I chose the horns because they had the exact radiation pattern I wanted: Constant directivity.
From my POV the horn radiation pattern is the most desirable characteristic, as opposed to a deficiency. Once I have that in place I can solve the frequency response problems with DSP.
From my POV the horn radiation pattern is the most desirable characteristic, as opposed to a deficiency. Once I have that in place I can solve the frequency response problems with DSP.
Last edited:
@AllenB If I might tease out what some people might be thinking as they read this convo:
Many times horns sound nasal, honky, shouty, megaphone-like. No EQ in the world can fix that. People who like traditional high end speakers especially hate that kind of sound.
When you go into the hallway while the music is playing, that kind of speaker sounds even more artificial. That's because the reverberant response in the room doesn't match the speaker at all. The directivity of the horn is working against you, while the woofer is 360 degrees.
The first dipoles I ever built sounded wonderful, but something sounded immediately "off" because the ambient bass and midrange didn't match the narrowly focused treble.
I added a compression horn tweeter on the back and that fixed it.
That's why my dipole designs are full range constant directivity on both sides from 20-20K.
I find they sound best with the back side +3dB to +6dB hotter above 5K, flat response on the front side. My room is fairly absorbent and the rear treble boost sounds most neutral to my ears.
Question for @Balthazarp who built the Bitches Brews: Let's say you are only allowed to play music at quiet levels. 80dB max, not 100dB and you have a friend over who loves super traditional "laid back" soft dome speakers like LS3/5As or Harbeths, and he's very sensitive to listener fatigue. He listens to chamber music and jazz quartets.
How would your friend like the Bitches Brews?
Many times horns sound nasal, honky, shouty, megaphone-like. No EQ in the world can fix that. People who like traditional high end speakers especially hate that kind of sound.
When you go into the hallway while the music is playing, that kind of speaker sounds even more artificial. That's because the reverberant response in the room doesn't match the speaker at all. The directivity of the horn is working against you, while the woofer is 360 degrees.
The first dipoles I ever built sounded wonderful, but something sounded immediately "off" because the ambient bass and midrange didn't match the narrowly focused treble.
I added a compression horn tweeter on the back and that fixed it.
That's why my dipole designs are full range constant directivity on both sides from 20-20K.
I find they sound best with the back side +3dB to +6dB hotter above 5K, flat response on the front side. My room is fairly absorbent and the rear treble boost sounds most neutral to my ears.
Question for @Balthazarp who built the Bitches Brews: Let's say you are only allowed to play music at quiet levels. 80dB max, not 100dB and you have a friend over who loves super traditional "laid back" soft dome speakers like LS3/5As or Harbeths, and he's very sensitive to listener fatigue. He listens to chamber music and jazz quartets.
How would your friend like the Bitches Brews?
Horns that sound honky or have the cupped hands sound are just badly designed horns and I get the impression there have been plenty of those over the years. I doubt any of Joseph Crowe’s horns sound like that. We should also distinguish between waveguides and true horns that actually load the driver; they represent different points on a continuum.
You know a lot more than I do so my information may be unhelpful. However, I modeled my speaker design in ABEC then imported the FRDs into VituixCAD (thanks to Marcel Batik's ATH software). I'd imagine you'd be able to model your open baffle below 500Hz even though you have the live edge because the wavelengths are longer by that point. Then combine your high frequency measurements with the low frequency simulation. Here's an example of what that looks like when combined in VituixCAD.
The horn portion is measured in REW. The enclosure portion is a simulation from ABEC merged with a nearfield measurement from the driver in the enclosure.
I also used polar radiation as step #1. Simulated in ABEC first. Then I proceeded from #1 in this order:
#2 Measured two different compression drivers on the waveguide then chose which one to use.
#3 Designed the enclosure to match the polar pattern of the waveguide within the bounds of my limited knowledge and experience.
#4 Used DSP (VituixCAD convolution file) to flatten both drivers on a specific axis determined by my waveguide design.
#2 was basically the same as power/dynamic range considerations but I did it because I was learning and had no experience with how a compression driver would behave on my waveguide in the real world. Modeling an open baffle with ABEC would be a lot more difficult compared to what I did. But you'd only need to model the driver that plays below 500Hz and then import it into VituixCAD to match against your real world measurements of the coaxial driver. Or maybe I'm wrong about that.
In the ATH thread they discovered termination of the constant directivity waveguide is important. A coaxial woofer will never have proper termination into a baffle face. One possibility is designing a waveguide to fit over the woofer with ports like a Multiple Entry Horn. But that creates differences between your front and rear radiation if you are designing an open baffle. You'd need to design a rear chamber to somehow match the front because you can't do it with DSP.
