I see we now have a special audiophile feedback theory - "not intended to be factual"
I'm also glad of wavebourn's diagram - it leaves Thorsten without any wiggle room when he identiifes
I find Boyk and Sussman's feedback math at odds with Thorsten's claim:
as can be read from their paper
I'm also glad of wavebourn's diagram - it leaves Thorsten without any wiggle room when he identiifes
I find Boyk and Sussman's feedback math at odds with Thorsten's claim:
as can be read from their paper
Attachments
Last edited:
If the looped feedback is series derived, what does the output impedance look like? I just want to make sure before we go further that common engineering terms are commonly understood.
Source and load inmpedances are always in equations. Are we kids, Scott? That diagram was drawn to show kids major basics: feedback by current or by voltage, applied in parallel or in series with input.
Last edited:
This does rather seem to me to be totally ****-about-face. In the absence of any claims, feedback is feedback, by observation alone. Your claim that its not in the case of degeneration is special pleading.
Downloads | Get The Latest Version of EAGLE | CadSoft EAGLE |
Free for personal use.
Open source:
Main Page - KiCad EDA Suite Official Wiki
Just to name two.
@Thorsten - that emittor resistor is called since the beginning of the solid-state era feedback resistor. Curent feedback.
The input voltage on the base-emitter is reduced by the negative amount produced via voltage drop on the resistor by the amplified current. It the control theory, any deduction from the input signal (reference) of the output signal, in order to create the error signal, is called fedback. Call it "negative" if you want to point that at the sumation point will be with negative sign/phase compared with the input.
Control theory - Wikipedia, the free encyclopedia
I think you are tempted to call "feedback" only the voltage feedback.
No "denegeration" takes place inside those devices, their parameters remain the same.
The input voltage on the base-emitter is reduced by the negative amount produced via voltage drop on the resistor by the amplified current. It the control theory, any deduction from the input signal (reference) of the output signal, in order to create the error signal, is called fedback. Call it "negative" if you want to point that at the sumation point will be with negative sign/phase compared with the input.
Control theory - Wikipedia, the free encyclopedia
I think you are tempted to call "feedback" only the voltage feedback.
No "denegeration" takes place inside those devices, their parameters remain the same.
Last edited:
Sonic, you make a good point, but please try to understand what we audio designers are talking about. We have found that LOCAL feedback can sound, and 'maybe' sometimes measure better than GLOBAL LOOP feedback. We are familiar with feedback, and I certainly have been for the last 50 years, when I took my first course in electronics. We have 'been there' and 'tried that'. Only it did not work as well as we predicted it would. That is WHY we have reverted to LOCAL feedback, in many cases, and minimized loop feedback.
The primary difference is that LOCAL feedback is high bandwidth, equal at least, over the entire audio band and perhaps much more. Loop feedback is run on a slope of 6dB/octave, at least somewhere if not everywhere in the audio band. Think it through, and you might see what we have found.
The primary difference is that LOCAL feedback is high bandwidth, equal at least, over the entire audio band and perhaps much more. Loop feedback is run on a slope of 6dB/octave, at least somewhere if not everywhere in the audio band. Think it through, and you might see what we have found.
stability of the "loop" is a requirement for local feedback as well - you couldn't use degeneration around a device that had 180 degree phase shift and unity gain - we certainly can see local oscillations with a single gain device
a "classic" is the input negative resistance oscillation of a emitter follower
since the "harmonic multiplication" math applies to local feedback/degeneration too (why do “anti-feedback” audiophiles bother making such an obviously false claim that it doesn't?)
basically the argument only comes down to speed?
I can get GHz op amps much easier than discrete GHz Q
a "classic" is the input negative resistance oscillation of a emitter follower
since the "harmonic multiplication" math applies to local feedback/degeneration too (why do “anti-feedback” audiophiles bother making such an obviously false claim that it doesn't?)
basically the argument only comes down to speed?
I can get GHz op amps much easier than discrete GHz Q
where does this particular "degeneration" myth come from?
