Hi,
It was not apparent to me and more to the point, if there are options, does this justify multiple entries?
My time is limited as well, but I still found it opportune to use two sources.
Well, at that I would not call any high performance audio product "mainstream". However once we accept that High End Audio is a market in it's own right Tube Amplifiers are part of the high end mainstream.
Well a certain kind of tube amplifier anyway, best embodied by the long serving Manufacturers Audio Research, Conrad Johnson, EAR/Yoshino, Quad, VTL and a few European brands with little recognition outside Europe.
Really, well it seems at least one reader wishes to point out the assumption is invalid.
Well, again you show you have not bothered to read.
I note that the MATERIAL effects that deliniate current drive from voltage drive is the cancellation of eddy current distortion, which is as I remarked a cubic function, forming a long train of odd harmonics that are mathematically related to the third harmonic, which is the one that is commonly easily measured in speakers and hence can be used as "diagnostic" indicator for eddy current distortion, and the cancellation of thermal compression.
Other effect exists but they tend to be not even of secondary importance (even the Eddy Current Distortion and Thermal compression are really only of secondary importance in speakers compared to sound perception), but of tertiary and lower...
Okay, lets have a look at your list of "current distortion effects" and estimate the magnitude...
Let us take them each in turn (which I neglected to do earlier on...
Okay, you list these first. Let us be clear. If the cone (or voicecoil) is actually allowed to be modulated by external signals (such as those from inside the cabinet or from other drivers) the result will be that the signal is distorted.
In fact, I would note that, if the effect that you mention was sufficiently material to cause problems, it would form the greatest argument AGAINST current drive, as current drive fails to oppose these effects, which are mechanical in nature.
The mechanical and pneumatic non-linearities remain unaffected by current drive. In fact, if drivers are carefully designed to have low distortion with voltage drive (which many current day drivers are) switching to current drive will substantially increase the overall non-linearity, as experienced in one of the reviews of your book where current driver substantially raised the drivers distortion, not lowered it.
Modern drivers tend to carefully control these problems, in fact even many vintage drivers are surprisingly good in this context.
BL variations are caused by varying field stength for different postions of the voice coil. The effect on the impedance in the midrange is minimal, the actual reason the impedance is modulated is varying inductance with the position of the voicecoil respective to the polepiece.
The effect remains minimal in drivers that do not have great excursion and remains still small in the midrange as the inductive reactance of the voicecoil is low at midrange frequency.
It is a problem in full-range drivers however, which is why many (even vintage ones) used some form of inductance control (copper cap on pole piece, short circuit rings etc...) thus minimising these effects at the outset.
By comparison the direct distortion produced by this varying field strength is very large (and 2nd order HD and related IMD) and remains unaffected by current drive.
In fact, if careful design is used to offset mechanical non-linearities in the suspension with those of the magnetic field in Drivers (which modern FEM analysis allows and which modern speaker designs often apply) using current drive will materially increase this primary distortion mechanism.
Now we are at one of the biggies, where current drive does indeed produce large changes.
Pulleeese. I have had my in audio transformer design. I have a better then fair idea at what field levels this becomes a problem. Let's say it is not an issue in MC transformers.
Plus, current drive changes neither the inherent magnetic system of the speaker, nor does it change the magnet field from the voice coil so any negative interactions remain unaffected, as far as they impact the output from the Driver.
From series production of speakers I can assure you the manufacturing tolerances on DCR are quite tight, with reputable makers.
The thermal compression is the other biggie and to someone coming from a pro-audio background a clearly observable problem that is well addressed by current drive.
Funny you mention this. I can measure the effects of connectors on cables on my AP2. It's usually well over 120dB down on the signal, but can be measured. The change to that from current drive can be real, but we are already so far down that the negtive impact is unlikely to have dramatic effects.
So, some of your additional effects where current drive changes performance are real, however their actual level is sufficiently small I am willing to dismiss them, in other areas it would seem the problems you describe are worsned by current drive, not improved.
Well, forgive me, I have also not done a doctoral thesis on it, but I have compared systems using drivers producting a non-flat (overdamped bass response) driven by SE Amplifiers with highish output impedance and PP Amplifiers with low output impedance (both of my own design and not conventional).
Once digitally equalised to the same frequency response most of the subjective differences disappeared (but not when using solid state amplifiers), the remaining ones I considered at the time mainly the result of the difference in harmonic distortion between the Amplifers.
My experiements with current driven Speakers in the 80's (incidentally using TDA2030 in a fully active setup, dual 20cm woofer with bridged TDA2030 plus current boost transsitors for bass to 250Hz, 12cm wideband driver as midrange 250Hz to 5KHz, 2cm supronyl plastic dome as tweeter above 5KHz) did not produce a speaker that was subjectively preferable to either of my other "references" I used at the time.
The references where an O18 Studio Monitor Pair (12" Coaxial Schulze TH315 in large vented boxes with matching solid state Amp and build in EQ for the speakers) or my 3-Way EV Horn System (15" EVM Woofer vented, 8HD Midrange Horn and T35 Tweeter) driven by 50W Studio Tube Amplifiers with the build in EQ defeated, output impedance of that tube amp was very low, it used large amounts of NFB.
Of course, this was not a controlled experiment and only my personal preference. Using our limited measurement gear at the time it was hard to penetrate why the active, current drive speaker did not outperform the others. It is worth noting that both my references had midrange/treble drivers using Alnico magnets AND respectively 8dB and 15dB greater efficiency.
I will also add that going to current drive in the Mid and Treble and to microphone based MFB for the bass significantly improved the active speaker compared to the original simple active voltage drive, just not enough to make it overtake "real He-Man Speakers (TM)" that where many time it's size and used pro-drivers, not hifi drivers. These experiences resulted in me discontinuing experiments with current drive, after I escaped from east germany in early 1989, it just did not seem worth the extra effort.
Ciao T
The reason why the number of samples is greater than of the amps becomes apparent from my text (Some amplifiers had options.);
It was not apparent to me and more to the point, if there are options, does this justify multiple entries?
I have had to omit many other sources too, as my time for this is limited. The choice of Soundstage was random.
My time is limited as well, but I still found it opportune to use two sources.
I wouldn't call any tube amplifiers 'mainstream' - curiosities they are all.
Well, at that I would not call any high performance audio product "mainstream". However once we accept that High End Audio is a market in it's own right Tube Amplifiers are part of the high end mainstream.
Well a certain kind of tube amplifier anyway, best embodied by the long serving Manufacturers Audio Research, Conrad Johnson, EAR/Yoshino, Quad, VTL and a few European brands with little recognition outside Europe.
I haven't 'claimed'. I made a realistic assumption. It is left to the reader to assess whether the assumption is valid.
Really, well it seems at least one reader wishes to point out the assumption is invalid.
I think I have now found an explanation why you are talking what you are talking. You somehow consider the 'current distortion' to mean only or mostly the 3rd HD due to eddy currents, whereas I have considered all of the nonharmonic, harmonic and other, unclassified, effects of distortion and interference. This may explain much of your objection of the title and claims of ignorance.
Well, again you show you have not bothered to read.
I note that the MATERIAL effects that deliniate current drive from voltage drive is the cancellation of eddy current distortion, which is as I remarked a cubic function, forming a long train of odd harmonics that are mathematically related to the third harmonic, which is the one that is commonly easily measured in speakers and hence can be used as "diagnostic" indicator for eddy current distortion, and the cancellation of thermal compression.
Other effect exists but they tend to be not even of secondary importance (even the Eddy Current Distortion and Thermal compression are really only of secondary importance in speakers compared to sound perception), but of tertiary and lower...
I have always known that tube amps produce 3rd HD along with other harmonics if that is the great knowledge you thought I was lacking. I am also fully aware that harmonics originating from different stages in the signal path may strengthen or weaken each other. However, the whole issue of 3rd HD is rather secondary in the whole scope of the current distortion effects, that are largely nonharmonic or phase muddling or even indefinite by nature.
Okay, lets have a look at your list of "current distortion effects" and estimate the magnitude...
Let us take them each in turn (which I neglected to do earlier on...
Voice coil acting as a microphone for the sound waves reflecting from the cabinet interior and passing through the diaphragm
- Voice coil acting as a microphone for the sound waves from adjacent drivers
Okay, you list these first. Let us be clear. If the cone (or voicecoil) is actually allowed to be modulated by external signals (such as those from inside the cabinet or from other drivers) the result will be that the signal is distorted.
In fact, I would note that, if the effect that you mention was sufficiently material to cause problems, it would form the greatest argument AGAINST current drive, as current drive fails to oppose these effects, which are mechanical in nature.
- Mechanical and pneumatic non-idealities of the moving parts causing unpredictable EMF-effects
The mechanical and pneumatic non-linearities remain unaffected by current drive. In fact, if drivers are carefully designed to have low distortion with voltage drive (which many current day drivers are) switching to current drive will substantially increase the overall non-linearity, as experienced in one of the reviews of your book where current driver substantially raised the drivers distortion, not lowered it.
Modern drivers tend to carefully control these problems, in fact even many vintage drivers are surprisingly good in this context.
- Bl-variation causing modulation of impedance's angle and hence phase modulation of current at middle frequencies
- Position-dependent inductance of voice coil causing both amplitude and phase modulation (as in post #66)
BL variations are caused by varying field stength for different postions of the voice coil. The effect on the impedance in the midrange is minimal, the actual reason the impedance is modulated is varying inductance with the position of the voicecoil respective to the polepiece.
The effect remains minimal in drivers that do not have great excursion and remains still small in the midrange as the inductive reactance of the voicecoil is low at midrange frequency.
It is a problem in full-range drivers however, which is why many (even vintage ones) used some form of inductance control (copper cap on pole piece, short circuit rings etc...) thus minimising these effects at the outset.
By comparison the direct distortion produced by this varying field strength is very large (and 2nd order HD and related IMD) and remains unaffected by current drive.
In fact, if careful design is used to offset mechanical non-linearities in the suspension with those of the magnetic field in Drivers (which modern FEM analysis allows and which modern speaker designs often apply) using current drive will materially increase this primary distortion mechanism.
- Voice coil inductance depends strongly on current level (without any displacement) causing non-harmonic distortion
Now we are at one of the biggies, where current drive does indeed produce large changes.
- Magnetic coarseness of iron causing harmonic distortion, hysteresis and Barkhausen noise
Pulleeese. I have had my in audio transformer design. I have a better then fair idea at what field levels this becomes a problem. Let's say it is not an issue in MC transformers.
Plus, current drive changes neither the inherent magnetic system of the speaker, nor does it change the magnet field from the voice coil so any negative interactions remain unaffected, as far as they impact the output from the Driver.
- Resistance changes caused by temperature variations and manufacturing tolerances
From series production of speakers I can assure you the manufacturing tolerances on DCR are quite tight, with reputable makers.
The thermal compression is the other biggie and to someone coming from a pro-audio background a clearly observable problem that is well addressed by current drive.
- Program-dependent contact resistance variations in connectors and switches
Funny you mention this. I can measure the effects of connectors on cables on my AP2. It's usually well over 120dB down on the signal, but can be measured. The change to that from current drive can be real, but we are already so far down that the negtive impact is unlikely to have dramatic effects.
So, some of your additional effects where current drive changes performance are real, however their actual level is sufficiently small I am willing to dismiss them, in other areas it would seem the problems you describe are worsned by current drive, not improved.
It's not only the bass. There also seems to happen something in the other ranges. No, I don't have a doctoral thesis to show for proof; it's just my general observation.
Well, forgive me, I have also not done a doctoral thesis on it, but I have compared systems using drivers producting a non-flat (overdamped bass response) driven by SE Amplifiers with highish output impedance and PP Amplifiers with low output impedance (both of my own design and not conventional).
Once digitally equalised to the same frequency response most of the subjective differences disappeared (but not when using solid state amplifiers), the remaining ones I considered at the time mainly the result of the difference in harmonic distortion between the Amplifers.
My experiements with current driven Speakers in the 80's (incidentally using TDA2030 in a fully active setup, dual 20cm woofer with bridged TDA2030 plus current boost transsitors for bass to 250Hz, 12cm wideband driver as midrange 250Hz to 5KHz, 2cm supronyl plastic dome as tweeter above 5KHz) did not produce a speaker that was subjectively preferable to either of my other "references" I used at the time.
