MethMan said:
Hi Mr. Meth (that has a funny ring to it!)
Correct for the most part and I was aware that the post had some caveats that might need explaination. First: the OS IS that singular device that creates the minimum diffraction for a plane wave at its circular entrance aperature. That much is correct. But there is still some diffraction, there always will be (I love how people turn "some diffraction" into "it diffracts just like everything else" - there is a lot of spining going on out there on this subject. None of it mine I hope.)
Now, it is possible to feed an OS with a wavefront that will completely surpress ALL diffraction - at any GIVEN FREQUENCY. This specific wavefront changes with frequency however. Now could a phase plug and driver be made that tracked the required wavefront with frequency? It is theoretically possible, but implimentation would be difficult. Could the phase plugs in drivers be improved in this context - an approximate solution - absolutely (I show how in one of my patents on phase plugs.)
The topic is incedibly complex and certainly the reason that there is so much misinformation and misunderstanding about it. I understand it and I'd be happy to answer any questions that you might have, but I will not argue to points with someone who has not studied them thouroughly.
MethMan said:
The subject of difraction is very complex. I could give you many references on it, but I'd need to know the level of math that you are comfortable with. At a low level of math there is not going to be much since the topic is not an easy one.
Let me give you some "rules of thumb"
1) the dffraction is proportional to sounce intensity - it is linear
2) The sharper the edge the more diffraction. It is actually proportional to the 2nd derivative of the surface. A "knife edge" therefor is a worst case.
3) The higher the frequency for a given edge the greater the diffraction intensity.
Hi earl:
Based upon your comments on diffraction, would you say that, after the throat expands smoothly into the conical section the wall geometry becomes less important? So, if the section were not fully circular, but only approximately , and had smooth walls, the performance would be largely the same up until fairly high-freq content at the edges of the pattern.
I'm also thinking about the foam plug- wouldn't this tend to damp any diffraction due to rough guide surfaces (within reason)
Essentially, if I have it right, small deviations from the profile, and small surface irregularities after the throat, and with the plug in place, would be largely irrelevant to performance. The throat is where the vast majority of diffraction is dealt with and the requirement of precision less high with the rest of the profile.
To be specific, I'm playing with molding them out of plaster and cloth, based upon a soft (flexible) foam mold with a higher precision throat mold attached. The flexible mold is nice because it comes free easily and is easy to fabricate, and deforms lightly around the base, expanding to match the flare-out in the profile. But because it's flexible, the weight leads to a small amound of deformation of the conical section, though the cross-sectional area remains constant.
After casting, I have a very smooth and precise throat, as it's wrapped tightly around a stronger mold section, and a "near" conical section with a flare, and some irregularities at the surface to be filled/sanded/repeated.
I will say, the rough take (still refining my technique, here!!) certainly passes the "doesn't mess up voices" test, where one speaks through the waveguide. This made voices a little breathier, but apart from tonal imbalance, didn't make any 'honky' sound, or any of the classical 'horn nasties'.
Based upon your comments on diffraction, would you say that, after the throat expands smoothly into the conical section the wall geometry becomes less important? So, if the section were not fully circular, but only approximately , and had smooth walls, the performance would be largely the same up until fairly high-freq content at the edges of the pattern.
I'm also thinking about the foam plug- wouldn't this tend to damp any diffraction due to rough guide surfaces (within reason)
Essentially, if I have it right, small deviations from the profile, and small surface irregularities after the throat, and with the plug in place, would be largely irrelevant to performance. The throat is where the vast majority of diffraction is dealt with and the requirement of precision less high with the rest of the profile.
To be specific, I'm playing with molding them out of plaster and cloth, based upon a soft (flexible) foam mold with a higher precision throat mold attached. The flexible mold is nice because it comes free easily and is easy to fabricate, and deforms lightly around the base, expanding to match the flare-out in the profile. But because it's flexible, the weight leads to a small amound of deformation of the conical section, though the cross-sectional area remains constant.
After casting, I have a very smooth and precise throat, as it's wrapped tightly around a stronger mold section, and a "near" conical section with a flare, and some irregularities at the surface to be filled/sanded/repeated.
I will say, the rough take (still refining my technique, here!!) certainly passes the "doesn't mess up voices" test, where one speaks through the waveguide. This made voices a little breathier, but apart from tonal imbalance, didn't make any 'honky' sound, or any of the classical 'horn nasties'.
