I'm kind of puzzled why mainstream audiophilia has lost its way. The measuring gear is way, way better than anything we had in Seventies and earlier, so we can seriously investigate what's going on with drivers, horns, amplifying elements, and various storage media. Enormous progress has been made here.
But mainstream audiophile drivers - hmm, are they much better than what was around 15 or 20 years ago? I'm not so sure. Efficiency hasn't gone up at all - 87~90 dB/metre/watt (real, not advertised) still rules the roost. In power terms, that's no better than 0.5% conversion efficiency - low by any standard. Impulse response (rapid settling time) and freedom from diaphragm resonance are still the big problems, and if anything are worse with the exotic diaphragm materials.
My only guess is that hard-won experience (particularly manufacturing trade secrets) has been lost amongst the mainstream driver manufacturers, and they're trying to substitute an assortment of clever marketing tricks instead. I am very grateful that RAAL, AESpeakers, Great Plains Audio and others are building good drivers today.
But mainstream audiophile drivers - hmm, are they much better than what was around 15 or 20 years ago? I'm not so sure. Efficiency hasn't gone up at all - 87~90 dB/metre/watt (real, not advertised) still rules the roost. In power terms, that's no better than 0.5% conversion efficiency - low by any standard. Impulse response (rapid settling time) and freedom from diaphragm resonance are still the big problems, and if anything are worse with the exotic diaphragm materials.
My only guess is that hard-won experience (particularly manufacturing trade secrets) has been lost amongst the mainstream driver manufacturers, and they're trying to substitute an assortment of clever marketing tricks instead. I am very grateful that RAAL, AESpeakers, Great Plains Audio and others are building good drivers today.
gedlee said:
Hi,Mr Geddes,I'm your fans
I learned more about acoustics through read your book <audio transducers> ,thanks!
KLIPPLE also release some article about the B field / VC wire height associats the high order distortion.
And since the FEMM or Magnet or other FEA tools can an analysis the B distributing of the gap,so we can use the Matlab to calculate the BL of the giving VC,but how to accurately transform these datas to X order distortion? I think if use the cure fit way,the high order problem would be hide......
Hope your reply,thanks.
I'm also searching for some good drivers. It's amazing that with all the information published, we should see more better drivers out there. I never thought I would have to dig into driver technology, but it seems a must now. Probably the settling time is where lots have trouble with, especially with more exotic material. Some people were recommending carbon fiber cones. I wonder whether you have experienced any and what your impression is.Lynn Olson said:
I would usually be the last guy to invoke Darwin in a discussion, but the transducer marketplace beastie we have today is the inevitable result of evolutionary pressures.
--Watts are cheap: ergo low efficiency.
--High wattage specs sell: ergo high power handling (low efficiency).
--Exotic high-Young's-modulus cone materials sell: ergo under-damped stop-band breakup.
--Crossover sophistication (complexity) sells: ergo steep rolloffs and rampant notching allowing for under-damped stop-band breakup.
--Small size sells: ergo high-excursion and more IMD.
--High excursion sells: ergo low efficiency and progressive suspensions that "bounce" at their limits and inject high-order harmonics.
The market at large will always be driven by what people think they want. Of necessity, it is what it is. A rose is a rose is a Bose.
In the country of the deaf, the one-eared kings must always look to the boutiques who cater to their tastes, or to the marketplace (pro sound) where the evolutionary pressures aren't so at odds with actual fidelity.
Bill
(To be completely fair, voice-coil motor design has come a long way, and the best of today measure a lot better than the best of yesteryear. But again, the design goals are usually at the mercy of marketing.)
--Watts are cheap: ergo low efficiency.
--High wattage specs sell: ergo high power handling (low efficiency).
--Exotic high-Young's-modulus cone materials sell: ergo under-damped stop-band breakup.
--Crossover sophistication (complexity) sells: ergo steep rolloffs and rampant notching allowing for under-damped stop-band breakup.
--Small size sells: ergo high-excursion and more IMD.
