Re: One Really Terrific Article
He's one of my hero's too Lynn -
Here is am Alnico Altec biflex with a large Aurum Cantus ribbon I built kinda 'Apogee' shaped - on the sides against the walls were horn loaded 18's for below 120 cycles
Lynn Olson said:
He's one of my hero's too Lynn -
Here is am Alnico Altec biflex with a large Aurum Cantus ribbon I built kinda 'Apogee' shaped - on the sides against the walls were horn loaded 18's for below 120 cycles
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
The FR of the RCF 12 looks difficult to use with a moderate-slope crossover - the 6 dB zero in the response at 2 kHz isn't going to be easy to use with 1.6~2 kHz crossover for the ribbon - and crossing the ribbon lower is asking for high distortion and potential destruction of the ribbon.
At a Theile/Small normalized efficiency of 99 dB/metre, I don't think the 12NDA520 is going to be sucking a lot of power. The lowpass filter I'll be trying first is a series inductor followed by a shunt notch filter broadly tuned to 3.5 kHz; this would directly feed the wideband driver, and the next driver down taps off the output of the lowpass filter and adds another series inductor tuned to about 200~250 Hz.
The driver(s) closest to the floor get their own solid-state power amp and dedicated room/dipole EQ, while the two topmost drivers share a common power amp and 1.6~2 kHz lowpass crossover. This is only a 4~8 ohm load, and if a power amp can't deliver a handful of watts into 4 ohms, its quality is open to question, whether SET, PP pentode, or transistor. That might sound a bit arrogant, but as an amp designer, I'm completely unsympathetic to amplifiers that are incapable of driving reactive loads. If amplifier designers haven't figured out by now that speakers are reactive, they should really start getting a clue.
An alternate topology would have the "quality" amplifier drive only the topmost widerange driver, with the two lower drivers powered as mentioned before with a dedicated power + EQ. This remains to be determined on audition.
The reason the 12NDA520 appeals to me is the rise in FR is not only benign, but a positive indication the VC inductance is low, a very desirable feature since VC inductance is quite nonlinear. A theoretically perfect driver with zero VC inductance would exhibit not flat response, but a fairly steep rise beginning at 500 Hz ~ 1 kHz, the region where VC inductance starts to roll off the response of the driver.
Conventional drivers intentionally balance off VC inductance against the normal HF directivity gain of the cone, the same as moving-magnet cartridges balance the LC rolloff of cable capacitance + cart inductance against mechanical peaking (Shure fans take note). The catch with balancing electrical and mechanical rolloffs, of course, happens when the terms are nonlinear, as they are with VC and MM phono-cartridge inductance. (By contrast, MC cartridges are mechanically flat, something not true of MM cartridges with their much higher inductance.)
If you're going to roll off a driver, it's a lot better to roll it off with an air-core inductor than the truly abominable VC inductance. The other thing I look for is absence of narrowband glitches near the crossover frequency - these are essentially impossible to equalize, and result in sharp changes in polar pattern thanks to narrowband phase shifts in the "glitch" region.
Of course, what really matters is how the wideband driver sounds on its own, without any crossover. If the driver passes that check, then I audition it with several lowpass filter topologies, and listen for "concentrations of energy" near the crossover frequency (this results in using low-Q filter networks).
After the midbass driver + lowpass crossover auditions satisfactorily, then I design the highpass filter for the tweeter, designing it to phase-match the already-existing midbass driver + crossover. I've found that crossover colorations can be masked when the tweeter is turned on, but are readily audible when listening to the midbass + crossover alone - this exposes midrange colorations of conventional 3rd and 4th-order filters, for example.
At a Theile/Small normalized efficiency of 99 dB/metre, I don't think the 12NDA520 is going to be sucking a lot of power. The lowpass filter I'll be trying first is a series inductor followed by a shunt notch filter broadly tuned to 3.5 kHz; this would directly feed the wideband driver, and the next driver down taps off the output of the lowpass filter and adds another series inductor tuned to about 200~250 Hz.
The driver(s) closest to the floor get their own solid-state power amp and dedicated room/dipole EQ, while the two topmost drivers share a common power amp and 1.6~2 kHz lowpass crossover. This is only a 4~8 ohm load, and if a power amp can't deliver a handful of watts into 4 ohms, its quality is open to question, whether SET, PP pentode, or transistor. That might sound a bit arrogant, but as an amp designer, I'm completely unsympathetic to amplifiers that are incapable of driving reactive loads. If amplifier designers haven't figured out by now that speakers are reactive, they should really start getting a clue.
An alternate topology would have the "quality" amplifier drive only the topmost widerange driver, with the two lower drivers powered as mentioned before with a dedicated power + EQ. This remains to be determined on audition.
The reason the 12NDA520 appeals to me is the rise in FR is not only benign, but a positive indication the VC inductance is low, a very desirable feature since VC inductance is quite nonlinear. A theoretically perfect driver with zero VC inductance would exhibit not flat response, but a fairly steep rise beginning at 500 Hz ~ 1 kHz, the region where VC inductance starts to roll off the response of the driver.
Conventional drivers intentionally balance off VC inductance against the normal HF directivity gain of the cone, the same as moving-magnet cartridges balance the LC rolloff of cable capacitance + cart inductance against mechanical peaking (Shure fans take note). The catch with balancing electrical and mechanical rolloffs, of course, happens when the terms are nonlinear, as they are with VC and MM phono-cartridge inductance. (By contrast, MC cartridges are mechanically flat, something not true of MM cartridges with their much higher inductance.)
If you're going to roll off a driver, it's a lot better to roll it off with an air-core inductor than the truly abominable VC inductance. The other thing I look for is absence of narrowband glitches near the crossover frequency - these are essentially impossible to equalize, and result in sharp changes in polar pattern thanks to narrowband phase shifts in the "glitch" region.
Of course, what really matters is how the wideband driver sounds on its own, without any crossover. If the driver passes that check, then I audition it with several lowpass filter topologies, and listen for "concentrations of energy" near the crossover frequency (this results in using low-Q filter networks).
After the midbass driver + lowpass crossover auditions satisfactorily, then I design the highpass filter for the tweeter, designing it to phase-match the already-existing midbass driver + crossover. I've found that crossover colorations can be masked when the tweeter is turned on, but are readily audible when listening to the midbass + crossover alone - this exposes midrange colorations of conventional 3rd and 4th-order filters, for example.
Thanks, Magnetar, for all your contributions - much appreciated. I come out of the Laurie Fincham/D.E.L. Shorter KEF/BBC school of design, so I don't design crossovers in the same way as either Altec/JBL old-timers or the modern throw-it-in-the-simulator approach. I optimize for low coloration (using subjective pink-noise assessment) and smoothest possible phase transfer between drivers.
A secondary goal of this system is taking advantage of the floor image with a close-to-floor driver, a widerange driver close to ear height, and a transitional driver midway between the two. This dictates a minimum of three drivers, with an option of four drivers (two of them close to the floor). The only true crossover is between the tweeter (most likely a RAAL ribbon) and the widerange driver.
The only function of the other drivers is to progressively increase the radiating area to offset the inherent dipole loss in efficiency. These drivers use cascaded low-pass filters to just offset the losses of the wideband driver for frequencies below 200~250 Hz. Net result, 99~100 dB/metre efficiency from 80 Hz to 20 kHz, with a monopole subwoofer below that.
A secondary goal of this system is taking advantage of the floor image with a close-to-floor driver, a widerange driver close to ear height, and a transitional driver midway between the two. This dictates a minimum of three drivers, with an option of four drivers (two of them close to the floor). The only true crossover is between the tweeter (most likely a RAAL ribbon) and the widerange driver.
The only function of the other drivers is to progressively increase the radiating area to offset the inherent dipole loss in efficiency. These drivers use cascaded low-pass filters to just offset the losses of the wideband driver for frequencies below 200~250 Hz. Net result, 99~100 dB/metre efficiency from 80 Hz to 20 kHz, with a monopole subwoofer below that.
Lynn Olson said:
Hey everyone... gee this is an active thread!
I generally don't have time during the week to look, so I'm a little behind the curve.
But to answer everyone's questions:
- the driver I use is not in production
- I run my compression drivers w/ 1st order filters @ <299Hz.
- it has some special and/or unique design aspects
- it is not a Cogent, Peavey, Vitavoxetc...
- I'm not saying because I'd like to be able to buy a few more if I can find them...
- If you want to understand compression drivers then learn all you can about the WE555 and the driver that came next from them (WE384??).
- The Cogent (an RCA type driver) is cool. Dunno how it sounds (yet).
- again, most compression drivers are fatally flawed; the causes are varied and multiple
- imho you can not EQ a compression driver back into sounding right
- I think that the driver in question could be reproduced but it is actually not a simple or cheap process, and I strongly doubt that the demand - even with money for advertising on top of the pre-production outlay makes it a viable business. I AM open to doing it anyhow if there is anyone reading this with some discretionary funds to put up (contact me privately please)? (not a huge pot full either...)
Lynn,
There was a site up with some baffle experiments for the BG ribbons. I suggest looking for that before you chose the "truncated pyramid" shape a la Apogee.
A general comment. If it was easy to make a system sound good, there would only be a few choices, sort of like an ipod? It's a balancing act with the room thrown in as a random variable (or a constant if it's ur room). There is no universal solution (thank goodness!).
My general philosophy of reproduction is to try to pick the compromises that do the least to come in the way of the "main spectrum" of our hearing - that being from <300 to >3000, and as much as you can get on either side of that - and at the same time keep the colorations and distortions to the minumum possible. That is the area where all the compromises come in. The idea, if I have one, is to try to keep the sound natural, so that the brain does the least possible work to decipher the sonic information. Less brain cycles to decipher = more cycles to detect small and subtle details = more better!
Obviously going from <300 to 3000+ is a difficult job for a single direct radiator dynamic driver.
Well, no matter what the availability of a given driver may be, I mention it only to say that everything may or may not be as it appears to be on the surface, and if there is something of significantly good performance out there it is at least a touchstone for comparisons and learning. Obviously not a choice for a commercial product or kit/project article though.
Regards,
_-_-bear
bear said:
The driver I use is not in production
- I run my compression drivers w/ 1st order filters @ <299Hz.
- it has some special and/or unique design aspects
- it is not a Cogent, Peavey, Vitavoxetc...
- I'm not saying because I'd like to be able to buy a few more if I can find them...
- If you want to understand compression drivers then learn all you can about the WE555 and the driver that came next from them (WE384??).
- The Cogent (an RCA type driver) is cool. Dunno how it sounds (yet).
- again, most compression drivers are fatally flawed; the causes are varied and multiple
- imho you can not EQ a compression driver back into sounding right
- I think that the driver in question could be reproduced but it is actually not a simple or cheap process, and I strongly doubt that the demand - even with money for advertising on top of the pre-production outlay makes it a viable business. I AM open to doing it anyhow if there is anyone reading this with some discretionary funds to put up (contact me privately please)? (not a huge pot full either...)
_-_-bear
Thanks for the info, much appreciated. I fully understand not wanting to make unobtanium even more so, thus the reticence to discuss XYZ driver.
The issue of non-flat or non-spherical wavefronts was discussed obliquely in the very interesting Earl Geddes/waveguide thread. What tiny amount I know of horn theory is that there is an assumption of a smooth, flat wavefront entering the horn - and I think Finite Element Analysis assumes this also. Unfortunately, what comes out the phasing plug is nothing like a planar or spherical wavefront - the concentric circular exits or "tangerine" face-plate gives that away.
So, if the underlying assumption of horn theory is violated, well, what comes out the horn is going to be different as well - the horn or waveguide is not going to automagically straighten out the wavefront - instead, I would expect it to be even more decorrelated and incoherent by the time it leaves the horn.
Re: Just a comment for thought...
That might be good for say a trombone recording, but HF driver has much harder job of reproducing upper harmonics of piano, strings and human voice. I don't think instrumental analogy is very valid approach for a speaker design.
I think RAAL is a quite safe bet for Lynn. Not many drivers can claim to be better (if any!), and virtually none to be SIGNIFICANTLY better.
That is, unless an extra wide vertical dispersion is mandatory. And I don't think it is, in this case.
moray james said:
That might be good for say a trombone recording, but HF driver has much harder job of reproducing upper harmonics of piano, strings and human voice. I don't think instrumental analogy is very valid approach for a speaker design.
I think RAAL is a quite safe bet for Lynn. Not many drivers can claim to be better (if any!), and virtually none to be SIGNIFICANTLY better.
That is, unless an extra wide vertical dispersion is mandatory. And I don't think it is, in this case.
You are probably right...
I was not attemptint to suggest that one should emulate a trombone mouth piece to create a good speaker. Just that for someone that doese not play a trombone the mouth piece woud seem to be a straight forward thing that one just blows into. That is not the case and it is interesting to see how a trombone mouth piece has to be played to make it do its job. So with comp drivers and thier horns there may well be methods of achieving excellent results that are not so obvious. As with the trombone mouthpiece the mouth/lips is the other half of the device. I do not have any revelations to share here just that thinking outside of conventional wisdom may yield superior results. I agree with Bear's comments regarding the We555 which is a masterpiece of design. In new designs I would imagine that the interface between the compression device and the horn will be critical to improving performance.
I was not attemptint to suggest that one should emulate a trombone mouth piece to create a good speaker. Just that for someone that doese not play a trombone the mouth piece woud seem to be a straight forward thing that one just blows into. That is not the case and it is interesting to see how a trombone mouth piece has to be played to make it do its job. So with comp drivers and thier horns there may well be methods of achieving excellent results that are not so obvious. As with the trombone mouthpiece the mouth/lips is the other half of the device. I do not have any revelations to share here just that thinking outside of conventional wisdom may yield superior results. I agree with Bear's comments regarding the We555 which is a masterpiece of design. In new designs I would imagine that the interface between the compression device and the horn will be critical to improving performance.
So how did you go from doing a "proper" horn to a simple ribbon Lynn? Ease of construction or cost? Or just for sonic reasons?
BTW I love this thread. The discussion has been a great learning experience for myself, even if I don't agree with some things. Some day I'll have to go back and distill all the things that interest me into a list topics for further consideration. Good job and good luck Lynn!
BTW I love this thread. The discussion has been a great learning experience for myself, even if I don't agree with some things. Some day I'll have to go back and distill all the things that interest me into a list topics for further consideration. Good job and good luck Lynn!
PRO speaker for OB
Hi
Simulations, measurements and listening for several weeks now under kind of "controlled conditions" to OB I find two issues not so clearly outspoken till now.
First to me it does not make sense to use BIG baffles – unless you extensively use wave guide technique. Simply because the upper mid octaves of the wide range unit are spoiled by the baffle itself rather unpredictable and are hard to correct / equalise.
The way out I am exploring right now is to stay with a small baffle for the mid and high speaker AND a second - much bigger one for the bass drivers. This has similar advantage like the " distributed speaker area " Lynn was pointing out several times in that it also doubles the efficiency by doubling the min distance ( with the SAME driver used ).
The second even more serious issue is that there are simply NO pro-speakers that really match for OB use IMO.
I cold not find a single on with phase plug or open VC – they ALL ventilate to the back which is nothing we really want for OB..
AND there is almost no one with Qts above 0,4- 0.5 in conjunction with Xpp-lin greater than 1-2 mm simply because they are all intended to be used in vented boxes OR strictly limited to mid applications only.
Last thing became an important aspect when searching for the reason why measurable much worse speaker ( FR and CSD and HD ) sound way better in my test OB.
Sure, you can equalise the falling FR to whatever Q we like but at least to me it does not seem to reconstruct the vividness of high q drivers in their mids.
When contacting manufacturers it seems that there is some progress in this but they are not out and ready available OR restricted to OEM use yet.
-------------------------
Such mouth piece are basically simple diffusers - optimised to decrease pressure of air flow under special conditions.
NO airflow in speakers hence a very different approach not really suitable for speakers.
Greetings
Michael
Hi
Simulations, measurements and listening for several weeks now under kind of "controlled conditions" to OB I find two issues not so clearly outspoken till now.
First to me it does not make sense to use BIG baffles – unless you extensively use wave guide technique. Simply because the upper mid octaves of the wide range unit are spoiled by the baffle itself rather unpredictable and are hard to correct / equalise.
The way out I am exploring right now is to stay with a small baffle for the mid and high speaker AND a second - much bigger one for the bass drivers. This has similar advantage like the " distributed speaker area " Lynn was pointing out several times in that it also doubles the efficiency by doubling the min distance ( with the SAME driver used
).The second even more serious issue is that there are simply NO pro-speakers that really match for OB use IMO.
I cold not find a single on with phase plug or open VC – they ALL ventilate to the back which is nothing we really want for OB..
AND there is almost no one with Qts above 0,4- 0.5 in conjunction with Xpp-lin greater than 1-2 mm simply because they are all intended to be used in vented boxes OR strictly limited to mid applications only.
Last thing became an important aspect when searching for the reason why measurable much worse speaker ( FR and CSD and HD ) sound way better in my test OB.
Sure, you can equalise the falling FR to whatever Q we like but at least to me it does not seem to reconstruct the vividness of high q drivers in their mids.
When contacting manufacturers it seems that there is some progress in this but they are not out and ready available OR restricted to OEM use yet.
-------------------------
Such mouth piece are basically simple diffusers - optimised to decrease pressure of air flow under special conditions.
NO airflow in speakers hence a very different approach not really suitable for speakers.
Greetings
Michael
Although the initial correspondence with Dr. Geddes got off to a rocky start, I've learned a lot from his postings here and in his own thread - that the foam in the EG waveguide fills the waveguide and has a net attenuation of 3 dB (less than I thought), that Earl is indeed looking very closely at the problem of the shape of the wavefront that enters the waveguide, and as I suspected, horns and waveguides have no ability to "straighten out" a deformed wavefront that enters the horn or waveguide.
Looking further, the Geddes waveguide might possibly be the only waveguide to be free of internal reflections - don't know how the LeCleac'h and Tractix profiles would compare in this respect, or if felt lining on the inside of a LeCleac'h or Tractrix would have a similar effect to the specially selected open-cell foam that Earl uses to fill his waveguide.
This endless process of getting horns or waveguides to perform as theory says they should takes me far away from the goals of the new speaker - I just want a good, efficient tweeter, not to re-invent horns. A lot of really bright people (starting with Bell Labs) have been working on these blasted things for 80 years now, and it's become obvious there is still quite a lot of important, fundamental information to be discovered - just getting a reasonably smooth wavefront out of the thing is a non-trivial exercise.
The shape of the real wavefront, not the simple spherical idealization that people usually think of, is a big deal for the operation of the crossover and the overall image quality. Narrow-beam spikes or holes are very undesirable - although hard to measure, they are easy to hear, and make balancing the system very tricky and time-consuming.
I've spent enough time twiddling and tweaking with the much smaller problem of cabinet-edge diffraction to take the shape of the wavefronts very seriously - and reading EG's writings closely and having extended phone chats with Alexander of RAAL have made it clear that deformed, uneven wavefronts are a serious problem with high-efficiency tweeters. Although the vertical height of a tall ribbon has implications for the overall frequency balance of the speaker at 2, 3, 4, and 5 meters (the narrow vertical beam of the ribbon falls off less with distance than the cone speakers), at least the impulse response is well-behaved over a wide lateral arc, an important criteria for image quality.
I also have admit to I like the sound of ribbon microphones (RCA 44B/BX) compared to condenser mikes, and expect similar virtues in the inverse device - sweetness, beautiful tonal character, and a relaxed, natural sound that is also revealing of small details.
So I'll leave it to others to explore what horns and waveguides can do - the choice of the tweeter, in all honesty, is a relatively small part of the overall design, and very much subject to personal taste. The most important decision - and this is subjective, again - is the choice of the wideband driver.
This driver must be musically consonant and pleasing to listen to full-range; the tweeter and LF drivers are add-ons that are chosen after the widerange driver is selected. This is the same protocol that I used for the Ariel - it took several years to settle on the 5.5" Vifa midbass driver, which then drove the rest of the design. BudP's EnABL process expands the choices, which is all to the good, since there are very few really beautiful-sounding drivers in the 96~100 dB/metre efficiency range.
The concept of an open-baffle system with a single widerange driver, assisted by a frequency-selective open-baffle bass array, is pretty open-ended. A ribbon vs waveguide+compression driver will give a very different HF presentation, which is a matter of taste as much as anything. Similarly, choices in widerange driver span the distance from a modified Visaton B200 to highpass-filtered Lowthers and AERs to different models of guitar and prosound speakers to PHL and Fertin field-coil drivers.
Looking further, the Geddes waveguide might possibly be the only waveguide to be free of internal reflections - don't know how the LeCleac'h and Tractix profiles would compare in this respect, or if felt lining on the inside of a LeCleac'h or Tractrix would have a similar effect to the specially selected open-cell foam that Earl uses to fill his waveguide.
This endless process of getting horns or waveguides to perform as theory says they should takes me far away from the goals of the new speaker - I just want a good, efficient tweeter, not to re-invent horns. A lot of really bright people (starting with Bell Labs) have been working on these blasted things for 80 years now, and it's become obvious there is still quite a lot of important, fundamental information to be discovered - just getting a reasonably smooth wavefront out of the thing is a non-trivial exercise.
The shape of the real wavefront, not the simple spherical idealization that people usually think of, is a big deal for the operation of the crossover and the overall image quality. Narrow-beam spikes or holes are very undesirable - although hard to measure, they are easy to hear, and make balancing the system very tricky and time-consuming.