The references where an O18 Studio Monitor Pair (12" Coaxial Schulze TH315 in large vented boxes with matching solid state Amp and build in EQ for the speakers) or my 3-Way EV Horn System (15" EVM Woofer vented, 8HD Midrange Horn and T35 Tweeter) driven by 50W Studio Tube Amplifiers with the build in EQ defeated, output impedance of that tube amp was very low, it used large amounts of NFB.
Of course, this was not a controlled experiment and only my personal preference. Using our limited measurement gear at the time it was hard to penetrate why the active, current drive speaker did not outperform the others. It is worth noting that both my references had midrange/treble drivers using Alnico magnets AND respectively 8dB and 15dB greater efficiency.
I will also add that going to current drive in the Mid and Treble and to microphone based MFB for the bass significantly improved the active speaker compared to the original simple active voltage drive, just not enough to make it overtake "real He-Man Speakers (TM)" that where many time it's size and used pro-drivers, not hifi drivers. These experiences resulted in me discontinuing experiments with current drive, after I escaped from east germany in early 1989, it just did not seem worth the extra effort.
Ciao T
Last edited:
Hi Thorsten,
raising efficiency in whatever way will impair linearity by the laws of physics. Statements and findings suggesting otherwise are wrong.
Mother Nature adopted sound level compression for human hearing with the ratio of 4:1. The purpose was exactly the same: to increase sensitivity, the side effect is also exactly the same: large distortion.
Why not start talking about the much more bothersome intermodulation distortion, for instance? Efficient speakers produce plenty of it. What observations did Martin Colloms make in that regard? (You should really choose your information sources carefully).
Nice to discuss with you, your argument is full of holes.
raising efficiency in whatever way will impair linearity by the laws of physics. Statements and findings suggesting otherwise are wrong.
Mother Nature adopted sound level compression for human hearing with the ratio of 4:1. The purpose was exactly the same: to increase sensitivity, the side effect is also exactly the same: large distortion.
The defects of voltage drive can only be cured by current drive.(BTW, the driver in one of my commercial speakers gets around this by placing the non-conductive magnet inside the voice coil and the solid steel part forming the magnetic circuit outside, a very efficient way to negate the issue in systems with voltage drive).
It´s not that I give a dime for measurements, but I find those figures way too optimistic. It`s impossible to accomplish heavy fivefold energy transformation without considerable distortion even with more promising (inefficient) techniques.I also have with the measured distortion of a Tannoy Autograph Speaker originating with MJ at around 100dB/1m SPL (1W power). Interestingly it shows around equal levels of 2nd and 3rd HD (3rd HD slightly lower) for the cone driver part between 100Hz and 1KHz (also quite flat distribution) at around 0.3%.
Above 1KHz where the compression driver takes over 2nd HD is increased (as one would expect from the non-linearity of the air in the compression chamber, to around 0.6% (eyballing), but 3rd HD (the one that is reduced by current drive) is dropped dramatically to well below 0.1%.
All this simply does not make sense to me. Besides, you can`t cancel anything. Of course, you can convert the 2nd HD (to something worse).Well, HD in SE Amplifiers will tend to change the original notes harmonic spectrum, however this will be mainly a preponderance of 2nd HD and the change will be small. Still, I found that cancelling 2nd HD in speakers with the SE Amplifiers 2nd HD (needs to be correct phase relationship, so measurements are needed) sounded notably cleaner.
You still continue messing with those second and third order HD numbers to no avail.Funny, I have been saying this for years.
Why not start talking about the much more bothersome intermodulation distortion, for instance? Efficient speakers produce plenty of it. What observations did Martin Colloms make in that regard? (You should really choose your information sources carefully).
I haven't had the opportunity to do that.I see you read my old article about SE Amplifiers where pointed all this out.
What is not measured does not mean it`s not there. In fact, it`s not much that can be measured - reliably anyway.The highest level harmonic caused by this is 3rd HD (the rest can be largely be implied by understanding the mechanisms and knowing the 3HD levels) and is often measured for speakers (whereas higher harmonics rearely are).
I don`t share your approach at all. Again, why are you so miffed about the exceedingly benign second and third order harmonics? Just because they appear on your scope, representing all evil?This is untrue. Devices in balanced configuration (or with even order HD cancellation between multiple stages add LESS 2nd HD (ideally non) to the signal, but the original harmonics of the signal are correctly amplified (and of course additionally distorted).
I agree with you here. Therefore feedback does upset the subjectively preferable harmonic distribution, and in addition, introduces various new distortions.Not as such. Feedback lowers the lower order harmonics at the expense of a multiplication process that generates higher order products, which have a much greater audibility.
Nice to discuss with you, your argument is full of holes.
Hi,
So, you say the many and varied sets of measurements that show significant reduction of 3rd HD (not 2nd HD though) at a given SPL for speaker systems with greatly above average efficiency, compared to average efficient speakers, including Soundsteage, Stereophile, Klang & Ton as well as Hobby Hifi are all actually wrong?
For fun I attach distortion measurements for the TAD 4001 compression driver with horn and other TAD Drivers (manufactures data, but generally accurate).
As you can see, at an SPL of 110dB/1M the 2nd HD is between 1 and 3% (depending on frequency) and 3rd HD is at no worse than 0.3% and averages around 0.15%. Remember, this is at 110dB and 2nd HD will fall linear (so for 90 dB we can expect 0.1-0.3% 2nd HD) and 3rd HD from eddy current distortion will fall with the square of the signal (so we can expect this to be 0.003% 3rd HD from eddy current, though some 3rd HD from other sources will remain, so I'd expect maybe 0.05% 3rd HD).
Equally, for the cone drivers above 100Hz we see at 96dB/SPL around 0.1% 2nd and 3rd HD evenly distributed.
So for this speaker at 90dB SPL using VOLTAGE DRIVE we will have around 0.1% 2nd HD to 2KHz and rising smoothly to 0.3% 2nd HD from this towards 20KHz and 3rd HD will be appreciably below 0.1%, both values that are VERY LOW for 90dB/1m from a Speaker and under consideration of the human auditory systems behaviour they are highly likely to be inaudible AND they are actually lower than the distortion from many SE Amplifiers at the 0.5W such a speaker would require for 90dB/1m.
So sorry, the data clearly contradicts your contention that an increase in efficiency MUST increase non-linearity. I could dig out more examples (there are many on my HDD) but I have better things to do, anyway, even a single example serves as absolute proof that your theory "more efficiency = more distortion" is not based in physical reality.
Actually, compression of air will lead to nonlinearities that follow the square of the air pressure. Yes, this is non-linearity, one of the various.
And yes, the human auditory system on a mechanical level has extremely high distortion. In my over a decade old article on SE Amplifiers I reproduced this graph:
http://i53.tinypic.com/2cmsvh4.gif
So I know about the compression and attendendant non-linearity in the human auditory system and have for years. But this does not mean compression drivers or indeed loudspeakers must follow the same pattern due to some physical law.
That is patently untrue.
ALL of the issues where current drive produces a material improvement in the speakers behaviour CAN be addressed in different ways at the very design of the driver.
For example, all the impedance variations caused by modulation of the inductance can be addressed by eliminating or minimising the inductive part of the voice coils impedance, something that many speaker manufacturers nowadays do. If the impedance does not vary with signal because the inductance is suppressed no distortion can be created this way.
Equally, thermal compression can be addressed by using suitable thermal design of the speaker and using alloys for the voice coil that show reduced thermal coefficient.
Some other benefits that are claimed for current drive I discussed above, their magnitude is either very small or they are completely imaginary.
As always, once we understand the nature of a given problem, we can elect many ways to deal with it.
Current drive is one of these options, designing drivers that behave correctly under voltage is another and the one many leading speaker driver manufacturers have taken, as far back as the 1960's and 1970's (perhaps unwittingly), though such approaches are now becoming more common.
Well, the figures are what was measured.
Here the (indepedent - not manufacturer) measurements of the Tannoy GRF Autograph enclosure loaded with Monitor Red Drivers (around 1960's vintage) from MJ:
If we scale to 90dB/1m we find around 0.1% 2nd HD up to 1KHz then rising to around 0.2..3% and 0.1% 3rd HD up to 1KHz and then lowering to around 0.05% 3rd HD.
So, the figures are not optimistic but realistic. Actually, it should be blindingly obvious to anyone who understands how speaker drivers operate that for WELL IMPLEMENTED speaker systems the distortion for a given SPL must fall as efficiency is increased.
Well, clearly there are more possibilities between heaven and earth than are dreamt of in your philosophy, WuYit.
It seems your philosophy is in dire need of rectification.
VISITA INTERIORA TERRAE RECTIFICANDO INVENIS OCCVLTVM LAPIDEM
I am discussing them because they are DOCUMENTED and they serve as indicators (at least to those who understand how Speakers (or amplifiers etc) actually work of what is happeing with higher harmonics produced by the same processes.
Because IMD and THD in simple systems (those that do not use looped feedback, such as speaker drivers) are linked to the same underlying nonlinearities and have around the same maginitude.
I have measured enough of this stuff to always get suspicious if I get IMD measurements that are more than a few dB different from the HD measurements. It indicates something afoot.
Efficient Speakers that have low levels of HD also have low levels of IMD.
So I do NOT understand where you get these wired ideas from, which appear in clear opposition to the observable facts.
If you have actual measured data that shows speakers with an IMD level that is dramatically higher than their HD, why not produce them?
Martin in his review did not elaborate, nor did he have to. He also understands the mechanisms that cause HD and IMD in Speakers and hence is aware that if you know how one of these behaves, you know the other.
I am not miffed by them. I am merely observing what is happening.
Had you read my article you would know I am more concerned regarding 3rd/odd order HD and the related IMD. As 3rd order levels are indicative of what happens at higher (odd) orders AND are often well documented I am going about them at length.
raising efficiency in whatever way will impair linearity by the laws of physics. Statements and findings suggesting otherwise are wrong.
So, you say the many and varied sets of measurements that show significant reduction of 3rd HD (not 2nd HD though) at a given SPL for speaker systems with greatly above average efficiency, compared to average efficient speakers, including Soundsteage, Stereophile, Klang & Ton as well as Hobby Hifi are all actually wrong?
For fun I attach distortion measurements for the TAD 4001 compression driver with horn and other TAD Drivers (manufactures data, but generally accurate).

As you can see, at an SPL of 110dB/1M the 2nd HD is between 1 and 3% (depending on frequency) and 3rd HD is at no worse than 0.3% and averages around 0.15%. Remember, this is at 110dB and 2nd HD will fall linear (so for 90 dB we can expect 0.1-0.3% 2nd HD) and 3rd HD from eddy current distortion will fall with the square of the signal (so we can expect this to be 0.003% 3rd HD from eddy current, though some 3rd HD from other sources will remain, so I'd expect maybe 0.05% 3rd HD).
Equally, for the cone drivers above 100Hz we see at 96dB/SPL around 0.1% 2nd and 3rd HD evenly distributed.
So for this speaker at 90dB SPL using VOLTAGE DRIVE we will have around 0.1% 2nd HD to 2KHz and rising smoothly to 0.3% 2nd HD from this towards 20KHz and 3rd HD will be appreciably below 0.1%, both values that are VERY LOW for 90dB/1m from a Speaker and under consideration of the human auditory systems behaviour they are highly likely to be inaudible AND they are actually lower than the distortion from many SE Amplifiers at the 0.5W such a speaker would require for 90dB/1m.
So sorry, the data clearly contradicts your contention that an increase in efficiency MUST increase non-linearity. I could dig out more examples (there are many on my HDD) but I have better things to do, anyway, even a single example serves as absolute proof that your theory "more efficiency = more distortion" is not based in physical reality.
Mother Nature adopted sound level compression for human hearing with the ratio of 4:1. The purpose was exactly the same: to increase sensitivity, the side effect is also exactly the same: large distortion.
Actually, compression of air will lead to nonlinearities that follow the square of the air pressure. Yes, this is non-linearity, one of the various.
And yes, the human auditory system on a mechanical level has extremely high distortion. In my over a decade old article on SE Amplifiers I reproduced this graph:
http://i53.tinypic.com/2cmsvh4.gif
So I know about the compression and attendendant non-linearity in the human auditory system and have for years. But this does not mean compression drivers or indeed loudspeakers must follow the same pattern due to some physical law.
The defects of voltage drive can only be cured by current drive.
That is patently untrue.
ALL of the issues where current drive produces a material improvement in the speakers behaviour CAN be addressed in different ways at the very design of the driver.
For example, all the impedance variations caused by modulation of the inductance can be addressed by eliminating or minimising the inductive part of the voice coils impedance, something that many speaker manufacturers nowadays do. If the impedance does not vary with signal because the inductance is suppressed no distortion can be created this way.