--High excursion sells: ergo low efficiency and progressive suspensions that "bounce" at their limits and inject high-order harmonics.
The market at large will always be driven by what people think they want. Of necessity, it is what it is. A rose is a rose is a Bose.
In the country of the deaf, the one-eared kings must always look to the boutiques who cater to their tastes, or to the marketplace (pro sound) where the evolutionary pressures aren't so at odds with actual fidelity.
Bill
(To be completely fair, voice-coil motor design has come a long way, and the best of today measure a lot better than the best of yesteryear. But again, the design goals are usually at the mercy of marketing.)
Rant: ON
Lynn has several times underlined the necessity of transducers having well-behaved stop bands. Lest anyone poo-poo this, here is an audio parable:
The intrepid designer starts with a nice little high-end metal-coned midbass that shows beautifully pistonic behavior up to 4KHz followed by a 12dB peak where it rings like a bell. No problem, right? Just lowpass it with a 24dB/octave electrical crossover somewhere around 2.5KHz. Voila. Zee peek ees gone. Maybe 10dB down. Heck. Let's throw in a notch and knock it flat. Smugly, he finishes the mini-monitor off with a smokin' tweeter, blessed crossover components, wires of loving grace, and a Bybee Quantum Purifier. Seriously.
Now, palms clammy, he plays music through it--softly at first--girl-and-guitar. Beautiful! Everything's great. At this point, he springs up from his seat in a paroxysm of joy and runs outside to hang up his mini-monitor manufacturer's shingle.
While he's out, let's be mean. Let's turn the volume up a bit and spin one of Ray Kimber's IsoMike uncompressed recordings of a marching band. As the voice coil leaps from the gap only to be caught in the iron grip of the progressive suspension a moment later, the indignant little driver sneezes out a broad spray of harmonics, let's say 10dB below the reference level.
Remember that little 12dB peak sleeping with the fishes in the stop band? Well, the driver-generated harmonic distortion in the mechanical realm cares not a whit about the steep crossover upstream in the electrical realm. It marches unencumbered right into the stop band and shakes that peak awake. Not only that, but maybe its harmonic sequence matches the peak's ring-down, so peak is exciting peak.
Instantaneously, with each heavy transient, the previously dead-flat "pistonic" monitor's frequency response transforms. Now it features a big angry bump at 4kHz followed upward by a spray of nasty hash.
It's a Jekyll-n-Hyde sort of transformation--from song to song and track to track--that can take your breath away. And if you don't understand the mechanisms behind it, it can be a head-scratcher to diagnose.
The moral of the story is that *electrical* EQ, be it notches or the crossover lowpass, is no replacement for good *mechanical* behavior, even in the stop band.
Rant: off
Lynn has several times underlined the necessity of transducers having well-behaved stop bands. Lest anyone poo-poo this, here is an audio parable:
The intrepid designer starts with a nice little high-end metal-coned midbass that shows beautifully pistonic behavior up to 4KHz followed by a 12dB peak where it rings like a bell. No problem, right? Just lowpass it with a 24dB/octave electrical crossover somewhere around 2.5KHz. Voila. Zee peek ees gone. Maybe 10dB down. Heck. Let's throw in a notch and knock it flat. Smugly, he finishes the mini-monitor off with a smokin' tweeter, blessed crossover components, wires of loving grace, and a Bybee Quantum Purifier. Seriously.
Now, palms clammy, he plays music through it--softly at first--girl-and-guitar. Beautiful! Everything's great. At this point, he springs up from his seat in a paroxysm of joy and runs outside to hang up his mini-monitor manufacturer's shingle.
While he's out, let's be mean. Let's turn the volume up a bit and spin one of Ray Kimber's IsoMike uncompressed recordings of a marching band. As the voice coil leaps from the gap only to be caught in the iron grip of the progressive suspension a moment later, the indignant little driver sneezes out a broad spray of harmonics, let's say 10dB below the reference level.
