Talking of snake oil though, I feel for those who payed $1000s to buy a system than needs totally transforming, and at a cost twice that of my Sweet Peach amp
A miniature nuclear power station inside of the plug?
PS: I can't correct my previous post. I should write "that requite higher power amplification factor", not "have".
Less need for corrective feedback is better engineering, I don't dispute that. Switching to local negative feedback when appropriate is not snake oil. Avoiding GNFB as a matter of religion, or always using GNFB as a matter of religion, is snake oil in my opinion.
In the battle between the "subjectivists" and the "engineers" I find I have some sympathies and some criticism of both camps. The subjectivists can't accept that some of them actually prefer a little distortion. The engineers can't accept that THD is not the whole story. Both camps seem to talk a mixture of sense and nonsense. If something sounds better, it doesn't necessarily mean it is better. If something measures better, it doesn't necessarily mean it is better. This makes real hi-fi quite a challenge, as we can't fully trust either our ears or our instruments! So mostly we separate into two groups, each of which trusts one of the these and largely ignores the other. However, I know there are people (including on this forum) who are trying to find better measurements which correlate with what we hear; this is the way forward. The aim should be to get a system which sounds like a real violin, not better or worse.
I have been reading "John Curl on Audio" and I have to say, this sounds a lot like his talk on PIM... Although he describes it as FM distortion, I don't think this is is true; Doppler distortion happens in air because there is a delay from one molecule to another. I don't think this delay mechanism exists in electrons. However PIM, phase modulation distortion, which may at first seem like FM distortion, is actually only that I think: phase modulation distortion. Barrie Gilbert's article here describes how it works:
http://archive.chipcenter.com/analog/c014.htm
Wavebourn mentions measuring IM sidebrand strips, well it is in a combination of tones that PIM distortion would be revealed.
Hoping I don't catch fire, albeit on good intentions,
- keantoken
http://archive.chipcenter.com/analog/c014.htm
Wavebourn mentions measuring IM sidebrand strips, well it is in a combination of tones that PIM distortion would be revealed.
Hoping I don't catch fire, albeit on good intentions,
- keantoken
Hmm, I'm not sure the term 'snake oil' is working for you here, I take it to mean some wonder product that solves everything - which is inappropriate for both cases (GNFB and no GNFB).
I think the term 'Religion' fits better what you are trying to say, although I do not see it that way. GNFB is great for a servo, I just do not think it does anything for Hi-Fi, although for general audio it provides a good way to use cheap stuff more effectively.
Hi-fi however is like steak, where audio is like processed meat or a burger. Re-processing the sound round and round feedback loops damages it more than just leaving it alone. In fact I'd say that is it's greatest triumph - GNFB allows people to use cheap components in audio. Hi-fi (and your ears) however require more, just like I prefer steak over burgers. The irony of course is people spend a great deal buying cheap components dressed up with GNFB - burgers disguised as steak. Like a Naim burger.
You can chose to ignore the evidence and take your own time to realise that GNFB removes the realism from hi-fi if you like, but for people who have already reached that conclusion (after many years of experience, experimentation and reading) the argument is pointless and was over a long time ago: GNFB will stop you achieving realistic hi-fi.
Lots of other things do too of course, but GNFB is a major one. You don't have to believe just me either, Lynn Olson has a very good article mentioning this subject Here and perhaps more directly here.
In fact you should read Lynn's whole site - you will learn more about sound and hi-fi there than most hi-fi people do in a lifetime. BTW I'm an engineer, and so is Lynn, this is not subjective.
keantoken: The electric electron level effect is not what causes the delay, it's usually the discharge of parasitic capacitors etc that does that. The idea of a feedback that is too late and a long time delay in the amp is a magnification of what happens - in practice I suspect a frequency dependent phase modulation of the error signal is what is heard. Of course the more non-linear the amp, the bigger (and more non linear) the error signal, which will be riding along on the output slightly behind and a few dB down from the main signal, gently sabotaging the realism and life of the sound. Also the more the amp is fighting to keep the output 'right', the more voltage swing happens internally which again hurts the linearity.
The main problem with no feedback (in a decent amp) is varying speaker impedances, but again the correct engineering approach is to use zobels to straighten them, rather than whack in loads of GNFB to ignore the issue.
Last edited:
I have realized that a changing phase shift as opposed to a constant one, will cause frequency modulation.
I think we need to give the fundamental special treatment where feedback is concerned. There are ways to null the fundamental from the feedback loop, so the LTP will only respond to errors. Given the OL design is done well and harmonics are well below the fundamental, this eliminates much of the input stage loading. Would this have a significant effect?
- keantoken
I think we need to give the fundamental special treatment where feedback is concerned. There are ways to null the fundamental from the feedback loop, so the LTP will only respond to errors. Given the OL design is done well and harmonics are well below the fundamental, this eliminates much of the input stage loading. Would this have a significant effect?
- keantoken
OK, remember that the poster of the original question, sdinfo, is the editor of a home theater publication.
He (or she) is likely to publish what they "learned" on this forum.
I must state at this time that there is no scholarly work in this body of "evidence" that tries to teach that all GNFB is bad for realistic sound.
Folks, these blogs and promos cited in this thread don't make a good case. There is something heard, some speculation as to what might be causing it, then some pseudo science and technobabble created to try to explain it.
Examples:
GNFB dos not "go round and round" unless you have ringing or oscillation.
There is not any distortion of an error signal involved in audio amp gnfb. The adder/subtractor is not a nonlinear component. it's usually a resistor voltage divider.
"Temporal distortion" is not described in the AES literature; but appears all over in audio blogs and promo literature. There's no precise definition, other than as applies to jitter in a digital system.
Negative feedback can't cause compression. If anything, it can linearize the gain/amplitude function.
There is a lot more; just go read these articles with a critical eye.
IMO, these blogs present no scientific evidence, no repeatable data, just more opinion with this undertone of "of course, everyone knows..." and are delivered using some charged language.
One of them even shows some hand-drawn waveforms, and follows with scope pictures of something else. If the phenomenon exists, than why can't it be physically demonstrated?
Try this: Put yourself in the position of a reviewer of papers submitted for a technical journal or conference. Which of the cited articles in this thread woud you accept? (my answer, none of them)
I'm an engineer, and my impression of this is that a whole self-reinforcing mythology has been built up around audio that defies scientific proof on the basis of "I heard it". No doubt something was heard, but IMO it hasn't been explained in any concrete way.
To restate my conclusion, I have found that GNFB can be a way to make a bad amp measure good, and it was abused a lot in both vacuum tube and SS audio amps, but many claims being made about why a GNFB amp can measure good and sound bad don't hold water.
GNFB certainly changes the sound of an amp and also changes the way it interacts with loudspeakers, but the blanket generalizations made in these blogs are IMO are technically weak and biased, and they don't help designers or purchasers or DIYers.
Of course it's harder and more expensive to build an amp that sounds good without NFB, but that is by no means a blanket condemnation of the technique!
So sdinfo is either going to write an article teaching that all gNFB is bad for sound and it's only used to make a bad amp measure good, or he could present this blog information as opinion. I think it would be more honest to present it as opinion unless some verifiable science can be found.
Cheers!
Michael
He (or she) is likely to publish what they "learned" on this forum.
I must state at this time that there is no scholarly work in this body of "evidence" that tries to teach that all GNFB is bad for realistic sound.
Folks, these blogs and promos cited in this thread don't make a good case. There is something heard, some speculation as to what might be causing it, then some pseudo science and technobabble created to try to explain it.
Examples:
GNFB dos not "go round and round" unless you have ringing or oscillation.
There is not any distortion of an error signal involved in audio amp gnfb. The adder/subtractor is not a nonlinear component. it's usually a resistor voltage divider.
"Temporal distortion" is not described in the AES literature; but appears all over in audio blogs and promo literature. There's no precise definition, other than as applies to jitter in a digital system.
Negative feedback can't cause compression. If anything, it can linearize the gain/amplitude function.
There is a lot more; just go read these articles with a critical eye.
IMO, these blogs present no scientific evidence, no repeatable data, just more opinion with this undertone of "of course, everyone knows..." and are delivered using some charged language.
One of them even shows some hand-drawn waveforms, and follows with scope pictures of something else. If the phenomenon exists, than why can't it be physically demonstrated?
Try this: Put yourself in the position of a reviewer of papers submitted for a technical journal or conference. Which of the cited articles in this thread woud you accept? (my answer, none of them)
I'm an engineer, and my impression of this is that a whole self-reinforcing mythology has been built up around audio that defies scientific proof on the basis of "I heard it". No doubt something was heard, but IMO it hasn't been explained in any concrete way.
To restate my conclusion, I have found that GNFB can be a way to make a bad amp measure good, and it was abused a lot in both vacuum tube and SS audio amps, but many claims being made about why a GNFB amp can measure good and sound bad don't hold water.
GNFB certainly changes the sound of an amp and also changes the way it interacts with loudspeakers, but the blanket generalizations made in these blogs are IMO are technically weak and biased, and they don't help designers or purchasers or DIYers.
Of course it's harder and more expensive to build an amp that sounds good without NFB, but that is by no means a blanket condemnation of the technique!
So sdinfo is either going to write an article teaching that all gNFB is bad for sound and it's only used to make a bad amp measure good, or he could present this blog information as opinion. I think it would be more honest to present it as opinion unless some verifiable science can be found.
Cheers!
Michael
Member
Joined 2009
Paid Member
There may be many good articles on GNFB but the articles cited here are IMO not good examples of science. If anyone can find factual articles on GNFB it would be good to cite those also in this thread.
Before a discussion ensues about how science has failed us, let me remind that science, or rather the scientific method, is only a tool to reduce uncertainty in establishing cause and effect relationships. Not much more really. My complaint is the articles don't apply the scientific method to successfully reduce the uncertainty in their theories. Not much more really.
I'm not saying what they claim is false, I'm saying it has not been scientifically established through anyting that qualifies IME as evidence.
I also have serious doubts about what "eveyone knows" and claims that are presented without any backup whatsoever, which one will also find in abundance in the cited articles.
Cheers,
Michael
Many home theatre systems sound dreadful with music, although they do have the advantage of having DVD sound - often much better than CD quality (clipped and massively compressed). Perhaps stuff learned here can help tip the balance of this growing segment into higher sound quality
This is obviously an area of disagreement, so it would be fair to say that some people think GNFB is a great idea, and those who think it has subtle negative effects to the sound quality, and the more in use, the worse the effect.
Actually it must go round and round (because it's fed back, in a loop), in normal operation it will get much quieter each time round, if it gets only slightly quieter each time around you get ringing, larger and you get oscillation.
Where does the signal go after the resistor divider? Into the same non-linear amplifier you are trying to correct..?
I thought the phrase 'temporal distortion' refers to feeding back a signal that is older than the new signal going in. This will never occur with an amplifier with zero delay of course, but I've never found one of those! Perhaps it's not a well defined phrase, I'm open to suggestions.
I'm not sure which article said this - can you let me know? Certainly I can't see how it can compress anything! Remember too that we are talking of global negative feedback, not any negative feedback - i.e. the more stages the feedback is applied to, the more delay and the worse the effect (in my and others opinion).
Certainly an emotive subject this one, like in fact most other aspects of hi-fi
. Lets all remember it could be worse, we could be arguing about cables..
And another analysis here, in fact I'll quote a bit as maybe this link is not working properly for some:
Lynn Olson said:
It may seem unscientific to some but to me it's pretty clear cut, now to find that article in Glass Audio!
So are you saying we now have 2 votes for pure math and 2 for snake oil, pseudo science and technobabble?
Didn't know about the resistors tempco but that makes sense - carbon only resistors from now on in that usage (which for my amps is very rare of course because GNFB is the work of the devil
).. I'll have to think about these tempcos.Does that also mean an anode resistor will also vary with temperature - or is the thermal mass too large? Mind you - speaker wadding behaves in a similar temperature related way, making the box appear bigger.
Why don't use a science to study snake oil effects?
If smart sharp thinking people say they hear something, may be they really hear, even if they are not fluent with scientific models used by those who design audio equipment? What if designers would try to discover what do they hear, instead of dismissing what does not fit into their scientific models?
If smart sharp thinking people say they hear something, may be they really hear, even if they are not fluent with scientific models used by those who design audio equipment? What if designers would try to discover what do they hear, instead of dismissing what does not fit into their scientific models?
I did a lot, actually, and discovered some structure of magic. And still keep doing...
I never saw anything wrong with Barrie Gilbert's article: As the operating point changes, so does the OLG and phase response. This causes dynamically changing phase shift, causing lower frequency tones to FM modulate higher frequency ones.
The device comparing must be an active one, if you want to have any gain. And all known active devices distort. As they distort, their speed and gain changes, which changes all the characteristics of the feedback loop.
- keantoken
The device comparing must be an active one, if you want to have any gain. And all known active devices distort. As they distort, their speed and gain changes, which changes all the characteristics of the feedback loop.
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.