Two junctions, one amplifier
Picture a garden hose with a foot pedal. The water in the hose comes from the mains at full pressure — that's the *big* energy. But you don't control it by wrestling the mains; you press a pedal with one toe, and that gentle nudge releases a torrent. A transistor is electricity's version of that pedal: a weak control signal gates a powerful flow. The bipolar junction transistor (BJT) was the very first device that could do this, and it changed everything.
An NPN BJT is a single semiconductor crystal doped into three layers, like a sandwich: a heavily-doped N-type emitter, a thin, lightly-doped P-type base in the middle, and a moderately-doped N-type collector. That makes *two* back-to-back PN junctions sharing the base: the base–emitter junction and the base–collector junction. The whole art of the BJT is in how those two junctions cooperate.
collector (C)
│
┌───┴───┐
│ N │ ← collector
├───────┤ ← base–collector junction
B ────┤ P │ ← base (thin & lightly doped!)
├───────┤ ← base–emitter junction
│ N │ ← emitter (heavily doped)
└───┬───┘
│
emitter (E)
schematic symbol (NPN): C
│
B ───┤< arrow on emitter
│ ╲ points OUT = NPN
EBeta: how a trickle commands a torrent
Here is the central fact of the BJT, stated plainly. Forward-bias the base–emitter junction (push the base about 0.7 V above the emitter, just like the diode you already know) and electrons flood from the emitter into the thin base. A *few* of them get caught and exit as base current I_B. But the vast majority — 99% or more — sail straight through and are collected as collector current I_C. The ratio between them is a number called beta.
β = I_C / I_B (current gain, dimensionless)
I_C = β · I_B collector follows base, amplified by β
I_E = I_C + I_B emitter carries the sum (KCL)
= (β + 1) · I_B
Example: β = 100, I_B = 20 µA
→ I_C = 100 × 20 µA = 2 mA
→ I_E = 2 mA + 20 µA ≈ 2.02 mAA typical small-signal BJT has a beta somewhere between 100 and 300. That's the lever: wiggle the base current by a microamp and the collector current swings by a hundred microamps. Feed the collector current through a resistor and that swing becomes a voltage swing — and now you have an amplifier, the first one in this whole track.
Three moods: cutoff, active, saturation
A BJT isn't always amplifying. Depending on how the two junctions are biased, it lives in one of three regions of operation — think of them as three moods. Which one you're in is decided entirely by the two junction voltages, and knowing the mood is the first thing you check on any transistor circuit.
- Cutoff — the valve is shut. The base–emitter junction is *not* forward-biased (V_BE below ~0.7 V). No base current, so no collector current: I_C ≈ 0. The transistor looks like an open switch. This is the 'OFF' state in digital logic.
- Active — the valve is amplifying. Base–emitter is forward-biased (V_BE ≈ 0.7 V) *and* base–collector is reverse-biased. Now the beta relation holds cleanly: I_C = β · I_B. This is the region you want for *amplifiers* — the collector current obeys the base like a faithful, amplified echo.
- Saturation — the valve is wide open. You push so much base current that *both* junctions become forward-biased. The collector can't pull any more current (the external resistor limits it), so I_C < β · I_B and V_CE collapses to a small floor (~0.2 V). The transistor looks like a closed switch. This is the 'ON' state in digital logic.
Worked example: finding the bias point
An amplifier only works if its transistor is sitting comfortably in the active region *before any signal arrives* — a resting state engineers call the bias point or Q-point (Q for 'quiescent', meaning 'at rest'). Get it wrong and your amplifier clips, distorts, or doesn't turn on at all. Let's compute one for the classic voltage-divider bias circuit, the workhorse of every common-emitter amplifier.
Vcc = 12 V
│
┌───┴───┐
[R1] [Rc] R1 = 47k, Rc = 2k
47k 2k
│ │
├────────┤────── collector node (Vc)
│ C │
│ B ┌─┴─┐
├──────┤ Q │ NPN, β = 150
│ └─┬─┘
[R2] E │
10k │
│ [Re] Re = 500 Ω
│ │
─┴────────┴───── ground (0 V)- Find the base voltage from the divider. Ignoring the small base current, R1 and R2 form a voltage divider: V_B = Vcc · R2 / (R1 + R2) = 12 · 10k / (47k + 10k) = 12 · 0.175 = 2.1 V.
- Drop 0.7 V across the base–emitter junction. The emitter sits one diode-drop below the base: V_E = V_B − 0.7 = 2.1 − 0.7 = 1.4 V.
- Get the current from the emitter resistor. Ohm's law across Re: I_E = V_E / Re = 1.4 V / 500 Ω = 2.8 mA. Since β is large, I_C ≈ I_E ≈ 2.8 mA.
- Find the collector voltage and V_CE. V_C = Vcc − I_C·Rc = 12 − 2.8 mA · 2k = 12 − 5.6 = 6.4 V. Then V_CE = V_C − V_E = 6.4 − 1.4 = 5.0 V.
- Confirm the mood. V_CE = 5.0 V is comfortably above the ~0.2 V saturation floor and well below the 12 V cutoff ceiling — so the transistor sits squarely in the active region, with room to swing both up and down. The Q-point is (I_C, V_CE) = (2.8 mA, 5.0 V).
The load line: seeing the Q-point on the graph
There's a beautiful picture that ties it all together. Plot the transistor's output characteristics — I_C versus V_CE, with one curve for each value of base current — and you get a fan of nearly-flat lines. Each curve says: 'for this much base current, the collector pushes about this much I_C, almost regardless of V_CE.' That flatness *is* the active region: I_C set by the base, not by V_CE.
But the transistor doesn't get to pick any point on those curves — the *external circuit* (Vcc, Rc, Re) imposes its own constraint. Apply Kirchhoff's voltage law around the output loop and you get a single straight line: Vcc = I_C·(Rc + Re) + V_CE. That's the load line, and the actual operating point must lie on both the curves and the line — so it sits exactly where they cross.
I_C
(mA)
6 ┤╲ ── I_B = 40µA
│ ╲___________________________
5 ┤ ╲
│ ╲________________________ ── I_B = 30µA
4 ┤ ╲
│ ╲______________________ ── I_B = 20µA (our Q!)
2.8 ┤●Q····╲·····················
│ ╲___________________ ── I_B = 10µA
1 ┤ ╲
│ load ╲_________________ ── I_B = 0 (cutoff)
0 └────┬─────┬───╲───┬──────┬──► V_CE (V)
0 2 5.0 8 10 12=Vcc
↑Q-point ↑ V_CE-intercept = Vcc
Load line endpoints:
V_CE = Vcc = 12 V when I_C = 0 (cutoff end)
I_C = Vcc/(Rc+Re) when V_CE = 0 (saturation end)
= 12 / 2500 = 4.8 mANow watch what a signal does. When an input nudges the base current up and down, the operating point *slides along the load line* — up toward saturation, down toward cutoff — and V_CE swings in response. Place your Q-point near the middle of the line and the signal has maximum room to swing both ways before it clips. Place it too near either end, and one half of your waveform gets flattened. The Q-point isn't bookkeeping — it sets how loud your amplifier can get before it distorts.
A different valve: the JFET
The BJT is current-controlled: to run it you must *feed* the base a real current, and that current is the price of admission. But there's a whole second family of transistors that asks for almost no input current at all — they're steered by voltage. The gentlest member to meet first is the JFET (junction field-effect transistor), and it tells a different physical story.
Picture a garden hose you squeeze with your fingers. The water flows from source to drain through a channel of semiconductor; a reverse-biased junction at the gate acts like your pinching fingers. Make the gate voltage more negative and its depletion region swells inward, squeezing the channel narrower until the current chokes off entirely (at the 'pinch-off' voltage). Because the gate junction is *reverse*-biased, it draws virtually no current — so a JFET's input is an almost perfect open circuit.