Interesting Paper from 18Sound
Inductance Minimization in Loudspeakers, an AES paper by 18Sound on minimization not just of inductance, but more significantly, delta-modulation effects that result in 2nd and 3rd-harmonic midrange distortion. Thorough discussion of the various modulation effects that occur in moving-coil loudspeakers, and the techniques used to address it. Since inductance modulation acts as a type of phase modulation of the signal, audibility may be greater than simple 2nd and 3rd-harmonic measurements indicate.
Lots of yummy graphs of developmental prototypes of the 10NDA520 and the 10NDA610, showing acoustically measured distortion curves and another set of distortion measurements of the voice-coil currents. An intriguing alternative to copper-plated pole-pieces - in fact, the A.I.C. system is compared to conventional non-plated pole pieces, and the copper-shielded versions.
The 610 has a saturated pole piece and top plate, and shows somewhat less improvement than the 520, which has a pole piece and a top plate that is not as close to saturation. I should note that very few drivers claim to have saturated pole pieces and top plates - a handful of professional drivers, the Lowther, the AER, the Feastrex, and maybe a few other exotics.
Inductance Minimization in Loudspeakers, an AES paper by 18Sound on minimization not just of inductance, but more significantly, delta-modulation effects that result in 2nd and 3rd-harmonic midrange distortion. Thorough discussion of the various modulation effects that occur in moving-coil loudspeakers, and the techniques used to address it. Since inductance modulation acts as a type of phase modulation of the signal, audibility may be greater than simple 2nd and 3rd-harmonic measurements indicate.
Lots of yummy graphs of developmental prototypes of the 10NDA520 and the 10NDA610, showing acoustically measured distortion curves and another set of distortion measurements of the voice-coil currents. An intriguing alternative to copper-plated pole-pieces - in fact, the A.I.C. system is compared to conventional non-plated pole pieces, and the copper-shielded versions.
The 610 has a saturated pole piece and top plate, and shows somewhat less improvement than the 520, which has a pole piece and a top plate that is not as close to saturation. I should note that very few drivers claim to have saturated pole pieces and top plates - a handful of professional drivers, the Lowther, the AER, the Feastrex, and maybe a few other exotics.
Charles,
I have heard Vivid speakers...and know the former importer. They are nice dynamic loudspeakers, but a little overpriced. I am aware of the B&W model you discussed, and the inherent differences between the snail and the subsequent N8XX models that followed. My point was about crossovers and assuming that because a driver has a diameter of x" it will ring at y hz. I am sure there is some truth to that regardless of material, as long as there is a dustcap, no phase plug, and a rubber surround which are all commonly found on most modern woofers.
Best,
Chris
I have heard Vivid speakers...and know the former importer. They are nice dynamic loudspeakers, but a little overpriced. I am aware of the B&W model you discussed, and the inherent differences between the snail and the subsequent N8XX models that followed. My point was about crossovers and assuming that because a driver has a diameter of x" it will ring at y hz. I am sure there is some truth to that regardless of material, as long as there is a dustcap, no phase plug, and a rubber surround which are all commonly found on most modern woofers.
Best,
Chris
gedlee said:
Charles Hansen said:
gedlee said:
Well, Charles already has answered about different forms of deformation in better words than I was able.
To some extent my arguments may suffer from semantics – as I'm no native speaker - so I always have to keep on the layman's side.
My point on "plastic deformation" in that discussion was to make clear that dampening is heavily linked to "plastic deformation" and hence the approach of linearity at brake-up is questionable.
This is a basic principle Earl, and – of course – DO " apply to dynamic loads ".
-------------
Charles, to be more precise with "You would potentially have to include an explicit time scale." means – that there isn't "hysteresis" for plastic (material wise) in the pure form we are familiar with from magnetising – at least I didn't come across that.
So Earl's "Hysteresis" – and yours too – does not seem to be of the same scientific definition value as "Hysteresis" is in electrical / magnetic context.
There is always – more or less - creaping (?) going on to release tension stored AND you wouldn't get back all – part if the invested energy is gone forever.
