gedlee said:
Hand waving, I think not. I'm not sure what you have a problem with, the idea of limiting the vertical pattern at HF or my descriptions of how the response curves look.
In any case, here are on-axis and off-axis measurements of a matched-directivity speaker using an asymmetrical horn:
Nulls form vertically off-axis for any loudspeaker with vertically spaced drivers. Even the novel approaches are nothing more than staggering or offset to move the null angles around, an attempt to get them outside the horn's coverage pattern. Danley's whole Unity/Synergy concept is based on that. I always felt it was a different compromise than I would make, but his goals place arrayability at extreme importantance. The point is, I would think most people agree that having the nulls placed outside the horn's pattern would be a good thing.
To me, it always made sense to make directivity-matched loudspeakers with nulls just outside the tweeter's pattern. You're not going to stop the null, but if the horn is directional enough, you may reduce it. Likewise, you can reduce the lobes outside the nulls. And even if the pattern control down low isn't good, by high frequencies it should be so you can prevent high frequencies at large off-axis angles from being much higher in amplitude than the crossover band. That has always been my goal, in addition to matching directivity in the horizontal plane.
Please explain yourself. What exactly do you have a problem with? What don't you like about limiting the HF pattern, keeping it inside the nulls? Do you not agree that it is a desirable goal?
Member
Joined 2003
Of course there are nulls. Never said there wouldn't be. What there is though is HF output in the upper frequencies that is about the same level as the average of the crossover band. Output rises slightly just above crossover because the horn doesn't have vertical pattern control yet. Then it falls back to approximately the same amplitude as the crossover region.
An axisymmetrical horn doesn't do this. There is a dip through the crossover region, followed by a rise back up to same level as before the crossover region. Why would you want that? I prefer the vertical pattern to be reduced at the crossover frequency and stay down. In fact, a larger midrange horn below the tweeter with vertical control that goes lower is even better. But to work with what we have, the directivity-matched two-way loudspeaker, I don't want the vertical pattern to increase at HF.
About my charts, they're unsmoothed. I don't know that I want to "brag", but I will say I'm pleased with it. I'd like to measure your Summas outdoors on an LMS system to see how they compare. The only data I've seen on your speakers was with 1/3rd octave smoothing, and I'm not sure what system you've used but I think it's a homebrew, isn't it?
I can tell you that the LMS system tends to be cruel to most speakers, some so much they don't even want to publish the charts. Perhaps one day, we'll see one of your speakers measured with it. Publish the charts unsmoothed, so the tiny little response details are exposed.
Beyond that, what other data would you want? I'll see if I can't measure it and put it online. I want to measure distortion and impulse response, but I don't think you care about distortion. I do, and in other tests have measured response and distortion starting at some level and then doubling it and measuring again and again until I reach the limits of the device. This gives a good picture of how the speaker acts at all power levels. It shows how parameters shift at high temperature and excursion. Are you interested in that?
That's not true.
Look at my response curves. The ones shown at the null angles show reduced output in the crossover region averaging 6dB. Output above 3kHz is also reduced about the same amount, about 6dB. The horn's pattern roughly equals the null angle, so it will be -6dB at that point.
Maybe we have a misunderstanding. Explain what you mean by saying the data doesn't support my claims. What I claim is the output of the horn at HF is approximately equal to the output of the system through the crossover region.
Look at my response curves. The ones shown at the null angles show reduced output in the crossover region averaging 6dB. Output above 3kHz is also reduced about the same amount, about 6dB. The horn's pattern roughly equals the null angle, so it will be -6dB at that point.
Maybe we have a misunderstanding. Explain what you mean by saying the data doesn't support my claims. What I claim is the output of the horn at HF is approximately equal to the output of the system through the crossover region.
You originally claimed that an axisymmetric device would have a deeper power response (or DI) null through the crossover than one with a narrower vertical pattern. I don't see that. I'm not saying that they don't I'm just saying that your arguement is too simplistic (there are a lot of other factors that you have ignored) and your data is too incomplete to draw any conclusions. You have not proved your contention.
I would need to see complete polar maps out to 90° in both axis. But that still doesn't prove anything unless you do this with an axisymmetric horn of the same basic design as a comparison. Then, if this all works as you claim, I'll accept it.
But as it stands now I only see a weak argument and superficial data that doesn't even seem to agree with your argument
I would need to see complete polar maps out to 90° in both axis. But that still doesn't prove anything unless you do this with an axisymmetric horn of the same basic design as a comparison. Then, if this all works as you claim, I'll accept it.
But as it stands now I only see a weak argument and superficial data that doesn't even seem to agree with your argument
gedlee said:
Then why not use a larger coaxial? They come in sizes all the way up to 18". The main problem is the discontinuities in the formed horn..excuse me, waveguide. The VC former attachment and of course the surround. Both of which are horrid in approximately 99% of coaxials, because they are designed as fullranged. They don't have to be. I was discussing this with PE and someone you might know, Jerry McNutt of Eminence, just last weekend. If they were designed to operate as true midranges, say above 400hz (far lower than even your largest WG), there would be no need for pleated or rolled surrounds. And there are ways to attach the former so that a smooth transition is made.
If you recall, it was a coax that was judged to be at least the equal of the Summa in Dukes listening tests, level notwithstanding
. Perhaps even greater uniformity in the polars are a good thing.Cheers,
AJ
You're putting words in my mouth. I never said DI was necessarily higher through the crossover region when using an axisymmetrical horn compared to an asymmetrical one. In fact, I specifically said "It's not that the null is any less deep, it's that the pattern above the crossover doesn't widen back up."
