P-A wrote> Dimitri, now you are talking rubbish. You are saying the same as I did, a couple of posts up and I talked rubbish.
Dear P-A, have I ever argued with your above mentioned post? 😕
Dear P-A, have I ever argued with your above mentioned post? 😕
I've noticed that this always happens when the topic of current feedback pops up. After a while, nobody remembers the position of anyone, and people just argue with those they usually argue with.
Maybe we could just call the different kinds of feedback Huey, Dewey and Louie😀. Would be equally confusing!
This was an excellent thread until CFB was brought into the discussion. I suggest P-A, as a moderator
, splits this of to a separate thread.
Rune
Maybe we could just call the different kinds of feedback Huey, Dewey and Louie😀. Would be equally confusing!
This was an excellent thread until CFB was brought into the discussion. I suggest P-A, as a moderator
, splits this of to a separate thread.Rune
Re: let us come to the point
No, no, no! What was called CFB is voltage feedback because the input voltage to the amp is created by subtracting the feedback VOLTAGE from the input VOLTAGE, the resulting VOLTAGE is multiplied by the open loop gain and creates the output voltage.
It has nothing to do with the impedance at the feedback point. The low impedance allows some tricks like class AB bias to quickly charge/discharge capacitance, high, symmetrical slewrate, to extend the bandwidth or make it less depending on the closed loop gain, but doesn't all of a sudden change the conceptual circuit!
And as I said before, I understand that this (CFB) has now become a de facto standard, and this will stay until another marketeer invents an even catchier phrase. I know that, I accept that (what else can I do
). In THAT sense I agree with you.
Jan Didden
dimitri said:
No, no, no! What was called CFB is voltage feedback because the input voltage to the amp is created by subtracting the feedback VOLTAGE from the input VOLTAGE, the resulting VOLTAGE is multiplied by the open loop gain and creates the output voltage.
It has nothing to do with the impedance at the feedback point. The low impedance allows some tricks like class AB bias to quickly charge/discharge capacitance, high, symmetrical slewrate, to extend the bandwidth or make it less depending on the closed loop gain, but doesn't all of a sudden change the conceptual circuit!
And as I said before, I understand that this (CFB) has now become a de facto standard, and this will stay until another marketeer invents an even catchier phrase. I know that, I accept that (what else can I do
). In THAT sense I agree with you. Jan Didden
Now it's my turn to say no!
In a CFB amp, the error signal inside the feedback loop is originally a CURRENT. (whether it is generated by subtracting voltages is not relevent).
This internal error current is converted to an error voltage in the trans-impedance stage. (This trans-impedance stage usually defines the dominant pole at the same time). The resulting voltage feeds an output stage as with any other amp.
In a CFB amp, the error signal inside the feedback loop is originally a CURRENT. (whether it is generated by subtracting voltages is not relevent).
This internal error current is converted to an error voltage in the trans-impedance stage. (This trans-impedance stage usually defines the dominant pole at the same time). The resulting voltage feeds an output stage as with any other amp.
Ouroboros, is totally right. Jan, you must forget the thing about voltage divider which is distracting you. A transimpedance amp must have current as input signal and feedback signal.
Ouroboros said:
I have read like this. And, the closed loop gain was driven from the initail idea of the internal error current. Then, the gain fomular was compared with the voltage-feedback case. The point was that the bandwidth might be kept constant in case of the current feed back, independent of the closed loop gain value. Anyhow, I have understood that the voltage and the current always exist together inside conducting paths.
Offtopic:
Sad all day... Bad people ... He was nothing but an innocent young man, ...
I think this whole discussion is complete nonsense. When I studied electrical engineering I specialized in analog design. The terms voltage feedback or current feedback were never introduced. It's all arbitrary.
When examining a feedbackloop you have some freedom to define feedback. In every amplifier you can choose to define the feedback-factor as a voltage or as a current, it doesn't make a difference and the calulations yield exactly the same result. This is not a suprise as everybody here knows that there is no such thing as a current or voltage source (remember Thevenin and norton equivalents?).
It really is semantics. That's what this discussion is all about. I don't think we can end this discussion unless we agree on the term we'll use for every topology around.
When examining a feedbackloop you have some freedom to define feedback. In every amplifier you can choose to define the feedback-factor as a voltage or as a current, it doesn't make a difference and the calulations yield exactly the same result. This is not a suprise as everybody here knows that there is no such thing as a current or voltage source (remember Thevenin and norton equivalents?).
It really is semantics. That's what this discussion is all about. I don't think we can end this discussion unless we agree on the term we'll use for every topology around.
I agree with most of the previous posts. It's all currents inside whatever. And you can define it arbitrarily if you can agree on the definition.
But there is a slight distraction. Making the amp input signal by subtracting feedback voltage from input voltage (as in the simple case, Vbe = Vb - Ve) has been defined as voltage feedback, ohh, I guess just after the allied invasion on France.
So, yes, you can make a lot of cases for any other definition you want, but unfortunately you have been beaten by about 50 years. That is why this is NOT about semantics, just about knowing your stuff.
Last post from me on this.
Jan Didden
But there is a slight distraction. Making the amp input signal by subtracting feedback voltage from input voltage (as in the simple case, Vbe = Vb - Ve) has been defined as voltage feedback, ohh, I guess just after the allied invasion on France.
So, yes, you can make a lot of cases for any other definition you want, but unfortunately you have been beaten by about 50 years. That is why this is NOT about semantics, just about knowing your stuff.
Last post from me on this.
Jan Didden
ABO said:
I think we agree
http://www.diyaudio.com/forums/showthread.php?s=&postid=413535&highlight=#post413535
However, please note that certain older textbooks actually do
use the term current feedback for feedback that senses the
load current and that the term is nowadays widely used for
another purpose, especially in the context of op amps. We
have however already discussed this conflicting terminology
to death in numerous previous threads.
ABO said:
An error....easily made....:
Some illumination is required here...methinks:
1) If the amplifier's forward path takes a voltage as it's input and delivers a voltage at it's output,(viz: voltage gain)...
..... then the global feedback loop applied to such an amp. must neccesarily be defined by a voltage transfer function...
...as signal sampled by the feedback network is a voltage,....and this is then subtracted from the input voltage to produce the error voltage needed to drive the amps. forward path....
That the voltage signal delivered by the feedback network can be converted to its Norton equivalent is academic, as we cannot subtract a current from the input voltage......a simple matter of inconsistent units....
2) If the amplifier's forward path takes a voltage as it's input, and delivers a current at it's output, (viz: transadmittance gain),
....then the global feedback loop applied to such an amp. must necessarily be defined by a transimpedance transfer function....
...as the signal sampled by the feedback network is a current,....a fraction of which must be returned as a voltage before it can be subtracted from the input voltage to produce the error voltage needed to drive the amps. forward path....
3)If the amplifier's forward path takes a current as it's input, and delivers a voltage at it's output, (viz: transimpedance gain),.....
.........then feedback applied to such an amp. is a transadmittance....
...as the signal sampled by the feedback network is a voltage,....a fraction of which must be returned as a current before it can be subtracted from the input current to produce the error current needed to drive the amps. forward path....
4) If the amplifier's forward path takes a current as it's input, and delivers a current at it's output, (viz: current gain),.....
.........then the feedback network's transfer funtion is a current....
...as the signal sampled by the feedback network is a current,....a fraction of which must be returned as a current before it can be subtracted from the input current to produce the error current needed to drive the amps. forward path....
It is only with the later that the term current feedback correctly applies.....Period.
Mikeks,
by your statement regarding point 4 above, you deviate both
from the modern usage of the term by op amp manufacturers
and from how it is used in older textbooks. We now have
three conflicting definitions of current feedback, instead of
the previous two we have discussed endlessly. 🙂
by your statement regarding point 4 above, you deviate both
from the modern usage of the term by op amp manufacturers
and from how it is used in older textbooks. We now have
three conflicting definitions of current feedback, instead of
the previous two we have discussed endlessly. 🙂
I agree with ABO
I don't really care what it is called, or how it is sampled. It may be wrong, but it is lot easier than trying to remember voltage-series or current-shunt, or whatever terminology that our text books from university days used. I had two texts that each used different terminology.
I'm all for simplicity.
Huey, Dewey and Louie would also work if it was industry standard.
Jocko
I don't really care what it is called, or how it is sampled. It may be wrong, but it is lot easier than trying to remember voltage-series or current-shunt, or whatever terminology that our text books from university days used. I had two texts that each used different terminology.
I'm all for simplicity.
Huey, Dewey and Louie would also work if it was industry standard.
Jocko
Christer said:
Hi Christer...,
....the type of feedback is defined by the feedback loop's transfer function.......This is sacrosanct.....
Mike.
mikeks said:
According to you, perhaps. There has been plenty of discussion
before on this topic and both I and others have referred to
old textbooks using the term current feedback in a different
way than the op amp manufacturers. However, none of these
books seem to agree with your definition. There seems to be
a rich and diversified terminology in the literature, though, so
perhaps one can find a book that agrees with you too.
- Status
- Not open for further replies.