This creaping is something not found with materials having a molecule grid structure – like metals – beyond a certain stress limit.
(Fibre-) glass for example does not have that kind of "plastic deformation" at all – at least not at a time scale (or temperatures) we would have to worry about .
Charles Hansen said:
Charles, I prefer to "rather not" - if you don't mind.
I remember the first laser interferometry pix published from Celeston – but would have to dig them out to revise them in the light of your argumentation.
If I remember right, there were some kind of not-uniform membrane movement way down in frequency, but I wouldn't like to confuse that with the rather well defined term of "cone brake-up"
It's ok to me if you don't want to share – though this is what this place is good for – and also – not sharing your experience wasn't exactly what I found arrogant.
Greetings
Michael
"plastic deformation"
I studied material science once upon a time, so that the present topic is familiar to me. I guess the term you are looking for is viscoelasticity.
The following comes from Wikipedia (http://en.wikipedia.org/wiki/Viscoelasticity)
A viscoelastic material has the following properties:
- hysteresis is seen in the stress-strain curve.
- stress relaxation occurs: step constant strain causes decreasing stress
- creep occurs: step constant stress causes increasing strain
The behaviour of viscoelastic materials is time dependant
When we apply a small oscillatory strain and measure the resulting stress.
- Purely elastic materials have stress and strain in phase, so that the response of one caused by the other is immediate.
- In purely viscous materials, strain lags stress by a 90 degree phase lag.
- Viscoelastic materials exhibit behavior somewhere in the middle of these two types of material, exhibiting some lag in strain.
And maybe you should start to call "plastic materials" by their real name: polymer materials, this might help avoiding some confusions...
I hope this will help you.
Regards,
Etienne
I studied material science once upon a time, so that the present topic is familiar to me. I guess the term you are looking for is viscoelasticity.
The following comes from Wikipedia (http://en.wikipedia.org/wiki/Viscoelasticity)
A viscoelastic material has the following properties:
- hysteresis is seen in the stress-strain curve.
- stress relaxation occurs: step constant strain causes decreasing stress
- creep occurs: step constant stress causes increasing strain
The behaviour of viscoelastic materials is time dependant
When we apply a small oscillatory strain and measure the resulting stress.
- Purely elastic materials have stress and strain in phase, so that the response of one caused by the other is immediate.
- In purely viscous materials, strain lags stress by a 90 degree phase lag.
- Viscoelastic materials exhibit behavior somewhere in the middle of these two types of material, exhibiting some lag in strain.
And maybe you should start to call "plastic materials" by their real name: polymer materials, this might help avoiding some confusions...
I hope this will help you.
Regards,
Etienne
Re: "plastic deformation"
Thanks, I am completely comfortable with these deffinitions.
The area inside of the loop in the stress-strain curve would then be the damping. This is also consistant with the way we calculate the energy output from motors and engines. The area inside a loop traversed in one direction is the energy output and loops in the other direction are called pumping losses. Gas engines have very large pumping loses which is why diesels have better efficiency.
Etienne88 said:I studied material science once upon a time, so that the present topic is familiar to me. I guess the term you are looking for is viscoelasticity.
The following comes from Wikipedia (http://en.wikipedia.org/wiki/Viscoelasticity)
A viscoelastic material has the following properties:
- hysteresis is seen in the stress-strain curve.
- stress relaxation occurs: step constant strain causes decreasing stress
- creep occurs: step constant stress causes increasing strain
The behaviour of viscoelastic materials is time dependant
When we apply a small oscillatory strain and measure the resulting stress.
- Purely elastic materials have stress and strain in phase, so that the response of one caused by the other is immediate.
- In purely viscous materials, strain lags stress by a 90 degree phase lag.
- Viscoelastic materials exhibit behavior somewhere in the middle of these two types of material, exhibiting some lag in strain.
And maybe you should start to call "plastic materials" by their real name: polymer materials, this might help avoiding some confusions...
I hope this will help you.