Another option is to start a waveguide at the edge of the woofer cone like KEF and Genelec. I have no idea if you'd need to design an edge guide for the rear to compensate in an open baffle or not. But without a good termination of the waveguide into either a baffle face or into a freestanding space you'll always be stuck with an inferior horn. And maybe that's ok if the negative consequences of poor termination are inaudible in a given system.
They also discovered you can use some compression drivers from 500Hz (or lower) up to 20kHz (or higher) while not losing directivity by starting the waveguide from the phase plug exit and then adding an extension before the waveguide starts. So maybe you could just place two of them back to back to approximate what you want from an open baffle design.
@perrymarshall Thanks for the comments. Here are a couple images of my franken baffles. I am a total amateur. The tweeters are Celestion compression drivers, not truly rear facing, connected in parallel to the Lii Song FR with a Eminence 3500 Hz high pass filter. The high pass filter takes the tweeter volume down a few dB so the efficiency mismatch between the FR and tweeter is not so pronounced. The upward firing tweeters makes a huge difference in the speaker presentation, way more airy and open. I use the Eminence 250 Hz low pass simply because its a work around to allow me to drive both coils of each Lii Song 15" woofer from the 2 channel woofer amp I am using. I like the simplicity of the Amazon table leg back supports and make up for the bass loss by increasing the miniDSP woofer gain. I am using VTV Hypex class d amplifiers, 125 watts on top and 800 W for woofer. The speakers are super revealing and spacious in sound. Probably leaning toward clinical as they are unforgiving to poor recordings. I stream all my music off a Wiim Pro and Amazon HD. Thanks again for the inspiration and I am looking forward to my next project!
Kinda funny you took up Harbeths as an example. I had a friend over that has Harbeths, powered them with 300b tubes and a dac with old Philips 1541 chip. He does not really care about bass or high detail, he is all for voices and midrange. That said, he loved the bitches brew. Best speakers he have heard and it definitely changed his way at looking different speaker designs. We sat and listened to music for six hours that day. Only a break for some food in the middle. We did not play very loud, but over 80db for at least a couple of hours straight.
We did not listen to a lot of chamber and jazz unfortunately. But lots of different types of music. I have always had trouble to listening to punk/rock but I don't think the BB are any worse than anything else.
@perrymarshall ; I know you have a preference for higher sensitivity woofers but I have a pair of Aurasound NS15-992 4A sat on a shelf looking mean, moody and ready to rumble. They have a 1W sensitivity of 86dB, QTS 0.37, high Xmax, high power handing, FS of 20hz. So they ain’t ideal for OB, but I’d use DSP and a high power amp to drive them. I’d welcome your thoughts.
I have a pair of old 12" Titanic subwoofers with 19mm xmax. They're in small 30 liter boxes and if you have a big amp they can shake the room. Gotta think they're a lot like your Aurasounds - ready to rumble.
I've used them as subwoofers with all my different dipoles and they're 95% fine.
They don't quite blend perfectly with the dipoles, which could be because they're in the corners, or because they're not dipoles, or because they're heavy low efficiency drivers with thick rubber surrounds, or for all of those reasons. A trained ear can tell there's just a tad of mismatch. Most people would never notice. But in my opinion the low bass sounds a little 2-dimensional and thumpy (regardless of EQ).
If you use the Aurasounds as dipoles I think that's fine, just a couple of thoughts:
-A dipole by its very nature is optimum with a large fairly lightweight cone. (The SB Acoustics 15OB350 is a perfect example of this.) Most subs are have heavy cones because they're designed for boxes. So you're throwing away efficiency. Of course with a big amp you may not care.
-I anticipate it may not blend quite perfectly with high efficiency midrange dipoles, or at least the kinds of pro drivers I typically use which have pleated surrounds and the like. It may blend very nicely with an average efficiency midrange driver with a rubber surround.
Paper has a signature sound, metal has a different signature sound, rubber surrounds have a signature sound, etc, and no EQ in the world changes that. The match between your drivers' "signature sound" may be more important for your application than efficiency per se.
Would love to hear your thoughts when you try it out.
I've used them as subwoofers with all my different dipoles and they're 95% fine.
They don't quite blend perfectly with the dipoles, which could be because they're in the corners, or because they're not dipoles, or because they're heavy low efficiency drivers with thick rubber surrounds, or for all of those reasons. A trained ear can tell there's just a tad of mismatch. Most people would never notice. But in my opinion the low bass sounds a little 2-dimensional and thumpy (regardless of EQ).
If you use the Aurasounds as dipoles I think that's fine, just a couple of thoughts:
-A dipole by its very nature is optimum with a large fairly lightweight cone. (The SB Acoustics 15OB350 is a perfect example of this.) Most subs are have heavy cones because they're designed for boxes. So you're throwing away efficiency. Of course with a big amp you may not care.
-I anticipate it may not blend quite perfectly with high efficiency midrange dipoles, or at least the kinds of pro drivers I typically use which have pleated surrounds and the like. It may blend very nicely with an average efficiency midrange driver with a rubber surround.
Paper has a signature sound, metal has a different signature sound, rubber surrounds have a signature sound, etc, and no EQ in the world changes that. The match between your drivers' "signature sound" may be more important for your application than efficiency per se.
Would love to hear your thoughts when you try it out.
Thanks for the pics! The upward pointing tweeter is pretty cool. Hope to see you in person at a cool meetup someplace.
Here I measured the individual drivers installed in baffles at ~0.5M distance, so these measurements have the Open Baffle behavior built in. I measured on-axis only. Then I loaded the data into VituixCad. The software in this model is assuming that these drivers are on the same plane (as though this were infinite baffle) and it is showing the polar plot in RED. There are no crossover components or DSP in any of these measurements.
This only shows how the front-facing cone surfaces interfere with each other. I am NOT modeling the additional effect of Open Baffle cancellation coming from the rear, such as would be picked up by a microphone placed above the speaker at +90 degrees. I think it's safe to assume that the vertical radiation pattern above the speaker is otherwise the same as the horizontal radiation I've already shown elsewhere.
BELOW: Live Edge Dipoles - Eminence Kappa 18LF subwoofer + 8" Radian 5208 midwoofer. On the baffle the two drivers are 27"/69cm apart. Acoustic crossover frequency ~200Hz:
In the red "Directivity" graph which plots vertical polar response across +/-90 degrees, there are narrow bands of off-axis anomalies beyond -10 / +30 degrees. Overall nicely behaved. I'm not sure that dip at 200Hz is reliable, it could be a measurement or room artifact. The drivers are 1/2 wavelength apart at the 200Hz crossover.
BELOW: Bitches Brew, 2x 15OB350 subwoofers and 1x 15CXN88 midwoofer (each about 16" or 40cm apart):
Acoustic crossover is ~100Hz. In the red "Directivity" portion which plots vertical polar response across +/-90 degrees, the only anomalies are below -30 degrees, so if you're laying on the floor you'll hear a mild suckout around 200Hz. Elsewhere in the room they exhibit Constant Directivity behavior across the entire band. At high frequencies, vertical and horizontal directivity are identical because of the coax drivers. @CinnamonRolls thank you for your very astute and sophisticated model and suggestions.
The drivers are 1/7 wavelength apart at the 100Hz crossover.
So both systems give excellent vertical behavior at any realistic listener position, and power response in the room is consistent.
Subjectively both of these systems have a seamless blend between woofer and midwoofer. Radiation patterns match, signature sounds match, and when you drive the subwoofers really hard the midrange remains clean.
Last edited:
Yep. You have to pick your poison: Proper horn termination, but with polar pattern interference via the separate woofer in the crossover region; or imperfect horn termination of a coax with associated peaks and dips, and near-perfect integration of woofer and tweeter polar patterns.
Because DSP can more or less compensate for the imperfect horn termination but can't fix polar pattern, at this time I opt for coaxes with compression drivers.
Hi
Since you are already talking subwoofers I'm thinking about building a pair aswell. The bass from the BB is plenty enough for music even with 22w amp that powers them but I am a little scared for clipping while I'm watching movies.
I really love the bass from the OB so I am thinking about 2xH-frame with a sb audience 18sw450 in each. Not a dedicated open baffle woofer but seems to be quite suitable basing on spec only. It will also work in closed boxes if I don't get any satisfactory results from OB. Any inputs on my idea? Dsp crossover and a crown xls2002 amp.
Regards Jakob
Since you are already talking subwoofers I'm thinking about building a pair aswell. The bass from the BB is plenty enough for music even with 22w amp that powers them but I am a little scared for clipping while I'm watching movies.
