Beyond the Ariel

Hi

I have to disagree with you here. The resonance in most drivers are a linear phenomenon and as such can be equalized.


JohnK, thanks for correcting me and giving some theoretical background.
From plain measurement its hard to tell whether the resonance is really corrected or just disappears in the dust. You are sure the break up is within linear operation as the whole area looks rather chaotic ?


The measurements below show an ancient 9" cone driver – pure paper, no exotic blend - no coating. - more relevant to this thread than metal cones I think.



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.


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.


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.



As can be seen, there is break up resonance not only with metal cones. Though for clarity I had to bring it up with a shelving filter as this driver rolls off at about 800 Hz thanks to the simple motor with no faraday rings. The resonance peak normally is around 10 – 15 dB down.

The tail of the resonance disappears with a notch at around –9 dB.


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.


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.


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.




What can be seen here is that the adjustment of the notch filter is VERY critical.
The tuning frequency in the Behringer DCX allows for center frequencies of 2,85 kHz / 2,79 kHz / 2,73 kHz which is not stepped fine enough and further more the Q is limited to 10 what also is not sufficient in this case.

You can not achieve a perfect linearisation with this tool instead the peak and the two dips at its side are only lowered. Still an audible improvement.

Any better tools out there?


Greetings
Michael
 
Hi

I agree from looking at this data, resonance in most drivers can be equalized; however, due to the fact that resonances occur after the first refleced wave in the cone, an equalized driver will never sound as good as the driver without this resonance. Since we may never be able to do this kind of comparison in reality, knows what we may be missing.


Soongsc, if I understood JohnK right this isn't the case and the correction is theoretically perfect.

It's not kind of feedback loop but rather a complimentary IN TIME balancing.

Greetings
Michael
 
Variac, beg your pardon.

Hi Jim, I was not so much referring to that speaker so much as I was the design.
IIRC, this was the OB speaker you heard which had some sort of malfunction in the bass. If it was corrected, perhaps a re-audition would be beneficial. Or perhaps an audition of an Orion to hear dipole radiation extended further down into the rooms acoustics and put things in perspective.

John, the NaO that I remember from the halcyon days of Madisound was the W18 version. When the inevitable Orion comparisons were made, it was not with the SS version. When I first viewed your Music and Design site, it was the W18 version for sale, with the SS option IIRC. Hindsight is 20/20, although I still can't see how you could consider this a "mistake". An option maybe.

john k... said:
Anyway, the Seas metal cones are very smooth in the pass band. There is no question that the W18 and the SS sound different and you’re not the only one who prefers the W18. But I bet you like a Steinway compared to a Baldwin. :)

That answer sits much better with me. It's more a subjective preference than measured distortion components. We each have our ideas of what things should sound like as referenced against our memory of live sound.

Tom Danley said:
Several answers;
While at Intersonics my job was to develop new kinds of transducers, “officially” these were for acoustic levitation but I also develop the Servodrive woofer and Phoenix Cyclone rotary woofer and some others while there. I guess from that perspective, I cannot under emphasize that damping is a very important quantity and is not associated with some of the cool looking cone materials.
If looks weren’t apparently more important than sound in some cases, then drivers with nasty behavior wouldn’t be sold based on a “New cone technology”.
It has been a while since I measured any, but I don’t see that they have changed fundamentally since then and if I were building something that used them, I would sample the ones I was interested in.
Your work shows ingenuity, from the rotary to the unity. However, this has little to do with soft vs hard cone behavior and driver design in general. Do you have any data on the drivers you are critical of from the Madisound catalog that you believe are all looks? John K presented his summary that it is a largely subjective choice, rather than a deficiency in the driver design.
Tom Danley said:

Your mention of a blaring pa system is spot on in a way.
Most of all the things that are wrong with speakers, get worse as the volume is increased and as the size and complexity of the speaker system increases and also as the room size increases. Often the sale of speakers in this kind of situation is based on hearing side by side demonstrations in the venue. If you can demonstrate better sound, you get the sale most of the time. That kind of puts the focus on better sound and is why I went the way I did.
My point was that even if a PA driver has 10X the headroom of some dainty Home Hifi unit, does that automatically make it the better choice in a living room? I agree that for my own audio requirements, a mixture of pro and home drivers are required for the headroom target, but my listening room and requirements may not be my neighbors. Who cares if a PA driver can achive 130+db with minimal compression if it sounds like s***?
Tom Danley said:
Well, all the horn style speakers I build, like the SH-50, only operate the drivers in the piston band. In fact, to reduce distortion, each cone driver feeds into the horn through an acoustic low pass filter comprised of the air volume trapped beneath the cone and a hole which connects to the horn. This low pass filter in front of the driver is above the low pass crossover point and has the effect of attenuating the harmonic distortion the driver produces which comes out above crossover frequency.

With a TEF, one can see a driver which is transitioning to non-piston motion, the old
TEF-10 /12 machines (which were much better in time than the modern ones) one could see the small change in time taking place in a mid range soft dome that had nice smooth response, as the dome de-coupled.
I would think that if you can, you would want to avoid that kind of operation (non-piston) at least where your ears are most sensitive, more or less 300Hz – 3KHz
Bottom line, if you have a mound of acoustic gain between the motor and the net radiation, you magnify the nonlinearity and sub multiples of that F.
On the other hand, if that acoustic gain is in the middle of the band and is broad (low Q) enough, then that can (if eq’d flat) can lower distortion by reducing the drive / excursion of the motor (who’s non-linearity is tied largely to motion).
This is what horns do (one thing), a bass horn that raises the drivers sensitivity by 10dB can reduces its excursion for a given SPL to 1/3 the direct radiator case.
You did not address my question. Are soft, flexing, lossy paper cones without sin? How you chose to use them in your synergy design is the same as how someone else, like a Toole, Thiel, Linkwitz, John K, etc. decides upon their design. I see no crossover frequencies for the SH-50. Nor do I see a laser scan of the mids, whatever they may be. As a matter of fact, based on the Yorkville design, I see what appears to be closed back mid drivers. Is this the same in the Synergy? If so, can you tell me how the rear wave off the driver cone is 100% absorbed in that minuscule chamber and not reflected and re-radiated through the paper thin membrane, delayed in time to the original?
Tom Danley said:
Your comment about compression drivers is related to using metal dome tweeters too.
A dome in a compression driver is one case where strength to weight is a controlling factor as the radiator as mass is more important than in a direct radiator tweeter.
Here, metal or mylar domes are used. A good one inch exit dome type compression driver will have a dome stiff enough (to act like a piston) to reach 17-20KHz before the first resonant mode. Realistically, if you’re an adult, your hearing rolls off before 20KHz and then drops like a rock above 20KHz. I’m not talking about can you detect anything up there under the best case but rather can you hear anything up there under normal conditions, no.
As a result, distortion produced at 5KHz and above has to be at an enormous level to be detectable at all. Similarly peaks and dips associated with breakup, that high, are not generally detectable.
So ringing like a bell tweeters are benign (and usable in your design, as opposed to say John K's) due to Fletcher Munson, but metal mids, with no corroborating data, are not?
Paper is this perfect, lovely material that folks wax poetically about, but no one uses in tweeters outside of Dr B (twiddlers, to be exact) and some hideously expensive fringe "fullrange" drivers with piezos?
Tom Danley said:
This is not what Earl Geddes is on about, that (HOM’s) are a variable that on one given horn, may make 5 different compression drivers with similar end responses all sound different.
Not sure how Earl got dragged into this, but he uses BC drivers, which I looked at and IIRC the woofer has that nice 'ol soft cone edge resonance. Is that benign also?
Tom Danley said:
“What if the goal is reproduction of acoustic instruments?”
An ideal reproducer reproduces whatever it is asked to, without alteration in any way.
Actually, the alteration of the signal begins at the speaker terminals through the drivers 3D radiation of acoustic power.
For me, rigid low loss cones offer the least alteration to my illusion of what a live acoustic instrument sounds like. To me, paper coloration is an association of amplified music.
Tom Danley said:

Like John K. said so well “Door bells and tuning forks should ring, not driver cones.”
No one can convince me that what we have no is so close that we can all give up.
Then there is the recording process and how the speaker interacts with the room, layers which make what the ideal speaker needs to do, less clear.
He also said this is based on subjective preference, rather than any objective data yet ascertained. Agree with the rest.
BTW, there was another recent thread about cone materials cones and data so as not to rehash it all. I have heard good and bad designs in both soft and rigid cones. I simply put a greater emphasis on transparency and not wanting to hear the loudspeakers whatsoever, than something more romantic, a filter that makes poor recordings more palatable.

cheers,

AJ
 
AJinFLA said:


I simply put a greater emphasis on transparency and not wanting to hear the loudspeakers whatsoever, than something more romantic, a filter that makes poor recordings more palatable.

cheers,

AJ

Ah! I think we come to the heart of the matter here.