I've spent enough time twiddling and tweaking with the much smaller problem of cabinet-edge diffraction to take the shape of the wavefronts very seriously - and reading EG's writings closely and having extended phone chats with Alexander of RAAL have made it clear that deformed, uneven wavefronts are a serious problem with high-efficiency tweeters. Although the vertical height of a tall ribbon has implications for the overall frequency balance of the speaker at 2, 3, 4, and 5 meters (the narrow vertical beam of the ribbon falls off less with distance than the cone speakers), at least the impulse response is well-behaved over a wide lateral arc, an important criteria for image quality.
I also have admit to I like the sound of ribbon microphones (RCA 44B/BX) compared to condenser mikes, and expect similar virtues in the inverse device - sweetness, beautiful tonal character, and a relaxed, natural sound that is also revealing of small details.
So I'll leave it to others to explore what horns and waveguides can do - the choice of the tweeter, in all honesty, is a relatively small part of the overall design, and very much subject to personal taste. The most important decision - and this is subjective, again - is the choice of the wideband driver.
This driver must be musically consonant and pleasing to listen to full-range; the tweeter and LF drivers are add-ons that are chosen after the widerange driver is selected. This is the same protocol that I used for the Ariel - it took several years to settle on the 5.5" Vifa midbass driver, which then drove the rest of the design. BudP's EnABL process expands the choices, which is all to the good, since there are very few really beautiful-sounding drivers in the 96~100 dB/metre efficiency range.
The concept of an open-baffle system with a single widerange driver, assisted by a frequency-selective open-baffle bass array, is pretty open-ended. A ribbon vs waveguide+compression driver will give a very different HF presentation, which is a matter of taste as much as anything. Similarly, choices in widerange driver span the distance from a modified Visaton B200 to highpass-filtered Lowthers and AERs to different models of guitar and prosound speakers to PHL and Fertin field-coil drivers.
I could quote Lynn's website, which I highly recommend reading through and through. It's a great deal of valuable information:
"Returning to the midrange, I cannot recommend any other midbass driver as a substitute. If you want to use any other drivers you are completely on your own."
found at http://www.nutshellhifi.com/Arieltxt1.html#td
I recommend buying two full sets of drivers (8 midrange, 4 tweets) if you're serious about having a set this late in the game. The midranges are cheap from Madisound. Tweets are pricey, though you could buy two of those and order two more of just the voicecoils (only $23.50 USD, http://www.madisound.com/catalog/pr...d=911&osCsid=43ce6fe2f3943064ed479f06618f7973)
Good luck.
"Returning to the midrange, I cannot recommend any other midbass driver as a substitute. If you want to use any other drivers you are completely on your own."
found at http://www.nutshellhifi.com/Arieltxt1.html#td
I recommend buying two full sets of drivers (8 midrange, 4 tweets) if you're serious about having a set this late in the game. The midranges are cheap from Madisound. Tweets are pricey, though you could buy two of those and order two more of just the voicecoils (only $23.50 USD, http://www.madisound.com/catalog/pr...d=911&osCsid=43ce6fe2f3943064ed479f06618f7973)
Good luck.
Lynn Olson said:
A link to the discussion that took place here would be nice.
Dr. Geddes appeared to have some "rocky" starts with attempts at discussion in other forums as well - which is too bad.
He talks about HOM - higher order modes in particular. Which is interesting to me.
Also he mentions that he designs his wavguides to fit particular driver geometry - it's not general purpose.
Measurements published in the British magazine Studio Sound many years ago showed Cepstrum results that indicated that the axysymmetric Tractrix horn that they tested had characteristics on par with or better than the Quad ESL - which is very good (within its dynamic range limits) compared to 99% of all speakers.
I had an interesting read/discussion with another designer who uses the matrix method to solve certain speaker design issues. Admittedly I am not very good at mathematics, having stunted training and being many years separated from that! But, I was able to grok how Geddes does his stuff as a result.
Lynn, I think ur overthinking this. Few multiway dynamic speakers have a wavefront shape that is worth a darn!
But, properly done, horns have several superior qualities - small size is not one of them, if you keep to my philosophy of not having an xover in the "voice range". And, matching them to dynamic drivers in that range, if you want to keep it small is problematic as well.
Agreed.
But few dynamic speaker systems pay attention to this at all!
I am also a fan of tall speakers - ribbons or ESLs.
I own some!
I guess much has to do with personal preferences in that tricky department of compromises, as well as one's own hearing & perceptions. My own experience is that the tweeter is absolutely critical, although no one component or factor trumps all the others.
It's interesting to hear how you conceptualize your system's factors. We're not very far apart in concept, perhaps slightly different in terms of the solutions we'd employ to solve the problem?
_-_-bear