GM said:
There is no advantage to round or square either way. Its all how you achieve the mouth contour, not about cross section. But since the throat for ALL compression drivers is round, there is an advantage to keeping the cross section round. But square can be dealt with, its just not as easy to do. The idea that the "mouth is bigger" for a sqare waveguide is a misnomer, I mean it is bigger, but so what?
Yep, you're 'preaching to the choir' with me.
BTW, a quick search didn't pan out, so I'd still like the link to the tweeter 'proof' thread I asked for here: http://www.diyaudio.com/forums/showthread.php?postid=1495117#post1495117
GM
BTW, a quick search didn't pan out, so I'd still like the link to the tweeter 'proof' thread I asked for here: http://www.diyaudio.com/forums/showthread.php?postid=1495117#post1495117
GM
gedlee said:
I use foam in my Unitys (which I got from Dr. Geddes). They did have a slight "horn" character, which the foam completely eliminates - to my ear. I found that I did not have to completely fill the horn. My driver is a TAD 2001. The foam goes from the phase plug of the driver to about halfway between the entrance and mouth of the horn. Filling the horn completely sounds no different to me. The SPL loss is not a problem, as it maybe falls off a couple of DB above 8-10k. I don't hear much there, but I've adjusted for it anyway.
Sheldon
edit: I've tried the my Unity's with without the mid drivers, and with the entrance holes covered. Of course, the crossover point was adjusted to about 1k. My bass driver (Lambda Apollo TDx) can easily cover this range smoothly. But there is something about the speaker when using the mids that I like a little better - a bit more fullness. Certainly my experience suggests that very fine sound can be had with a good compression upper mid/tweeter crossed to a good 15 in driver. It's simpler, and where I starting over, I'd probably go that way.
Sheldon
Patrick Bateman said:
Dear Patrick,
I also want to cross the 2" drivers around 200-250Hz.
Question is, how high they can play??
I just would like to use the 8" speakers as a third way below 200Hz...
Do you know any 3 way experiment on the Unity horn?
Like the SH50 for example play down to around 50Hz...
Thanks,
Tamas
How high they can play really depends on how close you can get them to the throat. For instance, if you can get the holes for the midranges ONE INCH from the throat, you'll see cancellation come into effect at 3,450hz. (13800hz / 1in / 4).
Basically your upper frequency limit is dictate by where those holes go. The soundwave from the midrange is going to reflect off the throat, and those reflections will obliterate any output above that frequency.
You can play around with it in horn response; versions 19 and higher can model an offset throat (i.e. a Unity horn.)
Basically your upper frequency limit is dictate by where those holes go. The soundwave from the midrange is going to reflect off the throat, and those reflections will obliterate any output above that frequency.
You can play around with it in horn response; versions 19 and higher can model an offset throat (i.e. a Unity horn.)
Hi Earl
You said “First, the Unity horn is designed for extremely high SPLs and it is good at that. In the Pro market sound quality is not a major (or certainly not THE major) driving factor (from what I have heard of late its not even a consideration).”
Actually the vast majority of our sales are in permanent installations like big Churches, LasVegas musical shows large multi-media rooms and such.
Maximum loudness isn’t a criteria but the best sound they can get is and this is often judged in side by side comparisons.
It turns out it is very hard to get good sound for a large number of people.
What was important to me in refining the design was eliminating the lobes that multi way speakers normally have and as much as possible make them emulate a single source in time and radiation pattern.
In concert sound it is easy to hear that the more sources you have interfering with each other, the less like music it sounds.
In large rooms, lobes from a single box or from multiple boxes are undesirable and least there is some distinction to be well enough behaved in time to preserve wave shape over a decade (unusual for a multi way speaker in Pro and hifi both).
Also, most of the SH products have at least one array able side where they can be placed adjacent to another unit or a physical boundary without the normal interference two cabinets sitting side by side normally produce. There is predicted and measured data for a pair for example.
You said “So are HOM an issue? No not really, there would be lots of them created by the midranges etc. and those from the throat would not be a big additive issue.
Again, is there a way to quantify this or is it a matter of mathematical criteria here?
Frankly, based on working the levitation sources, I don’t see a way for an “acoustically small” source to produce / propagate HOM’s in the throat of a straight sided conical horn. How would this happen?
On the other hand, the “Kit” unity horns had a throat profile was altered from the original, did have a mode at 4KHz and the interference from the mid holes made a “black dot” in the exact center of the pattern at 4KHz.
Four absorbers placed on the horn mouth reduced the depth to a few dB, using the original “less pretty” throat shape avoids the issue.
“But from a sound quality standpoint for lower SPL and smaller rooms, I would contend that the OS waveguides with foam would have a far better sound quality than the Unity. The unity will have lots more SPL. Its a tradeoff.”
I would like to hear that comparison myself, personally I would normally like to hear both cases before making a sweeping judgment about sound quality.
They might be ugly black boxes and not aimed at “hifi’ but I wouldn’t mind putting a pair of any of our SH horns in a side by side demo with anybodies hifi speakers, yours too.
So, I’m game, it’s summer, I can measure outside, can you describe a way to measure / quantify the “lots of HOM’s” you refer to ?.
Patrick,
I saw your car adaptation; you are a wild man, pretty cool.
Let the location of the cancellation notches steer your direction.
Best,
Tom Danley
You said “First, the Unity horn is designed for extremely high SPLs and it is good at that. In the Pro market sound quality is not a major (or certainly not THE major) driving factor (from what I have heard of late its not even a consideration).”
Actually the vast majority of our sales are in permanent installations like big Churches, LasVegas musical shows large multi-media rooms and such.
Maximum loudness isn’t a criteria but the best sound they can get is and this is often judged in side by side comparisons.
It turns out it is very hard to get good sound for a large number of people.
What was important to me in refining the design was eliminating the lobes that multi way speakers normally have and as much as possible make them emulate a single source in time and radiation pattern.
In concert sound it is easy to hear that the more sources you have interfering with each other, the less like music it sounds.
In large rooms, lobes from a single box or from multiple boxes are undesirable and least there is some distinction to be well enough behaved in time to preserve wave shape over a decade (unusual for a multi way speaker in Pro and hifi both).
Also, most of the SH products have at least one array able side where they can be placed adjacent to another unit or a physical boundary without the normal interference two cabinets sitting side by side normally produce. There is predicted and measured data for a pair for example.
You said “So are HOM an issue? No not really, there would be lots of them created by the midranges etc. and those from the throat would not be a big additive issue.
Again, is there a way to quantify this or is it a matter of mathematical criteria here?
Frankly, based on working the levitation sources, I don’t see a way for an “acoustically small” source to produce / propagate HOM’s in the throat of a straight sided conical horn. How would this happen?
On the other hand, the “Kit” unity horns had a throat profile was altered from the original, did have a mode at 4KHz and the interference from the mid holes made a “black dot” in the exact center of the pattern at 4KHz.
Four absorbers placed on the horn mouth reduced the depth to a few dB, using the original “less pretty” throat shape avoids the issue.
“But from a sound quality standpoint for lower SPL and smaller rooms, I would contend that the OS waveguides with foam would have a far better sound quality than the Unity. The unity will have lots more SPL. Its a tradeoff.”
I would like to hear that comparison myself, personally I would normally like to hear both cases before making a sweeping judgment about sound quality.
They might be ugly black boxes and not aimed at “hifi’ but I wouldn’t mind putting a pair of any of our SH horns in a side by side demo with anybodies hifi speakers, yours too.
So, I’m game, it’s summer, I can measure outside, can you describe a way to measure / quantify the “lots of HOM’s” you refer to ?.
Patrick,
I saw your car adaptation; you are a wild man, pretty cool.
Let the location of the cancellation notches steer your direction.
Best,
Tom Danley
Tom Danley said:
Hi Tom
As in any horn the HOM bounce off the walls. The fact that they are straight makes no difference. If a sound wave on its way down the horn encounters any "aberation" such as another driver or such, then it will diffract off of said aberation. The diffraction will then bounce off of the walls creating a wave that travles not down the horn in a straight line but angular to the axis bouncing off of the walls.
Take data at 2 to 3 meters out, not 10 and not ground plane, every 7.5° and send me the raw data by recording white noise (or pink, it doesn't matter) on one stereo channel and the SPL on the other (at the same time of course) as a wav file, about 10-15 seconds, then I will plot out the data just as you see for all my stuff. The HOM and diffraction, will appear as irregular polar responses. If your systems, measured in that way, have comparable polar pattern control to mine, then I will be very impressed indeed, since I have yet to see polars as good as what I show. And I think that the quality of the sound is reflected by the measurements.
I assume that you have a test stand at least four feet high as the ground bounce will be the limiting factor.
Hi Earl
“As in any horn the HOM bounce off the walls. The fact that they are straight makes no difference. If a sound wave on its way down the horn encounters any "aberation" such as another driver or such, then it will diffract off of said aberation. The diffraction will then bounce off of the walls creating a wave that travels not down the horn in a straight line but angular to the axis bouncing off of the walls.”
Yes I am familiar with this, a 2 inch driver on a wide angle horn, a compression driver with a typical ring type phase plug, all these can typically produce interference in the horn.
The mid hole / throat shape also cause a mode to appear in the Kit version of the unity horn at 4Khz too.
On the other hand, if one has a simple conical horn driven at the apex with an acoustically small dimension point source, there is essentially no way for anything but a partial spherical patch to radiate. To do else wise, would require the horns walls to be at something other than perpendicular to the direction of propagation or vise versa.
That condition doesn’t exist if the source is at the apex and is acoustically small (too small for the origin to have directionality on its own).
In other words, I would contend that if one drives a simple conical horn from a point where the horn is less than about 1 / 4 to 1/3 wl across, that HOM’s of any significance, can’t be produced or radiated. Sure you might be able to present that throat area with a complex phase but the distance between the phases is so small they all sum.
You can’t produce a complex radiation pattern in space or have directivity without a minimum acoustic size.
In practice, if one has a source which already produces a diverging wavefront which is “near enough” to the angle of the conical horn it attaches to for the frequency and size, that the conical internal horn in the driver is simply an extension of the outer horn you see.
I speak in black and white here, obviously the math for all this doesn’t reach asymptotic “zero” at these points rather, that for practical purposes, the effects go away.
I can measure up to 20 feet high, 2 or 3 meters is no problem.
My turntable has clicks at 2.5, 5 and 10 degrees, would 5 degrees be better instead?
Do you want from 0 to 90 degrees, half or full circle?
Hey, I’m not saying they will be as nice as yours but I am curious how they would look on your visual’s and willing to take the data. I will do this after the trade show next week.
Best,
Tom
“As in any horn the HOM bounce off the walls. The fact that they are straight makes no difference. If a sound wave on its way down the horn encounters any "aberation" such as another driver or such, then it will diffract off of said aberation. The diffraction will then bounce off of the walls creating a wave that travels not down the horn in a straight line but angular to the axis bouncing off of the walls.”
Yes I am familiar with this, a 2 inch driver on a wide angle horn, a compression driver with a typical ring type phase plug, all these can typically produce interference in the horn.
The mid hole / throat shape also cause a mode to appear in the Kit version of the unity horn at 4Khz too.
On the other hand, if one has a simple conical horn driven at the apex with an acoustically small dimension point source, there is essentially no way for anything but a partial spherical patch to radiate. To do else wise, would require the horns walls to be at something other than perpendicular to the direction of propagation or vise versa.
That condition doesn’t exist if the source is at the apex and is acoustically small (too small for the origin to have directionality on its own).
In other words, I would contend that if one drives a simple conical horn from a point where the horn is less than about 1 / 4 to 1/3 wl across, that HOM’s of any significance, can’t be produced or radiated. Sure you might be able to present that throat area with a complex phase but the distance between the phases is so small they all sum.
You can’t produce a complex radiation pattern in space or have directivity without a minimum acoustic size.
In practice, if one has a source which already produces a diverging wavefront which is “near enough” to the angle of the conical horn it attaches to for the frequency and size, that the conical internal horn in the driver is simply an extension of the outer horn you see.
I speak in black and white here, obviously the math for all this doesn’t reach asymptotic “zero” at these points rather, that for practical purposes, the effects go away.
I can measure up to 20 feet high, 2 or 3 meters is no problem.
My turntable has clicks at 2.5, 5 and 10 degrees, would 5 degrees be better instead?
Do you want from 0 to 90 degrees, half or full circle?
Hey, I’m not saying they will be as nice as yours but I am curious how they would look on your visual’s and willing to take the data. I will do this after the trade show next week.
Best,
Tom
Tom Danley said:
Hi Tom
Unfortunately what you are saying is not true, IF you could excite a cone at its apex, which is a point, with a point source, then what you say would be true. But you can't. You are always some distance from the apex and the true wavefront is always somewhat different than a spherical wave at that location. Hence these "mismatches" will always generate HOMs. Now below the 1/3 - 1/2 lambda that you mention these effects are evanescent, (i.e. complex wavenumber k) and they propagate with an exponentiallly decreasing amplitude. If this amplitude is negligable by the time that they reach the mouth then they will be insignificant in the radiated field. But the evanescent waves diminish to about 1 / eth (e^-1) in about lambda / 2, thus they can reach the mouth at any frequency. Above some frequency these aberations will be significant as they no longer propagate with an exponential decrease (i.e. the wavenumber becomes real).
I'm not saying that they will always be significant, I'm just saying that you can't wave your arms and say that they won't exist because they will always exist. They exist even in a perfect OS waveguide. The issue is to what extent do they exist.
Thats the whole idea behind the foam. It will attenuate these modes at an even greater rate thus diminising them even more. As anyone who has experimented with the foam will tell you this is not an insigificant audible effect. SO while the HOM may not be significant in measured terms all data points to them being significantly audible effects even at very low levels.
In my reasearch an axial hole is due to the mouth diffraction not HOM. HOMs should only affect off axis responses - they should not be able to create a hole on axis.
I can deal with 5° - that would be great. I'll look forward to seeing the data and I'll surely post it here no matter how it comes out - you agree with that? I'm always interested in the truth as this is the only way to guide research. Hiding or manipulating the data is never to ones advantage (unless you are in marketing.)
Thanks for the cooperation. I look forward to it.
Over on another thread, someone posted this:
"Hi Patrick,
i am not a horn specialist, but as far as i understood the
unity it is a conical horn with slow rolloff at the lower end,
when attaching the tweeter only.
This is compensated by the augmenting midrange drivers.
Is it necessary to have the wholes at all ?
Can one integrate some small midrange units with nearly
flat diaphragms into the wall without having wholes in it ?
Maybe my idea is completely wrong ...
__________________
Oliver
________________
www.dipol-audio.de"
Everyone has this idea, but it won't work.
There's a couple of reasons. First, if you mounted the midranges to the horn itself, without the coupling chamber, there would be a lot of reflections due to the shape of the cone. These reflections would screw up the frequency response. Reflections also muddy the imaging of a horn by disturbing the step response.
More seriously, you just can't get close enough to the throat. Even with the two-inch drivers I'm using, you'd be lucky to mount them within three inches of the throat. There's a cancellation which occurs in a unity horn due to to reflections from the mids to the throat. With a 3" gap from the mid to the throat, your mids would be unusable above 1150hz. The formula for that is ((13800/3)/4) = 1150hz.
"Hi Patrick,
i am not a horn specialist, but as far as i understood the
unity it is a conical horn with slow rolloff at the lower end,
when attaching the tweeter only.
This is compensated by the augmenting midrange drivers.
Is it necessary to have the wholes at all ?
Can one integrate some small midrange units with nearly
flat diaphragms into the wall without having wholes in it ?
Maybe my idea is completely wrong ...
__________________
Oliver
________________
www.dipol-audio.de"
Everyone has this idea, but it won't work.
There's a couple of reasons. First, if you mounted the midranges to the horn itself, without the coupling chamber, there would be a lot of reflections due to the shape of the cone. These reflections would screw up the frequency response. Reflections also muddy the imaging of a horn by disturbing the step response.
More seriously, you just can't get close enough to the throat. Even with the two-inch drivers I'm using, you'd be lucky to mount them within three inches of the throat. There's a cancellation which occurs in a unity horn due to to reflections from the mids to the throat. With a 3" gap from the mid to the throat, your mids would be unusable above 1150hz. The formula for that is ((13800/3)/4) = 1150hz.
Magical Curves
I have a new set of unities running in my car. This morning I spent about three hours listening. My listening session had me pondering why my reference speakers do some things that the Unity doesn't.
First, I have to preface this by saying that I have a Unity clone. I'm 110% certain that a real Unity or Synergy horn is in a whole 'nother ballpark than the mess I've made. Having said that...
My reference speakers (the Summas) are able to "disappear" in a way that my unities haven't. My unities have dynamics and a big soundstage, but it's solidly confined to the width of the dash. My Summas are able to image well beyond the boundaries of the room that they're in.
Pondering why this is, I think it has to do with the Summas ability to minimize any diffraction produced in the waveguide.
Once you get past the math, the fundamental idea behind an OS waveguide is to avoid diffraction like the plague. The idea is that there are not ANY sharp transitions from the diaphragm, all the way to your ear. Of course there's a point of diminishing returns; it doesn't make sense to use a 30" waveguide because the compression driver doesn't play low enough to warrant one.
But in the Unity, it DOES. Because the midranges are coupled to the waveguide, we need directivy and diffraction control down L-O-W.
To complicate matters further, I'm putting this in a CAR!
In my car, the "mouth" is formed by the dash and the windshield. Because of this, I'm now convinced that the best path to a solid image is waveguide which mimics the shape of the dash as closely as possible.
In my first iteration of a Unity in the car I used a elliptical mouth.
In my 2nd iteration, which is in the car right now, I used a rectangular mouth.
After thinking about this, I'm convinced that the best solution (for a car) is TRIANGULAR. Basically a triangular mouth to mate with the dash and the windshield, which gradually morphs to circular as it gets closer to the throat.
I have a new set of unities running in my car. This morning I spent about three hours listening. My listening session had me pondering why my reference speakers do some things that the Unity doesn't.
First, I have to preface this by saying that I have a Unity clone. I'm 110% certain that a real Unity or Synergy horn is in a whole 'nother ballpark than the mess I've made. Having said that...
My reference speakers (the Summas) are able to "disappear" in a way that my unities haven't. My unities have dynamics and a big soundstage, but it's solidly confined to the width of the dash. My Summas are able to image well beyond the boundaries of the room that they're in.
Pondering why this is, I think it has to do with the Summas ability to minimize any diffraction produced in the waveguide.
Once you get past the math, the fundamental idea behind an OS waveguide is to avoid diffraction like the plague. The idea is that there are not ANY sharp transitions from the diaphragm, all the way to your ear. Of course there's a point of diminishing returns; it doesn't make sense to use a 30" waveguide because the compression driver doesn't play low enough to warrant one.
But in the Unity, it DOES. Because the midranges are coupled to the waveguide, we need directivy and diffraction control down L-O-W.
To complicate matters further, I'm putting this in a CAR!
In my car, the "mouth" is formed by the dash and the windshield. Because of this, I'm now convinced that the best path to a solid image is waveguide which mimics the shape of the dash as closely as possible.
In my first iteration of a Unity in the car I used a elliptical mouth.
In my 2nd iteration, which is in the car right now, I used a rectangular mouth.
After thinking about this, I'm convinced that the best solution (for a car) is TRIANGULAR. Basically a triangular mouth to mate with the dash and the windshield, which gradually morphs to circular as it gets closer to the throat.
This might be of interest Patrick:
http://aa.peavey.com/downloads/pdf/qwp1.pdf
It's a paper which includes how to make a waveguide type device. Earls says not optimal, but then again, pretty easy to make....
I would be tempted to add a 3" roll at the mouth instead of the foam...
http://aa.peavey.com/downloads/pdf/qwp1.pdf
It's a paper which includes how to make a waveguide type device. Earls says not optimal, but then again, pretty easy to make....
I would be tempted to add a 3" roll at the mouth instead of the foam...
John, now that you have unity waveguides in the car again, what configuration are you using for the mids? I have built one waveguide and am using the same compression driver you are. However, I need direction for the midranges. What drivers, crossover points, enclosures are you using now for the mids and do you have a new freq response plotted? BTW, my project is for in car use as well.
I'm using the TB midranges which I recommended earlier in the thread. As far as I know, there's nothing better for an in-car application. The Aurasound mids which I used in 2006 have an FS which is too low, and a QMS which is too high (due to the heavy aluminum cone.)
The aurasound woofers have way more excursion, but you can't get them anyways.
I haven't mounted compression drivers to the waveguide yet; my goal is to get the mids right first.
