Hi
We use a lot of the 4592’s (some of the speakers at work have1, 3 and 4 of them) but the 4592 has an internal geometry and exit large enough to already be defining the directivity up high and the high on axis sensitivity is partly a result of that driver beaming at 20KHz even without a horn.
Another issue I have found it that the acoustic phase or time relationship makes it very hard to have a summation without crossover phase shift like the normal Synergy horn crossovers (which sum into what looks like a single driver w/o crossover).
You might look at the 4550 which has an unusually low resonant frequency and impedance bump while also having good hf response.
We use hundreds of those at work and they are also a very reliable driver.
The one inch exit will fill out up to about a 50 degree conical horn at 20KHz too and radiates a curved (expanding) wavefront due to its internal geometry.
Best,
Tom Danley
Danley Sound Labs
Hello Jack,
As Tom Danley, explained, the HF part of the 4592 compression driver defines itself its own beaming so the beaming in HF will be the same on the Azura and on the Minphase.
We have a lot of experience here in France with large Le Cléac'h round horns. We succeed in reducing the effect of the beaming inherent to that kind of horn rotating the axis of the horns to cross in front of the listeners head. Depending on the distance between the 2 loudspeakers , the crossing point of the horns axis can be from 80centimetres (30 inches) to 120 centimeters (50 inches).
Such horns should never be listened with parallel axis or small angle axis.
Additionally, that kind of set up allows a nearfield listening with 3D image effect.
Best regards from Paris, France
Jean-Michel Le Cléac'h
Jean
As Tom Danley, explained, the HF part of the 4592 compression driver defines itself its own beaming so the beaming in HF will be the same on the Azura and on the Minphase.
We have a lot of experience here in France with large Le Cléac'h round horns. We succeed in reducing the effect of the beaming inherent to that kind of horn rotating the axis of the horns to cross in front of the listeners head. Depending on the distance between the 2 loudspeakers , the crossing point of the horns axis can be from 80centimetres (30 inches) to 120 centimeters (50 inches).
Such horns should never be listened with parallel axis or small angle axis.
Additionally, that kind of set up allows a nearfield listening with 3D image effect.
Best regards from Paris, France
Jean-Michel Le Cléac'h
Jean
That's anyway how BD-Design recommends their Oprhean Mk2 horns for BMS4592 to be used - crossed in front of the listener. In a small room you can get a good result in any other way.
Tom, agree with all you say. What I meant with the horn determining dispersion is that you need a smaller horn for the HF, or a diffraction slot to control dispersion with the BMS dual concentric drivers.
Large horns and 2" exit drivers have their benefits. Incredibly gorgeous midrange is one. But they certainly have their problems. BMS has perhaps fixed the bandwidth issue -- there is no droop or breakup (maybe) like the classic large format drivers, but dispersion is still a problem.
If you choose not to use the 2" driver up high, then the crossover becomes a problem. Too much separation between the sources both vertically and in depth. The compromise I've landed on is a 1.4" driver and the SEOS-24.
I have tried the 4550 and it is an excellent driver. But it just doesn't have the LF extension I'm looking for. TAD-2001 is a good candidate, but I'm not paying that much for a driver.
Of course, your Synergies will solve all these problems, and I wanted to try and build a Synergy this summer. Alas, I haven't been able to the time. I'm hoping BWaslo will provide an update soon on his new build. And one can always hope for a kit from DSL
Last edited:
In the quite small listening room where the single 15" 416-Alnico and AH425/Radian745 were auditioned, the optimum listening angle seemed to be with half or less of the phase plug visible from the sitting position, which resulted in the axis crossing about the same distance in front of listener as mentioned in Jean-Michel's post above.
Since my strongest design influence has been the BBC research of the late Fifties, Sixties, and Seventies, I've always followed the BBC-stereo practice of the axis crossing a moderate distance in front of the listener. The exact distance is a function of dispersion and tolerance of diffraction artifacts.
I started designing low-diffraction loudspeakers in 1979, as a result of building a prototype for Audionics that looked like a pair of really big vitamin capsules. That was a fun experiment: if the room was darkened, you might actually walk into the speakers while they were playing, since the music seemed to float in the air without any apparent physical source.
That taught me that the apparent "pinning" of music to the loudspeaker is a result of diffraction around the sharp edges of the front baffle; if you reduce the diffraction below a perceptual threshold, not only does a lot of tonal coloration go away, but the spatial impression of sound coming from a speaker-like enclosure goes away too.