Badman
Yes, your understanding is correct. (Not too hard when you "read to learn" instead of "disrupt".) I have to be careful with some IP, but I would not strongly disagree with anything that you said. Your understanding is basically correct.
Mathematically speaking, the modes in an OS waveguide ALL become identical to the spherical modes in a Conical waveguide. This means that, since we know that the Conical waveguide does not generate HOM through modal coupling, that no more HOM are created after the point where the OS wave functions match the Conical ones, which one can pretty easily guess might occur when the OS contour closely matchs the Conical one.
However, the above is not quite so simple, because the higher the order of the mode, the further away from the throat it is before it matches the conical wave functions. What this means is the higher the order of the mode the more sensitive its going to be to small perturbations for a longer length along the device. This makes it very complicated to make a simple statement like "once the OS contour matches the Conical one there is no more diffraction". It all depends on the HOM content at the throat to begin with.
Yes, your understanding is correct. (Not too hard when you "read to learn" instead of "disrupt".) I have to be careful with some IP, but I would not strongly disagree with anything that you said. Your understanding is basically correct.
Mathematically speaking, the modes in an OS waveguide ALL become identical to the spherical modes in a Conical waveguide. This means that, since we know that the Conical waveguide does not generate HOM through modal coupling, that no more HOM are created after the point where the OS wave functions match the Conical ones, which one can pretty easily guess might occur when the OS contour closely matchs the Conical one.
However, the above is not quite so simple, because the higher the order of the mode, the further away from the throat it is before it matches the conical wave functions. What this means is the higher the order of the mode the more sensitive its going to be to small perturbations for a longer length along the device. This makes it very complicated to make a simple statement like "once the OS contour matches the Conical one there is no more diffraction". It all depends on the HOM content at the throat to begin with.
gedlee said:
I take it this is in reference to some of our previous disagreements. I think you'll find I'm fairly agreeable most of the time, much of what we've argued on has been driven by the poor quality of some of your early deliverables, and your extreme level of confidence in your measures and methods. I take big issue with premium pricing on substandard product and we have philosophical differences. You like to optimize to "good enough", with an eye towards cost controls, and I believe that "good enough" never is, within the context of this hobby.
If in these discussions we can stay within the discussions on hand rather than referencing any past argumentation, I think it would be best for our communications.
BTW: I know I'm abrasive. 😉
gedlee said:
So, then, can we take it a bit further? It seems to me that the most important component would be a smooth transition into a conical profile, rather than the specific contour of that transition. A gently rounded transition, relatively short, and matching the conical waveguide section to the driver throat section seem to be the most critical component of the profile. Getting a high degree of precise roundness here, with a very smooth surface and as gentle a transition as possible, seems like the most critical component.
With a 90 degree conical section, you're able to do that, it'd be even easier with a smaller dispersion, and require a much larger 'transition' section with a wide-profile guide.
Have you tested the difference between a precise profile in the throat and a close approximation (of equivalent smoothness)? I'd suspect the difference to be minimal.
You misunderstood. The reference was never to you, but others who want to argue about pointless issues and muddy the really important discussions, like your question. I never put you together with any previous arguments - I don't hold grudges. At any rate, no one complains about the kits anymore - minor parts mis-count, dents from shiping, thats about it. That the problems get fixed is all that can be expected. As to the "premium pricing" of my speakers ... 
all I can say is that we view this differently because I see them as a real bargain.
No I have not tested for the sensitivity of the throat tollerence accept as regards the match of the throat to the driver. Here I found some very measureable differences. There would be no point in my testing "almost OS" because I don't do "almost OS".
all I can say is that we view this differently because I see them as a real bargain.No I have not tested for the sensitivity of the throat tollerence accept as regards the match of the throat to the driver. Here I found some very measureable differences. There would be no point in my testing "almost OS" because I don't do "almost OS".
gedlee said:You misunderstood. The reference was never to you, but others who want to argue about pointless issues and muddy the really important discussions, like your question. I never put you together with any previous arguments - I don't hold grudges. At any rate, no one complains about the kits anymore - minor parts mis-count, dents from shiping, thats about it. That the problems get fixed is all that can be expected. As to the "premium pricing" of my speakers ...
No I have not tested for the sensitivity of the throat tollerence accept as regards the match of the throat to the driver. Here I found some very measureable differences. There would be no point in my testing "almost OS" because I don't do "almost OS".
