some interesting read, starting here: http://www.diyaudio.com/forums/multi-way/120505-significance-high-qms-2.html#post1639010
Kindhornman, thanks for the descriptions of the difficult tradeoffs between damping, rigidity, mass, and HF breakup modes.
After I designed the Ariel, I expected the European driver manufacturers to build on their experiences gained during the Seventies and Eighties and manufacture midbass drivers with better and better self-damping and increased efficiency. It didn't happen.
There was a mad chase for more and more rigid cone materials, which satisfied the demands of the idiot reviewers for "more bass slam" but did terrible things for HF breakup and did not yield any improvement in efficiency. In other words, distortion in the 80 to 400 Hz range was improved at the expense of everything else.
Thus, no direct successor for the Ariel, but more importantly, high-end loudspeakers as a group became harsher sounding over time. The overall FR measurements looked fine (aside from Wilson and other audio-magazine favorites), but closer examination revealed that the newer midbass drivers did not have good performance in the breakup region ... they were in fact worse than their polypropylene, bextrene, and paper-cone predecessors.
The magazine reviewers who made a fetish of "accuracy" did not notice, or deliberately overlooked, the gradual degradation in sound quality. Not many reviewers will admit (in print) that an expensive new product is in fact inferior to an older product; it doesn't attract advertisers, or readers, when you say that the latest-and-greatest ... isn't. This has particular force for magazines (or Web-pages) where the majority of the revenue is from advertisers.
I don't think it's wonderful that older stuff sometimes is better; I'm not a collector of unobtanium, and have no desire to increase already high prices. I see it as a failure of the contemporary audio-design community to pay attention to the basics of good sound quality, and more seriously, corruption in the mainstream audio-review press, which bears a large responsibility for mediocre or outright defective products getting good reviews.
One of the best things about a serious DIY forum is we can have an honest discussion of what design techniques work and which are less successful. Rather than being worthless armchair critics, we can put our ideas into practice and see how the physical realization measures and sounds. Although the Ariel is kind of an antique by now, a fair number of people built them, and know what they sound like. Same for the Amity and Karna amplifiers. When I get my own set of the new speakers going, you'll see the ARTA measurements here, although I can point to the sites that have measured the AH425 Azurahorn.
I'm finding this discussion of which drivers have good low-level detail/resolution/tonality really interesting. A consistent thread is they tend to be pretty old-school in design ... Lowther, Altec, Supravox, et al. As mentioned above, I don't feel "old" is necessarily good; it just means that something ... and we're not sure what ... has been lost in "mainstream" contemporary drivers.
I want to thank Kindhornman, and all the others, who have offered insight into what might be happening with the low-level detail/resolution/tonality issue. I have a personal interest in this since I want to choose the optimum direct-radiator driver to match the HF horn.
After I designed the Ariel, I expected the European driver manufacturers to build on their experiences gained during the Seventies and Eighties and manufacture midbass drivers with better and better self-damping and increased efficiency. It didn't happen.
There was a mad chase for more and more rigid cone materials, which satisfied the demands of the idiot reviewers for "more bass slam" but did terrible things for HF breakup and did not yield any improvement in efficiency. In other words, distortion in the 80 to 400 Hz range was improved at the expense of everything else.
Thus, no direct successor for the Ariel, but more importantly, high-end loudspeakers as a group became harsher sounding over time. The overall FR measurements looked fine (aside from Wilson and other audio-magazine favorites), but closer examination revealed that the newer midbass drivers did not have good performance in the breakup region ... they were in fact worse than their polypropylene, bextrene, and paper-cone predecessors.
The magazine reviewers who made a fetish of "accuracy" did not notice, or deliberately overlooked, the gradual degradation in sound quality. Not many reviewers will admit (in print) that an expensive new product is in fact inferior to an older product; it doesn't attract advertisers, or readers, when you say that the latest-and-greatest ... isn't. This has particular force for magazines (or Web-pages) where the majority of the revenue is from advertisers.
I don't think it's wonderful that older stuff sometimes is better; I'm not a collector of unobtanium, and have no desire to increase already high prices. I see it as a failure of the contemporary audio-design community to pay attention to the basics of good sound quality, and more seriously, corruption in the mainstream audio-review press, which bears a large responsibility for mediocre or outright defective products getting good reviews.