Regards,
Etienne
Thanks, I am completely comfortable with these deffinitions.
The area inside of the loop in the stress-strain curve would then be the damping. This is also consistant with the way we calculate the energy output from motors and engines. The area inside a loop traversed in one direction is the energy output and loops in the other direction are called pumping losses. Gas engines have very large pumping loses which is why diesels have better efficiency.
Thanks Etienne - your explanations and the link are very helpful.
Membrane coatings will not fall into the category of "polymer material" but for sure into the category of viscoeslastic materials – hence I used the term "plastic (material wise)" throughout my argumentation.
Paper ?- what blend, color, coating etc...
But to be serious, the discussion about linearity within brake-up wasn't the starting point.
Lynn mentioned a "gray coloration" with highly equalised systems and I mentioned kind if "dead coloration" with highly dampened membrane materials.
There is some interest in this as you obvious want a ruler flat FR and very fast CSD decay but NOTHING of the sonic patterns described above.
Usually you (the speaker manufacturer) only can accomplish that to some extent by dampening.
Charles – in a more radical approach – advised to restrict the operating range of a speaker more rigorously – leading to something like +7 speakers necessary to cover audio range I guess.
Greetings
Michael
Membrane coatings will not fall into the category of "polymer material" but for sure into the category of viscoeslastic materials – hence I used the term "plastic (material wise)" throughout my argumentation.
miksin said:What's this all fuss about the materials?
As everyone should already know by now, paper just sounds best.
Paper ?- what blend, color, coating etc...
But to be serious, the discussion about linearity within brake-up wasn't the starting point.
Lynn mentioned a "gray coloration" with highly equalised systems and I mentioned kind if "dead coloration" with highly dampened membrane materials.
There is some interest in this as you obvious want a ruler flat FR and very fast CSD decay but NOTHING of the sonic patterns described above.
Usually you (the speaker manufacturer) only can accomplish that to some extent by dampening.
Charles – in a more radical approach – advised to restrict the operating range of a speaker more rigorously – leading to something like +7 speakers necessary to cover audio range I guess.
Greetings
Michael
Charles,
Can you point to commercial examples of full bandwidth 2 ways that acheive 20hz at nominal levels? I'm not trying to bust your chops, and am sincerely interested in what you come up with. I know you own JBL K2 9800's, (3-way) that is down ?? db @20hz, and the TAD Model 1 (4 way), which is also down ?? db at 20 hz. I appreciate the time and knowledge you lend to the forum freely.
Thank you,
Chris
Can you point to commercial examples of full bandwidth 2 ways that acheive 20hz at nominal levels? I'm not trying to bust your chops, and am sincerely interested in what you come up with. I know you own JBL K2 9800's, (3-way) that is down ?? db @20hz, and the TAD Model 1 (4 way), which is also down ?? db at 20 hz. I appreciate the time and knowledge you lend to the forum freely.
Thank you,
Chris
By the way, the TAD's we have are on loan. They are Reference Ones and not Model Ones. The Model Ones were a four-way design using a stacked plywood cabinet. They only made a handful of prototype pairs of these, most of which are in recording studios. The current production model is the Reference One and it is a three-way. The cabinet is presumably made from MDF, although I don't know for certain.
You can tell them apart by two methods:
a) The Model Ones had an extra upper-range woofer between the two lower-range woofers and the co-axial driver.
b) The co-axial driver on the Model One was mounted on a silver hemispherical "pod". On the Reference One it is mounted on a shaped extension of the main baffle.
You can tell them apart by two methods:
a) The Model Ones had an extra upper-range woofer between the two lower-range woofers and the co-axial driver.
b) The co-axial driver on the Model One was mounted on a silver hemispherical "pod". On the Reference One it is mounted on a shaped extension of the main baffle.
Re: Re: Interesting Paper from 18Sound
Sharp reader, Magnetar. Yes, I thought this is one of those papers where there is a lot below the surface - items to be picked up those "skilled in the art", as the patent attorneys like to say.
