Diffraction, Stored Energy, and Illusions
Not off-topic as far as I'm concerned. The whole subject of dipole and cardioid speakers needs much more work and attention. For one thing, almost everyone is using idealized models for polar patterns that don't pay enough attention to diffraction effects at the edges of the baffle. As mentioned below, a flat baffle is merely a special case of a conical horn - and like a horn, suffers from diffractive energy storage which falls in the critical 0-2 mSec time window.
Similarly, speakers with vertical-array drivers suffer from non-coincident arrival time, although this at least is a one-shot error versus the slowly decaying succession of reflections created by edge diffraction. Although difficult to measure (it just barely appears in MLS waterfall displays), the ear is extremely sensitive to this succession of closely-spaced echoes. If you're curious, they can be auditioned directly by playing pink-noise at a modest level, and walking around the front of a conventional speaker. If you listen closely, you'll hear what appear to be small tweeters spraying noise off the corners of the cabinet - this is at a maximum at a 45-degree angle with respect to the front panel.
I think of panel-edge diffraction being a bit like the partially-reflecting Brewster windows in a gas laser, which reflect light back and forth as the energy is built up with each pass inside the laser. In a speaker, thank goodness, nothing builds up the energy with each pass, but the successive reflections do take a while to die down below audibility.
Good reminder that a cone has horn-like behaviour at the top of the passband, with the de facto driver being a ring radiator close to the voice coil, a hard dust cap, or an unfortunate combination of both.
This image certainly calls to mind the famous RCA LC-1A quasi-coax driver and the clever dimples that HF Olson put on the cone. When I was at the last Rocky Mountain show, I found out the principals of Cogent actually owned (!) a pair of LC-1A drivers, and I urged them - in the strongest terms - to put them back in production.
The patents on all of the classics - LC-1A, 604 Duplex, and the Tannoy Dual Concentric - ran out decades ago, so there's no reason at all they can't be built with much, much better materials. My dream driver would have a hemp-composite cone, diamond-diaphragm compression driver, and field-coil magnets. (Oh, did I mention that particular LC-1A is the RCA lab prototype and uses field-coil magnets?)
I've been thinking quite a lot about dispersion characteristics - most horn-fans don't know it, but unless they've got an all-horn system, the dispersion is not in fact constant with frequency. The direct-radiation bass unit radiates over 360 degrees (omnidirectional), then gradually narrows down to 180 to 90 degrees (depending on crossover point), then hands off to a horn with the specified radiation pattern (typically 120 to 90 degrees), which then narrows further as the frequency increases (unless it's a constant-directivity horn, which have their own problems). The narrow sidelobes that appear in the polar-pattern curves also appear as ripples in the time-domain, frequency-domain, and impedance curves - this is a consequence of antenna theory, where errors in one domain must appear in the other domains as well.
A conventional direct-radiator box also has a 360 to 90 degree transition, but the frequencies are much higher. The typical dome tweeter is 180 degrees over most of its useful range, and the direct-radiator woofer, just like the direct-radiator woofer in a horn system, makes the 360 to 180 to 90 degree transition over its working passband. And like a horn system, the radiation pattern gets very sloppy and complex at and above the crossover frequency, as one or more main lobes move up and down with frequency, nulls sweep close to the listener position, and many small sidelobes are created by cabinet-edge diffraction. Most of the "phase trims" we like to do with the HF crossover are actually doing little more that moving the main lobes up and down, and pushing the nasty-sounding null zone further away, or at least not on top of the floor reflection.
With a dipole or cardioid system, we're not starting with 360 degree radiation from the bass box, but 180 degrees instead. (The choice of subwoofer - dipole or closed-box - is a separate issue.) There's a better chance of keeping the radiation pattern fairly constant, especially if we twiddle with felt lining behind to the driver to adjust the rear energy level (thus choosing between figure-8, limacon, or cardioid radiation patterns). Not to mention standing waves from inside the box no longer exist - now we just have to contend with standing waves on the cone, which occur several octaves higher. This is where Bud Purvine's ENBL patterns, or other termination techniques, come in handy.
Room size is an essential consideration here. When the sound is several wavelengths smaller than the smallest room dimension, the listener can separate the sound into direct-arrival and total-room-energy components. At wavelengths approaching room size, direct-arrival and total-room-energy merge together. What sounds unnatural are the direct-arrival and total-room-energy spectra diverging from each other, since this is something that rarely happens in nature (or a concert hall).
As mentioned earlier, the ear/brain/mind processes sound in different "slots" using cross-correlation techniques (the first 20 or so reflections are compared to the direct arrival), and can compare the direct-arrival spectra to the earliest reflections. This happens automatically all the time - in fact, it's working when you're asleep or awake, constantly analyzing the environment for threat or safety. The environment-processor is not only connected to the frontal lobes (the "adult" sense of self - responsible for planning-ahead) but to the limbic system, which reacts with emotion and activates the fight-or-flight system.
When a hifi system tampers with the early-arrival information, the ear/brain/mind discounts the artificiality and categorizes the sound as just another artificial creation, mimicking life but most certainly not the real thing, just as you'd never confuse a TV picture or movie with real-life. When you get that hair-raising "spooky" feeling with a hifi system, that means early-arrival information is arriving intact, and the auditory illusion of being somewhere else is working. The machine is really generating an illusion, something HDTV and IMAX are still a long way from doing successfully.
The reason that movies and TV shows "pull you in" is the light hypnotic trance generated by the quiet, darkened room - a modern replica of the environment where thousands of previous generations listened to stories told round the tribal campfire. For a really fun experience, listen to the ZBS radio-dramas on a top-quality system - the illusion of "being there" is far more intense than watching a movie, as your ears open up to the environmental sounds of Morocco at night, and the narrator weaves the story-line.
Cloth Ears said:
Not off-topic as far as I'm concerned. The whole subject of dipole and cardioid speakers needs much more work and attention. For one thing, almost everyone is using idealized models for polar patterns that don't pay enough attention to diffraction effects at the edges of the baffle. As mentioned below, a flat baffle is merely a special case of a conical horn - and like a horn, suffers from diffractive energy storage which falls in the critical 0-2 mSec time window.
Similarly, speakers with vertical-array drivers suffer from non-coincident arrival time, although this at least is a one-shot error versus the slowly decaying succession of reflections created by edge diffraction. Although difficult to measure (it just barely appears in MLS waterfall displays), the ear is extremely sensitive to this succession of closely-spaced echoes. If you're curious, they can be auditioned directly by playing pink-noise at a modest level, and walking around the front of a conventional speaker. If you listen closely, you'll hear what appear to be small tweeters spraying noise off the corners of the cabinet - this is at a maximum at a 45-degree angle with respect to the front panel.
I think of panel-edge diffraction being a bit like the partially-reflecting Brewster windows in a gas laser, which reflect light back and forth as the energy is built up with each pass inside the laser. In a speaker, thank goodness, nothing builds up the energy with each pass, but the successive reflections do take a while to die down below audibility.
ScottG said:Ah Lynn.. my favorite in the HiFi "world"..
While I'm generally no fan of horns .. consider that a good bit of a cone driver's bandwidth at higher freq's is in fact horn loaded. The larger the driver the lower the freq.. and the worse it is - Bastini's in particular. Just about any coax has a similar reality. Worse still, the "horn" for these cone drivers is moving around.
Heck.. even a baffle acts as a horn.
Also note that most pro drivers have increased distortion at lower spl's near the top of their passband (at higher freq.s) due to larger VC's (when compared to better hifi drivers). Both linear and nonlinear. Additionally, while they do exhibit less thermal compression.. they often suffer from mechanical compression that won't allow a lower freq. response anywhere near their max spl rating.
Despite the above.. I'm pretty much lock-step with your views. I just like hi eff. drivers better, more energetic/dynamic and IF they are of lower mass then they seem to offer greater clarity. Also, I've noted that linear decay is critical at least 9-10 db when factoring time - the shorter the better. Oddly though, HOW you achieve a cleaner decay is extremely important. Drivers with high internal loss and only moderate or worse ability to react to change, despite having *overall* cleaner decays (i.e. are "cleaner" further in time and level) sound significantly less clear. Likewise, drivers with acoustic resistance (particularly near time acoustic resistance), often via box stuffing and panel friction, kill ambiance and make the sound more monophonic (and "attached to the speaker").
I however specifically do NOT prefer a dipole pattern above the modal region - it creates a "sameness of sound" from recording to recording.. and in comparision to a really good enclosure falls short on depth. My preference is for a radial in the midrange and a wide dispersion pattern higher in freq. approaching 180 degrees (..when considering the front wall in a listening room). This does make several demands on both the "box" and the driver..
Good reminder that a cone has horn-like behaviour at the top of the passband, with the de facto driver being a ring radiator close to the voice coil, a hard dust cap, or an unfortunate combination of both.
This image certainly calls to mind the famous RCA LC-1A quasi-coax driver and the clever dimples that HF Olson put on the cone. When I was at the last Rocky Mountain show, I found out the principals of Cogent actually owned (!) a pair of LC-1A drivers, and I urged them - in the strongest terms - to put them back in production.
The patents on all of the classics - LC-1A, 604 Duplex, and the Tannoy Dual Concentric - ran out decades ago, so there's no reason at all they can't be built with much, much better materials. My dream driver would have a hemp-composite cone, diamond-diaphragm compression driver, and field-coil magnets. (Oh, did I mention that particular LC-1A is the RCA lab prototype and uses field-coil magnets?)
I've been thinking quite a lot about dispersion characteristics - most horn-fans don't know it, but unless they've got an all-horn system, the dispersion is not in fact constant with frequency. The direct-radiation bass unit radiates over 360 degrees (omnidirectional), then gradually narrows down to 180 to 90 degrees (depending on crossover point), then hands off to a horn with the specified radiation pattern (typically 120 to 90 degrees), which then narrows further as the frequency increases (unless it's a constant-directivity horn, which have their own problems). The narrow sidelobes that appear in the polar-pattern curves also appear as ripples in the time-domain, frequency-domain, and impedance curves - this is a consequence of antenna theory, where errors in one domain must appear in the other domains as well.
A conventional direct-radiator box also has a 360 to 90 degree transition, but the frequencies are much higher. The typical dome tweeter is 180 degrees over most of its useful range, and the direct-radiator woofer, just like the direct-radiator woofer in a horn system, makes the 360 to 180 to 90 degree transition over its working passband. And like a horn system, the radiation pattern gets very sloppy and complex at and above the crossover frequency, as one or more main lobes move up and down with frequency, nulls sweep close to the listener position, and many small sidelobes are created by cabinet-edge diffraction. Most of the "phase trims" we like to do with the HF crossover are actually doing little more that moving the main lobes up and down, and pushing the nasty-sounding null zone further away, or at least not on top of the floor reflection.
With a dipole or cardioid system, we're not starting with 360 degree radiation from the bass box, but 180 degrees instead. (The choice of subwoofer - dipole or closed-box - is a separate issue.) There's a better chance of keeping the radiation pattern fairly constant, especially if we twiddle with felt lining behind to the driver to adjust the rear energy level (thus choosing between figure-8, limacon, or cardioid radiation patterns). Not to mention standing waves from inside the box no longer exist - now we just have to contend with standing waves on the cone, which occur several octaves higher. This is where Bud Purvine's ENBL patterns, or other termination techniques, come in handy.
Room size is an essential consideration here. When the sound is several wavelengths smaller than the smallest room dimension, the listener can separate the sound into direct-arrival and total-room-energy components. At wavelengths approaching room size, direct-arrival and total-room-energy merge together. What sounds unnatural are the direct-arrival and total-room-energy spectra diverging from each other, since this is something that rarely happens in nature (or a concert hall).
As mentioned earlier, the ear/brain/mind processes sound in different "slots" using cross-correlation techniques (the first 20 or so reflections are compared to the direct arrival), and can compare the direct-arrival spectra to the earliest reflections. This happens automatically all the time - in fact, it's working when you're asleep or awake, constantly analyzing the environment for threat or safety. The environment-processor is not only connected to the frontal lobes (the "adult" sense of self - responsible for planning-ahead) but to the limbic system, which reacts with emotion and activates the fight-or-flight system.
When a hifi system tampers with the early-arrival information, the ear/brain/mind discounts the artificiality and categorizes the sound as just another artificial creation, mimicking life but most certainly not the real thing, just as you'd never confuse a TV picture or movie with real-life. When you get that hair-raising "spooky" feeling with a hifi system, that means early-arrival information is arriving intact, and the auditory illusion of being somewhere else is working. The machine is really generating an illusion, something HDTV and IMAX are still a long way from doing successfully.
The reason that movies and TV shows "pull you in" is the light hypnotic trance generated by the quiet, darkened room - a modern replica of the environment where thousands of previous generations listened to stories told round the tribal campfire. For a really fun experience, listen to the ZBS radio-dramas on a top-quality system - the illusion of "being there" is far more intense than watching a movie, as your ears open up to the environmental sounds of Morocco at night, and the narrator weaves the story-line.
Lynn,
It is really good reading your words again - some people just have a way of putting their ideas across so that others can understand them.
I can get this effect on some tracks in my home (in spite of pair of sealed woofers and a pair of Elacs x-ed at 120Hz - all boxes). But the tracks are all fairly light (Diana Krall, Keb 'Mo, Aaron Neville, and a recording I did of my brother and brother-in-law one night at a cafe). You can get the "you are there" feel with some of these. But I'm looking to the time my dog does more than look up (and you can hear him thinking "you're trying to fool me again, aren't you?") when I put on rainstorms or fireworks I've recorded off the internet. I guess I won't be able to play them again after that - but he's currently my arbiter of realism.
It is really good reading your words again - some people just have a way of putting their ideas across so that others can understand them.
Lynn Olson said:
I can get this effect on some tracks in my home (in spite of pair of sealed woofers and a pair of Elacs x-ed at 120Hz - all boxes). But the tracks are all fairly light (Diana Krall, Keb 'Mo, Aaron Neville, and a recording I did of my brother and brother-in-law one night at a cafe). You can get the "you are there" feel with some of these. But I'm looking to the time my dog does more than look up (and you can hear him thinking "you're trying to fool me again, aren't you?") when I put on rainstorms or fireworks I've recorded off the internet. I guess I won't be able to play them again after that - but he's currently my arbiter of realism
.Lynn;
Good to hear that you are on the mend.
I am a fan of your tube amps, and look forward to your new speaker.
With the Coax driver, do you think the Dipole to monopole / waveguide transition is important? There has been discussions of extra realism with a second tweeter.
I'm following this with interest.
Doug
Good to hear that you are on the mend.
I am a fan of your tube amps, and look forward to your new speaker.
With the Coax driver, do you think the Dipole to monopole / waveguide transition is important? There has been discussions of extra realism with a second tweeter.
I'm following this with interest.
Doug
Hi Lynn,
I have some questions about drivers and mounting surfaces, for anyone with information they would care to share.
1. Would it be helpful to narrow the null zone between front and back sides of the drivers and would this benefit or detract from their activities after mounting?
2. Assuming that it would not be beneficial to narrow that null zone, would it help to slow the back wave down with a light mass damping?
3. If the mass damping on the back is beneficial would speeding up the front even further be beneficial?
4. Has there been any even semi rigorous evaluation of mounting board shapes and perhaps phase directing side panels. used to force a wider dispersion of the front wave VS the back?
5. Assuming the existence of technology for making the edge termination effectively invisible to an expanding wave front and it's lateral wave creator, are there analogs in other wave propagation devices, that are used to bend and shape emitted waves, that we might draw upon to refine our ideas about mounting boards and cone drivers.
6. Does anyone know what drivers Bob Carver used in his large dipole speakers and are there any sites devoted to this speaker where folks are dissecting Bob's thoughts?
Bud
I have some questions about drivers and mounting surfaces, for anyone with information they would care to share.
1. Would it be helpful to narrow the null zone between front and back sides of the drivers and would this benefit or detract from their activities after mounting?
2. Assuming that it would not be beneficial to narrow that null zone, would it help to slow the back wave down with a light mass damping?
3. If the mass damping on the back is beneficial would speeding up the front even further be beneficial?
4. Has there been any even semi rigorous evaluation of mounting board shapes and perhaps phase directing side panels. used to force a wider dispersion of the front wave VS the back?
5. Assuming the existence of technology for making the edge termination effectively invisible to an expanding wave front and it's lateral wave creator, are there analogs in other wave propagation devices, that are used to bend and shape emitted waves, that we might draw upon to refine our ideas about mounting boards and cone drivers.
6. Does anyone know what drivers Bob Carver used in his large dipole speakers and are there any sites devoted to this speaker where folks are dissecting Bob's thoughts?
Bud
Re: Diffraction, Stored Energy, and Illusions
Hi Lynn,
Sorry to hear about your leg! Hope you mend quickly. I had a nasty fall not far from you in Niwot when I lived in Colorado. Fortunately, I landed squarely on my right kidney and didn't break anything.
I'm somewhat surprised to see you lusting after a diamond diaphragm given your stance on carbon-fibre and kevlar woofers. The same high Q resonant behaviours happen with stiff CD diaphragms. I reckon some people can hear this hence the continued popularity of aluminium over titanium and berylium. The polypropolene of CD driver materials, if you will.
Jeff
Lynn Olson said:
Hi Lynn,
Sorry to hear about your leg! Hope you mend quickly. I had a nasty fall not far from you in Niwot when I lived in Colorado. Fortunately, I landed squarely on my right kidney and didn't break anything.
I'm somewhat surprised to see you lusting after a diamond diaphragm given your stance on carbon-fibre and kevlar woofers. The same high Q resonant behaviours happen with stiff CD diaphragms. I reckon some people can hear this hence the continued popularity of aluminium over titanium and berylium. The polypropolene of CD driver materials, if you will.
Jeff
Big Thanx to JoshK
Many thanks to JoshK for the link to the Isisris on the HTGuide forum:
here
This is a wonderful thread - and the gradual evolution from a Linkwitz-style speaker to a more pro-audio approach is interesting and entertaining. I personally feel the high-end driver manufacturers have dropped the ball over the last 15 years, ignoring the growing clamor for higher-efficiency drivers, and just making more and more exotic-coned low-efficiency stuff at prices that now equal serious pro-audio gear.
Although now mentioned explicitly in the previous threads, yes, I think edge treatment in a dipole is a good thing. In the Ariel the difference between 3/4" radiusing and 1 3/4" radiusing on the cabinet edges was clearly audible as a doubling of soundstage depth (no exaggeration here), several feet more of extra-width effects when present on the recording, and a substantial improvement of vocal timbre. Yet the difference was only just barely visible on MLSSA at the highest resolution I could get.
I surmise that a kind of teardrop shape on the edge of the flat panel might be ideal, but failing that, a simple round-over on the front edge. Maybe a 3" diameter cardboard tube glued to the edge of the panel might do the job. The overall shape of the front panel would mimic the profile of the Apogee, similar to the Isisris on the thread referenced above.
The kind of diffraction-profiling I've done for years would very much apply to the new speaker. Golden-section ratio for the centerline-to-L/R-edge spacing (1:1.618), slanted edges, and slanting the baffle backward so the arrival times from the bass-reinforcement woofer and fullrange driver are synchronized. The rearward slant is required because the lowpass inductor for the bass-reinforcement driver delays the electrical signal to the driver 1/4 wavelength at the -3dB frequency, acoustically moving the driver back in space several inches.
The coax itself is going to need modification. All the ones I've seen so far have the conventional horn High-Order-Modes (HOM) as described by Geddes, as shown by the ripples in the impedance curve for the horn. These ripples correspond to reflections in the time domain and variations in the polar pattern when the horn goes in and out of the primary modes.
Geddes technique relies on partially filling the horn with foam, with an unspecified internal profile for the foam inside the horn. (Somehow I doubt the entire horn is filled, I suspect the foam depth is greatest in the center and decreases towards the edge, thus compensating for narrowing directivity at the highest frequencies.)
But ... there are other ways of improving the acoustic termination at the horn mouth (the edge of the driver for a coax). Triangular strips of felt coming to a point deep inside the horn, surface treatments of the driver, and avoiding the Altec Duplex style of horn would all help the horn-termination issue.
The Duplex and its successors represent a worst-case solution, with a sharp 180-degree reflection from the horn edge sitting only an inch or so away from the partially reflecting surface of the cone. This degrades the HOM of the coax horn IN ADDITION TO partially shading the cone surface, which creates additional standing waves on the bass cone. So the bass cone and coax horn get in trouble in the same frequency range, about 1 kHz to 4 kHz. This unfortunately falls right in the middle of the ear/brain/mind's most sensitive frequency range to frequency response aberrations (1/2 dB threshold), distortion, and localization errors.
That's why I lean towards the Tannoy and Tannoy-alike compromise - at least the horn termination doesn't have any really sharp edges, and the cone isn't shaded by the horn. People who design loudspeakers keep forgetting these things operate at the speed of sound, and sharp edges anywhere create big problems in the time domain (just as they do for airplanes).
My mental model for a loudspeaker is to visualize the spherical wavefront that a firecracker would create as it expands out of the voice-coil region, and then think of all the reflections that occur as the wavefront grows larger and larger, eventually expanding past the front face of the speaker and into the room. Every imperfection and boundary creates secondary reflections, what you could think of as parasitic virtual sound sources that create ripples in the time, frequency, and polar-pattern domains.
The most undesirable scenario are full-blown standing waves, which appear when reflecting surfaces are parallel and have very hard or sharp edges - this happens on the inside of a spherical or cubical enclosure, horns with hard reflecting phase plugs and sharp edges on the horn mouth (or with internal kinks in the profile), or a front-panel baffle that has sharp edges (or worse, overhanging surfaces) and a square or rectangular profile.
When you get standing waves, ripples appear in the impedance and frequency-response curves, a series of reflections appears in the time domain, and the polar pattern has sharp, very narrow sidelobes. All of these are VERY audible as colorations, and don't respond well to equalization because the frequency correction actually makes the time error worse. (You can't correct a time error with a frequency correction.)
This kind of thing gets glossed over when the designer doesn't have access to time-domain measuring equipment - I don't want to start a flame-war, dammit, but the old-school horn guys from the Fifties were the worst here, ignoring time domain for decades after it became important to the hifi guys. There are still plenty of famous hifi designers today who ignore the time domain, or won't take a good look at the MLS waterfall curves - this has became an odd sort of ideology in the loudspeaker biz.
What has muddies the waters are the linear-phase proponents, who make a big deal out of perfect square waves. This has more to with choice of crossover (low-order) than cleaning up the time domain. Pretty-looking square waves and clean time domain response are not in fact the same - you can build a speaker with decent square-wave response that still has lots of problems with energy storage and standing waves - and in fact most square-wave-perfect commercial loudspeakers have indifferent treatment of standing waves in the cabinet and edge diffraction problems.
I avoid both the frequency-response-only and perfect-square-wave camps: I'm much more interested in reducing energy storage, standing waves, and diffraction. This comes back to driver selection and modification, or custom drivers when possible, edge treatment, and paying attention to arrival times at the listening position.
Many thanks to JoshK for the link to the Isisris on the HTGuide forum:
here
This is a wonderful thread - and the gradual evolution from a Linkwitz-style speaker to a more pro-audio approach is interesting and entertaining. I personally feel the high-end driver manufacturers have dropped the ball over the last 15 years, ignoring the growing clamor for higher-efficiency drivers, and just making more and more exotic-coned low-efficiency stuff at prices that now equal serious pro-audio gear.
Although now mentioned explicitly in the previous threads, yes, I think edge treatment in a dipole is a good thing. In the Ariel the difference between 3/4" radiusing and 1 3/4" radiusing on the cabinet edges was clearly audible as a doubling of soundstage depth (no exaggeration here), several feet more of extra-width effects when present on the recording, and a substantial improvement of vocal timbre. Yet the difference was only just barely visible on MLSSA at the highest resolution I could get.
I surmise that a kind of teardrop shape on the edge of the flat panel might be ideal, but failing that, a simple round-over on the front edge. Maybe a 3" diameter cardboard tube glued to the edge of the panel might do the job. The overall shape of the front panel would mimic the profile of the Apogee, similar to the Isisris on the thread referenced above.
The kind of diffraction-profiling I've done for years would very much apply to the new speaker. Golden-section ratio for the centerline-to-L/R-edge spacing (1:1.618), slanted edges, and slanting the baffle backward so the arrival times from the bass-reinforcement woofer and fullrange driver are synchronized. The rearward slant is required because the lowpass inductor for the bass-reinforcement driver delays the electrical signal to the driver 1/4 wavelength at the -3dB frequency, acoustically moving the driver back in space several inches.
The coax itself is going to need modification. All the ones I've seen so far have the conventional horn High-Order-Modes (HOM) as described by Geddes, as shown by the ripples in the impedance curve for the horn. These ripples correspond to reflections in the time domain and variations in the polar pattern when the horn goes in and out of the primary modes.
Geddes technique relies on partially filling the horn with foam, with an unspecified internal profile for the foam inside the horn. (Somehow I doubt the entire horn is filled, I suspect the foam depth is greatest in the center and decreases towards the edge, thus compensating for narrowing directivity at the highest frequencies.)
But ... there are other ways of improving the acoustic termination at the horn mouth (the edge of the driver for a coax). Triangular strips of felt coming to a point deep inside the horn, surface treatments of the driver, and avoiding the Altec Duplex style of horn would all help the horn-termination issue.
The Duplex and its successors represent a worst-case solution, with a sharp 180-degree reflection from the horn edge sitting only an inch or so away from the partially reflecting surface of the cone. This degrades the HOM of the coax horn IN ADDITION TO partially shading the cone surface, which creates additional standing waves on the bass cone. So the bass cone and coax horn get in trouble in the same frequency range, about 1 kHz to 4 kHz. This unfortunately falls right in the middle of the ear/brain/mind's most sensitive frequency range to frequency response aberrations (1/2 dB threshold), distortion, and localization errors.
That's why I lean towards the Tannoy and Tannoy-alike compromise - at least the horn termination doesn't have any really sharp edges, and the cone isn't shaded by the horn. People who design loudspeakers keep forgetting these things operate at the speed of sound, and sharp edges anywhere create big problems in the time domain (just as they do for airplanes).
My mental model for a loudspeaker is to visualize the spherical wavefront that a firecracker would create as it expands out of the voice-coil region, and then think of all the reflections that occur as the wavefront grows larger and larger, eventually expanding past the front face of the speaker and into the room. Every imperfection and boundary creates secondary reflections, what you could think of as parasitic virtual sound sources that create ripples in the time, frequency, and polar-pattern domains.
The most undesirable scenario are full-blown standing waves, which appear when reflecting surfaces are parallel and have very hard or sharp edges - this happens on the inside of a spherical or cubical enclosure, horns with hard reflecting phase plugs and sharp edges on the horn mouth (or with internal kinks in the profile), or a front-panel baffle that has sharp edges (or worse, overhanging surfaces) and a square or rectangular profile.
When you get standing waves, ripples appear in the impedance and frequency-response curves, a series of reflections appears in the time domain, and the polar pattern has sharp, very narrow sidelobes. All of these are VERY audible as colorations, and don't respond well to equalization because the frequency correction actually makes the time error worse. (You can't correct a time error with a frequency correction.)
This kind of thing gets glossed over when the designer doesn't have access to time-domain measuring equipment - I don't want to start a flame-war, dammit, but the old-school horn guys from the Fifties were the worst here, ignoring time domain for decades after it became important to the hifi guys. There are still plenty of famous hifi designers today who ignore the time domain, or won't take a good look at the MLS waterfall curves - this has became an odd sort of ideology in the loudspeaker biz.
What has muddies the waters are the linear-phase proponents, who make a big deal out of perfect square waves. This has more to with choice of crossover (low-order) than cleaning up the time domain. Pretty-looking square waves and clean time domain response are not in fact the same - you can build a speaker with decent square-wave response that still has lots of problems with energy storage and standing waves - and in fact most square-wave-perfect commercial loudspeakers have indifferent treatment of standing waves in the cabinet and edge diffraction problems.
I avoid both the frequency-response-only and perfect-square-wave camps: I'm much more interested in reducing energy storage, standing waves, and diffraction. This comes back to driver selection and modification, or custom drivers when possible, edge treatment, and paying attention to arrival times at the listening position.
Hiya Bud, Good Questions
1. I don't see the width of the null-zone is that important. Narrow sidelobes created by diffraction, though, is a big deal, since there's significant delayed energy created by the reflections, and this is audible.
2. Ah, now THAT'S a most interesting question. Yes, there are ways to reduce the backwave. If we are clever, a -6 dB reduction in the backwave gives us a cardioid radiation pattern instead of a figure-8. However, an absorbing material that gives -6 dB attenuation over a multi-octave range probably doesn't exist, although recycled cotton waste and wool felt are a good place to start.
2B. As the backwave gets quieter, going from 0 dB to -3 dB to -6 dB, the null zone moves backward as well. This is the same as a variable-pattern microphone that can be steered between figure-8 and cardioid. If you're really clever, yes, the polar pattern couid make a gentle transition from the figure-8 pattern created by the cone drivers to the quasi-cardioid created by the compression driver. This would be a skillful combination of crossover design, just the right amount of thin cotton batting behind the driver, and cone treatment so the driver has a very smooth rolloff region so the phase transition is well-managed.
2C. We also have to be quite careful not to mass-load the cone. This odd effect appears when damping material gets within an inch or so of the cone - it then starts acting like additional cone mass, decreasing efficiency and HF bandwidth of the driver. This has to determined by measurement and audition. I'm wary of using music in the early stages of design, preferring repeatable test signals like impulses and pink noise. I'm planning on buying the Old Colony Sound
Sound Strobe for doing just this sort of subjective dialling-in, and to complement the MLSSA system.
3. Don't understand the question. How do I speed up the sound on the front? Can't think of any way to do this. As mentioned earlier, termination of the coax is important, and I can see use of ENBL or similar techiques to clean up cone modes for the big 12 or 15-inch driver. Thin wool felt strips that slightly overhang the cone surround and fill the area between the cone edge and the beginning of the baffle might be useful.
4. The thread mentioned earlier about the ISIRIS shows around Page 5 or so some sort of modelling program that simulates the frequency response for the given baffle shape. This looks interesting, and the freq response shown is entirely consistent with what I've measured on MLSSA for open baffles.
4B. I'm a little cautious about techniques to shape the front/rear polar patterns, since most of what comes to mind involves hornlike techniques that have a price to be paid in terms of HOM and time-response effects. For me, the whole point of open-baffle and dipole designs is cleaner time-domain response that is free of standing waves and box modes - the "UN-Horn" if you will.
5. Boy, Bud, you really got me there. I dunno. No clue. We can look at what the microwave guys do to terminate the microwave strips, and apply that to the acoustic domain. It helps that microwaves are about the same size as the acoustics we work with. They use carbon-loaded foam, we use wool felt. Same effect, a gradual broadband absorbent strip material.
5B. One thing to be looked into more closely is the odd behaviour of acoustic surface waves - sound appears to "stick" to surfaces as it expands, and surface roughening can actually improve its ability to "unstick" itself and expand into space. This is an analogue to aerodynamic surface treatments for aircraft wings. The hydrodynamics and acoustics of surfaces gets really weird really fast.
6. I vaguely remember these things when I was at Audionics in the late Seventies. Rumor had it that Carver just ordered a custom version of a generic large-diaphragm speaker (Eminence?), with a tiny little magnet so the Qts went up to a high value like 1.4, so the 40~50 Hz boom of the driver array neatly offset the 1/f rolloff of the fairly narrow dipole baffle. Carver certainly wasn't afraid of EQ or lots of amplifier power, so I'd guess there was plenty of bass-lift equalization as well.
7. It's snowing outside as I look out my window. Fortunately, only an inch or so, nothing like the several feet that sat on the ground for all of January and February. Is it raining in Seattle?
BudP said:
1. I don't see the width of the null-zone is that important. Narrow sidelobes created by diffraction, though, is a big deal, since there's significant delayed energy created by the reflections, and this is audible.
2. Ah, now THAT'S a most interesting question. Yes, there are ways to reduce the backwave. If we are clever, a -6 dB reduction in the backwave gives us a cardioid radiation pattern instead of a figure-8. However, an absorbing material that gives -6 dB attenuation over a multi-octave range probably doesn't exist, although recycled cotton waste and wool felt are a good place to start.
2B. As the backwave gets quieter, going from 0 dB to -3 dB to -6 dB, the null zone moves backward as well. This is the same as a variable-pattern microphone that can be steered between figure-8 and cardioid. If you're really clever, yes, the polar pattern couid make a gentle transition from the figure-8 pattern created by the cone drivers to the quasi-cardioid created by the compression driver. This would be a skillful combination of crossover design, just the right amount of thin cotton batting behind the driver, and cone treatment so the driver has a very smooth rolloff region so the phase transition is well-managed.
2C. We also have to be quite careful not to mass-load the cone. This odd effect appears when damping material gets within an inch or so of the cone - it then starts acting like additional cone mass, decreasing efficiency and HF bandwidth of the driver. This has to determined by measurement and audition. I'm wary of using music in the early stages of design, preferring repeatable test signals like impulses and pink noise. I'm planning on buying the Old Colony Sound
Sound Strobe for doing just this sort of subjective dialling-in, and to complement the MLSSA system.
3. Don't understand the question. How do I speed up the sound on the front? Can't think of any way to do this. As mentioned earlier, termination of the coax is important, and I can see use of ENBL or similar techiques to clean up cone modes for the big 12 or 15-inch driver. Thin wool felt strips that slightly overhang the cone surround and fill the area between the cone edge and the beginning of the baffle might be useful.
4. The thread mentioned earlier about the ISIRIS shows around Page 5 or so some sort of modelling program that simulates the frequency response for the given baffle shape. This looks interesting, and the freq response shown is entirely consistent with what I've measured on MLSSA for open baffles.
4B. I'm a little cautious about techniques to shape the front/rear polar patterns, since most of what comes to mind involves hornlike techniques that have a price to be paid in terms of HOM and time-response effects. For me, the whole point of open-baffle and dipole designs is cleaner time-domain response that is free of standing waves and box modes - the "UN-Horn" if you will.
5. Boy, Bud, you really got me there. I dunno. No clue. We can look at what the microwave guys do to terminate the microwave strips, and apply that to the acoustic domain. It helps that microwaves are about the same size as the acoustics we work with. They use carbon-loaded foam, we use wool felt. Same effect, a gradual broadband absorbent strip material.
5B. One thing to be looked into more closely is the odd behaviour of acoustic surface waves - sound appears to "stick" to surfaces as it expands, and surface roughening can actually improve its ability to "unstick" itself and expand into space. This is an analogue to aerodynamic surface treatments for aircraft wings. The hydrodynamics and acoustics of surfaces gets really weird really fast.
6. I vaguely remember these things when I was at Audionics in the late Seventies. Rumor had it that Carver just ordered a custom version of a generic large-diaphragm speaker (Eminence?), with a tiny little magnet so the Qts went up to a high value like 1.4, so the 40~50 Hz boom of the driver array neatly offset the 1/f rolloff of the fairly narrow dipole baffle. Carver certainly wasn't afraid of EQ or lots of amplifier power, so I'd guess there was plenty of bass-lift equalization as well.
7. It's snowing outside as I look out my window. Fortunately, only an inch or so, nothing like the several feet that sat on the ground for all of January and February. Is it raining in Seattle?
Re: Re: Re: Diffraction, Stored Energy, and Illusions
Horns were sourced from Azurahorn. They are the 204Hz units:
http://www.azurahorn.com/
Drivers are Yamaha units with 1.4" throat exits. These are of very unique construction. Meyersound was the major customer for these and Yamaha stopped making them when Meyersound began making their own compression drivers.
That's enough of a threadjack. Apologies to Lynn. If there are further questions I'll start another thread.
Jeff
Curly Woods said:
Horns were sourced from Azurahorn. They are the 204Hz units:
http://www.azurahorn.com/
Drivers are Yamaha units with 1.4" throat exits. These are of very unique construction. Meyersound was the major customer for these and Yamaha stopped making them when Meyersound began making their own compression drivers.
That's enough of a threadjack. Apologies to Lynn. If there are further questions I'll start another thread.
Jeff
Welcome Back
So good to read your posts, Lynn. Your contribution to the DIY realm has been tremendous, and greatly appreciated (thanks to diyAudio also, of course). The ME-2's will be completed this weekend (believe it or not, people are still building these puppies). Raven/Amity soon to follow hopefully.
I read what you write twice, and understand half of what I read.
So good to read your posts, Lynn. Your contribution to the DIY realm has been tremendous, and greatly appreciated (thanks to diyAudio also, of course). The ME-2's will be completed this weekend (believe it or not, people are still building these puppies). Raven/Amity soon to follow hopefully.
I read what you write twice, and understand half of what I read.
Here's Bud's EnABL Paper & the Mamboni Mods
Over at Positive Feedback magazine, my old stomping ground back when I lived in Portland.
As mentioned in earlier posts, the smoothness of the rolloff region is CRITICAL for high-quality sound, as well as a well-behaved radiation pattern in the crossover region. The EnABL treatment might be especially important for pro-grade coax drivers, which from the published curves have real challenges in the rolloff region (2 to 10 kHz).
The thread by Mamboni elsewhere in the forum about the Ohm F also some interesting felt-strip treatments for modern 10" woofers, allowing them to act as Walsh bending-mode transducers up to 10 kHz.
Both techniques would be useful for chasing out those nasty peaks and ripples in the 2 to 10 kHz region, and would certainly improve the sound of a coax driver. With dipoles free of box colorations, yes, you'll hear even small improvements in the driver.
Over at Positive Feedback magazine, my old stomping ground back when I lived in Portland.
As mentioned in earlier posts, the smoothness of the rolloff region is CRITICAL for high-quality sound, as well as a well-behaved radiation pattern in the crossover region. The EnABL treatment might be especially important for pro-grade coax drivers, which from the published curves have real challenges in the rolloff region (2 to 10 kHz).
The thread by Mamboni elsewhere in the forum about the Ohm F also some interesting felt-strip treatments for modern 10" woofers, allowing them to act as Walsh bending-mode transducers up to 10 kHz.
Both techniques would be useful for chasing out those nasty peaks and ripples in the 2 to 10 kHz region, and would certainly improve the sound of a coax driver. With dipoles free of box colorations, yes, you'll hear even small improvements in the driver.
It Gets Better
Further on in the Ohm F thread, Bud and Mamboni start collaborating together on combining the techniques. Following this thread to the end makes for some VERY interesting reading on sound propagation through solid materials and methods to control standing waves.
The Bud/Mamboni techniques are especially useful for any dipole speaker - once again, because these designs are entirely free of cabinet coloration, so you can hear the driver as as they really are. Yes, I guess you could combine a dipole with a Walsh - maybe if the baffle were slanted or curved towards the top - but you guys are on your own here. I still want to explore a really good coax - and yeah, not too many of these - on a dipole baffle.
Further on in the Ohm F thread, Bud and Mamboni start collaborating together on combining the techniques. Following this thread to the end makes for some VERY interesting reading on sound propagation through solid materials and methods to control standing waves.
The Bud/Mamboni techniques are especially useful for any dipole speaker - once again, because these designs are entirely free of cabinet coloration, so you can hear the driver as as they really are. Yes, I guess you could combine a dipole with a Walsh - maybe if the baffle were slanted or curved towards the top - but you guys are on your own here. I still want to explore a really good coax - and yeah, not too many of these - on a dipole baffle.
Dear Lynn,
Could you please describe in a few words your subjective impressions of listening Summa Cum Laude designed by Dr Geddes? The measurements are impressive.
Mr Linkwitz is now changing from dipole to omni exactly like Yoshii Hiroyuki from Time Domain Corp. and constructor of Grand Scepter by Onkyo.
It is really hard to judge what is and what is not an artefact in audio, especially imaging. It would be nice to point which type of mic arrangement should be used for subjective evaluation eg. exact Blumelin (two dipole ribbon microphones, 90 deg, speakers at the same angle). James Boyk from Caltech favours ribbons over condesers. It not so easy - for example my first binaural recordings doesn't tell me if the source is forward or backside (25mV/Pa electret, open ear canal).
Please take a look at my speaker CSD graph:
http://audiostereo.lukarnet.com/gfx/700000/708676_1.gif
The box (2 litres) is made of steel (O profile, 100x100x200x3) with a lot of BAF. Do you see any box colorations?
Have you ever used cepstrum analysis for the resonances and echo detection?
Could you please describe in a few words your subjective impressions of listening Summa Cum Laude designed by Dr Geddes? The measurements are impressive.
Mr Linkwitz is now changing from dipole to omni exactly like Yoshii Hiroyuki from Time Domain Corp. and constructor of Grand Scepter by Onkyo.
It is really hard to judge what is and what is not an artefact in audio, especially imaging. It would be nice to point which type of mic arrangement should be used for subjective evaluation eg. exact Blumelin (two dipole ribbon microphones, 90 deg, speakers at the same angle). James Boyk from Caltech favours ribbons over condesers. It not so easy - for example my first binaural recordings doesn't tell me if the source is forward or backside (25mV/Pa electret, open ear canal).
Please take a look at my speaker CSD graph:
http://audiostereo.lukarnet.com/gfx/700000/708676_1.gif
The box (2 litres) is made of steel (O profile, 100x100x200x3) with a lot of BAF. Do you see any box colorations?
Have you ever used cepstrum analysis for the resonances and echo detection?
Hi Lynn,
When I asked about the potential benefits of "speeding up" a front side surface, as opposed to the back side, I was thinking about the Mamboni process of small triangular light felt necklaces applied at the cone terminus, on the back side, just as he does now for the Walsh style woofer drivers, and the EnABL process on the front side. Using a conformal coating that will transmit boundary layer energy faster than the speed of sound through air will allow a fairly large increase in useful emitting surface when the EnABL pattern is underneath it to kill the transient standing waves. The three together seemed likely to be beneficial here in dipole land.
I am going to assume, without ANY proof that the Mamboni process on the back side will allow us to naturally bend the null zone back like a skirt and if we provide a skirt plate for those waves to attach to we could probably shape the entire dipole emission into the room with that lever.
Do not count the Walsh bending event out with respect to dipole emission either. When EnABL treated, the things emit as true point source radiators, right up and over the magnets if those surfaces have also been eliminated as diffraction edges. Scary looking in a living room filled with wife... anyway.
I am also interested in a Walsh style upper mid and tweeter in combination with a modified dipole emission lower mid and bass driver setup. That led me to question whether any of our participants had investigated sounding board shapes other than the usual straight edged planar surface. I have seen one interesting one, from the land of OZ participants, that shares its perimeter shape with that of musical instruments. I wonder if that French Curve derived edge treatment helps to kill transient standing waves. Seems likely to me. See BobF's pics here.
http://home.vicnet.net.au/~macinc/clubproj.htm
Bud
When I asked about the potential benefits of "speeding up" a front side surface, as opposed to the back side, I was thinking about the Mamboni process of small triangular light felt necklaces applied at the cone terminus, on the back side, just as he does now for the Walsh style woofer drivers, and the EnABL process on the front side. Using a conformal coating that will transmit boundary layer energy faster than the speed of sound through air will allow a fairly large increase in useful emitting surface when the EnABL pattern is underneath it to kill the transient standing waves. The three together seemed likely to be beneficial here in dipole land.
I am going to assume, without ANY proof that the Mamboni process on the back side will allow us to naturally bend the null zone back like a skirt and if we provide a skirt plate for those waves to attach to we could probably shape the entire dipole emission into the room with that lever.
Do not count the Walsh bending event out with respect to dipole emission either. When EnABL treated, the things emit as true point source radiators, right up and over the magnets if those surfaces have also been eliminated as diffraction edges. Scary looking in a living room filled with wife... anyway.
I am also interested in a Walsh style upper mid and tweeter in combination with a modified dipole emission lower mid and bass driver setup. That led me to question whether any of our participants had investigated sounding board shapes other than the usual straight edged planar surface. I have seen one interesting one, from the land of OZ participants, that shares its perimeter shape with that of musical instruments. I wonder if that French Curve derived edge treatment helps to kill transient standing waves. Seems likely to me. See BobF's pics here.
http://home.vicnet.net.au/~macinc/clubproj.htm
Bud
Impressions
Compare the CSD to the
Ariel
Note how fast things drop off above 2kHz. That's what low diffraction looks like. Also, note the speed of the impulse response, despite the up-and-down wiggle caused by the 360-degree phase rotation of the crossover (the Ariel is not a linear-phase speaker).
I am surprised you used steel for a cabinet material - it's one of the most resonant materials you could have chosen, will subtly interact with the stray field from the speaker magnet, and will certainly affect the linearity of the inductors in the crossover.
As for the sonics of the Summa, I really couldn't tell, since Geddes for reasons of his own had chosen a $200 Pioneer HT receiver and a no-name Costco CD player. I think he believes that all consumer-quality electronics sounds the same, and put his money where his mouth is by demo-ing his speakers at the RMAF with low-end generic HT gear.
To answer your question as truthfully as possible, I heard what sounded exactly like a c**p mass-market receiver and a c**p CD player, but with fabulous macro-dynamics and almost no horn artifacts. Micro-dynamics were impossible to assess due to grainy transistor and cheap-CD-player sound - poorly filtered switching supplies in the CD player probably didn't help.
If the Summa were even as good as an Altec A7, bottom-of-the-market electronics would sound dreadful, gritty, two-dimensional, and very harsh. Just from knowing a little about the Summa, it is far better than an A7, which after all is a late Forties-vintage entry-level theater speaker intended for small movie houses that couldn't afford the bigger and better Altecs.
The Summa by contrast is a very modern design, using Geddes' latest design protocol, and employing the best prosound drivers he could find. So the Summa, being the accurate speaker it is designed to be, should faithfully reflect the quality of the source.
As for auditioning the Summa on high-end electronics, good luck. Earl has strong beliefs on the subject, and is convinced the high-end business is a scam. I am still puzzled why he spent a good amount of money exhibiting at what was after all a high-end show, with very little home-theater on display.
As for assessing subjective image quality, well ... I've been doing this a long time, starting with designing a variable-parameter SQ decoder in 1973, followed by a line of speakers for Audionics, and now the stuff I do for myself. The folks who taught me the basics were the BBC quadraphonic research team and Laurie Fincham at KEF, who I met in 1975, and have read most of their research papers. I know what many speaker and electronic artifacts sound like from direct experience - the sound of resonant drivers, box modes, edge diffraction, blurring due to poor-quality crossover parts, insufficient current in the driver stage, slewing in a phono preamp, unstable Clas AB bias, switching noise in the power supply, etc. All of these affect image quality in one way or another. I'm not aware of any artifact that would actually improve image quality.
What surprises me more than a little is just how clueless magazine reviewers are. The US school of reviewers have no idea at all about image quality - what they consider "good imaging" is an extremely artificial cookie-cutter image with angular edges, and almost no depth or realism at all. Wrong, wrong, wrong.
Good imaging is actually extremely simple. It sounds real, wide-open, like the walls of your room have fallen down and you have dropped into another room. You can even hear this before the music starts, from the ambient room noises alone! There is a quality of tactility, of palpability, where you can hear textures, sizes, and dimensions, as well as the spaces between things.
At the limit of the art, with discrete multichannel sound and phase-matched equipment of the highest quality, the listening room completely disappears, and subtle visual impressions are at the threshold of perception. I have heard this at the BBC research labs and occasionally with my own system, so it is real and repeatable, although very rare. I have never heard this with commercial equipment at any price, up to and including $250,000. That, by the way, is why I design my own equipment. I can't buy the sound I want.
