Op amps, negative feedback or not?

[kougou]

Homework Statement

What is the exact definition of negative feedback? the instructor told us that negative exists if the output is connected back to the inverting input through a passive element. But what exactly is precise the definition ? (say parallel connect, same two nodes, I want the definition for negative feedback).

https://d1b10bmlvqabco.cloudfront.n...6vo/h8qxj4erju92rw/h8qxjbmfsg830v/photo_3.JPG
is this negative feedback for the first op amp?
How do you know?

 

[Simon Bridge]

http://en.wikipedia.org/wiki/Negative_feedback
That would be pretty precise for a definition.

In your example - the op amp in the dashed lines is in negative feedback because the inverting input is connected to the output via a passive element (the resistor). The other op-amp is not because it's output has to pass through an active element (an op-amp) first. That help?

 

[nsaspook]

http://en.wikipedia.org/wiki/Miller_theorem

 

[skeptic2]

the instructor told us that negative exists if the output is connected back to the inverting input through a passive element.

I think your instructor's definition is a bit too restrictive but perhaps he was trying to keep it simple in order not to get too confusing.

I would suggest that negative feedback is when the output is combined with the input in such a way as to oppose the action of the input. For instance, suppose you have a circuit in which an opamp is driving a transistor in order to increase the opamp's current capability. If the transistor inverts the signal then the signal at the collector would be connected to the non-inverting input for negative feedback. And since the transistor is part of the feedback loop, the feedback element does not have to be passive.

 

[haruspex]

I think your instructor's definition is a bit too restrictive

It's not clear to me from the OP whether the instructor defined it that way or was merely stating a sufficient condition.

 

[Simon Bridge]

I doubt it is a definition, I think the instructor tying to tell the students how to tell if an op-amp is configured for negative feedback. Anybody else looked at the example yet?

 

[aralbrec]

http://en.wikipedia.org/wiki/Negative_feedback
That would be pretty precise for a definition.

In your example - the op amp in the dashed lines is in negative feedback because the inverting input is connected to the output via a passive element (the resistor). The other op-amp is not because it's output has to pass through an active element (an op-amp) first. That help?

The wikipedia link is correct but the business about passing through an active element is not. I can tell you signals pass through many active elements in negative feedback electronic systems all the time. When your instructor included the active element caveat, I am sure he meant there was no opportunity for signal inversion (ie making it negative).

As usual, the wikipedia article is correct but often clear as mud because of the language used. The diagram on the page showing a general feedback model is what you should look at. If the loop gain (ie the gain from the output of the adder, passed through A to the output and then passed through B back to the point where it gets added to the input again) is negative, you have negative feedback. If it is positive, you have positive feedback.

If the loop gain is negative, the output of the adder can be thought of as an error between the output and the input. B in this case takes some sort of measurement of the output and compares it to what is desired by *subtracting* it from the input (this for negative feedback). The error is then amplified by some huge amount** in A. This huge amount guarantees that the error will stay close to zero. This is the idea behind negative feedback.

(** a huge amount may have undesirable consequences such as overshoot and ringing as the system seeks zero error)

As mentioned, for negative feedback to occur, the loop gain must be negative (again this is the gain from the error summer through A and B). In your notes, the loop gain is the gain from the output of the first amplifier (Voa), which is calculating an error through subtraction, to Vna. This gain must be *positive* to have negative feedback occurring. It's positive this time because the first amp is subtracting this number from the input. So a positive error from the first error amp gets positive gain from the second amp and is subtracted from the input. Higher error acts to reduce the error, the definition of negative feedback.

When you start adding capacitors or inductors into the loop, you find that your loop gain will become frequency dependent. This means it is possible for negative feedback to turn into positive feedback for certain signal frequencies (eg, a formerly positive quantity could become negative with sufficient phase shift). Several of your courses will expend considerable time on making sure the overall system will be stable despite this frequency dependence of the loop gain.

Positive feedback is not always bad. Oscillators, for example, depend on positive feedback to start (via thermal noise) and maintain an oscillating signal without input. In this case the loop gain is designed to generate the right amount of positive feedback to sustain oscillation at a specific frequency.

 

[kougou]

It's not clear to me from the OP whether the instructor defined it that way or was merely stating a sufficient condition.

sorry guys. The instructor doesn't give precise definition for negative feedback. The text says negative feedback is used to maintain the input voltage ( the different between inverting and non-inverting) to zero by feeding back a portion of voltage back to the inverting terminal of the op amp. I have trouble understanding the description above.

Also, in the image I gave, the second op amp (inside the dot line) is actually a inverting op amp. the inverting op amp is then connecting back to the inverting terminal of the first op amp. The TA says negative * negative = position. So we have positive feedback. This even makes the whole thing more complicate.

But actually, for negative feedback, does the the output has to connect back to inverting terminal through a passive element? Say if I change the problem and replace the op amp inside the dot line with a non-inverting op amp. This should give negative feedback to the first op amp?

 

[kougou]

I doubt it is a definition, I think the instructor tying to tell the students how to tell if an op-amp is configured for negative feedback. Anybody else looked at the example yet?

That is exactly what I thought. I don't think it defines properly. Probably because we only deal with op amp having negative feedback, so we ignore the rest.

 

[kougou]

The wikipedia link is correct but the business about passing through an active element is not. I can tell you signals pass through many active elements in negative feedback electronic systems all the time. When your instructor included the active element caveat, I am sure he meant there was no opportunity for signal inversion (ie making it negative).

As usual, the wikipedia article is correct but often clear as mud because of the language used. The diagram on the page showing a general feedback model is what you should look at. If the loop gain (ie the gain from the output of the adder, passed through A to the output and then passed through B back to the point where it gets added to the input again) is negative, you have negative feedback. If it is positive, you have positive feedback.

If the loop gain is negative, the output of the adder can be thought of as an error between the output and the input. B in this case takes some sort of measurement of the output and compares it to what is desired by *subtracting* it from the input (this for negative feedback). The error is then amplified by some huge amount** in A. This huge amount guarantees that the error will stay close to zero. This is the idea behind negative feedback.

(** a huge amount may have undesirable consequences such as overshoot and ringing as the system seeks zero error)

As mentioned, for negative feedback to occur, the loop gain must be negative (again this is the gain from the error summer through A and B). In your notes, the loop gain is the gain from the output of the first amplifier (Voa), which is calculating an error through subtraction, to Vna. This gain must be *positive* to have negative feedback occurring. It's positive this time because the first amp is subtracting this number from the input. So a positive error from the first error amp gets positive gain from the second amp and is subtracted from the input. Higher error acts to reduce the error, the definition of negative feedback.

When you start adding capacitors or inductors into the loop, you find that your loop gain will become frequency dependent. This means it is possible for negative feedback to turn into positive feedback for certain signal frequencies (eg, a formerly positive quantity could become negative with sufficient phase shift). Several of your courses will expend considerable time on making sure the overall system will be stable despite this frequency dependence of the loop gain.

Positive feedback is not always bad. Oscillators, for example, depend on positive feedback to start (via thermal noise) and maintain an oscillating signal without input. In this case the loop gain is designed to generate the right amount of positive feedback to sustain oscillation at a specific frequency.

I read it very careful. Thank you! Very helpful!

 

[skeptic2]

But actually, for negative feedback, does the the output has to connect back to inverting terminal through a passive element? Say if I change the problem and replace the op amp inside the dot line with a non-inverting op amp. This should give negative feedback to the first op amp?

Exactly, and in the case of the outer loop, you have positive feedback even though it is connected to the inverting input.

This is also why I say negative feedback does not have to be through a passive element. Likewise, and more importantly, positive feedback can occur through active elements and with complex circuits you have to be very careful there are no positive feedback paths. I once had a power controller, using negative feedback, oscillate because of a positive feedback path through the power supply.