Remember that little 12dB peak sleeping with the fishes in the stop band? Well, the driver-generated harmonic distortion in the mechanical realm cares not a whit about the steep crossover upstream in the electrical realm. It marches unencumbered right into the stop band and shakes that peak awake. Not only that, but maybe its harmonic sequence matches the peak's ring-down, so peak is exciting peak.
Instantaneously, with each heavy transient, the previously dead-flat "pistonic" monitor's frequency response transforms. Now it features a big angry bump at 4kHz followed upward by a spray of nasty hash.
It's a Jekyll-n-Hyde sort of transformation--from song to song and track to track--that can take your breath away. And if you don't understand the mechanisms behind it, it can be a head-scratcher to diagnose.
The moral of the story is that *electrical* EQ, be it notches or the crossover lowpass, is no replacement for good *mechanical* behavior, even in the stop band.
Rant: off
I definitely agree Bill. One of the things that Nick had always mentioned is that we want any kind of resonance to first of all be well damped. Secondly we want it to be at least 48dB down from the signal level. So if you plan on a 24dB filter you want a response that is smooth and has any resonances well damped for a full 2 octaves above the Xover point. This greatly helps avoid those issues.
John
John
More High-SPL Weirdness
I'm a terrible golfer, but I've always thought of it as quiet game, the group equivalent of fishing - meditative, but also demanding a high degree of skill. Well, all that's changed now - there are golf clubs that produce a stunning 130 dB peak (!!!) when the ball is struck, using my least favorite material, titanium. Check out this BBC medical report.
Darwinian market-driven de-evolution indeed. What's wrong with these people - the manufacturers, the reviewers, and the buyers. Do they want to be deaf?
Calming down a bit from my own little rant, Bill F is telling is something extremely important. Electrical equalization (passive or active) merely controls the drive power going to the driver. The driver, though, is a complex electro-magnetic-mechanical-acoustical device, and will do whatever it wants with the energy that is sent to it. It is perfectly free to distort (all drivers distort) and cross-modulate the incoming driving energy into its preferred resonant modes, which may be quite poorly damped. The amplifier has no say about this; an infinite damping factor, or even exotic tricks like negative impedances, has very little effect on a mechanical high-Q resonance (or cabinet modes).
Assuming drivers are distortionless is a very serious mistake - and what distortion does is move spectral components around (most often upwards). And real-world drivers have much more complex distortion spectra than merely 2nd and 3rd harmonic - seek, and ye shall find.
By looking at small-signal resonances via CSD, and then separately looking at swept 2nd and 3rd harmonic distortion, it is easy for the designer to forget these two domains strongly interact, and the resulting cross-modulation doesn't appear in the standard types of measurements. In the absence of zero-distortion drivers, electrical equalization needs to be reserved for excursion control (to keep distortion down), gentle in-band equalization, and smooth phase transfers to the other drivers.
This is the real reason hifi gear is demoed with spare-spectrum program material. The BBC found out more than 40 years ago (D.E.L. Shorter Wireless World articles) that with real-word music, sum-and-difference IM artifacts greatly outnumber simple THD components. In other words, the more complex the signal source, the greater the number of sum-and-difference terms, and that IM distortion complexity increases much faster than simple THD terms. With very simple material (little more than test tones), THD dominates, but with dense spectra, IM distortion dominates - and the hallmark of IM distortion is spectrally uncorrelated distortion (buzzes and hash sounds).
So if you have a high-distortion driver with an additional bonus of multiple poorly-damped resonances, avoid dense program material played at medium to high volumes, since that will trigger the inherent high-Q resonances of the driver (through crossmodulation). Thus the material we hear at hifi shows.
If you want to quickly remove bad drivers (and designs) from contention, listen to spectrally dense material (symphonic and choral) at medium to high volumes. If you hear grossly unmusical sounds - buzzes, hash, whistling sounds, metallic clanging sounds - those are driver artifacts. Weirdly enough, many audiophiles have trained themselves to ignore these sounds, since they are so common in the top-reviewed $30,000 to $100,000 speakers.