Equally, thermal compression can be addressed by using suitable thermal design of the speaker and using alloys for the voice coil that show reduced thermal coefficient.
Some other benefits that are claimed for current drive I discussed above, their magnitude is either very small or they are completely imaginary.
As always, once we understand the nature of a given problem, we can elect many ways to deal with it.
Current drive is one of these options, designing drivers that behave correctly under voltage is another and the one many leading speaker driver manufacturers have taken, as far back as the 1960's and 1970's (perhaps unwittingly), though such approaches are now becoming more common.
It´s not that I give a dime for measurements, but I find those figures way too optimistic.
Well, the figures are what was measured.
Here the (indepedent - not manufacturer) measurements of the Tannoy GRF Autograph enclosure loaded with Monitor Red Drivers (around 1960's vintage) from MJ:

If we scale to 90dB/1m we find around 0.1% 2nd HD up to 1KHz then rising to around 0.2..3% and 0.1% 3rd HD up to 1KHz and then lowering to around 0.05% 3rd HD.
So, the figures are not optimistic but realistic. Actually, it should be blindingly obvious to anyone who understands how speaker drivers operate that for WELL IMPLEMENTED speaker systems the distortion for a given SPL must fall as efficiency is increased.
It`s impossible to accomplish heavy fivefold energy transformation without considerable distortion even with more promising (inefficient) techniques.
Well, clearly there are more possibilities between heaven and earth than are dreamt of in your philosophy, WuYit.
It seems your philosophy is in dire need of rectification.
VISITA INTERIORA TERRAE RECTIFICANDO INVENIS OCCVLTVM LAPIDEM
You still continue messing with those second and third order HD numbers to no avail.
I am discussing them because they are DOCUMENTED and they serve as indicators (at least to those who understand how Speakers (or amplifiers etc) actually work of what is happeing with higher harmonics produced by the same processes.
Why not start talking about the much more bothersome intermodulation distortion, for instance?
Because IMD and THD in simple systems (those that do not use looped feedback, such as speaker drivers) are linked to the same underlying nonlinearities and have around the same maginitude.
I have measured enough of this stuff to always get suspicious if I get IMD measurements that are more than a few dB different from the HD measurements. It indicates something afoot.
Efficient speakers produce plenty of it.
Efficient Speakers that have low levels of HD also have low levels of IMD.
So I do NOT understand where you get these wired ideas from, which appear in clear opposition to the observable facts.
If you have actual measured data that shows speakers with an IMD level that is dramatically higher than their HD, why not produce them?
What observations did Martin Colloms make in that regard? (You should really choose your information sources carefully).
Martin in his review did not elaborate, nor did he have to. He also understands the mechanisms that cause HD and IMD in Speakers and hence is aware that if you know how one of these behaves, you know the other.
I don`t share your approach at all. Again, why are you so miffed about the exceedingly benign second and third order harmonics? Just because they appear on your scope, representing all evil?
I am not miffed by them. I am merely observing what is happening.
Had you read my article you would know I am more concerned regarding 3rd/odd order HD and the related IMD. As 3rd order levels are indicative of what happens at higher (odd) orders AND are often well documented I am going about them at length.
WuYit;2457036Nice to discuss with you said:So are all of them. Mine does appear however to at least be in line with observable reality and has few large holes.
I leave it to others to consider the holes in your philosophy, Mr Yit...
Ciao T
Thorsten,
Lowering the level of the low order harmonics gives protruded high orders and unpleasing sound.
It`s reasonable to expect high distortion in speakers and output stages, but the sound still can be appealing if the distortion is right kind. Measurements cannot give a hint. Low measured distortion does not ensure ear-friendly sound, actually, often designates the opposite. High measured distortion does not necessarily mean high perceived distortion, likewise, low measured distortion can be highly objectionable.
Physics and mathematics do not cover this issue, it has deep-seated biological, psychological, psychoacoustical and philosophical aspects.
Your concern is misplaced. Rather the high order harmonics are accompanied by several times higher level of IM products.Had you read my article you would know I am more concerned regarding 3rd/odd order HD and the related IMD.
There´s nothing magical or special about the 3rd harmonic whatsoever, its just one of a large number playing the same role of equal importance. Low order harmonic distortion is much less audible.As 3rd order levels are indicative of what happens at higher (odd) orders AND are often well documented I am going about them at length.
Lowering the level of the low order harmonics gives protruded high orders and unpleasing sound.
Possibilities do exist, but have explicit limits due to the ubiquitous presence of conflicting factors and for other reasons.Well, clearly there are more possibilities between heaven and earth than are dreamt of in your philosophy, WuYit.
Reproduced sound must unconditionally follow the ear`s self-distortion pattern to be transparent, not due to some physical law, but because the auditory system has no separate rules for reproduced sound, which definitely won´t be processed according to made-up scales. Hearing is strongly nonlinear in every respect, so measurements are insignificant even when accurate. Measurements data are useful to convince customers, but in an academic context, we don`t want to cheat ourselves, do we?But this does not mean compression drivers or indeed loudspeakers must follow the same pattern due to some physical law.
It`s reasonable to expect high distortion in speakers and output stages, but the sound still can be appealing if the distortion is right kind. Measurements cannot give a hint. Low measured distortion does not ensure ear-friendly sound, actually, often designates the opposite. High measured distortion does not necessarily mean high perceived distortion, likewise, low measured distortion can be highly objectionable.
Physics and mathematics do not cover this issue, it has deep-seated biological, psychological, psychoacoustical and philosophical aspects.
Hi,
There is no magic IMD. If you have X% of XHD then any intermodulation product of the same order will have approximately the same level, otherwise we are dealing with a system that is not a simple "forward system", like a speaker, feedbackless amplifier or simple PCM DAC without noiseshaping.
Any system that shows substantial differences has in essence "hidden" non-linearities, such as those exposed (if present) by such testing in Amplifiers with looped feedback or DAC's that use noiseshaping and other processes that are analogous to feedback.
I repeat, in "simple" non-looped systems having the 2nd and 3rd HD levels usually allows a very good prediction of the higher order even and odd harmonics. Clearly you fail to understand both why I am saying this and how I arrive at such an conclusion.
As are the IMD products caused by it.
However, we must be CLEAR with what we mean by "less audible".
It means that an equal level of a higher harmonics (and the unavoidably linked intermodulation) is much more audibly in direct relation of the order of the harmonic, compared to 2nd harmonics.
The precise weighting is still being heavily debated.
My own proposal for an axiom (that is backed by my experience so far) is that in order to maked by the human hearings process any harmonic (and IMD component) must be significantly (around 20dB or more) below the levels of HD/IMD present in the human hearing mechanism.
As long as this is obeyed distortion is masked and hence remains inaudible.
So, as long as 3rd HD (for example) remains around 20dB lower than that in the human hearing system it will remain inaudible, REGARDLESS of the level of 2nd HD and even harmonics.
Only once we have levels of HD/IMD that exceed this audibility treshold is the distribution of the harmonics relevant and then a preponderance of odd harmonics gives a sound best described as "dissonant", while a preponderance of even harmonics gives an audible alteration, but one that is even at times judged as improvement by the listeners.
The way you are stating it, axiomatically, it is plainly wrong.
Patently NOT. It must merely ensure all distortion present remains low enough to be masked.
Actually, in academic context there is much good work on the subject of how and what we hear and how the actual mechanisms (including the ones in the nervous system and brain) operate. Most of these are nowadays in the context of research into hearing prosthetics (the rest relates to perceptual coding), it may be illuminating should you apply yourself to some such.
I would rephrase this as:
"The sound still can be appealing if the distortion is NOT OF THE WRONG kind.
Measurements are merely a quantification of certain qualities.
They are meaningless without being able to interpret them and to correlate with the qualities desired/required in practice.
Once we actually can interpret measurements correctly they become a meaningful tool.
Surely you would not suggest to abandon all quantitative measurements in all walks of life (such as how much beer is in a glass, how much money is in your account or how much a certain piece of steak weighs)?
I have made the same point many times. The difference is that I seem to know a bit how the measurements relate to what we hear, while you do not.
The term philosophy denotes a love of knowledge. Any Knowledge. I have always understood myself as a philosopher in this antique sense. So understanding substantially the general nature of the problems in ALL the domains you mention (and some you do not) is natural for me.
And I can tell you ex cathedra that physics and mathematics cover the magisteria they are applicable to perfectly well. And that these magisteria have substantial overlap with ones you seem to wish to consider non-overlapping, as in the fallacy of S.J. Gould.
In other words, the physics and mathematics of distortion, harmonics etc. are essential in the understanding of pysio- and psychoacoustical impact of them (incidentally, biological and physiological cover the same area, you may as well omit one) and as Philosophy is all encompassing they all actually form facets of the Philosophy of hearing and good sound.
Ciao T
Your concern is misplaced. Rather the high order harmonics are accompanied by several times higher level of IM products.
There is no magic IMD. If you have X% of XHD then any intermodulation product of the same order will have approximately the same level, otherwise we are dealing with a system that is not a simple "forward system", like a speaker, feedbackless amplifier or simple PCM DAC without noiseshaping.
Any system that shows substantial differences has in essence "hidden" non-linearities, such as those exposed (if present) by such testing in Amplifiers with looped feedback or DAC's that use noiseshaping and other processes that are analogous to feedback.
There´s nothing magical or special about the 3rd harmonic whatsoever, its just one of a large number playing the same role of equal importance.
I repeat, in "simple" non-looped systems having the 2nd and 3rd HD levels usually allows a very good prediction of the higher order even and odd harmonics. Clearly you fail to understand both why I am saying this and how I arrive at such an conclusion.
Low order harmonic distortion is much less audible.
As are the IMD products caused by it.
However, we must be CLEAR with what we mean by "less audible".
It means that an equal level of a higher harmonics (and the unavoidably linked intermodulation) is much more audibly in direct relation of the order of the harmonic, compared to 2nd harmonics.
The precise weighting is still being heavily debated.
My own proposal for an axiom (that is backed by my experience so far) is that in order to maked by the human hearings process any harmonic (and IMD component) must be significantly (around 20dB or more) below the levels of HD/IMD present in the human hearing mechanism.
As long as this is obeyed distortion is masked and hence remains inaudible.
So, as long as 3rd HD (for example) remains around 20dB lower than that in the human hearing system it will remain inaudible, REGARDLESS of the level of 2nd HD and even harmonics.
Only once we have levels of HD/IMD that exceed this audibility treshold is the distribution of the harmonics relevant and then a preponderance of odd harmonics gives a sound best described as "dissonant", while a preponderance of even harmonics gives an audible alteration, but one that is even at times judged as improvement by the listeners.
Lowering the level of the low order harmonics gives protruded high orders and unpleasing sound.
The way you are stating it, axiomatically, it is plainly wrong.
Reproduced sound must unconditionally follow the ear`s self-distortion pattern to be transparent,
Patently NOT. It must merely ensure all distortion present remains low enough to be masked.
but in an academic context, we don`t want to cheat ourselves, do we?
Actually, in academic context there is much good work on the subject of how and what we hear and how the actual mechanisms (including the ones in the nervous system and brain) operate. Most of these are nowadays in the context of research into hearing prosthetics (the rest relates to perceptual coding), it may be illuminating should you apply yourself to some such.
It`s reasonable to expect high distortion in speakers and output stages, but the sound still can be appealing if the distortion is right kind.
I would rephrase this as:
"The sound still can be appealing if the distortion is NOT OF THE WRONG kind.
Measurements cannot give a hint.
Measurements are merely a quantification of certain qualities.
They are meaningless without being able to interpret them and to correlate with the qualities desired/required in practice.
Once we actually can interpret measurements correctly they become a meaningful tool.
Surely you would not suggest to abandon all quantitative measurements in all walks of life (such as how much beer is in a glass, how much money is in your account or how much a certain piece of steak weighs)?
Low measured distortion does not ensure ear-friendly sound, actually, often designates the opposite. High measured distortion does not necessarily mean high perceived distortion, likewise, low measured distortion can be highly objectionable.
I have made the same point many times. The difference is that I seem to know a bit how the measurements relate to what we hear, while you do not.
Physics and mathematics do not cover this issue, it has deep-seated biological, psychological, psychoacoustical and philosophical aspects.
The term philosophy denotes a love of knowledge. Any Knowledge. I have always understood myself as a philosopher in this antique sense. So understanding substantially the general nature of the problems in ALL the domains you mention (and some you do not) is natural for me.