the false claim is that "degeneration" doesn't create "new" harmonics - al la Baxendal, Crowhurst, ect
degeneration, any linear feedback around a "square law" device creates an output harmonic distortion spectrum containing all orders, even and odd
an interesting case is the pure odd order case - a gain device with only a cubic nonlinear term only produces odd order distortion harmonics in its output
do you have a problem with the Boyk and Sussman I posted above for anyone to read? - Thorsten identified their source resistor local feedback as "degeneration" - as I think we all agree
then simply reading the text in the 2nd paragraph shows they found all orders of distortion harmonics in the output of their "square law" model of the fet circuit
I've posted sims, there are pages and pages of the math, plots in the Cordell Negative Feedback thread
Even Nelson Pass sees the “new” third harmonic in degenerated fet output spectrum – it only requires looking
if you "learned" something different maybe you misremembered, or are not revealing conditions, limitations, assumptions - which is it, "enquiring minds want to know"
the false claim is that "degeneration" doesn't create "new" harmonics - al la Baxendal, Crowhurst, ect
degeneration, any linear feedback around a "square law" device creates an output harmonic distortion spectrum containing all orders, even and odd
an interesting case is the pure odd order case - a gain device with only a cubic nonlinear term only produces odd order distortion harmonics in its output
do you have a problem with the Boyk and Sussman I posted above for anyone to read? - Thorsten identified their source resistor local feedback as "degeneration" - as I think we all agree
then simply reading the text in the 2nd paragraph shows they found all orders of distortion harmonics in the output of their "square law" model of the fet circuit
I've posted sims, there are pages and pages of the math, plots in the Cordell Negative Feedback thread
Even Nelson Pass sees the “new” third harmonic in degenerated fet output spectrum – it only requires looking
if you "learned" something different maybe you misremembered, or are not revealing conditions, limitations, assumptions - which is it, "enquiring minds want to know"
Last edited:
The degeneration myth seems to come from people who like degeneration but who wish to claim (for some purpose) that they don't like/use feedback. Hence all the solid-state threads by people offering 'zero feedback' designs which abound in emitter followers. I used to point out their mistake, then I laughed at them, now I ignore them.
I suppose part of the problem is that some people only think of voltage as the signal, so degeneration (which samples a proxy for the output current) looks like something different. It isn't different. You can feedback voltage, current, temperature, gas pressure or whatever else you are interested in - it all depends on what your feedback loop/servo is trying to do. The maths is the same.
You don't even need amplification to get the effect. Take an ideal square-law diode and add a resistor to ground. Input to the top of the diode, and output from the junction of the diode and resistor. You get distortion at all orders, not just second. Just algebra - the inverse of a polynomial can be an infinite series.
I suppose part of the problem is that some people only think of voltage as the signal, so degeneration (which samples a proxy for the output current) looks like something different. It isn't different. You can feedback voltage, current, temperature, gas pressure or whatever else you are interested in - it all depends on what your feedback loop/servo is trying to do. The maths is the same.
You don't even need amplification to get the effect. Take an ideal square-law diode and add a resistor to ground. Input to the top of the diode, and output from the junction of the diode and resistor. You get distortion at all orders, not just second. Just algebra - the inverse of a polynomial can be an infinite series.
sorry for the misinterpretation of your comment John, its good to know some of our "Gurus" can get (some of) the basics correct - why not help Thorsten with this?
I am still curious how competent designers can get worse measured distortion from high feedback amps under reasonable conditions for comparison to local feedback, degeneration only amps for audio frequencies
because it is also a common result presented in EE courses on feedback systems that global loop gain gives the least sensitivity/most distortion reduction compared to the same total gain implemented with only local feedback around the same gain blocks
Block diagram or Mason Signal flow graph representations also show you can "move" local feedbacks to the outer loop with appropriate scaling, compensation for the same system response with only the global feedback
these results are based in "linear" modeling but do extend to "weakly nonlinear" systems - I hope all of us are trying to "minimize nonlinearity before applying feedback"
Otala's concerns, "TIM" is easily avoided in high feedback amps by knowledgeable designers - the conditions are spelled in JAES articles and elsewhere
or are you making the usual "strawman" comparison of uA741 class op amp used for Av 10, 10 V output as the limit of what global feedback can do
I am still curious how competent designers can get worse measured distortion from high feedback amps under reasonable conditions for comparison to local feedback, degeneration only amps for audio frequencies
because it is also a common result presented in EE courses on feedback systems that global loop gain gives the least sensitivity/most distortion reduction compared to the same total gain implemented with only local feedback around the same gain blocks
Block diagram or Mason Signal flow graph representations also show you can "move" local feedbacks to the outer loop with appropriate scaling, compensation for the same system response with only the global feedback
these results are based in "linear" modeling but do extend to "weakly nonlinear" systems - I hope all of us are trying to "minimize nonlinearity before applying feedback"
Otala's concerns, "TIM" is easily avoided in high feedback amps by knowledgeable designers - the conditions are spelled in JAES articles and elsewhere
or are you making the usual "strawman" comparison of uA741 class op amp used for Av 10, 10 V output as the limit of what global feedback can do
Last edited:
This result falls out of the simplest possible statement of the problem, you could probably find someone who noticed this in 1930.
If you guys are all done ranting now, you might like to go back and read what Thorsten actually wrote. He did NOT say "degeneration isn't feedback". What he said was:
Notice the difference between the following two sentences:
A) "Carrots taste different to cabbage."
B) "Carrots are not vegetables."
- Status
- Not open for further replies.