By now it should be evident we hear - or at least, expect to hear - reproduced music in different ways. Speaking only for myself, I find all hifi systems not very close to the real thing - which for me, is live, un-amplified symphonic and choral music. When I go to a concert, it's at least a couple of days before I can stand to turn the hifi on again, it sounds so grossly artificial. Once I re-adapt, there is always at least some suspension of disbelief when listening to mechanical sound reproducers - and I use the word "mechanical" quite deliberately, since that's exactly what's wrong with hifi in general - a mechanical, "canned" quality that is always there.

This isn't to say all hifi sounds the same - hardly. Some, and a lot of the very expensive mainstream systems, are completely unacceptable and artificial sounding, with timbres and tone colors that don't sound like any acoustic instrument at all. I can tolerate these for no more than a few minutes, fighting an urgent desire to leave the exhibition room immediately, while right next to me, an audiophile (or magazine reviewer) is grinning from ear to ear.

What can I say? There's no accounting for taste. Since I find even the very finest systems removed from reality to a fairly obvious degree, I prefer systems that have colorations that land in the "musical" direction, as opposed to metallic, sizzly, grainy, flat-sounding, or other "electronic" or "mechanical" sounding colorations. In presence of obvious and hard-to-ignore colorations - which, again, I find in all systems compared to the real thing - the whole discussion of "transparency" doesn't seem as relevant.

Oddly enough, when it comes to amplifiers - where colorations aren't as gross and severe - then transparency as a desideratum becomes more relevant. In my limited experience, I find bigger transparency differences amongst amplifiers, which really can have a sort of remarkable MP3-like ability to erase fine detail and subtleties of musical expression.

In technical terms, though, not so surprising. Speaker drivers add distortion and resonances - lots of them - but don't have the ability to actually erase resolution. Electronics can perform this feat, thanks to Class AB-transition crossover distortion, nonmonotonicity in ADC and DAC conversion, digital jitter, and assorted low-level signal bending mechanisms that just don't exist in passive electroacoustic transducers.

Measurements confirm this - in most electronics with Class AB and/or quasi-complementary elements (this includes nearly all op-amps), distortion (not just noise!) starts to rise at low levels, while this is not seen in electracoustic transducers, where distortion falls monotonically with level (and of course there is no addition of noise, either).

P.S. One area where we could legitimately discuss loss-of-resolution in transducers is "stiction" in the spider - that's plenty real, but I haven't seen any measurements of the phenomenon. Spiders could certainly stand a re-design to get rid of stiction and other low-level nonlinearities, as well as unwanted re-emission and self-resonances. Maybe that's why some prosound drivers use stacked spiders with a layer of silicone damping between them, to help control some of the spider nonlinearities and break-in weirdnesses.
 
arend-jan said:
Hi Lynn,

have you considered an electrostatic driver for the mid/highs? It may be tough to match with dynamic drivers in an OB and no, you won't get high efficiency. But to my ears an electrostatic driver, like for example the treble panel from a Quad ESL, has the most natural sound I've heard yet.

Regards,
Arend-Jan

Oh so tempting, I agree about the wonderful sound of electrostats. But ... since the mids and bass are being designed for a 110~115 dB headroom with many dB to spare, I think the electrostat would get crushed, or more accurately, zapped.

The efficiency mismatch is also great - more than 10 dB - the system would require yet another amplifier, a dismaying prospect. The bass array amplifier and associated equalizer don't need to be the last word in subtlety and beauty, since they only cover 80~400 Hz, but the quality requirements for an electrostatic-tweeter amplifier are not trivial.

The best ribbons have nothing to apologize for sonically, and it looks like the double-high RAAL at 100dB/metre is in a class of its own in terms of impulse response, very wide horizontal dispersion, and peak SPL capability.
 
Some Subjective Measurements

Made with the best-quality measuring tape, but unfortunately not traceable to NIST standards.

Sitting on the floor meditation-style (but without the fancy leg moves, which I can't do), my ears are 32.5~33" high, depending on position.

