Here you have pos. and neg.CFB, also (neg.)VFB.All going to the cathode of the first tube = called CFB today 😕
So there is voltage to current feedback and current to current feedback 😀
Mona
Attachments
If you look at votages and currents in an amplifier then you'll see that the voltage modulation you put into the inputnode is converted into current modulation directly after the first device and only transformed into voltage modulation as the VAS bias spread is modulated by the alternating currents in the VAS stage, so essentially all feedback is summation of currents.
From the outside of a circuit the feedback has the characteristics of Voltage to voltage summation regardless what you may call the circuit VFA or CFA. Difference between the two is only topology and the fact the one is passive V/I and in the other (VFA) the V toI is through a device.
From the outside of a circuit the feedback has the characteristics of Voltage to voltage summation regardless what you may call the circuit VFA or CFA. Difference between the two is only topology and the fact the one is passive V/I and in the other (VFA) the V toI is through a device.
as soon as you get to current-mode operation of circuits, will CFB make better sense.
Until then, currents can be converted to voltages and voltage converted to current and around and around we go.
-RM
Until then, currents can be converted to voltages and voltage converted to current and around and around we go.
-RM
Last edited:
Exactly, the only thing that counts is the result.Do you want to regulate the current or the voltage (or both) of the result (output).
Mona
Feedback is based on subtraction of two signals. In your valve amp, it is realized by V1 which senses the voltage between g and k to control the current though it. You can do a lot of things with the feedback networks, the fundamental, physical, law to which V1 obeys is unique.
The same way, global feedback loads cathode of the 1'St pentode, adding a local FB dependent on global loop including the OL gain, the same CFB as you are trying to digest here. But the example contains also a combination of PFB by current and NFB by voltage!
It might be cleaner to think of feedback in terms of amplifying an error signal. That error signal could be a difference of two voltages, or a non-zero current into or out of a node. The second case is more like a low impedance inverting current input - there is no explicit 'subtraction' needed to generate the error signal, the error signal _is_ the current flowing into the inverting current input. Or, viewed another way, the 'subtraction' takes place inside of the amplifier as part of the inverting current input node.
Well said. Feedback is a concept, and people come up all the time with interesting ways to implement it. Adding or subtraction, any combination of current or voltage, what will they think of next. Saying that current feedback 'must be subtraction of currents' and referring to a fundamental physical law is very narrow and limited and does not do justice to a rich electronic engineeering design space.
Jan
Yes. Complex values, frequency dependent. Also, the majority of residents in this thread analyse ideal cases as if forgetting that the purpose of feedbacks is to correct errors.
Just my 
, defining whether it is a "current" or "voltage" that is fed back into the feedback node may be somewhat fuzzy, ie. when is the FB becoming C or V signal, but perhaps it could be defined by a rough ballpark figure ratio of ca. 50-500, that is the hfe within where vast majority of bjt's fall within, hence when we use the emitter as feedback node(TIS), instead of the base(TAS), the FB current difference needed is on the order of the hfe ratio, and then we have mosfets too.....
Maybe my crude assumption doesn't hold in all cases, but I generally used to think of "CFA" being simply a TIS feedback node. (.. irrespective whether we are sampling/aspiring for a C or V output signal through the load by means of manipulated FB regimes, and/or are dealing with an amp which in its forward path alone is a V/I converter or else.)
Anyhow, Bonsai and Jan already provided slides and tables for the 4 classic topologies, they are fine for me, not a big deal, just sink into philosophy mode where "everything is plausible" and try listen to the sound of one hand clapping...
, defining whether it is a "current" or "voltage" that is fed back into the feedback node may be somewhat fuzzy, ie. when is the FB becoming C or V signal, but perhaps it could be defined by a rough ballpark figure ratio of ca. 50-500, that is the hfe within where vast majority of bjt's fall within, hence when we use the emitter as feedback node(TIS), instead of the base(TAS), the FB current difference needed is on the order of the hfe ratio, and then we have mosfets too.....Maybe my crude assumption doesn't hold in all cases, but I generally used to think of "CFA" being simply a TIS feedback node. (.. irrespective whether we are sampling/aspiring for a C or V output signal through the load by means of manipulated FB regimes, and/or are dealing with an amp which in its forward path alone is a V/I converter or else.)
Anyhow, Bonsai and Jan already provided slides and tables for the 4 classic topologies, they are fine for me, not a big deal, just sink into philosophy mode where "everything is plausible" and try listen to the sound of one hand clapping...
Last edited:
@Forr: Are you going to comment on MK's (and maybe your) current position on this CFA issue?
Jan
Jan
What intrigues me with, as Jan puts it, the concept of feedback, is that the errors are corrected before they happen. This to me is the fundamental paradox of negative feedback.
They are not. It is fully causal. Think of riding a bike. As soon as you detect that you start to fall over, you turn the handle bars slightly and/or shift your weight sleightly to start moving to the other side. As a result, you stay upright.
A very nice example of feedback control.
Jan
Ok, it's a matter of speed, they happen so quickly it's instantaneous, they never get the chance to accumulate
Exactly! The signal transit time in an audio circuit is sub-microsecond, and in comparison, the feedback reacts instantaneously.
Jan
Thanks Jan, I knew that I knew that I knew that.....
I expressed it badly, as it appears in a simplistic way
I expressed it badly, as it appears in a simplistic way
If there was really no delay why than the problems with phase shift ? Sometimes the feedback has to over-react to compensate for the delay.
Mona
Stop! This is getting painful again. A circuit reaches a steady state solution, even thinking of an entire 15min song on repeat is still a steady state signal. The delay comment leads to the feedback goes round and round and very fuzzy thinking.
Overreact? You make it sound as if the feedback thinks about what it does.
If there is phase shift, the feedback is no longer exactly 'negative' (in opposite phase). If the phase accumulates 180 degrees extra, you have now positive feedback - the stuff oscillators are made of.
But before that point you start to see ringing and prolonged settling time of course.
All in most good feedback and control systems books.
Jan