Graham Maynard said:
So you’re trying to tell me that you don't accept that performance compromises would be made if one decided on a basic topology like that of Carlos Mergulhao’s amplifier (which, for the sake of the argument, I’ll just assume is capacitive-load-stable without an inductor) instead of designing along the lines of a Halcro, which requires an output inductor?
Re: It is a provocation Graham..do not enter this game
Graham,
Though it has been destroyer X who steped in, this is addressed to both.
You like me and everyone else here are inhabiting an edifice of technology that affords us to build amplifiers and enjoy countless other benefits, which has been built on two basic tennets.
- Contrast what we cleverly theorize with the hard facts of reality thorugh honest, cold measurement.
- Expose our incredibly smart ideas to others thus finding out the embarrassing blunders that went unnoticed, we are humans after all.
Experiment and peer review, that is what set apart science in the last centuries from religion, from bumping in the dark for millenia. Not that science is all there is to be, but when trying to deal with the real world, there is only one way to do it that has proven to be workable.
Back to audio.
There are a couple of central concepts in your line or reasoning which sound perfectly acceptable at face value.
That negative feedback being an "after the fact" attempt to correct, and being not instantaneous of necessity (first cycle), then may under some circumstances be worse than nothing at all.
You also stress the voltage / current time relationships resulting from reactive loads (back EMF) and how this intrudes in the corrective performance of negative feedback.
To address the relevance of this arguments, one must first quantify the effect within the expected operating environment. A crane designed to lift 1 ton will not lift 10 tons and is not a bad crane for that, it can happily live a life of a good crane lifting up to 1 ton.
No GNF amplifier can be expected to correct instantaneously deviations, for no amplifier has infinite bandwidth. Better than first cycle, a delta input (infinitely narrow, infinitely tall pulse) in fact exposes at the amplifier's output all there is to know about it, that is, in the frequency domain, the amplitude and phase response.
No amplifier is expected to pass transparently a delta for there is no expected audio signal to be a delta. More appropriately, one can pass a delta first through a lowpass filter in such a way as to restirct the spectral components to lie within the expected frequency band where audio program resides.
Now, if the first test, that is with a delta, reveals that within the expected usable frequency range the amplifier responds with negligible amplitude and phase deviations, then the second test - with the band limited signal - will reveal the amplifier is transparent to it, that is, the waveform and spectral signature are essentially identical at input and output.
This holds no matter how the amplifier was built, it is a "black box" approach and it is measurable.
It can allways be possible to argue there are "x-factors" as Bob likes to put it, that reveal themselves only with auditioning. I am not to argue with this, for if something has been learned is dogma in one way or another only leads to failure.
But I have strong feelings anyway in this respect, for to accept there are unmeasurable factors that translate into audible differences, goes precisely against the basic tennets of science, of what I remarked in the beginig about which are the procedures that separated millenia of bumping in the dark from success in dealing with material reality.
I am much more inclined to suspect either in that we are not doing the correct, complete measurements, or that we are being mislead by our senses, or a combination of both.
We must try not to be complacent neither in an entrenched objetivist deny-it-all stand, nor in an equally obtuse what-we-listen-is-all-that-matters one.
Rodolfo
PS Apologies for the long post
destroyer X said:
Graham,
Though it has been destroyer X who steped in, this is addressed to both.
You like me and everyone else here are inhabiting an edifice of technology that affords us to build amplifiers and enjoy countless other benefits, which has been built on two basic tennets.
- Contrast what we cleverly theorize with the hard facts of reality thorugh honest, cold measurement.
- Expose our incredibly smart ideas to others thus finding out the embarrassing blunders that went unnoticed, we are humans after all.
Experiment and peer review, that is what set apart science in the last centuries from religion, from bumping in the dark for millenia. Not that science is all there is to be, but when trying to deal with the real world, there is only one way to do it that has proven to be workable.
Back to audio.
There are a couple of central concepts in your line or reasoning which sound perfectly acceptable at face value.
That negative feedback being an "after the fact" attempt to correct, and being not instantaneous of necessity (first cycle), then may under some circumstances be worse than nothing at all.