I really love the bass from the OB so I am thinking about 2xH-frame with a sb audience 18sw450 in each. Not a dedicated open baffle woofer but seems to be quite suitable basing on spec only. It will also work in closed boxes if I don't get any satisfactory results from OB. Any inputs on my idea? Dsp crossover and a crown xls2002 amp.
Regards Jakob
There's a simpler idea that maybe you can try first: simply disconnect the bottom woofer from the current circuit. Leave everything else the same. Power the bottom woofer with your crown amplifier. Reduce the 25Hz bass boost in the DSP channel that feeds the woofers and create a new channel that drives the Crown amp. Needs to have the same EQ curve as the original Bitches Brew woofer channel, but add a 6dB low pass at 100Hz in the new channel.
Now the bottom woofer receives the high power from the Crown, and the middle woofer and tube amp have less burden to carry.
Hi, Nice clear layout of your priorities.
I very much agree with that order.
Although I group #1 and #2 together as rather inseparable, given specified SPL & headroom & bass extension goals.
I see the two together as comprising the acoustic design.
I also group #3 and #4 together as inseparable, when working at the individual driver level.
But at the complete speaker level, i too put #3 ahead of #4.
And I totally agree how DSP lets the acoustic design be the starting point.
It simply removes all the shackles associated with passive crossovers/EQs.
Can even remove the much milder shackles of DSP IIR crossovers/EQs, if FIR is part of the DSP.
Anyway, I've never built an open-baffle due to my high demands on bass extension. It's about the only type speaker I haven't tried.
Must admit you have me interested ! Always looking for the best sound I've yet to hear
That was my first idea aswell, but I did the "mistake" of calculating the new layout in vituixCAD and thought I would need different crossover parts.
I was not brave enough to try for myself, but with your blessing I'll make a weekend project of it.
Thanks for your input and help.
Regards Jakob
Your exposition of your design described so many advanced techniques of driver matching and crossover design I have been trying to understand from other texts, including Vance's, in a way that is so understandable and also impressive, in that you show the improvements made of each technique. Someone could build this design as cookbook and then walk away having a master's comprehension of so many design techniques. Wow, just wow!
Hello,
not at least Perry, but also other in this thread. Now i finally recieved the 4 bass woofers and horn tweeter components for this “crazy” project which by all means is an overstretch of my normal doings. Never build a speaker before or never heard them play either - why spend 5k euro then…
Well as everyone else 12-14 years old kid back in the early eighties I had a cousin that I admired endlessly. He had for instance the first cd player - a golden one i think… priced nearly 1000 usd that time…
Also amazing big highend speaker.
I remember some us product - a kind of combo with big woffer and kind of eg on top with midletone and and horn. And offcourse a brand new 1.8 Opel Manta in the driveway. The perfect man idol for a young kid in the early eighties…
And then the music he played wery load with big enthusiasm and patos. Was Saga in transit - live edition…
Also we heard Toto offcourse -
And to spice thing a bit up maybee some “live in japan” with - was it Led Zeppelin- no offcourse it was Deep purple…
So guess what - why do i want to spend 8k in total on this music set…
Wiimp dsp amp inclouded…
To hear SAGA in transit on 120db full volume… you the start with the Moog keyboard running wild…
Is this a good plan - Perry, Jacob and others please stop me if im wrong…
Have a great evening…
Best from Peter
Ps. Sorry for going down memory lane…
not at least Perry, but also other in this thread. Now i finally recieved the 4 bass woofers and horn tweeter components for this “crazy” project which by all means is an overstretch of my normal doings. Never build a speaker before or never heard them play either - why spend 5k euro then…
Well as everyone else 12-14 years old kid back in the early eighties I had a cousin that I admired endlessly. He had for instance the first cd player - a golden one i think… priced nearly 1000 usd that time…
Also amazing big highend speaker.
I remember some us product - a kind of combo with big woffer and kind of eg on top with midletone and and horn. And offcourse a brand new 1.8 Opel Manta in the driveway. The perfect man idol for a young kid in the early eighties…
And then the music he played wery load with big enthusiasm and patos. Was Saga in transit - live edition…
Also we heard Toto offcourse -
And to spice thing a bit up maybee some “live in japan” with - was it Led Zeppelin- no offcourse it was Deep purple…
So guess what - why do i want to spend 8k in total on this music set…
Wiimp dsp amp inclouded…
To hear SAGA in transit on 120db full volume… you the start with the Moog keyboard running wild…
Is this a good plan - Perry, Jacob and others please stop me if im wrong…
Have a great evening…
Best from Peter
Ps. Sorry for going down memory lane…
Last edited: