Hi,
I'm confused why there should be local feedback and global feedback. If I add global feedback, can I throw out any local feedback (in each modul/stage), so that there is only 1 feedback (global feedback) in the system, taken from output end to input (right after potentiometer/volume), make it negative feedback (off course, polarity must be inverted/inverting amp)?
E.g. I have 2 amp module, 1st modul = inverting with local (its own) feedback, 2nd modul = non inverting with its own local feedback, could I change 1st module input feedback to outer end (2nd output/line out), and throw all both local feedback for simplicity? Which one is better (global + local or global only)?
Thx,
Ervin L
I'm confused why there should be local feedback and global feedback. If I add global feedback, can I throw out any local feedback (in each modul/stage), so that there is only 1 feedback (global feedback) in the system, taken from output end to input (right after potentiometer/volume), make it negative feedback (off course, polarity must be inverted/inverting amp)?
E.g. I have 2 amp module, 1st modul = inverting with local (its own) feedback, 2nd modul = non inverting with its own local feedback, could I change 1st module input feedback to outer end (2nd output/line out), and throw all both local feedback for simplicity? Which one is better (global + local or global only)?
Thx,
Ervin L
Local feedback has fewer parasitic phase shifts, helps to prevent instability.
Sometimes harder to fix error after the fact, than prevent in the first place.
Local feedback will either be direct coupled with negligible phase shift,
or with single pole of no more than 90deg phase shift. Thus considered
"unconditionally stable".
Much more difficult with 2 or three Miller effects + intentional coupling caps,
and no local feedback, to say there is no frequency with -unity gain or more,
with phase shift not greater than 120degrees.
What is inverting or non-inverting in the context of a global feedback only?
At some frequency you just won't know... Much easier to control the roll of
such things in the smaller picture of a single stage.
Even when you *think* the whole thing is direct coupled, there are many
parasitic factors beyond our control. Far too many to play guessing games
with a single global loop..
Sometimes harder to fix error after the fact, than prevent in the first place.
Local feedback will either be direct coupled with negligible phase shift,
or with single pole of no more than 90deg phase shift. Thus considered
"unconditionally stable".
Much more difficult with 2 or three Miller effects + intentional coupling caps,
and no local feedback, to say there is no frequency with -unity gain or more,
with phase shift not greater than 120degrees.
What is inverting or non-inverting in the context of a global feedback only?
At some frequency you just won't know... Much easier to control the roll of
such things in the smaller picture of a single stage.
Even when you *think* the whole thing is direct coupled, there are many
parasitic factors beyond our control. Far too many to play guessing games
with a single global loop..
Last edited:
Does it mean I should keep local feedback, and add global feedback to improve sound quality? If so, what should be global feedback ratio vs local feedback?
E.g.:
Local feedback 1 = 10 (gain = 10x)
Local feedback 2 = 1 (gain = 1x --> make it buffer stage)
I'd like to add global feedback, from output 2 to input 1. What should be the R ratio range (<10, >10, or exactly 10)?
Will this global feedback improve the sound quality (e.g. reduce noise/harmonic, add damping factor, correction etc)?
Thx,
Ervin L
E.g.:
Local feedback 1 = 10 (gain = 10x)
Local feedback 2 = 1 (gain = 1x --> make it buffer stage)
I'd like to add global feedback, from output 2 to input 1. What should be the R ratio range (<10, >10, or exactly 10)?
Will this global feedback improve the sound quality (e.g. reduce noise/harmonic, add damping factor, correction etc)?
Thx,
Ervin L
As you describe it, I see no need for a global at all.
All the qualities you describe (damping etc) come
from the local 1x of the output buffer stage.
You could make first stage a 20x, then add global to
knock the overall gain back down to 10x. I don't see
how this improves anything, just cause you can...
Global is generally an evil, you want little as possible.
There are exceptions, but not many...
All the qualities you describe (damping etc) come
from the local 1x of the output buffer stage.
You could make first stage a 20x, then add global to
knock the overall gain back down to 10x. I don't see
how this improves anything, just cause you can...
Global is generally an evil, you want little as possible.
There are exceptions, but not many...
Last edited:
Hi ervini,
There are a few well written technical papers and several genius amp designers that claim a special magical sound can only be achieved by amplifiers which only use local circuit degenerative feedback and only use JFETS and MOSFETS with square law gain. No global feedback and no bipolars are allowed since these produce large percentages of odd harmonic distortion. I built one and also believe, but hand matching dozens of JFETs and avoiding speakers with capacitive loads comes with this territory.
You may find this thread interestng.
GR-25
http://www.diyaudio.com/forums/solid-state/120673-gr-25-a.html
There are a few well written technical papers and several genius amp designers that claim a special magical sound can only be achieved by amplifiers which only use local circuit degenerative feedback and only use JFETS and MOSFETS with square law gain. No global feedback and no bipolars are allowed since these produce large percentages of odd harmonic distortion. I built one and also believe, but hand matching dozens of JFETs and avoiding speakers with capacitive loads comes with this territory.
