1. Home
2. Academy
3. Electronics from Zero
4. Volume 05

5.1 Why switch with a transistor

A transistor is a switch with no moving parts. A small current or voltage on one lead turns a much bigger current on or off through the other two. That lets a tiny control signal run lamps, motors and relays.

A pin is not a power supply

A microcontroller controls things through its pins. Each pin can switch between 0 V and its supply voltage, often 3.3 V. But it can only supply a little current, usually about 10 mA to 20 mA before it is damaged. Many loads want far more, often at a higher voltage:

A mechanical switch could handle those currents, but a program cannot press it. A relay could, but it is slow, and it wears out. A transistor switches in a millionth of a second, uses little power to control, and never wears out.

Three leads

Every transistor has three leads. Two of them carry the big current, and the third controls it. There are two main families:

1. The bipolar transistor. A small current into its control lead lets a current about a hundred times bigger flow. Module 2.
2. The MOSFET. A voltage on its control lead switches it on, and it draws no steady current at all. Module 3.

5.2 The bipolar transistor as a switch

In an NPN bipolar transistor, current into the base lets a bigger current flow from collector to emitter - up to about a hundred times bigger. Give it enough base current, and it turns fully on, with only about 0.2 V across it.

Collector, base and emitter

The three leads are the collector, the base and the emitter. The load goes between the supply and the collector, and the emitter goes to ground. The base is the control. From base to emitter, the transistor behaves like a diode, so the base sits about 0.7 V above the emitter while current flows into it.

circuit npn-switch: pin at 3.3 V: base current 2.6 mA; lamp current 98.3 mA; Q1 is fully on with 200 mV across it
pin at 0 V: base current 0 A; lamp current 0 A; Q1 is off with 12 V across it

RB sets the base current, by Ohm's law:

base current = (3.3 V - 0.7 V) / 1 kΩ = 2.6 mA

A 2.6 mA control current switches 98.3 mA at 12 V. The pin never meets the 12 V at all.

Gain, and fully on

The transistor multiplies its base current by its current gain. It is about 100 for a small transistor, though it varies a lot from one part to the next. As the base current grows, the lamp current grows a hundred times faster - until the lamp has the whole supply across it. Then no more can flow. The transistor is fully on, or saturated:

fully on needs a base current of at least 98.3 mA / 100 = 983 µA

Designers give the base two to five times the least it needs, because the gain varies. Here 2.6 mA is plenty.

Why a switch must be fully on

A fully-on transistor has only 0.2 V across it, so it wastes very little:

Q1 turns 0.2 V x 98.3 mA = 19.7 mW into heat

With too little base current, it is only part on. Then it has a large voltage across it and a large current through it at the same time - and that makes heat:

with RB = 10 kΩ: base current 260 µA; lamp current 26 mA; Q1 is only part on, with 8.88 V across it, and turns 231 mW into heat
with RB = 100 kΩ: base current 26 µA; lamp current 2.6 mA; Q1 is only part on, with 11.7 V across it, and turns 30.4 mW into heat

5.3 The MOSFET as a switch

A MOSFET is switched by a voltage, not a current. Raise its gate above its threshold voltage and it conducts with a tiny resistance; the gate itself takes no steady current.

Gate, drain and source

An N-channel MOSFET has a gate, a drain and a source. The load goes between the supply and the drain, and the source goes to ground. The gate is a metal plate insulated from the rest - a small capacitor, as in Volume 03. Once it is charged, no more current flows into it.

When the gate is more than the threshold voltage above the source, the drain-to-source path conducts, with a small on-resistance. This MOSFET's threshold is 2 V and its on-resistance 0.05 Ω:

circuit nmos-switch: pin at 3.3 V: the gate sits at 3.3 V; lamp current 100 mA; M1 has 5 mV across it
M1 turns 500 µW into heat
pin at 0 V: lamp current 0 A; pin not driving at all: R2 holds the gate at 0 V, lamp current 0 A

Five millivolts across it, where the bipolar transistor had 200 mV. That difference grows with the current:

at 2 A: a MOSFET of 0.05 Ω turns 200 mW into heat; a transistor at 0.2 V turns 400 mW

That is why MOSFETs switch most heavy loads today.

The two resistors on the gate

RG, the gate resistor, limits the brief current that flows each time the pin charges or empties the gate's capacitance. R2 holds the gate at 0 V when nothing drives it. For a moment after power-up, a microcontroller's pins are not driving yet. Without R2, the gate would float, and the load might switch on by itself. Module 5 has more on resistors like R2.

Logic-level MOSFETs

Many MOSFETs need 10 V on the gate to turn fully on. A 3.3 V pin needs a logic-level MOSFET, whose datasheet promises a low on-resistance at 3.3 V or less. Volume 08 shows where to find that number.