Fair enough. It seemed strange and out of place. I only bring up the past arguments because it seemed so odd, like a grudge (that, to your credit, I'd previously not thought you held). And you're right, you made right on the early kits, from what I tracked, so I certainly have no issue with that anymore, just wondering where the comment came from. Thanks for clarification.
As far as "Almost OS", fair enough. My throat profile is a close, but not 'perfect' match for the OS, but is very smooth and precise in its contour, and some sanding around and about the throat brings it pretty darned close. We'll see how they perform, though I don't have nearly the measurement setup I'd need to duplicate some of your test regimen.
Do you have a set of throat angle measures on different compression drivers? My initial plan was to use some JBL 2426H on these, since they're what I have on hand. It's still some weeks out. I figure I can do some final "throat matching" contouring with careful sanding and your clay method.
badman said:
No, I don't have a lot of drivers measured. From what I have done, there is not a large difference, the waveguide tends to dominate the response and the driver differences show up mostly at the high end because of diaphragm breakup. In some drivers there is a voice coil leak "hole" (like the TAD drivers) and this always shows up, but for similarly constructed drivers like the JBLs, Beymas, B&Cs, etc. the results are very very similar.
Sorry, I edited as you were responding.
I was specifically asking about the throat angle measure, not the on/off axis content.
The JBL is likely to show more top end rolloff than the B&C because it's highly damped. I'll probably be using supertweeters in conjunction with them to get the last octave.
I was specifically asking about the throat angle measure, not the on/off axis content.
The JBL is likely to show more top end rolloff than the B&C because it's highly damped. I'll probably be using supertweeters in conjunction with them to get the last octave.
Throat angle data can be very hard to come by. I find it very interesting that its never published and according to B&C almost nobody but me has ever asked for it. Tells you something about where the mass market is at WRT waveguide/driver matching. I have complained to B&C about the big step in the throat on the DE250. They say that no one has ever commented on this before. It is surprising (and then not really) that so few pay any attention to this point in the driver/waveguide system when I find it to be the most critical.
I've emailed JBL, hopefully I'll hear back from them.
Unrelated- an elliptical OS is just a 'squashed' conical, is that correct? My particular casting method actually makes an elliptical every bit as easy as a conical. I have to retrofit the mouth roundouts on either geometry using styrofoam forms and TLC anyway.
Unrelated- an elliptical OS is just a 'squashed' conical, is that correct? My particular casting method actually makes an elliptical every bit as easy as a conical. I have to retrofit the mouth roundouts on either geometry using styrofoam forms and TLC anyway.
badman said:
Sorry, but the correct way to make an elliptical has to remain my trade secret. There was some previous discussions here not long ago and they were pretty good, but I can't say if they were what I do or not. I can't give away ALL of my secrets now can I!?
Well, I'll post what I did (and what someone else posted I believe in another thread). Plot an OS profile for the horizontal and vertical planes of the waveguide. Draw ellipses that go through each of the OS profiles at various points. Connect the ellipses together smoothly. You can see this here:
An externally hosted image should be here but it was not working when we last tested it.
gedlee said:
Speaking of trade secrets, I did some polar measurements of an OS waveguide and an XT1086 that I cloned, and was surprised that there was such a difference, considering how similar the curves are.
Then again, I did the measurements in my car, so ymmv.
But it's still curious that two similar waveguides could behave so differently.
" the dffraction is proportional to sounce intensity - it is linear"
I must be misunderstanding, but doesn't that mean that directivity will vary with level?
I must be misunderstanding, but doesn't that mean that directivity will vary with level?
noah katz said:
I think that what you are missing is that the diffraction field increases at the exact same rate as the source field. Hence the sum of the two fields is independent of level. Understand?
Patrick Bateman said:
I honestly don't know how sensitive the contours are to the curve. I have never seen any point in not just doing the OS exactly so I have no experince with approximations.
gedlee said:
The strange thing is that the XT1086 changes coverage angles slower than the OS curve. The coverage angle in an OS waveguide changes from six degrees to ninety degrees in the span of 1/2". The XT1086 changes it's coverage angle very VERY slowly.
I would think that this wouldn't be a bad thing, but the polars are inferior.
Having said that, the data is a bit suspect, due to where it was measured.
I'll take a picture of an elliptical OS waveguide and the XT1086 and post it, it will make more sense then.
gedlee said:
Not in terms of the way I normally think of diffraction, which is as a local change in the direction of the wavefront.
However you think of diffraction, it is a linear effect and the results cannot change with the source level. As a percentage of the source level the diffraction is constant, but it IS NOT constant with source level.