Tenson said:
In theory, if you match your target slopes perfectly from DC to infinity and then you time align the drivers, then yes you will have a perfect phase match. Of course matching your target curves perfectly from DC to infinity is not possible with real drivers.
You are correct it is the acoustic slopes that matter. If the impedance doesn't change the frequency response, it won't impact the acoustic phase.
Regards.
Dennis
This would work great if the tweeter can handle a 1350hz crossover point with a 2nd order electrical filter. I think it's asking too much of the HDS, but the charts look purdy 
Response barely varies over a 30 degree vertical window. My sim assumes the woofer is 3.5cm back and 20cm down, and the listening axis is 10cm higher than the tweeter at 2m. Link to csp
Dan
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Response barely varies over a 30 degree vertical window. My sim assumes the woofer is 3.5cm back and 20cm down, and the listening axis is 10cm higher than the tweeter at 2m. Link to csp
Dan
Asymmetrical radiation pattern
Speakers that are designed to cover a target pattern with constant directivity have angular coverage specifications, like 40x40 or 60x40 or 90x40. Getting the response curve smooth straight on-axis is certainly an important first step, but I think it is good to look at response off-axis too.
If the target listening area is such that a symmetrical pattern is required, fine. But in many cases, the desired coverage angle is wider than tall. In this case, symmetrical sound sources would have to be splayed, which cause interference nulls to form.
In my opinion, this is the perfect application of asymmetrcial horns. In a home hifi system for example, you generally want a wide pattern for coverage throughout the room. But you don't need (or want) the pattern to be tall. By using an asymmetrical horn with vertical pattern less than the null angle, the vertical nulls actually serve to punctuate pattern control.
This is my favorite implementation, and why I prefer asymmetrical horns to round or square ones. Not only is the horn flare asymmetrical, but the vertical placement of sound sources tends to work in your favor if you choose such a pattern. I like a consistent radiation pattern in both horizontal and vertical planes, or at least a uniformly collapsing one. But with vertically stacked sound sources, one must pay attention to coverage angle and driver placement in order to avoid having an off-axis null right in the middle of the band. In my opinion, the best way to handle this is to limit the vertical pattern to the location of the first null. In fact, the vertical nulls actually help provide pattern control for the horn, because without them the horn would be having trouble with pattern control because of its dimensions.
This is an excellent method for getting uniform response over an asymmetrical coverage pattern. Within the coverage angle, the drivers acoustic outputs combine perfectly, acting as a single sound source.
Speakers that are designed to cover a target pattern with constant directivity have angular coverage specifications, like 40x40 or 60x40 or 90x40. Getting the response curve smooth straight on-axis is certainly an important first step, but I think it is good to look at response off-axis too.
If the target listening area is such that a symmetrical pattern is required, fine. But in many cases, the desired coverage angle is wider than tall. In this case, symmetrical sound sources would have to be splayed, which cause interference nulls to form.
In my opinion, this is the perfect application of asymmetrcial horns. In a home hifi system for example, you generally want a wide pattern for coverage throughout the room. But you don't need (or want) the pattern to be tall. By using an asymmetrical horn with vertical pattern less than the null angle, the vertical nulls actually serve to punctuate pattern control.
This is my favorite implementation, and why I prefer asymmetrical horns to round or square ones. Not only is the horn flare asymmetrical, but the vertical placement of sound sources tends to work in your favor if you choose such a pattern. I like a consistent radiation pattern in both horizontal and vertical planes, or at least a uniformly collapsing one. But with vertically stacked sound sources, one must pay attention to coverage angle and driver placement in order to avoid having an off-axis null right in the middle of the band. In my opinion, the best way to handle this is to limit the vertical pattern to the location of the first null. In fact, the vertical nulls actually help provide pattern control for the horn, because without them the horn would be having trouble with pattern control because of its dimensions.
This is an excellent method for getting uniform response over an asymmetrical coverage pattern. Within the coverage angle, the drivers acoustic outputs combine perfectly, acting as a single sound source.
Wayne wrote:
“Speakers that are designed to cover a target pattern with constant directivity have angular coverage specifications, like 40x40 or 60x40 or 90x40. Getting the response curve smooth straight on-axis is certainly an important first step, but I think it is good to look at response off-axis too.”
Among other’s, Floyd Tool’s research shows that very thing, that people prefer a listening situation where the reverberant field has essentially the same spectrum as the direct sound, meaning a constant directivity condition (directivity which does not change with frequency).
Further, on axis, resolution of stereo details and words is improved when the direct sound is significantly stronger than the reverberant again suggesting directivity of a horn or large panel is your friend.
