Home / Learn / Exam prep / GATE Physics / Lesson 9.2
All lessons

Module 9: Electronics · Lesson 9.2 · about 45 min

Amplifiers and feedback

Transistor amplifiers and negative feedback at GATE depth: common-emitter, common-base and follower stages, emitter degeneration, the differential pair and CMRR, low- and high-frequency cut-offs with the Miller effect, the feedback equation with desensitivity and the constant gain-bandwidth product, the four feedback topologies and their input and output resistances, and stability through the loop gain and phase margin.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 9 · about 45 minutes. Lesson 9.1 gave the small-signal parameters gm, rπ and ro. Here they become amplifiers, and then feedback turns an amplifier with a large but unreliable gain into one with a smaller, precise gain. The JAM transistors lesson covers biasing and the load line.

Negative feedback spends gain to buy almost everything else. Divide the gain by the loop factor 1 + Aβ, and the gain becomes insensitive to the transistor, the bandwidth grows by the same factor, distortion shrinks by it, and input and output resistances move by it in the useful direction. The price is a risk: if the loop phase reaches −180° while the loop gain is still above 1, the amplifier oscillates.

1. Single-transistor stages

With the hybrid-π model, three BJT configurations (and their FET twins) cover nearly every question:

Stage Voltage gain Input resistance Output resistance Use
common emitter (source) −gm(RC∥ro) rπ (FET: ∞) RC∥ro voltage gain
common base (gate) +gmRC ≈1/gm ≈RC low input resistance, wide bandwidth
common collector (drain), follower gmRE1+gmRE≈1 rπ+(β+1)RE ≈1/gm+Rs/(β+1) buffer

Emitter degeneration. An unbypassed resistor RE in the emitter is local series feedback:

Av=−gmRC1+gmRE≈−RCRE(gmRE≫1),Rin=rπ+(β+1)RE

Worked example. IC=1 mA (gm=40 mS), RC=5 kΩ, RE=200 Ω: Av=−200/9=−22.2, close to −RC/RE=−25. Without RE the gain is −200, but it would change with temperature and with every transistor; with RE it depends mostly on two resistors.

The differential pair. Two matched transistors share a tail current source of resistance REE. A difference signal sees the emitters as a virtual ground; a common signal sees 2REE in each emitter:

Ad=gmRC2  (single-ended),Acm≈−RC2REE,CMRR=|AdAcm|≈gmREE

With IC=0.5 mA per side (gm=20 mS), RC=10 kΩ and REE=50 kΩ: Ad=100, Acm=0.1, CMRR ≈ 1000 = 60 dB. A transistor current source in the tail makes REE huge, which is why op-amp input stages reject common-mode signals so well.

2. Frequency response

Low frequencies. Coupling and bypass capacitors make high-pass corners at f=1/2πRseenC, with Rseen the resistance the capacitor sees with the signal sources set to zero. The emitter bypass capacitor sees the small resistance 1/gm+Rs/(β+1) and usually sets fL.

High frequencies. The device capacitances Cπ and Cμ make low-pass poles. Cμ connects the input to an output that swings −Av times as much, so the current through it is (1+Av) times larger than for a grounded capacitor. This is the Miller effect:

Cin=Cπ+Cμ(1+|Av|),fH≈12πRs′Cin

A common-base stage or a cascode (common emitter feeding common base) avoids it: the input transistor then sees a gain of only about −1, so its Cμ is barely multiplied.

3. The feedback equation

Feed a fraction β of the output back and subtract it from the input. With forward gain A:

Af=A1+Aβ→Aβ≫11β,dAfAf=11+AβdAA
`$T = A\beta$`
loop gain: the gain around the loop with the input set to zero
`$1 + A\beta$`
the amount of feedback, or desensitivity factor: everything good about feedback scales with it
`$A_f \approx 1/\beta$`
the closed-loop gain is set by the feedback network, usually two resistors
bandwidth
for a single-pole amplifier, `$f_{H,f} = f_H(1 + A_0\beta)$` while the gain falls by the same factor: the gain-bandwidth product `$A_0f_H$` is constant
distortion and interference
disturbances added inside the loop are divided by `$1 + A\beta$`; noise added at the input is amplified like the signal and is not reduced

Worked example. A=105, β=0.01: Af=105/1001=99.90. Halve A to 5×104: Af=99.80, a change of 0.1% for a 50% change in A. If the amplifier has a single pole at 10 Hz, the closed-loop bandwidth is 10×1001≈10 kHz, and the product 1 MHz is unchanged.

Four topologies. Feedback can sample the output voltage (shunt at the output) or current (series), and mix at the input as a voltage (series) or a current (shunt). Each moves the resistances by 1+Aβ:

Topology (mixing-sampling) Stabilises Rin Rout Ideal amplifier
series-shunt voltage gain × (1 + Aβ) ÷ (1 + Aβ) voltage amplifier
shunt-series current gain ÷ × current amplifier
series-series transconductance × × voltage to current
shunt-shunt transresistance ÷ ÷ current to voltage

The rule: series connections raise the resistance at that port, shunt connections lower it. A non-inverting op-amp amplifier is series-shunt; an inverting one with its input resistor removed is shunt-shunt (a transimpedance amplifier); the degenerated common emitter is series-series.

4. Stability

A real amplifier has several poles. Each contributes up to −90° of phase, so with two or more the loop gain A(jω)β can reach −180°. There the “negative” feedback has become positive. If |Aβ|≥1 at that frequency, the loop sustains a signal by itself: oscillation (lesson 9.4 uses this on purpose).

phase margin=180∘+∠Aβ  at |Aβ|=1,gain margin=−20log10⁡|Aβ|  at ∠Aβ=−180∘

On a Bode plot the loop gain is the distance between the |A| curve and the horizontal line 1/β (in dB). They cross where |Aβ|=1. If they cross on a −20 dB/decade stretch, the phase there is near −90° and the margin is comfortable; on a −40 dB/decade stretch, the margin is small and the closed-loop response peaks.

Worked example. A0=105 with poles at 10 Hz and 1 MHz. For a gain of 100 (1/β = 40 dB) the crossover is at 10 kHz, where the second pole adds only 0.6°: phase margin 89°. For a unity-gain follower (β=1) the crossover moves to 786 kHz, and the phase margin drops to 52°. Op-amps are compensated with a dominant pole low enough that even the follower, the hardest case, has a margin of about 45° or more.

Bode plot of an amplifier with two or three poles: open-loop gain, the line 1/β, and the closed-loop gain, which follows 1/β until it meets the open-loop curve. The lower panel shows the loop phase and the phase margin at the crossing. Lower the closed-loop target or bring the second pole down to see the margin collapse and the closed-loop response peak.

“More feedback is always better.” More loop gain gives more precision, but it pushes the crossover frequency up into the region where the higher poles add phase. A follower is the most demanding configuration for stability, not the easiest: it has the largest loop gain. Capacitive loads, which add a pole at the output, make it worse.

  • CE: −gm(RC∥ro); degenerated −gmRC/(1+gmRE)≈−RC/RE; CB: input 1/gm; follower: gain ≈ 1, input (β+1)RE.
  • Differential pair: Ad=gmRC/2 single-ended, CMRR ≈ gmREE.
  • Miller: Cin=Cπ+Cμ(1+|Av|); cascode and CB avoid it.
  • Af=A/(1+Aβ)→1/β; sensitivity divided by 1+Aβ; single pole: gain × bandwidth constant.
  • Series raises, shunt lowers the resistance at that port, by 1+Aβ.
  • Phase margin = 180° + ∠Aβ where |Aβ|=1; the follower is the worst case.

Practise: exam questions on this lesson

Next: lesson 9.3 applies these ideas to the op-amp, where Aβ is so large that the closed-loop gain is set by the resistors alone. The filter designer shows Bode plots of real filters.

Going on to CSIR NET? Noise, filtering and lock-in detection asks how small a signal these amplifiers can recover.

Check yourself

1An amplifier with open-loop gain 10⁴ is placed in a loop with β = 0.01. If the open-loop gain falls by 20%, the closed-loop gain falls by about
2Series (voltage) feedback at the input and shunt (voltage) sampling at the output, with loop gain 1 + Aβ = 50, change the input and output resistances by
3A common-emitter stage has gain −100 and C_μ = 2 pF between base and collector. The Miller capacitance seen at the base is

Sources

  • A. S. Sedra and K. C. Smith, Microelectronic Circuits, 7th ed. (2015), chapters 7, 9, 10 and 11
  • P. Horowitz and W. Hill, The Art of Electronics, 3rd ed. (2015), chapters 2 and 4
  • P. R. Gray, P. J. Hurst, S. H. Lewis and R. G. Meyer, Analysis and Design of Analog Integrated Circuits, 5th ed. (2009), chapters 8 and 9
  • GATE Physics (PH) syllabus, Electronics section, published by the organising IIT each year

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. Not affiliated with the IITs, IISc or the GATE organising institute. The general aptitude part of GATE is not covered here. Check the official GATE website for the current pattern. All questions are original.