Sitting on different chairs and couches, my ears are 40~49" high, depending on how much I lean back and the recline angle and height of the couch or chair.

Generic 1" dome + 7" midbass speakers (Magico Mini to Cambridge SoundWorks Model Six) typically aim the crossover lobe downward, unless the designer took special steps with the phasing to aim the beam upward (not very common). Since these things usually sit on short stands, I can see why they usually sound better when I sit on the floor, instead of a chair, looking down on them. Pointing them upwards is probably a good idea.

P.S. I'll take the CSW Model Six at $299/pair instead of the Magico Mini at $20,000/pair, thank you very much. I'm a big fan of Henry Kloss and his work.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
My reference axis on my current system (the one that you saw as OT earlier in this thread when some more info was asked), is 47'' high.
I am with you in that high axis thing obviously.

Earlier in the posts I have shown the Bastanis naked back compression driver 'Gemini' dipole tweeter approach. What about something like that for HF? Not strictly for the ''Beyond'' speaker that needs a fairly low cross. Your opinion in general. Bastanis's wide range crosses high and there are many (almost FR) 8 inch options if making a less dynamic OB. What do you think?
 
soongsc said:
I agree from looking at this data, resonance in most drivers can be equalized; however, due to the fact that resonances occur after the first refleced wave in the cone, an equalized driver will never sound as good as the driver without this resonance. Since we may never be able to do this kind of comparison in reality, knows what we may be missing.

Well it all comes down to whether the resonance is minimum phase or not. And I will suggest that is based on my measurements. Given that, minimum phase equalization of a minimum phase resonance corrects the response. The equalized driver and the ideal electrical bandpass filter have the same CSD. Since the CSD is just another way of post processing the impulse response then the two must also have the same impulse response. If its a linear system, then it can be corrected linearly.

However there are caveats. Although the breakup is primarily a linear phenomenon there are certainly nonlinear components, which, while not dominant, can not be corrected by application of such equalization. There are numerous other potential sources of problems too. As you say though, it's academic and there really isn't any way to definitive test it. All you can do is listen.
 
AJinFLA

Your focus on “soft” vs hard was not addressed.

I mentioned damping, mechanical loss within the radiator is very important.
One cannot avoid chaotic behavior past the upper limit of the piston range unless one has lots of loss. A good soft dome is an example of this, the driver is mostly in non-piston mode but has no pronounced resonances.
The best driver (in time) I ever measured was a Manger bending wave transducer, it is a resistively dominated device, very high mechanical losses, low stiffness.
This sounds like the way to go except that mechanical resistances tend to change with time, change with temperature and change with use and are progressively non-linear.
The down side of the Manger was that at 1 volt is that it is already pretty high in distortion (a few %) at 200Hz. They (with a subwoofer) would make a superb desktop stereo though.

For a radiator, it is difficult to keep the mass low enough and have a high enough break up AND have enough damping to kill the Q’s.
Not having a perfect radiator, one can say a low Q resonance is better than a high Q one, a 3 dB peak is better than a 10dB peak and so on. All this you can see when/if a response curve is provided.
Also, while one can filter the drivers breakup with DSP, one cannot fix the fact that the acoustic gain magnifies the natural distortion at sub-multiple frequencies.
That is not to say such a driver can’t sound great, what you hear is so much more than that one aspect of its performance.

“I see no crossover frequencies for the SH-50. Nor do I see a laser scan of the mids, whatever they may be. As a matter of fact, based on the Yorkville design, I see what appears to be closed back mid drivers. Is this the same in the Synergy? If so, can you tell me how the rear wave off the driver cone is 100% absorbed in that minuscule chamber and not reflected and re-radiated through the paper thin membrane, delayed in time to the original?”

The crossovers are about 350Hz and 1400Hz in the SH-50.
The mids are made to spec, they are closed back type.
Again the answer is in thinking of the acoustic dimensions, the rear radiation is not an issue, here is why. At the highest frequency the mid drivers go to (lets say 1500Hz), the wavelength is about 9 inches. The back cover is about 1 / 2 inch away from the cone which is far less than the 2 ¼ inch distance which could cause the first reflection notch (at 1500Hz). The rear cover is so close, it can’t cause a reflection in its operating band.

“To me, paper coloration is an association of amplified music.”
Yes but if you make speakers that are also to produce natural sounds (not generally harmonic in nature) are to be faithful to the recording, you don’t want any coloration.

