And a Very, Very Merry Christmas to you too, Reodor!
There nothing like a little ad hominen to put us all in the holiday spirit!
There nothing like a little ad hominen to put us all in the holiday spirit!
Last edited:
Ha ha��
I’ll think forr struggles with someone who’s will not let him divert from answering the question.
Merry Christmas everyone �� ��
I’ll think forr struggles with someone who’s will not let him divert from answering the question.
Merry Christmas everyone �� ��
I agree on all this (apart the last sentence ?)
This said, what are the threshold of audibility of humans for harmonic distortions, IM and TIM distortions, phase turns at 20kHz, minimum damping factor etc. for an amplifier ? (yes it will depend on the speakers too).
How can-you (all) explain so many people pretend they can feel differences between amplifiers in the way they reproduce music, while their are way better than the speakers used to compare them ?
How can you explain we can be so tolerant to speakers defects, and so sensible, in the same time to so little amplifiers ones ?
All questions I have no answer but for sure, we better work speaker side than trying to reach 1ppm or 1000V/µs in our designs, while it is, indeed an intellectual fun (all of us guilty).
This said, some personal questions to you, Bonsai.
Did you feel too what I call a slightly different "character" in the way CFAs reproduce music ? And, if yes, can-you explain why, on your opinion ?
Their "expansive" dynamic behavior ?
Having some time to test my own audibility thresholds on distortions, I would like to play again with what you called AFEC.
So can-you answer my previous question: did-you build one IRL, and, if yes, what are your subjective listening conclusions if any ? Audible improvements ?
And, in order to save my time, which model of OPA between the ones that are provided with LTSpice seems the best compromises between bandwidth (phase turns/stability margin) and distortion for this purpose ?
Thanks in advance for your answers.
Andrew, here's (according to my understanding) a logically equivalent seasonal question: is Santa's reindeer more brown than wide?
Seriously, I would appreciate if somebody would bother to put together some rigorous general definitions of "CFA" and "VFA" and then explain what's the contention here. Not holding my breath, though 🙂.
BTW, forr is not the only one avoiding what appear to be inconvenient questions 🙂.
I try to and believe that I do respond to challenges addressed to me, so please don't be coy if you think that I haven't.
As to your second question, first of all, I'm not sure that there is a rigorous answer. Of course, even allowing for the difficulty of herding cats like those of us on this post, that is clearly not a sufficient reason to avoid an argument!
I do believe that a useful distinction worth considering is whether emitters/sources/cathodes are connected to the inverting feedback input, or whether bases/gates/grids are. CFAs are generally associated with the former and VFAs with the latter. I do not, however, wish to defend this association as a rigorous definition against the Mongol Hordes which might arise to attack it.
I would also suggest (note my reluctance to be the target of the same or other Golden Horde) that the controversy is whether a CFA accepts current feedback, or whether it does not, working as VFA does with voltage feedback. Please be aware that, to buttress arguments, each side has employed consultants who supported one or the other Lilliputian camp during the Little/Big Endian War.
And now, in danger of being seen as someone dares to summarize the thoughts of others, please excuse me as I change my name and enter the Federal Witness Protection Program. For surely there has been a crime committed here - that of two groups of people wasting their time trying to change the minds of the other group.
As to your second question, first of all, I'm not sure that there is a rigorous answer. Of course, even allowing for the difficulty of herding cats like those of us on this post, that is clearly not a sufficient reason to avoid an argument!
I do believe that a useful distinction worth considering is whether emitters/sources/cathodes are connected to the inverting feedback input, or whether bases/gates/grids are. CFAs are generally associated with the former and VFAs with the latter. I do not, however, wish to defend this association as a rigorous definition against the Mongol Hordes which might arise to attack it.
I would also suggest (note my reluctance to be the target of the same or other Golden Horde) that the controversy is whether a CFA accepts current feedback, or whether it does not, working as VFA does with voltage feedback. Please be aware that, to buttress arguments, each side has employed consultants who supported one or the other Lilliputian camp during the Little/Big Endian War.