You may find this thread interestng.
GR-25
http://www.diyaudio.com/forums/solid-state/120673-gr-25-a.html
we like "good stories" - and find logic, math difficult
actually while the situation is complex "global feedback" with high excess loop gain is easily shown to reduce distortion and frequency response variations due to the gain stage imperfections much more strongly than any partitioning of local feedback around the same gain stages
within some limits you should also be able to obtain the same stability, output impedance with only global feedback
local feedback may simplify circuit design by reducing interaction between circuit blocks, and can improve nonlinear stability/overload recovery which is probably why the most popular amp circuits use a mix of local and global feedback
while respecting their circuit design prowess I find most "no global feedback" or "low feedback is a barely acceptable evil" proponents profoundly ignore/misrepresent feedback theory and resist Cambrell, Cordell, Vanderkooy and Lipshitz and Cherry's works that contradict their prejudices
when the supposed evils of high global feedback in audio amplifiers have been reduced to concrete, testable hypothesis the cited authors have convincingly shown there is little or no substance to the most common complaints or that adjustments to forestall say “slew rate induced distortion” may be incorporated in high gain global feedback amplifiers without need to adopt “no feedback” topologies to fix the purported problems
feedback cannot “fix” bad amplifiers- the open loop gain stages must be able to power the load without deadbands, hysterisis, clipping, current limiting either static or dynamic – once these conditions are obtained then (global) negative feedback can reduce smooth nonlinearities and stabilize frequency response, gain with respect to device and operating point variations in the frequency range where the excess loop gain can be large – with modern devices and advanced loop gain compensation it isn’t a huge challenge to have 60 dB loop gain at 20KHz
it is a major disservice to the audio community that popular magazines, advertorials, and “just listen” naive subjectivism ignoring known psychoacoustics and perceptual testing realities make the above the least bit controversial or dismissible as "only an alternative opinion"
actually while the situation is complex "global feedback" with high excess loop gain is easily shown to reduce distortion and frequency response variations due to the gain stage imperfections much more strongly than any partitioning of local feedback around the same gain stages
within some limits you should also be able to obtain the same stability, output impedance with only global feedback
local feedback may simplify circuit design by reducing interaction between circuit blocks, and can improve nonlinear stability/overload recovery which is probably why the most popular amp circuits use a mix of local and global feedback
while respecting their circuit design prowess I find most "no global feedback" or "low feedback is a barely acceptable evil" proponents profoundly ignore/misrepresent feedback theory and resist Cambrell, Cordell, Vanderkooy and Lipshitz and Cherry's works that contradict their prejudices
when the supposed evils of high global feedback in audio amplifiers have been reduced to concrete, testable hypothesis the cited authors have convincingly shown there is little or no substance to the most common complaints or that adjustments to forestall say “slew rate induced distortion” may be incorporated in high gain global feedback amplifiers without need to adopt “no feedback” topologies to fix the purported problems
feedback cannot “fix” bad amplifiers- the open loop gain stages must be able to power the load without deadbands, hysterisis, clipping, current limiting either static or dynamic – once these conditions are obtained then (global) negative feedback can reduce smooth nonlinearities and stabilize frequency response, gain with respect to device and operating point variations in the frequency range where the excess loop gain can be large – with modern devices and advanced loop gain compensation it isn’t a huge challenge to have 60 dB loop gain at 20KHz
it is a major disservice to the audio community that popular magazines, advertorials, and “just listen” naive subjectivism ignoring known psychoacoustics and perceptual testing realities make the above the least bit controversial or dismissible as "only an alternative opinion"
Last edited:
Member
Joined 2009
Paid Member
To the original poster - local feedback is hard to avoid even if you wanted to have only global feedback as the use of a source/emitter follower is popular for making a driver stage with low output impedance and this topology is inherently high local feedback. Global feedback is very useful, it allows you to control the gain (and matching of it between channels), and it gives a flat frequency response. As far as I'm concerned, both local and global feedback can be used to reduce distortion low enough that it's mostly acadamic to argue that one is better than the other for reducing distortion.
Feedback around the output stage is a special category - because the output stage interacts with the non-linear speaker load via a long cable. Therefore, the impact of the load and any parasitic r.f. interference picked up by the speaker cable can find it's way back into the amplifier through the feedback loop. There may be benefits in excluding the output stage from a feedback loop in some applications. So perhaps one approach to feedback implementation does not fit all amplifiers.
Feedback around the output stage is a special category - because the output stage interacts with the non-linear speaker load via a long cable. Therefore, the impact of the load and any parasitic r.f. interference picked up by the speaker cable can find it's way back into the amplifier through the feedback loop. There may be benefits in excluding the output stage from a feedback loop in some applications. So perhaps one approach to feedback implementation does not fit all amplifiers.
- Status
- Not open for further replies.