The crossover for the compression driver which I listed on the thread at audiogroupforum.com should still work.
The defect in the original design I posted in 2006 wasn't the compression driver; it was the mids which were defective.
Needles to say, I've fixed that this time around.
The aurasound woofers have way more excursion, but you can't get them anyways.
I haven't mounted compression drivers to the waveguide yet; my goal is to get the mids right first.
The crossover for the compression driver which I listed on the thread at audiogroupforum.com should still work.
The defect in the original design I posted in 2006 wasn't the compression driver; it was the mids which were defective.
Needles to say, I've fixed that this time around.
John, doesn't the pointy bit of the conical waveguide have to be truncated so it has a 1" mouth also?
I think you're right about minimizing diffraction in a waveguide, as mentioned in earlier post. Have you smoothed the edge of the mid's ducts where they enter the horn and have you given them that smoothed "frustoconical" shape on the driver side?
I think you're right about minimizing diffraction in a waveguide, as mentioned in earlier post. Have you smoothed the edge of the mid's ducts where they enter the horn and have you given them that smoothed "frustoconical" shape on the driver side?
Patrick Bateman said:In the process of doing a sanity check between a conical and an oblate spheroidal waveguide, I threw together this pic which compares the two.
Re: Magical Curves
"Disappearing" is one of my most important subjective requirements. If you can localize the source then something is wrong - you want to hear the recording not the speakers.
This is a big aspect of why I put my speakers behind screens. When you can't see them you really have no idea where they are and this adds to the "illusion" of good sound stage. When they are visible your mind will always say "I know where they are". even if you close your eyes. I always ask people if they can tell where the speakers actually are and they almost never can. This is a very very big factor in sound quality assesment and very few speakers are capable of it.
This is exactly where you diffraction hypothesis is dead on. The diffraction will virtually always be localizable because it is delayed from the main signal. Only with near zero diffraction can the speakers "disappear".
Patrick Bateman said:
"Disappearing" is one of my most important subjective requirements. If you can localize the source then something is wrong - you want to hear the recording not the speakers.
This is a big aspect of why I put my speakers behind screens. When you can't see them you really have no idea where they are and this adds to the "illusion" of good sound stage. When they are visible your mind will always say "I know where they are". even if you close your eyes. I always ask people if they can tell where the speakers actually are and they almost never can. This is a very very big factor in sound quality assesment and very few speakers are capable of it.
This is exactly where you diffraction hypothesis is dead on. The diffraction will virtually always be localizable because it is delayed from the main signal. Only with near zero diffraction can the speakers "disappear".
You know what would be fun?
At my house I should build a false wall, and put the Summas behind it. Then I should buy a pair of Paradigm Atoms and put them out in the room.
Then I could invite friends over, and have fun convincing them that the Paradigms are playing.
It's the best of both worlds - big speaker dynamics behind the veil, and a lil' tiny speaker out in the room.
At my house I should build a false wall, and put the Summas behind it. Then I should buy a pair of Paradigm Atoms and put them out in the room.
Then I could invite friends over, and have fun convincing them that the Paradigms are playing.
It's the best of both worlds - big speaker dynamics behind the veil, and a lil' tiny speaker out in the room.
Question.
Thinking about Tom Danley's synergy speakers.... When the circular aperture of the driver is opened into the rectangular conical waveguide, there is a smoothing of the circular shape of driver exit to the rectangular shape of the waveguide, using putty or something.
Now, my question is this, if I think of the driver/waveguide as a single system, then with the modification of shape, is the aperture into the waveguide really circular or has it been modified into some other shape?
That is, looking down the waveguide from the mouth direction a millimeter or so beyond the driver, the cross section doesn't seem to be a rectangle or a circle. How might this be described?
Thinking about Tom Danley's synergy speakers.... When the circular aperture of the driver is opened into the rectangular conical waveguide, there is a smoothing of the circular shape of driver exit to the rectangular shape of the waveguide, using putty or something.
Now, my question is this, if I think of the driver/waveguide as a single system, then with the modification of shape, is the aperture into the waveguide really circular or has it been modified into some other shape?
That is, looking down the waveguide from the mouth direction a millimeter or so beyond the driver, the cross section doesn't seem to be a rectangle or a circle. How might this be described?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.