And now, in danger of being seen as someone dares to summarize the thoughts of others, please excuse me as I change my name and enter the Federal Witness Protection Program. For surely there has been a crime committed here - that of two groups of people wasting their time trying to change the minds of the other group.
Hi Tour - you said -->
On my side, I tend to prefer class D for sub bass, VFA for bass reproduction and CFA for treble. On a pure subjective way that cannot be disputed.
The best of the 3 worlds for me and everybody is free to have different preferences.
Again this is the same as so many other people who listen to the two - VMA and CMA. It is the reason i asked why so many people hear it the same way you and I do. Even manufacturers.
No one knows for sure. But the audible playback characteristics are not exactly the same...... in many cases the distortion remains constant and low under dynamic conditions.
I dont think we can say it is clearly due to one thing... except maybe the current on demand aspect, maybe.... But, it is interesting. Same as when so many people hear passive parts in the same way but distortion was also supposed to be too low under the tests conducted.
There is a reason. We should try to find it.
THx-RNMarsh
On my side, I tend to prefer class D for sub bass, VFA for bass reproduction and CFA for treble. On a pure subjective way that cannot be disputed.
The best of the 3 worlds for me and everybody is free to have different preferences.
Again this is the same as so many other people who listen to the two - VMA and CMA. It is the reason i asked why so many people hear it the same way you and I do. Even manufacturers.
No one knows for sure. But the audible playback characteristics are not exactly the same...... in many cases the distortion remains constant and low under dynamic conditions.
I dont think we can say it is clearly due to one thing... except maybe the current on demand aspect, maybe.... But, it is interesting. Same as when so many people hear passive parts in the same way but distortion was also supposed to be too low under the tests conducted.
There is a reason. We should try to find it.
THx-RNMarsh
Beware, Mr Marsh. You are referring to listening impressions. You will be immediately accused of "non sense" on this forum.
You would not dare to judge the sharpness of a photography by looking at it ?
Last edited:
Regrettably, there will always be those that deny the existence of CFA’s, or their right to stand independently alongside VFA’s as a viable amplifier topology, especially suited to audio power amplifiers.
However, I take comfort from the fact that these deniers are the same ones that claim to have endured the proctological probing of aliens of as yet undetermined origin; or that Kennedy was terminated by a Russian religious sect; or that there is a race of men that live under the ground and secretly run the US government.
I am personally exhausted by these deliberations. We argued to a standstill on the original CFA vs VFA thread and it descended into such a farce and with so much vitriol that it was eventually closed down.
However, this thread is far more gentlemanly. So, we should simply leave it at that and accept we are one community divided by two topologies.
😀
However, I take comfort from the fact that these deniers are the same ones that claim to have endured the proctological probing of aliens of as yet undetermined origin; or that Kennedy was terminated by a Russian religious sect; or that there is a race of men that live under the ground and secretly run the US government.
I am personally exhausted by these deliberations. We argued to a standstill on the original CFA vs VFA thread and it descended into such a farce and with so much vitriol that it was eventually closed down.
However, this thread is far more gentlemanly. So, we should simply leave it at that and accept we are one community divided by two topologies.
😀
I believe no one deny the advantages of VFA, in the CFA camp. Specially what it brings in matter of PSR. Not so sure about the contrary from the VFA camp. And, yes, why those two camps ?
OT:
This said, it seems you don't want to help-me, avoiding twice to answer my precise questions to you ?
May-be somebody else will help-me to find the good model of an OPA, in those provided with LTSpice, that are not the ones I know and use by habit. An ultra fast OPA model, with low distortion. As it is to play again with AFEK for a CFA amp, i should tend to prefer a CFA OPA to keep everything in the same topology.
Last edited:
Wait... what? You begin with the proliferation of derogatory ad Hominem's against the "unbelievers" and then take the lead in welcoming them into the gentleman community...!
A Christmas present : at last, I found thatsome time ago, you gave a very good definition for CFA and which everbody can agree with. I will give it later. It is not a joke.
