Besides Lynn Olson, show me someone else who really understands THD...
No salesman does and I'm not sure any hi-fi journalists understand it either..
As for phase distortion being audible I'm not sure what you mean. No one will hear a time delay, but you will quickly hear it mixing in when you start adding feedback: one reason the 741 op-amp didn't do very well as an audio amp.
In most amps, high range odd order harmonics are far more audible than low order ones. High range odd harmonics may look smaller on the FFT but your ears hear them better - so they are just as loud.
THD is meaningless, I want to see the FFT, or at least a table going out the the 20th harmonic of a 1kHz test tone.
SGD stands for Spectral Growth Distortion, which is the term I use to describe the effect that Baxandall discussed. It denotes the additional higher-order harmonics (or IM products) that can result under certain conditions when negative feedback is placed around an amplifier stage or an amplifier (e.g., distortion resulting from the re-mixing phenomena).
I actually focus largely on amplifiers where the NFB is applied to the "other" side of the input LTP, as it is the most common approach. I do cover common mode distortion and ways to reduce it in some depth.
Cheers,
Bob
Thanks for the comment.
Did you notics "harshness" with the high feedback amp ?
What about the attacks: how where they better in the low feedback amp ?
What about the music itself, did you find it better sounding on the low feedback amp ?
Bear with me as I am quite curious about personal experiences with different amplifiers.
Anyone else who has done such comparative listening, please post your impressions here.
Bob, since you have a lot of experience with amplifiers, would you tell us what you hear different with a high feedback amplifier ? In comparing the sound of different amplifiers, did it ever happen to you to like less an amp even though it measured better. I'm talking about amps with THD<0.1% such that their distortions are inaudible, at least on paper. I'm not talking about amps that introduce distortions beacuse they are considered "musical", like some tube amps.
I have compared several amps.
I do not think however that you can lump together 'high feedback' amps together as a group, the effects of feedback vary dramatically between amplifiers and it is impossible to generalise.
For instance my current tube amp has a triode/pentode section with a feedback of 37dB around it, and it sounds better than most amplifiers I have heard - some with less and some with more feedback. It just depends on too many factors.
The reason feedback (and SS/transistor amps) get a bad press is because many high feedback SS amps sound awful. This however is not because of either the SS or the feedback, rather of appalling design involving throwing lots of non-linear components in there and tying it all down with lots of negative feedback. Then you get a wonderful servo, but a soulless tiring harsh sound with a compressed polite 'hi-fi'ness to it all.
These days you can walk into most hi-fi stores and listen to any SS amplifier and it will sound the same. Tube amps survive and sound much better for the simple reason that tubes are more forgiving and so not punish a poor designer so hard.
A good rule of thumb is that feedback can move the distortion for you, so if you start with lots in the first place (at open loop) you'll end up with something sounding like cr*p no matter what you do with the feedback to make it measure better. And yes - the more delay or lower the slew rate the worse it will sound too
I didn't tell you that. If there is a dominant pole then phase distortion reduces with frequency. This is not equivalent to a fixed time delay. It is a variable time delay, with the delay being highest where it does least damage - at low frequencies. That was my point; it is not a time delay, it is an integrator. It does create problems, but not in the way that you keep saying.
I am not confused, although I could have been clearer for the
benefit of the reader.
Like its inventor Harold Black and my favorite text Horowitz
and Hill, I regard negative feedback as "the process of
coupling the output back in such a way as to cancel some of
the input".
I don't see the two as the same, although some of the effects
are the same. Some aren't, for example degeneration does
not particularly alter the output impedance, nor does it suffer
the same sorts of bandwidth stability issues as loop feedback.
spice jockey makes quick sim check:
degeneration in a ce transistor amp increases and linearizes input impedance, reduces and linearizes gm
the output Z is increased over some frequency range but there is competing internal Zcb feedback holding down output Z improvement
the gm(s) shows increased phase shift with degen
degeneration in a ce transistor amp increases and linearizes input impedance, reduces and linearizes gm
the output Z is increased over some frequency range but there is competing internal Zcb feedback holding down output Z improvement
the gm(s) shows increased phase shift with degen
It's easy to get into a semantic argument over this. I think
clarity is achieved by referring to the two differently, such as
"Degeneration" or "Degenerative feedback" versus "Loop
feedback or "Negative feedback"
Of course if you want to get really picky about it, there's no
such thing as "no feedback", since real devices have a finite
Source/Emitter/Cathode resistance. Real devices also have a
finite Drain/Collector/Plate "resistance" which emulates loop
negative feedback.
Not to mention capacitances....
clarity is achieved by referring to the two differently, such as
"Degeneration" or "Degenerative feedback" versus "Loop
feedback or "Negative feedback"
Of course if you want to get really picky about it, there's no
such thing as "no feedback", since real devices have a finite
Source/Emitter/Cathode resistance. Real devices also have a
finite Drain/Collector/Plate "resistance" which emulates loop
negative feedback.