And I can tell you ex cathedra that physics and mathematics cover the magisteria they are applicable to perfectly well. And that these magisteria have substantial overlap with ones you seem to wish to consider non-overlapping, as in the fallacy of S.J. Gould.
In other words, the physics and mathematics of distortion, harmonics etc. are essential in the understanding of pysio- and psychoacoustical impact of them (incidentally, biological and physiological cover the same area, you may as well omit one) and as Philosophy is all encompassing they all actually form facets of the Philosophy of hearing and good sound.
Ciao T
Quite well. What, then, should I have done with the amps that had impedance options - disqualify them out?ThorstenL said:It was not apparent to me and more to the point, if there are options, does this justify multiple entries?
I only tried to introduce my own perspective on the distortive effects. If I don't tediously repeat every detail of your remarks, that does not mean that I haven't read them. You should also know that I was not comparing the effects to 'sound perception' but relative to each other.Well, again you show you have not bothered to read.
I note that the MATERIAL effects that deliniate current drive from voltage drive is the cancellation of eddy current distortion, which is as I remarked a cubic function, forming a long train of odd harmonics that are mathematically related to the third harmonic, which is the one that is commonly easily measured in speakers and hence can be used as "diagnostic" indicator for eddy current distortion, and the cancellation of thermal compression.
Other effect exists but they tend to be not even of secondary importance (even the Eddy Current Distortion and Thermal compression are really only of secondary importance in speakers compared to sound perception), but of tertiary and lower...
Reading your notions on the issues here and below, I only have to be concerned: how can even those actively involved in design and consulting in this field be so ignorant and misinformed in these fundamentals? What is your education in circuit theory?Originally Posted by ETM
* Voice coil acting as a microphone for the sound waves reflecting from the cabinet interior and passing through the diaphragm
* Voice coil acting as a microphone for the sound waves from adjacent drivers
Okay, you list these first. Let us be clear. If the cone (or voicecoil) is actually allowed to be modulated by external signals (such as those from inside the cabinet or from other drivers) the result will be that the signal is distorted.
In fact, I would note that, if the effect that you mention was sufficiently material to cause problems, it would form the greatest argument AGAINST current drive, as current drive fails to oppose these effects, which are mechanical in nature.
The effect is quite well measurable and very material at lower mid-frequencies. It corresponds to placing a poor-quality microphone inside the enclosure and mixing the microphone signal thus gained with the original one. Perceiving it requires only application of Kirchhoff's and Ohm's laws. Figure 'a' (attached) represents the case under voltage drive, where u0 is the amplifier output voltage, Rc is voice coil resistance, and voltage source e represents here the microphone EMF, that appears in series with the applied voltage and hence modulates the current through the voice coil. (The actual motional EMF due to intended diaphragm motion also appears in series with the other voltages but for simplicity is not drawn here.)
Instead, in Figure b, where the amplifier is now represented by a current source, the same microphone EMF appearing in the circuit is now unable to modulate the current, which is entirely determined by the feeding current source. Thus, the microphone feedback modulation exists only on voltage drive, and it does not exist on current drive.
Someone is now surely remarking: Hey, the damping factor takes care of all that on voltage drive by controlling diaphragm motion! The truth is, however, stranger than wishful thinking. As I explained earlier in post #202, electrical damping is nonexistent outside the resonance region (above 200 Hz or so for woofers) and therefore is of no help in this issue.
The magnitude of the effect is proportional to the square of driver sensitivity. For low-efficiency woofers of 87-88 dB/W, the feedback ratio is at least 3%, but in poorly damped cabinets considerably more. When the sensitivity is raised just to 93 dB/W, feedback ratios over 10% result; and I don't even go to PA:s.
The outward microphone coupling from adjacent woofers is rather appreciable too. It is lower in magnitude than the inwardly coupled microphone EMF but extends higher in frequency.
In a low-efficiency coaxial driver, EMF coupling from the treble unit to the bass unit varied between 1 and 2 % in a wide frequency range, and the effect doubled when the cone was moved a few millimeters. The square law with sensitivity also holds in these.
(The leakage of enclosure noise through the cone is of course a problem in itself (Only open baffle designs are free from it.) and can also be determined quite reliably. It can only be addressed by heavy use of effective damping material which also serves to lower the mechanical Q.)
This is about non-idealities at mid-frequencies. While the non-idealities in themselves are not affected by current-drive, they have a corruptive effect on the motional EMF, which is yet significant at these frequencies, this degraded EMF then further degrading the current on voltage drive just as happens with the microphone EMF (see the figures).Originally Posted by ETM
* Mechanical and pneumatic non-idealities of the moving parts causing unpredictable EMF-effects
The mechanical and pneumatic non-linearities remain unaffected by current drive. In fact, if drivers are carefully designed to have low distortion with voltage drive (which many current day drivers are) switching to current drive will substantially increase the overall non-linearity, as experienced in one of the reviews of your book where current driver substantially raised the drivers distortion, not lowered it.
Modern drivers tend to carefully control these problems, in fact even many vintage drivers are surprisingly good in this context.
By a certain technique, this EMF can be extracted from the driver, and its behavior is really nothing such that one would like to blend into one's audio signal. In many mid-woofers, there can be seen humps in the impedance graph resulting from these EMF irregularities, and notably in high-efficiency drivers they can be quite pronounced.
Not surprisingly, the reviewer of the book focused on bass region damping issues, when my message has been that current-drive is not so necessary at the low end. The increase in THD was observed near and below the resonant frequency and should not be a wonder with a high Q value and with a box so large that, against my ideas, the driver is left operating at the mercy of its own nonlinear spring and damping forces. I also wonder whether the currents were made equal at each point for the two modes.
No, the angle modulation is a separate effect that stems from the geometry of the vectors of the three components of impedance (DC resistance, motional impedance, and inductive impedance) in the range where the impedance modulus is low. The outcome is a frequency modulation akin to that produced by the Doppler effect but greater in magnitude.Originally Posted by ETM
* Bl-variation causing modulation of impedance's angle and hence phase modulation of current at middle frequencies
* Position-dependent inductance of voice coil causing both amplitude and phase modulation (as in post #66)
BL variations are caused by varying field stength for different postions of the voice coil. The effect on the impedance in the midrange is minimal, the actual reason the impedance is modulated is varying inductance with the position of the voicecoil respective to the polepiece.
WRONG. Considering a typical hifi driver at 2 kHz frequency, the impedance magnitude varies by +/-10% as a function of cone displacement within the rated Xmax limits. This means that on voltage drive a 2 kHz tone exhibits +/- 10% amplitude modulation with accompanied phase modulation also. According to basic theory of amplitude modulation, the distortion components thus generated amount to 5% of the original amplitude, giving rise to a full 7% of total intermodulation distortion. Unlike the Bl variation, this effect becomes material already at relatively low excursions.The effect remains minimal in drivers that do not have great excursion and remains still small in the midrange as the inductive reactance of the voicecoil is low at midrange frequency.
It is a problem in full-range drivers however, which is why many (even vintage ones) used some form of inductance control (copper cap on pole piece, short circuit rings etc...) thus minimising these effects at the outset.
The inductive reactance of the voice coil IS NOT LOW in the midrange. This cannot be inferred from the driver impedance curve because of the masking effect of the motional impedance. I have actual measured data on this from a typical low-efficiency hifi woofer with 32 mm voice coil diameter. The inductive impedance equals voice coil resistance Rc slightly above 1 kHz. At 500 Hz, the inductive impedance is yet more than half of Rc; and still at 100 Hz 15% of Rc. And this was only a 6.5-inch Vifa. When the size and efficiency increase, the proportion of inductive impedance also increases. Thus, the inductance modulation effect is VERY MATERIAL throughout the middle and treble regions.
Yes, it can be mitigated by copper caps and rings but at what cost; and in what percentage of all consumer applications this is actually being done? I'm not only considering the audiophiles.
In Seas Excel line for example, where these rings are employed, the drop in inductance is not very striking. Perhaps a half or so. Enough to make a difference but inadequate to solve the problem.
Assuming the variation in Bl in a quality driver is of the same order of magnitude than the variation in impedance at the high mid-frequencies (about +/-10%), the intermodulation distortion due to these effects will also be of the same magnitude. The inductance modulation though decreases towards low frequencies, but otherwise they are well comparable.By comparison the direct distortion produced by this varying field strength is very large (and 2nd order HD and related IMD) and remains unaffected by current drive.
There are rather high field levels in the pole pieces, not very far from saturation. It would be a real wonder if the well-known nonlinear magnetization properties of the materials didn't have any impact on the additional field introduced by the voice coil.Originally Posted by ETM
* Magnetic coarseness of iron causing harmonic distortion, hysteresis and Barkhausen noise
Pulleeese. I have had my in audio transformer design. I have a better then fair idea at what field levels this becomes a problem. Let's say it is not an issue in MC transformers.
How do you know it is merely eddy currents that are responsible for the HD in current? Eddy currents as such are linear effects and unable to produce distortion unless there exists some distortion mechanism that deflects these currents from linear behavior. These mechanisms are complicated but always they come down to the magnetic non-idealities of the structures.
Again, the attached equivalent circuits illustrate the principle. The voltage source e represents now the voltage across the voice coil inductance, that is, the inductive EMF produced. Due to the magnetization nonlinearities occurring in the system, e is not a linear function of the original signal but reflects all the iron-based aberrations, that on voltage drive become injected to the voice coil current.Plus, current drive changes neither the inherent magnetic system of the speaker, nor does it change the magnet field from the voice coil so any negative interactions remain unaffected, as far as they impact the output from the Driver.
For audiophiles who regularly refresh their contacts this may not be an issue, but with aged and oxidized contacts in layman use, with poor connectors and switches or relays on the path, the situation is entirely different and therefore worth giving a mention. Mechanical vibration is also a degrading factor, and sometimes outputs are protected by fuses, that itself are also nonlinear.Originally Posted by ETM
* Program-dependent contact resistance variations in connectors and switches
Funny you mention this. I can measure the effects of connectors on cables on my AP2. It's usually well over 120dB down on the signal, but can be measured. The change to that from current drive can be real, but we are already so far down that the negtive impact is unlikely to have dramatic effects.
Thorsten,
You may use the term "masking", however I´m referring to the ear`s distortion due to cochlear compression. The masking mechanism is something else serving different purpose, low level tones in close proximity to higher level tones remaining unheard, this effect is not included in the single tone tests. For example, the self-generated second harmonic at 90 dB is 10%. In other words, 10% second harmonic distortion at normal listening level (and higher) is not audible regardless of any circumstances.
Why in the world would it be wrong? Harmonics constitute a oneness.The way you are stating it, axiomatically, it is plainly wrong
I do know. They do not relate at all. I`m not against measurements so long as they are kept separated from perception.I have made the same point many times. The difference is that I seem to know a bit how the measurements relate to what we hear, while you do not.
Why on earth would it be "Patently NOT."?It must merely ensure all distortion present remains low enough to be masked.
You may use the term "masking", however I´m referring to the ear`s distortion due to cochlear compression. The masking mechanism is something else serving different purpose, low level tones in close proximity to higher level tones remaining unheard, this effect is not included in the single tone tests. For example, the self-generated second harmonic at 90 dB is 10%. In other words, 10% second harmonic distortion at normal listening level (and higher) is not audible regardless of any circumstances.
This segment is very muddy again. You should better accept existing knowledge, you can unlikely make valuable research findings in this area. Your measurements mislead you hideously.As are the IMD products caused by it.
However, we must be CLEAR with what we mean by "less audible".
It means that an equal level of a higher harmonics (and the unavoidably linked intermodulation) is much more audibly in direct relation of the order of the harmonic, compared to 2nd harmonics.
The precise weighting is still being heavily debated.
My own proposal for an axiom (that is backed by my experience so far) is that in order to maked by the human hearings process any harmonic (and IMD component) must be significantly (around 20dB or more) below the levels of HD/IMD present in the human hearing mechanism.
As long as this is obeyed distortion is masked and hence remains inaudible.
So, as long as 3rd HD (for example) remains around 20dB lower than that in the human hearing system it will remain inaudible, REGARDLESS of the level of 2nd HD and even harmonics.
Only once we have levels of HD/IMD that exceed this audibility treshold is the distribution of the harmonics relevant and then a preponderance of odd harmonics gives a sound best described as "dissonant", while a preponderance of even harmonics gives an audible alteration, but one that is even at times judged as improvement by the listeners.