The deniers contest the naming of CFA and the way this naming suggests how the topology works. Most of them do not contest the usefulness of the topology.
As already said, the subjectivity orientation as supported by Tournesol and D. Marsh has nothing to do with the topic. It will be wellcome that they create a new thread to express their audio feelings.
Last edited:
forr, perhaps it would please you to know that the Middlebrook dual insertion test shows that significant voltage feedback does occur in most CFAs. It's just that the test also shows that current feedback is generally predominant. This is part of the reason that CFAs were given that name. On the other hand, the same test shows little current feedback in VFAs.
Would you be willing to state what you consider to be the usefulness of the CFA topology?
Andrew, I don't see around any example of those deniers; all I see is a (rather sterile) debate about the legitimacy of the "CFA" name. Myself, the only thing I'm claiming is that the CFA topology does not bring anything new to the feedback theory and practice (as the name would suggest); it's just another example of shunt output, serial input feedback topology example, which happens to have some combined remarkable (in particular given the simplicity of the circuit signal path) small and large signal properties, very useful in certain applications, and that's about it. To me it is obvious that the "CFA" name and concept, even if not fully academically correct, are here to stay, and no further amount of academic discussion will ever move it out of the picture.
And yes, I do see some that are claiming CFAs have an edge in audio reproduction; to me this is a big pile of bull chips, since not only could nobody justify the CFA edge by anything but subjective metrics, but I also can't stop remembering how 10 years ago, on this very forum, members of the same subjectivist team were chastising CFAs (Alexander amp, case in point) as having a horrible sound.
I really don't see any "community division" around the CFA concept, but then I'm after a 8 years long break, so I may have missed a lot.
Last edited:
I don't see folks claiming that the name or concept is new; arguably, it existed the first time feedback was returned to a cathode in a two-triode cascade. However, returning feedback to emitters/sources/cathodes is topologically distinct from doing so to bases/gates/grids, and since different properties tend to emerge in each case, it's useful to have a shorthand for distinguishing between the circuits. I see no academic problem at all with an assignment of the name that the Middlebrook Dual Insertion Test provides objective justification for. The problem, if anything, is historical in nature, and quite unfortunate.
The CFA name was applied to certain circuits which controlled a current through a load. Often, the output was connected to said load which was itself connected to ground through a sense resistor. The load/sense resistor junction was tapped and applied to the grid of a differential pair to provide feedback. The term was misappropriated, as the only thing that current was fed to was ground - and there was no place for it to go "back" to. Topologically, it was indistinguishable from a non-inverting VFA. The re-purposing of this regrettable name assignment, while more accurate than the original, is part of the reason for those objections which have arisen.
By the way, despite what you may have heard, have no fear of discussing things with me. You need go back only a few posts to see two cases in which Hans Polak and I started out in disagreement and /or miscommunication, and resolved our differences amicably. I would ask anyone who believes me to be unreasonable to find other examples in this thread of people who started in opposition and ended in agreement.
I am delighted at that.
Predominant ?
You know that the name has been, and still is, contested. The big problem at the time was the interpretation even when the particularity of low impedance of the inverting input was in use since forty years. That's an argument that the CFA defenders refuse to hear.
My simulation shows the output currents of CFA and VFA of the input stage in same operating conditions are equal in CFA and VFA.
Defender of the concept, you know it better than myself. Did'nt you recently speak with Hans Polak of a working situation at 400 MHz ?
Yes. There is measurably more current than voltage feedback.
Of course I know of the regrettable and inaccurate prior mis-assignment 80 or more years ago of that name to a circuit . I've even mentioned it within the prior two or three posts. But I know of no instance of those who believe that the CFA employs c.f. asserting that this is something new, with the possible exception of its appearance in an IC. Can you point to one?
It is the input rather than the output currents where the biggest distinctions occur.
Yes, but I'm not sure what you're getting at.
For some reason you object to my comparison of a VFA and a CFA with similar bandwidths, arbitrarily chosen to be 400MHz. I don't see that the frequency matters, but you do, and so I invited you to select a new frequency. I found it very difficult to find op amps with the 100Hz bandwidth you specified. A 55MHz bandwidth is more common for audio applications. Would that be acceptable?
Chris, you really need a crash course in feedback theory, since you have some serious misunderstandings in how the four types of feedback are working. But rest assured, you are not alone, the topic is indeed confusing.
Here's a very short primer:
Feedback circuits are classified by the:
a) sampling type and sampled value
b) summing type and summed value
Separating the feedback network from the open loop (aka "base") amplifier is a critical step in defining the above, it is not always obvious at the first glance. A parallel connection of the feedback network to the base amplifier at the output is considered a "voltage" sampling, a serial connection at the output is considered a "current" sampling. To add insult injury, a parallel summing node (at the input) is considered "current" summing while a serial summing node (at the input) is considered "voltage" summing.
This naming convention is the result of the analysis of what the feedback is doing to the closed loop amplifier. A parallel feedback connection tends to lower the input/output impedance, so the amplifier can be either more of a voltage output amplifier (parallel output sampling) or a current input amplifier (parallel input summing). So for example an amplifier with a parallel (voltage) output sampling, parallel (current) input summing feedback leads to a closed loop amplifier that closer to an ideal transimpedance amplifier (voltage out/current in).
Now, what the proponents call a "CFA" is an amplifier that has parallel (voltage) output sampling and serial (current) error signal summing. However, the input connection leads to an increased input impedance, so the "CFA" in fact makes the closed loop amplifier closer to an ideal voltage amplifier. Why increased input impedance? Because the error signal (whatever way you want to consider it, directly a current or a voltage through the Ohm law) simply subtracts from the open loop amplifier input current, so that the input impedance increases.
Following the same logic, you can easily identify the closed loop amplifier type, based on the sampling/summing. They are the transconductance amplifier and the current amplifier. That's all you can do out of a two-port device and two sampling/summings values (current/voltage).
The "problem" is that perhaps the simplest base (open loop, before feedback is applied) amplifier circuit topology (emitter/source/cathode summing input) has natural remarkable small and large signal properties, for a range of closed loop gains. These remarkable properties arise because the feedback resistor (or more general impedance) is not only part of the feedback network, but also affects the open loop gain of the base amp, in the range of closed loop gains where the feedback network loading (please check out the concept of "feedback loading") cannot be ignored. This closed loop gains range is anywhere where R1||RF cannot be considered constant when RF is varied.
Another "problem" is that what is generically called a "CFA" topology is not the only parallel output serial input feedback topology. As much as you may be surprised, what is called by it's proponents a "VFA" is a feedback amplifier topology that is of exactly but EXACTLY the same feedback type as a "CFA", that is parallel output, input serial, leading to a closed loop amp that is closer to an ideal voltage amplifier. However, the "VFA" does NOT have the "CFA" small signal remarkable properties (by some clever changes, the large signal "current on demand" can be implemented in a VFA). This is why I find rather funny the current CFA vs. VFA debate, which is in fact a debate about two base (open loop) amplifier topologies, that should be obvious for everybody and their mothers they are different as day and night!
So what is the bottom line of this messy, and mostly academic discussion? That CFAs and VFAs do not identify any feedback types, but two base amp (open loop) circuit topologies, each with it's own properties. The whole mess shortly described above also fully justifies finding a short and suggestive name to put in a product description leaflet or datasheet. Somebody choosed CFA, the rest is history 🙂
BTW, the CFA properties were indeed known for 80 years, but the CFA name was not choosen before 1982 or so, when Comlinear integrated the first amplifier with the CFA properties. I thought I already mentioned, I personally used the benefits of the CFA properties in discrete designs, in the late 70's, without calling them anywhere "CFA". And nobody cared to call the John Curl topology for audio (which is also a CFA topology), even earlier than that. These amplifiers were known at the time (at least in my academic work at the time) as "maximum bandwidth voltage amplifiers", in the sense they were maximizing the gain-bandwidth product.
A few useful references:
http://in.ncu.edu.tw/ncume_ee/harvard-es154/lect_18_feedback.pdf
http://cas.ee.ic.ac.uk/people/dario/files/E22/L3-feedback amplifiers.pdf
These are elementary, undergraduate level, descriptions. If you are interested, I can dig for some references treating the general/formal analysis of feedback circuits, graduate level, where the base amplifier and feedback networks are "black boxes" characterized by two port parameters (Hij, Yij,..). This "black box" type descriptions (of mostly academic importance) is arid and hard to digest, though it is offering some good insight of why the above messy naming conventions to describe the feedback circuits were chosen.
I did my best above, but I cannot hope this would close an otherwise pretty useless debate. If anybody can put the above messy prose in a more concise way, it would be appreciated. But I'm afraid there is no way around in depth studying these concepts and facts before further debating ad nauseam...
Last edited:
So please, explain the "not only" you put in the title of this thread ?
Because if your question was "CFA, a semantic problem", i would have never even opened-it. So much the name or, worse, the acronym -a disease of our time- that one can give to things that we use since decades before the name appears does not interest me.
CF Mr. Jourdain in "Le bourgeois Gentilhomme".
As well I consider that any comment in "audio" without reference to how things can "sound" is, for me, like talking about fashion without a word about how it looks.
You can keep dancing around in circles with loud noises like Indians around the torture post, I will not bother you anymore. Have fun, but be nice, do not kill anyone for once.
A few changes, corrections, and clarifications in the above...
Chris, you really need a crash course in feedback theory, since you have some serious misunderstandings in how the four types of feedback are working. But rest assured, you are not alone, the topic is indeed confusing.
Here's a very short primer:
Feedback circuits are classified by the:
a) sampling type and sampled value
b) summing type and summed value
Separating the feedback network from the open loop (aka "base") amplifier is a critical step in defining the above, it is not always obvious at the first glance. A parallel connection of the feedback network to the base amplifier at the output is considered a "voltage" sampling, a serial connection at the output is considered a "current" sampling. To add insult to injury, for not really obvious reasons (until you get into the grueling details of the two-port formal modelling and analysis) a parallel summing node (at the input) is considered "current" summing while serial summing (at the input) is considered "voltage" summing.
This naming convention of the closed loop amplifier is the result of the analysis of feedback effect. A parallel (voltage) output feedback connection tends to lower the output impedance, so the amplifier can be more of a voltage output amplifier, while a parallel (current) input summing also tends to lower the input impedance, making the input more "current friendly" (ideally a short). As such, a closed loop amplifier as above will be closer to an ideal transimpedance amplifier (voltage out/current in).
Now, what the proponents call a "CFA" is an amplifier that has parallel (voltage) output sampling and serial (voltage) input summing of error signal. The fact that the impedance at the serial input summing is low is completely irrelevant from a feedback type perspective. The input serial summing leads to an increased input impedance, so the "CFA" in fact makes the closed loop amplifier closer to an ideal voltage amplifier. Why increased input impedance? Because the error signal (whatever way you want to consider it, directly a voltage or a current through the Ohm law) simply adds to the input voltage, or subtracts from the input current, so that the input impedance increases, anyway.
Following the same logic, you can easily identify the other closed loop amplifier types, based on the sampling/summing. They are the ideal transconductance amplifier and the ideal current amplifier. That's all you can do out of a two-port device and two sampling/summings values (current/voltage).
The "problem" is that perhaps the simplest base (open loop, before feedback is applied) amplifier circuit topology (emitter/source/cathode summing input) has natural remarkable small and large signal properties, for a range of closed loop gains. These remarkable properties arise because the feedback resistor (or more general impedance) is not only part of the feedback network, but also affects the open loop gain of the base amp, in the range of closed loop gains where the feedback network loading (please check out the concept of "feedback loading") cannot be ignored. This closed loop gains range is anywhere where R1||RF cannot be considered constant when RF is varied.
Another "problem" is that what is generically called a "CFA" topology is not the only parallel output serial input feedback topology. As much as you may be surprised, what is called by it's proponents a "VFA" is a feedback amplifier topology that is of exactly but EXACTLY the same feedback type as a "CFA", that is parallel (voltage) output sampling, input serial (voltage) summing, leading as well to a closed loop amp that is closer to an ideal voltage amplifier. However, the "VFA" does NOT have the "CFA" small signal remarkable properties, essentially because the feedback network impedance does not load the base (open loop) amplifier, or otherwise affect its open loop gain (by some clever changes, the large signal "current on demand" can be implemented in a VFA). This is why I find rather funny the current CFA vs. VFA debate, which is in fact a debate about two base (open loop) amplifier topologies, that should be obvious for everybody and their mothers they are as different as day and night!
So what is the bottom line of this messy, and mostly academic discussion? That CFAs and VFAs do not identify any feedback types, but two base amp (open loop) circuit topologies, each with it's own properties. The whole mess shortly described above also fully justifies finding a short and suggestive name to put in a product description leaflet or datasheet. Somebody choosed "CFA", the rest is history 🙂
BTW, the CFA properties were indeed known for 80 years, but the CFA name was not choosen before 1982 or so, when Comlinear integrated the first amplifier with the "CFA" properties. I thought I already mentioned, I personally used the benefits of the CFA topology in discrete designs, in the late 70's, without calling them anywhere near "CFA". And nobody cared to call the John Curl topology for audio (which is also a CFA topology), even earlier than that. These amplifiers were known at the time (at least in my academic work at the time) as "maximum bandwidth voltage amplifiers", in the sense they were maximizing the gain-bandwidth product.
A few useful references:
http://in.ncu.edu.tw/ncume_ee/harvard-es154/lect_18_feedback.pdf
http://cas.ee.ic.ac.uk/people/dario/files/E22/L3-feedback amplifiers.pdf
These are elementary, undergraduate level, descriptions. If you are interested, I can dig for some references treating the general/formal analysis of feedback circuits, graduate level, where the base amplifier and feedback networks are "black boxes" characterized by two port parameters (Hij, Yij,..). This "black box" type descriptions (of mostly academic interest, related to circuit design as much as the Residue Theorem in complex analysis is related to the circuit stability criteria, is arid and hard to digest, though it is offering some good insight of why the above messy naming conventions to describe the feedback circuits were chosen, and about the meaning of add/substract ref. the conventional voltage/current signs in the two-port theory.
I did my best above, but I cannot hope this would close an otherwise pretty useless debate. If anybody can put the above messy prose in a more concise way, it would be appreciated. But I'm afraid there is no way around in depth studying these concepts and facts before further debating ad nauseam...
Chris, you really need a crash course in feedback theory, since you have some serious misunderstandings in how the four types of feedback are working. But rest assured, you are not alone, the topic is indeed confusing.
Here's a very short primer:
Feedback circuits are classified by the:
a) sampling type and sampled value
b) summing type and summed value
Separating the feedback network from the open loop (aka "base") amplifier is a critical step in defining the above, it is not always obvious at the first glance. A parallel connection of the feedback network to the base amplifier at the output is considered a "voltage" sampling, a serial connection at the output is considered a "current" sampling. To add insult to injury, for not really obvious reasons (until you get into the grueling details of the two-port formal modelling and analysis) a parallel summing node (at the input) is considered "current" summing while serial summing (at the input) is considered "voltage" summing.
This naming convention of the closed loop amplifier is the result of the analysis of feedback effect. A parallel (voltage) output feedback connection tends to lower the output impedance, so the amplifier can be more of a voltage output amplifier, while a parallel (current) input summing also tends to lower the input impedance, making the input more "current friendly" (ideally a short). As such, a closed loop amplifier as above will be closer to an ideal transimpedance amplifier (voltage out/current in).
Now, what the proponents call a "CFA" is an amplifier that has parallel (voltage) output sampling and serial (voltage) input summing of error signal. The fact that the impedance at the serial input summing is low is completely irrelevant from a feedback type perspective. The input serial summing leads to an increased input impedance, so the "CFA" in fact makes the closed loop amplifier closer to an ideal voltage amplifier. Why increased input impedance? Because the error signal (whatever way you want to consider it, directly a voltage or a current through the Ohm law) simply adds to the input voltage, or subtracts from the input current, so that the input impedance increases, anyway.
Following the same logic, you can easily identify the other closed loop amplifier types, based on the sampling/summing. They are the ideal transconductance amplifier and the ideal current amplifier. That's all you can do out of a two-port device and two sampling/summings values (current/voltage).
The "problem" is that perhaps the simplest base (open loop, before feedback is applied) amplifier circuit topology (emitter/source/cathode summing input) has natural remarkable small and large signal properties, for a range of closed loop gains. These remarkable properties arise because the feedback resistor (or more general impedance) is not only part of the feedback network, but also affects the open loop gain of the base amp, in the range of closed loop gains where the feedback network loading (please check out the concept of "feedback loading") cannot be ignored. This closed loop gains range is anywhere where R1||RF cannot be considered constant when RF is varied.
Another "problem" is that what is generically called a "CFA" topology is not the only parallel output serial input feedback topology. As much as you may be surprised, what is called by it's proponents a "VFA" is a feedback amplifier topology that is of exactly but EXACTLY the same feedback type as a "CFA", that is parallel (voltage) output sampling, input serial (voltage) summing, leading as well to a closed loop amp that is closer to an ideal voltage amplifier. However, the "VFA" does NOT have the "CFA" small signal remarkable properties, essentially because the feedback network impedance does not load the base (open loop) amplifier, or otherwise affect its open loop gain (by some clever changes, the large signal "current on demand" can be implemented in a VFA). This is why I find rather funny the current CFA vs. VFA debate, which is in fact a debate about two base (open loop) amplifier topologies, that should be obvious for everybody and their mothers they are as different as day and night!
So what is the bottom line of this messy, and mostly academic discussion? That CFAs and VFAs do not identify any feedback types, but two base amp (open loop) circuit topologies, each with it's own properties. The whole mess shortly described above also fully justifies finding a short and suggestive name to put in a product description leaflet or datasheet. Somebody choosed "CFA", the rest is history 🙂
BTW, the CFA properties were indeed known for 80 years, but the CFA name was not choosen before 1982 or so, when Comlinear integrated the first amplifier with the "CFA" properties. I thought I already mentioned, I personally used the benefits of the CFA topology in discrete designs, in the late 70's, without calling them anywhere near "CFA". And nobody cared to call the John Curl topology for audio (which is also a CFA topology), even earlier than that. These amplifiers were known at the time (at least in my academic work at the time) as "maximum bandwidth voltage amplifiers", in the sense they were maximizing the gain-bandwidth product.
A few useful references:
http://in.ncu.edu.tw/ncume_ee/harvard-es154/lect_18_feedback.pdf
http://cas.ee.ic.ac.uk/people/dario/files/E22/L3-feedback amplifiers.pdf
These are elementary, undergraduate level, descriptions. If you are interested, I can dig for some references treating the general/formal analysis of feedback circuits, graduate level, where the base amplifier and feedback networks are "black boxes" characterized by two port parameters (Hij, Yij,..). This "black box" type descriptions (of mostly academic interest, related to circuit design as much as the Residue Theorem in complex analysis is related to the circuit stability criteria, is arid and hard to digest, though it is offering some good insight of why the above messy naming conventions to describe the feedback circuits were chosen, and about the meaning of add/substract ref. the conventional voltage/current signs in the two-port theory.
I did my best above, but I cannot hope this would close an otherwise pretty useless debate. If anybody can put the above messy prose in a more concise way, it would be appreciated. But I'm afraid there is no way around in depth studying these concepts and facts before further debating ad nauseam...
Last edited: