Graham Maynard said:
The variable-geometry example I gave above was the theoretical principle, in essence an alternative to the Linkwitz approach, with all of its power-hungry equalization requirements. It trades equalization and amplifier power for an increased number of drivers, connected in parallel or series-parallel, depending on impedances and crossover realization.
In practice, it should be simpler than the initial suggestion. There's no reason the "jumps" between lowpass filters and groups of drivers have to be one octave apart - this is determined by real-world room-gain measurements, instead of theory.
Also, the lowest-frequency set of driver(s) can really be anything you like: H or W-box dipoles, closed-box monopoles, heck why not something really different, like a TQWT! That would go from the ridiculous inefficiency of dipoles to the impressive efficiency of horn-loading with a quarter-wave resonance thrown in.
I'm expecting a real-world system has whatever kind of driver you like for the wideband driver - Lowther/AER, a line array of Fostexes, the notorious/famous Visaton B200, something gorgeous from 18Sound, the mythical good-sounding coaxial unicorn, the creme-de-la-creme field-coil Fertin, whatever. Then directly below that is a high-Q physically large driver for midbass. And below that whatever kind of woofer you like. With an oddball array like this, the ideal crossover spacing could be anything - again, to be determined by audition and measurement.
Hi Lynn,
Yes so many variables, and I think the big problem we eventually hit is spectral coherence from different drivers.
You suggested tilting the baffle slightly backwards; I assumed for coincidental driver timing in the vertical plane.
Hi John,
You stated that tilting the rear-wave downwards has a significant effect upon the soundstage.
I don't dispute the soundstage shift....
but I still can't get my head properly around the statement,
because what we hear of rear output is always delayed, dispersed and reflected.
So I ask - does the soundstage become altered because it is actually the baffle edge, and thus the 'zero reference' launch plane that becomes tilted wrt the driven source we hear in our vertical plane ?
If yes,
then cannot soundstage shifting due to baffle back-tilting with HF drivers at the top, be countered by wings having equal but opposite slope, thus gradually extending backwards towards the floor - this going towards the 2D LF dimensioning you suggested ?
Cheers ....... Graham.
Yes so many variables, and I think the big problem we eventually hit is spectral coherence from different drivers.
You suggested tilting the baffle slightly backwards; I assumed for coincidental driver timing in the vertical plane.
Hi John,
You stated that tilting the rear-wave downwards has a significant effect upon the soundstage.
I don't dispute the soundstage shift....
but I still can't get my head properly around the statement,
because what we hear of rear output is always delayed, dispersed and reflected.
So I ask - does the soundstage become altered because it is actually the baffle edge, and thus the 'zero reference' launch plane that becomes tilted wrt the driven source we hear in our vertical plane ?
If yes,
then cannot soundstage shifting due to baffle back-tilting with HF drivers at the top, be countered by wings having equal but opposite slope, thus gradually extending backwards towards the floor - this going towards the 2D LF dimensioning you suggested ?
Cheers ....... Graham.
Here is a study that suggests this is a good idea (look at test "H": not a back wing, but it should be similar) http://www.troelsgravesen.dk/OBS.htm
I like Lynn´s "variable geometry" concept very much. I see it as the setting of general design principles, which as Lynn says, are then to be realized accordingly to real world measurements with real drivers. His side and back firing tweeters idea is also very good.
A second and most important design principle still to be determined is how high (at least) the wide-range mid should do. We have spoken of the 500-3kHz "telephone bandwidth" as a most critical, but also about the wish of crossing over to a supertweeter at 10kHz. Determining the minimum critical midband requirements on the base of theoretical consideration could help narrowing down the selection of the mid.
A very interesting idea that could be discussed and that could complement Lynn´s variable geometry concept was brought up by Thorsten some time ago in this forum. If one uses a transimpedance -current source amp- with a driver which has an appropriate Qms, the use of such an amplifier together with the driver´s rising impedance on the lower end could be a very simple way of equalizing the 1/f losses. John Swenson has spoken here of a very nice transimpedance tube amp of his. Hawksford paper on current driven speakers shows a marked distortion reduction, from memory something like -20dB.
I like the idea of thinking about the system as a whole. People like Lynn that have heavenly good tube amps would perhaps not be very inclined to set them aside and design a whole new system. But maybe biamping can be an alternative.
H and W subs have the resonance peaks at 200-300Hz. One "advantage" of the dipole subs concerns room modes below ~300hz (depending on rooms dimensions) to about 60hz: in that range one would have to decide between more "kick" and better linearity of the upper low end. Below 60Hz I would support Lynn´s option of doing a monopole or a TQWT.
Hello,
Well this is not realy related to the current discusion but someone mentionde earlier the use of two 6" 18sound drivers. Of course these shoud have a low cross point with the tweeter because of the polar response (but they actualy beacame almost useless because the bandwith they are handling is very narrow) and so on but: Why highly praised speakers like Zu are using 10" drivers in what it seems to me a d'Appolito arangement except the very high cross point of 10Khz. The two 10" units are runing fullrange almost. I've seen other speakers (wich again, were well reviewed) using these strange D'Appolito configurations. Everything says something is not normal (from the theory point of view) but they sound well it seems.
So... what's the answer... am I missing something or it's just well.. marketing?
Well this is not realy related to the current discusion but someone mentionde earlier the use of two 6" 18sound drivers. Of course these shoud have a low cross point with the tweeter because of the polar response (but they actualy beacame almost useless because the bandwith they are handling is very narrow) and so on but: Why highly praised speakers like Zu are using 10" drivers in what it seems to me a d'Appolito arangement except the very high cross point of 10Khz. The two 10" units are runing fullrange almost. I've seen other speakers (wich again, were well reviewed) using these strange D'Appolito configurations. Everything says something is not normal (from the theory point of view) but they sound well it seems.
So... what's the answer... am I missing something or it's just well.. marketing?
swak said:
Yes, I think H and W subs are my last choice - they combine the appalling inefficiency of dipoles at long wavelengths with the resonances of parallel-sided boxes. That just feels wrong at an intuitive level, something I trust in speaker design.
I think you can do a lot with a flat baffle and one side wing - with a C-shaped cutaway profile as seen from the side, so it has the greatest depth at floor level, fading away to zero by the time it reaches the bottom of the widerange driver. If the dipole can make it down to 60 ~ 80 Hz (which isn't asking a whole lot) a completely different kind of (stereo) sub can take over from there. TQWT's and bass horns certainly qualify, with tremendous efficiency and physical "kick". That would be a great complement to the large-area dipole sound.
As for amps, the deluxe amps can be reserved for the wideband driver and the tweeters. Everything below that is going to need room EQ anyway, since the desirable region for room EQ is 300 Hz on down. Back in the days of the SAE parametric EQ (I still have one!), I used it to EQ rooms - and found out quite early that midband EQ, or attempting to "fill in" holes, is a prescription for disaster. But using the variable-Q notch filters to chase out the two biggest bass bumps - on a per-channel basis - really improved the sound a lot. It also did wonders for the imaging, something I didn't expect.
It was interesting to see at the last RMAF that Wally of Wallytractor fame is also using bi-amped speakers (of his own design) with pro EQ for the 240 Hz on down region, and goes out to client's houses to personally EQ their systems. If you've ever met Wally, he's an old-school European with excellent taste in music and sound, so I have no doubt he tunes the system to sound beautiful. What I heard at the RMAF certainly sounded agreeable and musical.
SunRa said:
Don't be that impressed by the Zu's - I heard them, eh, another speaker made by a cable company. They are good at marketing, though, they could teach Nike a thing or two.
After struggling with the Ariel crossover for six months, and accepting a response curve I wasn't completely happy with, I became sensitive to an odd kind of coloration that MTM's make. For lack of a better word - this is going to sound weird - they sound hollow and forward at the same time. I could only tame this in the Ariel by intentionally allowing the mids to sink into a mild "BBC dip", which gives the speaker a subjectively neutral balance.
If the Ariel uses a conventional "measure-flat" 6 or 18 dB/octave crossover the coloration returns, so this is some kind of weird conflict between the direct-arrival spectra, the "total power into the room" spectra, inter-driver phase angles, and tight vertical lobing.
This coloration is much worse in WMTMW's or large-diaphragm MTM's. Admittedly, not everyone is aware of the coloration - but, like the notorious "rainbow effect" of microdisplay TV's with color shutters, once you see (or hear) it, you never forget it. (I was stupid enough to train myself to see it - it's most obvious on high-contrast black-and-white content - and now I see it everywhere on big-screen HDTVs as an almost subliminal flicker.)
I enjoy the Ariels because I pretty much "dialled out" the coloration with very small phase trims in the response of midbass drivers - I actually sat in the listening position with a long cable going to the crossover and slowly adjusted a damping resistor while listening to pink-noise. At one setting the coloration suddenly dropped out; I measured the value of the resistance, replaced it with a precision Ohmite power resistor, and did more evaluation with music from there. A tedious business.
I did the Ariels as a kind of a goof - it was originally intended as a quick-n-dirty transmission-line speaker with enough efficiency to satisfy the local triode nuts. I was fooled by years of reading about the myriad advantages of MTM's. The tech literature blinded me to the fact that I had yet to hear any MTM - from any manufacturer - that sounded musical or even listenable - they all had gross and unacceptable midrange colorations and almost non-existent portrayal of depth (a bad sign).
Looking back, I was overconfident and a bit cocky, thinking I could solve problems where nobody else in the industry had succeeded - or even acknowledged existed. Well, I did more or less solve it, but I still don't think MTM's are the way to go - the midband colorations are too strange, and too hard to remove with drivers that are too big, crossovers that are too complex, or drivers that are too rough. There are a lot of compromises in the Ariel that I wouldn't do now - but then again, good-quality efficient drivers were pretty hard to find back at the time I designed it, and MTM was about the only way to get into the 91~92 dB/metre region.
The Ariel experience forcibly reminded me that it is dangerous - or at least a huge time-waster - to solve problems that the rest of the industry has failed to solve. That's partly why horns scare me - so many brilliant people have worked on them for almost a century, and they still sound like horns, every blessed one of them, no matter what the pedigree, Geddes, Azura, Tractrix, JBL, Oris, Avante-Garde Duos and Trios, TAD, Altec, etc. etc. I've heard all of these at considerable length, in many different systems, in many different locations around the world.
Yes, some are much better than others, and the designs are getting better with powerful new analytic techniques like finite-element analysis. But it's ridiculous to say any of these horns sound like a direct-radiator or an electrostat. They just plain don't. All of them, even the newest and most exotic, sound very much like what they are - horns. It is obvious in the first minute of listening.
The problem of horn coloration is very deep and very serious, and nothing to be solved with a little weekend math and a dash of creativity.
I'm not as concerned about dipoles, since I've heard three completely different but wonderful systems from Linkwitz, Bastani, and Gary Pimm. This is great news, that dipoles built from such radically different set of design principles all sound so wonderful - and free of the obnoxious box colorations that plague just about every high-end speaker. The relatively minor "issues" I heard with each came down to choice of driver, crossover topology, the usual mundane stuff.
No show-stoppers, unlike horns, which seem to be one show-stopper after another (the more you discover, the worse the problems get). This is very encouraging to me, and why I've started this thread in the first place, to stimulate ideas and find out what others are doing.
jzagaja said:
Don't buy it. There are plenty of AES Journal studies showing that this or that digital-compression scheme is "inaudible" to the (unspecified) group of listeners - which to me says more about the design of the study than the conclusion the authors want you to reach. My degree is in psychology, and I can tell you there are many ways to make a study reach a misleading conclusion - selecting the group of listeners is a good start, and then you use a test protocol that confuses and stresses the listener.
Also, pink-noise is the first stimulus you use to assess subjective speaker coloration, not some kind of minor afterthought. The BBC started using pink-noise as a subjective tool in the late Fifties, so it wasn't exactly new when Holland and Newille wrote their paper in the mid-Eighties.
On a personal level, I can't accept the study's conclusion. I've heard many different horn systems, with every imaginable profile, and the horn "sound" was immediately audible. After a few minutes of listening, you come to terms with it, if it isn't too strong, but it never goes away completely.
Music with a broad and dense spectrum like a massed chorus, or massed violins, are what bring out the coloration most severely, to the point of unacceptability for many horn systems. Of course, many listeners today never listen to classical music, and many audiophiles only listen to jaz or blues with very few instruments, thus a sparse spectrum that doesn't excite the horn coloration.
The audibility of horn coloration seems to strongly depend on the listener - I'm convinced that some people just don't hear it at all, and think that everyone else is complaining about nothing. I think other audiophiles enjoy it as a euphonic coloration, one they positively enjoy as a musical enhancement. Other people detest it - I'm closer to this group, but not all the way over in their corner. I surmise this is a lot like color perception, with significant differences in individual perception of overall color gamut, resolution of subtle hue-differences, and color harmony.
When listening to horns, I'd rather listen to the "new-generation" horn designs, such as the Azura/Lowther, Oris/AER, or the Geddes Summa, instead of the old-school Altec/JBL aluminum-diaphragm CD's with an exponential horn. But ... all of them, new or old, sound more like each other than any direct-radiator or electrostatic speaker, which have a completely different presentation.
Lynn I have had my stint with horns in the pro world, that was my job, even messed with large line arrays, and with MTMs since I have designed decades of home speakers.
I must agree that MTMs have the trait you mention. My idea is that by mutually canceling lower than crossover frequency with each midwoofer, they eat out power response there, hence the power response in the room is always stronger around cross. By dipping them in cross I can ameliorate that but I always hear the hollow then upfront transition.
A BBC dip in my knowledge isn't a mid dip but a -2dB slant from 100Hz to 10Khz though.
As for horns the only thing I can put in a word even for the most refined ones (EAW NT series catering with DSP for horn honk named Gunness Focusing) is that they sound 'projected'. Well that's why they are made for, to mainly save and project energy to a defined area.
I must agree that MTMs have the trait you mention. My idea is that by mutually canceling lower than crossover frequency with each midwoofer, they eat out power response there, hence the power response in the room is always stronger around cross. By dipping them in cross I can ameliorate that but I always hear the hollow then upfront transition.
A BBC dip in my knowledge isn't a mid dip but a -2dB slant from 100Hz to 10Khz though.
As for horns the only thing I can put in a word even for the most refined ones (EAW NT series catering with DSP for horn honk named Gunness Focusing) is that they sound 'projected'. Well that's why they are made for, to mainly save and project energy to a defined area.
Lynn Olson said:
I think that I agree with you there. Because the most neutral, uncolored, realistic systems I have ever heard were horn based. Compared with the very best planars, e.g., Quad ESLs and Magnaplanars, the horns always sounded more like real music to me. All other direct radiating systems sound like speakers to me, as nice as they may be.
When I lived in Paris I was lucky enough to hear, use and live with a number of high end speaker technologies. At the time I also worked in the music business and lived with a cellist
. My ears were full of live music every day, all day long. As great as some of the systems I heard were, none sounded more like real music to me than the horns. It was not a matter of nuance, the horns nailed it.The very big horn systems that we put together at the Kiron Theatre demos (designed by J.Hiraga) were by far the best imaging of any system I've heard before or since. Absolutely no problem with mass choral or strings. In fact, that's where they really shone.
So when I was in the thick of live music every day, all day, the good horn systems were by far the most life-like, the most realistic to my ears. Lack of coloration and dynamic ease where where they stood out. No direct radiator was as good, tho many were still quite wonderful to listen to.
But maybe horn colorations just don't bother me. I certainly did not hear them in those systems. I do hear them in lesser horn systems, i.e., most P.A. systems. Is it a matter of taste, or do we actually hear differently?
Graham,
We cheated and were planning on using a chip amp module mounted right next to each driver. All actively filtered.
There is a lot to like about this approach. The combined output of all the small amps adds up to pretty massive power, especially when you add the efficiency of the active crossovers, without the problems of High power amps- especaially DIY high power amps! Being so close to the drivers, they should have good control over the cone, and as you mention, no issues with impedence.
The amp modules wouldn't have to be chip amps. Nelson Pass used a column of class A amps in his Rushmore speakers. Of course for Lynn this won't do for all the drivers- maybe the lower frequency drivers though...
panomaniac said:
I think that I agree with you there. Because the most neutral, uncolored, realistic systems I have ever heard were horn based. Compared with the very best planars, e.g., Quad ESLs and Magnaplanars, the horns always sounded more like real music to me. All other direct radiating systems sound like speakers to me, as nice as they may be.
When I lived in Paris I was lucky enough to hear, use and live with a number of high end speaker technologies. At the time I also worked in the music business and lived with a cellist
The very big horn systems that we put together at the Kiron Theatre demos (designed by J.Hiraga) were by far the best imaging of any system I've heard before or since. Absolutely no problem with mass choral or strings. In fact, that's where they really shone.
So when I was in the thick of live music every day, all day, the good horn systems were by far the most life-like, the most realistic to my ears. Lack of coloration and dynamic ease where where they stood out. No direct radiator was as good, tho many were still quite wonderful to listen to.
But maybe horn colorations just don't bother me. I certainly did not hear them in those systems. I do hear them in lesser horn systems, i.e., most P.A. systems. Is it a matter of taste, or do we actually hear differently?
There's nothing wrong with your hearing, there's nothing wrong with mine, and similarly for Linkwitz and Geddes. People really do hear differently, although the academic community sometimes says otherwise - again, it's very easy to make a study reach a pre-determined conclusion, especially if Philips or Sony are backing it.
The default bias of many academic studes - especially industry-funded ones - is that this or that sonic characteristic is not audible. Thus we are stuck with a prematurely introduced 44.1/16 sampling system, and following that, a crudely designed digital-compression system originally meant for ATSC/HDTV soundtracks (that was then retrofitted to movie soundtracks, thus Dolby Digital), MP3, and so on. A lot more money is spent on marketing than on R&D for the simple reason that the Return On Investment (ROI) is better.
Of course, the best ROI of all, unmatched by any other, is buying Congressmen and regulatory agencies. Multinational companies all over the world have discovered this is the surest and most cost-effective route to long-term profits in the USA. I know the EU folks have issues with weird technical directives coming out Brussels (RoHS, etc.), but I don't know enough to comment on the probably inevitable money=politics issues in the EU. Money and political favors probably change hands under the table over there as well, it's probably just not as obvious.
(off-topic rant off) My guess is that working with certain types of audio systems sensitizes us to certain faults, just as people who make color prints in a photo studio can see color differences that are invisible to most of us. I know I am very sensitive to stored energy, having battled against it in loudspeakers since 1975. I'm much less critical of IM distortion and high-level compression, an area where horns are dramatically superior to all other systems.
I find the horn coloration gross and immediately obvious, and not a pleasant euphonic one. But I am aware the dynamic compression and IM distortion congestion of direct-radiators (and electrostats) is completely unacceptable to horn enthusiasts. For them, horns are the only way to go.
In another example, Bob Sickler, designer of the Audionics CC-2 amplifier, living with a violin player who played at the Portland Symphony, fell in love with the sound of the first CD's from DGG. This on a big TAD horn system with all solid-state electronics and the truly notorious Sony CDP-101 CD player (yes, the first-generation unit). I thought it was some of the worst sound I had ever heard in my life - yet he and the professional violin player went on and on how wonderful the first CD's sounded. And these were people who I had great respect for their taste, sensitivity, and design talents.
Obviously, good people, with sharp hearing and decades of musical experience, are hearing different things. There is no other explanation.
What I don't accept is that XYZ or ABC type of design is "free of coloration", or the ONLY way to design a "good" speaker. This nothing more than fundamentalism (my way or the highway) applied to audio design, and is no appealing than it is in politics or religion.
Variac said:
From 300 Hz on down, sounds like a great idea! The chip amps can share a common split-rail power supply, which can sit in the bottom of the bass module. As mentioned above, the chips themselves can be mounted directly on the speaker - although you don't want the heat sinks to obstruct the air-flow or create reflections. Maybe have the draft from the speaker cool the heat sinks?
It also occured to me this is an ideal use for my old HT multichannel receiver, powering these narrowband bass drivers. It's hardly a demanding application, due to the band-limiting and cautious application of EQ (notching-out, not peaking).
