Transistor Guide

A transistor has a control terminal and two power terminals. A small signal on the control terminal sets how much current flows between the two others. That one idea serves two jobs: as a switch, the transistor is either fully off or fully on and turns a load on and off; as an amplifier, it sits in between, and a small change at the control makes a larger copy at the output.

The kinds differ in what the control is. In a bipolar transistor (BJT) a current into the base controls the current from collector to emitter. In a field-effect transistor (FET) a voltage on the gate controls a channel between drain and source, and the gate draws almost no current. The rest of the family are combinations and specialists: the IGBT puts a MOSFET gate in front of a bipolar output, GaN and SiC are field-effect transistors in new materials, the UJT is an old timing part.

Three questions for any transistor

1. What turns it on? A current (BJT) or a voltage (FET, IGBT), and of which polarity, counted from the emitter or the source: N-type parts turn on with a positive control, P-type parts with a negative one.
2. Is it on or off with nothing on the control? Most are off. A JFET and a depletion MOSFET conduct until the gate turns them off.
3. What does fully on look like? A BJT keeps a small drop, $V_{CE(sat)}$, 0.05 to 0.4 V, that grows with the current. A MOSFET becomes a resistor, $R_{DS(on)}$. An IGBT keeps a knee of about one diode drop plus a little resistance.

The same three regions, two vocabularies

The word "saturation" names opposite regions in the two families: the fully-on switch for a BJT, the current-source region for a FET. Datasheets and websites mix them up. This guide names a region by what it does, with the textbook name next to it.

With the gate at the source voltage, $V_{GS} = 0$, a JFET and a depletion MOSFET already conduct: they are normally on and need a gate voltage of the other polarity to turn off. An enhancement MOSFET, like a BJT, is normally off. The symbols say it: a solid channel line for normally on, a broken one for normally off.

Most datasheets follow the same layout, and two of their sections are often confused. Absolute maximum ratings are limits not to be exceeded even for an instant; running a part at them guarantees nothing. Electrical characteristics are what the part does, each line guaranteed only at the test condition printed next to it. "Typ" is a typical value, not a promise: a design uses min and max.

1. The pinout drawing. The same part number can have different pinouts from different manufacturers, and the package drawing is the only reference.
2. The voltage rating ($V_{CEO}$, $V_{DSS}$), with margin for spikes.
3. The on-state line at the conditions of the job: $V_{CE(sat)}$ at the collector and base currents of the circuit, or $R_{DS(on)}$ at the gate voltage the driver really has.
4. The gain or the threshold, with its spread: $h_{FE}$ min at the current of the job; $V_{GS(th)}$ min and max, measured at a tiny current (often 250 µA or 1 mA), which is where the part only starts to conduct.
5. The thermal resistance, to turn power into temperature: $T_J = T_A + P R_{\theta JA}$. The current rating on page one usually assumes a case held at 25 °C, which no breadboard does.
6. The graphs: they show how every number moves with current and temperature, and the safe operating area of a power part.

The calculation that decides whether a switch survives, for a MOSFET:

An example with real numbers: a 2N7000 switching 75 mA from a 5 V pin. Its datasheet guarantees $R_{DS(on)} \le 5.3\ \Omega$ at $V_{GS} = 4.5$ V, $I_D = 75$ mA and 25 °C; at 125 °C the resistance can be up to 1.8 times larger, some 9.5 Ω. So $P \approx 0.075^2 \times 9.5 = 0.053$ W, and with $R_{\theta JA} = 312.5$ °C/W for its TO-92 the junction runs about 17 °C above the room: fine. At its 200 mA rating the 4.5 V line no longer applies, since it is tested at 75 mA, and even 9.5 Ω gives 0.38 W and some 120 °C of rise: time for a larger MOSFET.

A multimeter on its diode range finds the junctions. A bipolar transistor reads like two diodes back to back: 0.5 to 0.7 V from the base to each of the other two pins (red lead on the base for an NPN, black for a PNP), nothing between collector and emitter. A power MOSFET shows its body diode, 0.4 to 0.7 V from source to drain for an N-channel part, and nothing else, unless the gate holds charge from the last touch: shorting gate to source first discharges it.

Two parts in the same TO-92 can have mirror-image pinouts: a 2N3904 is E-B-C seen from the flat face, a BC547 is C-B-E. The 2N7000 is S-G-D in TO-92, while the SOT-23 2N7002 in the same datasheet is G-S-D. The drawings below come from each manufacturer's datasheet.

Most real circuits use transistors in pairs or groups. These five building blocks run in the simulator, and each is the core of a larger circuit: the pair and its current mirror at the input of an op-amp, the push-pull at the output of most audio amplifiers, the multivibrator in an LED flasher, and the CMOS inverter in CMOS logic gates.