I'm a terrible golfer, but I've always thought of it as quiet game, the group equivalent of fishing - meditative, but also demanding a high degree of skill. Well, all that's changed now - there are golf clubs that produce a stunning 130 dB peak (!!!) when the ball is struck, using my least favorite material, titanium. Check out this BBC medical report.
Darwinian market-driven de-evolution indeed. What's wrong with these people - the manufacturers, the reviewers, and the buyers. Do they want to be deaf?
Calming down a bit from my own little rant, Bill F is telling is something extremely important. Electrical equalization (passive or active) merely controls the drive power going to the driver. The driver, though, is a complex electro-magnetic-mechanical-acoustical device, and will do whatever it wants with the energy that is sent to it. It is perfectly free to distort (all drivers distort) and cross-modulate the incoming driving energy into its preferred resonant modes, which may be quite poorly damped. The amplifier has no say about this; an infinite damping factor, or even exotic tricks like negative impedances, has very little effect on a mechanical high-Q resonance (or cabinet modes).
Assuming drivers are distortionless is a very serious mistake - and what distortion does is move spectral components around (most often upwards). And real-world drivers have much more complex distortion spectra than merely 2nd and 3rd harmonic - seek, and ye shall find.
By looking at small-signal resonances via CSD, and then separately looking at swept 2nd and 3rd harmonic distortion, it is easy for the designer to forget these two domains strongly interact, and the resulting cross-modulation doesn't appear in the standard types of measurements. In the absence of zero-distortion drivers, electrical equalization needs to be reserved for excursion control (to keep distortion down), gentle in-band equalization, and smooth phase transfers to the other drivers.
This is the real reason hifi gear is demoed with spare-spectrum program material. The BBC found out more than 40 years ago (D.E.L. Shorter Wireless World articles) that with real-word music, sum-and-difference IM artifacts greatly outnumber simple THD components. In other words, the more complex the signal source, the greater the number of sum-and-difference terms, and that IM distortion complexity increases much faster than simple THD terms. With very simple material (little more than test tones), THD dominates, but with dense spectra, IM distortion dominates - and the hallmark of IM distortion is spectrally uncorrelated distortion (buzzes and hash sounds).
So if you have a high-distortion driver with an additional bonus of multiple poorly-damped resonances, avoid dense program material played at medium to high volumes, since that will trigger the inherent high-Q resonances of the driver (through crossmodulation). Thus the material we hear at hifi shows.
If you want to quickly remove bad drivers (and designs) from contention, listen to spectrally dense material (symphonic and choral) at medium to high volumes. If you hear grossly unmusical sounds - buzzes, hash, whistling sounds, metallic clanging sounds - those are driver artifacts. Weirdly enough, many audiophiles have trained themselves to ignore these sounds, since they are so common in the top-reviewed $30,000 to $100,000 speakers.
If one is testing drivers/speakers for low IMD manually (by ear, that is), it seems wise to check oneself's intrinsic ear IM distortions in advance, in order to prevent false conclusions:
Perception of mid frequency and high frequency intermodulation distortion in loudspeakers, and its relationship to highdefinition audio, a David Griesinger paper. The important stuff is in the second part of this document. I tested myself along these guidelines and had to find out that the dominant IM at 1.6kHz or so (xover freq) that I hear with my (stub) horn loaded CD in my Tannoys, they don't distort that much in reality as I perceive it, using the suggested dense narrowband signals. Using high quality headphones confirmed that, too: It may well be more often the ear itself that distorts than we think.
- Klaus
Perception of mid frequency and high frequency intermodulation distortion in loudspeakers, and its relationship to highdefinition audio, a David Griesinger paper. The important stuff is in the second part of this document. I tested myself along these guidelines and had to find out that the dominant IM at 1.6kHz or so (xover freq) that I hear with my (stub) horn loaded CD in my Tannoys, they don't distort that much in reality as I perceive it, using the suggested dense narrowband signals. Using high quality headphones confirmed that, too: It may well be more often the ear itself that distorts than we think.