The comb-filtering of diffraction (as seen by ripples on the FR graph) is the least of the problems with diffraction; the time-domain reflections are far worse, and severely degrade the spatial impression in stereophonic playback.
This is also why I never seriously audition a loudspeaker in (single-speaker) mono; the spatial impression is completely different with a single-speaker presentation, and there's no way of telling how bad diffraction is. In stereo, if you're familiar with the sound of a very low diffraction loudspeaker, you can easily hear how much diffraction is impairing the spatial presentation. Most commercial speakers fail this test rather badly, but then again, very few reviewers and audiophiles have ever heard a loudspeaker with low diffraction and a fast-decay time signature.
Although magnetic-planars and electrostats can have fast-decay time signatures, the planar shape with hard edges isn't necessarily low diffraction. So the sound often seems to come from somewhere behind the planar, as if it were a window, instead of floating in the air, like it does with a truly low-diffraction loudspeaker.
Conversely, you can have a speaker with little or no diffraction, like the MBL and Walsh omni-radiators, but the time signature can be stretched-out and lengthy, thanks to diaphragm and structural resonances. This results in a diffuse and "swimmy" spatial impression. Room treatments can reduce the "swimmy", too-large sensation, but cannot improve the time response of the loudspeaker itself. (A true-omni radiator is a particularly good candidate for digital FIR correction, since the corrected time response is good in all (horizontal) directions, unlike a conventional array of drivers.)
The goal with this system is to have a rapid settling time, low diffraction, and high efficiency, with a slightly warm spectral balance (about 1~2 dB tilt from 100 Hz to 10 kHz).
Last edited:
I did just such a comparison using the SEOS-18 horn.
1 inch CD Comparison (SEOS18)
The BMS has slightly less ripple, more output and less distortion in the lower octaves. Both are excellent, IMO. I was using them with a 800 Hz LR 6th order target. This is too low for either driver, even in the home.
Attachments
Last edited:
Hi Lynn, all
I can expand on Jean-Michel’s observation so far as aiming and the Stereo image.
If you think of a polar pattern as an “equal loudness contour”, most loudspeakers or horns that have directivity, have the maximum level at the zero zero on axis. In the olden days in commercial sound, the rule was you put the speaker in the air and aim the speaker at the farthest seat
Once the “polar pattern” came to be, the next step was to figure out the optimum height and angle so that the listeners were partly on the bottom side of the lobe. If one had the lobe angle and height “right” the spl in the audience plane did not change with the inverse square law but much slower.
In the listening room, if you have a couch, one can aim he right speaker at the left seat and vis versa and this puts the R listener in the R seat in a slightly lower level than normal and since part of the stereo image is loudness based (as in a pan pot say), one has a wider sweet spot.
I would offer too that your impression of “sound through a window” IS what one wants if reproducing a stereo image or in commercial sound maximizing voice intelligibility (the latter is a property that can be measured with a STIpa measurement).
The reason is the radiation of spatial q’s are what let you hear the source’s physical depth if your eyes are closed, are not part of the input signal. This works best with one speaker. Take the loudspeakers physical distance clues away and the sound, sounds “up close” on a dry mic recording or off somewhere in a reverberant recording. Listen in stereo and the speaker with spatial q’s produces a mono signal as a center phantom AND a right and left source. Take away the loudspeakers spatial q’s and you have a strong mono phantom and no apparent right or left source.
If you had planar speakers well enough behaved to produce that lack of spatial identity, then in spite of whatever edge diffraction, the direct energy is far greater than the spatial ques. Keep in mind that to the degree such a source can replicate true piston motion; it would have a LARGE amount of directivity up high. Also, it’s what you measure where you’re sitting that reflects what you hear where you’re sitting, a high degree of directivity delivers something to the listening position much closer to what one measured at one meter say.
While you can’t measure stereo imaging as that is entirely subjective, what one can predict is intelligibility of words as that depends on preserving the signal information / keeping it intact at the listening position.
The STIpa measurement I mentioned here and a number of posts elsewhere here is an audio Modulation Transfer Function Measurement, actually the STIpa figure is the result of 7 MTF’s across the speech band.
