small mouth. A large difference in mouth size. But, the impedance curves are almost exactly the same. This is unexpected to me also, and I see that the response graph, Fig. 14 does show a difference in response, as I would expect.
If Fig. 14 shows a response difference near the cutoff region, an acoustic impedance dominated region, doesn't the impedance plot need to show a difference also?
I don't know what to add, these are simply not very intuitive concepts, but all you mention is just true - so are the figures in the paper. I still have more questions than answers myself.
An example: Quite recently, I would tend to think that a one smooth curve (with continuous curvature) as a horn profile is desirable, starting right at the phase plug exit, but now I'm not so sure even about that anymore. The quadratic throat shown above has a large curvature discontinuity at the point where the arc connects to the straight wall, causing a reflection. Is this bad in the end? I don't know anymore. I'm only pretty sure it wasn't meant as an advantage.
An example: Quite recently, I would tend to think that a one smooth curve (with continuous curvature) as a horn profile is desirable, starting right at the phase plug exit, but now I'm not so sure even about that anymore. The quadratic throat shown above has a large curvature discontinuity at the point where the arc connects to the straight wall, causing a reflection. Is this bad in the end? I don't know anymore. I'm only pretty sure it wasn't meant as an advantage.
Last edited:
mabat, thank you. I have also been struggling with these geometry questions about that short throat section. Even though Geddes states clearly, or emphatically as an editorial comment to his scientific work, that the acoustic impedance is predominantly, or I think he even says solely determined by the mouth and throat diameters by his ray acoustic math, this short straight or exponential expansion section at the throat of the horn appears throughout designs for diffraction optimized horn curves. We see this in Don Keele's designs for EV in his paper "What's So Sacred About Exponential Horns?" where he explicitly addresses the cutoff loading improvement of having a short exponential expansion section prior to the waveguide section and then the mouth termination profile is also explicitly addressed as a separate section of more rapid flare. I think others like Kohlbrek have also shown that this short section improves cutoff loading and Kohlbrek even points to the deep internal throat of CD's like the Altec 288 as having a beneficial influence on use for LF horns. The paper by Hughes about the Peavey horn also has this short section without a solid acoustic justification except something about the acoustic source at the vertex. It appears to be a kind of acoustic loading patch to constant directivity expansions.
It has me coming back around to the Le Cleach approach where a hyper exponential expansion of some value of T = .7 then has the Le Cleach iterative math applied to it to get an expansion that keeps the wave front normal to the sidewalls all the way past the mouth. A designer can then simulate the tradeoffs of loading, directivity and diffraction all in one operation simultaneously in Hornresp. For example, if a hyper exponential expansion uses a value T = infinity, it becomes a conical horn, which looks a lot like a waveguide to me. With the Le Cleach expansion transform math as applied in Hornresp to a hyper exponential expansion model you can essentially dial up or down constant directivity geometry as much or as little as you like and get a comprehensive axisymmetric profile including throat and mouth termination and simulation. True? That's a nice escape hatch from a what's best discussion about pure profile categories.
I'm also curious, does the BEM simulation approach and your software allow for using any mesh model shape for the horn? You used a parametrically defined geometry, is that required by the BEM software?
Cheers,
Jamie
It has me coming back around to the Le Cleach approach where a hyper exponential expansion of some value of T = .7 then has the Le Cleach iterative math applied to it to get an expansion that keeps the wave front normal to the sidewalls all the way past the mouth. A designer can then simulate the tradeoffs of loading, directivity and diffraction all in one operation simultaneously in Hornresp. For example, if a hyper exponential expansion uses a value T = infinity, it becomes a conical horn, which looks a lot like a waveguide to me. With the Le Cleach expansion transform math as applied in Hornresp to a hyper exponential expansion model you can essentially dial up or down constant directivity geometry as much or as little as you like and get a comprehensive axisymmetric profile including throat and mouth termination and simulation. True? That's a nice escape hatch from a what's best discussion about pure profile categories.
I'm also curious, does the BEM simulation approach and your software allow for using any mesh model shape for the horn? You used a parametrically defined geometry, is that required by the BEM software?
Cheers,
Jamie
Ath cannot produce every horn shape you might want to, but it can certainly do a lot and once learned it is easy to iterate.
ABEC or AKABAK which is the BEM software that works with Ath can use whatever mesh you give it. The quality of the mesh is important and Ath really helps to get that right. Anything else can be simulated if you can model the curve in CAD or through nodes and points.
Like this or many others
https://www.diyaudio.com/community/...ifications-and-bem-simulation-results.382115/
I think this improvement has not much to do with the short section being exponential per se, as it turns out that the exact shape is of little importance. All my experiments indicate that it is the wavefront mismatch, the reflection and resonance that actually helps to increase the throat resistance on the lower end of the passband in this kind of design (although the mechanism itself it still quite obscured to me). What I find important is that this extending section doesn't open too much itself, because then it causes HF beaming. Basically, it can be even cylindrical.
BTW, when talking about wavefront mismatch, we could as well talk about a flare-rate mismatch, as it seems this is actually one and the same thing:
https://www.researchgate.net/public...aping_using_novel_single-parameter_waveguides
Last edited:
Mabat, I don't understand what you mean when you say:
"it is the wavefront mismatch, the reflection and resonance that actually helps to increase the throat resistance on the lower end of the passband in this kind of design"
Shouldn't the reflection and resonance products present themselves as the reactive component of the complex impedance? For purposes of loading, I think we would only expect the real part of the acoustic impedance to increase loading and gain. Any reflections or resonances would show up as the reactive or imaginary component of the acoustic impedance which we might deduce causes problems, but not directly related to LF loading. Right?
But, this is really academic, since if you simulate and measure an improved loading or avoid poor loading, you really don't need to know the cause of that when using BEM or FEA. You can iterate the design by any geometry change you like and then see the result. I see a lot of speculation about the causes of these results which might not be supported by the physics or acoustic theory, and the simulation is agnostic to all these theories and models.
When I read the thread provided by fluid about the BEM simulations of the Yuichi horn, I saw a lot of simulations of impedance and response including directivity, but no modeling of distortion. Are the BEM simulations limited to impedance and directivity?
If they are limited to impedance and response, we may want to recognize that designing a horn is still a very iterative process where empirical testing is paramount for assessing distortion, resonances and coloration of sound. We'd want to really pay attention to folks like Joseph Crowe who are doing good IMD and Gedlee GM measurements of distortion on drivers and horns and the old guys who've made many horns may posses empirical data in their heads that just can't be modeled, yet.
"it is the wavefront mismatch, the reflection and resonance that actually helps to increase the throat resistance on the lower end of the passband in this kind of design"
Shouldn't the reflection and resonance products present themselves as the reactive component of the complex impedance? For purposes of loading, I think we would only expect the real part of the acoustic impedance to increase loading and gain. Any reflections or resonances would show up as the reactive or imaginary component of the acoustic impedance which we might deduce causes problems, but not directly related to LF loading. Right?
But, this is really academic, since if you simulate and measure an improved loading or avoid poor loading, you really don't need to know the cause of that when using BEM or FEA. You can iterate the design by any geometry change you like and then see the result. I see a lot of speculation about the causes of these results which might not be supported by the physics or acoustic theory, and the simulation is agnostic to all these theories and models.
When I read the thread provided by fluid about the BEM simulations of the Yuichi horn, I saw a lot of simulations of impedance and response including directivity, but no modeling of distortion. Are the BEM simulations limited to impedance and directivity?
If they are limited to impedance and response, we may want to recognize that designing a horn is still a very iterative process where empirical testing is paramount for assessing distortion, resonances and coloration of sound. We'd want to really pay attention to folks like Joseph Crowe who are doing good IMD and Gedlee GM measurements of distortion on drivers and horns and the old guys who've made many horns may posses empirical data in their heads that just can't be modeled, yet.
No, they do increase the real part of the complex impedance as well. I have shown an example in the above links.
As for nonlinear distortion, all the BEM calculations are strictly linear stuff (we don't care at all about absolute levels in these simulations). For practical horns, this is most probably all we need.
Last edited:
Modeling non linearities would require knowledge of the materials and a Finite Element model. I don't think there is a huge amount of benefit to that when designing a waveguide or horn unless you have very detailed simulation parameters for the actual driver going to be used. Distortion increases with excursion and designs that lose radiation impedance at lower frequencies do not support the driver and leave it to fend for itself. If you want to use a driver to lower frequencies the radiation impedance will give you a good idea of how much help the horn will give.
I don't think anyone should suggest that you can simuate and design without the need for actual measurement and listening to decide what is best or preferred for any specific situation. The information that Joseph includes in his reviews may be useful as a comparison but it has to be remembered that the device may well perform quite differently on another horn/waveguide. Joseph's devices are all quite similar to each other at a basic level and are not representative of that many commercial devices that many people want to use.
Thanks Mabat, I see the imaginary or reactive component of acoustic impedance is phase. I think besides some very basic fundamentals, I need to stop trying to figure out the math, unless it's math I can truly calculate. It's very complex and my intuitive leaps are all wrong. I guess this is why we have simulation.
And thanks for the clarification on the state of all simulation currently and I think Fluid's point about the requirement of specificity of analysis is very important. Maybe this actually does come back around to Pano's original proposition for discussion.
What I think I have learned is that the only good way to predict a CD's performance on a horn (unless someone with experience tells us) is to simulate that driver on a prospective horn. The simulation will address both the electrical and mechanical properties of the driver and the acoustic impedance and gain on the horn. As mabat says, the results might be surprising. And, to go out on a limb again, people who have experience can probably infer distortion behaviour from the impedance and response plots even if they are not explicitly modeled.
I do think Joseph Crowe's measurements are more useful than has been characterized here. You many not prefer his horns over others, but they are certainly well enough designed for loading and directivity to compare drivers and I have not seen any other source that provides such extensive distortion analysis for so many drivers all measured in consistently the same way and published in a consistent format. Not only do we have the data, but also the ongoing development of the idea of correlation between distortion measurement and subjective perception of sound quality. It is a tutorial series on measurement AND a database of measurements referenced to a standard horn design.
And, measurements are also published for other horns like the Yuichi 290 performed in exactly the same way as done on the JC horns. So, if you really like the Yuichi and know it, you could compare results. Nice.
I think it is the sheer size of the population of drivers all measured in the same way that is huge value to the DIY community, in my opinion.
Compare this to the Parts Express methods. They don't really do IMD, certainly not IMD at multiple SPL levels progressively increased to show when nonlinear distortion emerges. And, the Parts Express reviews usually use the MFG supplied horn. This means there is no way to directly compare drivers on the same horn to try to isolate driver characteristics from the horn.
If there is anyone else publishing these kinds of measurements except for a few pet projects, I'd be happy to learn about them.
And thanks for the clarification on the state of all simulation currently and I think Fluid's point about the requirement of specificity of analysis is very important. Maybe this actually does come back around to Pano's original proposition for discussion.
What I think I have learned is that the only good way to predict a CD's performance on a horn (unless someone with experience tells us) is to simulate that driver on a prospective horn. The simulation will address both the electrical and mechanical properties of the driver and the acoustic impedance and gain on the horn. As mabat says, the results might be surprising. And, to go out on a limb again, people who have experience can probably infer distortion behaviour from the impedance and response plots even if they are not explicitly modeled.
I do think Joseph Crowe's measurements are more useful than has been characterized here. You many not prefer his horns over others, but they are certainly well enough designed for loading and directivity to compare drivers and I have not seen any other source that provides such extensive distortion analysis for so many drivers all measured in consistently the same way and published in a consistent format. Not only do we have the data, but also the ongoing development of the idea of correlation between distortion measurement and subjective perception of sound quality. It is a tutorial series on measurement AND a database of measurements referenced to a standard horn design.
And, measurements are also published for other horns like the Yuichi 290 performed in exactly the same way as done on the JC horns. So, if you really like the Yuichi and know it, you could compare results. Nice.
I think it is the sheer size of the population of drivers all measured in the same way that is huge value to the DIY community, in my opinion.
Compare this to the Parts Express methods. They don't really do IMD, certainly not IMD at multiple SPL levels progressively increased to show when nonlinear distortion emerges. And, the Parts Express reviews usually use the MFG supplied horn. This means there is no way to directly compare drivers on the same horn to try to isolate driver characteristics from the horn.
If there is anyone else publishing these kinds of measurements except for a few pet projects, I'd be happy to learn about them.
Positive imaginary part corresponds to acoustic mass and negative imaginary part corresponds to acoustic compliance - quite hard to imagine. But it's never without a non-zero real part, i.e. the throat impedance is never purely imaginary.
Both parts (Re and Im) can have high values at the same time. They are in fact tied together somehow, I just don't recall the details now. At a peak of the Re part, the impedance is typically close to being purely real, the same as for Re being constant.
Last edited:
Yeah, this is kind of what I mean by intuitive leaps from the math. Complex numbers which have a real and imaginary part are a mathematical device to allow for SQRT of -1, or for equations that have a vector or magnitude plus direction or angle of rotation of phase component. Our reactive acoustic impedance component may not have a direct translation into a physical or observable acoustic phenomenon. It is a necessary component of complex acoustic impedance computations, but I"m not sure it's useful as a separate value as an intuitive or directly relatable quantity to response or distortion or observable phenomenon.
If I'm looking at an acoustic impedance plot, I'm just thinking I don't want to operate a horn in a range where there are big swings or steep slope changes in Re or where the Im value is large and I am guessing these are tending to be related mathematically just as response, impulse response and phase are related and can be transformed with derivative equation solutions like the La Place or Fourier.
If I'm looking at an acoustic impedance plot, I'm just thinking I don't want to operate a horn in a range where there are big swings or steep slope changes in Re or where the Im value is large and I am guessing these are tending to be related mathematically just as response, impulse response and phase are related and can be transformed with derivative equation solutions like the La Place or Fourier.
Like this?
I can assure you it's absolutely fine in every regard.(It's one and the same device - throat impedance and raw acoustic responses at roughly 0/10/20/30 deg.)
http://www.at-horns.eu/exar-story.html
Last edited: