Hi Gerrit.
Is something wrong with the tube amplifier voltage trace shown in Post#2 ?
Looking back to Post#1 it is there too.
As if something is saturating or current/voltage limiting;
maybe this also affecting the following settlement !
Regarding Post#5.
The NAD current response shows an increase in mid-range current draw, essentially an increase in the 'shout' normally attributable to a widerange driver.
I suggest that if you put a resistor in series with the output of the NAD having a value the same as output impedance of the triode amplifier the increased mid draw will even out between the amplifiers.
The current curve for the triode shows increased output at higher frequencies, in line with the results observed by Nelson Pass in his Current Source Amplifier .pdf, which I cannot find on his Website, and so I cannot reference it at the moment.
Undamped impedance peaks due to the Replikon cabinet show very clearly.
Hi Klaus,
I have already seen SS amplifier NFB circuits which tailor amplifier output impedance according to frequency, though I do not have a reference to hand.
Maybe it is the mid-range driver circuit transduction 'efficiency' which changes with source impedance - or observed 'shoutiness' due to mid-range and break-up region current draws increasing with NFB corrected drive.
SS NFB amplifiers drive whatever current is necessary through the loudspeaker circuit in order to minimise output terminal voltage error with respect to input, in other words, the amplifier fights against electrical back-EMF AFTER it was generated by an electromechanical event. The amplifier 'correction' arises after the the audible event has finished or is still in progress, and the current flow becomes deleteriously modified in 'music-time' in a manner which cannot similarly arise with current drive or via a well designed non-NFB tube amplifier.
Voltage/current drive can be equalised via series/parallel networks respectively, but mechanical driver characteristics arising after electrical waveform energisation can only be optimally damped via the voice coil and series drive resistance, which the tube amp has but the SS amp does not until fitted with networks or NFB circuitry (complex) being applied to match a specific loudspeaker composite.
Cheers ............ Graham.
Is something wrong with the tube amplifier voltage trace shown in Post#2 ?
Looking back to Post#1 it is there too.
As if something is saturating or current/voltage limiting;
maybe this also affecting the following settlement !
Regarding Post#5.
The NAD current response shows an increase in mid-range current draw, essentially an increase in the 'shout' normally attributable to a widerange driver.
I suggest that if you put a resistor in series with the output of the NAD having a value the same as output impedance of the triode amplifier the increased mid draw will even out between the amplifiers.
The current curve for the triode shows increased output at higher frequencies, in line with the results observed by Nelson Pass in his Current Source Amplifier .pdf, which I cannot find on his Website, and so I cannot reference it at the moment.
Undamped impedance peaks due to the Replikon cabinet show very clearly.
Hi Klaus,
I have already seen SS amplifier NFB circuits which tailor amplifier output impedance according to frequency, though I do not have a reference to hand.
Maybe it is the mid-range driver circuit transduction 'efficiency' which changes with source impedance - or observed 'shoutiness' due to mid-range and break-up region current draws increasing with NFB corrected drive.
SS NFB amplifiers drive whatever current is necessary through the loudspeaker circuit in order to minimise output terminal voltage error with respect to input, in other words, the amplifier fights against electrical back-EMF AFTER it was generated by an electromechanical event. The amplifier 'correction' arises after the the audible event has finished or is still in progress, and the current flow becomes deleteriously modified in 'music-time' in a manner which cannot similarly arise with current drive or via a well designed non-NFB tube amplifier.
Voltage/current drive can be equalised via series/parallel networks respectively, but mechanical driver characteristics arising after electrical waveform energisation can only be optimally damped via the voice coil and series drive resistance, which the tube amp has but the SS amp does not until fitted with networks or NFB circuitry (complex) being applied to match a specific loudspeaker composite.
Cheers ............ Graham.
Hi forr,
Very interesting, I think it's time to join the AES so I have access to the papers.
Hi Graham,
I don't think so. The peak for the voltage is probably between two samples at 2.5 microseconds, a higher sample rate would be required to confirm my suspicion.
I have the paper by Nelson Pass, this was also one of the things that started me on this journey.
I would have to measure the driver in isolation to see if all the nastiness is caused by the Replikon. While I'm at it, I can compare the impedance plot of a stock FE208S and my Phase-plugged and ENABLed FE208S.
You guys certainly have given me a lot to think about, time to make up a schedule for the weekend.
Very interesting, I think it's time to join the AES so I have access to the papers.
Hi Graham,
I don't think so. The peak for the voltage is probably between two samples at 2.5 microseconds, a higher sample rate would be required to confirm my suspicion.
I have the paper by Nelson Pass, this was also one of the things that started me on this journey.
I would have to measure the driver in isolation to see if all the nastiness is caused by the Replikon. While I'm at it, I can compare the impedance plot of a stock FE208S and my Phase-plugged and ENABLed FE208S.
You guys certainly have given me a lot to think about, time to make up a schedule for the weekend.
Hi Sreten,
I disagree that testing amplifiers with a resistance
>> e.g. Z= 6 ohm / phase angle = 45 is ~ equivalent to 4R
loading. <<
calculated to match as you suggest is at all adequate.
The reason being that very few SS amplifiers themselves have a coherent *NFB* characteristic.
Due to stabilisation considerations a NFB amplifier often has a phase shifted element to its correction of driver/inductor generated back-EMF, and the amplifier cannot prevent fractional quadrature (correction group delay) error arising at the output terminal before it attempts to correct.
Good amplifiers might hold coherence to 10kHz, but some high NFB examples might be in quadrature as low as 100Hz, with that amplifier 90 degrees error component additionally being in series with load angle, and the error potential increasing linearly with frequency above its NFB turnover.
With resistor testing the load current is always in phase with voltage and this cannot reveal any new quadrature components which develop in music-time at a reactive/dynamic NFB-loudspeaker interface.
I believe the audible effects of this aspect has contributed to renewed interest in tube and current source amplifiers, but when the problems are properly addressed a SS NFB design can be equally useful.
I concur with others that it is the overall reproduction system which counts, for although there is a current/voltage duality around the resistance of a voice coil, the same cannot be said for cabinet and all driver characteristics - capacitive/inductive.
I note in Post#8 that the dynamically induced current variations modify the voltage trace of the tube output (in series with Ra), though modify the current amplitude of the NAD due to its output voltage being NFB controlled.
Also,
this being something which would need to be considered by those who rely upon CSDs etc., and as considered by Klaus above -
the electrical phase response has been modified by the change in drive impedance,
and with the transduction response being modified, so too must be the reproduction characteristics as well.
Of course this is far from new, for others have shown how drive impedance/damping affects LS response and thus reproduction, and so I wish Gerrit the best of Luck in being better able to analyse the technical aspects of what we hear via our loudspeakers.
Cheers ......... Graham.
I disagree that testing amplifiers with a resistance
>> e.g. Z= 6 ohm / phase angle = 45 is ~ equivalent to 4R
loading. <<
calculated to match as you suggest is at all adequate.
The reason being that very few SS amplifiers themselves have a coherent *NFB* characteristic.
Due to stabilisation considerations a NFB amplifier often has a phase shifted element to its correction of driver/inductor generated back-EMF, and the amplifier cannot prevent fractional quadrature (correction group delay) error arising at the output terminal before it attempts to correct.
Good amplifiers might hold coherence to 10kHz, but some high NFB examples might be in quadrature as low as 100Hz, with that amplifier 90 degrees error component additionally being in series with load angle, and the error potential increasing linearly with frequency above its NFB turnover.
With resistor testing the load current is always in phase with voltage and this cannot reveal any new quadrature components which develop in music-time at a reactive/dynamic NFB-loudspeaker interface.
I believe the audible effects of this aspect has contributed to renewed interest in tube and current source amplifiers, but when the problems are properly addressed a SS NFB design can be equally useful.
I concur with others that it is the overall reproduction system which counts, for although there is a current/voltage duality around the resistance of a voice coil, the same cannot be said for cabinet and all driver characteristics - capacitive/inductive.
I note in Post#8 that the dynamically induced current variations modify the voltage trace of the tube output (in series with Ra), though modify the current amplitude of the NAD due to its output voltage being NFB controlled.
Also,
this being something which would need to be considered by those who rely upon CSDs etc., and as considered by Klaus above -
the electrical phase response has been modified by the change in drive impedance,
and with the transduction response being modified, so too must be the reproduction characteristics as well.
Of course this is far from new, for others have shown how drive impedance/damping affects LS response and thus reproduction, and so I wish Gerrit the best of Luck in being better able to analyse the technical aspects of what we hear via our loudspeakers.
Cheers ......... Graham.
Hello Gerrit,
you are absolutely right: a loudspeaker is a current controlled device.
There is an historical perspective to explain that today: voltage control is the norm.
- the after WWII generalisation of large negative feedback in commercial amplifiers
- the generalization of optimisation methods to design enclosures, methods which are simplified when the hypothesis of a zero ohm source impedance is done.
May be there is other reasons...
Most people has forgotten that Hilliard and Lansing used to design the VOT to be used for "improved perfomance" on an amplifer the output impedance of which is equal to the impedance of the loudspeaker.
Give a look to my reply to Lynn Olson here:
http://www.diyaudio.com/forums/showthread.php?postid=1485847#post1485847
A more recent and very interesting reading is Nelson Pass 's paper:
http://www.firstwatt.com/downloads/cs-amps-speakers.pdf
The control in current of the loudspeaker results in a very important decrease of distortion.
(Nelson Pass's "First Watt" amplifier is a transconductance amplifier with a very large output impedance).
Best regards from Paris, France
Jean-Michel Le Cléac'h
you are absolutely right: a loudspeaker is a current controlled device.
There is an historical perspective to explain that today: voltage control is the norm.
- the after WWII generalisation of large negative feedback in commercial amplifiers
- the generalization of optimisation methods to design enclosures, methods which are simplified when the hypothesis of a zero ohm source impedance is done.
May be there is other reasons...
Most people has forgotten that Hilliard and Lansing used to design the VOT to be used for "improved perfomance" on an amplifer the output impedance of which is equal to the impedance of the loudspeaker.
Give a look to my reply to Lynn Olson here:
http://www.diyaudio.com/forums/showthread.php?postid=1485847#post1485847
A more recent and very interesting reading is Nelson Pass 's paper:
http://www.firstwatt.com/downloads/cs-amps-speakers.pdf
The control in current of the loudspeaker results in a very important decrease of distortion.
(Nelson Pass's "First Watt" amplifier is a transconductance amplifier with a very large output impedance).
Best regards from Paris, France
Jean-Michel Le Cléac'h
Gerrit Boers said:
Graham Maynard said:
Hi,
It is certainly not realistic but that does not mean it is inadequate.
An amplifier that is allegedly "4 ohm" compatable that shuts down
with ~ 3R or less loads is not realistically "4 ohm" compatible.
Which is the sort of thing I was alluding to.
Testing into low resistive loads tells you about reactive capabilities.
Your pet theories on feedback are exactly that, not fact.
🙂/sreten.
Hi Gerrit,
--- I think it's time to join the AES so I have access to the papers.---
This may help you.
Stahl's patent US 4118600 freely downloadable :
http://www.pat2pdf.org/pat2pdf/foo.pl
Hawksford-Mills papers, freely downloadable :
http://www.essex.ac.uk/dces/researc...J12 Distortion reduction MC current drive.pdf
http://www.essex.ac.uk/dces/researc...bdocs/J14 Mills-Hawksford power amplifier.pdf
Greiner, Sims & Travis JAES paper :
JAES Volume 32 Issue 12 pp. 956-963; December 1984
Nonlinear distortion and frequency response aberrations are the major weaknesses of low-frequency loudspeaker systems. A multiple-loop feedback system is presented that deals effectively with both of these problems. The feedback system, utilizing current and velocity feedback, also decouples the system Q factor from enclosure and driver parameters. A theoretical analysis is presented which has been the basis for several successful system designs. Extensive data on several practical systems have been taken to show the usefulness of this approach. Data for two of these systems are given here.
Authors: Greiner, R.A., Sims, Jr., Travis M.
Getting any desired positive output impedance from a high negative feedback amplifier is a rather academic exercise. Just a matter of feedback connections.
If voltage amplifiers have become the norm, it is because it's the most handy and predictable way to drive loudspeakers.
As seen in the above papers, current driving them has some advantages. But it requires that you prefectly know what you do.
--- I think it's time to join the AES so I have access to the papers.---
This may help you.
Stahl's patent US 4118600 freely downloadable :
http://www.pat2pdf.org/pat2pdf/foo.pl
Hawksford-Mills papers, freely downloadable :
http://www.essex.ac.uk/dces/researc...J12 Distortion reduction MC current drive.pdf
http://www.essex.ac.uk/dces/researc...bdocs/J14 Mills-Hawksford power amplifier.pdf
Greiner, Sims & Travis JAES paper :
JAES Volume 32 Issue 12 pp. 956-963; December 1984
Nonlinear distortion and frequency response aberrations are the major weaknesses of low-frequency loudspeaker systems. A multiple-loop feedback system is presented that deals effectively with both of these problems. The feedback system, utilizing current and velocity feedback, also decouples the system Q factor from enclosure and driver parameters. A theoretical analysis is presented which has been the basis for several successful system designs. Extensive data on several practical systems have been taken to show the usefulness of this approach. Data for two of these systems are given here.
Authors: Greiner, R.A., Sims, Jr., Travis M.
Getting any desired positive output impedance from a high negative feedback amplifier is a rather academic exercise. Just a matter of feedback connections.
If voltage amplifiers have become the norm, it is because it's the most handy and predictable way to drive loudspeakers.
As seen in the above papers, current driving them has some advantages. But it requires that you prefectly know what you do.
sreten said:
To test an amplifier into a reactive load you have to put a reactance on it -- ie a load where the imaginary part is not equal to 0. A Resistor has imaginary = 0 so will tell you nothing about an amplifiers ability to drive a reactive load.
Testing an amp into a low resistance only tells you how much current (assuming a voltage source amp) it can deliver into a resistive load.
dave
forr,
Thank you very much for the provided links. I will need some time to digest this.
Jean-Michel,
I didn't know that either. I quickly scanned the paper and was surprised by the stuff they did in 1945. An example: Edge wound aluminum ribbon voice-coil with copper beryllium leads.
Thank you very much for the provided links. I will need some time to digest this.
Jean-Michel,
I didn't know that either. I quickly scanned the paper and was surprised by the stuff they did in 1945. An example: Edge wound aluminum ribbon voice-coil with copper beryllium leads.
Without going thru the papers linked to, does anyone have an idea as to what kind of decrease in distortion can be achieved with current drive?
I have seen measurements of a midbass with and without a series coil that indicates a 6dB, or thereabouts, reduction of harmonic distortion. Wonder if this is about the limit or if it's possible to reduce distortion even more with a higher drive impedance, passive or active.
/Peter
I have seen measurements of a midbass with and without a series coil that indicates a 6dB, or thereabouts, reduction of harmonic distortion. Wonder if this is about the limit or if it's possible to reduce distortion even more with a higher drive impedance, passive or active.
/Peter
planet10 said:
Hi,
Paragraph 1 , no you don't unless your looking at stability.
Paragraph 2 generally is simply wrong, how much current
is generally applicable to any load, not just resistive.
🙂/sreten.
Hi Pan,
Current drive might be observed improving steady sine amplitude response wrt frequency, but does it lead to reduced distortion during *audio* reproduction in 'music-time'?
Voltage drive provides better transient response than current where loudspeakers exhibit capacitive characteristics; current drive is better where loudspeakers exhibit inductive characteristics.
Current drive to a dynamic loudspeaker will be less able to drive against air/cone support springs and move cone mass than current flowing due to voltage drive which does not become reduced until coil motion generates back-EMF.
Fig.22 in Forr's last link above examples current flow with voltage drive. It shows not only the increased current which Sreten looks for in an amplifier, but also shows current reversals due to the imaginary component effects Dave mentions.
As illustrated, such delayed back-EMF causes the output current of a voltage amplifier to become reversed, and thus a non-class-A output stage can become load instead of input driven, which in SS class-AB designs can lead to a switch like reversal of current generated output stage voltage bias in music time as the load current alternates between lag and lead.
Phase ~ time; and the tiny ripples on Gerrits NAD phase trace deserve further investigation, for the Fostex/Replicon did not cause tube amplifier current to become simarly 'time' distorted.
Cheers ........... Graham.
PS. Hi Sreten, you posted after I last checked.
By default, making a high gain NFB SS class-AB amplifier stable introduces a phase change in its damping characteristic. Then there can be an inability of the output stage to maintain coherent control of current wrt voltage which becomes a problem during music reproduction.
This effect will not show up in conventional steady sine and loading tests because it is not related to current driving capabilities; thus increasing the current drive safety margin is not a solution.
Current drive might be observed improving steady sine amplitude response wrt frequency, but does it lead to reduced distortion during *audio* reproduction in 'music-time'?
Voltage drive provides better transient response than current where loudspeakers exhibit capacitive characteristics; current drive is better where loudspeakers exhibit inductive characteristics.
Current drive to a dynamic loudspeaker will be less able to drive against air/cone support springs and move cone mass than current flowing due to voltage drive which does not become reduced until coil motion generates back-EMF.
Fig.22 in Forr's last link above examples current flow with voltage drive. It shows not only the increased current which Sreten looks for in an amplifier, but also shows current reversals due to the imaginary component effects Dave mentions.
As illustrated, such delayed back-EMF causes the output current of a voltage amplifier to become reversed, and thus a non-class-A output stage can become load instead of input driven, which in SS class-AB designs can lead to a switch like reversal of current generated output stage voltage bias in music time as the load current alternates between lag and lead.
Phase ~ time; and the tiny ripples on Gerrits NAD phase trace deserve further investigation, for the Fostex/Replicon did not cause tube amplifier current to become simarly 'time' distorted.
Cheers ........... Graham.
PS. Hi Sreten, you posted after I last checked.
By default, making a high gain NFB SS class-AB amplifier stable introduces a phase change in its damping characteristic. Then there can be an inability of the output stage to maintain coherent control of current wrt voltage which becomes a problem during music reproduction.
This effect will not show up in conventional steady sine and loading tests because it is not related to current driving capabilities; thus increasing the current drive safety margin is not a solution.
Attachments
Hi Graham!
Yes of course. If the driver is non-linear distortion will be there with music. The high impedance driven voice coil becomes more linear. If HD and IMD is decreased with sine/s then music will be reproduced with less distortion as well.
Actually music often is more steady state like than many realise. Vocals, wind instruments, a ringing guitar and so on have a envelope that shifts slowly and contains a series of sines, not much different form a multitone stady state measurement situation.
/Peter
Yes of course. If the driver is non-linear distortion will be there with music. The high impedance driven voice coil becomes more linear. If HD and IMD is decreased with sine/s then music will be reproduced with less distortion as well.
Actually music often is more steady state like than many realise. Vocals, wind instruments, a ringing guitar and so on have a envelope that shifts slowly and contains a series of sines, not much different form a multitone stady state measurement situation.
/Peter
Hi Peter,
It is only a dynamic loudspeaker voice coil force which becomes more linear with current drive, not the physical displacement, and this brings us back to optimum impedance drive to match driver characteristics; driver types, enclosure types, frequency ranges.
What I most often find lacking after a loudspeaker has been 'optimised' to exhibit a flat frequency response are its dynamic timing and transient capabilities. The reproduction just does not 'sound' as real any more.
Cheers .......... Graham.
It is only a dynamic loudspeaker voice coil force which becomes more linear with current drive, not the physical displacement, and this brings us back to optimum impedance drive to match driver characteristics; driver types, enclosure types, frequency ranges.
What I most often find lacking after a loudspeaker has been 'optimised' to exhibit a flat frequency response are its dynamic timing and transient capabilities. The reproduction just does not 'sound' as real any more.
Cheers .......... Graham.
JMMLC
From the Altec paper referenced here
http://www.diyaudio.com/forums/show...847#post1485847
"In the past it has been customary to adjust the amplifier output impedance to a value of approximately one-half to one-third of the average loudspeaker impedance. Improved performance can be obtained with the new loudspaker when the amplifier and loudspeaker impedance are approximately equal."
What the claimed improvement of performance consists of, the authors do not say.
From the Altec paper referenced here
http://www.diyaudio.com/forums/show...847#post1485847
"In the past it has been customary to adjust the amplifier output impedance to a value of approximately one-half to one-third of the average loudspeaker impedance. Improved performance can be obtained with the new loudspaker when the amplifier and loudspeaker impedance are approximately equal."
What the claimed improvement of performance consists of, the authors do not say.
Hello Forr,
Knowing the characters (J.B. Lansing and Hilliard) , I guess it is not just visual. 😉
Best regards from Paris, France
Jean-Michel Le Cléac'h
Knowing the characters (J.B. Lansing and Hilliard) , I guess it is not just visual. 😉
Best regards from Paris, France
Jean-Michel Le Cléac'h
I had noticed the irregularities in the current phase response that Graham mentioned. It seemed worth 'zooming into', so I repeated the measurement with a lower sample rate setting (96KHz) but a higher FFT size (1536000) resulting in a 62.5 mHz resolution. The output level was 250mV for both amps.