Like with optics, an MTF is a measurement of resolution, “how fast” the image can transition from black to white. In the audio version it is the rate of amplitude modulation of what ever signal F one is interested in. If one measures the MTF’s of a loudspeaker (with say ARTA), one picks the carrier F and then it measures how deep the modulation depth is at higher and higher rates.
Things like ringing, energy storage as well as distortion and reflected sounds (and more) limit the rate of modulation. This MTF view is a relatively new way of looking at loudspeakers, there is no “figure of quality” yet it is easy to tell if something you did, made it better or worse. On the other hand, the STIpa measurement which is a number of MTF’s has been proven to be a reliable language independent way of predicting the intelligibility in larger scale sound. In the home, one can clearly see the difference between different loudspeakers such as you mentioned.
I couldn’t agree more about the point source, it doesn’t have to be Omni, it can have directivity but it HAS TO radiate simply as if it’s origin were an acoustically small source and was producing simple (clue free) spherical radiation.
At work, our business is large sound systems but I have applied the same hifi goals to those systems. AS you might imagine, as the loudspeaker system gets larger and the acoustic power goes up, this is more and more difficult to do, but it is still possible. If you go to football games, you have heard just “how good” the sound quality is when you have 50,000 - 100,000 seats and a concert sound system. Anyway, if you want to hear some “single point sources” in a larger room, you can get some idea with headphones and the Video’s below (from LSU tiger stadium). Using that CD radiation lobe and choosing aligning the pattern, the spl only varies about + - 3dB over the entire stadium and it “sounds the same everywhere”.
Best,
Tom Danley
https://www.dropbox.com/sh/nmmmdtum82lyig9/QnEaYWlnDE
I can expand on Jean-Michel’s observation so far as aiming and the Stereo image.
If you think of a polar pattern as an “equal loudness contour”, most loudspeakers or horns that have directivity, have the maximum level at the zero zero on axis. In the olden days in commercial sound, the rule was you put the speaker in the air and aim the speaker at the farthest seat
Once the “polar pattern” came to be, the next step was to figure out the optimum height and angle so that the listeners were partly on the bottom side of the lobe. If one had the lobe angle and height “right” the spl in the audience plane did not change with the inverse square law but much slower.
In the listening room, if you have a couch, one can aim he right speaker at the left seat and vis versa and this puts the R listener in the R seat in a slightly lower level than normal and since part of the stereo image is loudness based (as in a pan pot say), one has a wider sweet spot.
I would offer too that your impression of “sound through a window” IS what one wants if reproducing a stereo image or in commercial sound maximizing voice intelligibility (the latter is a property that can be measured with a STIpa measurement).
The reason is the radiation of spatial q’s are what let you hear the source’s physical depth if your eyes are closed, are not part of the input signal. This works best with one speaker. Take the loudspeakers physical distance clues away and the sound, sounds “up close” on a dry mic recording or off somewhere in a reverberant recording. Listen in stereo and the speaker with spatial q’s produces a mono signal as a center phantom AND a right and left source. Take away the loudspeakers spatial q’s and you have a strong mono phantom and no apparent right or left source.
If you had planar speakers well enough behaved to produce that lack of spatial identity, then in spite of whatever edge diffraction, the direct energy is far greater than the spatial ques. Keep in mind that to the degree such a source can replicate true piston motion; it would have a LARGE amount of directivity up high. Also, it’s what you measure where you’re sitting that reflects what you hear where you’re sitting, a high degree of directivity delivers something to the listening position much closer to what one measured at one meter say.
While you can’t measure stereo imaging as that is entirely subjective, what one can predict is intelligibility of words as that depends on preserving the signal information / keeping it intact at the listening position.
The STIpa measurement I mentioned here and a number of posts elsewhere here is an audio Modulation Transfer Function Measurement, actually the STIpa figure is the result of 7 MTF’s across the speech band.
Like with optics, an MTF is a measurement of resolution, “how fast” the image can transition from black to white. In the audio version it is the rate of amplitude modulation of what ever signal F one is interested in. If one measures the MTF’s of a loudspeaker (with say ARTA), one picks the carrier F and then it measures how deep the modulation depth is at higher and higher rates.
Things like ringing, energy storage as well as distortion and reflected sounds (and more) limit the rate of modulation. This MTF view is a relatively new way of looking at loudspeakers, there is no “figure of quality” yet it is easy to tell if something you did, made it better or worse. On the other hand, the STIpa measurement which is a number of MTF’s has been proven to be a reliable language independent way of predicting the intelligibility in larger scale sound. In the home, one can clearly see the difference between different loudspeakers such as you mentioned.
I couldn’t agree more about the point source, it doesn’t have to be Omni, it can have directivity but it HAS TO radiate simply as if it’s origin were an acoustically small source and was producing simple (clue free) spherical radiation.
At work, our business is large sound systems but I have applied the same hifi goals to those systems. AS you might imagine, as the loudspeaker system gets larger and the acoustic power goes up, this is more and more difficult to do, but it is still possible. If you go to football games, you have heard just “how good” the sound quality is when you have 50,000 - 100,000 seats and a concert sound system. Anyway, if you want to hear some “single point sources” in a larger room, you can get some idea with headphones and the Video’s below (from LSU tiger stadium). Using that CD radiation lobe and choosing aligning the pattern, the spl only varies about + - 3dB over the entire stadium and it “sounds the same everywhere”.
Best,
Tom Danley
https://www.dropbox.com/sh/nmmmdtum82lyig9/QnEaYWlnDE
Hi Tom, I'm a little confused by the phrase "spatial q’s". Do you mean "spatial cues" in the subjective sense, or are does "q" refer to loudspeaker directivity in some way?
Going back to the "picture window" metaphor, I've heard speakers render the overall spatial impression (not the same as imaging, which refers to direct instrument locations) in several different ways.
If the speaker has serious issues with diffraction and long time decays, there's not much impression of a space at all. Everything hangs on a clothesline stretched between the two speakers, with no depth, height, or extra-width at all. Each instrument is a paper-thin cutout. If the driver layout is especially inept (old loudspeakers with scattershot driver locations and no mirror-imaging), even that is not accomplished, and the instruments blur and move around.
Modern designs with average time decays and slight attention paid to diffraction create a moderate space impression; extra-width impression happens on a few recordings with antiphase content (typically reverb), there's a bit of height (which is an artifact of two-speaker playback and not in the original recording), and depth impression correlates with lack of diffraction and a reasonable absence of low-level Class AB switching artifacts (which destroy space impression). MP3 and other lossy algorithms are impressively good at removing depth impression as well; I'm guessing the algorithm treats low-level, broadband reverb as noise, and removes it.
The better planar loudspeakers present a "through-the-window" impression, thanks to fast decay times, which preserves the time signature of the more important reverb cues in the recording, which in turn allows the ear/brain to cross-correlate the reverb cues with the first-arrival sounds, and derive distance information.
Point-source radiators with very low diffraction signatures can pull off an additional trick: disappearing completely as a sound source, since there's no cabinet-edge reflection for the ear/brain to detect. With the cabinet-edge reflection removed, there's no apparent cabinet, either in the original recording (microphone housings are far smaller than loudspeakers), or in the playback system. The picture-window impression disappears, and whole front half of the room merges with the acoustic of the recording environment.
It's been my experience that intelligibility is partly orthogonal to spatial impression: I've heard speakers with exceptional intelligibility and almost no spatial qualities at all, and speakers with truly remarkable spatial realism and only moderately good intelligibility. There are a handful of loudspeakers that combine both.
Last edited:
Hi Lynn
Yes, sorry i was channeling my old boss who was a WWII acoustical from England and used that expression, read Cues.
It isn't the first time a couple of his expressions raise eyebrows haha.
Best,
Tom
Beyond the Ariel (as the thread unravels)
Lynn, and all others:
I have found relevance in this thread because I am also designing my "next" speaker system. I really want it to be my last. I'm not exactly sure where Lynn is at in his design process, but I wanted to introduce a thought, and that is that the mid-bass section of any speaker system (multi-way, of course) is going to be the predominant determination of sound quailty.
For discussion purposes, let's assume mid-bass includes lower mid-range such that 60-500Hz is what I am referring to. I have concluded that, in order to provide a realistic presentation, this area has to then again be covered by multiple drivers, perhaps even some sort of mechanical crossover as I feel a 4-way is the limit of complexity I care to indulge in.
Here's my thoughts: Since I am using a dedicated mid-range horn, it makes sense to me to also use a horn for lower mid-range. I am planning on using my pair of Altec 515-8G's, but since I am somewhat space limied I do not want a long horn;only enough to cover 150-600Hz max. (yes I may move my 500Hz x-over up to 600Hz, as I reckognize the center of energy in a musical programme as being 300Hz; 150-600 covers the octave either side). Because of this, I'll need more drivers to carry the "bass" section of the mid-bass. Perhaps two more 15 inchers per side, as I love the Grand Piano, but I really don't want to fool with the enourmous amount of wood/and/size required for a horn that by all rights, unless massively dampened, will have the propensity to resonate.
