No, I don't think I follow your reasoning.
The debate is whether an emitter follower is a negative feedback system with 100% negative feedback (B=1). But you have started your argument by defining B=1.
That's a problem because I'm not sure how we work the problem if we start with the answer.
Ok. Say again which node you are defining as the input and which node as the output. You seem to be implying they are the same node...but that doesn't make sense to me. Why isn't the input node the base and the output node the emitter?
That site says: "The negative feedback effect due to $R_E$ can be shown qualitatively: ".
Qualitatively is the root of many frustrating threads in diyAudio.
I need to see a mathematical proof. Call me picky.
Dear Moderator, I think you just called me snide and lazy? Sorry if I am mis-interpreting you.
In the forum rules page I see "Trolling is posting inflammatory, extraneous, or off-topic messages in an online community with the intent of provoking other users into an emotional response or of otherwise disrupting normal on-topic discussion."
I would welcome your comments on emitter followers...please do join in on that.
100% negative feedback means that the whole output signal, unattenuated by any resistive divider etc., is applied to the input. Now 'applied to the input' does not necessarily mean 'applied to the same point as the input', but it does mean 'applied to the input signal'. In the case of the emitter follower (and the non-inverting opamp) the input signal goes to one connection - base (or non-inverting input) - and the feedback goes to another connection - emitter (or inverting input). In both cases the input signal seen by the active device - BJT (or opamp) - is the difference between these two points. Hence, mathematically, the feedback signal is subtracted from the input signal.traderbam said:
Does this help?
Not yet. I am not happy to treat the input signal as the base-emitter voltage. You see, if I understand you correctly, Wouldn't this explanation equally apply to a simple resistor divider (where the transistor is just a resistor)? I don't consider a resistor divider to be a negative feedback circuit.
You will know it when I am posting as the moderator.
But seriously, if you have read the linked page in its entirety, then do you have any objection to the content? If so, then perhaps you can enlighten us with your interpretation of the emitter follower.What do you propose as the input to a BJT? You need two connections, as voltage always needs two points.traderbam said:
The input to the circuit is between base and ground, but the transistor unavoidably has its input signal with feedback already subtracted.
Generally we don't talk about feedback when there are no active devices. However, if you treat a resistor as being a transconductance device (voltage in, current out) then the algebra for feedback should still work. Please be clear: I am not saying that feedback is the only way to analyse an emitter follower but it it is a valid way and it correctly predicts all the outcomes: reduced gain, raised input impedance, lowered output impedance etc. It is certainly false to say that an emitter follower does not involve feedback, as some seem to claim.
You are always in uniformYou will know it when I am posting as the moderator.
. I don't object to that site in general except that I am unable, at the moment, to understand its qualitative claim that the emitter follower uses "deep negative feedback". I'll either get there or I'll try to show the error of that sites ways. I first want to listen to what other folks think so I don't overlook anything.
Agreed. Before I forget my thread, one argument I would make is that if this method of imbuing a circuit with 100% NFB also imbues a simple resistor divider with 100% NFB then I suggest the method is invalid. And if it is invalid then we cannot use it to claim an emitter follower uses 100% NFB. We need another way.
I'll have a go at answering your question about inputs and outputs soon.
That site that Jan linked to is Harvey Mudd College, it's no MIT, but probably still have professors/TA's capable of explaining how an emitter follower works. You can also search the terms yourself and you will find many course syllabus/handouts from various colleges - they will all say pretty much the same thing, because that's what they teach in undergrad EE courses.
You can also search the terms yourself and you will find many course syllabus/handouts from various colleges - they will all say pretty much the same thing, because that's what they teach in undergrad EE courses.I am not saying that a circuit has feedback because the feedback equation works. I am saying that the feedback equation working is consistent with feedback being present. We normally talk about feedback when amplifiers are in mind. An emitter follower contains an amplifier (the BJT) and is itself an amplifier; a resistive divider does not.traderbam said:
The output signal (at the emitter) is 100% applied to the input (at the emitter). In what sense is this not feedback, and not 100% feedback?
I agree with the gist but I don't yet agree this is a definitive argument. I would say we talk about NFB when there is forward gain and a portion of the output is subtracted from the input to the amplifier. The emitter follower does not have forward voltage gain from base to emitter. It has current gain but that is not what we are talking about. Nor does it have a discrete feedback path from emitter to base via a subtractor.
There is obviously no discrete feedback path from emitter to base. If we redefine what a feedback path is and say it is the same thing as the forward path you have to conclude that a resistor is a voltage feedback device. So this is not useful.
I would say if the output signal is also the input signal, if both are at the emitter node, then the requirement of a portion of the output being subtracted from the input does not apply and it is not a NFB circuit. I know you will say that the input is not actually the emitter node but is Vbe; I don't think you can argue it both ways.
I feel the (non-algebraic) answer lies in the argument that if a resistor is not a NFB device then an emitter follower isn't either. Eg: if an emitter follower behaves like a pair of resistors, for small signals.
You must have led a very sheltered life. Business is nearly synonymous with BS in my experience
Both are true of the emitter follower, in exactly the same way that both are true of the non-inverting opamp. Take a unity gain non-inverting opamp (so a wire from output pin to inverting input pin) and you have the same signal circuit as the emitter follower. The only difference is that in the emitter follower you cannot remove the wire.traderbam said:
The BJT has forward transconductance gain. As soon as you add a resistor you get voltage gain. You also get voltage feedback. It is the fact that these two features come along together which seems to confuse people.
Don't fall into the trap of thinking that feedback is only feedback if it is applied to the same wire as the signal input. If you insist on this then the non-inverting opamp configuration does not use feedback, neither does the common 'feedback to first cathode' used in many valve amps.
The input to the BJT is the voltage from base to emitter. Hence feedback can be applied at either base or emitter. Feedback theory classifies these as shunt and series feedback, respectively.
As I said, there is a sense in which a resistor is voltage feedback device. We don't use the term 'feedback' because it is a passive component.
I am not arguing it both ways. I am saying that the input to the BJT is the difference between base and emitter voltages. Hence signal and/or feedback can be applied at either. In the case of the emitter follower the signal goes to the base and the feedback goes to the emitter. Now think about what signal the BJT sees; it is Vbase-Vemitter i.e. (input signal)-(feedback signal). What is the feedback signal? It is the output signal, so we have 100% feedback.
This is a false argument because you are assuming that a circuit cannot be described in more than one way. It is true that an emitter follower can be considered to be a pair of resistors (at low frequencies where device parasitics play no role). Similarly, a pair of resistors can be considered using feedback theory. Go to higher frequencies where the BJT has more complex behaviour and you must use feedback theory; you can even get instability due to phase shift around the feedback loop - this is well known for any follower circuit, especially with a capacitive load. Your simple resistors theory cannot handle this; feedback theory can.
My argument that an emitter follower does not use negative feedback depends on the premise that a resistor is not a feedback device.
Here is a definition of feedback:
https://en.wikipedia.org/wiki/Negative_feedback
It follows that:
Ohm's Law is not caused by feedback.
Ohm's Law does not depend on feedback in any way.
A resistors function does not require feedback and resistors are not negative feedback systems.
If you still believe that resistors are feedback systems then I think we lack the common ground to usefully discuss this.
Ohm's Law is caused by random collisions. The circuit effects of Ohm's Law may be handled by using the algebra of feedback. Consider two resistors forming a potential divider, R1 and R2. The incoming signal is applied to the top of R1. The bottom of R2 is grounded.
R1 is a transconductance device, because the current through it is set by the voltage across it. R2 is a resistance device, because the voltage across it is set by the current through it. Now apply the feedback equation, with A=R2/R1 (the forward gain) and B=1 (100% negative feedback).
closed loop gain = A/(1+AB) = R2/R1 x 1/(1+R2/R1) = R2/(R1+R2)
This is the result you could obtain by using Ohm's Law and Kirchoff's laws.
This does not prove that a resistor is a feedback device. It merely shows that the theory of feedback can be properly applied to circuits which are not usually considered to use feedback.
Now of course those who know that a BJT can be modelled by a (sort of) resistor also know that an emitter follower can be modelled as a potential divider. Note that it isn't really a resistor because a real resistor would consume from the signal through the base the same current as flows out of the emitter. It is a transresistor - usually shortened to transistor - as the current here depends on the voltage somewhere else.
Your argument seems to be that as an emitter follower can be modelled as a potential divider (although only at low frequencies) then it must be modelled as a potential divider and cannot involve feedback. But that same argument shows that a potential divider (which can be modelled using feedback theory) must be a feedback system. In a sense it is, as the current through R1 is modified and reduced by the voltage developed across R2.
So we have to agree that an emitter follower (at low frequencies) and a potential divider are both feedback systems and Ohm's Law systems. As always in science, you use the model which best fits and gives the easiest sums.
Now, as I said, the non-inverting opamp buffer is the same circuit as an emitter follower. By your logic the non-inverting opamp is not a feedback system. At high frequencies the potential divider model for an emitter follower breaks down, but the feedback model still gives the right results. The active device sees only the difference between the input signal and the output signal: that is feedback.
R1 is a transconductance device, because the current through it is set by the voltage across it. R2 is a resistance device, because the voltage across it is set by the current through it. Now apply the feedback equation, with A=R2/R1 (the forward gain) and B=1 (100% negative feedback).
closed loop gain = A/(1+AB) = R2/R1 x 1/(1+R2/R1) = R2/(R1+R2)
This is the result you could obtain by using Ohm's Law and Kirchoff's laws.
This does not prove that a resistor is a feedback device. It merely shows that the theory of feedback can be properly applied to circuits which are not usually considered to use feedback.
Now of course those who know that a BJT can be modelled by a (sort of) resistor also know that an emitter follower can be modelled as a potential divider. Note that it isn't really a resistor because a real resistor would consume from the signal through the base the same current as flows out of the emitter. It is a transresistor - usually shortened to transistor - as the current here depends on the voltage somewhere else.
Your argument seems to be that as an emitter follower can be modelled as a potential divider (although only at low frequencies) then it must be modelled as a potential divider and cannot involve feedback. But that same argument shows that a potential divider (which can be modelled using feedback theory) must be a feedback system. In a sense it is, as the current through R1 is modified and reduced by the voltage developed across R2.
So we have to agree that an emitter follower (at low frequencies) and a potential divider are both feedback systems and Ohm's Law systems. As always in science, you use the model which best fits and gives the easiest sums.
Now, as I said, the non-inverting opamp buffer is the same circuit as an emitter follower. By your logic the non-inverting opamp is not a feedback system. At high frequencies the potential divider model for an emitter follower breaks down, but the feedback model still gives the right results. The active device sees only the difference between the input signal and the output signal: that is feedback.
If we can't say an amplifier circuit having "no feedback", then the same could be argued to apply for when "open loop gain" is discussed, because as someone mentioned:
"in the emitter follower you cannot remove the wire"
hence, from that, one could conclude we can not open up all loops, so therefore such thing as "open loop" can not exit.
We therefore could also argue that we can't fully know what the true open loop gain would be for example an op-amp even if the manufacturer state some still useful figures in their datasheet, if it can be open loop at all.
Loops and feedbacks goes hand in hand.
Also as other have mentioned, resistors have feedback too, then lets bring forward an example where a BJT is used in a CE configuration, we connect E to GND, C to V+, and B is signal input, in this case the transistor is "shorted" so no voltage gain but current gains could be observed.
Now let's add a collector resistor, we may now ask is the voltage over the resistor produced by the BJT, or is it the resistor which is trying to maintain a low current through it by increasing the voltage fall? Which one is providing for the feedback mechanism so there will be a voltage gain? But it doesn't stop only here.
What I am asking above is not the answer, but where do we draw the boundaries in such discussion regarding "feedbacks" and "loops", think of a Mandelbrot, we dive deeper into it and the pattern repeats over and over again, down to an atomic level, and who knows maybe there are even smaller bits and pieces well beyond the atoms which provides for the possibility to maintain feedback actions, eg. we reach to a new bottom of "academic Asperger, drivel and nonsense" when peoples vocabulary is exempt of the word "colloquial" and start nitpicking and going on a n/\zi barrage because of some people are using the "non feedback" term to explain their audio amplifier circuit for an intrigued public, talk about party pooper.
I would still like to maintain what I earlier said, the colloquial use of "no feedback" amplifiers are still valid and widely understood, and it is not meant to be confused with academic textbook examples wherein for example the intrinsic feedback mechanisms in BJT's is discussed, which I have never opposed to and understand very will its use and value.
I have seen even Mr. Pass colloquially using "no feedback" when describing circuits in the context of audio amplifiers.
"in the emitter follower you cannot remove the wire"
hence, from that, one could conclude we can not open up all loops, so therefore such thing as "open loop" can not exit.
We therefore could also argue that we can't fully know what the true open loop gain would be for example an op-amp even if the manufacturer state some still useful figures in their datasheet, if it can be open loop at all.
Loops and feedbacks goes hand in hand.
Also as other have mentioned, resistors have feedback too, then lets bring forward an example where a BJT is used in a CE configuration, we connect E to GND, C to V+, and B is signal input, in this case the transistor is "shorted" so no voltage gain but current gains could be observed.
Now let's add a collector resistor, we may now ask is the voltage over the resistor produced by the BJT, or is it the resistor which is trying to maintain a low current through it by increasing the voltage fall? Which one is providing for the feedback mechanism so there will be a voltage gain? But it doesn't stop only here.
What I am asking above is not the answer, but where do we draw the boundaries in such discussion regarding "feedbacks" and "loops", think of a Mandelbrot, we dive deeper into it and the pattern repeats over and over again, down to an atomic level, and who knows maybe there are even smaller bits and pieces well beyond the atoms which provides for the possibility to maintain feedback actions, eg. we reach to a new bottom of "academic Asperger, drivel and nonsense" when peoples vocabulary is exempt of the word "colloquial" and start nitpicking and going on a n/\zi barrage because of some people are using the "non feedback" term to explain their audio amplifier circuit for an intrigued public, talk about party pooper.
I would still like to maintain what I earlier said, the colloquial use of "no feedback" amplifiers are still valid and widely understood, and it is not meant to be confused with academic textbook examples wherein for example the intrinsic feedback mechanisms in BJT's is discussed, which I have never opposed to and understand very will its use and value.
I have seen even Mr. Pass colloquially using "no feedback" when describing circuits in the context of audio amplifiers.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.