One of the best things about a serious DIY forum is we can have an honest discussion of what design techniques work and which are less successful. Rather than being worthless armchair critics, we can put our ideas into practice and see how the physical realization measures and sounds. Although the Ariel is kind of an antique by now, a fair number of people built them, and know what they sound like. Same for the Amity and Karna amplifiers. When I get my own set of the new speakers going, you'll see the ARTA measurements here, although I can point to the sites that have measured the AH425 Azurahorn.
I'm finding this discussion of which drivers have good low-level detail/resolution/tonality really interesting. A consistent thread is they tend to be pretty old-school in design ... Lowther, Altec, Supravox, et al. As mentioned above, I don't feel "old" is necessarily good; it just means that something ... and we're not sure what ... has been lost in "mainstream" contemporary drivers.
I want to thank Kindhornman, and all the others, who have offered insight into what might be happening with the low-level detail/resolution/tonality issue. I have a personal interest in this since I want to choose the optimum direct-radiator driver to match the HF horn.
Last edited:
Lynn,
You are indeed correct about so much of the nonsensical information that has come out over time about cones and extremely rigid materials. We have aluminum cones that are anodized and called ceramic cones. Carbon fiber and epoxy using the highest modulus materials that may be great for an airplane but sound terrible for a cone driver and even vapor deposited diamond materials for dome tweeters. How is that diamond material suppose to flex above its pistonic range without making some nasty noises?
While I found that I could easily increase the stiffness of my cone material and increase the upper frequency response up into the 10khz region it just sounds awful and had no advantage if I was going to use the cone with a dome tweeter over 2.5khz. It would just ring like crazy and keep on going long after the impulse was removed. It becomes a careful balance to get it right and that is what paper seems to do so well. I have looked at lowering the moving mass of the cone, I have some very exotic carbon fiber that I can do that with but then you get to the point that you need such soft spiders and surround that they become unobtanium or something that you have to develop on your own. That material along with the Kevlar I match it with could be used for a midrange driver instead of the wide range bass mid that I have developed. I am looking at making some custom molded surrounds as I just can't find a surround for a small driver that has a very long excursion capability. Mostly we are stuck with either a half roll or perhaps an M-roll which are generally very stiff. So I do have my work cut out for me. Some would say to just increase the diameter of the surround but those people do not understand that half the weight becomes part of the moving mass and that part of the surround is in anti-phase to the cone and why would you want to increase this area? Rubber surrounds have way to much hysteresis and return much to much energy back into the cone and that is to be avoided. I've been working with TPE-PU materials for surrounds and they are probably some of the most expensive materials to use in this application but have the best mechanical and acoustical properties. The trick in spiders is to develop a spider that has both the return properties and also a linear excursion instead of the rising rate of most standard spider designs. Springs are interesting components to work with and this is what a spider is in reality. Creating a spider that has no break-in mode and is stable over time is one of my goals that I will be working on. Many people seem to think that the fibers in the spider are breaking while working in but the real mechanical change is in the phenolic resins used that micro crack in the folds as the spider is worked over time.
You are indeed correct about so much of the nonsensical information that has come out over time about cones and extremely rigid materials. We have aluminum cones that are anodized and called ceramic cones. Carbon fiber and epoxy using the highest modulus materials that may be great for an airplane but sound terrible for a cone driver and even vapor deposited diamond materials for dome tweeters. How is that diamond material suppose to flex above its pistonic range without making some nasty noises?