The attached image shows the coherence (top) and phase (bottom) for output voltage and current. Notice that the voltage coherence is perfect for both amps, just a light glitch at around 150Hz in case of the SET amp.
The attached image shows the coherence (top) and phase (bottom) for output voltage and current. Notice that the voltage coherence is perfect for both amps, just a light glitch at around 150Hz in case of the SET amp.
Attachments
Hi Gerrit,
I guess we've all waited a long time to be able study something like this.
THANK YOU.
The main difference between SET and SS NFB amplifiers is that the latter has two inputs, with that second input being the NFB sensing node which is connected, hopefully directly, to the output terminal. Many SS designs have series output chokes between the output terminal and the control sensing node, which means that the music waveform at the output terminal can be fractionally modified by back-EMF with respect to the sensing node!
When I simulated amplifier circuits with equivalent dynamic loads (complex R+L+C) I found errors were much greater when the virtual loudspeakers had crossovers which introduce current phase change with respect to voltage around the crossover frequency.
In view of SS output stage class-AB crossover effects being exacerbated by load phase change I ended up examining how the amplifiers behaved in response to 1V and 10V step inputs fed into their output terminal via a nominal load impedance resistor (8 ohm). Some would allow >>100mV of output terminal swing before settling with damped oscillation.
I cannot presume to know exactly what your current investigation is revealing, though clearly measurable phase jitter (noise arising due to NFB correction?) is arising which cannot fail but affect waveform coherence in time.
One significant aspect here is that this is for just one channel.
Where two channels are similtaneously but independently creating such jitter, then any stereo imagery location must be rendered fuzzy and less precise than that provided by the SET.
I also equate this response jitter as being the SS aspect which impairs the 'blackness' behind reproduction, the 'blackness' which immediately deepens with tube amplifiers and competent SS class-A.
I trust that the amplifier input is truly clean such that the input stage is not switching under some circumstances.
Any chance of you repeating this examination with a 2 or 3 way LS with crossover to check for linearity about the upper crossover frequency where most SS NFB designs have fully developed quadrature damping characteristics.
Cheers ....... Graham.
I guess we've all waited a long time to be able study something like this.
THANK YOU.
The main difference between SET and SS NFB amplifiers is that the latter has two inputs, with that second input being the NFB sensing node which is connected, hopefully directly, to the output terminal. Many SS designs have series output chokes between the output terminal and the control sensing node, which means that the music waveform at the output terminal can be fractionally modified by back-EMF with respect to the sensing node!
When I simulated amplifier circuits with equivalent dynamic loads (complex R+L+C) I found errors were much greater when the virtual loudspeakers had crossovers which introduce current phase change with respect to voltage around the crossover frequency.
In view of SS output stage class-AB crossover effects being exacerbated by load phase change I ended up examining how the amplifiers behaved in response to 1V and 10V step inputs fed into their output terminal via a nominal load impedance resistor (8 ohm). Some would allow >>100mV of output terminal swing before settling with damped oscillation.
I cannot presume to know exactly what your current investigation is revealing, though clearly measurable phase jitter (noise arising due to NFB correction?) is arising which cannot fail but affect waveform coherence in time.
One significant aspect here is that this is for just one channel.
Where two channels are similtaneously but independently creating such jitter, then any stereo imagery location must be rendered fuzzy and less precise than that provided by the SET.
I also equate this response jitter as being the SS aspect which impairs the 'blackness' behind reproduction, the 'blackness' which immediately deepens with tube amplifiers and competent SS class-A.
I trust that the amplifier input is truly clean such that the input stage is not switching under some circumstances.
Any chance of you repeating this examination with a 2 or 3 way LS with crossover to check for linearity about the upper crossover frequency where most SS NFB designs have fully developed quadrature damping characteristics.
Cheers ....... Graham.
Hi forr,
I cannot open your last link, and a Search throws up too much to go through.
Cheers ...... Graham.
I cannot open your last link, and a Search throws up too much to go through.
Cheers ...... Graham.
Hi Graham,
Glad to be of service.
Your description matches my observations, to my ears the SET has a more precise stereo image and better localization. I've been looking for measurements that correlate with this observation. Could this be it?
Alas, I do not have such a system at the moment. I do have a pair of Audax woofers and two Seas tweeters lying around, maybe I can construct something with those.
It was my intention to repeat this experiment with a higher Q driver (like the Audax) and see if there are any significant changes in the results.
Attached you'll find the 'zoomed in' plot for coherence and amplitude. Again, the amplitude plot for the voltage is clean. The nastiness is only visible in the current plot.
Glad to be of service.
Your description matches my observations, to my ears the SET has a more precise stereo image and better localization. I've been looking for measurements that correlate with this observation. Could this be it?
Alas, I do not have such a system at the moment. I do have a pair of Audax woofers and two Seas tweeters lying around, maybe I can construct something with those.
It was my intention to repeat this experiment with a higher Q driver (like the Audax) and see if there are any significant changes in the results.
Attached you'll find the 'zoomed in' plot for coherence and amplitude. Again, the amplitude plot for the voltage is clean. The nastiness is only visible in the current plot.
Attachments
Hi Gerrit,
Another aspect I reasoned regarding NFB SS amplifiers, is that because they match the voltage output waveform to input waveform regardless of current, the amplifier countered driver back-EMF alternations become trapped within the composite loudspeaker circuitry.
Thus with an underhung single driver like most Fostex types, all current flow errors are trapped in series with amplifier/cable/voice coil/cone impedances and cannot fail to modify on-going reproduction after the event which generated them - a phased haze where there should be clarity.
Thus an underhung widerange driver cannot fail but to 'shout' much more acutely at higher frequencies when driven by a SS amplifier - unless - it is driven via a series impedance having values similar to the transformed anode resistance of a tube amplifier via which some of the back-EMF and re-transduced mechanical and microphonic output due to cone break up and secondary air-side events can become passively dissipated externally; as with the triode plots.
Many years ago, observations here with tweeters on extended leads showed the audibility of SS class-AB amplifier exacerbated noise events was most notable because they modified higher audio frequencies (not clipping), even though they were often caused by low and mid frequency related inductions. Tube amp tweeter output always seemed just plain normal, even via AB1 with NFB (the application of which is of course limited by circuit phase changes).
Lynn Olsen's Ariel LS is known to reveal weaknesses in SS amplifiers, I believe because it crosses over to the tweeter at a higher frequency where most NFB responses are already in full quadrature.
Of course all SS amps do not behave similarly, so maybe here is a more relevent method for studying bipolar/Mosfet and different classes of output stages, maybe even with different R+L+C circuits to (quietly) represent realistic LS loads.
Maybe an equivalent circuit for the Ariel LS would be worth testing as a LS load, though of course complex cone and air induced events will be completely missing.
Cheers ......... Graham.
Another aspect I reasoned regarding NFB SS amplifiers, is that because they match the voltage output waveform to input waveform regardless of current, the amplifier countered driver back-EMF alternations become trapped within the composite loudspeaker circuitry.
Thus with an underhung single driver like most Fostex types, all current flow errors are trapped in series with amplifier/cable/voice coil/cone impedances and cannot fail to modify on-going reproduction after the event which generated them - a phased haze where there should be clarity.
Thus an underhung widerange driver cannot fail but to 'shout' much more acutely at higher frequencies when driven by a SS amplifier - unless - it is driven via a series impedance having values similar to the transformed anode resistance of a tube amplifier via which some of the back-EMF and re-transduced mechanical and microphonic output due to cone break up and secondary air-side events can become passively dissipated externally; as with the triode plots.
Many years ago, observations here with tweeters on extended leads showed the audibility of SS class-AB amplifier exacerbated noise events was most notable because they modified higher audio frequencies (not clipping), even though they were often caused by low and mid frequency related inductions. Tube amp tweeter output always seemed just plain normal, even via AB1 with NFB (the application of which is of course limited by circuit phase changes).
Lynn Olsen's Ariel LS is known to reveal weaknesses in SS amplifiers, I believe because it crosses over to the tweeter at a higher frequency where most NFB responses are already in full quadrature.
Of course all SS amps do not behave similarly, so maybe here is a more relevent method for studying bipolar/Mosfet and different classes of output stages, maybe even with different R+L+C circuits to (quietly) represent realistic LS loads.
Maybe an equivalent circuit for the Ariel LS would be worth testing as a LS load, though of course complex cone and air induced events will be completely missing.
Cheers ......... Graham.
- Status
- Not open for further replies.