You also stress the voltage / current time relationships resulting from reactive loads (back EMF) and how this intrudes in the corrective performance of negative feedback.
To address the relevance of this arguments, one must first quantify the effect within the expected operating environment. A crane designed to lift 1 ton will not lift 10 tons and is not a bad crane for that, it can happily live a life of a good crane lifting up to 1 ton.
No GNF amplifier can be expected to correct instantaneously deviations, for no amplifier has infinite bandwidth. Better than first cycle, a delta input (infinitely narrow, infinitely tall pulse) in fact exposes at the amplifier's output all there is to know about it, that is, in the frequency domain, the amplitude and phase response.
No amplifier is expected to pass transparently a delta for there is no expected audio signal to be a delta. More appropriately, one can pass a delta first through a lowpass filter in such a way as to restirct the spectral components to lie within the expected frequency band where audio program resides.
Now, if the first test, that is with a delta, reveals that within the expected usable frequency range the amplifier responds with negligible amplitude and phase deviations, then the second test - with the band limited signal - will reveal the amplifier is transparent to it, that is, the waveform and spectral signature are essentially identical at input and output.
This holds no matter how the amplifier was built, it is a "black box" approach and it is measurable.
It can allways be possible to argue there are "x-factors" as Bob likes to put it, that reveal themselves only with auditioning. I am not to argue with this, for if something has been learned is dogma in one way or another only leads to failure.
But I have strong feelings anyway in this respect, for to accept there are unmeasurable factors that translate into audible differences, goes precisely against the basic tennets of science, of what I remarked in the beginig about which are the procedures that separated millenia of bumping in the dark from success in dealing with material reality.
I am much more inclined to suspect either in that we are not doing the correct, complete measurements, or that we are being mislead by our senses, or a combination of both.
We must try not to be complacent neither in an entrenched objetivist deny-it-all stand, nor in an equally obtuse what-we-listen-is-all-that-matters one.
Rodolfo
PS Apologies for the long post
Hi Glen,
I have not been trying to tell you anything in the hope that you will actually go and re-check for yourself, and thus have a basis for your challenges about what is, or what is not, a compromise, because it is you who has not explained to me about what you think should be the ? compromises ? I consider.
If a design by Halcro uses an output inductor then I will not give it a second look. I thought I had clarified that I am simply not interested in any design that uses a series output inductor, other than that of a few feet of LS cable.
Hi Rodolpho,
I do not have a laboratory, nor a proper workbench at home, nor even the health to travel to someone else's to work at it without suffering.
I understand what you are writing, but see the many theoretically self assured here (other than Andy_C some time back) fighting with me rather than attempting to investigate what I cannot prove to your satisfaction.
I have been barred from saying that suddenly starting waves are relevent because it would need infinite bandwidth; so you deny the validity of my voice. I have also been ridiculed for suggesting high (NFB) amplifier bandwidths.
There is a need to time delay an original music waveform and compare it with the output of an amplifier to check for real loudspeaker - amplifier interface induced artifacts, but it will need to cover the supersonic components related to leading edges and harmonics, which I will probably be told cannot be heard anyway!
Only maybe they already are, but the theoretically self assured are refusing to accept that NFB around delaying amplifiers is causing deviation in real time.
Maybe there has been an assumption here that I am writing about every NFB amplifier in existence ? What I am writing about are those which cannot operate properly unless a series output choke allows the NFB loop to operate and register excellent THD before driving the loudspeaker. Also those which need to have their modified internal open loops 'protected' by bandwidth limiting input filters which themselves are not inaudible !
Why has the NFB thread gone quiet since I illustrated variable (including negative) group delay, which cannot fail but apply to NFB amplifiers as well, and yet which can be investigated by suddenly starting (first cycle) reverse sine testing.
The higher the degeneration, the greater the NFB and the lower the THD, but alas, the lower the frequency of unavoidable internal phase shift, the greater the necessary internal compensation and the greater the susceptibility to NFB problems with lagging/leading load current alternation; hence filters and chokes.
Some designers have specific solutions.....
Cheers ......... Graham.
I have not been trying to tell you anything in the hope that you will actually go and re-check for yourself, and thus have a basis for your challenges about what is, or what is not, a compromise, because it is you who has not explained to me about what you think should be the ? compromises ? I consider.
If a design by Halcro uses an output inductor then I will not give it a second look. I thought I had clarified that I am simply not interested in any design that uses a series output inductor, other than that of a few feet of LS cable.
Hi Rodolpho,
I do not have a laboratory, nor a proper workbench at home, nor even the health to travel to someone else's to work at it without suffering.
I understand what you are writing, but see the many theoretically self assured here (other than Andy_C some time back) fighting with me rather than attempting to investigate what I cannot prove to your satisfaction.
I have been barred from saying that suddenly starting waves are relevent because it would need infinite bandwidth; so you deny the validity of my voice. I have also been ridiculed for suggesting high (NFB) amplifier bandwidths.
There is a need to time delay an original music waveform and compare it with the output of an amplifier to check for real loudspeaker - amplifier interface induced artifacts, but it will need to cover the supersonic components related to leading edges and harmonics, which I will probably be told cannot be heard anyway!
Only maybe they already are, but the theoretically self assured are refusing to accept that NFB around delaying amplifiers is causing deviation in real time.
Maybe there has been an assumption here that I am writing about every NFB amplifier in existence ? What I am writing about are those which cannot operate properly unless a series output choke allows the NFB loop to operate and register excellent THD before driving the loudspeaker. Also those which need to have their modified internal open loops 'protected' by bandwidth limiting input filters which themselves are not inaudible !
Why has the NFB thread gone quiet since I illustrated variable (including negative) group delay, which cannot fail but apply to NFB amplifiers as well, and yet which can be investigated by suddenly starting (first cycle) reverse sine testing.
The higher the degeneration, the greater the NFB and the lower the THD, but alas, the lower the frequency of unavoidable internal phase shift, the greater the necessary internal compensation and the greater the susceptibility to NFB problems with lagging/leading load current alternation; hence filters and chokes.
Some designers have specific solutions.....
Cheers ......... Graham.
Graham, I don't know the thread that criticizes you for specifying an 'infinite bandwidth' test signal, but this is nonsense. A fast rate of change is not an instantaneous rate of change. If that were so, then what about square wave testing?
Personally, I like to use a test square wave with a 1nS risetime, because it stresses the circuits, if they cannot properly handle it. That is FAR FASTER than anything that you have implied, to the best of my knowledge.
Personally, I like to use a test square wave with a 1nS risetime, because it stresses the circuits, if they cannot properly handle it. That is FAR FASTER than anything that you have implied, to the best of my knowledge.
Hi John Curl
What was the symptom of an output inductor which degrades sound?
Have you or can you quantify what the issue is?
Is it just simply the constraint on bandwidth?
In the old days it was common to see 6 uH specified. This gives about a 150 kHz bandwidth with 8 ohm load. Perhaps this is enough, when coupled with other roll-offs, such as an input RC filter, to seriously limit the bandwidth of what might be a good amp. Individually a single roll-off should not impair the performance of a good amp, but I suspect that the problem with inductors is the simple bandwidth limitation even if not apparently obvious from the time-constants involved.
Or is the audibility you refer to the partial-resonance if there is an output capacitor as well? But such resonances are above audio for typical values of components (3 uH, 10 ohm parallel; and 100 nF 10 ohm series across the output terminals)...
Or is it due to magnetic pickup or other nonlinearity due to other signals as suggested by others?
cheers
John
What was the symptom of an output inductor which degrades sound?
Have you or can you quantify what the issue is?
Is it just simply the constraint on bandwidth?
In the old days it was common to see 6 uH specified. This gives about a 150 kHz bandwidth with 8 ohm load. Perhaps this is enough, when coupled with other roll-offs, such as an input RC filter, to seriously limit the bandwidth of what might be a good amp. Individually a single roll-off should not impair the performance of a good amp, but I suspect that the problem with inductors is the simple bandwidth limitation even if not apparently obvious from the time-constants involved.
Or is the audibility you refer to the partial-resonance if there is an output capacitor as well? But such resonances are above audio for typical values of components (3 uH, 10 ohm parallel; and 100 nF 10 ohm series across the output terminals)...
Or is it due to magnetic pickup or other nonlinearity due to other signals as suggested by others?
cheers
John
Hi John,
You wrote >>
"If that were so, then what about square wave testing?" <<
I have asked exactly the same, but that does not cut.
Of course 1nS is impossibly fast and thus irrelevent for audio, but studying responses through such short time intervals does reveal how a circuit copes.
Rodolfo wrote >>
We must try not to be complacent neither in an entrenched objetivist deny-it-all stand, nor in an equally obtuse what-we-listen-is-all-that-matters one. <<
Thanks for the (not) long post ...... Graham.
You wrote >>
"If that were so, then what about square wave testing?" <<
I have asked exactly the same, but that does not cut.
Of course 1nS is impossibly fast and thus irrelevent for audio, but studying responses through such short time intervals does reveal how a circuit copes.
Rodolfo wrote >>
We must try not to be complacent neither in an entrenched objetivist deny-it-all stand, nor in an equally obtuse what-we-listen-is-all-that-matters one. <<
Thanks for the (not) long post ...... Graham.
What we listen is all that matters.
Someone need more?
Sound equipment was made for humans or for instruments?
Can you imagine humans buying equipment, not using them...power off...and reading the amplifier data sheet having brain orgasms?...yesss..yeeeeess....
Non sense is not to observe this reality.... sound is for humans...human ears...instruments are auxiliary, meters, for better communications and competition between engineers.
The winner of the Enginneers competition will sell the equipment to a Scope..if the machine had the chance to negotiate price and had interest in sound.
Sorry Rodolfo..i have understand you wrong...my point of view, and beliefs are centered in humans, my degree is from Psychology..and my aproach is subjectivist not deliroid, anti snake oil, reserved about those ones that believe in unobtanium connector tips and a true oposition related instrumentologists.
Watch my profile folks..you will see that whole life i worked with instruments..but for audio they loose some meaning in many cases..as what you measure is not exactly what you listen related values....good meterings are not necessarily good audio.
Sony Japan used to evaluate doing meeting with their staff to listen...of course they know how to use instruments and they did in advance...but the decision was taken by ears!...nobody told me that..i was there, as their employee...i made part of that and i know how those things works.
I love Doctor Self, and i use to read his book, laughing a lot about a lot of things he say there.... having my agreement and beeing funny how subjectivists can be so...so...so....you understood what i mean....but Blameless amplifier receive a shot from a double barreled shot gun..so scandalized i was with that sonics.
An excelent engineer can produce stérile sonics too....in special if the produce those things to instruments.
Good for metering that one....the blameless...go to listen to it!
John Curl supports the remotion of the output coil because disturb sonics..it is audible...why discuss things so clear as this one.
regards,
Carlos
Someone need more?
Sound equipment was made for humans or for instruments?
Can you imagine humans buying equipment, not using them...power off...and reading the amplifier data sheet having brain orgasms?...yesss..yeeeeess....
Non sense is not to observe this reality.... sound is for humans...human ears...instruments are auxiliary, meters, for better communications and competition between engineers.
The winner of the Enginneers competition will sell the equipment to a Scope..if the machine had the chance to negotiate price and had interest in sound.
Sorry Rodolfo..i have understand you wrong...my point of view, and beliefs are centered in humans, my degree is from Psychology..and my aproach is subjectivist not deliroid, anti snake oil, reserved about those ones that believe in unobtanium connector tips and a true oposition related instrumentologists.
Watch my profile folks..you will see that whole life i worked with instruments..but for audio they loose some meaning in many cases..as what you measure is not exactly what you listen related values....good meterings are not necessarily good audio.
Sony Japan used to evaluate doing meeting with their staff to listen...of course they know how to use instruments and they did in advance...but the decision was taken by ears!...nobody told me that..i was there, as their employee...i made part of that and i know how those things works.
I love Doctor Self, and i use to read his book, laughing a lot about a lot of things he say there.... having my agreement and beeing funny how subjectivists can be so...so...so....you understood what i mean....but Blameless amplifier receive a shot from a double barreled shot gun..so scandalized i was with that sonics.
An excelent engineer can produce stérile sonics too....in special if the produce those things to instruments.
Good for metering that one....the blameless...go to listen to it!
John Curl supports the remotion of the output coil because disturb sonics..it is audible...why discuss things so clear as this one.
regards,
Carlos
Rodolfo has taken a centre line by which all views and auditions might be considered.
The 'Blameless' (with the original design having a series output choke) was the amplifier I wrote a letter to WW/EW about, suggesting that Douglas Self should examine circuit capabilities via its second input - the output terminal.
This design exemplifies all of the isolated-sine investigated optimisations, and yet it is still not widely acclaimed and used for audio reproduction. I was also puzzled that the Stocchino was presented as an apparently ideal amplifier, yet this too was fitted with an electrical 'coil spring' between the loudspeaker and its NFB node.
Whilst I applaud Rodolfo's balanced outlook, I must say that like Carlos, I have tried and rejected many designs, and that I base my own decisions upon audition rather than technical or simulated measurement.
Years ago I sat at my own bench and constructed circuits where I found it increasingly harder to measure their distortion, but they did not sound any better for it, and Carlos' useage of the word 'sterile' fairly describes some of the outputs. Another description I used was 'introverted', as if the sound had lost its ability to radiate dynamically from a loudspeaker.
There is a limit to the amount of global feedback that can be applied before an amplifier's output damping characteristic becomes phase shifted, I would say commonly about 40dB.
Exceed this figure and distortion can be reduced, but NFB control of the output terminal voltage with leading loudspeaker current runs increasingly into inductive quadrature as the enclosed open loop capabilities become degraded at AF, this simultaneously degrades class-AB crossover control capabilities and increases susceptibility to NFB loop induced early stage overdrive.
Nested resistive feedback and loading of high gain topologies can improve control by reducing the overall open loop gain/NFB, thus ensuring damping coherence throughout the AF range, often without needing an output choke either.
Cheers ......... Graham.
The 'Blameless' (with the original design having a series output choke) was the amplifier I wrote a letter to WW/EW about, suggesting that Douglas Self should examine circuit capabilities via its second input - the output terminal.
This design exemplifies all of the isolated-sine investigated optimisations, and yet it is still not widely acclaimed and used for audio reproduction. I was also puzzled that the Stocchino was presented as an apparently ideal amplifier, yet this too was fitted with an electrical 'coil spring' between the loudspeaker and its NFB node.
Whilst I applaud Rodolfo's balanced outlook, I must say that like Carlos, I have tried and rejected many designs, and that I base my own decisions upon audition rather than technical or simulated measurement.
Years ago I sat at my own bench and constructed circuits where I found it increasingly harder to measure their distortion, but they did not sound any better for it, and Carlos' useage of the word 'sterile' fairly describes some of the outputs. Another description I used was 'introverted', as if the sound had lost its ability to radiate dynamically from a loudspeaker.
There is a limit to the amount of global feedback that can be applied before an amplifier's output damping characteristic becomes phase shifted, I would say commonly about 40dB.
Exceed this figure and distortion can be reduced, but NFB control of the output terminal voltage with leading loudspeaker current runs increasingly into inductive quadrature as the enclosed open loop capabilities become degraded at AF, this simultaneously degrades class-AB crossover control capabilities and increases susceptibility to NFB loop induced early stage overdrive.
Nested resistive feedback and loading of high gain topologies can improve control by reducing the overall open loop gain/NFB, thus ensuring damping coherence throughout the AF range, often without needing an output choke either.
Cheers ......... Graham.
Graham, don't worry too much about Doug Self. I had a huge correspondence, back and forth in EW, (WW then), back in the 80's over caps. Now, he has discovered 'DISTORTION' in caps. 
I finally gave up on the LTE correspondence, since it was hopeless. Then he declared himself the 'winner' by default. I didn't see your interaction with him, but I can predict what it was like. Keep on Truckin!
I finally gave up on the LTE correspondence, since it was hopeless. Then he declared himself the 'winner' by default. I didn't see your interaction with him, but I can predict what it was like. Keep on Truckin!Graham Maynard said:
Excuse me? I am not attempting to set challenges for you. I am simply asking you to substantiate your objection to a position of mine. Instead you evade with this:
Graham Maynard said:
That’s a blindly dogmatic position in my opinion, from which one is destined to learn nothing.
Cheers,
Glen
Graham Maynard said:
Graham
I may be balanced in the way of addressing the subject, but I hope it is very clear I do not endorse auditioning as the sole benchmark to judge products with. I am a firm advocate for using technology to assess performance.
To accept an amplifier with lower performance parameters is nonetheless better than another, is akin to assume the music input includes some extra dimension apart from v(t) which somehow is treated differently and this is the reason for the perceived superiority.
When I speak about measurements, I am talking about comprehensive ones, the ultimate being the null test, whereby any alterations introduced can be exposed and analyzed. The electro mechanical interface and interaction between the amplifier and moving air ultimately, must not be left out of a more comprehensive objective evaluation approach.
Auditioning for its part, is out of necessity rooted in the listener's ability and personal preferences. When I asked Mr. Curl for a description of the audible features he looks to at the time of judgment, he answered with what may be arguably a joke I took nonetheless at face value, that he compared listening to tasting, which only reinforces dependence on the observer.
Yet, from a bussiness point of view, it is playing by the rules to tout unmeasurable qualities associated with such or which design philosophy and go ahead on the face of science.
Which remainds me why is it that as far as I know there are no germanium only solid state amplifiers. "Bridging the chasm between sterile, analytic silicon and the warm, visceral sound of tubes". Someone must think seriously about this, not me of course.
Rodolfo
PS. I am old enough to have built tube and germanium transistor amplfiers, young enough to use spice.
I heard about a new germanium design being made just last night. In my bathroom, I have a Telefunken quality portable radio that is all germanium and still sounds great, over 40 years later. I have been told by others that germanium sounds better that silicon, all else being equal. How about that?
I have listened that telefunken radio you said.
And the unit sounds extremelly good.... a hard to beat sonic quality....full germanium and assembled into a metalic chassis.
I also, evaluating by this unit, perceive the germanium sounding much better than silicon., despite noisy and sensitive to heat.
I remember one of the earliest factory matched AC187 and AC188 i have used into an amplifier....another incredible sonics amplifier.
Human ears did not evolute..we still listen in the same way we use to listen in the stone age.
If Germanium sonics were nice in the past..it is still nice those days.
regards,
Carlos
.......................................................................................................
Rodolpho
Because of a bond broken inside the skull, Graham feels enormous pains to move his neck....reason why he made a break into constructions.
And related the Pspice, i am sure Graham use this constantly..he is not that old folk loving a phonograph....he is a scientist..up to date with now a days technologie and probably has more hours doing or researching electronics that your life time.
Graham may have twice your age....well...i suppose you are near 35 years old.
And if you are older..do not think you are the only gray hair guy that use to read and study.
Now a days, Graham is in front of his computer, using Spice and simulator for more than 5 hours a day...and he is doing this for many years.... not working and using a break to use Simulator...he is exclusivelly doing researches using simulators.
regards,
Carlos
And the unit sounds extremelly good.... a hard to beat sonic quality....full germanium and assembled into a metalic chassis.
I also, evaluating by this unit, perceive the germanium sounding much better than silicon., despite noisy and sensitive to heat.
I remember one of the earliest factory matched AC187 and AC188 i have used into an amplifier....another incredible sonics amplifier.
Human ears did not evolute..we still listen in the same way we use to listen in the stone age.
If Germanium sonics were nice in the past..it is still nice those days.
regards,
Carlos
.......................................................................................................
Rodolpho
Because of a bond broken inside the skull, Graham feels enormous pains to move his neck....reason why he made a break into constructions.
And related the Pspice, i am sure Graham use this constantly..he is not that old folk loving a phonograph....he is a scientist..up to date with now a days technologie and probably has more hours doing or researching electronics that your life time.
Graham may have twice your age....well...i suppose you are near 35 years old.
And if you are older..do not think you are the only gray hair guy that use to read and study.
Now a days, Graham is in front of his computer, using Spice and simulator for more than 5 hours a day...and he is doing this for many years.... not working and using a break to use Simulator...he is exclusivelly doing researches using simulators.
regards,
Carlos
- Status
- Not open for further replies.