I think that applies to Fostex fans also. In fact that's what I like about some Fostex's- they have good and bad points the bad being frequency response.
There seems to be a law similar to Hoffman's Iron Law: box size, efficiency and deep bass- choose two
It seems that you can get efficiency, wide bandwidth, cone excursion, but not all together!
To show I'm not dissing Fostex:
Here's a Fostex with very flat response IMHO (according to their charts!)
http://www.madisound.com/pdf/fostexdrivers/f200a.pdf
but it's 90 dB efficient
Also $400
Here's another Fostex with very flat response:
http://www.madisound.com/pdf/fostexdrivers/ff225k.pdf
And it's more than 96dB efficient!
And it's $100!
But it's x-max is 0.3mm
It's a midrange that's well behave out of the usable bandwidth..
I've heard tell that it really does better than this, but with more distortion. Crossed over higher than we've been discussing on the low end, it might sound really good. but in an open baffle? I dunno.
Variac said:
Here's a Fostex with very flat response IMHO (according to their charts!)
http://www.madisound.com/pdf/fostexdrivers/f200a.pdf
but it's 90 dB efficient
Also $400
Here's another Fostex with very flat response:
http://www.madisound.com/pdf/fostexdrivers/ff225k.pdf
And it's more than 96dB efficient!
And it's $100!
But it's x-max is 0.3mm
I wouldn't trust those curves. When I measured several Fostex drivers in the mid-Nineties for a Oregon Triode Society project, there was no correlation between MLSSA and the factory curves at all. That caused considerable consternation among the OTS'ers. The Fostexes (specifically chosen for their good-looking factory curves) measured (and sounded) so rough the project had to be abandoned. This costly and somewhat disconcerting experience made me much more skeptical of factory curves. I am especially wary whenever I see signs of smoothing, which conceals a multitude of sins.
We need a third party to measure these drivers - I'm not volunteering (for physical reasons), but if somebody out there has a decent MLS system and a big flat baffle, you could do us all a favor and measure some of these things. No smoothing, please - show us the truth, none of this happy-talk 1/3 octave-smoothing stuff, or hand-drawn graphs made with a Radio Shack SPL meter.
P. S. If there are waterfall (CSD) graphs, look at the back of the graph to get the unsmoothed frequency response. Not everybody knows that, so you sometimes see comically-smoothed FR graphs alongside the not-so-pretty-looking CSD graphs.
Variac said:
On the back of the magnets?
Tempting, I know, but do you want the magnets (and voice coils) any hotter than they are?
Voice-coil heating is the reason the horn guys have the advantage they do - their stuff runs cool, ours doesn't, thanks to VC heating. The heating itself isn't so bad, it's the several second time constant that's troublesome (slowly varying dynamic compression due to changes in copper resistance with temperature).
Actually , I was referring to the people that are concerned about magnetic effects on electronic components and currents flowing through magnetic fields. BUT if all the components are non magnetic then less concern for them there. I'm not concerned at all about that at this point. Chip amps very seldom get more than warm. Often they are just attached to a small plate of aluminum rather than even an official heat sink. If we are using PA drivers, then I don't think that drivers designed for hundreds of watts are going to get very warm with what we are putting through them.. (sure less compression is always better ) We could just space the amps a 1/4 to 1/2" inch away from the magnets, with their own heatsinks, and be fine I think.
The heatsinks would be small enough to be completely behind the magnet. Now Class A amps- that's another story..
The heatsinks would be small enough to be completely behind the magnet. Now Class A amps- that's another story..
Lynn Olson said:
I have come to believe that. Perhaps it's at the root of so many audio arguments. Some of it is condtioning, some of it innate.
But it's a subject that mertis a whole thread to itself. I'll start one later today. (Funny you should mention color, that's what I do for a living - fine art printing - even more picky than photography).
Variac: If you are putting amps right at the drivers, there are some nice Class-D amps that would fit the bill. Some of the Tripath kits and/or the Hypex-UCD stuff. If the amp is right next to the driver then some of the H-bridge amps might be possible without the output filter. TBD.