So, where does this lead us to, in, "Beyond the Ariel" ?
Lynn, and all others:
I have found relevance in this thread because I am also designing my "next" speaker system. I really want it to be my last. I'm not exactly sure where Lynn is at in his design process, but I wanted to introduce a thought, and that is that the mid-bass section of any speaker system (multi-way, of course) is going to be the predominant determination of sound quailty.
For discussion purposes, let's assume mid-bass includes lower mid-range such that 60-500Hz is what I am referring to. I have concluded that, in order to provide a realistic presentation, this area has to then again be covered by multiple drivers, perhaps even some sort of mechanical crossover as I feel a 4-way is the limit of complexity I care to indulge in.
Here's my thoughts: Since I am using a dedicated mid-range horn, it makes sense to me to also use a horn for lower mid-range. I am planning on using my pair of Altec 515-8G's, but since I am somewhat space limied I do not want a long horn;only enough to cover 150-600Hz max. (yes I may move my 500Hz x-over up to 600Hz, as I reckognize the center of energy in a musical programme as being 300Hz; 150-600 covers the octave either side). Because of this, I'll need more drivers to carry the "bass" section of the mid-bass. Perhaps two more 15 inchers per side, as I love the Grand Piano, but I really don't want to fool with the enourmous amount of wood/and/size required for a horn that by all rights, unless massively dampened, will have the propensity to resonate.
So, where does this lead us to, in, "Beyond the Ariel" ?
AE
Thanks, Face
You have offered a relatively simple solution to what I have made complicated. I'm sure that driver would work fine in a smaller scaled loudspeaker system than what I am describing.
IMHO opinion, it will take at least 2x 15's per-side to generate enough acoustic power in the mid-bass to re-produce a Grand Piano realistically.
Thanks, Face
You have offered a relatively simple solution to what I have made complicated. I'm sure that driver would work fine in a smaller scaled loudspeaker system than what I am describing.
IMHO opinion, it will take at least 2x 15's per-side to generate enough acoustic power in the mid-bass to re-produce a Grand Piano realistically.
From my experience a pair per channel of high quality (four is better but has other compromises) 15's will not keep up with a good recording or a quality compression driver crossed over 600 Hz or lower. Sure the PA systems sold today are configured that way because we all don't want to haul around a truck full of bass horns. A good bass horn loaded with a high quality driver is where to look and what to build if you really want to produce the low mid range and bass with realism. If you crossover low enough to the compression driver you can fold the horn to save space. Another take is a VOTT or JBL front loaded horn type system with the bass reflex ports in conjunction with multiple 15's or 18's to supplement below 100 Hz or so. A carefully matched example of this is a better compromise then just using direct radiator 15's in my experience. The lower you go with a front horn the more it approaches realism.
too tired when I wrote
This is exactly what I was trying to describe in my ealier post. I was just too tired when I wrote it to explain it right. I plan on using one horn loaded 15, an
Altec 515-8G AND in addition at least one additional pair of 15's EQ'd such that they only augment where the Altec unloads below and down to where it crosses to the dedicated SUBS.
We pretty much agree, except I think there is a point to where a horn is too large and we arrive at the law of dimishing returns. Just too big, bulky and
too much wood suseptable to resonating.
A few days ago, at a local audiophile coffee meet, my friend pointed out that 4 15's per side have the equivalent cone radiating area as a VOTT horn:
540 square inches (approx). My thoughts were........ BINGO !!
This is exactly what I was trying to describe in my ealier post. I was just too tired when I wrote it to explain it right. I plan on using one horn loaded 15, an
Altec 515-8G AND in addition at least one additional pair of 15's EQ'd such that they only augment where the Altec unloads below and down to where it crosses to the dedicated SUBS.
We pretty much agree, except I think there is a point to where a horn is too large and we arrive at the law of dimishing returns. Just too big, bulky and
too much wood suseptable to resonating.
A few days ago, at a local audiophile coffee meet, my friend pointed out that 4 15's per side have the equivalent cone radiating area as a VOTT horn:
540 square inches (approx). My thoughts were........ BINGO !!