Not to mention capacitances....
exactly - "no feedback" is a marketing invention
why should people understanding feedback theory application to circuit analysis give a rhetorical "pass" to the ones coining gobbledygook such as "no feedback amplifier" and allow them to continue to push it onto the uneducated buying public as a marketing tool?
As taught in accredited university engineering courses, as used by circuit designers in Blackman’s theorem, analyzing distortion products, ect. – degeneration is local negative feedback
Clearly the marketing war has been lost in the glossy mags but why put up with this market driven distinction with no engineering justification in this forum?
the real question in amplifier design is where to apply feedback, as local, nested, multiloop, global... for what result and at what cost in attaining circuit performance goals
Cherry summed up the options to linearize a circuit as
increasing bias vs operating range (using less of the nonlinear transfer curve)
distortion cancellation as in diff pair or Class A complementary outputs
or negative feedback (in all of its forms)
why should people understanding feedback theory application to circuit analysis give a rhetorical "pass" to the ones coining gobbledygook such as "no feedback amplifier" and allow them to continue to push it onto the uneducated buying public as a marketing tool?
As taught in accredited university engineering courses, as used by circuit designers in Blackman’s theorem, analyzing distortion products, ect. – degeneration is local negative feedback
Clearly the marketing war has been lost in the glossy mags but why put up with this market driven distinction with no engineering justification in this forum?
the real question in amplifier design is where to apply feedback, as local, nested, multiloop, global... for what result and at what cost in attaining circuit performance goals
Cherry summed up the options to linearize a circuit as
increasing bias vs operating range (using less of the nonlinear transfer curve)
distortion cancellation as in diff pair or Class A complementary outputs
or negative feedback (in all of its forms)
Like its inventor Harold Black and my favorite text Horowitz
and Hill, I regard negative feedback as "the process of
coupling the output back in such a way as to cancel some of
the input".
I don't see the two as the same, although some of the effects
are the same. Some aren't, for example degeneration does
not particularly alter the output impedance, nor does it suffer
the same sorts of bandwidth stability issues as loop feedback.
Degeneration raises output impedance I think (link).
I also thought degeneration was transparent until I read this, which showed that although the phase shift/delay was not an issue the multiplication of harmonics was.
Ref post above:
So zero-feedback in marketing terms means just zero global negative feedback. It is however a big and important difference because local feedback does not have the multi-stage delay and consequent phase margin issues (and associated audible defects) of a long loop. My own preference is for local feedback only - although with some tubes you do not even need that - they are linear and quiet enough on there own, the D3a even for phono amps. In fact my current tube amp has a no feedback (not even degenerative) input driving a very high feedback driver/output tube stage which then passes on to the output transformer with no feedback. The difference between this design and the old global negative feedback loop design that was used on that amp is that now the people and instruments sound like they are in my room - not just some hi-fi sound in the speaker.
For SS I would suggest the correct way to engineer the amp is to choose some very linear very high gain devices and treat them like gold (i.e. sparingly) and run them at high voltage so you end up with a design that requires no global feedback. I also think the latest high voltage mosfets allow one to do this because tubes have another trick - for instance when you run a 1V signal through a tube sitting on 300V or so - so the transfer curve is only used in a tiny portion which again aids linearity.
While I largely agree with this there is scope for no feedback (of any kind - except the parasitic emitter/cathodic resistances in the device) designs too - like the late 1930's western electric designs for instance. To mimic them completely however you will have to travel back in time 70 years to wind back the hi-fi clock
While I largely agree with this there is scope for no feedback (of any kind - except the parasitic emitter/cathodic resistances in the device) designs too - like the late 1930's western electric designs for instance. To mimic them completely however you will have to travel back in time 70 years to wind back the hi-fi clock
We can conclude that checking a Ford T will help
improving a Mercedes SL...
Then read the article, it's like checking the Colosseum to help build a junior baseball stadium
While I largely agree with this there is scope for no feedback (of any kind - except the parasitic emitter/cathodic resistances in the device) designs too - like the late 1930's western electric designs for instance. To mimic them completely however you will have to travel back in time 70 years to wind back the hi-fi clock
And remember that at the time neg feedback wasn't known - Harry Black needed many years (7, IIRC) to get his nfb patent accepted by the patent office because they didn't believe it worked!
So the WE no-feedback wasn't a concious decision not to use fb, it's just that they didn't have any choice. Who knows how good it would have been if they had had a choice
!jd
I don't think it even means that insofar as many amplifiers
with zero global negative feedback have internal loops over
one or more stages.
I think the marketing term for zero-feedback allows for
degeneration only. Zero global negative feedback means
that there is no loop over the entire amplifier.
I read the paper you linked to: it's quite interesting but what I was most interested in is the idea that negative feedback creates a kind of "noise floor" that might sound bad to listeners.
Any more studies on that idea ? I'd especially like to know if someone tested this assertion experimentally, i.e. that this kind of effect is indeed audible and unpleasant to the listeners.