While it is easy to say 90 by 40 degrees coverage, when one measures where the sound really goes one finds it is a complicated issue.
“If the target listening area is such that a symmetrical pattern is required, fine. But in many cases, the desired coverage angle is wider than tall. In this case, symmetrical sound sources would have to be splayed, which cause interference nulls to form.”
While most speakers have them, Nulls and side lobes are not strictly required in a multiway speaker and in fact are to avoided if possible. Ours don’t have them for example and really do radiate as if they were one acoustic source.
Lobes and nulls (from the speaker system) are from self-interference caused by sources which are too far apart to add coherently which also usually makes it distinctly not constant directivity when measured, even if the sub components are.
Also while conceptually the idea of an asymmetric horn is simple, the element of “pattern Flip” makes what looks like a wide horn, radiate in a tall, narrow vertical pattern.
For example, when the EV T-35 was actually measured, they found that to have a wide horizontal radiation pattern, that it had to be mounted up and down, Pattern flip dominated its entire operating range.
Generally the more asymmetric the horn mouth / coverage angle is, the more pattern flip it will have.
“In my opinion, this is the perfect application of asymmetrcial horns. In a home hifi system for example, you generally want a wide pattern for coverage throughout the room. But you don't need (or want) the pattern to be tall. By using an asymmetrical horn with vertical pattern less than the null angle, the vertical nulls actually serve to punctuate pattern control.”
Like Don Davis said in a Synaudcon class ages ago “put the sound where the people are, not the walls, ceiling and floor”
Stand where the speaker is going to be, figure out where the audience will be.
Usually, this turns out to be much smaller angle than you would think, unfortunately if you keep the mouth the same dimension but reduce the angle by two, the pattern control loss frequency goes up an octave too. Null's and there partners, side lobes are not desirable.
If you’re going to Infocom / NSCA next month, you could stop by our booth and see two new acoustic “things” I came up with which deal with these issues (but for large scale sound however).
Tom Danley
“Speakers that are designed to cover a target pattern with constant directivity have angular coverage specifications, like 40x40 or 60x40 or 90x40. Getting the response curve smooth straight on-axis is certainly an important first step, but I think it is good to look at response off-axis too.”
Among other’s, Floyd Tool’s research shows that very thing, that people prefer a listening situation where the reverberant field has essentially the same spectrum as the direct sound, meaning a constant directivity condition (directivity which does not change with frequency).
Further, on axis, resolution of stereo details and words is improved when the direct sound is significantly stronger than the reverberant again suggesting directivity of a horn or large panel is your friend.
While it is easy to say 90 by 40 degrees coverage, when one measures where the sound really goes one finds it is a complicated issue.
“If the target listening area is such that a symmetrical pattern is required, fine. But in many cases, the desired coverage angle is wider than tall. In this case, symmetrical sound sources would have to be splayed, which cause interference nulls to form.”
While most speakers have them, Nulls and side lobes are not strictly required in a multiway speaker and in fact are to avoided if possible. Ours don’t have them for example and really do radiate as if they were one acoustic source.
Lobes and nulls (from the speaker system) are from self-interference caused by sources which are too far apart to add coherently which also usually makes it distinctly not constant directivity when measured, even if the sub components are.
Also while conceptually the idea of an asymmetric horn is simple, the element of “pattern Flip” makes what looks like a wide horn, radiate in a tall, narrow vertical pattern.
For example, when the EV T-35 was actually measured, they found that to have a wide horizontal radiation pattern, that it had to be mounted up and down, Pattern flip dominated its entire operating range.
Generally the more asymmetric the horn mouth / coverage angle is, the more pattern flip it will have.
“In my opinion, this is the perfect application of asymmetrcial horns. In a home hifi system for example, you generally want a wide pattern for coverage throughout the room. But you don't need (or want) the pattern to be tall. By using an asymmetrical horn with vertical pattern less than the null angle, the vertical nulls actually serve to punctuate pattern control.”
Like Don Davis said in a Synaudcon class ages ago “put the sound where the people are, not the walls, ceiling and floor”
Stand where the speaker is going to be, figure out where the audience will be.
Usually, this turns out to be much smaller angle than you would think, unfortunately if you keep the mouth the same dimension but reduce the angle by two, the pattern control loss frequency goes up an octave too. Null's and there partners, side lobes are not desirable.
If you’re going to Infocom / NSCA next month, you could stop by our booth and see two new acoustic “things” I came up with which deal with these issues (but for large scale sound however).