Hi,
You could have normalised to 8 Ohm and for either best or worst output impedance.
And I tried to give an indication of the relative magnitude of the effects, in absolute terms and in relative terms respective to voltage drive.
As you have written the book on it I would actually have expected some basic hard numbers, like at a level of 95dB/1m with such and such a speaker I get this result, distortion type A is reduced X dB with current drive, distortion type B is reduced by Y dB with current drive and so on. At least Messrs Hawkesford and Mills had the decency to include some actual data in their research.
I could counter that I am appalled by the lack of understanding of the relative magnitudes and relative audibility of effects that are being presented as being affected by "current drive" in a way that suggests they are significant, but I guess I shall not.
So, let me get this right. You are saying the microphone effect is significant (we come back to that later), yet in the train of tought you also comment on the non-existence of damping (correctly so, I may add, if using multiway speakers with passive crossovers). Now no electrical damping (or low electrical damping) is actually in our case caused by feeding the woofer from a high impedance, which approximates current-drive.
So first you are saying the effect is material and then dismiss your own notion (correctly I might add).
Thank you for doing my work for me.
This deserves a little more consideration.
First, the "poor quality microphone" is not anywhere near as bad as you make out. If it was, it would also be a poor speaker.
Secondly, the voice coil can only induce a voltage if the cone is being moved by the sound pressure of the sound bouncing around inside the box, of applied by another drive unit. So in the first instance the problem is that cone is being moved by an unwanted signal, that causes all sorts of interference.
This already has to have happened in order to for the voltage to appear. So, in fact, the voltage that appears is NOT actually the CAUSE for distortion, it is the RESULT of mechanical distortion.
With voltage drive (especially in active systems) the virtual short-circuit of the Amplifier will actually attempt to counteract this unwanted cone movement of the driver, by distorting the current in the voice coil in opposition to the unwanted cone movement.
However as electrical damping is minimal in speaker drivers (except at resonances), as you so astutely observed, the amplifier does not succeed too well. Hence essentially the distortion is only reduced by a small amount, when using voltage drive, compared to current drive, I would be surprised if it would even be measured reliably in the midrange (due to the drivers Z approaching the voice coil DCR closely - exception compression drivers).
So in this particular case the single-minded focus on not distorting the current in voice coil leads to greater distortion in the actual output of the speaker, even though it can be argued that the current in the voice coil is now less distorted. At any extent, my main point was actually that the effect is so marginal in magnitude regarding the driver output, when comparing it between current and voltage drive, that it can be safely consigned to the "insignificant" effects in the context.
First, these "humps" are not down some magical EMF non-linearity, but to cone/diaphragm resonances. And again you vapous statement that are empty. Something can be quite pronounced in something else, meaning it is unlikely in actual reality to be common, even though extreme cases happen.
Yes. To be clear. In the range where the impedance modulus is it is low precisely because the inductive reactance is low compared to the voice coil DCR and the motional impedance is low compared to the voice coil DCR. Therefore effects from modulating either in the very range you claim this effect takes place is the range where it is most difficult to demonstrate.
In this you may be right. Given that the magnitude of doppler distortion in speakers has been shown to be next to non, if any just about any distortion is larger...
Doppler Distortion in loudspeakers
I could now go and look through my issues of Voice coil, find a driver that has much less impedance modulation and claim my point proven. I shall not, however. Because the effect is quite real, though 10% is excessive for a well constructed driver.
What I would like to point out however is that any driver that has build in inductance control (copper caps on pole piece, short circuit rings in the magnet system etc. et all) shows much lower changes to impedance.
Yes, however you omit to consider how large the electromechanical non-linearity has to be (variation of magnet field with travel of the voice coil) with a magnet system that causes such a large modulation inductance and how much amplitude modulation and other problems that would create...
So, first, a driver that has +/-10% impedance modulation at 2KHz from travel within in Xmax is probably best not considered "high fidelity" with the best of intentions and further, the problems from that modulation are minor, next to the problems of magnet field non-linearity, which are not addressed by current drive.
In fact, the best way to deal with all this stuff is to simply make sure that the driver requires minimal movement for a given SPL, such as given with large format high efficiency cone drivers or compression drivers.
Ooops, now I have mananged again to illustrate why large format, high efficiency speaker system are so superior... Well, why not.
The effect becomes material at excursions that cause enough displacement of the cone to modulate the voice coil inductance appreciably. The BL modulation becomes material at much lower excursions in many drivers, ask Mr Klippel.
So, no matter how we view this, it is clearly desirable to construct speaker drivers that have minimal inductance to start with and a linear magnet field with cone travel. Happily enough, many of the top european and chinese driver manufacturers now apply such techniques in their higher grade models, pro audio driver manufacturers have done so already as a matter of routine for many decades.
Okay, first the definition of "midrange": 320-2,560 Hz according to one source...
Second, a "bogey" bass/midrange driver, Wavecor WF120BD04 (appx. 5" Diameter) from Voice Coil 9/2010, the latest on my HDD (I really need to download the later issues). For those that do not know Wavecor, they are a large, low cost OEM producer of drivers, both custom and OTS. They have a solid reputation for quality, but not neccesarily for leading edge technology or even very unusual technology.
This driver shows 0.2mH inductance at 0mm excursion and varies between 0.165mH and 0.215mH across the +/-4mm excursion range (rated Xmax) but only between 0.19 and 0.205mH for +/-1mm Excursion. The voice coil DCR is 6 Ohm.
So the impedances come out as follows at 1280Hz:
Now lets look in the same way at Bl, shall we?
At the same time BL is 5.4 at rest and drops to 4.5 at +/-4mm excursion.
So, first, the degree of current modulation at either level of excursion is less than that of the BL modulation.
Secondly, we can see that the BL modulation will lead to primarily 3rd and odd order harmonics and related IMD, while the inductance modulation will cause mainly 2nd HD.
Whichever and how much, clearly this driver does not show any majorly higher problem from inductance modulation than it suffers from BL Modulation.
Will benefit from current drive?
Complex question, I would need to add in the suspension stiffness vs. excursion in the mix and the suspension stiffness seems carefully tailored to offset some other non-linearities. I would need to look at the effect of that combined with inductance modulation and Bl modulation and we could then derive meaningful data on how much difference current vs. voltage drive really makes.
But I leave this exercise to the reader, I have spend enough time.
Big assumption, guvenor, innit?
How about instead actually quoting real numbers, as I have done above? I assume you have them, or are you talking about more hearsay, just like with tube amp's.
Precisely. You should of course be aware that at saturation Barkhausen noise cannot exist and with a nearly saturated piece of iron there are also minimal numbers of magnetic dipoles available to cause Barkhausen noise. Any EE101 student would know that.
Where barkhausen noise becomes a potential problem is where the fields are VERY low (e.g. MC stepup transformers).
Oh, they will have an effect. Mainly from the fact that solid iron is employed which causes very significant levels of distortion due to eddy current losses (following a cubic function to the current incidentally).
The remainder can be also measured with a Klippel analyser, if one is so inclined.
It is not "merely", but "mainly.
And I know that eddy currents losses as such are NONLINEAR effects (as should anyone who has completed EE101). You may wish to look an easy introduction, like:
"Relationship among eddy current loss, frequency, maximum flux density"
IEEE transactions on electromagnetics Vol.Mag-17 No4 July 1981
Yo Sakaki, Shin Ichi Imagi
Well, perhaps there is a certain degree, still, it would take VERY BAD connections to amount to heap of big eyed beans from venus or a chinese penny's worth of tea...
I thin k I will leave it here. It is not my job to educate you.
I merely took exception to one specific claim and really do not have the time to deconstruct your whole house of cards.
Finally, current drive does have the ability to improve some areas of performance in Speaker Drivers, on that we are actually in agreement.
What we disagree on is that first of all these play any material role in "Tube Amplifier Sound" and secondly we strongly disagree on which areas are materially improved by current drive and which are minor.
As you wrote the book, please illustrate your claims with actual numbers.
Ciao T
Quite well. What, then, should I have done with the amps that had impedance options - disqualify them out?
You could have normalised to 8 Ohm and for either best or worst output impedance.
I only tried to introduce my own perspective on the distortive effects.
And I tried to give an indication of the relative magnitude of the effects, in absolute terms and in relative terms respective to voltage drive.
As you have written the book on it I would actually have expected some basic hard numbers, like at a level of 95dB/1m with such and such a speaker I get this result, distortion type A is reduced X dB with current drive, distortion type B is reduced by Y dB with current drive and so on. At least Messrs Hawkesford and Mills had the decency to include some actual data in their research.
Reading your notions on the issues here and below, I only have to be concerned: how can even those actively involved in design and consulting in this field be so ignorant and misinformed in these fundamentals?
I could counter that I am appalled by the lack of understanding of the relative magnitudes and relative audibility of effects that are being presented as being affected by "current drive" in a way that suggests they are significant, but I guess I shall not.
The effect is quite well measurable and very material at lower mid-frequencies.
....
.... [Thor, much snipped for clarity]
....
...electrical damping is nonexistent outside the resonance region (above 200 Hz or so for woofers) and therefore is of no help in this issue.
So, let me get this right. You are saying the microphone effect is significant (we come back to that later), yet in the train of tought you also comment on the non-existence of damping (correctly so, I may add, if using multiway speakers with passive crossovers). Now no electrical damping (or low electrical damping) is actually in our case caused by feeding the woofer from a high impedance, which approximates current-drive.
So first you are saying the effect is material and then dismiss your own notion (correctly I might add).
Thank you for doing my work for me.
It corresponds to placing a poor-quality microphone inside the enclosure and mixing the microphone signal thus gained with the original one.
This deserves a little more consideration.
First, the "poor quality microphone" is not anywhere near as bad as you make out. If it was, it would also be a poor speaker.
Secondly, the voice coil can only induce a voltage if the cone is being moved by the sound pressure of the sound bouncing around inside the box, of applied by another drive unit. So in the first instance the problem is that cone is being moved by an unwanted signal, that causes all sorts of interference.
This already has to have happened in order to for the voltage to appear. So, in fact, the voltage that appears is NOT actually the CAUSE for distortion, it is the RESULT of mechanical distortion.
With voltage drive (especially in active systems) the virtual short-circuit of the Amplifier will actually attempt to counteract this unwanted cone movement of the driver, by distorting the current in the voice coil in opposition to the unwanted cone movement.
However as electrical damping is minimal in speaker drivers (except at resonances), as you so astutely observed, the amplifier does not succeed too well. Hence essentially the distortion is only reduced by a small amount, when using voltage drive, compared to current drive, I would be surprised if it would even be measured reliably in the midrange (due to the drivers Z approaching the voice coil DCR closely - exception compression drivers).
So in this particular case the single-minded focus on not distorting the current in voice coil leads to greater distortion in the actual output of the speaker, even though it can be argued that the current in the voice coil is now less distorted. At any extent, my main point was actually that the effect is so marginal in magnitude regarding the driver output, when comparing it between current and voltage drive, that it can be safely consigned to the "insignificant" effects in the context.
In many mid-woofers, there can be seen humps in the impedance graph resulting from these EMF irregularities, and notably in high-efficiency drivers they can be quite pronounced.
First, these "humps" are not down some magical EMF non-linearity, but to cone/diaphragm resonances. And again you vapous statement that are empty. Something can be quite pronounced in something else, meaning it is unlikely in actual reality to be common, even though extreme cases happen.
No, the angle modulation is a separate effect that stems from the geometry of the vectors of the three components of impedance (DC resistance, motional impedance, and inductive impedance) in the range where the impedance modulus is low.
Yes. To be clear. In the range where the impedance modulus is it is low precisely because the inductive reactance is low compared to the voice coil DCR and the motional impedance is low compared to the voice coil DCR. Therefore effects from modulating either in the very range you claim this effect takes place is the range where it is most difficult to demonstrate.
The outcome is a frequency modulation akin to that produced by the Doppler effect but greater in magnitude.
In this you may be right. Given that the magnitude of doppler distortion in speakers has been shown to be next to non, if any just about any distortion is larger...
Doppler Distortion in loudspeakers
WRONG. Considering a typical hifi driver at 2 kHz frequency, the impedance magnitude varies by +/-10% as a function of cone displacement within the rated Xmax limits.
I could now go and look through my issues of Voice coil, find a driver that has much less impedance modulation and claim my point proven. I shall not, however. Because the effect is quite real, though 10% is excessive for a well constructed driver.
