There`s not so many threads about nested feedback round here.
There are some examples of such a design like AKSA`s "Lifeforce" and Greg BAll`s "SKA" but the schematics are not available. Everyone saying that these amps sound wonderfull.
ANyone have schematic or example?
It would be nice of someone to tell how it realy works and wath`s the trick.
Cheers!
There are some examples of such a design like AKSA`s "Lifeforce" and Greg BAll`s "SKA" but the schematics are not available. Everyone saying that these amps sound wonderfull.
ANyone have schematic or example?
It would be nice of someone to tell how it realy works and wath`s the trick.
Cheers!
I didnt ment Aksa`s and Greg`s schematics (I know that only kit buyers have them and I respect that).
I meant some other design exaples of amp with nested feedback? Anyone round here designed such a amp?
I meant some other design exaples of amp with nested feedback? Anyone round here designed such a amp?
This should get you started:
http://www.sx.ac.uk/ese/research/au...s/C39 NS and nested differential feedback.pdf
http://www.personal.reading.ac.uk/~shsmchlr/pof/pof20.pdf
NDF is quite complex, and the Cherry paper is difficult to read.
NDF gives you a bit more feedback gain at the freq extremes and so measurably improves linearity there.
Whether it is audible of course depends on a LOT of other factors.
Due to the complexitity in stabilizing each FB loop and the limited advantages it never really caught on.
Jan Didden
http://www.sx.ac.uk/ese/research/au...s/C39 NS and nested differential feedback.pdf
http://www.personal.reading.ac.uk/~shsmchlr/pof/pof20.pdf
NDF is quite complex, and the Cherry paper is difficult to read.
NDF gives you a bit more feedback gain at the freq extremes and so measurably improves linearity there.
Whether it is audible of course depends on a LOT of other factors.
Due to the complexitity in stabilizing each FB loop and the limited advantages it never really caught on.
Jan Didden
Nested feedback implies certain feedback loops (AKA: local feedback) in an amplifier of some sort that also has global negative feedback.
An example of nested feedback in an amp using global feedback would be a voltage gain stage using an emitter/source/cathode resistance to degenerate the gain of the stage.
Another possibility would be degeneration resistances in the legs of a differential pair. Resistors would be added to the emitter/source/cathode of each side of a long/short tailed pair to degenerate the active devices and reduce distortion.
The major concept here is to reduce the inherent distortion of each important part of the amplifier separately so that the reliance on global feedback is not so heavy. This is important for reducing the dynamic distortions like intermodulation distortion and others.
Hopefully that clears a bit up. Others will have more on this.
An example of nested feedback in an amp using global feedback would be a voltage gain stage using an emitter/source/cathode resistance to degenerate the gain of the stage.
Another possibility would be degeneration resistances in the legs of a differential pair. Resistors would be added to the emitter/source/cathode of each side of a long/short tailed pair to degenerate the active devices and reduce distortion.
The major concept here is to reduce the inherent distortion of each important part of the amplifier separately so that the reliance on global feedback is not so heavy. This is important for reducing the dynamic distortions like intermodulation distortion and others.
Hopefully that clears a bit up. Others will have more on this.
Local feedback loops are “nested” in the sense of being enclosed in the global feedback loop
if “local feedback” is defined as feedback around single stage then of course only a single level of “nesting” is possible, however I know of no rule that “local” is restricted to only one stage - but rather is just any feedback loop that isn’t global
some common audio amp circuits can be seen as more levels of nesting:
emitter degeneration of a VAS with a conventional Miller C is “inside” both the Miller loop and the Global loop for 2 levels of nesting, as would be either input or output unity gain followers buffering the VAS Q
Sziklai pair/triple output stages also add one or two more feedback nesting levels beyond the one level of nesting that results from unity gain follower output stage being considered a local feedback itself enclosed in the global loop
these examples are not regularly structured and extensible as is the Cherry scheme but I think it is still fair to call them examples of nested feedback
if “local feedback” is defined as feedback around single stage then of course only a single level of “nesting” is possible, however I know of no rule that “local” is restricted to only one stage - but rather is just any feedback loop that isn’t global
some common audio amp circuits can be seen as more levels of nesting:
emitter degeneration of a VAS with a conventional Miller C is “inside” both the Miller loop and the Global loop for 2 levels of nesting, as would be either input or output unity gain followers buffering the VAS Q
Sziklai pair/triple output stages also add one or two more feedback nesting levels beyond the one level of nesting that results from unity gain follower output stage being considered a local feedback itself enclosed in the global loop
these examples are not regularly structured and extensible as is the Cherry scheme but I think it is still fair to call them examples of nested feedback
JCX: I agree. This is what I had based my first post on. I understand that the nomenclature sways a little, but from the literal meaning of the word 'nested' it is plausible.
You are all correct, I suspect.
Nested fb is only intuitively understood not to include local feedback around gain elements. Nested refers more to two fb loops, one global from output to input, and another across two stages or more inside the global loop.
Jan is right. These designs are much more difficult to stabilise, and there are Zout issues because the global fb is diminished by the internal, nested fb. But the advantages are increased resolution, flatter power bandwidth, and higher tolerance to reactive loads.
Cheers,
Hugh
Nested fb is only intuitively understood not to include local feedback around gain elements. Nested refers more to two fb loops, one global from output to input, and another across two stages or more inside the global loop.
Jan is right. These designs are much more difficult to stabilise, and there are Zout issues because the global fb is diminished by the internal, nested fb. But the advantages are increased resolution, flatter power bandwidth, and higher tolerance to reactive loads.
Cheers,
Hugh
AKSA said:
Yes. Nested feedback as developed by Cherry is a very specific nest of feedback loops, starting all at the output and enclosing progressively more stages until the most outer loop goes back to the input stage as normal global feedback.
The aim is to tailor the feedback gain freq response in such a way that the max feedback is possibly where it is most needed: at the high freq end of the bandwidth.
I would really recommend to read the Cherry paper (I'l see if i can find it) before commenting on it.
Jan Didden
Such loop topologies are quite common in other fields of feedback control.
I think it isn't used more often for audio amps because many designers do still tune the "standard" amp topology by the use of trial and error. With multi-loop topolgies this would definitely not work that well anymore.
The only nested loop topology that can easily be trimmed by trial-and error is the "standard" amp with local degeneration (which is actually called current feedback in my mother-tongue BTW) as mentioned above.
Regards
Charles
P.S.: By "standard" I mean the ubiquitos topology with
differential, common-emitter VAS, driver and output stage.
I think it isn't used more often for audio amps because many designers do still tune the "standard" amp topology by the use of trial and error. With multi-loop topolgies this would definitely not work that well anymore.
The only nested loop topology that can easily be trimmed by trial-and error is the "standard" amp with local degeneration (which is actually called current feedback in my mother-tongue BTW) as mentioned above.
Regards
Charles
P.S.: By "standard" I mean the ubiquitos topology with
differential, common-emitter VAS, driver and output stage.
Hi all
Cherry wrote up the principle in a fairly easy to follow couple of articles in Electronics Today International back in the 80's.
He published a 60W amp using the principles in the second article.
Before Stochino wrote about high input stage currents (non-slewing amp) in Electronics World, Cherry also pointed out the use of relatively high input degeneration resistors in the input stages.
Interestingly, I found that the simulations of Cherry's amp showed slightly worse distortion than Self's blameless, at lower frequencies but there was a reduction in crossover artefacts.
The higher distortion is almost certainly due to the increased degen which reduces the open loop gain compared with Self's, which would explain the differences. Cherry also made a point about taking the "miller" capacitor from the output terminal instead of the VAS to reduce crossover distortion. This works - see my suggested JLH update and the quasi version.
cheers
John
Cherry wrote up the principle in a fairly easy to follow couple of articles in Electronics Today International back in the 80's.
He published a 60W amp using the principles in the second article.
Before Stochino wrote about high input stage currents (non-slewing amp) in Electronics World, Cherry also pointed out the use of relatively high input degeneration resistors in the input stages.
Interestingly, I found that the simulations of Cherry's amp showed slightly worse distortion than Self's blameless, at lower frequencies but there was a reduction in crossover artefacts.
The higher distortion is almost certainly due to the increased degen which reduces the open loop gain compared with Self's, which would explain the differences. Cherry also made a point about taking the "miller" capacitor from the output terminal instead of the VAS to reduce crossover distortion. This works - see my suggested JLH update and the quasi version.
cheers
John
I have the second part.
Can anyone post the first part with the design equations or even both parts?
You might check out the works of Hawksford
http://www.essex.ac.uk/ese/research/audio_lab/malcolms_publications.html
and an amplifier implementation by Cordell.
http://www.cordellaudio.com/papers/
http://www.essex.ac.uk/ese/research/audio_lab/malcolms_publications.html
and an amplifier implementation by Cordell.
http://www.cordellaudio.com/papers/
myhrrhleine said:
Thanks for the suggestion. I already have those articles. Cherry's design is different.
I might be wrong, but isn't Cherry's nested feedback defined like this:
-one feedback point (usually the amp's output)
-multiple points where you send back the feedback signal
imagine an amp with 3 stages, and instead of sending the output back to the input, you send it both to the input and to the 2nd gain stage's input.
so local feedback (degeneration) can't be considered as part of a nested feedback, except maybe if you degenerate the last stage
-one feedback point (usually the amp's output)
-multiple points where you send back the feedback signal
imagine an amp with 3 stages, and instead of sending the output back to the input, you send it both to the input and to the 2nd gain stage's input.
so local feedback (degeneration) can't be considered as part of a nested feedback, except maybe if you degenerate the last stage
somehow I doubt there will be much action in a DIY thread that is restricted to Cherry's nested differentiating feedback loops as the sole acceptable example of nested feedback - nor is it clear that that is the op's definition
I see no reason to define nested feedback by any other criteria than as feedback loops "inside" the global or other local feedback loops, Matryoshka style or tangent loops would both fit the simple "nested" concept
this certainly includes local followers or degeneration as (perhaps trivial) examples of "nested" feedback loops inside the global, miller and/or other local feedback loops enclosing multiple stages
You might note that the Extrema Class A amp circuit has a local feedback from T29,30 around the tripple darlington output
http://www.diyaudio.com/forums/showthread.php?s=&threadid=96853
I see no reason to define nested feedback by any other criteria than as feedback loops "inside" the global or other local feedback loops, Matryoshka style or tangent loops would both fit the simple "nested" concept
this certainly includes local followers or degeneration as (perhaps trivial) examples of "nested" feedback loops inside the global, miller and/or other local feedback loops enclosing multiple stages
You might note that the Extrema Class A amp circuit has a local feedback from T29,30 around the tripple darlington output
http://www.diyaudio.com/forums/showthread.php?s=&threadid=96853