Electrical Engineering › Section Q · Thyristors, operational amplifiers, and power conversion › Topic Q.5
Negative feedback and divided feedback
An op-amp on its own slams from rail to rail. Feed part of its output back to the inverting input and it settles instead at exactly the output that keeps its two inputs equal, giving gains set by two resistors and almost nothing else.
Closing the loop
Connect an op-amp's output to its inverting input and apply a signal to the noninverting input. If the output rises, it raises the inverting input and so shrinks the difference that drives it; if it falls, the difference grows and pushes it back up. The output quickly settles where it is just large enough to sustain itself. Feeding the output back so that it opposes the change that produced it is negative feedback, the same principle that steadies thermostats, cruise control and countless other control systems. It is what lets an op-amp work in its linear region instead of as the comparator of Single-ended, differential, and operational amplifiers.
How close is close?
With the output tied straight to the inverting input, \(V_- = V_{out}\), so \(V_{out} = A_{OL}(V_{in} - V_{out})\). Solving gives
With \(A_{OL} = 100{,}000\) and a 4 V input, the output is 3.99996 V, and the 40 µV left between the inputs is exactly what the op-amp needs to hold that output. If \(A_{OL}\) were 200,000, the output would move only 20 µV closer. For practical work, then, negative feedback holds the two inputs at the same voltage, however much the open-loop gain varies from part to part.
The two golden rules
Ideal op-amp analysis with negative feedback rests on two rules. First, no current flows into either input. Second, the output does whatever it must to make \(V_+ = V_-\); the inputs behave as if joined by a virtual short, although no current passes between them. Both hold only while the output stays between its saturation limits. A 741 on ±15 V supplies saturates at roughly ±13 V, while rail-to-rail op-amps get within millivolts of the rails; once the output hits a limit, the feedback can no longer keep the inputs equal, and the rules fail.
The voltage follower
The circuit with the output tied to the inverting input is the voltage follower: gain 1, very high input impedance and low output impedance, the op-amp version of the common-collector amplifier. Its value is buffering. A sensor with 100 kΩ of internal resistance connected straight to a 1 kΩ load delivers only \(1/101\), about 1%, of its voltage. Through a follower, the sensor sees the op-amp's nearly infinite input impedance and supplies almost no current, while the op-amp supplies the load's current from its own power rails. The load then receives essentially the full signal.
Divided feedback: the noninverting amplifier
Feed back only a fraction of the output through a divider: \(R_f\) from the output to the inverting input, and \(R_g\) from the inverting input to ground. The signal still drives the noninverting input. By the golden rules, \(V_- = V_{in}\), so a current \(V_{in}/R_g\) flows through \(R_g\) to ground, and because no current enters the input, the same current flows through \(R_f\). The output therefore sits at \(V_{in}\) plus the drop across \(R_f\):
The gain is positive and never less than 1; with \(R_f = 0\) the circuit is the follower again.
Worked example: a noninverting amplifier
An op-amp on ±12 V supplies has \(R_f = 10\,\text{k}\Omega\), \(R_g = 2.2\,\text{k}\Omega\) and a 1.5 V input. 1. Gain: \(A_V = 1 + 10/2.2 = 5.55\). 2. Current: \(I = 1.5/2200 = 0.682\,\text{mA}\) through \(R_g\) and the same through \(R_f\). 3. Output: \(1.5 + (0.682\,\text{mA})(10\,\text{k}\Omega) = 8.32\,\text{V}\), inside the ±12 V supply. 4. Finite gain: with the feedback fraction \(B = R_g/(R_g + R_f)\), written B as in Feedback, amplifier impedances, current mirrors, ratings, and quirks, and \(A_{OL} = 100{,}000\),
only 0.006% below \(1/B = 5.5455\). Check: the signal source supplies no current; the 0.682 mA comes from the op-amp's output.
The inverting amplifier
Now ground the noninverting input and apply the signal through \(R_{in}\) to the inverting input, with \(R_f\) from output back to that input. Feedback holds the inverting input at 0 V, a virtual ground: at ground potential but not connected to ground. The input current is \(V_{in}/R_{in}\); it cannot enter the op-amp, so it continues through \(R_f\), and the output must sit that far below ground:
The minus sign records the inversion. Unlike the noninverting circuit, the gain magnitude can be below 1. The source sees an input resistance of just \(R_{in}\), since its far end is at virtual ground.
Worked example: an inverting amplifier
An inverting amplifier on ±12 V supplies has \(R_{in} = 4.7\,\text{k}\Omega\), \(R_f = 22\,\text{k}\Omega\) and \(V_{in} = +0.80\,\text{V}\). 1. Virtual ground: \(V_- = 0\). 2. Input current: \(I = 0.80/4700 = 0.170\,\text{mA}\), flowing towards the virtual ground. 3. The same current flows on through \(R_f\) towards the output, so
4. Gain: \(-22/4.7 = -4.68\), and \(-4.68 \times 0.80 = -3.74\,\text{V}\). 5. Limit: if the output saturates near −11 V, the largest usable input is \(11/4.68 = 2.35\,\text{V}\). Check: positive in, negative out.
A lever picture
A mechanical lever makes the two gains easy to remember. Draw the feedback divider as a bar whose lengths stand for resistances, with the pivot at the point held at zero volts. In the noninverting amplifier the pivot is the grounded end of \(R_g\); the input point is \(R_g\) from the pivot and the output end is \(R_g + R_f\) away, so the output moves further in the same direction, by \((R_g + R_f)/R_g\). In the inverting amplifier the pivot is the virtual ground between the resistors, so the ends move in opposite directions in the ratio \(R_f/R_{in}\), which can be more or less than 1.
Common mistakes to avoid
First, writing the noninverting gain as \(R_f/R_g\): it is \(1 + R_f/R_g\). Second, forgetting the sign of the inverting gain. Third, assuming the inverting input is connected to ground: it is a virtual ground, held at 0 V by feedback. Fourth, trusting the golden rules when the output is saturated: then the inputs are no longer equal. Fifth, thinking the input source supplies the load current: in every one of these circuits the current comes from the op-amp's supply rails.
Key takeaways
Negative feedback returns the output to the inverting input so that it opposes changes, which makes an op-amp linear and self-correcting. With it, no current enters the inputs and the output forces V+ = V−, as long as the output is not saturated. The follower has gain 1 and buffers weak sources. The noninverting amplifier has gain 1 + R_f/R_g, never below 1; the inverting amplifier has gain −R_f/R_in, a virtual ground and input resistance R_in. The closed-loop gain A_OL/(1 + A_OL B) is almost exactly 1/B, so resistors, not the op-amp, set the gain.
Flashcards
Flashcard
Output returned to the inverting input so that it opposes the change that produced it
Test your understanding
Question 1 of 5
A noninverting amplifier has R_f = 33 kilohms and R_g = 3.3 kilohms. What is its output for a +0.25 V input?
Question 2 of 5
An inverting amplifier has R_in = 10 kilohms and R_f = 4.7 kilohms. What is its output for a −2.0 V input?
Question 3 of 5
A voltage follower uses an op-amp with an open-loop gain of 50,000 and has a 2.0 V input. About how much voltage remains between the op-amp's two inputs?
Question 4 of 5
An inverting amplifier has R_in = 4.7 kilohms and R_f = 47 kilohms. What input resistance does the signal source see?
Question 5 of 5
Why does the gain of a negative-feedback op-amp amplifier hardly change if the op-amp's open-loop gain doubles?