Have I heard sound that wide-open from horns? No, never. There is a strange myth going around that "controlled directivity" is necessary for good imaging. Uh, according to who? Considering the abominable image quality of modern home theatre - well below the quadraphonic systems in the research labs during the Seventies - I sure wouldn't cite any HT "experts" on this matter.
I'm in the old-school camp that supports very low energy storage and smooth polar patterns distribution as the starting point for good image quality, along with precise phase and amplitude matching for all speakers in the system - especially the center speaker if there is one.
Does this philosophy make the dipole challenging to design - well, yes, since I also want 97dB/metre efficiency as well. Prosound stuff isn't optimized for smooth response, since the pro attitude is that everything is going to be digitally equalized anyway, so it doesn't matter as much as high power output and low distortion. This is why Bud's and Mamboni's work is so exciting to me - it would be truly wonderful to have prosound ultra-high SPL and low distortion combined with low energy storage and flat response.
jzagaja said:
Compare the CSD to the
Ariel
Note how fast things drop off above 2kHz. That's what low diffraction looks like. Also, note the speed of the impulse response, despite the up-and-down wiggle caused by the 360-degree phase rotation of the crossover (the Ariel is not a linear-phase speaker).
I am surprised you used steel for a cabinet material - it's one of the most resonant materials you could have chosen, will subtly interact with the stray field from the speaker magnet, and will certainly affect the linearity of the inductors in the crossover.
As for the sonics of the Summa, I really couldn't tell, since Geddes for reasons of his own had chosen a $200 Pioneer HT receiver and a no-name Costco CD player. I think he believes that all consumer-quality electronics sounds the same, and put his money where his mouth is by demo-ing his speakers at the RMAF with low-end generic HT gear.
To answer your question as truthfully as possible, I heard what sounded exactly like a c**p mass-market receiver and a c**p CD player, but with fabulous macro-dynamics and almost no horn artifacts. Micro-dynamics were impossible to assess due to grainy transistor and cheap-CD-player sound - poorly filtered switching supplies in the CD player probably didn't help.
If the Summa were even as good as an Altec A7, bottom-of-the-market electronics would sound dreadful, gritty, two-dimensional, and very harsh. Just from knowing a little about the Summa, it is far better than an A7, which after all is a late Forties-vintage entry-level theater speaker intended for small movie houses that couldn't afford the bigger and better Altecs.
The Summa by contrast is a very modern design, using Geddes' latest design protocol, and employing the best prosound drivers he could find. So the Summa, being the accurate speaker it is designed to be, should faithfully reflect the quality of the source.
As for auditioning the Summa on high-end electronics, good luck. Earl has strong beliefs on the subject, and is convinced the high-end business is a scam. I am still puzzled why he spent a good amount of money exhibiting at what was after all a high-end show, with very little home-theater on display.
As for assessing subjective image quality, well ... I've been doing this a long time, starting with designing a variable-parameter SQ decoder in 1973, followed by a line of speakers for Audionics, and now the stuff I do for myself. The folks who taught me the basics were the BBC quadraphonic research team and Laurie Fincham at KEF, who I met in 1975, and have read most of their research papers. I know what many speaker and electronic artifacts sound like from direct experience - the sound of resonant drivers, box modes, edge diffraction, blurring due to poor-quality crossover parts, insufficient current in the driver stage, slewing in a phono preamp, unstable Clas AB bias, switching noise in the power supply, etc. All of these affect image quality in one way or another. I'm not aware of any artifact that would actually improve image quality.
What surprises me more than a little is just how clueless magazine reviewers are. The US school of reviewers have no idea at all about image quality - what they consider "good imaging" is an extremely artificial cookie-cutter image with angular edges, and almost no depth or realism at all. Wrong, wrong, wrong.
Good imaging is actually extremely simple. It sounds real, wide-open, like the walls of your room have fallen down and you have dropped into another room. You can even hear this before the music starts, from the ambient room noises alone! There is a quality of tactility, of palpability, where you can hear textures, sizes, and dimensions, as well as the spaces between things.
At the limit of the art, with discrete multichannel sound and phase-matched equipment of the highest quality, the listening room completely disappears, and subtle visual impressions are at the threshold of perception. I have heard this at the BBC research labs and occasionally with my own system, so it is real and repeatable, although very rare. I have never heard this with commercial equipment at any price, up to and including $250,000. That, by the way, is why I design my own equipment. I can't buy the sound I want.
Have I heard sound that wide-open from horns? No, never. There is a strange myth going around that "controlled directivity" is necessary for good imaging. Uh, according to who? Considering the abominable image quality of modern home theatre - well below the quadraphonic systems in the research labs during the Seventies - I sure wouldn't cite any HT "experts" on this matter.
I'm in the old-school camp that supports very low energy storage and smooth polar patterns distribution as the starting point for good image quality, along with precise phase and amplitude matching for all speakers in the system - especially the center speaker if there is one.
Does this philosophy make the dipole challenging to design - well, yes, since I also want 97dB/metre efficiency as well. Prosound stuff isn't optimized for smooth response, since the pro attitude is that everything is going to be digitally equalized anyway, so it doesn't matter as much as high power output and low distortion. This is why Bud's and Mamboni's work is so exciting to me - it would be truly wonderful to have prosound ultra-high SPL and low distortion combined with low energy storage and flat response.
But the Linkwitz did the exercise with dipoles and now he does like more the omnis. I'm not sure which is more right - directional point source or 180 deg stereolith, Moulton or Geddes et al.
I've chosen steel box as a temporary enclosure made of scrap.
You wrote that PA drivers require linearisation. What's your opinion about using linearisation FIR filters for music playback? Mentioned Yoshii Hiroyuki had abandoned this technique after several trials. My linearised Jordan JXR6 speaker (real measurement not a convolution):
http://audiostereo.lukarnet.com/gfx/690000/698197_1.gif
Now there are two things left:
1) Manger transducer has advertised nearly null storage energy but I can't see it on the CSD plots.
2) As a reference transducer I'm using Shure E2 headphone (full range driver) which is a in-ear insert. Jordan is so similar and DSP doesn't improve much on programme material. It is probably very like as your beloved ESL57.
PS. Spatial impression of having no walls strongly depends on room size and amount of ambience isn't it?
Best Regards,
I've chosen steel box as a temporary enclosure made of scrap.
You wrote that PA drivers require linearisation. What's your opinion about using linearisation FIR filters for music playback? Mentioned Yoshii Hiroyuki had abandoned this technique after several trials. My linearised Jordan JXR6 speaker (real measurement not a convolution):
http://audiostereo.lukarnet.com/gfx/690000/698197_1.gif
Now there are two things left:
1) Manger transducer has advertised nearly null storage energy but I can't see it on the CSD plots.
2) As a reference transducer I'm using Shure E2 headphone (full range driver) which is a in-ear insert. Jordan is so similar and DSP doesn't improve much on programme material. It is probably very like as your beloved ESL57.
PS. Spatial impression of having no walls strongly depends on room size and amount of ambience isn't it?
Best Regards,