While I found that I could easily increase the stiffness of my cone material and increase the upper frequency response up into the 10khz region it just sounds awful and had no advantage if I was going to use the cone with a dome tweeter over 2.5khz. It would just ring like crazy and keep on going long after the impulse was removed. It becomes a careful balance to get it right and that is what paper seems to do so well. I have looked at lowering the moving mass of the cone, I have some very exotic carbon fiber that I can do that with but then you get to the point that you need such soft spiders and surround that they become unobtanium or something that you have to develop on your own. That material along with the Kevlar I match it with could be used for a midrange driver instead of the wide range bass mid that I have developed. I am looking at making some custom molded surrounds as I just can't find a surround for a small driver that has a very long excursion capability. Mostly we are stuck with either a half roll or perhaps an M-roll which are generally very stiff. So I do have my work cut out for me. Some would say to just increase the diameter of the surround but those people do not understand that half the weight becomes part of the moving mass and that part of the surround is in anti-phase to the cone and why would you want to increase this area? Rubber surrounds have way to much hysteresis and return much to much energy back into the cone and that is to be avoided. I've been working with TPE-PU materials for surrounds and they are probably some of the most expensive materials to use in this application but have the best mechanical and acoustical properties. The trick in spiders is to develop a spider that has both the return properties and also a linear excursion instead of the rising rate of most standard spider designs. Springs are interesting components to work with and this is what a spider is in reality. Creating a spider that has no break-in mode and is stable over time is one of my goals that I will be working on. Many people seem to think that the fibers in the spider are breaking while working in but the real mechanical change is in the phenolic resins used that micro crack in the folds as the spider is worked over time.
Yes, I suspected that something a lot smaller than the fibers were breaking; cracks developing in the phenolic resin makes a lot more sense. The break-in stimulus might well make a difference: large excursions make one sort of pattern, small HF excursions yet another.
Now that's an unhappy thought: the frequency spectra and duty-cycle of the break-in stimulus might permanently "set" the sound of the driver! Maybe play Bach for a week and leave it at that.
Since we were playing around with expensive Surpravox drivers a few posts ago, I revisited Wolf von Langa's Kilimanjaro Field-Coil Loudspeaker site. Presumably Mr. von Langa rebuilds JBL, Altec, and JBL frames with new field-coil magnets.
Last edited:
In the pricey category don't forget:
Home
Here are the driver's in use in an open baffle with valve amps:
A Trip to Switzerland and Da Vinci Audio Labs | The Absolute Sound
..of particular note:
"..Virtùs were uncannily lifelike; moreover, their realism was of a different order than that of other great speakers.. the Virtus and the Da Vinci electronics recreated the sound of a singer and accompanists in a church as if I were sitting before that singer and those accompanists in that church. It was one of the most jaw-dropping listening experiences I’ve ever had. I’m used to the more-or-less-realistic reproduction of singers and musicians; I’m also accustomed to hearing the ambience of the hall they play in. What I’m not used to—and what the Virtùs were capable of—was feeling as if I had been transported to an entirely different venue, as if I weren’t hearing reproduced music in a listening room but were present at the event. The illusion of “being there” was that much more complete.."
Though not inexpensive, Enviee makes a nice small midbass driver (with a pretty wide variety of applications) with a very low Rms for an 8" driver:
http://www.enviee.de/en/bass/data.html
Home
Here are the driver's in use in an open baffle with valve amps:
A Trip to Switzerland and Da Vinci Audio Labs | The Absolute Sound
..of particular note:
"..Virtùs were uncannily lifelike; moreover, their realism was of a different order than that of other great speakers.. the Virtus and the Da Vinci electronics recreated the sound of a singer and accompanists in a church as if I were sitting before that singer and those accompanists in that church. It was one of the most jaw-dropping listening experiences I’ve ever had. I’m used to the more-or-less-realistic reproduction of singers and musicians; I’m also accustomed to hearing the ambience of the hall they play in. What I’m not used to—and what the Virtùs were capable of—was feeling as if I had been transported to an entirely different venue, as if I weren’t hearing reproduced music in a listening room but were present at the event. The illusion of “being there” was that much more complete.."
Though not inexpensive, Enviee makes a nice small midbass driver (with a pretty wide variety of applications) with a very low Rms for an 8" driver:
http://www.enviee.de/en/bass/data.html
Last edited:
Any idea of the prices?
Here is another interesting candidate with a low Rms (and its bi cone brother has klippel measurements), but with some breakups to tame...
Dayton Audio PM220-8 8" Wideband Midbass Neo Driver 295-345
Here is another interesting candidate with a low Rms (and its bi cone brother has klippel measurements), but with some breakups to tame...
Dayton Audio PM220-8 8" Wideband Midbass Neo Driver 295-345
Last edited:
This discussion has strayed quite close to the 'Leak Sandwich' speaker design at times !