What I would like to point out however is that any driver that has build in inductance control (copper caps on pole piece, short circuit rings in the magnet system etc. et all) shows much lower changes to impedance.
This means that on voltage drive a 2 kHz tone exhibits +/- 10% amplitude modulation with accompanied phase modulation also.
Yes, however you omit to consider how large the electromechanical non-linearity has to be (variation of magnet field with travel of the voice coil) with a magnet system that causes such a large modulation inductance and how much amplitude modulation and other problems that would create...
So, first, a driver that has +/-10% impedance modulation at 2KHz from travel within in Xmax is probably best not considered "high fidelity" with the best of intentions and further, the problems from that modulation are minor, next to the problems of magnet field non-linearity, which are not addressed by current drive.
In fact, the best way to deal with all this stuff is to simply make sure that the driver requires minimal movement for a given SPL, such as given with large format high efficiency cone drivers or compression drivers.
Ooops, now I have mananged again to illustrate why large format, high efficiency speaker system are so superior... Well, why not.
According to basic theory of amplitude modulation, the distortion components thus generated amount to 5% of the original amplitude, giving rise to a full 7% of total intermodulation distortion. Unlike the Bl variation, this effect becomes material already at relatively low excursions.
The effect becomes material at excursions that cause enough displacement of the cone to modulate the voice coil inductance appreciably. The BL modulation becomes material at much lower excursions in many drivers, ask Mr Klippel.
So, no matter how we view this, it is clearly desirable to construct speaker drivers that have minimal inductance to start with and a linear magnet field with cone travel. Happily enough, many of the top european and chinese driver manufacturers now apply such techniques in their higher grade models, pro audio driver manufacturers have done so already as a matter of routine for many decades.
The inductive reactance of the voice coil IS NOT LOW in the midrange.
Okay, first the definition of "midrange": 320-2,560 Hz according to one source...
Second, a "bogey" bass/midrange driver, Wavecor WF120BD04 (appx. 5" Diameter) from Voice Coil 9/2010, the latest on my HDD (I really need to download the later issues). For those that do not know Wavecor, they are a large, low cost OEM producer of drivers, both custom and OTS. They have a solid reputation for quality, but not neccesarily for leading edge technology or even very unusual technology.
This driver shows 0.2mH inductance at 0mm excursion and varies between 0.165mH and 0.215mH across the +/-4mm excursion range (rated Xmax) but only between 0.19 and 0.205mH for +/-1mm Excursion. The voice coil DCR is 6 Ohm.
So the impedances come out as follows at 1280Hz:
Code:
X L (mH) R Z I Delta I Delta I % Delta I dB
-4mm 0.165 6 8.3 0.34 -0.02 -5.92% -0.50
-1mm 0.19 6 8.7 0.33 -0.01 -1.62% -0.14
0mm 0.2 6 8.8 0.32 0.00 0.00% 0.00
1mm 0.205 6 8.9 0.32 0.00 0.79% 0.07
4mm 0.215 6 9.0 0.31 0.01 2.34% 0.21
Now lets look in the same way at Bl, shall we?
At the same time BL is 5.4 at rest and drops to 4.5 at +/-4mm excursion.
Code:
X Bl Delta BL Delta BL % Delta BL dB
-4mm 4.50 0.90 16.67% 1.58
-1mm 5.30 0.10 1.85% 0.16
0mm 5.40 0.00 0.00% 0.00
1mm 5.30 0.10 1.85% 0.16
4mm 4.50 0.90 16.67% 1.58
So, first, the degree of current modulation at either level of excursion is less than that of the BL modulation.
Secondly, we can see that the BL modulation will lead to primarily 3rd and odd order harmonics and related IMD, while the inductance modulation will cause mainly 2nd HD.
Whichever and how much, clearly this driver does not show any majorly higher problem from inductance modulation than it suffers from BL Modulation.
Will benefit from current drive?
Complex question, I would need to add in the suspension stiffness vs. excursion in the mix and the suspension stiffness seems carefully tailored to offset some other non-linearities. I would need to look at the effect of that combined with inductance modulation and Bl modulation and we could then derive meaningful data on how much difference current vs. voltage drive really makes.
But I leave this exercise to the reader, I have spend enough time.
Assuming the variation in Bl in a quality driver is of the same order of magnitude than the variation in impedance at the high mid-frequencies
Big assumption, guvenor, innit?
How about instead actually quoting real numbers, as I have done above? I assume you have them, or are you talking about more hearsay, just like with tube amp's.
There are rather high field levels in the pole pieces, not very far from saturation.
Precisely. You should of course be aware that at saturation Barkhausen noise cannot exist and with a nearly saturated piece of iron there are also minimal numbers of magnetic dipoles available to cause Barkhausen noise. Any EE101 student would know that.
Where barkhausen noise becomes a potential problem is where the fields are VERY low (e.g. MC stepup transformers).
It would be a real wonder if the well-known nonlinear magnetization properties of the materials didn't have any impact on the additional field introduced by the voice coil.
Oh, they will have an effect. Mainly from the fact that solid iron is employed which causes very significant levels of distortion due to eddy current losses (following a cubic function to the current incidentally).
The remainder can be also measured with a Klippel analyser, if one is so inclined.
How do you know it is merely eddy currents that are responsible for the HD in current? Eddy currents as such are linear effects and unable to produce distortion unless there exists some distortion mechanism that deflects these currents from linear behavior.
It is not "merely", but "mainly.
And I know that eddy currents losses as such are NONLINEAR effects (as should anyone who has completed EE101). You may wish to look an easy introduction, like:
"Relationship among eddy current loss, frequency, maximum flux density"
IEEE transactions on electromagnetics Vol.Mag-17 No4 July 1981
Yo Sakaki, Shin Ichi Imagi
For audiophiles who regularly refresh their contacts this may not be an issue, but with aged and oxidized contacts in layman use, with poor connectors and switches or relays on the path, the situation is entirely different and therefore worth giving a mention. Mechanical vibration is also a degrading factor, and sometimes outputs are protected by fuses, that itself are also nonlinear.
Well, perhaps there is a certain degree, still, it would take VERY BAD connections to amount to heap of big eyed beans from venus or a chinese penny's worth of tea...
I thin k I will leave it here. It is not my job to educate you.
I merely took exception to one specific claim and really do not have the time to deconstruct your whole house of cards.
Finally, current drive does have the ability to improve some areas of performance in Speaker Drivers, on that we are actually in agreement.
What we disagree on is that first of all these play any material role in "Tube Amplifier Sound" and secondly we strongly disagree on which areas are materially improved by current drive and which are minor.
As you wrote the book, please illustrate your claims with actual numbers.
Ciao T
Hi,
Actually, the distortion in the human auditory system (due to ultimatly air non-linearity, incidentally) has a masking effect on distorted signals. I cannot be bothered to dig out the reference now, maybe another day.
Ciao T
I´m referring to the ear`s distortion due to cochlear compression. The masking mechanism is something else serving different purpose,
Actually, the distortion in the human auditory system (due to ultimatly air non-linearity, incidentally) has a masking effect on distorted signals. I cannot be bothered to dig out the reference now, maybe another day.
Ciao T
My own proposal for an axiom (that is backed by my experience so far) is that in order to maked by the human hearings process any harmonic (and IMD component) must be significantly (around 20dB or more) below the levels of HD/IMD present in the human hearing mechanism.
As long as this is obeyed distortion is masked and hence remains inaudible.
So, as long as 3rd HD (for example) remains around 20dB lower than that in the human hearing system it will remain inaudible, REGARDLESS of the level of 2nd HD and even harmonics.
I read this some times ago in this thesis : http://dancheever.com/main/cheever_thesis_final.pdf
page 28.
That makes sense even if it may be hard to experience. How can we calibrate the ears distorsion easily? We have to include amplifier power and speaker efficiency.
Anyway, I still imagine one day measurement will show what a good sound is.
Your reading comprehension fails. I wrote: "As I explained earlier in post #202, electrical damping is nonexistent outside the resonance region (above 200 Hz or so for woofers) and therefore is of no help in this issue." That means the electrical damping is nonexistent regardless of the driving mode, that is, on voltage drive as well as on current drive. Thus, the electrical damping, that is commonly believed to 'control' cone motion on voltage drive, is factually nonexistent at the frequencies of interest and therefore of no help in this issue of EMF feedback on voltage drive.ThorstenL said:So, let me get this right. You are saying the microphone effect is significant (we come back to that later), yet in the train of tought you also comment on the non-existence of damping (correctly so, I may add, if using multiway speakers with passive crossovers). Now no electrical damping (or low electrical damping) is actually in our case caused by feeding the woofer from a high impedance, which approximates current-drive.
So first you are saying the effect is material and then dismiss your own notion (correctly I might add).
Thank you for doing my work for me.
You still hold on to the belief that the short-circuit of the amplifier attempts to counteract this unwanted cone movement. It doesn't even attempt. This is also a thing that can be determined by simulation; no need to argue about it.ThorstenL said:...This already has to have happened in order to for the voltage to appear. So, in fact, the voltage that appears is NOT actually the CAUSE for distortion, it is the RESULT of mechanical distortion.
With voltage drive (especially in active systems) the virtual short-circuit of the Amplifier will actually attempt to counteract this unwanted cone movement of the driver, by distorting the current in the voice coil in opposition to the unwanted cone movement.
However as electrical damping is minimal in speaker drivers (except at resonances), as you so astutely observed, the amplifier does not succeed too well. Hence essentially the distortion is only reduced by a small amount, when using voltage drive, compared to current drive, I would be surprised if it would even be measured reliably in the midrange (due to the drivers Z approaching the voice coil DCR closely - exception compression drivers).
Below is shown the electrical equivalent circuit of a driver (so-called mobility analogy), representing the electrical and mechanical parts. (The dependent sources Em and Id make up the transformer between the electrical and mechanical sides.) The example values given represent a typical smallish woofer in a closed enclosure with the following parameters: Bl = 6 Tm, m = 0.008 kg, spring constant = 2000 N/m, mechanical resistance = 1.5 Ns/m, and Rc = 6 Ω. I this analogy, current flowing to the resonator L-R-C represents applied mechanical force and voltage across the resonator (U) represents diaphragm velocity. Current source Ix represents here the external mechanical force that now attempts to move the diaphragm.
An externally hosted image should be here but it was not working when we last tested it.
When the switch is closed, the operation corresponds to voltage drive, while an open switch corresponds to current drive. By running an AC analysis in both states with Ix as the input and recording voltage U, that indicated the velocity, the difference in diaphragm motion is that shown in the graph. This is the amount of electrical damping that is obtained when the driver operates short-circuited. At the same time, it gives the difference in frequency response between pure voltage and current modes.
At the resonant frequency, the damping reaches here 14 dB, but already one octave away from the resonance it has dropped below 4 dB. Two octaves from the resonance, only less than a decibel is left. This basic model still omitted voice coil inductance. If an inductance model is added in series with Rc, the falloff will be even steeper.
Thus, the belief that the short-circuit of the amplifier would counteract unwanted movement outside the resonance region is fallacious. It is a widespread imagination indeed and a basis of prejudice against current-drive.
Hence, the unwanted cone movement caused by the bouncing pressure inside the enclosure is independent of the driving mode, and so is the microphone EMF voltage component induced according to the law e=Blv. The only difference is that only on voltage drive the current becomes distorted.
The effect can be measured from a speaker that has two identical drivers. One is used to create pressure and the other as the microphone. If it can be assumed that both cones see equal pressure level, the voltage measured from the mic driver also tells the EMF component developed in the active driver. The mic voltage is then related to the active driver's voltage from about 200 Hz upwards.
While the effect is not likely to produce much HD, it forms a totally unnecessary uncertainty factor to impair time domain behavior.
You haven't understood the basic equivalent circuit in my post #267 and its relation to driver impedance. Everything that appears in the impedance curve in addition to the DC resistance is due to EMF:s, either motional or inductive. Resonances itself cannot affect the curve unless they modulate the motional EMF, and all irregularities in the EMF do not give rise to visible changes. Examples of how likely they are can be found in this Pass articleThorstenL said:First, these "humps" are not down some magical EMF non-linearity, but to cone/diaphragm resonances. And again you vapous statement that are empty. Something can be quite pronounced in something else, meaning it is unlikely in actual reality to be common, even though extreme cases happen.