Although 2nd and 3rd harmonic distortion are the obvious first-order effects of inductance modulation, that may not be the most audible effect. In other parts of the audio chain, program-induced time modulation is extremely undesirable, and something good transformer designers are very aware of. There's a lot more to a transformer than the THD specs, the overload point, and the small-signal bandwidth.
This is an area where moving-coil loudspeakers overlap with the problems of transformers. The choice of core materials and the pattern of the field lines points to the parts of the core that are saturated first (transformer cores don't saturate all-at-once, it happens in small regions of high flux density and grows outward). The parts of the paper that simulate the flux density inside the pole-piece and top plate are very most interesting reading.
The biggest problem I have with the commonly used overhung voice-coil is the curved shape of the field lines that intercept the parts of the coil that are physically outside the gap. JBL and others have used heroic measures to linearize these out-of-gap field lines, but it is obvious from inspection that the lines in the gap are much straighter.
Transformer designers go to a lot of trouble to design the field-line structure in their transformers, since it has an important effect on saturation, HF performance, and problems with program-driven changes in group delay. This is the same problem AIC is addressing - in this case, with the counter-wound fixed coil structure in electrical parallel with the moving coil.
It's convenient to stick with simple models of cone behavior and RLC lumped models of the electrical system, but that's a long way from the full story of what the physical device is doing. Some of these more complex delta-inductance effects are not easy to measure, and the effect of FM distortion on sonics are not well known or documented in the literature.
The most common way of thinking of loudspeaker distortion is from simple excursion effects as the coil moves out of the gap, but that only accounts for distortion at the lowest frequencies. Distortion in the midband - where it is much more audible - is determined by delta-inductance and complex distortion mechanisms in the cone itself. The rise in HF distortion, for example, isn't in the acceleration domain (direct-radiators are constant-acceleration devices), but the third derivative of motion, "jerk".
The shape of the distortion curve vs frequency is significant, since it points to the underlying mechanism that creates it - motion out of the gap for the lowest frequencies, and delta-inductance and complex cone motions at the highest frequencies. As a rule of thumb, regions of rapidly rising distortion are a warning sign not to use the driver in that region.
I find it particularly appalling that modern theater practice uses heavy boost equalization in exactly the same region where compression-driver distortion is rising rapidly - from 7 to 20 kHz. This is a recipe for massive IM distortion that creates sum-and-difference artifacts that beat down into the more audible 2~5 kHz region. I hope that multitone IM distortion testing becomes more widespread - the communications industry has been using this measure for years. It always has to be kept in mind that IM-distortion artifacts can fall in regions where the driver is ordinarily supposed to be "rolled off". The crossover controls the energy going into the driver, but it has no effect on where the IM sum-and-difference products fall.
Although there is nothing in the literature to support my hunch, I have a feeling that drivers have some sort of slew-limiting going on at the top of their range, which we can indirectly measure as rapidly rising HF distortion. That, by the way, is the real reason I do not want to use a compression driver at the top of the audio range - the diaphragms just seem too big for that kind of speed and acceleration, no matter what kind of miracle material they are made of.
Direct-radiator dome tweeters avoid the problem two ways - by progressive decoupling and inherently high losses in treated-fabric soft domes, and with rigid-dome tweeters, trying to put the worst of the artifacts above the top of the audio band.
The brickwall filtering of Red Book CD's has favored the latter approach, since energy above 20 kHz is steeply rolled-off before it hits the driver, but IM distortion sum-and-difference products can still splatter up and down the audio spectrum. Once the energy arrives at the driver, any kind of distortion will result in spectral broadening. Performance that is usually considered "out of band" does in fact matter if IM distortion is a consideration.
If you're curious what IM distortion sounds like, any nearby Bose store will be happy to give a demonstration with their "Lifestyle" cube speakers. Turn up the treble to get the full effect.
Magnetar said:
Sharp reader, Magnetar. Yes, I thought this is one of those papers where there is a lot below the surface - items to be picked up those "skilled in the art", as the patent attorneys like to say.