Leak Sandwich Loudspeaker
I once spent a very fine afternoon listening to many LPs played on Leak Sandwich speakers, driven by a pair of Single-ended 4212 amplifiers ( the STC one, aka. 'v1505' is the best audio tube I have ever heard ) .
IslandPink,
Yes composites made with a core and skins have been used in many application that I know of. There are others who do similar speakers using alternative materials but we always come down to a fairly thick cone with a combination of low mass core and stiff skins.
Many years ago when I first started making horns I built a massive horn that was about 7 foot long on axis, it was a curved horn design with a rear sealed enclosure and I think it had a 75hz mouth cutoff size. We used a similar construction on that horn. It had skins of molded fiberglass and it was filled with low density urethane foam as a core. The horn worked great but trying to mass produce the parts would have been more than a bit extravagant for a production piece. It did sound awesome in bass reproduction and I have used that technique to build other things besides speakers.
Yes composites made with a core and skins have been used in many application that I know of. There are others who do similar speakers using alternative materials but we always come down to a fairly thick cone with a combination of low mass core and stiff skins.
Many years ago when I first started making horns I built a massive horn that was about 7 foot long on axis, it was a curved horn design with a rear sealed enclosure and I think it had a 75hz mouth cutoff size. We used a similar construction on that horn. It had skins of molded fiberglass and it was filled with low density urethane foam as a core. The horn worked great but trying to mass produce the parts would have been more than a bit extravagant for a production piece. It did sound awesome in bass reproduction and I have used that technique to build other things besides speakers.
I believe that the 12" Fertin is a little less than 3k US each, and the fullrange a bit less than that.
Last time I looked the Enviee (bass driver) is a bit over 300 US each. Note that the Enviee has about half of the "fullrange" version's Rms value (which is listed under small signal) - and the Dayton is still reasonably low.
At a lower price-point (133 US each), there is the Audax PR240M0 with still a fairly low Rms for its size and decent sensitivity:
The Madisound Speaker Store
Last edited:
I just don't see the utility of a electro magnetic speaker over a permanent magnet design and why anybody would pay such a premium for that technology? The Audax speakers I have to say I have found to be very good designs for the simplicity that they use in there motor designs and use of components. I have used Audax and Polydax drivers for a long time and am rarely disappointed in their overall sound quality. Yes there may be better drivers but for the price they are hard to beat.
Ok, so for the 8" to 12" drivers discussed so far we have:
Audax PR240mo: Rms=0.78, $133
Dayton PM220-8: Rms=0.46, $138
Enviee bass: Rms=0.45, $300 (based on Mmd and not Mms...)
Fertin 30 model 7: Rms=0.44, $3000
Fertin 21 model 7: Rms=0.24 (!), unknown price
That new dayton midbass looks like a good deal to test this low Rms theory...
On the other hand the design decision that lead the choice of the cone material is exactly what Lynn described above as a bad trend in the hifi industry:
Announcement: Dayton Audio PM Series Midbass Drivers
But would that really be a problem when crossed over (possibly sharply) and EQed properly (ie actively) in the 200Hz-1kHz range for example?...
Those breakups above 2kHz do not look very different than what you get with typical vintage paper cones...
You could say the same for Alnico.. or even at a much lower cost level neo..
It also depends on the overall design of the motor itself, both field integrity under continuous operation and the strength of the magnet itself, as well as it's interaction with the VC.
If manufacturers would start using a counter-wound fixed coil a'la 18sound "Active Impedance" - then most of the issues pertaining to magnet material (or electromagnet operation) would largely be avoided. Then it would be left to all those other issues.
Most of the Audax drivers (ACC) need some technical "updates" though to lower non-linear effects. Of course the same could be said for Supravox or other niche manufacturers that rely on some of the better qualities of decades-old design.
Yes, the "EQ fixes everything" belief. A true perennial, going back to the days of mass-market 10-band equalizers, and the B*** 901, with 9 radio-TV speakers and lots of equalization.
I guess it depends on the reviewer and the listener. If the only music they've been exposed to is of electronic origin, and already heavily processed, maybe EQ really does fix everything. EQ'ing isn't as successful on music of acoustic origin; it can compensate for broadband peaking (if the Q is low enough), but cannot undo cone breakup.
This is revealed on ARTA and MLSSA as a sharp departure from minimum-phase radiation; if you look at the driver by itself, and have either a large IEC baffle or a low-diffraction conventional baffle, you can see the frequency where the driver departs from uniform cone motion. Drivers with good self-damping have a controlled entry into this region, and the transition region can be an octave or more. Rigid drivers (titanium, aluminum, carbon-fiber, Kevlar, etc.) have abrupt transitions into the breakup region, and the behavior in the breakup region is chaotic, with 20 dB peak-to-dips, and sharp changes in FR with small shifts in microphone position.
As far as I can tell, some audiophiles and probably the majority of magazine and webzine reviewers cannot hear driver breakup, unless it is very severe. The giveaway in the reviews is all the music referenced is of electronic origin or FX-driven movie soundtracks. You're not going to hear driver breakup on a movie soundtrack dominated by phaser blasts, gunshots, and explosions.
Breakup changes the sound of massed violins to a metallic screech, and no amount of EQ can get rid of it ... merely dull the sound, and make the screech a little less offensive. Same thing for massed choral. If it sounds like chainsaws are revving up behind the choir, yes, that's driver breakup. It's not subtle.
Sharp 24 dB/octave crossovers can help by pushing the breakup region 20 to 30 dB below the main signal ... but if you listen carefully to the driver, with the rest of the drivers silenced, it's still audible. It's just masked by the sound of the other drivers in normal operation ... but can still lead to a sensation of fatigue and a subtle metallic coloration to the overall sound. This is what drives me out of the room at most hifi shows.
Relying on masking is something speaker designers like to do (I include myself here), but it's a dangerous practice. The speaker will sound "good" in the usual sense, low coloration, musical, all the things we like, but over time, you'll get a little uneasy with the sound. You'll want to turn it down, or shut it off completely. This is a warning sign that something's been left undone. The onset of fatigue points to residual audibility of breakup, resonances, or low-level problems in the electronics.
Last edited:
I was almost about to go to youtube to look for a performance with actual chainsaws behind the choir. I'm certain such thing exists.
Scott,
I am aware of the winding technique that 18sound is using and besides a little higher cost it is available to anyone who is willing to pay the price. Yes it increases the heat transfer over a normal wound double layer coil but I don't see any other changes that can be attributed to that winding technique. I don't know if it is used with an edge wound wire but I will look into that.
PS. Today I just received my beryllium dome samples from Brush Wellman. On to the dome tweeter development program........
I am aware of the winding technique that 18sound is using and besides a little higher cost it is available to anyone who is willing to pay the price. Yes it increases the heat transfer over a normal wound double layer coil but I don't see any other changes that can be attributed to that winding technique. I don't know if it is used with an edge wound wire but I will look into that.
PS. Today I just received my beryllium dome samples from Brush Wellman. On to the dome tweeter development program........
All you'd have to do is truncate PCM to 8 or 10-bit resolution, leave off dithering at the encode end, and I think you'd get about the same effect. The bigger the choir, the more obvious it would be.
You could then run the truncated PCM through a lossy codec, like MP3 or Dolby Digital, to get the full effect.
I think the real economic downside of the 18Sound technique is the requirement for extra space in the gap to make room for the additional windings, and the additional cost of a bigger magnet to offset the loss of BL product. But neo magnets aren't that expensive, compared to the cost of labor of building the driver.
Have fun with the beryllium domes, I think you'll like them. Be prepared to re-EQ the crossover, though; the subjective balance is quite a bit different than the usual materials.
Lynn,
One of the major premises of my design philosophy is to do as little damage as possible. Without giving the secrets away to what I am doing with cone development was to leave the thinking of extreme stiffness behind. It does cause the problems that you are talking about and it really doesn't matter if it is done with a metallic or organic material, they all have the same problems. What that is is a loss of damping qualities, the materials though extremely stiff have very high Q resonances. This follows with your observation that you can have 20db peaks and even dips and that much of this is in band and not out of band. Using eq to correct this is a fools paradise, you can never actually remove the sound, as you say you can only try to hide or mask it at best. I have stayed away from 1/3 octave eq since my days in PA and understanding how these units worked, they are the worst possible way to correct the problems as they introduce such terrible phase shifts in not only the band being manipulated but in the adjacent bands that are affected also by the interaction of the circuit. If you have to do it use a parametric eq, they are less prone to this problem.
