Why you need a buffer
The common-emitter amplifier from the last guide is a fine voltage amplifier, but it has a weak handshake: it presents its signal through a high output resistance of roughly the collector resistor (a few kilohms). Connect a small load to that and the loading effect divides your hard-won gain straight back down. The amplifier is like a strong pump bolted to the end of a thin pipe — the pressure is genuinely there, but the moment anything tries to draw real flow, the pressure sags.
What you want is a stage that adds no voltage gain at all — gain of one — but has the muscle the previous stage lacked: a very high input resistance so it barely loads whatever feeds it, and a very low output resistance so it drives the next stage stiffly. That impedance-transforming buffer is the emitter follower. Think of it as impedance matching: choosing a hose that neither chokes the source nor wastes the flow at the load.
The emitter follower: gain of one, all muscle
The wiring is almost embarrassingly simple. Feed the signal into the base; take the output from the emitter, across an emitter resistor; the collector ties straight to the supply. There is no collector resistor and no inversion. The emitter always sits one base-emitter drop (about 0.7 V) below the base, and it tracks the base faithfully up and down — so the output is just the input minus a fixed 0.7 V offset. Voltage gain is a hair under one, and the output is in phase with the input.
Why does the emitter follow so obediently? Local negative feedback. Suppose the output tries to lag behind the rising base: that instantly widens the base-emitter voltage, which (recall the active region) swells the collector and emitter current, which pulls the emitter voltage back up to where it should be. The emitter chases the base like a thermostat forever nudging the room back to the setpoint. In small-signal terms this is full emitter degeneration with no bypass capacitor — every bit of the output is fed back to oppose the input.
Feedback is what gives the follower its muscle on both ends, and here beta works for you instead of against you. Looking into the base, the load on the emitter appears multiplied by beta, so the input resistance is high. Looking back into the emitter, the source resistance appears divided by beta, so the output resistance is low (down to the intrinsic emitter resistance re plus that divided-down source). One transistor turns a feeble high-impedance source into a stiff low-impedance one.
Designing one, with real numbers
Let us buffer a signal that comes from a weak 10 k source. Bias the transistor at a collector current of 1 mA, take beta = 100, use an emitter resistor Re = 4.7 k, and drive a load of RL = 10 k. The one new quantity is the intrinsic emitter resistance re = 25 mV / Ic, which comes from the transconductance (re is just 1/gm). Watch how feedback transforms the impedances.
- Intrinsic emitter resistance: re = 25 mV / 1 mA = 25 ohm. Small, and it sets the floor for the output resistance.
- Output resistance: looking into the emitter, Rout ≈ re + Rsource / beta = 25 + 10000 / 100 = 25 + 100 = 125 ohm. The 10 k source has been shrunk to about 125 ohm.
- Input resistance: the emitter sees Re in parallel with RL, about 3.2 k, so Rin ≈ beta times (re + 3.2 k) ≈ 100 times 3225 ≈ 320 k. The previous stage now barely feels the load.
- Voltage gain: gain ≈ (Re||RL) / (Re||RL + re) ≈ 3200 / 3225 ≈ 0.99 — just under one, exactly as promised.
Read the result like a translator: a 10 k source going in, a 125 ohm source coming out, with the signal essentially untouched. The same 10 k load that would have crushed a bare common-emitter stage is now driven without sag, because the follower's 125 ohm output dwarfs it — the loading effect is tamed. The price you pay is honest and small: no voltage gain, and a fixed 0.7 V DC drop you must allow for in the bias.
The current mirror: copy a current
Now flip the problem: instead of a stiff voltage, you often want a stiff current — a near-ideal current source for biasing. The trick is the current mirror: two matched transistors that share the exact same base-emitter voltage. One of them, diode-connected (base tied to collector), is forced to pass a reference current set by a resistor, and in doing so it 'measures out' the Vbe that produces that current. The second transistor, handed that identical Vbe, has no choice but to pass the same collector current. You measure a current once and clone it.
+Vcc
|
[ R ] Iref = (Vcc - 0.7 V) / R
|
Iref | Iout (the copy)
---------> +-------+ +-------> to load
| | |
(collector) (collector)
Q1 |=====|====| Q2 bases tied together;
| | Q1 is diode-connected
(emitter) (emitter) (its base ties to its collector)
| |
GND GND
Q1 sets the Vbe; Q2, sharing it exactly, copies the current.
Iout = Iref / (1 + 2/beta) -> beta = 100 gives Iout ~2% low.Notice what the mirror does and does not depend on. There is a small base-current error — the two bases steal a little from the reference node, so Iout = Iref / (1 + 2/beta), about 2 percent low at beta = 100 (a beta-helper or a cascode shrinks even that). But crucially, the mirror leans on matching, not on the absolute value of beta: the two transistors only have to be the same as each other — same die, same temperature — and the copy is faithful even though beta itself is sloppy and drifts. That is the whole philosophy of analog IC design in miniature.
Differential pair, Darlington, push-pull
The same building blocks combine into the heart of every op-amp. A differential pair is two transistors standing on a shared tail current source; it amplifies the difference between its two inputs and rejects whatever they have in common (the basis of common-mode rejection). The tail is usually a current mirror, and a second mirror often serves as the active load that converts the pair's output to a single wire and boosts the gain. If you have ever wondered what is behind an op-amp's two inputs, this is it.
When you need enormous current gain, the Darlington pair cascades two transistors so the first one's collector current becomes the second one's base current — the betas multiply, giving an effective beta of ten thousand or more. The honest costs are real: it now takes two base-emitter drops (about 1.4 V) to turn on, it switches more slowly, the first transistor's leakage gets amplified by the second, and its saturation voltage is higher. Use it for raw current gain, not for precision or speed.
Finally, the push-pull output is just two emitter followers back to back: an NPN follower pushes the output up on the positive half of the swing, and a PNP follower pulls it down on the negative half. Each device handles one half, so the stage can drive heavy loads in both directions efficiently. The catch is the 0.7 V dead zone where neither transistor is on yet — that gap creates crossover distortion, cured by biasing both slightly on (class AB). This is the output stage inside audio amplifiers and op-amps alike, and it is why the humble follower you met earlier is everywhere.
The honest core: don't lean on beta
Step back and notice the thread running through every circuit in this rung. Beta is a terrible number to trust: it ranges from maybe 50 to 400 across parts from the same bin, it climbs and falls with temperature, and it shifts with collector current. So every robust technique you have learned quietly sidesteps it — voltage-divider bias and emitter degeneration set the operating point by resistors, the emitter follower's near-unity gain barely cares about beta, the current mirror leans on matching, and feedback regulates the rest. The rule of thumb: if your circuit only works at beta = 200, it does not actually work.
You now hold the whole BJT toolkit: switch, amplifier, biasing, the inverting common-emitter, and the buffering follower with its mirror, diff-pair, Darlington, and push-pull cousins. The next rung swaps in the MOSFET, whose gate draws essentially no current — an even cleaner switch and buffer — and after that the op-amp, which packages a differential pair, a mirror, and a push-pull output into a single part you bias with two resistors. When you meet the op-amp voltage follower, greet it as exactly what it is: the polished, beta-free cousin of the emitter... follower you just built by hand.