Above the crossover range, I prefer to have the vertical directivity remain narrow so the average off-axis amplitude is matched as closely as possible. I'd rather that the vertical pattern not widen back up at HF after it dips through the crossover region because of the nulls.
You say you'd like to see a comparison of a system of the same basic design using an axisymmetric horn for comparison. I have one for you: your Summa. Look at the response curve at the null angle. I'm sure the output above about 3kHz is just as strong as it is below 800Hz. The crossover region dips, but HF does not. Outside the null angle, your horn should be strong too, which is sending a lot of HF energy to reflect off the ceiling and floor. These features of axisymmetrical horns are not particularly attractive to me.
The way I see it, horns with tall vertical pattern are at least 6dB too hot at large vertical angles. Their off-axis coverage angle is tall, so they provide full output at angles outside the null angle. My position is, and always has been, that it would be better to limit the vertical angle. Output at large vertical angles serves no useful purpose.
Wayne Parham said:
This is symantic, you still saying that the null will be less deep relative to the passband. I don't see any proof or data to support this.
You are forgetting that at the crossover the smaller vertical dimension will cause the vertical response to widen relative to the wider axisymmetric device, thus defeating your argument. The reality depends on relative and absolute sizes of all components and cannot be generalized. If the devices were all very large compared to the wavelengths then what you say has a stronger case, but NOT when the device size (and driver spacing) is comparable to a wavelength at the crossover. Then its a more complex situation and I think that you are guessing at what will happen.
AJinFLA said:
Then why not use a larger coaxial? They come in sizes all the way up to 18". The main problem is the discontinuities in the formed horn..excuse me, waveguide. The VC former attachment and of course the surround. Both of which are horrid in approximately 99% of coaxials, because they are designed as fullranged. They don't have to be. I was discussing this with PE and someone you might know, Jerry McNutt of Eminence, just last weekend. If they were designed to operate as true midranges, say above 400hz (far lower than even your largest WG), there would be no need for pleated or rolled surrounds. And there are ways to attach the former so that a smooth transition is made.
If you recall, it was a coax that was judged to be at least the equal of the Summa in Dukes listening tests, level notwithstanding
Cheers,
AJ
I don't follow. How does one make a coax where the waveguide is about the same size as the woofer?
Duke did no such test that I am aware of. What are you referring to? The test that Duke and I did with the Summas had no coax.
gedlee said:
Yes, I'm saying the average energy through the crossover region at the null angle will be less deep relative to the passband, absolutely. That almost goes without saying. An horn with a tall vertical pattern radiates more off axis energy than one with a more narrow pattern.
gedlee said:
I'm not forgetting anything. I have been making DI-matched speakers for a long time and have studied this issue even longer. I have always been able to use either axisymmetrical or asymmetrical horns. I chose asymmetrical horns because I do not think there is anything to be gained having a tall vertical pattern in a speaker like this. Even with a relatively small horn, one that doesn't get vertical pattern control until 3kHz, you still get the benefit of reduced vertical HF from that point up. To me, that's worthwhile.
I can remember discussions like this at Audio Asylum probably 8 years ago. The round horn guys there weren't concerned about directivity so much, they all liked tractrix "salad" bowl horns. So my arguments fell on deaf ears there. But I came to the conclusion then that the strongest benefit of round horns is they're easy to make.
Earl, the Gradient Revolution we used in the blind test at your house has a coaxial mid/tweet module.
Here are some measurements of the Revolution from twelve years ago:
http://www.stereophile.com/floorloudspeakers/616/index4.html
Duke
Here are some measurements of the Revolution from twelve years ago:
http://www.stereophile.com/floorloudspeakers/616/index4.html
Duke
Dr Geddes,
could you answer a question about the mouth termination of an OS waveguide?
Your speakers waveguides have a mouth flare after the part with the constant wall angle, which looks like a circular shape. It doesnt seem to be part of what the OS equation gives. You stated earlier in this thread, that the intention of this is to attenuate diffraction and mouth reflection.
My question is: How do i determine the correct shape for this mouth termination?
could you answer a question about the mouth termination of an OS waveguide?
Your speakers waveguides have a mouth flare after the part with the constant wall angle, which looks like a circular shape. It doesnt seem to be part of what the OS equation gives. You stated earlier in this thread, that the intention of this is to attenuate diffraction and mouth reflection.
My question is: How do i determine the correct shape for this mouth termination?
Wayne Parham said:
This has clearly been your assumption, but it is not correct. It depends on where the size of the horn lies relative to the wavelengths. A narrower vertical coverage horn has a narrower vertical dimension. Hence it will be diffraction limited to a far higher frequency than a round one. You are making assumptions for simplicity that are simply not correct. The narrower coverage horn COULD have a wider pattern at the lower frequencies than the wider coverage horn. It all depends on relative sizes as I said before.
audiokinesis said:
This was not the subject of the discussion. We were talking about horn tweeter coaxes and the size of the horn. Those speakers did not have horns. Coaxes without horns are not subject to the same limitations as trying to fit a waveguide in the cone of the woofer. The best approach is to use the cone body as the waveguide, but this is not ideal either. Remember that the Revelations were some 10 dB less efficient than the Summas.