Although 2nd and 3rd harmonic distortion are the obvious first-order effects of inductance modulation, that may not be the most audible effect. In other parts of the audio chain, program-induced time modulation is extremely undesirable, and something good transformer designers are very aware of. There's a lot more to a transformer than the THD specs, the overload point, and the small-signal bandwidth.
This is an area where moving-coil loudspeakers overlap with the problems of transformers. The choice of core materials and the pattern of the field lines points to the parts of the core that are saturated first (transformer cores don't saturate all-at-once, it happens in small regions of high flux density and grows outward). The parts of the paper that simulate the flux density inside the pole-piece and top plate are very most interesting reading.
The biggest problem I have with the commonly used overhung voice-coil is the curved shape of the field lines that intercept the parts of the coil that are physically outside the gap. JBL and others have used heroic measures to linearize these out-of-gap field lines, but it is obvious from inspection that the lines in the gap are much straighter.
Transformer designers go to a lot of trouble to design the field-line structure in their transformers, since it has an important effect on saturation, HF performance, and problems with program-driven changes in group delay. This is the same problem AIC is addressing - in this case, with the counter-wound fixed coil structure in electrical parallel with the moving coil.
It's convenient to stick with simple models of cone behavior and RLC lumped models of the electrical system, but that's a long way from the full story of what the physical device is doing. Some of these more complex delta-inductance effects are not easy to measure, and the effect of FM distortion on sonics are not well known or documented in the literature.
The most common way of thinking of loudspeaker distortion is from simple excursion effects as the coil moves out of the gap, but that only accounts for distortion at the lowest frequencies. Distortion in the midband - where it is much more audible - is determined by delta-inductance and complex distortion mechanisms in the cone itself. The rise in HF distortion, for example, isn't in the acceleration domain (direct-radiators are constant-acceleration devices), but the third derivative of motion, "jerk".
The shape of the distortion curve vs frequency is significant, since it points to the underlying mechanism that creates it - motion out of the gap for the lowest frequencies, and delta-inductance and complex cone motions at the highest frequencies. As a rule of thumb, regions of rapidly rising distortion are a warning sign not to use the driver in that region.
I find it particularly appalling that modern theater practice uses heavy boost equalization in exactly the same region where compression-driver distortion is rising rapidly - from 7 to 20 kHz. This is a recipe for massive IM distortion that creates sum-and-difference artifacts that beat down into the more audible 2~5 kHz region. I hope that multitone IM distortion testing becomes more widespread - the communications industry has been using this measure for years. It always has to be kept in mind that IM-distortion artifacts can fall in regions where the driver is ordinarily supposed to be "rolled off". The crossover controls the energy going into the driver, but it has no effect on where the IM sum-and-difference products fall.
Although there is nothing in the literature to support my hunch, I have a feeling that drivers have some sort of slew-limiting going on at the top of their range, which we can indirectly measure as rapidly rising HF distortion. That, by the way, is the real reason I do not want to use a compression driver at the top of the audio range - the diaphragms just seem too big for that kind of speed and acceleration, no matter what kind of miracle material they are made of.
Direct-radiator dome tweeters avoid the problem two ways - by progressive decoupling and inherently high losses in treated-fabric soft domes, and with rigid-dome tweeters, trying to put the worst of the artifacts above the top of the audio band.
The brickwall filtering of Red Book CD's has favored the latter approach, since energy above 20 kHz is steeply rolled-off before it hits the driver, but IM distortion sum-and-difference products can still splatter up and down the audio spectrum. Once the energy arrives at the driver, any kind of distortion will result in spectral broadening. Performance that is usually considered "out of band" does in fact matter if IM distortion is a consideration.
If you're curious what IM distortion sounds like, any nearby Bose store will be happy to give a demonstration with their "Lifestyle" cube speakers. Turn up the treble to get the full effect.
mige0 said:
You're mixing things up here...
Polymer is the type of material, ceramic and metals would fall in the same category.
Viscoelastic describes the mechanical properties of a material.