- Klaus
It's amazing that lots of designs and consumers don't take into consideration what kind of SPL a system would normally perform at. This makes a significant amount of difference. There will be no super system that will perform best at all listening levels since expected listeing environment and distance will drive the design trad-offs.
One thing why complicated music will reveal more defficiencies is that comlicated music stimulates continuous existance of these difficiencies which makes them more detectible. If we look at masking effects that noise related industry are more familiar with, the knowledge is there; it's just that people just think it's a different area of technology and often ignor research done in that area.
One thing why complicated music will reveal more defficiencies is that comlicated music stimulates continuous existance of these difficiencies which makes them more detectible. If we look at masking effects that noise related industry are more familiar with, the knowledge is there; it's just that people just think it's a different area of technology and often ignor research done in that area.
Lynn Olson said:
I also work as an inspector after big disasters. After these last 2 hurricanes I inspected just over 400 homes. In 1 home out of that I saw anything close to an audiophile system and he liked the older 50's stuff. Maybe in 2 homes I saw a turntable and they were older low fi stuff. In 3/4 of these homes I saw large flat panel TV however.
High quality audio is dead for the masses, if the speakers take up more space than a shoe box they are large now. The TV however needs to cover the wall. I know of nobody local that has a high end audio system that I haven't also know since the Lambda days. I don't remember the last time I went into a local home for the first time and saw even a medium end system.
I grew up when you could still buy records, everybody's home I went to growing up had at least a decent sized stereo/record player combo. Now they all have Mp3 from their computer and the TV playing all the time and talk about Blue Rays.
nickmckinney said:
hmmm! That's a rare perspective.
I hadn't realized things had got quite to that point, nor had I realized to what extent I was out of step with the masses.
I have never watched TV on a long term basis, never having had one in the house for more than 2 months at a time because, not being exposed to it, I am pretty much mesmerized by it and find I can't tear myself away. Four hours later I am kicking myself for wasting time watching nothing of consequence.
Now I waste time online - but at least I have some control - and what a stunning resource!
Indeed.
When seeing my speakers, most visitors just can't help spitting out "what the xxxx!!!" or similar words. Others who hold on being silent are just in the shock or despising. A friend said to his girl friend (in front of me and my speakers), "See, there're many kinds of maniac.". The OB speakers at that time were also critized as "walls"....
But what's wrong with big speakers? They SHOULD be big. That's just plain physics, isn't it?
When seeing my speakers, most visitors just can't help spitting out "what the xxxx!!!" or similar words. Others who hold on being silent are just in the shock or despising. A friend said to his girl friend (in front of me and my speakers), "See, there're many kinds of maniac.". The OB speakers at that time were also critized as "walls"....
But what's wrong with big speakers? They SHOULD be big. That's just plain physics, isn't it?
nickmckinney said:
<start off topic>
It's funny that this comes up over and over and over.... and that the people who favor big speakers (and make big messes building them) talk so kindly about their wives. Look at the way Lynn talks about his S.O., or Earl, or (nice to be in high-falutin' company) me, or Shinobiwan.
True love goes with understanding of the other one's crazies. We're all a little nuts, but the tolerant or even enthusiastic wives get a lot of credit and appreciation from their men.
Of course, if mine tried to control me the way some guys in hifi describe, she'd get a walking ticket, not for the speakers but the principle.
<end off topic>
It's valuable to have a number of different types of speakers. There's better and worse speakers, but the inherent advantages of a single driver for small scale music with modest levels are pretty hard to overcome with even the best multiway. They don't work for the bulk of my music, but they're what are playing in my room now (Zigmahornets with Merrills), replacing some modded JBL 2 ways (plus supertweeters). They don't do bass, but there is a lot going for a driver like this, lower treble dynamics are some of the best I've heard, so long as it's not accompanied by large demands in the mids/bass.
And knowing ones hearing, per another point here, is absolutely critical. My right ear distorts, level dependent, at about 1.8kHz. This makes this region extremely sensitive to loudspeaker distortion mechanisms, for whatever reason. But when I first noticed it, I spent many hours trying to pinpoint it within the stereo.