Workhorse said:
Hi Kanwar,
what is by definition the difference between Hi-End and Hi-Fi?
Workhorse said:
Oh come on, please. Even a novice can see that this feedbackless thingy will produce a lot of thd. Normally, I run a sim before commenting on a design. In this case, however, that seems a useless exercise.
And for once and for all, there ain't no decent solid state amp without (a lot of) feedback.
Cheers,
SY said:
Glen, though I am not unsympathetic to your technical argument, the personal stuff has no place here. Please confine yourself to the subject at hand, not your opinion of Hugh. That is most unwelcome.
Okkkkaaayyyyyyy SY...... My response to Hugh probably was a little bit snide.
I'll try to be a good boy from now on
Honest.Cheers,
Glen
AKSA said:
Hi Hugh,
Theoretically, global or local, it doesn't matter, but I prefer a combination of both, that is, NDFL and global feedback.
See for example:
http://www.diyaudio.com/forums/show...=96634&perpage=10&highlight=ndfl&pagenumber=3
post #29
Well, if that is the case, such amplifier is probably unstable, i.e. suffering from a bad frequency compensation or saturating some stages. As for a sound amp, global feedback will reduce all kinds of distortion, but, admittedly, high order artefacts to a lesser extend, caused by a diminishing loop gain at higher frequencies.
But perhaps you had a different "mechanism" in mind:
There are some people who believe that a supposedly 'time delay' from a RC filter or a compensating cap compromises the global feedback. As a matter of fact, there ain't no such time delay. Let's take a RC filter for example, as soon as signal is applied to the input, the output will respond immediately, without any delay. Indeed, there is a phase shift that looks like a delay, but has nothing to do with a propagation time as can be seen with a (coax) cables or so. If it was, then feedback couldn't work at all.
Cheers,
AKSA said:
I am sorry, but this is ludicrous, as it suggests engineering decisions and valid critical evaluation of circuitry must and can only be done "by ear".
Indeed, the contrary position is the ONLY tenable one: engineering decisions in audio circuit design and critical evaluation of the same must NEVER be done "by ear".
Why?
Because, as Sam9 eloquently puts it,
http://www.diyaudio.com/forums/showthread.php?postid=1185187#post1185187
estuart said:
Actually I disagree.
I am a fan of solid state and I am also a fan of negative feedback. The latter has gotten an undeserved bad rap over the years, and some bad designs have been made by people who either do not understand negative feedback, or abuse it as a tool in trying to make a purse out of a sow's ear circuit.
That having been said, I'm sure that I and many others so inclined could design and build a very good solid state amplifier with no global negative feedback and no local negative feedback other than degeneration or shunt FB around one transistor. It might not achieve THD below 0.01%, and it might not achieve damping factor much above 20, and it might have to run warm with modest Class AAB or Class A bias, but I'm sure it would still qualify as a very decent solid state amp.
Cheers,
Bob
AKSA said:
Hi Hugh,
You are referring to re-entrant distortion caused by the application of negative feedback. If you had a rather magical open-loop amplifier that created ONLY second harmonic distortion, then when you applied negative feedback the second order distrotion components fed back would mix with the fundamental and create thirds; so the phenomenon is real.
However, it is a bit academic for two reasons: the new distortions are quite small created by this mechanism in any reasonably linear open-loop design, and they may in fact be smaller than the higher order distortions that were present in most practical open loop amplifiers in the first place. Finally, as negative feedback is increased, it decreases the second order products' amplitude, so less re-entrant IM can take place. Thus there is an amount of negative feedback where the results of this mechanism is maximized, and larger amounts of applied negative feedback above that amount reduce ALL higher-order distortions, whether from the re-entrant mechanism or from the open loop characteristic.
Finally, the caveat that the amount of negative feedback at high frequencies is less must always be taken into account AND that the amount of drive to the input stage at higher frequencies may often be larger at high frequencies, creating larger amounts of open loop nonlinearity at high frequencies than at low frequencies. We should remember that all of these things are relative, and that when we are talking about distortion with and without negative feedback applied, we should be talking about an input stage with the same open loop dynamic range and signal-handling capability in both cases.
Cheers,
Bob
Bob Cordell said:
Hi Bob,
I disagree with your 'feedback'.
Apparently you missed my point, as I mean ANY form of feedback, including the examples around one transistor you are referring to. Even an emitter/source follower should be considered as a (current) gain stage with (voltage) feedback, at least that is my perception of negative feedback.
Cheers,
BTW, did you have a look at my NDFL amp?
estuart said:
I know, we're getting too technical here. Charles Hanson has also fought the battle over feedback semantics. OK, forget the shunt feedback around one stage and just leave me with emitter degeneration. I'll still make a decent SS amplifier with it. Only a fool would not use emitter degeneration and nobody would care about the result anyway.
Cheers,
Bob
Bob Cordell said:
Another rash statement, Bob...
I do like your phrase: "no local negative feedback other than
degeneration or shunt FB around one transistor", although I
don't exclude instances of cascoding, parallel transistors or
differential pairs.
I don't think there's any point to beating the definition of
feedback any more - it's dead. The propeller-head definition
tells you that there's no such thing as no feedback whatsoever -
even the stray internal resistance and capacitance of devices
gives them feedback.
The marketplace insists on a phrase which differentiates global
feedback from local, so the words "No Feedback" is not going
away anytime soon.
Thanks Bob, Edmond,
Yes, I am referring to re-entrant distortion, which is responsible for the spray of high order artefacts arising out of low order distortions. These are certainly low level, but my understanding is that the human ear can pick high order distortion at remarkably low levels, better than 0.05% for example for the 9th harmonic of a trumpet.
Mikeks wrote:
No, Mike, I'm not suggesting that at all, but thank you for the earnest apology. You are certainly courteous today!! I am not suggesting sprinkling holy water and animal blood around a celtic cross; merely that after all the engineering is done, the circuit is debugged and the stability considerations met, we need to do listening tests to attempt to correlate what is in the schematic with what we hear. Owing to the intense subjectivity of the listening preference, however, there do need to be a sample of five or six people to verify, in carefully controlled AB, that people are hearing the same thing. Once identified, the real skill comes in lining up what is heard with what is designed; we can do the engineering with distinction, but we must not forget this machine will be used for listening to music by human beings, and while it's not possible to precisely test music in any sort of objective, meaningful test (yet!!) it does behove us to attempt to assess the sound as the market will, right or wrong.
An analogous argument applies to the automotive industry. I'm confident that the designers drive the cars exhaustively, and if 'Top Gear' is any indication, they'd better, because the subjective issues appear to dominate the choices made by the buying public.
Cheers,
Hugh
Yes, I am referring to re-entrant distortion, which is responsible for the spray of high order artefacts arising out of low order distortions. These are certainly low level, but my understanding is that the human ear can pick high order distortion at remarkably low levels, better than 0.05% for example for the 9th harmonic of a trumpet.
Mikeks wrote:
No, Mike, I'm not suggesting that at all, but thank you for the earnest apology. You are certainly courteous today!! I am not suggesting sprinkling holy water and animal blood around a celtic cross; merely that after all the engineering is done, the circuit is debugged and the stability considerations met, we need to do listening tests to attempt to correlate what is in the schematic with what we hear. Owing to the intense subjectivity of the listening preference, however, there do need to be a sample of five or six people to verify, in carefully controlled AB, that people are hearing the same thing. Once identified, the real skill comes in lining up what is heard with what is designed; we can do the engineering with distinction, but we must not forget this machine will be used for listening to music by human beings, and while it's not possible to precisely test music in any sort of objective, meaningful test (yet!!) it does behove us to attempt to assess the sound as the market will, right or wrong.
An analogous argument applies to the automotive industry. I'm confident that the designers drive the cars exhaustively, and if 'Top Gear' is any indication, they'd better, because the subjective issues appear to dominate the choices made by the buying public.
Cheers,
Hugh
AKSA said:
Hugh,
Trouble is once you depart from objective criteria, you must make a choice as to wich type of performance anomalies - "features" in IT parlance - have acceptance and in which market you may be after.
This is dangerous, and quite frankly from an engineering standpoint I should stick with accuracy. I do not agree either that there are not yet meaningful objective tests in this regard. Some posts earlier I mentioned the null test with music under working load as a quite irrefutable means to prove or disprove an amplifier is true to input.
Of course there is then the speaker, preamplifier if present, source components etc. But if we at least can be assured the ampifier is accurate, there are less simultaneously variable parameters to juggle with.
If on the other hand lead designers choose on purpose to trade accuracy for some distinctive marketing edge, be it "no feedback", "tube sound" or whatever, just fine, there will allways be niche customers for that.
Rodolfo
Hi Rodolfo,
Thanks for your post.
Ah yes, thank you, this slipped by. Good point.
Perhaps you misunderstand my intent. I suggest a two stage development here; first get the engineering right, focus on low distortion, accuracy, flat response, THEN address the subjective issues within market constraints and try to make the connection between the 'coloration' we hear in a null test as you describe and the topology/dimensioning/component choice. This is the aspect which requires deep experience of the market and the practicalities to my way of thinking, although I do not decry the engineering. It's tough to get it right and clear proof is a quiet walk through the silicon crematorium.
You know, I might come across as mystical in my approach, but I have never in my design experience seen an improvement in linearity without a commensurate improvement in sound quality. This is a clear and obvious proof that accuracy is good. But while I appreciate that audio engineers stand their ground without apology I do not decry the objective approach and do not intend to fight city hall, I merely believe there is more to it than engineering, a little art, if you will. After all, pcb layout is called artwork, this is no coincidence.....
This is recognised in other fields of engineering; bridges, autos, architecture, even chemistry. I would offer that the difference between a competent design and a great one is art and passion. Why not audio? And I have noticed that even the most hard-bitten technologists here like to describe how their amps 'sound'. This is subjective, and demonstrates we are all artists at heart.
Cheers,
Hugh
Thanks for your post.
Ah yes, thank you, this slipped by. Good point.
Perhaps you misunderstand my intent. I suggest a two stage development here; first get the engineering right, focus on low distortion, accuracy, flat response, THEN address the subjective issues within market constraints and try to make the connection between the 'coloration' we hear in a null test as you describe and the topology/dimensioning/component choice. This is the aspect which requires deep experience of the market and the practicalities to my way of thinking, although I do not decry the engineering. It's tough to get it right and clear proof is a quiet walk through the silicon crematorium.
You know, I might come across as mystical in my approach, but I have never in my design experience seen an improvement in linearity without a commensurate improvement in sound quality. This is a clear and obvious proof that accuracy is good. But while I appreciate that audio engineers stand their ground without apology I do not decry the objective approach and do not intend to fight city hall, I merely believe there is more to it than engineering, a little art, if you will. After all, pcb layout is called artwork, this is no coincidence.....
This is recognised in other fields of engineering; bridges, autos, architecture, even chemistry. I would offer that the difference between a competent design and a great one is art and passion. Why not audio? And I have noticed that even the most hard-bitten technologists here like to describe how their amps 'sound'. This is subjective, and demonstrates we are all artists at heart.
Cheers,
Hugh