So my intention is to give you a speaker that is flat in its passband without correction from an outside electrical correction. I don't expect someone to have to build a crossover that has to eq the passband and I have designed the motor in such a way as to cause the least impedance rise possible even though that does cause some decrease in efficiency. It is such an art of balance in designing a working speaker. Trying to do everything with one unit or attempting to correct any anomalies with corrective electrical correction is just the wrong approach as far as I am concerned. When I hear someone say they can correct a high Q resonance or counter a mechanical resonance with an inverse electrical function I know we are going to have problems. I repeat what I have said and was pooh poohed by a know designer that you can not remove a mechanical self resonance with an electrical circuit. The resonant property of the device does not change and any acoustical output at that resonant frequency will again excite the problem frequency. So unless someone is willing to live with a large hole in the program material the problem will always exist.
One of the major premises of my design philosophy is to do as little damage as possible. Without giving the secrets away to what I am doing with cone development was to leave the thinking of extreme stiffness behind. It does cause the problems that you are talking about and it really doesn't matter if it is done with a metallic or organic material, they all have the same problems. What that is is a loss of damping qualities, the materials though extremely stiff have very high Q resonances. This follows with your observation that you can have 20db peaks and even dips and that much of this is in band and not out of band. Using eq to correct this is a fools paradise, you can never actually remove the sound, as you say you can only try to hide or mask it at best. I have stayed away from 1/3 octave eq since my days in PA and understanding how these units worked, they are the worst possible way to correct the problems as they introduce such terrible phase shifts in not only the band being manipulated but in the adjacent bands that are affected also by the interaction of the circuit. If you have to do it use a parametric eq, they are less prone to this problem.
So my intention is to give you a speaker that is flat in its passband without correction from an outside electrical correction. I don't expect someone to have to build a crossover that has to eq the passband and I have designed the motor in such a way as to cause the least impedance rise possible even though that does cause some decrease in efficiency. It is such an art of balance in designing a working speaker. Trying to do everything with one unit or attempting to correct any anomalies with corrective electrical correction is just the wrong approach as far as I am concerned. When I hear someone say they can correct a high Q resonance or counter a mechanical resonance with an inverse electrical function I know we are going to have problems. I repeat what I have said and was pooh poohed by a know designer that you can not remove a mechanical self resonance with an electrical circuit. The resonant property of the device does not change and any acoustical output at that resonant frequency will again excite the problem frequency. So unless someone is willing to live with a large hole in the program material the problem will always exist.
Last edited:
Practicality aside, I'm pretty sure that there is no technical reason why a mechanical resonance can't be "corrected" back to linear as long as it's minimum phase, and I'm pretty sure I've seen a whole bunch of people prove that it is, so maybe that's why you were pooh poohed. There are plenty of other explanations for ending up with poor results after doing so.
dumptruck,
What I am saying is that yes you can make it look like a flat frequency response with an electrical correction to a steady state condition for a mechanical resonance. But, big But here, the mechanical resonance of the device does not change. So even if you remove or have an inverse electrical function that makes it look like the resonance is gone it is not. The mechanical excitation will still be there, that doesn't change, so the problem really doesn't go away. If you produce the acoustical frequency that excites that frequency the device will resonate. You can cut a hole in the frequency response to make it look like it is gone but it is still there mechanically. The mechanical and electrical functions are separate, one can not correct the other. I am not talking about just the FR, I am talking about the mechanical function.
What I am saying is that yes you can make it look like a flat frequency response with an electrical correction to a steady state condition for a mechanical resonance. But, big But here, the mechanical resonance of the device does not change. So even if you remove or have an inverse electrical function that makes it look like the resonance is gone it is not. The mechanical excitation will still be there, that doesn't change, so the problem really doesn't go away. If you produce the acoustical frequency that excites that frequency the device will resonate. You can cut a hole in the frequency response to make it look like it is gone but it is still there mechanically. The mechanical and electrical functions are separate, one can not correct the other. I am not talking about just the FR, I am talking about the mechanical function.