I don't know for sure what membrane coating are made of, but I know for sure that most of the polymer materials (if not fully crystalline) will show a degree of viscoelasticity.
The following comes from Wikipedia:
All materials exhibit some viscoelastic response. (...) Synthetic polymers, wood, and human tissue as well as metals at high temperature display significant viscoelastic effects.
If you want to go further, paper can be considered as a polymer since it is made of chains of cellulose. Like most of the polymer, paper do show a viscoelastic behaviour. In fact paper is more a composite material since it contains additives. But I think I will stop here for today in order to not confuse you further!
Regards,
Etienne
Re: Interesting Paper from 18Sound
Interesting technique but does not seem to work well enough for 18sound to display distortion graphs at their datasheets.
If I translate this crude text and plots right the measurement was taken at roughly 95dB SPL.
The third order harmonics – no higher orders shown sadly – seems to peak at 800 Hz with a 55dB attenuation to the fundamental translating to somewhere at around 0,18 %.
Not that impressive especially considering the natural roll off at higher frequencies of that driver - if I did interpret wrong would anybody please give the correct numbers?
-----
Over the fence – Usher offers comparable distortion numbers
http://www.zaphaudio.com/6.5test/8945P-HD.gif
-----
Another AES Paper – quite different principe and FR range though – shows figures roughly 0,05 % (3rd) downwards taken at comparable SPL levels – with NO roll off to be considered (until 7kHz / 3rd order) and really EXTREME LOW higher order distortion in addition.
taken from:
http://home.comcast.net/~neilandbarbaradavis/DIYHeil/ART.pdf
Greetings
Michael
Lynn Olson said:
Interesting technique but does not seem to work well enough for 18sound to display distortion graphs at their datasheets.
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.
If I translate this crude text and plots right the measurement was taken at roughly 95dB SPL.
The third order harmonics – no higher orders shown sadly – seems to peak at 800 Hz with a 55dB attenuation to the fundamental translating to somewhere at around 0,18 %.
Not that impressive especially considering the natural roll off at higher frequencies of that driver - if I did interpret wrong would anybody please give the correct numbers?
-----
Over the fence – Usher offers comparable distortion numbers
http://www.zaphaudio.com/6.5test/8945P-HD.gif
-----
Another AES Paper – quite different principe and FR range though – shows figures roughly 0,05 % (3rd) downwards taken at comparable SPL levels – with NO roll off to be considered (until 7kHz / 3rd order) and really EXTREME LOW higher order distortion in addition.
An externally hosted image should be here but it was not working when we last tested it.
taken from:
http://home.comcast.net/~neilandbarbaradavis/DIYHeil/ART.pdf
Greetings
Michael
The average SPL levels in the 18Sound measurement are at 115 dB, while the 2nd harmonic (gray) is around 82 dB, and the 3rd harmonic (blue) peaks around 70 dB (in the 500 Hz to 2 kHz region). That's about 2.2% 2nd harmonic, and 0.5% 3rd harmonic - at 115 dB. (A typical 88 dB/metre audiophile driver would need an input of 500 watts to achieve 115 dB - I don't think the voice coil would last long enough for a swept-sinewave distortion test.)
If we make a big assumption that the distortion decreases in proportion with level (well, it does some of the time), then at the approximate 100 dB test level of the Klaus Heinz transducer, then the distortion components of the 18Sound 10NDA520 prototype should be about 15 dB lower relative to the hypothetical 100 dB SPL test level. This would put them at 52 dB (0.4%) for the 2nd harmonic, and 40 dB (0.1%) for the largest peak of the 3rd harmonic.
If the 10NDA520 was tested close to its published limit of 123~126 dB SPL, then the expected 100 dB distortion figures might be a little bit lower - but that's strictly a guess.
Those seem like pretty reasonable distortion levels to me, considering the low audibility of 2nd harmonic. The Klaus Heinz HF driver has good distortion figures, though, no question about it. (Note: this is probably known by all who read diyAudio, but dB's are converted into voltage ratios for calculations of distortion percentages, while power ratios are used for calculations of required watts from the power amplifier.)
If we make a big assumption that the distortion decreases in proportion with level (well, it does some of the time), then at the approximate 100 dB test level of the Klaus Heinz transducer, then the distortion components of the 18Sound 10NDA520 prototype should be about 15 dB lower relative to the hypothetical 100 dB SPL test level. This would put them at 52 dB (0.4%) for the 2nd harmonic, and 40 dB (0.1%) for the largest peak of the 3rd harmonic.
If the 10NDA520 was tested close to its published limit of 123~126 dB SPL, then the expected 100 dB distortion figures might be a little bit lower - but that's strictly a guess.
Those seem like pretty reasonable distortion levels to me, considering the low audibility of 2nd harmonic. The Klaus Heinz HF driver has good distortion figures, though, no question about it. (Note: this is probably known by all who read diyAudio, but dB's are converted into voltage ratios for calculations of distortion percentages, while power ratios are used for calculations of required watts from the power amplifier.)
That refers to the other data set, made with the voice coil glued into the gap so it is motionless. The authors then measure the distortion of the current going into the driver terminals - in effect, seeing how linear the voice-coil inductance is, but without any motion from the VC to confuse the issue. In effect, they are seeing how linear the inductor is, just like measuring an iron-core inductor that might be used in a crossover.
Since these distortion effects are at a lower level, the distortion harmonics need to be raised 30 dB so they may be seen on a graph with an easy-to-read 10 dB/div scale. By contrast, the graphs with the distortion artifacts raised 20 dB are acoustic measurements, using a microphone in an anechoic chamber.
In a power amplifier, the assumptions I made about the distortion (as a percentage) falling as power is reduced works well most of the time. In a loudspeaker driver, though, we are on shakier ground, since the underlying distortion mechanisms are a lot more complex, and have different curves vs level. The assumptions I was making about what the 18Sound driver would do at 100 dB SPL is a crude approximation at best - but I would expect the real-world distortions to be lower, not higher, since the driver was operating close to the top of its dynamic range.
By comparing the acoustic vs electrical distortion graphs, you can get a good idea of what the cone, spider, and surround are adding to the distortion. Note how peaks appear at half the frequency of the acoustic peak in the 2nd harmonic, and one-third the frequency of the acoustic peak in the 3rd harmonic, along with a lot of other clutter. The clutter, in fact, is telling us a lot about "noise" in the mechanical system - spider breakup, cone decoupling, all kinds of things.
Interesting fireworks over on the "Horn vs Waveguide" thread - lots of good measurements, especially the rarely-seen CSD's and impulse data on various horns-n-waveguides.
Since these distortion effects are at a lower level, the distortion harmonics need to be raised 30 dB so they may be seen on a graph with an easy-to-read 10 dB/div scale. By contrast, the graphs with the distortion artifacts raised 20 dB are acoustic measurements, using a microphone in an anechoic chamber.
In a power amplifier, the assumptions I made about the distortion (as a percentage) falling as power is reduced works well most of the time. In a loudspeaker driver, though, we are on shakier ground, since the underlying distortion mechanisms are a lot more complex, and have different curves vs level. The assumptions I was making about what the 18Sound driver would do at 100 dB SPL is a crude approximation at best - but I would expect the real-world distortions to be lower, not higher, since the driver was operating close to the top of its dynamic range.
By comparing the acoustic vs electrical distortion graphs, you can get a good idea of what the cone, spider, and surround are adding to the distortion. Note how peaks appear at half the frequency of the acoustic peak in the 2nd harmonic, and one-third the frequency of the acoustic peak in the 3rd harmonic, along with a lot of other clutter. The clutter, in fact, is telling us a lot about "noise" in the mechanical system - spider breakup, cone decoupling, all kinds of things.
Interesting fireworks over on the "Horn vs Waveguide" thread - lots of good measurements, especially the rarely-seen CSD's and impulse data on various horns-n-waveguides.