In the range in question, the inductive impedance and motional impedance are typically some 30% of DCR, but due to their phase relationship they largely mask each other in the impedance curve. The effect can be demonstrated by simple reasoning as the motional impedance varies according to (Bl)^2.ThorstenL said:Yes. To be clear. In the range where the impedance modulus is it is low precisely because the inductive reactance is low compared to the voice coil DCR and the motional impedance is low compared to the voice coil DCR. Therefore effects from modulating either in the very range you claim this effect takes place is the range where it is most difficult to demonstrate.
The Doppler effect is quite a hard physical fact, and when combined with the above-mentioned phase modulation, frequency variation of the order of +/-1% can typically be expected. I that means nothing, obviously comparable amounts of clock jitter would also be acceptable to you.ThorstenL said:In this you may be right. Given that the magnitude of doppler distortion in speakers has been shown to be next to non, if any just about any distortion is larger...
The inductance variation is quite a linear function of displacement while the Bl variation remains often minor at low excursions. Thus the inductance variation may become material first.ThorstenL said:The effect becomes material at excursions that cause enough displacement of the cone to modulate the voice coil inductance appreciably. The BL modulation becomes material at much lower excursions in many drivers, ask Mr Klippel.
Below are shown the impedance variations at 2 kHz for three randomly picked hifi drivers from three manufacturers (unlike your carefully selected sample where special techniques are employed and for which you use simplistic way to calculate impedance at a lower frequency). The drivers tested were A: Vifa M13SG-09-08, B: Seas P14RC4Y/DC, and C: Peerless 833429. (You cannot claim these manufacturers are not "high fidelity".) In each driver, the measurement has been extended to the rated Xmax limit.
An externally hosted image should be here but it was not working when we last tested it.
The variation is in all cases about +/-10%, giving rise to the over 7% IMD I mentioned. I don't have the Bl profiles of these, but if the span is as high as 20%, like in the impedance, the IMD due to Bl variation will be quite the same as due to the impedance variation. If the Bl the variation is 16.7%, like in your example, IMD due to impedance variation will even dominate at 2 kHz and above. The main difference is that the modulation frequency of the Bl variation is twice that of the Z variation (assuming symmetric Bl profile).
So we see that for mainstream hifi woofers, distortion due to impedance variation is on voltage drive well comparable to that caused by the Bl variation. With the copper caps and rings recently being employed in some expensive types, the problem can be significantly reduced but usually not solved.
I haven't emphasized that effect. As well I could have left it out, but it was mentioned for completeness mostly for those who may be after the high-resolution formats.ThorstenL said:Where barkhausen noise becomes a potential problem is where the fields are VERY low (e.g. MC stepup transformers).
Surely the end result is nonlinear; I was only saying there is always a cause behind an effect.ThorstenL said:And I know that eddy currents losses as such are NONLINEAR effects (as should anyone who has completed EE101).
Feel free to spend your time in a useful way. Educators also usually have an education in their respective field.ThorstenL said:I thin k I will leave it here. It is not my job to educate you.
I merely took exception to one specific claim and really do not have the time to deconstruct your whole house of cards.
I'm trying to figure out how one would design a non-GNFB current-drive power amplifier. I would use a cascoded common-emitter output stage. But without feedback, getting offset as low as 10mV would be a challenge. 10mv/8R=1.25mA, this is a high art to ask of bulky 10A output transistors. Perhaps an FET cascoded by an FET? And then you have to worry about thermals... *shudder*
And even then it would have to be class A. Class AB, common emitter? *shudder*
Even in a feedback amp, the typical emitter follower output stage is not well suited to this because it is designed specifically for low output impedance, and this would undermine the current feedback circuitry, which would have to correct for it.
To achieve high output impedance without using a common-emitter or common-base output stage, a series resistor would be needed, and the amp would need very high-voltage rails. This would be very inefficient.
An interesting challenge...
- keantoken
And even then it would have to be class A. Class AB, common emitter? *shudder*
Even in a feedback amp, the typical emitter follower output stage is not well suited to this because it is designed specifically for low output impedance, and this would undermine the current feedback circuitry, which would have to correct for it.
To achieve high output impedance without using a common-emitter or common-base output stage, a series resistor would be needed, and the amp would need very high-voltage rails. This would be very inefficient.
An interesting challenge...
- keantoken
Hi,
I would not call it "non-existent", but limited by the voicecoil DCR.
I hold no beliefs.
However, I do remember the basics of electromagnetic I learned (again) in long and incredibly boring classes many years ago.
Hence, IF I move a coil in a magnetic field I MUST induce a Voltage. If I apply a "load" to the coil (in our case short circuit plus voice coil DCR) I will remove energy from the moving coil, requiring more force for the same movement.
This is the basic physics.
Now we can look at the various impedance in common speakers and conclude that mid/high frequencies the impedances in the crossover are as large or larger as the voice coil DCR and so on, but that does not change the physics.
What it does change, however is the degree of difference we obtain from current drive.
Actually, the induced voltage is the same, but the cone motion MUST be reduced. This is in fact easily determined by experiment. How much, this a different story.
Now, I previously asked you to actually quantify the effects in terms of order of magnitude. You have placed loads of virtual Ink on virtual pages to proove what is already known and what was never really in dispute.
What you have not done is put an actual number on the distortion created, derived from experiment.
Your bottom line is:
Okay, let's do this again.
1) You say the effect is not going to produce much distortion, so let's agree, minimal distortion...
2) You say that the effect impairs time domain behaviour. Well, does it? NO. The "problem" is actually a symptom of impairment of the timedomain reponse due to reflection in the MECHANICAL DOMAIN. You are completely confusing CAUSE AND EFFECT!
No matter of voltage or current drive it still shows up in the acoustical output, because no material damping happens either way.
And yes, the current is subject to a little distortion in voltage drive. You would find that this distortion in current is in opposition to cone movement. So where cone is being moved in ways other than that by the drive signal ANY distortion of the drive current in the voice coil will OPPOSE the mechanical movement causing it in the first place and hece reduce the impact on the acoustic output (not by much, that I'll grant).
Now you call this "distortion" when using voltage because it shows up in a domain you focus on and can easily measure/model, the voice coil current. I call it "reduction in distortion" with voltage drive as I focus on the acoustic output of the driver.
So with voltage drive it is reduced in magnitude (if only minimally, as you pointed out) over current drive, so while the "timedomain uncertainty factor" (WTF, why don't you just call it what it is, reflections) in the voice coil current is greater with voltage drive, it MUST be reduced by an equal amount in the acoustical output from the driver.
I have. What you fail to understand however is my main point.
The same above WRT the "microphone effects" in speakers.
Yes. However it does not exist in moving coil speakers in any appreciable degree, which would be obvious anyway if one makes the effort to calculate the approximate amounts.
Both what little Doppler Effect is there and the Phase Intermodulation (which incidentally also exists in all amplifiers with looped feedback) are swamped by the levels of non-linearity in the Speakers magnet field and mechanical parameters. So as far as real speakers in the real world are concerned, the effect can be dismissed as insignificant.
You are again making completely empty statements.
You expect +/-1% Doppler and PIM Distortion at (say) 60dB SPL? No? 70dB?
At what SPL in a normal driver do you expect 1%?
You have looked at the output of Mr. Klippel's speaker analysis system before, yes?
My specifically selected sample was selected by opening the latest voice coil issue on my Hard-Drive and picking the Driver that was in there.
All drivers are discontinued. I would only consider Seas among these three manufacturers to make "state of the art" Drivers, the others turned out dross no better than cheap chinese drivers (e.g. the one from Wavecor I referenced). Many well designed cheap chinese drivers do better.
With "recently introduced" you mean 10 to 20 Years ago, where high performance, state of the art drivers are concerned, yes?
So, shall we agree that with cheap and poorly designed drivers current drive does help somewhat, modern well designed drivers significantly reduce this advantage.
Your original list had Barkhausen noise right after the eddy current losses in the polepiece.
Now I understand that your list was NOT in order of the level of Impact these effects have... May I suggest you actually indicate with each an every problem the approximate size of the problem in respect to each other and to the main causes of distortion in speakers and in how they relate, not to the current in the voice coil, but to the acoustic output of the speaker.
Sure, they are. So are the results of thewire in the voice coil only being only of limited purity or of the solder joins between voice coil and leadouts.
The issue I am taking is one of:
ARE THE EFFECTS MATERIAL IN THE ACOUSTIC OUTPUT OF THE DRIVER.
Unless it suits you otherwise, then you are perfectly happy to treat the effect as cause (see above).
So, my bottom line remains the following:
1) The major contributors to driver non-linearity are mechanical in nature, or down to the magnet system non linearity.
2) The two largest areas where current drive and voltage drive cause differences in performance are in eddy current loss induced distortion (cubic function) due to the solid metal nature of the pole piece and thermal compression, followed by voice coil inductance modulation.
3) Suitable magnet system design can minimise all the issues except thermal compression.
4) Increasing the efficiency of the driver by using a larger surface and lighter diaphragm reduces these issues all proportionately to the order of the function they follow and the increase in efficiency.
If you disagree please illustrate with real numbers that allow comparisons (shall we normalise to 95dB SPL @ 1m).
Ciao T
Thus, the electrical damping, that is commonly believed to 'control' cone motion on voltage drive, is factually nonexistent at the frequencies of interest and therefore of no help in this issue of EMF feedback on voltage drive.
I would not call it "non-existent", but limited by the voicecoil DCR.
You still hold on to the belief that the short-circuit of the amplifier attempts to counteract this unwanted cone movement. It doesn't even attempt.
I hold no beliefs.
However, I do remember the basics of electromagnetic I learned (again) in long and incredibly boring classes many years ago.
Hence, IF I move a coil in a magnetic field I MUST induce a Voltage. If I apply a "load" to the coil (in our case short circuit plus voice coil DCR) I will remove energy from the moving coil, requiring more force for the same movement.
This is the basic physics.
Now we can look at the various impedance in common speakers and conclude that mid/high frequencies the impedances in the crossover are as large or larger as the voice coil DCR and so on, but that does not change the physics.
What it does change, however is the degree of difference we obtain from current drive.
Below is shown the electrical equivalent circuit of a driver (so-called mobility analogy),
....
Hence, the unwanted cone movement caused by the bouncing pressure inside the enclosure is independent of the driving mode, and so is the microphone EMF voltage component induced according to the law e=Blv. The only difference is that only on voltage drive the current becomes distorted.
Actually, the induced voltage is the same, but the cone motion MUST be reduced. This is in fact easily determined by experiment. How much, this a different story.
Now, I previously asked you to actually quantify the effects in terms of order of magnitude. You have placed loads of virtual Ink on virtual pages to proove what is already known and what was never really in dispute.
What you have not done is put an actual number on the distortion created, derived from experiment.
Your bottom line is:
While the effect is not likely to produce much HD, it forms a totally unnecessary uncertainty factor to impair time domain behavior.
Okay, let's do this again.
1) You say the effect is not going to produce much distortion, so let's agree, minimal distortion...
2) You say that the effect impairs time domain behaviour. Well, does it? NO. The "problem" is actually a symptom of impairment of the timedomain reponse due to reflection in the MECHANICAL DOMAIN. You are completely confusing CAUSE AND EFFECT!
No matter of voltage or current drive it still shows up in the acoustical output, because no material damping happens either way.
And yes, the current is subject to a little distortion in voltage drive. You would find that this distortion in current is in opposition to cone movement. So where cone is being moved in ways other than that by the drive signal ANY distortion of the drive current in the voice coil will OPPOSE the mechanical movement causing it in the first place and hece reduce the impact on the acoustic output (not by much, that I'll grant).
Now you call this "distortion" when using voltage because it shows up in a domain you focus on and can easily measure/model, the voice coil current. I call it "reduction in distortion" with voltage drive as I focus on the acoustic output of the driver.
So with voltage drive it is reduced in magnitude (if only minimally, as you pointed out) over current drive, so while the "timedomain uncertainty factor" (WTF, why don't you just call it what it is, reflections) in the voice coil current is greater with voltage drive, it MUST be reduced by an equal amount in the acoustical output from the driver.
You haven't understood the basic equivalent circuit in my post #267 and its relation to driver impedance.
I have. What you fail to understand however is my main point.
The same above WRT the "microphone effects" in speakers.
The Doppler effect is quite a hard physical fact,
Yes. However it does not exist in moving coil speakers in any appreciable degree, which would be obvious anyway if one makes the effort to calculate the approximate amounts.
Both what little Doppler Effect is there and the Phase Intermodulation (which incidentally also exists in all amplifiers with looped feedback) are swamped by the levels of non-linearity in the Speakers magnet field and mechanical parameters. So as far as real speakers in the real world are concerned, the effect can be dismissed as insignificant.
and when combined with the above-mentioned phase modulation, frequency variation of the order of +/-1% can typically be expected.
You are again making completely empty statements.
You expect +/-1% Doppler and PIM Distortion at (say) 60dB SPL? No? 70dB?
At what SPL in a normal driver do you expect 1%?
The inductance variation is quite a linear function of displacement while the Bl variation remains often minor at low excursions. Thus the inductance variation may become material first.
You have looked at the output of Mr. Klippel's speaker analysis system before, yes?
Below are shown the impedance variations at 2 kHz for three randomly picked hifi drivers from three manufacturers (unlike your carefully selected sample where special techniques are employed and for which you use simplistic way to calculate impedance at a lower frequency).
My specifically selected sample was selected by opening the latest voice coil issue on my Hard-Drive and picking the Driver that was in there.
The drivers tested were A: Vifa M13SG-09-08, B: Seas P14RC4Y/DC, and C: Peerless 833429. (You cannot claim these manufacturers are not "high fidelity".) In each driver, the measurement has been extended to the rated Xmax limit.
All drivers are discontinued. I would only consider Seas among these three manufacturers to make "state of the art" Drivers, the others turned out dross no better than cheap chinese drivers (e.g. the one from Wavecor I referenced). Many well designed cheap chinese drivers do better.
So we see that for mainstream hifi woofers, distortion due to impedance variation is on voltage drive well comparable to that caused by the Bl variation. With the copper caps and rings recently being employed in some expensive types, the problem can be significantly reduced but usually not solved.
With "recently introduced" you mean 10 to 20 Years ago, where high performance, state of the art drivers are concerned, yes?
So, shall we agree that with cheap and poorly designed drivers current drive does help somewhat, modern well designed drivers significantly reduce this advantage.
I haven't emphasized that effect. As well I could have left it out, but it was mentioned for completeness mostly for those who may be after the high-resolution formats.
Your original list had Barkhausen noise right after the eddy current losses in the polepiece.
Now I understand that your list was NOT in order of the level of Impact these effects have... May I suggest you actually indicate with each an every problem the approximate size of the problem in respect to each other and to the main causes of distortion in speakers and in how they relate, not to the current in the voice coil, but to the acoustic output of the speaker.
Surely the end result is nonlinear;
Sure, they are. So are the results of thewire in the voice coil only being only of limited purity or of the solder joins between voice coil and leadouts.
The issue I am taking is one of:
ARE THE EFFECTS MATERIAL IN THE ACOUSTIC OUTPUT OF THE DRIVER.
I was only saying there is always a cause behind an effect.
Unless it suits you otherwise, then you are perfectly happy to treat the effect as cause (see above).
So, my bottom line remains the following:
1) The major contributors to driver non-linearity are mechanical in nature, or down to the magnet system non linearity.
2) The two largest areas where current drive and voltage drive cause differences in performance are in eddy current loss induced distortion (cubic function) due to the solid metal nature of the pole piece and thermal compression, followed by voice coil inductance modulation.
3) Suitable magnet system design can minimise all the issues except thermal compression.
4) Increasing the efficiency of the driver by using a larger surface and lighter diaphragm reduces these issues all proportionately to the order of the function they follow and the increase in efficiency.
If you disagree please illustrate with real numbers that allow comparisons (shall we normalise to 95dB SPL @ 1m).
Ciao T
Hi,
The precisely OPPOSITE way you do one for current drive. So common emitter stages (preferably cascoded) in the output, probably with a floating Split Rail PSU and the speaker connected between PSU centertap and amplifier signal ground...
Ciao T
I'm trying to figure out how one would design a non-GNFB current-drive power amplifier.
The precisely OPPOSITE way you do one for current drive. So common emitter stages (preferably cascoded) in the output, probably with a floating Split Rail PSU and the speaker connected between PSU centertap and amplifier signal ground...
Ciao T
Even in a feedback amp, the typical emitter follower output stage is not well suited to this because it is designed specifically for low output impedance, and this would undermine the current feedback circuitry, which would have to correct for it.
To achieve high output impedance without using a common-emitter or common-base output stage, a series resistor would be needed, and the amp would need very high-voltage rails. This would be very inefficient.
An interesting challenge...
- keantoken
See the figure in Post #1. With this configuration, the output impedance is approx. the sense resistor times the open loop gain, nearly independent of the open loop output impedance of the amplifier. So, yes, you can design your amplifier nearly as you would do for voltage drive.
But, I do agree with you when it comes to a non global feedback amplifier suited for current drive (a transconductance amplifier). I have found it have to be class A with quite a large quiescent current, cascode would be an advantage, but not necessary? But what to do with the offset...?
DC servo for almost no GNFB..?...but what to do with the offset...?
Without taking sides on the noise, there are still plenty of magnetic dipoles available for the voice coil. Otherwise the inductance would be only that obtained in free air.ThorstenL (post #269) said:You should of course be aware that at saturation Barkhausen noise cannot exist and with a nearly saturated piece of iron there are also minimal numbers of magnetic dipoles available to cause Barkhausen noise.
I effect, you are still entertaining the myth that the distorted current by some 'must be so' law would be able to reduce or counteract the mechanical impact that originally created the distortion.ThorstenL said:Actually, the induced voltage is the same, but the cone motion MUST be reduced. This is in fact easily determined by experiment. How much, this a different story.
...
And yes, the current is subject to a little distortion in voltage drive. You would find that this distortion in current is in opposition to cone movement. So where cone is being moved in ways other than that by the drive signal ANY distortion of the drive current in the voice coil will OPPOSE the mechanical movement causing it in the first place and hece reduce the impact on the acoustic output (not by much, that I'll grant).
Now you call this "distortion" when using voltage because it shows up in a domain you focus on and can easily measure/model, the voice coil current. I call it "reduction in distortion" with voltage drive as I focus on the acoustic output of the driver.
So with voltage drive it is reduced in magnitude (if only minimally, as you pointed out) over current drive, so while the "timedomain uncertainty factor" (WTF, why don't you just call it what it is, reflections) in the voice coil current is greater with voltage drive, it MUST be reduced by an equal amount in the acoustical output from the driver.
When an external mechanical force is applied on the cone at frequencies above the resonant region, the force translates into the cone's acceleration (F=ma). The resulting velocity is, however, as the time integral of acceleration, in 90 degree phase lag with the acceleration (and the interfering force). Thus the resulting EMF, electrical current and consequent 'corrective' force are also in phase quadrature with the interfering force and therefore unable to effect ANY real reduction in the unwanted mechanical motion.
I already showed by appropriate simulation that the mechanical damping effect produced by short-circuiting the driver is at the frequencies of interest a fraction of a decibel (and when inductance is regarded, not even that), which is negligible for all practical purposes. So, my point is proven that the distortion of current by the microphone coupling is merely harmful and there is nothing beneficial or corrective effects related to it.
When a thing has been proven, you should either accept the proof and live with it or show that the proof is invalid and present your own data. But instead, you continue to mess things up by recycling the same fallacy by wishful argumentation.
If effect A produces say 5 units of distortion and effect B respectively 1 unit of similar distortion, yes, the contribution of B easily becomes swamped by that of A; but as it is still better to have only 5 units of distortion than 6, it may still be worth going after both of them.ThorstenL said:Both what little Doppler Effect is there and the Phase Intermodulation (which incidentally also exists in all amplifiers with looped feedback) are swamped by the levels of non-linearity in the Speakers magnet field and mechanical parameters. So as far as real speakers in the real world are concerned, the effect can be dismissed as insignificant.
These are not directly related to SPL. The Doppler modulation is proportional to cone velocity and the phase modulation I listed stems from Bl variation. The number is an estimated magnitude in the area of minimum impedance at maximum rated excursion.ThorstenL said:You are again making completely empty statements.
You expect +/-1% Doppler and PIM Distortion at (say) 60dB SPL? No? 70dB?
At what SPL in a normal driver do you expect 1%?
You go even so far as to extract a sentence totally out of its original context (which was eddy currents) and use it in a new context to make it seem as I would be interested in wire impurities. Trolling?ThorstenL said:Sure, they are. So are the results of thewire in the voice coil only being only of limited purity or of the solder joins between voice coil and leadouts.Originally Posted by ETM
Surely the end result is nonlinear;
The issue I am taking is one of:
ARE THE EFFECTS MATERIAL IN THE ACOUSTIC OUTPUT OF THE DRIVER.
Are vaguely coupled parasitic electrical signals at levels of -20 - -30 dB throughout the midrange material? Depends much on one's real objectives (natural sound or something else). Useless to dispute that here.
I have already discussed the magnitudes of the microphone effects like also of the inductance modulation. You should also know it is very difficult to isolate every effect from the whole and determine which portion of which distortion or anomaly is attributable to which effect. I don't owe any numbers to you, and it is also not meaningful to reproduce essential parts of a book in a discussion forum.ThorstenL said:If you disagree please illustrate with real numbers that allow comparisons (shall we normalise to 95dB SPL @ 1m).
I will try to take this into account when making revisions to the text. Maybe some can feel that way as my aspect has been mostly objective engineering.Joe Rasmussen said:...on page 111-112 you take a dismissive attitude toward tube amplifiers as effectively as "distortion generators" (my phrase, not yours). This is likely to cause friction with some of us, admittedly me included, and especially here on diyaudio.com that is not the most palatable position to take.
Yes, it is quite possible to compensate the total impedance to nearly resistive so that the frequency responses become fairly equal in both modes, but as I see it, the sound with voltage amplifiers will actually only improve in proportion to the increase in the impedance seen by the drivers. If we don't want to compromise the current mode operation, only series mode compensation networks are usable, which can only increase the total impedance. The bass peak may, however, be leveled by a shunt network, as the driving mode is not so important there.Joe Rasmussen said:In terms of compromise, it is not a bad solution and at least avoid voltage drive. But, then there is the challenge to loudspeaker designers to make their speakers compatible with both current and voltage drive. It as can be done. And there is potential there as they will sound better with people who are stuck with voltage amps.
I have lately worked on a speaker concept that can be used with ordinary voltage amplifiers and where sensitivity and efficiency are willingly traded for current and sound purity. Here, a series resistor is added for each driver and placed (along with filter capacitors) in a separate heat sinking rear box. The loss in sensitivity can be partially compensated by using two or more drivers in parallel fashion and by using 4-ohm drivers instead of 8. The idea has now been registered as a Utility Model in Finland but is free to be utilized elsewhere.
The two last figures in the drawings describe the concept with a simple 1st-order crossover. (The RCL networks are for resonance damping.) The series resistances 3a, 3b and 4 are preferably about 5 times the nominal impedance of the driver, giving a fairly good current-drive. With such an impedance ratio, the SPL of a driver will decrease to 1/6, i.e. 15-16 dB, compared to direct feed. By using two parallel drivers, like in the figures, the on-axis SPL is, however, increased by 6 dB, leaving a net attenuation of less than 10 dB. Further, by using a 4-ohm variant as the driver we can draw 2-fold current from the amplifier which in practice increases the SPL yet by about 3 dB compared to 8-ohm use. (With equal current, a 4-ohm variant is usually 3 dB weaker.) Thus, the resulting loss in sensitivity will only be some 6-7 dB, not a very big deal.
An important advantage is also that we can use any passive crossover topology without further compromises to the driving mode. The load presented to the amplifier is easy, here with a minimum of some 12 ohms; and due to high amplitudes, crossover distortion is also less an issue.
If effect A produces say 5 units of distortion and effect B respectively 1 unit of similar distortion, yes, the contribution of B easily becomes swamped by that of A; but as it is still better to have only 5 units of distortion than 6, it may still be worth going after both of them.
I agree, but this is also an area of controversy. Some say that there is no point in making an audio amp with distortion below .01% because we can't possibly hear the difference. Meanwhile, we continue to have fun by designing amplifiers with .0001% distortion...
I think this is half subjective and half objective. It is up to you whether you pursue anything more than "good enough". This is relegated mostly to hobbies because "good enough" tends to be the most reliable source of income. It also advances science, even though practical gain can't be anticipated.
It all comes down to that one question I guess: What is the point?
I am not making any arguments about the validity of the ideas discussed by the quote, so I may be taking it out of context.
- keantoken
- Status
- Not open for further replies.
- Home
- Vendor's Bazaar
- The Secret of Tube Amplifiers Revealed - and much more!