Reducing closed loop gain in any series (voltage applied) shunt (voltage derived) negative feedback amplifier, including so-called "CFAs", means either
1) increasing the value of the feedback resistor connected to ground
or
2) reducing the value of the resistor connected to the output.
The first of these, in a so-called "CFA", reduces first stage gain and therefore reduces bandwidth even as it reduces closed loop gain. This is because the effective impedance at the emitter of the input stage is increased.
The second option, in a so-called "CFA" slightly increases first stage gain and therefore increases bandwidth even as it reduces closed loop gain.
The second option is therefore preferable with a so-called "CFA" as it increases loop transmission if stability margins are not compromised by the increase in bandwidth.
However, users of so-called "CFAs" are often exhorted to alter closed loop gain by using option (1), presumably because option (2) may compromise stability.
Last edited:
I haven't used any particular model for so-called "current-mode feedack".
To me it's abundantly obvious that so-called "current-mode feedack" amplifiers are nothing of the sort: they're self-evidently series (voltage applied) shunt (voltage derived) feedback amplifiers, i.e. voltage amplifiers.
All these attributes are readily attainable in audio applications with VFAs that use a differential stage for their input stage.
I am not sure what is meant by "more consistent sound quality" though.
Last edited:
I am not sure what is meant by "more consistent sound quality" though.
Now its becoming interesting!
Well, I'm not going to defend the sound character of anything.... I will just put forth for all to consider -->
The typical THD curve with Voltage feedback toplogy is one of higher distortion in the high audio freqs than at the bass and mids. Rising thd as freq goes up.
This doesnt happen with CMode feedback circuits. The harmonics structure - relation of one harmonic to another -- their amplitutes and the freqs relationship of the harmonics stays the same across the BW when using current-mode or compimentary push-pull current-mode fb circuits. The sonic descriptions given to this class of amplifier topology is unattainable any tradional way. The diff/vfb circuits sound one way in the bass and other in the highs.
What is clear to you isnt so clear to the IC industry, it appears.
Noise can be lower, thd can be as low and etc. PSR and CMR can be boosted by cascoding and better Power supplies (which even vfb opamps benefit from) et al....
Evryone can study up on it and find out for themselves the pro-con of each and the means to overcome each cons. So, thanks for the input and thoughts and I'll leave the rest for others to explore for themselves in the published literature. The references are also interesting.
Thx-RNMarsh
The typical THD curve with Voltage feedback toplogy is one of higher distortion in the high audio freqs than at the bass and mids. Rising thd as freq goes up.
This doesnt happen with CMode feedback circuits. The harmonics structure - relation of one harmonic to another -- their amplitutes and the freqs relationship of the harmonics stays the same across the BW when using current-mode or compimentary push-pull current-mode fb circuits. The sonic descriptions given to this class of amplifier topology is unattainable any tradional way. The diff/vfb circuits sound one way in the bass and other in the highs.
What is clear to you isnt so clear to the IC industry, it appears.
Noise can be lower, thd can be as low and etc. PSR and CMR can be boosted by cascoding and better Power supplies (which even vfb opamps benefit from) et al....
Evryone can study up on it and find out for themselves the pro-con of each and the means to overcome each cons. So, thanks for the input and thoughts and I'll leave the rest for others to explore for themselves in the published literature. The references are also interesting.
Thx-RNMarsh
Last edited:
Trolling its is. Combined with a solid measure of the old Gadfly ointment. Some of us will go on believing what we decide is fact, no matter what the evidence is to the contrary or despite perfectly valid alternative interpretations. And there we should leave it. For to soldier on in the face of such obstinance is pure folly. 
"This shows that the (CFB's) bandwidth becomes independent of the set gain G, because bandwidth is defined as the frequency at which the gain drops by 3 dB, no matter how large the gain is."
To that, add an open loop gain level as much as you can and you'll get the best amplifier as money can buy.
Where is you evidence?
True. You and a few others have chosen to believe, without adducing a single shred of truth in rebuttal, that so-called "CFAs" are not VFAs.
Sad.
I really am afraid one shouldn't be able to tell the difference between two competently designed amplifiers in a controlled listening test, regardless of differences in their design philosophies.
I am positive this need not be the case with VFAs with traditional differential pair as input stage.
Scott Wurcer's AD797 and the NE5534 spring immediately to mind.
please refer to attached schematic
Folks I posted this schematic in 2006, I ask that you refer to it. Where is the error signal in this simplified CFA topology? I will comment but those who deny the current mode natire of attached topology should state the error signal and how it controls the loop. For the record I explained in 2006:
http://www.diyaudio.com/forums/soli...eedback-not-suitable-audio-21.html#post955535
Please review and comment if you disagree. The critics keep saying it's Vin-Vfb, so I ask that the mechanism be explained. A serious understanding of control theory and amp topology is needed to address this issue. The error signal is In, the current at the inverting input, which is low Z. BR.
Claude
Folks I posted this schematic in 2006, I ask that you refer to it. Where is the error signal in this simplified CFA topology? I will comment but those who deny the current mode natire of attached topology should state the error signal and how it controls the loop. For the record I explained in 2006:
http://www.diyaudio.com/forums/soli...eedback-not-suitable-audio-21.html#post955535
Please review and comment if you disagree. The critics keep saying it's Vin-Vfb, so I ask that the mechanism be explained. A serious understanding of control theory and amp topology is needed to address this issue. The error signal is In, the current at the inverting input, which is low Z. BR.
Claude
It is generally thought that the same amount of NFB across the whole audio range would provide this feature (a dubious requirement which drives us back to year 1980).
The idea of a constant distortion's harmonic structure, beyond the fact that it supposes that the harmonic distorsion is so high that it is audible, has to be exposed to the human acoustic physiology. Looking at the equal loudness contours, the distorsion harmonic components are certainly not perceived in the same manner at all frequencies. For example, for signals at 1 kHz and 3.33 kHz, considering the third harmonic, the ear sensitivity is about 3 dB higher at 3 Khz than at 1 kHz but looses 15 dB between 3 kHz and 10 kHz.
An other argument against this concern about a uniform harmonic distorsion across the whole audio range is that the dominating mechanism in the generation of distorsion is in the output stage and that it usually completely masks the effects of the non-linearties of the previous stages. As the output non-linearity worsens with frequency, even a constant amount of NFB would not avoid changes in distorsion and distorsion structure across the audio band. What would be needed is an increase of NFB with frequency without compromising the NFB amount at low frequencies in any way. Electronic circuits do not allow this.
Last edited:
Forr--- generally i agree with you. As I indicated, I wouldnt defend it. It does have a unique characteristic sound.... not that all harmonics are heard the same sensitivity.... even if the harmonic are the same levels across the BW.
Also thd below -100dB from a single tone is quite possible with vfb or cfb but I cant hear it. But power amps are harder than preamps to get that low at all output/load levels/conditions. Especially if you do multitone testing... the total level rises further making it more likely audible.
And, in the whole chain there are many opamps etc from mic to playback speaker in the home.... the total accumulated thd/IM et al adds up to much higher total levels with all those multitones being played thru those stages such that the sound is more like hi-fi than what real live music sound like. Thus, you are back to -100dB or more required for all stages in the chain to stay below audibility at the end of the chain. If that could be accomplished, we would not have to worry what the topology was.
Thx- RNMarsh
What is interesting, even entertaining at times, are all the various explainations here of how the circuitry actually works. It appears in literature going back to the 1990's that it IS actualy known and understood.
Also thd below -100dB from a single tone is quite possible with vfb or cfb but I cant hear it. But power amps are harder than preamps to get that low at all output/load levels/conditions. Especially if you do multitone testing... the total level rises further making it more likely audible.
And, in the whole chain there are many opamps etc from mic to playback speaker in the home.... the total accumulated thd/IM et al adds up to much higher total levels with all those multitones being played thru those stages such that the sound is more like hi-fi than what real live music sound like. Thus, you are back to -100dB or more required for all stages in the chain to stay below audibility at the end of the chain. If that could be accomplished, we would not have to worry what the topology was.
Thx- RNMarsh
What is interesting, even entertaining at times, are all the various explainations here of how the circuitry actually works. It appears in literature going back to the 1990's that it IS actualy known and understood.
Last edited:
A good tutorial in clear, easy to understand language is given by Analog Device members on both voltage and current feedback circuits:
www.analog.com/library/AnalogDialog/archives/39-05/Web_Ch1_final_R.pdf
Enjoy. -RNMarsh
www.analog.com/library/AnalogDialog/archives/39-05/Web_Ch1_final_R.pdf
Enjoy. -RNMarsh
Last edited:
The munn-fletcher curves shows the audibility of sounds, not of distortion.
What I mean is that IME hf distortion is more audible to most people. Take two amps, all equal, both with thd1 of 0,01% but with thd20 of 0,15% the first and 0,02% the second and guess which one sounds better.
About the other argument, a sufficiently high bw and plenty of feedback should achieve very good HF distortion specs.
For vfb-cfb argument I think is not needed to expand the discussion on the quality of the harmonic spectrum.
This is a very interesting observation.
I have noticed this "hifi" sound also in chains that didn't have a series of opamps... Lots of active devices produce the same degradation and maybe even sooner (they are less matched to each other than in ICs).
I remember an article of Gerhard where up to 8 opamps were not audible to him.
Less opas but of limited bandwith should also deteriorate the sound quicker. I'm very sensitive to HF distortion and I have a certain hate for this kind of hifi sound, so I'm biased (in class A I suppose
).I am sorry, Claude, but Control Theory was one of my majors, and I am completely satisfied that the error signal driving the amplifier's forward path in a so-called "CFA" is Vin-Vfeedback because the quantity being sampled by the feedback network is the ouput voltage and NOT the output current. This is elementary.
Last edited:
Fascinating! I requested a schematic so we could examine your "CFA" circuit that you assert has all these wonderful characteristics but you're yet to provide it:
http://www.diyaudio.com/forums/soli...k-how-do-i-see-difference-52.html#post3425571
If these characteristics truly exist, then wouldn't it be fair to expect that so-called "CFAs" would be as ubiquitous as the NE5532/4 in the audio industry?
Last edited:
Michael Kiwanuka wrote
"I am sorry, Claude, but Control Theory was one of my majors, and I am completely satisfied that the error signal driving the amplifier's forward path in a so-called "CFA" is Vin-Vfeedback because the quantity being sampled by the feedback network is the ouput voltage and NOT the output current. This is elementary. "
100% agree with you that it is the output voltage that is being sampled in these new fangled CFA thingies. And, I dont think I have held a different position on this point since we started this discussion.
So, if we agree on this point, tell me where I am going wrong and exactly what it is that we disagree on.
"I am sorry, Claude, but Control Theory was one of my majors, and I am completely satisfied that the error signal driving the amplifier's forward path in a so-called "CFA" is Vin-Vfeedback because the quantity being sampled by the feedback network is the ouput voltage and NOT the output current. This is elementary. "
100% agree with you that it is the output voltage that is being sampled in these new fangled CFA thingies. And, I dont think I have held a different position on this point since we started this discussion.
So, if we agree on this point, tell me where I am going wrong and exactly what it is that we disagree on.
100% agree with you that it is the output voltage that is being sampled in these new fangled CFA thingies. And, I dont think I have held a different position on this point since we started this discussion.
So, if we agree on this point, tell me where I am going wrong and exactly what it is that we disagree on.
Correct me if am wrong, but I reckon where you're going wrong is assuming "these new fangled CFA thingies" are current feedback amplifiers; they are not, by any strech of the imagination; they are, in fact, voltage feedback amplifiers because the signal fed into the emitter(s) of the input stage is a voltage.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.