Tom Danley
Mouth diffraction widens the pattern below the frequency where the mouth is acoustically small. If the mouth is larger in one dimension than the other, then pattern control will be lost at two different frequncies. There is a transition region where the pattern narrows briefly before it widens, so some horns are made with the last section slightly wider to counter this. Above this, where the mouth is acoustically large, the wall angle of the flare sets the pattern.
Hi Tenson
Thanks, it has taken a while but they are catching on.
It’s kind of funny how these came about too, I had been collecting compression drivers with the thought that someday I would build up a large horn speaker system.
I had not thought of a way to combine the ranges that wouldn’t produce the self-interference separate horns produce.
They sat for years lacking an inspiration until my old partner teased me about my Trip to Egypt to measure the acoustics in the Great Pyramid (a brief write up)
http://www.livesoundint.com/archives/2000/julyaug/pyramid/pyramid.php
I ignored the Pyramid comment about “sharpening the sound” but later started to think about the pyramid shaped horn and its pro’s and con’s. It dawned on me one night what Don Davis meant in an old lecture on conical CD horns and I went out to the garage and made the first one and began to try to figure out how to make it work.
Anyway, we are what 10 years later now.
As the sound is really important to me, with the exception of the really big stuff, everything takes a turn in my listening room.
This is particularly useful during the crossover development process. A combination of anechoic like measurements using the TEF machine, a tower and listening has proven to be a good route for me at least.
Right now, I am using a pair of SH-50’s and a pair of TH-50’s right and left with an electronic crossover between them.
It’s kind of funny my kids have grown up with this stuff and think nothing of it.
We have recently shipped TH-50’s and a new big brother to the SH-50’s to the Imax cinema in Chicago which is being refurbished. When its open, I am going to take them down there to see what this stuff does there.
I do think there is a future in some home systems for those willing to listen to them.
The down side is that although there is a significantly higher driver and parts cost compared to “hifi” speakers, the pricing structure now doesn’t produce a price would be credible in the mainstream hifi market, no thought went into boutique attractions or “what people expect” and lastly most at work are leery, not familiar with or not at all impressed with “the state of hifi”.
Hi Wayne
Pattern narrowing as you approach pattern loss F is real and addressed with a second flare angle as don Keele described.
What I was talking about is “Pattern Flip” however, a problem with most asymmetric horns.
Pretend you have a simple horn that has a 80 degree wide and 40 degree tall pattern, a 2:1 aspect ratio, it is twice was wide as it is high.
When you go through the math, one finds to avoid Pattern flip, that the height needs to be twice the width instead of the way the horn is, twice as wide as it is tall.
For this reason, one finds that with normal horns, the more asymmetric the horn pattern is, the more pattern flip it will also have.
Tom Danley
Thanks, it has taken a while but they are catching on.
It’s kind of funny how these came about too, I had been collecting compression drivers with the thought that someday I would build up a large horn speaker system.
I had not thought of a way to combine the ranges that wouldn’t produce the self-interference separate horns produce.
They sat for years lacking an inspiration until my old partner teased me about my Trip to Egypt to measure the acoustics in the Great Pyramid (a brief write up)
http://www.livesoundint.com/archives/2000/julyaug/pyramid/pyramid.php
I ignored the Pyramid comment about “sharpening the sound” but later started to think about the pyramid shaped horn and its pro’s and con’s. It dawned on me one night what Don Davis meant in an old lecture on conical CD horns and I went out to the garage and made the first one and began to try to figure out how to make it work.
Anyway, we are what 10 years later now.
As the sound is really important to me, with the exception of the really big stuff, everything takes a turn in my listening room.
This is particularly useful during the crossover development process. A combination of anechoic like measurements using the TEF machine, a tower and listening has proven to be a good route for me at least.
Right now, I am using a pair of SH-50’s and a pair of TH-50’s right and left with an electronic crossover between them.
It’s kind of funny my kids have grown up with this stuff and think nothing of it.
We have recently shipped TH-50’s and a new big brother to the SH-50’s to the Imax cinema in Chicago which is being refurbished. When its open, I am going to take them down there to see what this stuff does there.
I do think there is a future in some home systems for those willing to listen to them.
The down side is that although there is a significantly higher driver and parts cost compared to “hifi” speakers, the pricing structure now doesn’t produce a price would be credible in the mainstream hifi market, no thought went into boutique attractions or “what people expect” and lastly most at work are leery, not familiar with or not at all impressed with “the state of hifi”.
Hi Wayne
Pattern narrowing as you approach pattern loss F is real and addressed with a second flare angle as don Keele described.
What I was talking about is “Pattern Flip” however, a problem with most asymmetric horns.
Pretend you have a simple horn that has a 80 degree wide and 40 degree tall pattern, a 2:1 aspect ratio, it is twice was wide as it is high.
When you go through the math, one finds to avoid Pattern flip, that the height needs to be twice the width instead of the way the horn is, twice as wide as it is tall.
For this reason, one finds that with normal horns, the more asymmetric the horn pattern is, the more pattern flip it will also have.
Tom Danley
Asymmetrical horns
There are various ways to make asymmetrical horns, some that have mouth dimensions that are roughly proportional to the aspect ratio, others that aren't. I won't go into the details because they can be found elsewhere. An interested reader can study radials, sectional, the early CD horns like the Manta Ray, etc.
My point is that a DI matched speaker with tweeter horn positioned above a midwoofer is best made with an asymmetrical tweeter pattern. The reasons are shown in the link below.
The direct radiating midwoofer becomes directional at higher frequencies, and reaches 90 degrees around 1kHz to 2kHz, depending on its size. This makes a convenient place to crossover to a horn with 90 degree horizontal coverage angle.
In the crossover range, both the tweeter and midwoofer cover the same frequencies so nulls will form at angles above and below the speaker. The angle between these nulls is usually 40-50 degrees, depening on the size and distance between woofer and tweeter and the crossover frequency. So if a tweeter with vertical pattern of 40 degrees is used, then the pattern will not increase at high frequencies. It will narrow in the crossover region and stay at approximately the same pattern throughout the high frequency band.
The desired result of the DI matched speaker is uniformly collapsing directivity from bass to midrange frequencies. If the tweeter pattern is symmetrical, then the pattern can be matched in the horizontal plane but not the vertical. It will narrow through the crossover range and then widen again above it. But if an asymmetrical flare is used with horizontal angle equal to the midwoofer pattern at crossover, and the vertical pattern within the null angles, the the directivity collapse is uniform in both planes. I think this is the best way to make a DI matched speaker.
There are various ways to make asymmetrical horns, some that have mouth dimensions that are roughly proportional to the aspect ratio, others that aren't. I won't go into the details because they can be found elsewhere. An interested reader can study radials, sectional, the early CD horns like the Manta Ray, etc.
My point is that a DI matched speaker with tweeter horn positioned above a midwoofer is best made with an asymmetrical tweeter pattern. The reasons are shown in the link below.
The direct radiating midwoofer becomes directional at higher frequencies, and reaches 90 degrees around 1kHz to 2kHz, depending on its size. This makes a convenient place to crossover to a horn with 90 degree horizontal coverage angle.
In the crossover range, both the tweeter and midwoofer cover the same frequencies so nulls will form at angles above and below the speaker. The angle between these nulls is usually 40-50 degrees, depening on the size and distance between woofer and tweeter and the crossover frequency. So if a tweeter with vertical pattern of 40 degrees is used, then the pattern will not increase at high frequencies. It will narrow in the crossover region and stay at approximately the same pattern throughout the high frequency band.
The desired result of the DI matched speaker is uniformly collapsing directivity from bass to midrange frequencies. If the tweeter pattern is symmetrical, then the pattern can be matched in the horizontal plane but not the vertical. It will narrow through the crossover range and then widen again above it. But if an asymmetrical flare is used with horizontal angle equal to the midwoofer pattern at crossover, and the vertical pattern within the null angles, the the directivity collapse is uniform in both planes. I think this is the best way to make a DI matched speaker.
Hi Wayne
You said: “There are various ways to make asymmetrical horns, some that have mouth dimensions that are roughly proportional to the aspect ratio, others that aren't. I won't go into the details because they can be found elsewhere. An interested reader can study radials, sectional, the early CD horns like the Manta Ray, etc.”
You missing the point, these (radial, conical, pyramidal, sectoral, early CD, multi-cell) horns ARE the kinds that HAVE pattern flip. For an asymmetric horn not to have pattern flip, it needs to have the right proportions as I described earlier.
This is not visible looking at the horn itself, in fact it is counter intuitive and only seen by measuring where the sound actually goes.
“In the crossover range, both the tweeter and midwoofer cover the same frequencies so nulls will form at angles above and below the speaker.”
True when the sources are too far apart to add coherently.
Tom
You said: “There are various ways to make asymmetrical horns, some that have mouth dimensions that are roughly proportional to the aspect ratio, others that aren't. I won't go into the details because they can be found elsewhere. An interested reader can study radials, sectional, the early CD horns like the Manta Ray, etc.”
You missing the point, these (radial, conical, pyramidal, sectoral, early CD, multi-cell) horns ARE the kinds that HAVE pattern flip. For an asymmetric horn not to have pattern flip, it needs to have the right proportions as I described earlier.
This is not visible looking at the horn itself, in fact it is counter intuitive and only seen by measuring where the sound actually goes.
“In the crossover range, both the tweeter and midwoofer cover the same frequencies so nulls will form at angles above and below the speaker.”
True when the sources are too far apart to add coherently.
Tom
Is "Pattern Flip" evident in this JBL 90° x 50° asymmetric "Horn" as recently measured by Dr. Geddes?
http://140.174.80.168/downloads/Horn VS waveguide.pdf
http://140.174.80.168/downloads/Horn VS waveguide.pdf
There are several kinds of asymmetrical horns and not all exhibit pattern flip. This phrase is a generalization, because the true issue is mouth diffraction. When an aperture is acoustically small it forms a diffraction slot. That's why the pattern widens at low frequencies.
At high frequencies, the wall angle sets the pattern. At low frequencies, wall angle is irrelevant because mouth dimensions prevent the wall from setting the pattern. That's what is really happening. The larger the horn, the lower it has pattern control, whether it is symmetrical or not. It's just that an asymmetrical horn's aspect ratio tends to make the dimension you need the tightest pattern shorter, which tends to make it have less pattern control.
Pattern flip is not an issue if the horn is used above the frequency where pattern control is lost. It is also not an issue if the pattern is modified by something else outside the horn, as is the case of a loudspeaker with vertically stacked drivers. In this case, the vertical pattern isn't set by the horn at low frequencies anyway.
The thing that really matters in this issue is not how the horn acts in the vertical plane at low frequencies, but how it does at high frequencies. At low frequencies, the null angle sets the pattern more than anything else. The position of the drivers on the baffle is what sets the vertical pattern at low frequencies. Above the crossover point, it is good that the vertical angle remain within the angle set by the nulls. This is a good way to provide uniform directivity in the vertical plane.
About sound sources combining coherently, drivers stacked vertically on a baffle do combine coherently on-axis and off-axis in the horizontal plane. When the listener moves along the vertical plane, the issue becomes one of path length.
On-axis, the two sound sources are in phase. The path lengths remain within 1/4 wavelength of each other at small vertical angles. As the vertical angle is increased, the path length difference increases to the point where the sound sources arrive out of phase. This sets the angular limit of coherent summing in the vertical plane.
My point is this makes it desirable to limit the directivity of the sound sources to an angle less than the vertical null angle.
I don't need to re-state this anymore. It gets tiresome to go over and over the same thing, each time restating the relevant points. Readers here can study the isues and make up their own minds.
At high frequencies, the wall angle sets the pattern. At low frequencies, wall angle is irrelevant because mouth dimensions prevent the wall from setting the pattern. That's what is really happening. The larger the horn, the lower it has pattern control, whether it is symmetrical or not. It's just that an asymmetrical horn's aspect ratio tends to make the dimension you need the tightest pattern shorter, which tends to make it have less pattern control.
Pattern flip is not an issue if the horn is used above the frequency where pattern control is lost. It is also not an issue if the pattern is modified by something else outside the horn, as is the case of a loudspeaker with vertically stacked drivers. In this case, the vertical pattern isn't set by the horn at low frequencies anyway.
The thing that really matters in this issue is not how the horn acts in the vertical plane at low frequencies, but how it does at high frequencies. At low frequencies, the null angle sets the pattern more than anything else. The position of the drivers on the baffle is what sets the vertical pattern at low frequencies. Above the crossover point, it is good that the vertical angle remain within the angle set by the nulls. This is a good way to provide uniform directivity in the vertical plane.
About sound sources combining coherently, drivers stacked vertically on a baffle do combine coherently on-axis and off-axis in the horizontal plane. When the listener moves along the vertical plane, the issue becomes one of path length.
On-axis, the two sound sources are in phase. The path lengths remain within 1/4 wavelength of each other at small vertical angles. As the vertical angle is increased, the path length difference increases to the point where the sound sources arrive out of phase. This sets the angular limit of coherent summing in the vertical plane.
My point is this makes it desirable to limit the directivity of the sound sources to an angle less than the vertical null angle.
I don't need to re-state this anymore. It gets tiresome to go over and over the same thing, each time restating the relevant points. Readers here can study the isues and make up their own minds.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.