Simulation results
Johan
What is your simulation software?
What is the power amp circuit that you use in your simulation (or is it an ideal perfect amp gain block, if that can exist in real life!)?
What is not clear to me is that a speaker load is not entirely resistive. You have the crossover circuits and the speaker cones coil, so since all of these affect the amp waveform being delivered then an adaptation circuit (L, Rl, Cz, Rz) that tries to "correct" this damping should be an important factor, no?
Sorry if I might appear a little lost after all these posts but I thought that there was a generally accepted procedure for the design of the amp output filter.
Johan
What is your simulation software?
What is the power amp circuit that you use in your simulation (or is it an ideal perfect amp gain block, if that can exist in real life!)?
What is not clear to me is that a speaker load is not entirely resistive. You have the crossover circuits and the speaker cones coil, so since all of these affect the amp waveform being delivered then an adaptation circuit (L, Rl, Cz, Rz) that tries to "correct" this damping should be an important factor, no?
Sorry if I might appear a little lost after all these posts but I thought that there was a generally accepted procedure for the design of the amp output filter.
Hi,
I think that the general acceptance to set Rl=0r0 i.e no inductor, Cz=100nF and Rz=10r, shows that there is no agreed procedure.
Dr Cherry did a short paper looking at the Thiel network published in Electronics World (mid 90s?) that examined a range of values between the two limiting conditions proposed by Thiel.
It could all be converted to formulae and out popped the network values.
I have compared Cherry's computed values to a number of Thiel networks hung on the end of power amps and they rarely get close never mind agree.
The clever bit is, each of the designers believe their circuit works.
I think that the general acceptance to set Rl=0r0 i.e no inductor, Cz=100nF and Rz=10r, shows that there is no agreed procedure.
Dr Cherry did a short paper looking at the Thiel network published in Electronics World (mid 90s?) that examined a range of values between the two limiting conditions proposed by Thiel.
It could all be converted to formulae and out popped the network values.
I have compared Cherry's computed values to a number of Thiel networks hung on the end of power amps and they rarely get close never mind agree.
The clever bit is, each of the designers believe their circuit works.
Yes, sorry Fab, I have been doing you an injustice - shows what can happen if one tries to correct things at 3 a.m. So let me be more precise:
The error was that all the values are in ohms and not K.ohm; it should be 10 ohm at 10 KHz, 13.7 ohm at 20 KHz and 19 ohm at 50 KHz. I.e. the tweeter behaves increasingly as a lossy inductor at the frequencies in question.
I am using a general simulation programme called P.Spice. The tweeter equivalent circuit consists of several Ls, Cs and Rs, such that the actual measured impedance with frequency is closely approximated. I cannot post this circuit at present but will try to do so soon, sould you be interested.
I would repeat that it should be realised that the serie L (bypassed by an R to keep it from ringing) is there ONLY to keep mainly cable capacitance away from the amplifier output should the latter have a problem handling it. So also does the Zobel provide load only in addition to the rising impedance of the tweeter, ideally above 20 KHz for amplifiers which do not like working into high load resistors. These measures should not be seen as part of loudspeaker design proper although the Zobel is in a sense that, but at supersonic frequencies only. The damping and impedance normalising of loudspeakers is a task for the loudspeaker designer and can involve some quite complex circuitry. You must keep in mind that a loudspeaker impedance is not just a function of the voice coil inductance, but also includes cabinet effects which reflect as equivalent capacitive/inductive/resistive impedances. It is the total of this that is measured in an impedance vs. frequency graph. (Although not the subject of this thread, I might mention as an example, that in order to provide a reasonable load of between 6 and 10 ohms and phase angle deviating by no more than +/- 20 degrees over the audio range, I had to use a total of 4 x Cs, 3 x Ls and 3 resistors in the cross-over/conjugating network between the above tweeter and a woofer - compare that to the simple 4 component cross-overs mostly sold over-the-counter. One can find a real horror show plotting some loudspeaker systems Z/f-wise.)
In the previously mentioned illustration I used as load the "naked" tweeter only to keep the example simple - it is at least a step more realistic than just an 8 ohm resistor. If it is kept in mind that the L.R and Zobel measures are more or less tailored for a specific amplifier's stability, it follows that different designs would require different values for optimum working. This, Andrew T, would be the main reason for the different values encountered, although I rather suspect a liberal quantity of thumb-sucking in certain instances, with all due respect. In that sense there is no generally "best" set of values. In my own designs I have used very few Zobels because the amps were happy with no (i.e. infinite) load, but I do include an L of some 2,5 uH bypassed, to keep funny cable capacitances "at a distance".
Hope the situation is now clearer.
Johan Potgieter said:
The error was that all the values are in ohms and not K.ohm; it should be 10 ohm at 10 KHz, 13.7 ohm at 20 KHz and 19 ohm at 50 KHz. I.e. the tweeter behaves increasingly as a lossy inductor at the frequencies in question.
I am using a general simulation programme called P.Spice. The tweeter equivalent circuit consists of several Ls, Cs and Rs, such that the actual measured impedance with frequency is closely approximated. I cannot post this circuit at present but will try to do so soon, sould you be interested.
I would repeat that it should be realised that the serie L (bypassed by an R to keep it from ringing) is there ONLY to keep mainly cable capacitance away from the amplifier output should the latter have a problem handling it. So also does the Zobel provide load only in addition to the rising impedance of the tweeter, ideally above 20 KHz for amplifiers which do not like working into high load resistors. These measures should not be seen as part of loudspeaker design proper although the Zobel is in a sense that, but at supersonic frequencies only. The damping and impedance normalising of loudspeakers is a task for the loudspeaker designer and can involve some quite complex circuitry. You must keep in mind that a loudspeaker impedance is not just a function of the voice coil inductance, but also includes cabinet effects which reflect as equivalent capacitive/inductive/resistive impedances. It is the total of this that is measured in an impedance vs. frequency graph. (Although not the subject of this thread, I might mention as an example, that in order to provide a reasonable load of between 6 and 10 ohms and phase angle deviating by no more than +/- 20 degrees over the audio range, I had to use a total of 4 x Cs, 3 x Ls and 3 resistors in the cross-over/conjugating network between the above tweeter and a woofer - compare that to the simple 4 component cross-overs mostly sold over-the-counter. One can find a real horror show plotting some loudspeaker systems Z/f-wise.)
In the previously mentioned illustration I used as load the "naked" tweeter only to keep the example simple - it is at least a step more realistic than just an 8 ohm resistor. If it is kept in mind that the L.R and Zobel measures are more or less tailored for a specific amplifier's stability, it follows that different designs would require different values for optimum working. This, Andrew T, would be the main reason for the different values encountered, although I rather suspect a liberal quantity of thumb-sucking in certain instances, with all due respect. In that sense there is no generally "best" set of values. In my own designs I have used very few Zobels because the amps were happy with no (i.e. infinite) load, but I do include an L of some 2,5 uH bypassed, to keep funny cable capacitances "at a distance".
Hope the situation is now clearer.
Differences of opinions.
Hi Fab,
I don't wish to engage in a protracted disagreement here with anyone, but as you asked about "audible" effects relating to these 'artifices' used at the outputs of power amps, for your sake I must mention the following.
If only life was as simple as Johan seems to believe, it would make life a lot easier for many audio designers and manufacturers. Bear in mind that, for example, some well-known and most-likely adverse effects (phase and amplitude) can and do commence at a decade lower than the upper -3dB quoted figures for a power amp. Accordingly, if the -3dB upper limit with any amp is at say 100kHz (not unusual), then these adverse effects can already commence at 10kHz, which is well within the accepted audio range.
Theoretically, it could be said that *ideally* one needs to ensure that anything tacked onto the output of any amp by way of a low-pass filter, and to be *absolutely* certain "there can be no audible effect from these networks" it should not have a -3dB roll-off any lower than say 200kHz, or possibly even greater.
However, I am not claiming this to be sacrosanct as much as illustrating a simple point which sometimes appears to be overlooked by commentators.
What also matters, here, is that the entire 'wanted' audio signal passes through the series elements used in these filters, and it is also affected by the shunt elements to a lesser extent, albeit still audibly so.
Postulations, theories, or simulations, simply do not expose the entire story here with regard to the 'sonic' effects which will result from the components used in any filters, unfortunately.
Remember, science 'proves' conclusively that bees cannot fly, but maybe no one has explained this to all the bees I have ever seen in flight!
The late JLH, who was referred to before, and who was as set in his views as Johan now appears to be (so was I, incidentally, until maybe 35 yrs. ago!) was never one to accept any magic/snake-oil/smoke & mirrors, or whatever else, and was responsible for designing some of the finest audio circuits still in use to this day. As a consultant, he advised numerous UK audio companies throughout his working lifetime, and, as it happened, I was lucky enough to count him as one my personal friends (and electronics mentors) for very many years.
Approximately 25 (?) yrs. ago, I demonstrated some of these 'component' effects (I already mentioned in this thread before) to JLH, when he was on one of his many visits to listen to my audio 'progress'.
Whilst the concept was initially extremely 'uncomfortable' to JLH (and if I am to tell the truth, this must be said) he subsequently incorporated several changes in component choices in his own designs, directly as a result of what I had demonstrated to him.
I quote from his article "Audio Design" in ETI, July 1984.
" It has been demonstrated to me in respect of an earlier design of mine, that the *component types* employed can have a *considerable effect on audio quality*. In particular, the capacitor employed in the NFB loop (C7) is a very sensitive component, where *a considerable improvement*in sound quality-not readily measured instrumentally- can be gained...." etc.
In this case, C7 referred to here was not even in the direct signal path!
You might ask yourself why, if I had either 'cheated' him somehow, or was simply imagining these effects, JLH would take up this 'revised' stance, which for him was an absolute 'U' turn, and subsequently openly comment on it some several years after my demonstration to him.
Don't you think that an eminent and well-published audio designer would have carried out his own trials to satisfy himself here, long before risking the wrath of everyone who read his articles, because certainly in those days and from such a respected person, this was audio heresy!
Occasionally, I regret that I did not make the effort to give similar demonstrations to say Peter Walker, or maybe Peter Baxandall, but at that time I didn't feel any need to 'sell' something of which I had (quite accidentally) become aware, and I have since missed my chance.
Regrettably, with some less-experienced engineers, certain theoretically-inspired and very bigoted views still remain, but if you really are interested in obtaining the best sonic results from your labours, I urge you keep an open mind, and simply try this out for yourself. Where you are obliged to use certain components, anyway, try some alternatives of nominally similar values (or do what I have done and actually match them by measuring to fractions of a percent, to avoid any unwiitting self-deceit here) and just see what happens.
I know you may not wish to spend this time, but this is the only way to verify this matter for yourself, unless you are prepared to take the word of others who have actually experienced these differences, as opposed to those theoreticians and/or simulators who continually claim that it cannot be so.
What I have said you should find to be true, even if it is not intuitive, and the effects (generally) still cannot be measured as far as I am aware. Maybe one day we will find a way of doing this, but our existing knowledge/measuring-equipment doesn't make this prospect easy.
Regards,
Hi Fab,
I don't wish to engage in a protracted disagreement here with anyone, but as you asked about "audible" effects relating to these 'artifices' used at the outputs of power amps, for your sake I must mention the following.
If only life was as simple as Johan seems to believe, it would make life a lot easier for many audio designers and manufacturers. Bear in mind that, for example, some well-known and most-likely adverse effects (phase and amplitude) can and do commence at a decade lower than the upper -3dB quoted figures for a power amp. Accordingly, if the -3dB upper limit with any amp is at say 100kHz (not unusual), then these adverse effects can already commence at 10kHz, which is well within the accepted audio range.
Theoretically, it could be said that *ideally* one needs to ensure that anything tacked onto the output of any amp by way of a low-pass filter, and to be *absolutely* certain "there can be no audible effect from these networks" it should not have a -3dB roll-off any lower than say 200kHz, or possibly even greater.
However, I am not claiming this to be sacrosanct as much as illustrating a simple point which sometimes appears to be overlooked by commentators.
What also matters, here, is that the entire 'wanted' audio signal passes through the series elements used in these filters, and it is also affected by the shunt elements to a lesser extent, albeit still audibly so.
Postulations, theories, or simulations, simply do not expose the entire story here with regard to the 'sonic' effects which will result from the components used in any filters, unfortunately.
Remember, science 'proves' conclusively that bees cannot fly, but maybe no one has explained this to all the bees I have ever seen in flight!
The late JLH, who was referred to before, and who was as set in his views as Johan now appears to be (so was I, incidentally, until maybe 35 yrs. ago!) was never one to accept any magic/snake-oil/smoke & mirrors, or whatever else, and was responsible for designing some of the finest audio circuits still in use to this day. As a consultant, he advised numerous UK audio companies throughout his working lifetime, and, as it happened, I was lucky enough to count him as one my personal friends (and electronics mentors) for very many years.
Approximately 25 (?) yrs. ago, I demonstrated some of these 'component' effects (I already mentioned in this thread before) to JLH, when he was on one of his many visits to listen to my audio 'progress'.
Whilst the concept was initially extremely 'uncomfortable' to JLH (and if I am to tell the truth, this must be said) he subsequently incorporated several changes in component choices in his own designs, directly as a result of what I had demonstrated to him.
I quote from his article "Audio Design" in ETI, July 1984.
" It has been demonstrated to me in respect of an earlier design of mine, that the *component types* employed can have a *considerable effect on audio quality*. In particular, the capacitor employed in the NFB loop (C7) is a very sensitive component, where *a considerable improvement*in sound quality-not readily measured instrumentally- can be gained...." etc.
In this case, C7 referred to here was not even in the direct signal path!
You might ask yourself why, if I had either 'cheated' him somehow, or was simply imagining these effects, JLH would take up this 'revised' stance, which for him was an absolute 'U' turn, and subsequently openly comment on it some several years after my demonstration to him.
Don't you think that an eminent and well-published audio designer would have carried out his own trials to satisfy himself here, long before risking the wrath of everyone who read his articles, because certainly in those days and from such a respected person, this was audio heresy!
Occasionally, I regret that I did not make the effort to give similar demonstrations to say Peter Walker, or maybe Peter Baxandall, but at that time I didn't feel any need to 'sell' something of which I had (quite accidentally) become aware, and I have since missed my chance.
Regrettably, with some less-experienced engineers, certain theoretically-inspired and very bigoted views still remain, but if you really are interested in obtaining the best sonic results from your labours, I urge you keep an open mind, and simply try this out for yourself. Where you are obliged to use certain components, anyway, try some alternatives of nominally similar values (or do what I have done and actually match them by measuring to fractions of a percent, to avoid any unwiitting self-deceit here) and just see what happens.
I know you may not wish to spend this time, but this is the only way to verify this matter for yourself, unless you are prepared to take the word of others who have actually experienced these differences, as opposed to those theoreticians and/or simulators who continually claim that it cannot be so.
What I have said you should find to be true, even if it is not intuitive, and the effects (generally) still cannot be measured as far as I am aware. Maybe one day we will find a way of doing this, but our existing knowledge/measuring-equipment doesn't make this prospect easy.
Regards,
Fab,
The easiest answer to the previous (almost) diatribe is to go and listen, by all means, Just make sure that the listening test is devoid of any inadvertant possible bias.
To accentuate, this is the best single piece of advice I can give to anybody: DO LISTEN ... and do not buy or make a change unless YOU can hear a definite unambiguous difference.
It has been demonstrated to death that "Sonic effects" (sic) are subjective, i.e. they are there but not always repeatable and seldom indicative of what measures to take on how to overcome them. Relevant articles in many magazines over the decades, notably the Journal of the Audio Engineering Society, abound. (I am tempted to mention just 2 of the better ones: "Science vs Subjectivism in Audio Engineering" in Electronics and Wireless World some 18 years ago, and "Why I buy my Speaker Cable at Woolworths" in Electronics World some 5 years ago, both by Douglas Self - I imagine I do not need to explain who he is.)
I do not dispute Bobken's experience with "C7", I have often found such effects myself. The difference seems to be that in all cases the discrepancy could eventually be defined by proper investigative procedures with the right measuring instruments. Admittedly I was blessed to have worked in a laboratory where such facilities were available.
But that was 25 years ago. Recent analysis of e.g. capacitors in a series of articles by Bateman defined characteristics as extensive as any may desire, and is far from theory and postulation. So are the capabilities of modern analysers. Simulation shows the way there, analysis the proof.
It is my principle not to indulge in response to discussion veering to the personal, but I can hardly be blamed that in Bobken's 3rd last paragraph (this time I did count correctly!) I detect a certain allusion to myself - I have not noticed other engineers contributing here lately. So may I (against my desire and in all humility) point out that regarding experience, I have been a professional electronic engineer for the past 50+ years, much of it in audio. I designed and built many practical circuits, etc. Oh, and just for the record: I too have corresponded and exchanged ideas with the late JLH, as well as Dr Walker, Peter Baxandall, and others - let us leave the list at that....as said, just for the record.
So with respect, Bobby, I am getting a little tired of your personal innuendos on this forum regarding me and others; I do not notice the designation "Moderator" under your name. You have made your point (repeatedly) and I now mine (although many will notice that I am merely the messenger of wider audio practice). Let us move on. I shall not respond to further condescending personal overtones from you; I should not have done so now.
Apology and regards to all.
The easiest answer to the previous (almost) diatribe is to go and listen, by all means, Just make sure that the listening test is devoid of any inadvertant possible bias.
To accentuate, this is the best single piece of advice I can give to anybody: DO LISTEN ... and do not buy or make a change unless YOU can hear a definite unambiguous difference.
It has been demonstrated to death that "Sonic effects" (sic) are subjective, i.e. they are there but not always repeatable and seldom indicative of what measures to take on how to overcome them. Relevant articles in many magazines over the decades, notably the Journal of the Audio Engineering Society, abound. (I am tempted to mention just 2 of the better ones: "Science vs Subjectivism in Audio Engineering" in Electronics and Wireless World some 18 years ago, and "Why I buy my Speaker Cable at Woolworths" in Electronics World some 5 years ago, both by Douglas Self - I imagine I do not need to explain who he is.)
I do not dispute Bobken's experience with "C7", I have often found such effects myself. The difference seems to be that in all cases the discrepancy could eventually be defined by proper investigative procedures with the right measuring instruments. Admittedly I was blessed to have worked in a laboratory where such facilities were available.
But that was 25 years ago. Recent analysis of e.g. capacitors in a series of articles by Bateman defined characteristics as extensive as any may desire, and is far from theory and postulation. So are the capabilities of modern analysers. Simulation shows the way there, analysis the proof.
It is my principle not to indulge in response to discussion veering to the personal, but I can hardly be blamed that in Bobken's 3rd last paragraph (this time I did count correctly!) I detect a certain allusion to myself - I have not noticed other engineers contributing here lately. So may I (against my desire and in all humility) point out that regarding experience, I have been a professional electronic engineer for the past 50+ years, much of it in audio. I designed and built many practical circuits, etc. Oh, and just for the record: I too have corresponded and exchanged ideas with the late JLH, as well as Dr Walker, Peter Baxandall, and others - let us leave the list at that....as said, just for the record.
So with respect, Bobby, I am getting a little tired of your personal innuendos on this forum regarding me and others; I do not notice the designation "Moderator" under your name. You have made your point (repeatedly) and I now mine (although many will notice that I am merely the messenger of wider audio practice). Let us move on. I shall not respond to further condescending personal overtones from you; I should not have done so now.
Apology and regards to all.
Hi Johan,
I am very sorry to see that my comments have caused you so much apparent anxiety here, but I think this is quite unfounded if you read, properly, what I have said.
When I read your unequivocal "suffice to indicate that there *can* be no audible effects from these networks" ("can" being used here rather than perhaps "should") I saw little alternative but to comment as I did, knowing that this simplistic remark of yours is inadequate, and possibly misleading to Fab or others.
With respect, I did not criticise you in any way personally, but merely made it clear to Fab that *your comments/beliefs* did not cover the issue adequately.
Unlike yourself, I did not resort to any silly remarks about "Moderators", (what?), nor use any soppy 'nick-names' (or maybe this was an intended term of endearment?) when I simply pointed out that I didn't agree with all that you have said, and for the reasons which I politely described. There have been no "innuendos" nor "overtones", as you put it, on my part, and I have been completely open with all comments I have made here.
The only part I have not seen 'eye-to-eye' with you is over whether your *measurements and simulations* are sufficient to expose the entire story here, and I happen to know they are not.
I have not attempted to cast any doubt on the tests you carried out, themselves, their execution, nor their results, and neither have I ever suggested that they are not worthwhile. Regrettably, I don't see the same consideration being always shown in return.
The first sign of anything 'untoward' in this thread between us was when you quite wrongly accused me of 'emphasis-shifting' by bringing the matter of audible effects into the discussion, and quite rightly I pointed out that this was totally unfounded, simply because the thread-starter had specifically asked about this. I duly suggested you should take care over reading what has been posted by others here, but from this latest post of yours, you appear to be reluctant to just do this. It's all in the history above, and I am very happy for Fab to make his own judgements on the merits of our various comments, together with how useful in practice they are to him.
I regret that we don't agree on this matter, but there is no need to fall out over it, and I can only say that I am sad you haven't said at any time that you have actually carried out any such tests, and that you personally have found *no differences*, or whatever else, here.
I could understand your stance more, if that had been the case.
When I try out some previously uknown wine for the first time, I don't go to an analytical chemist and ask him to measure its various ingredients etc., or perhaps to what level the wine has become oxidised during keeping. Instead, because it is the taste & smell which matters (to most!) here, I simply try it for myself and I then know if it tastes/smells good, or if it is 'corked', or whatever. SY, the Mod. would be the person to confirm this, but I guess that most (conventional grape-based) wines will be composed of reasonably similar basic ingredients, but as with (some) components, I have found that there are many subtle and different 'flavours' involved.
Probably at aged 20 I couldn't tell one wine from another (except maybe red from white or rose), but I diligently applied myself to wine-tasting (someone's got to do this dirty job!) and, given many years of practice, I am now doing a lot better with this. It took me about 20yrs of 'practice/experience' before I realised I could reliably identify the differences when 'listening' to (many) electronic components, and the initiation into this particular sphere was entirely by accident (unlike wine-drinking!).
On a couple of minor points, "C7" was not the only component which JLH referred to, and there were many others, too.
Regarding "laboratories", their facilities, and your correspondence and ideas-exchanges with JLH, I am a little surprised he made no mention of this to me, as almost all new thoughts and ideas he had or swapped etc., he would run past me at one time or another, in the same way he would send me his latest designs for a second opinion, 'sonically'.
However, what is even more surprising is that you don't appear to know that for many years and until he retired from full-time working to commence writing, he was employed as the head of the Research Laboratories at British Cellophane, (at one time) the largest manufacturer of plastic films in Europe. Their 'basic' products were used almost universally in this part of the world by all capacitor manufacturers, and their research facilities lacked for nothing in this respect which was relevent to measuring parameters of passive devices like caps.
Part of JLH's overall responsibility was ensuring QC of the raw materials they manufactured. This involved every test on caps ever thought of (DA, ESR, etc.), and some more which were proprietary to them, so I think it is reasonable to assume that he was very familiar with all kinds of caps, together with their attributes, from the various makers' random samples which were sent back to him for evaluation.
However, until I made an accidental, but quite significant discovery, and shared this with JLH, he had never thought to try any 'listening' trials with different caps, as this was counter to all conventional wisdom abounding in those days. Until a certain European gentleman tried to sail the wrong way around to find a safer passage (than going around your country and some others) to get to East Asia, it was conventional wisdom (except for those already living there!) that America did not exist either!
I do hope you have a good weekend, in spite of our differing opinions. Opinions don't count for much, anyway, it is actuality which matters, and I hope that Fab can find the time and inclination to carry out some appropriate trials to satisfy himself in this regard.
Regards,
I am very sorry to see that my comments have caused you so much apparent anxiety here, but I think this is quite unfounded if you read, properly, what I have said.
When I read your unequivocal "suffice to indicate that there *can* be no audible effects from these networks" ("can" being used here rather than perhaps "should") I saw little alternative but to comment as I did, knowing that this simplistic remark of yours is inadequate, and possibly misleading to Fab or others.
With respect, I did not criticise you in any way personally, but merely made it clear to Fab that *your comments/beliefs* did not cover the issue adequately.
Unlike yourself, I did not resort to any silly remarks about "Moderators", (what?), nor use any soppy 'nick-names' (or maybe this was an intended term of endearment?) when I simply pointed out that I didn't agree with all that you have said, and for the reasons which I politely described. There have been no "innuendos" nor "overtones", as you put it, on my part, and I have been completely open with all comments I have made here.
The only part I have not seen 'eye-to-eye' with you is over whether your *measurements and simulations* are sufficient to expose the entire story here, and I happen to know they are not.
I have not attempted to cast any doubt on the tests you carried out, themselves, their execution, nor their results, and neither have I ever suggested that they are not worthwhile. Regrettably, I don't see the same consideration being always shown in return.
The first sign of anything 'untoward' in this thread between us was when you quite wrongly accused me of 'emphasis-shifting' by bringing the matter of audible effects into the discussion, and quite rightly I pointed out that this was totally unfounded, simply because the thread-starter had specifically asked about this. I duly suggested you should take care over reading what has been posted by others here, but from this latest post of yours, you appear to be reluctant to just do this. It's all in the history above, and I am very happy for Fab to make his own judgements on the merits of our various comments, together with how useful in practice they are to him.
I regret that we don't agree on this matter, but there is no need to fall out over it, and I can only say that I am sad you haven't said at any time that you have actually carried out any such tests, and that you personally have found *no differences*, or whatever else, here.
I could understand your stance more, if that had been the case.
When I try out some previously uknown wine for the first time, I don't go to an analytical chemist and ask him to measure its various ingredients etc., or perhaps to what level the wine has become oxidised during keeping. Instead, because it is the taste & smell which matters (to most!) here, I simply try it for myself and I then know if it tastes/smells good, or if it is 'corked', or whatever. SY, the Mod. would be the person to confirm this, but I guess that most (conventional grape-based) wines will be composed of reasonably similar basic ingredients, but as with (some) components, I have found that there are many subtle and different 'flavours' involved.
Probably at aged 20 I couldn't tell one wine from another (except maybe red from white or rose), but I diligently applied myself to wine-tasting (someone's got to do this dirty job!) and, given many years of practice, I am now doing a lot better with this. It took me about 20yrs of 'practice/experience' before I realised I could reliably identify the differences when 'listening' to (many) electronic components, and the initiation into this particular sphere was entirely by accident (unlike wine-drinking!).
On a couple of minor points, "C7" was not the only component which JLH referred to, and there were many others, too.
Regarding "laboratories", their facilities, and your correspondence and ideas-exchanges with JLH, I am a little surprised he made no mention of this to me, as almost all new thoughts and ideas he had or swapped etc., he would run past me at one time or another, in the same way he would send me his latest designs for a second opinion, 'sonically'.
However, what is even more surprising is that you don't appear to know that for many years and until he retired from full-time working to commence writing, he was employed as the head of the Research Laboratories at British Cellophane, (at one time) the largest manufacturer of plastic films in Europe. Their 'basic' products were used almost universally in this part of the world by all capacitor manufacturers, and their research facilities lacked for nothing in this respect which was relevent to measuring parameters of passive devices like caps.
Part of JLH's overall responsibility was ensuring QC of the raw materials they manufactured. This involved every test on caps ever thought of (DA, ESR, etc.), and some more which were proprietary to them, so I think it is reasonable to assume that he was very familiar with all kinds of caps, together with their attributes, from the various makers' random samples which were sent back to him for evaluation.
However, until I made an accidental, but quite significant discovery, and shared this with JLH, he had never thought to try any 'listening' trials with different caps, as this was counter to all conventional wisdom abounding in those days. Until a certain European gentleman tried to sail the wrong way around to find a safer passage (than going around your country and some others) to get to East Asia, it was conventional wisdom (except for those already living there!) that America did not exist either!
I do hope you have a good weekend, in spite of our differing opinions. Opinions don't count for much, anyway, it is actuality which matters, and I hope that Fab can find the time and inclination to carry out some appropriate trials to satisfy himself in this regard.
Regards,
BOBKEN
---the capacitor employed in the NFB loop (C7) is a very sensitive component, where *a considerable improvement*in sound quality-not readily measured instrumentally- can be gained...." etc.
In this case, C7 referred to here was not even in the direct signal path!---
Sorry, capacitor C7 of the JLH's amp published in ETI july 84, page 46, belongs to the lower arm of the feedback circuit and must be considered in the direct path. It's in direct path to the ground and ground is the reference for the signal voltages. Baxandall remarked that the first components deserving quality in an amp are, logically, those defining the negative feedback.
I think that JLH, probably without wanting it, can be made a little repsonsible for the subjectivity in the amps evaluation : back in 1969, he compared the Williamson tube amp and his own class A on a somewhat subjective basis.
I used to be a great fan of JLH designs but slowly became quite disappointed : he had clever ideas but also made some misconceptions, sorry to say. For example, it is quite contradictory to give so much importance to above capacitor C7 and to design a complex and unuseful regulated power supply having a bad transient behaviour.
~~~~~ Forr
§§§
---the capacitor employed in the NFB loop (C7) is a very sensitive component, where *a considerable improvement*in sound quality-not readily measured instrumentally- can be gained...." etc.
In this case, C7 referred to here was not even in the direct signal path!---
Sorry, capacitor C7 of the JLH's amp published in ETI july 84, page 46, belongs to the lower arm of the feedback circuit and must be considered in the direct path. It's in direct path to the ground and ground is the reference for the signal voltages. Baxandall remarked that the first components deserving quality in an amp are, logically, those defining the negative feedback.
I think that JLH, probably without wanting it, can be made a little repsonsible for the subjectivity in the amps evaluation : back in 1969, he compared the Williamson tube amp and his own class A on a somewhat subjective basis.
I used to be a great fan of JLH designs but slowly became quite disappointed : he had clever ideas but also made some misconceptions, sorry to say. For example, it is quite contradictory to give so much importance to above capacitor C7 and to design a complex and unuseful regulated power supply having a bad transient behaviour.
~~~~~ Forr
§§§
Hi Forr,
Thank you for this comment, and I have no intention of getting involved in another protracted 'difference of opinion', here.
I know very well where C7 is in this case, and it is *not* in the "direct signal path", precisely as I said.
It is, however in the *feedback path to ground* as you correctly suggest, but its location is of no consequence, whatsoever, to this particular discussion.
It was merely mentioned as an illustration, and was taken from a published article by this designer to indicate that JLH accepted that different *choices of components* can, and do, have some effect on the sound of any amplifier. This was in response to another's implications which suggested that component choices which I had referred to earlier (albeit in another part of an amp) were of no consequence, anyway. The particular comment was "there can be no audible effects from these networks", presumably because (as I also readily agree) they *should* be outside of the accepted normal hearing range.
However, notwithstanding the fact that they are outside of the hearing range, or for whatever other reason, they can, and certainly in my experience do, affect the 'sonics' of an amplifier, however unlikely this may seem.
Interestingly, if Peter Baxandall did say what you suggest here, and I don't doubt this, if there is no effect (one way or another) from the choice of components in the feedback path, for example, why did he suggest that it is important to use suitable quality components here? Surely just like the earlier "Woolworths wires" quote, anything will do.
I have no particular wish to disagree with such an authority, but in my experience, first in importance are the direct signal-path components, and secondly (but not far behind) come the feedback components, as Baxandall appears to accept.
You may be right in your assertion that JLH was "a little responsible for the subjectivity in the amps evaluation" but I don't see how this is particularly relevant to this duscussion, either.
With respect, the point which matters when discussing *audible effects*, is whether any such 'component' effects *are actually audible* or not, and not whether any postulations/theories/simulations would indicate that they are, or are not, audible.
In this case cited and in many other locations, they were/are audible to me, to many others, and to the late JLH, which was simply what I said.
I note your comments on your personal opinion of JLH's design 'foibles', but, once again, don't quite understand how these are relevant to this issue. Nevertheless, and regardless of some other areas like the PS you mentioned here, I can see no point in permitting any other area to be any less well-engineered or thought-out than is possible.
Regards,
Thank you for this comment, and I have no intention of getting involved in another protracted 'difference of opinion', here.
I know very well where C7 is in this case, and it is *not* in the "direct signal path", precisely as I said.
It is, however in the *feedback path to ground* as you correctly suggest, but its location is of no consequence, whatsoever, to this particular discussion.
It was merely mentioned as an illustration, and was taken from a published article by this designer to indicate that JLH accepted that different *choices of components* can, and do, have some effect on the sound of any amplifier. This was in response to another's implications which suggested that component choices which I had referred to earlier (albeit in another part of an amp) were of no consequence, anyway. The particular comment was "there can be no audible effects from these networks", presumably because (as I also readily agree) they *should* be outside of the accepted normal hearing range.
However, notwithstanding the fact that they are outside of the hearing range, or for whatever other reason, they can, and certainly in my experience do, affect the 'sonics' of an amplifier, however unlikely this may seem.
Interestingly, if Peter Baxandall did say what you suggest here, and I don't doubt this, if there is no effect (one way or another) from the choice of components in the feedback path, for example, why did he suggest that it is important to use suitable quality components here? Surely just like the earlier "Woolworths wires" quote, anything will do.
I have no particular wish to disagree with such an authority, but in my experience, first in importance are the direct signal-path components, and secondly (but not far behind) come the feedback components, as Baxandall appears to accept.
You may be right in your assertion that JLH was "a little responsible for the subjectivity in the amps evaluation" but I don't see how this is particularly relevant to this duscussion, either.
With respect, the point which matters when discussing *audible effects*, is whether any such 'component' effects *are actually audible* or not, and not whether any postulations/theories/simulations would indicate that they are, or are not, audible.
In this case cited and in many other locations, they were/are audible to me, to many others, and to the late JLH, which was simply what I said.
I note your comments on your personal opinion of JLH's design 'foibles', but, once again, don't quite understand how these are relevant to this issue. Nevertheless, and regardless of some other areas like the PS you mentioned here, I can see no point in permitting any other area to be any less well-engineered or thought-out than is possible.
Regards,
HI BOBKEN
I wonder what you call "direct path".
Current goes through the whole feedback network, upper and lower arm. At least, because of dielectric effects, voltage across the capacitor has quite an importance . Any misbehaviour of this capacitor will be reflected in the feedback voltage.
Baxandall said so : any non linearity of the feedback network will be reflected in the fedback voltage and then at the output of the amp. He only gave one example of a misbehaving component.
Incidentally, the greatest subjective difference I heard from an amp (a JLH's 1978 Hi-Fi News design ) was when changing the power rating of the feedback resistors from 1/4 W to minature 1/8 W. Sound became very flat, without life. I then changed them to 1 W and I thought that sound was better than with the 1/4 W.
Note that Robert Cordell in his feedforward Mosfet amp advises for 2 W (upper arm, 4220 Ohm) and 1 W resistors (lower arm, 215 Ohm) in the feedback network. He wrote : "Current flow and dissipation is not insignificant (100 mW in the 4220 Ohm at 50-Watt operating level). To avoid thermally-induced distorsion, these resistors should be oversized."
There is no mystery when components become audible.
~~~~~~ Forr
§§§
I wonder what you call "direct path".
Current goes through the whole feedback network, upper and lower arm. At least, because of dielectric effects, voltage across the capacitor has quite an importance . Any misbehaviour of this capacitor will be reflected in the feedback voltage.
Baxandall said so : any non linearity of the feedback network will be reflected in the fedback voltage and then at the output of the amp. He only gave one example of a misbehaving component.
Incidentally, the greatest subjective difference I heard from an amp (a JLH's 1978 Hi-Fi News design ) was when changing the power rating of the feedback resistors from 1/4 W to minature 1/8 W. Sound became very flat, without life. I then changed them to 1 W and I thought that sound was better than with the 1/4 W.
Note that Robert Cordell in his feedforward Mosfet amp advises for 2 W (upper arm, 4220 Ohm) and 1 W resistors (lower arm, 215 Ohm) in the feedback network. He wrote : "Current flow and dissipation is not insignificant (100 mW in the 4220 Ohm at 50-Watt operating level). To avoid thermally-induced distorsion, these resistors should be oversized."
There is no mystery when components become audible.
~~~~~~ Forr
§§§
Forr,
We are somewhat off-thread now but I hope it will be allowed to emphasize that what you (and Baxandall earlier) hinted at is vitally important but sometimes neglected. The fed-back signal at the point of encountering the input must be the exact replica of the output signal. Changes here are not corrected by feedback. While this is logical when concentrating on the matter, it is often overlooked in the total picture. When distortion of the order of 0,01% or such is relevant, the scale of this aspect becomes apparent.
It depends of course on the nature of the "non-linearity", and that not being the thread subject I will stop there. I also purposely referred to "encountering the input" - in designs where discreet active stages are used to combine feedback with input, even in a differential mode, increase in distortion can occur when these devices are not equal. For this reason I have on occasion used passive (resistive) mixing of these signals at the input (as per inverting mode), even though that meant suffering some noise degradation due to the finite input resistance. This is all the more true in the case of tube amplifiers (which I still occasionally dabble in) where the input is usually on a grid and feedback via a cathode circuit - but that is yet another thread!
Just to confirm especially for less experienced members, the importance of this aspect.
Regards.
We are somewhat off-thread now but I hope it will be allowed to emphasize that what you (and Baxandall earlier) hinted at is vitally important but sometimes neglected. The fed-back signal at the point of encountering the input must be the exact replica of the output signal. Changes here are not corrected by feedback. While this is logical when concentrating on the matter, it is often overlooked in the total picture. When distortion of the order of 0,01% or such is relevant, the scale of this aspect becomes apparent.
It depends of course on the nature of the "non-linearity", and that not being the thread subject I will stop there. I also purposely referred to "encountering the input" - in designs where discreet active stages are used to combine feedback with input, even in a differential mode, increase in distortion can occur when these devices are not equal. For this reason I have on occasion used passive (resistive) mixing of these signals at the input (as per inverting mode), even though that meant suffering some noise degradation due to the finite input resistance. This is all the more true in the case of tube amplifiers (which I still occasionally dabble in) where the input is usually on a grid and feedback via a cathode circuit - but that is yet another thread!
Just to confirm especially for less experienced members, the importance of this aspect.
Regards.
This is an interesting subject, because I have seen lots of poor PCB layouts where absolutely no care was taken to get a feedback signal resembling well the output signal. Instead, they take the feedback reference from whatever point that comes more handy, like from some emitter resistor in some corner of the PCB where asymmetric voltage drops across the output track are more than measurable. Note that in class B and all its derivatives only half the output current waveform flows through each set of output devices, so the designer must ensure that feedback is taken exactly from the mid point, preferably from the output connector).
Hi Forr,
For what this is worth, I originally used the term "direct signal path", in this case intending to mean the path which the signal takes 'directly' on its journey from input to output of the amp. i.e. the entire 'wanted' signal passes through this area.
As I think we both recognise, C7 in this case forms part of a high-pass network which determines the lower -3dB roll-off of the amp.
This network running from the feed-back path (connecting the amp's output back to the inverting input) together with the remaining network components R8 (and R9 // RV3) shunts a proportion of the feedback to ground, and also establishes the overal gain of the amp.
I agree with Baxandall, though, about the importance of the feedback components, as already mentioned, and they do need to be considered in much the same way as the 'direct' signal path parts here.
None of this is material to the main issue here, though, and it was only to indicate to Fab, that a 'dyed-in -the-wool' objectivist as JLH was at one time, accepted that component choices can, and do, have some effect on the sound, exactly as you appear to have done with these resistors you mention. And, as I also said, this designer was in a rather unique position with his main 'occupation' to be able to study/measure and consider from all aspects, the performances of individual components used in electronics.
I am very familiar with the 1972 HFN design (I think this must be what you refer to, not '78), which I also built, and I still have one in working order, although I don't use it except if I am feeling particularly 'nostalgic', and for a bit of fun.
Regrettably, it was far too complex with all of its technically- excellent filters/ tone controls etc., and especially the use of tantalum bead caps in various critical areas was not a good choice here, as the designer later accepted. These (then very new) tantalum bead caps I soon found to be poor 'subjectively', resulting in a 'grainy' coarse kind of effect, which I couldn't accept.
Merely substituting these tants with some ordinary almn. electrolytics, considerably improved the subjective results to my ears.
Incidentally, it was not a bad design, overall, in its day, and there were several unauthorised commercial 'rip-offs' by unscrupulous audio makers in the UK! This happened quite a lot over the years with various designs of JLH, as presently appears to be the case with Nelson Pass.
Eva and Johan have made some interesting and important points here relating to feedback and non-linearities etc.
For many months I ran my own 'derivative' of the 1984 circuit, with the input signal being taken directly to the inverting input of the amp. (Q6), as opposed to Q1. This 'sounded' to be an improvement to me, but (as always in life, and with its necessary compromises) there were some snags, which finally encouraged me to abandon this trial. Of course, this had the effect of inverting the signal at the output (which was of no practical consequence) but the consequentially much lower input impedance was less comfortable to deal with.
The most significant problem (IIRC) was that in this configuration, the output characteristics of any pre-amp had some effect on the overall frequency response of the amp, which was not acceptable to me.
After being impressed with the results I had (temporarily)achieved here, JLH took this amp back to his lab for some tests, as at that time my own test gear did not include the ability to measure THD etc., very accurately. When returning it later, he commented ".... it is superior in harmonic distortion and in handling transients.....this is because the signal and NFB are taken to the same point and this avoids errors from one limb to the other of the input long-tailed pair".
This is another example of when I initially 'heard' an improvement during empirical trials, but this was later confirmed by measurements following some suitable tests. Regrettably, so far I don't believe anyone has yet managed to measure many of the sonic effects attendant to different types of components, but maybe one day, this will come.
Apologies to Fab if this is rather straying from his original query, but it seems to be relevant to what has just been touched upon.
Regards,
For what this is worth, I originally used the term "direct signal path", in this case intending to mean the path which the signal takes 'directly' on its journey from input to output of the amp. i.e. the entire 'wanted' signal passes through this area.
As I think we both recognise, C7 in this case forms part of a high-pass network which determines the lower -3dB roll-off of the amp.
This network running from the feed-back path (connecting the amp's output back to the inverting input) together with the remaining network components R8 (and R9 // RV3) shunts a proportion of the feedback to ground, and also establishes the overal gain of the amp.
I agree with Baxandall, though, about the importance of the feedback components, as already mentioned, and they do need to be considered in much the same way as the 'direct' signal path parts here.
None of this is material to the main issue here, though, and it was only to indicate to Fab, that a 'dyed-in -the-wool' objectivist as JLH was at one time, accepted that component choices can, and do, have some effect on the sound, exactly as you appear to have done with these resistors you mention. And, as I also said, this designer was in a rather unique position with his main 'occupation' to be able to study/measure and consider from all aspects, the performances of individual components used in electronics.
I am very familiar with the 1972 HFN design (I think this must be what you refer to, not '78), which I also built, and I still have one in working order, although I don't use it except if I am feeling particularly 'nostalgic', and for a bit of fun.
Regrettably, it was far too complex with all of its technically- excellent filters/ tone controls etc., and especially the use of tantalum bead caps in various critical areas was not a good choice here, as the designer later accepted. These (then very new) tantalum bead caps I soon found to be poor 'subjectively', resulting in a 'grainy' coarse kind of effect, which I couldn't accept.
Merely substituting these tants with some ordinary almn. electrolytics, considerably improved the subjective results to my ears.
Incidentally, it was not a bad design, overall, in its day, and there were several unauthorised commercial 'rip-offs' by unscrupulous audio makers in the UK! This happened quite a lot over the years with various designs of JLH, as presently appears to be the case with Nelson Pass.
Eva and Johan have made some interesting and important points here relating to feedback and non-linearities etc.
For many months I ran my own 'derivative' of the 1984 circuit, with the input signal being taken directly to the inverting input of the amp. (Q6), as opposed to Q1. This 'sounded' to be an improvement to me, but (as always in life, and with its necessary compromises) there were some snags, which finally encouraged me to abandon this trial. Of course, this had the effect of inverting the signal at the output (which was of no practical consequence) but the consequentially much lower input impedance was less comfortable to deal with.
The most significant problem (IIRC) was that in this configuration, the output characteristics of any pre-amp had some effect on the overall frequency response of the amp, which was not acceptable to me.
After being impressed with the results I had (temporarily)achieved here, JLH took this amp back to his lab for some tests, as at that time my own test gear did not include the ability to measure THD etc., very accurately. When returning it later, he commented ".... it is superior in harmonic distortion and in handling transients.....this is because the signal and NFB are taken to the same point and this avoids errors from one limb to the other of the input long-tailed pair".
This is another example of when I initially 'heard' an improvement during empirical trials, but this was later confirmed by measurements following some suitable tests. Regrettably, so far I don't believe anyone has yet managed to measure many of the sonic effects attendant to different types of components, but maybe one day, this will come.
Apologies to Fab if this is rather straying from his original query, but it seems to be relevant to what has just been touched upon.
Regards,
Hi EVA,
---[...] poor PCB layouts [.(..]---
Subjectivists should have much more respect to the Douglas Self's writings than he receives, at least, because of his simple recommandations for the taking of the feedback point and the power supply - grounding schemes which apply to any kind of amps. I still have to see masters of subjectivism, who loved to spread upon the sound of components, to spare some time upon these points of primary importance.
Hi BOBKEN
Sorry, I made a mistake but it is not what you think. I am refering to a JLH design published in the Hi-Fi News issue of january 1980 . This is a very simple amp (+ preamp), using bipolar output triples capacitively coupled to the output and a singleton input. In the same magazine, in december 1980, JLH published an updated version using Hitachi Mosfets. I have built two bipolar examplaries of the bipolar version, they are still in use, and one mosfet. What fascinated me in these articles were the square waves on capacitive loads : no overshoot at all. I then thought it was mandatory for good sound ! D. Self explains what happens exactly. This drives us back to the subject of the thread : in most of his designs, JLH uses a 0R22 resistor instead of a small inductor, this partly explains the absence of overshoots. Quite funnily, the problem Baxandall encoutered with a passive component was with a small value sensing resistor in a (real) current feedback, high impedance output, power amplifier.
Belonging to the subjective camp at my beginnings, years of experience made me very cautious about sonic claims, albeit without rejecting them systematically. So I am quite interested by what you say about comparing amps in direct and inverting configurations. Bar the noise (unnoticeable in most conditions), the inverting configuration is always measurably better. I wonder why subjectivists do not explore this connexion which is objectively more significant than so many exotic receipes.
~~~~~~ Forr
§§§
---[...] poor PCB layouts [.(..]---
Subjectivists should have much more respect to the Douglas Self's writings than he receives, at least, because of his simple recommandations for the taking of the feedback point and the power supply - grounding schemes which apply to any kind of amps. I still have to see masters of subjectivism, who loved to spread upon the sound of components, to spare some time upon these points of primary importance.
Hi BOBKEN
Sorry, I made a mistake but it is not what you think. I am refering to a JLH design published in the Hi-Fi News issue of january 1980 . This is a very simple amp (+ preamp), using bipolar output triples capacitively coupled to the output and a singleton input. In the same magazine, in december 1980, JLH published an updated version using Hitachi Mosfets. I have built two bipolar examplaries of the bipolar version, they are still in use, and one mosfet. What fascinated me in these articles were the square waves on capacitive loads : no overshoot at all. I then thought it was mandatory for good sound ! D. Self explains what happens exactly. This drives us back to the subject of the thread : in most of his designs, JLH uses a 0R22 resistor instead of a small inductor, this partly explains the absence of overshoots. Quite funnily, the problem Baxandall encoutered with a passive component was with a small value sensing resistor in a (real) current feedback, high impedance output, power amplifier.
Belonging to the subjective camp at my beginnings, years of experience made me very cautious about sonic claims, albeit without rejecting them systematically. So I am quite interested by what you say about comparing amps in direct and inverting configurations. Bar the noise (unnoticeable in most conditions), the inverting configuration is always measurably better. I wonder why subjectivists do not explore this connexion which is objectively more significant than so many exotic receipes.
~~~~~~ Forr
§§§
Hi Forr,
I don't much care for the way D.Self espouses his doctrine; distortion is the be all etc.
However I agree with you totally that D.Self has given us a whole series of building/assembly advice that if carefully followed can and does reduce amplification errors that are avoidable. Hence his title blameless.
Read and adopt with care, you can probably improve most any amplifier.
I believe, if you ignored LTP current mirror and Miller compensation around VAS, his amps could be even better.
I don't much care for the way D.Self espouses his doctrine; distortion is the be all etc.
However I agree with you totally that D.Self has given us a whole series of building/assembly advice that if carefully followed can and does reduce amplification errors that are avoidable. Hence his title blameless.
Read and adopt with care, you can probably improve most any amplifier.
I believe, if you ignored LTP current mirror and Miller compensation around VAS, his amps could be even better.
Hi Forr,
No need for any apologies. I just assumed as you mentioned the '70s that it must be the amp I referred to as I was sure JLH had not published any design around the time you mentioned (1978).
I must be honest and say that this amp you refer to was about the only circuit of this designer which I never really had any involvement with. By that time, I had started experimenting with (Hitachi) mosfets, which you also mention, and I was therefore not really interested in bi-polar designs then, although I have nothing against bi-polars, themselves.
I was just more interested in seeing what I could get out of the newer devices, and this did present a few new challenges which were occupying me.
What you say is interesting though, and I take some pleasure in having encouraged JLH to reduce his 'component count' in some of his much later designs (not all of them having been published), for example his 're-hash' in 1996 of the much earlier and very simple Class 'A' design, going back to 1969, in WW.
I had discovered that, although technical improvements/complications seemed preferable on paper and often when being measured, there appeared to be a trade-off, sonically, with almost a law of diminishing returns here.
Frankly, this was an uncomfortable discovery, but it couldn't be ignored, and it may well relate to the fact that no component is quite perfect, and accordingly, the fewer parts used meant fewer imperfections in the chain. I cannot say for certain.
I demonstated this to JLH over a period of some years (with his encouragement, and advice) mainly with a design of his published in WW in 1982, which started out with such 'lily-gilding' as positive feedback within the neg. feedback loop etc. For whatever reason, and in spite of this apparently superior technical topology, it simply sounded better without this particular addition. It progressed along from this in every area I could experiment with, and provided the measured performance was not seriously compromised, it nearly always ended up with better sonics.
Over a few years, I tried out all kinds of other circuit variations, too, along the lines of that I mentioned before, and measured (or had measured, when I could not do this for myself) the effects, and spent month after month listening to what these actually did to the sound. Swapping things back and forth time after time, and discovering as best I could the correlation between various topologies and their resultant sounds.
In all, I have been doing this for over 35 yrs., and I am still learning all the time.
Regrettably, because of conventional wisdom and the lack of time and inclination, most people will never spend the time needed to carry out serious long-term listening tests, but nowadays most high-end manufacturers will make some effort to 'optimise' their products through some listening tests, before their products hit the shelves.
I fully agree with the comments from Douglas Self over care where the neg, feedback is taken from, and, of course, this is quite intuitive. However, when I read his articles in EW & WW some years ago, this was the first time I had seen any published reference to this simple, but very important, issue.
So many apparently insignificant issues contrive together to produce either a very good-sounding result or maybe otherwise, and attention to every last detail is vital in my experience if you are keen to get the very best 'sonic' results with any audio equipment.
Regards,
No need for any apologies. I just assumed as you mentioned the '70s that it must be the amp I referred to as I was sure JLH had not published any design around the time you mentioned (1978).
I must be honest and say that this amp you refer to was about the only circuit of this designer which I never really had any involvement with. By that time, I had started experimenting with (Hitachi) mosfets, which you also mention, and I was therefore not really interested in bi-polar designs then, although I have nothing against bi-polars, themselves.
I was just more interested in seeing what I could get out of the newer devices, and this did present a few new challenges which were occupying me.
What you say is interesting though, and I take some pleasure in having encouraged JLH to reduce his 'component count' in some of his much later designs (not all of them having been published), for example his 're-hash' in 1996 of the much earlier and very simple Class 'A' design, going back to 1969, in WW.
I had discovered that, although technical improvements/complications seemed preferable on paper and often when being measured, there appeared to be a trade-off, sonically, with almost a law of diminishing returns here.
Frankly, this was an uncomfortable discovery, but it couldn't be ignored, and it may well relate to the fact that no component is quite perfect, and accordingly, the fewer parts used meant fewer imperfections in the chain. I cannot say for certain.
I demonstated this to JLH over a period of some years (with his encouragement, and advice) mainly with a design of his published in WW in 1982, which started out with such 'lily-gilding' as positive feedback within the neg. feedback loop etc. For whatever reason, and in spite of this apparently superior technical topology, it simply sounded better without this particular addition. It progressed along from this in every area I could experiment with, and provided the measured performance was not seriously compromised, it nearly always ended up with better sonics.
Over a few years, I tried out all kinds of other circuit variations, too, along the lines of that I mentioned before, and measured (or had measured, when I could not do this for myself) the effects, and spent month after month listening to what these actually did to the sound. Swapping things back and forth time after time, and discovering as best I could the correlation between various topologies and their resultant sounds.
In all, I have been doing this for over 35 yrs., and I am still learning all the time.
Regrettably, because of conventional wisdom and the lack of time and inclination, most people will never spend the time needed to carry out serious long-term listening tests, but nowadays most high-end manufacturers will make some effort to 'optimise' their products through some listening tests, before their products hit the shelves.
I fully agree with the comments from Douglas Self over care where the neg, feedback is taken from, and, of course, this is quite intuitive. However, when I read his articles in EW & WW some years ago, this was the first time I had seen any published reference to this simple, but very important, issue.
So many apparently insignificant issues contrive together to produce either a very good-sounding result or maybe otherwise, and attention to every last detail is vital in my experience if you are keen to get the very best 'sonic' results with any audio equipment.
Regards,
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.