On the other hand, one form of coloration is one which I don’t know the name of, but can describe.
In one case, with music or voice you can with your eyes closed you can easily tell how far away the speaker is, in another, you have no idea how far away it is.
In one case, a pair of one kind and a pair of the other gives radically different stereo images even if the responses are similar.
The one with no clues as to distance makes a much better stereo image.
I think that the source interference that lets you localize the actual depth of the speaker harms or competes with the stereo clues.
Coming a full circle, one finds that a single full range driver (which usually has a low mass thin paper cone) is the simplest emulation of that non-interfering source, the goofy horns I make at work at the large end of the scale.
Hope that helps,

Tom Danley
 
Magico Mini

Did a little digging and found the driver used in the much-hyped Magico Mini - here. Want the awful truth? Here's the frequency response. Want to make any guesses whether that kind of curve is correctable with any kind of equalization?

Note carefully the published reference efficiency - 86.5 dB/metre. Now let us descend into the deep, dark underworld of high-end reviewing and read how impressed the reviewer was with the "dynamic range" in a very large listening room:

"The "stand-in" demo in Peter's neo-gothic wing cast such a capacious soundstage filling a completely counter-intuitive cubic volume -- and at very happy levels -- that if Alon had any reservations about me getting the concept of the mondo monitor, they were gone with the proverbial wind. Frankly my dear, I do give a damn! While expensive as sin without apologies or guilty confessions, my first impressions in Oakland have me convinced that even under serious fire -- i.e. John Corigliano's Chaconne for solo violin and orchestra based on his score for The Red Violin -- the Magico Minis don't falter to telegraph the usual limitations of two-way speakers. They act far more like good conventional three-ways except for the lowest bass."

Uh-huh, yup, I think I'm starting to understand how this works.
 
AJinFLA said:
Variac, beg your pardon.

[snip]

John, the NaO that I remember from the halcyon days of Madisound was the W18 version. When the inevitable Orion comparisons were made, it was not with the SS version. When I first viewed your Music and Design site, it was the W18 version for sale, with the SS option IIRC. Hindsight is 20/20, although I still can't see how you could consider this a "mistake". An option maybe.

[snip]





cheers,

AJ


Oh the good old days. :) I remember taking heat about the rear tweeter and a somewhat heated debate about the u-frame woofer. How the landscape has changed. Anyway, here is a link to the original NaO pages (preserved for historical reference).

http://www.geocities.com/kreskovs/NaO.html


The very first version of the NaO was back in October 2003 and used the SS mids, T25 front tweeter Vifa D27 rear, a PE 12" IB woofer and was intended to be use with an off-the-shelf active crossover (Like the Marchand XM9) between the woofer and panel with only a very simple active Eq circuit patched between preamp and active crossover. The AEP version came out about the time I offered the first generation custom PCBs with built in active crossover in the early 2004. John P. built his system and took it to the Iowa 2004 meeting. He also used a TC Sound woofer for which I design mods to the eq circuit. As far as I know, his is the only AEP version ever build other than my AEP prototype. He still seems happy with it.
 
mige0 said:
Hi




Soongsc, if I understood JohnK right this isn't the case and the correction is theoretically perfect.

It's not kind of feedback loop but rather a complimentary IN TIME balancing.

Greetings
Michael
What happens first will arrive at the ear first. Since electrical compensation happens first, you hear the wrongly compensated signal first. Unless you can make the compensation after the cone mode is excited and before the wave reaches the ear, you cannot ideally correct for it. I think if someone does a finite element analysis of the acoustic wave then you can clearly see it. I recall that Manger has a patent that does some kind of compensation acoustically, now this is possible, but the compensation will be perfect at specific measurement locations only.
 
Lynn Olson said:


Ah! I think we come to the heart of the matter here.

By now it should be evident we hear - or at least, expect to hear - reproduced music in different ways. Speaking only for myself, I find all hifi systems not very close to the real thing - which for me, is live, un-amplified symphonic and choral music. When I go to a concert, it's at least a couple of days before I can stand to turn the hifi on again, it sounds so grossly artificial. Once I re-adapt, there is always at least some suspension of disbelief when listening to mechanical sound reproducers - and I use the word "mechanical" quite deliberately, since that's exactly what's wrong with hifi in general - a mechanical, "canned" quality that is always there.


Hi Lynn,

Well, I have to agree with you here about symphonic concerts and reproduction realism. However, I also attend smaller 4-8 pc acoustic jazz jams and I find that I can occasionally get fooled by the illusion a bit more if the recording is just right.
I find that up close to real musical instruments, the sound has a much more raw, sharp nature far from many smoothed over creamy recordings and yes, paper and other non-rigid, soft materials are capable of. That last layer of transparency is obscured by non-rigid materials IMHO. Perhaps it is all in my mind when I both see and hear a high hat being tapped and a sax's visceral feel, who knows.

Hi Tom,

Thank you for the info you provided. Perhaps we will have to agree to disagree here. I believe both rigid and non-rigid diaphragms have their issues and I pick what I perceive as the lesser poison. What I see in the data (that I have seen) is that rigid cones tend to have less distortion and smoother, more controlled behavior within the pass band that I would use them in.
Coupled with the fact that rigid cones are non-existent in HE pro audio type drivers, I'm not really sure how you could make comparisons to the drivers you are using or obtain data.
I may definitely have to try your record the output and loopback a couple times method with my next speaker design which creeps along ever so slowly. Did you ever take any impulses of the SH-100? BTW, I hope you are not using any of the standard B&C 8" coaxes, as I noted some major breakup in the factory curves.

Hi John,

I hope I wasn't one of those who gave you grief about the rear tweeter, as I would have used one myself had I not used the waveguide for my tweeter to closer match the forward directivity of the mid. I had actually used a rear tweeter with separate, adjustable eq to shape the power response on a bipole circa late 90's and quite liked the sound, given enough distance to rear wall.

Mark K,

What's taking so long with those things;)

cheers,

AJ
 
I Have Been Warned

Karna is 62" high, and doesn't want the speakers to be bigger than she is. Can't say I blame her, a speaker that more than 74" high (my height) would intimidate me too. Her preference is she can at least see the tops of the speakers - again, I sympathize.

With the top of the front panel at 46" with a 12+15+15 layout (and 1" spacing), and 49" with a 12+15+18 layout, it'll be just barely within the limit with the 12" high double-panel RAAL tweeter. Now there is a bit of question about the RAAL - the front plate can be removed and the ribbons used singly, depending on how efficiency works out.

The answer to this isn't certain, since the narrow vertical dispersion means the energy from the ribbon tweeter falls off more slowly than the inverse-square law - in other words, at 3 to 5 meters, it'll be putting 2~3 dB more energy than the standard dB-at-1-meter specification implies. Alexander said in the last phone call "expect about 2.5 dB more at the listening position" than the standard spec implies. Also, the voltage sensitivity of the RAAL can fine-tuned by varying the ratio of the step-down transformer - the standard transformers have a 10:1 stepdown ratio and present a near-resistive 8 ohm impedance at the primary.

So I can't say whether I'll use 1, 2, or 3 RAAL ribbon sections (3 would be in the case of 2 stacked vertically in the front and 1 facing backwards). It seems like there's at least a reasonable chance for 1 in front and 1 in back, using the transformer to fine-tune the voltage sensitivity to match the sensitivity of the widerange driver at the 3~5 meter distance.
 
AJinFLA said:


Hi Lynn,

Well, I have to agree with you here about symphonic concerts and reproduction realism. However, I also attend smaller 4-8 pc acoustic jazz jams and I find that I can occasionally get fooled by the illusion a bit more if the recording is just right.
I find that up close to real musical instruments, the sound has a much more raw, sharp nature far from many smoothed over creamy recordings and yes, paper and other non-rigid, soft materials are capable of. That last layer of transparency is obscured by non-rigid materials IMHO. Perhaps it is all in my mind when I both see and hear a high hat being tapped and a sax's visceral feel, who knows.

cheers,

AJ

Hi there, AJinFLA.

I respect what you hear - our perceptions are our own, the deepest and most intimate part of personal reality. What you hear exists, and is real.

The whole symphony vs jazz discussion can get lost in the weeds of personal musical preferences, and is a diversion from a more important point that bears on what we've all been talking about. A jazz group has a much sparser spectra than a full-scale symphony orchestra, and the reverb time is much shorter as well.

This means the symphonic spectra superficially looks as dense as noise, but in reality is highly correlated with itself and the hall reflections. Any perturbation to the fine spectral and time structure does enormous violence to the performance, since so much is going on all at once - indeed, the sheer density, complexity, and fleeting spatial relationships are an integral part of the composer's and conductor's intentions.

Music with many fewer instruments has a much simpler and sparser spectrum, and is more about the vivid and pulsing tone colors of the individual instruments - the whole point of the trademark jazz solo. It's all about expression and tone colors.

The requirements of the two types of music for the hifi system are quite different. Even tiny amounts of IM distortion have a ruinous effect on a dense spectrum, creating all kinds of inharmonic sidebands that clash with the tone-colors of the ensemble. The effects of IM distortion (and driver resonances) are especially destructive to the subtle shimmering interplays of harmonic decay structures among different instruments.

This, I submit, is why symphonic and jazz fans tend to prefer different kinds of hifi systems. The spectral and dynamic requirements are dissimilar, with low dynamic IM distortion being especially important to classical music. Solid-state electronics typically have very low steady-state distortion, but can degrade by many orders of magnitude under fast transient conditions (memory effects in transistors causing on-chip thermal lags and transient mis-biasing). Similarly, the whole concept of dithering relies on a large number of samples to obtain the required distortion improvement in the LSB region - with fewer samples (shorter intervals), the benefits of dithering are greatly reduced.
 
Very interesting thread

john k... said:


Well it all comes down to whether the resonance is minimum phase or not. And I will suggest that is based on my measurements. Given that, minimum phase equalization of a minimum phase resonance corrects the response. The equalized driver and the ideal electrical bandpass filter have the same CSD. Since the CSD is just another way of post processing the impulse response then the two must also have the same impulse response. If its a linear system, then it can be corrected linearly.

However there are caveats. Although the breakup is primarily a linear phenomenon there are certainly nonlinear components, which, while not dominant, can not be corrected by application of such equalization. There are numerous other potential sources of problems too. As you say though, it's academic and there really isn't any way to definitive test it. All you can do is listen.

Makes me a bit nostalgic for this kind of thread that used to occur with some frequency at Madisound.

I'll throw in my $0.02 on driver minimum-phase. As you know John, I'm in full agreement, having modeled drivers for years to occasionally excruciating detail up to 24KHz, even 10" drivers. Every single one could be modeled perfectly with minimum-phase elements that resulting in essentially perfectly matching modeled/measured phase responses.

I'm also in your camp insofar as metal vs. doped paper, distinguished from the less accurate and unmodified term "paper", implying soft. I have yet to have fully successful result making a hard-coned driver the equivalent of a well done doped paper driver, especially in the midrange. I also suspect the amplification of the non-linear motor distortion by the peakiness exhibited in the raw driver.

That said, I am, however, slowly getting one hard-coned unit to finally sound good that I previously gave up on, though it's still not in the league of the SS midrange units. Also, the Accuton that you've heard at DIY, though a hard cone, doesn't have the excessive peak of a Seas and seems to be better as a result. But this is the larger 7" C95, the smaller ones still look to peaky to me, especially at the price.

AJ,

You keep referring to "soft, flexing, lossy paper cones", but this is a bit disingenuous. The truly good driver examples such as the SS units are not soft, they are relatively stiff doped-paper with significant internal damping and don't flex nearly as much as implied. There's also a very simple way to compare any two drivers in this regard. Make a range of distortion tests, such as those that Mark K and zaph make. All implications of "soft", "flexing", "lossy", etc., are really irrelevant without some sort of objective evaluation for comparison.

Better yet, make as set of these with crossover in place, since the driver's distortion profile will be altered when in use in a system. The non-linear aspect such as motor distortion amplification should be largely unmodified in only a lowpass is used whereas with a true midrange that has a bandpass applied should tend to show lower non-linear distortion products for the same acoustic output of a raw distortion measurement due to lessened driver displacement. In this case, a 3-way or more with a true midrange unit with a bandpass, the highpass could very easily, I suspect, result in a midrange output that exhibits lower non-linear distortion for a doped-paper than for a hard coned driver. At higher volumes, the doped paper could quite easily be perceived as being less dissonant since in general there will be less peak-driver amplification of distortion products.

With regard to the bandpass low-Q linear FR issues, these can, as John pointed out, be equalized, often with little difficulty. Those small perturbations that remain may be inaudible and for those that are not, they will still tend to be even order and euphonic, not dissonant. But for good doped-paper (and let's keep the conversation to good units of each type), these are usually not the big issue.

Dave