Low side and high side

Both switches here sit between the load and ground: a low-side switch. To switch the supply side of a load instead, designers use a P-channel MOSFET, which turns on when its gate is pulled below its source. Volume 06 uses both kinds.

5.4 Motors, relays and the flyback diode

Motors and relays are coils, and a coil switched off makes a huge voltage spike. A flyback diode across the coil gives its current somewhere to go, so the spike never happens.

Relays

A relay is a switch worked by an electromagnet. Current through its coil pulls a metal contact across, closing a separate circuit. The two circuits never touch, so a transistor on a safe, low voltage can switch a completely separate load.

circuit relay: Q1 on: the coil carries 98.3 mA; D1 is off

With Q1 on, D1 is backwards: its cathode is at 12 V and its anode near 0 V. It does nothing.

The moment of switching off

Volume 03 showed that an inductor's current cannot stop instantly. When Q1 turns off, the coil keeps pushing its 98.3 mA out of its lower end. D1 is now the only way round, so the current flows up through D1 and back into the top of the coil:

the instant Q1 turns off: D1 carries the coil's 98.3 mA, and the collector rises only to 12.7 V
the current then dies away with a time constant of L / R = 50 mH / 120 Ω = 417 µs

The collector rises just 0.7 V above the supply, which Q1 can take easily. Without the diode, the story is very different:

without D1: no current path at all - the coil forces the voltage as high as it takes
stopping 98.3 mA in 1 µs: V = 50 mH x 98.3 mA / 1 µs = 4.92 kV

Nearly five thousand volts would destroy Q1 at once. Motors are coils too, and need the same diode.

5.5 Pull-up and pull-down resistors

An input connected to nothing floats, and reads at random. A pull-up or pull-down resistor gives it a definite level whenever nothing else is driving it.

The floating input

A chip's input takes almost no current, so it follows whatever voltage it is connected to. Connected to nothing, it has no voltage of its own. It drifts, picks up noise from nearby wires, and can read 0 or 1 at random. That is a floating input.

A button alone does not fix this. When it is pressed it connects the input to something, but when it is released the input floats again.

The pull-up resistor

circuit pull-up: button open: the input is at 3.3 V and 0 A flows; pressed: 0 V and 330 µA flows

With the button open, no current flows through R1, so there is no voltage across it, and the input sits at 3.3 V. Pressing the button joins the input to ground. R1 then limits the current that flows, and the input reads 0. Note the logic is turned round: pressed reads 0.

A pull-down resistor is the mirror image: a resistor to ground, and a button to the supply. Then pressed reads 1.

Choosing the value

a 1 kΩ pull-up wastes 3.3 mA while the button is held
a 10 kΩ pull-up wastes 330 µA while the button is held
a 100 kΩ pull-up wastes 33 µA while the button is held

A small resistor wastes current while the button is held. A very large one pulls only weakly, so noise and leakage can overcome it. Around 10 kΩ is the usual compromise. Many microcontrollers have pull-up resistors built in, which a program can switch on.

Pull-ups appear all over digital electronics. The gate resistor R2 in Module 3 is a pull-down. Shared wires such as I²C use open-drain outputs with a pull-up on each line.

What you learned

• A transistor lets a small control signal switch a big current, far faster than any mechanical switch.
• An NPN transistor multiplies its base current by its gain, about 100, until it is fully on with about 0.2 V across it.
• A switch must be fully on or fully off; part on, it runs hot.
• A MOSFET is switched by its gate voltage, draws no steady gate current, and has a tiny on-resistance.
• A logic-level MOSFET turns fully on from a 3.3 V pin; a gate pull-down keeps it off at power-up.
• A flyback diode across a coil clamps the switch-off spike to 0.7 V above the supply.
• A pull-up or pull-down resistor stops an input floating; about 10 kΩ is typical.

Key words from this volume

Every word below has a plain-English entry in the glossary.

• Transistor
• Bipolar transistor (BJT)
• Collector
• Emitter
• Current gain (beta)
• Saturation (transistor)
• MOSFET
• Threshold voltage
• On-resistance
• Logic-level MOSFET
• Low-side switch
• Flyback diode
• Relay
• Pull-up resistor
• Pull-down resistor
• Floating input
• Open-drain

Practice

5 V, 100 Ω load, RB = 1.5 kΩ: base current 2.87 mA; load current 48 mA; Q1 fully on
fully on needs at least 48 mA / 100 = 480 µA of base current

a 4.7 kΩ pull-up on 3.3 V wastes 702 µA while the button is held

a 400 Ω relay coil on 12 V with Q1 on: 29.5 mA

Interview corner

Volume 06 builds logic gates from MOSFETs: the CMOS inverter, NAND and NOR gates, inside every chip.