The job nobody notices
Look at the little black brick on your charger, the bump on a laptop cable, or the silver box humming inside a server. They all do the same unglamorous job: take electrical energy that arrives in one shape and deliver it in another. The wall gives you 230 V of wobbling AC; your phone wants a steady 5 V of DC. A car battery gives 400 V; the headlights want 12 V and the infotainment chip wants 3.3 V. Power electronics is the art of converting electrical energy from one form to another — and doing it while wasting as little as possible.
Why obsess over waste? Because everything you don't deliver to the load comes out as heat, and heat is the enemy. Heat melts solder, ages capacitors, and forces you to bolt on bulky fins or noisy fans. A converter that is 50% efficient throws away as much energy as it delivers — and that wasted half has to go somewhere. So the whole field can be summed up in one stubborn question: how do I change the voltage without burning the difference?
The lazy way: a resistor that burns the difference
Suppose you have 12 V and need 5 V at 1 A for a small gadget. The most obvious idea is to put something in the path that 'eats' the extra 7 V. A resistor does exactly that: by Ohm's law, a resistor that drops 7 V while passing 1 A must be 7 Ω, and it dissipates P = V × I = 7 V × 1 A = 7 watts of heat. It works — your gadget sees 5 V — but only if the current stays exactly 1 A. Draw a bit more and the voltage sags; draw less and it rises. The fix is to make the resistor *adjustable*.
That adjustable resistor is the linear voltage regulator. Inside a part like the classic LM7805 sits a transistor used as a smart, variable resistor. A feedback loop watches the output: if it droops below 5 V, the transistor opens a little; if it climbs, it closes. The output stays rock-solid — but notice what the transistor *is*. It sits half-on, in its linear region, deliberately dropping voltage. And dropping voltage while carrying current is just another way of saying making heat.
Linear regulator dropping 12 V -> 5 V at 1 A
12V o----[ pass transistor ]----o 5V --> load (1 A)
acts like 7 ohm |
| drops 7 V @ 1 A |
| => 7 W of HEAT |
GND --------------------------- GND
Power IN = 12 V x 1 A = 12 W
Power to load= 5 V x 1 A = 5 W
Power wasted = 12 W - 5 W = 7 W (as heat)
Efficiency = 5 W / 12 W = 41.7 %The clever way: stop dissipating, start switching
The waste came from one choice: holding the transistor half-on, so it has both voltage across it *and* current through it at the same time. Power is V × I, so half-on is the worst place to be. What if we never let the transistor sit there? Run it as a pure on/off switch instead. When fully ON, it has nearly zero voltage across it (so V × I ≈ 0). When fully OFF, it carries nearly zero current (so V × I ≈ 0). Either way, an ideal switch burns *no* power. The only loss is during the brief flicker between states.
But a switch alone gives you a chopped-up, all-or-nothing voltage — 12 V, then 0 V, then 12 V — not a smooth 5 V. The magic is what we put after the switch: an inductor and a [[capacitor|capacitor]]. An inductor resists sudden changes in current; a capacitor resists sudden changes in voltage. Together they act like a flywheel and a reservoir, smoothing the chopped pulses into a steady DC level. The switch moves energy in discrete *packets*; the inductor and capacitor pour those packets into a smooth stream. This is the heart of a switched-mode power supply.
Switching converter (a buck): chop, then smooth
switch L (inductor)
12V o----/ ----+------UUUU------+----o 5V --> load
| |
diode C (cap) --- smooths
| | the ripple
GND ------------ GND
Switch ON : energy flows in, L current ramps up
Switch OFF : L keeps current flowing through diode
Output (V) : ___ ___ ___
before C : | |_____ | |_____| |__ (chopped)
after C : ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ (smooth 5 V)How do you choose the output voltage? By how *long* the switch stays on each cycle — the fraction of time it's ON is called the duty cycle. In an ideal buck converter, V_out = duty cycle × V_in. Want 5 V from 12 V? Hold the switch on about 42% of the time. Want 3 V? Drop to 25%. You're no longer trading voltage for heat; you're trading it for *time*, and time is free. This specific chopper is called a buck converter — the first real circuit you'll build later in this track.
Same job, run the numbers side by side
Let's settle the argument with the same task — 12 V in, 5 V out, 1 A to the load, so 5 W of useful power — and compare honestly. The linear regulator's input current equals its output current (a series resistor passes whatever goes through), so it pulls 1 A from the source the whole time. The switcher is sneakier: because power in must equal power out (minus small losses), pulling less voltage at the output means it can draw *less current* at the input. Energy conservation, working in your favour.
TASK: 12 V in -> 5 V out at 1 A (useful = 5 W)
------------------------------------------------
LINEAR regulator | SWITCHING (buck)
------------------------------------------------
Input voltage 12 V | 12 V
Output 5 V @ 1 A | 5 V @ 1 A
Input current 1.00 A | ~0.46 A (*)
Power IN 12.0 W | ~5.5 W
Power to load 5.0 W | 5.0 W
Power WASTED 7.0 W heat | ~0.5 W heat
Efficiency 41.7 % | ~91 %
------------------------------------------------
(*) 5 W out / 0.91 eff = 5.5 W in; 5.5 W / 12 V = 0.46 ASeven watts versus half a watt isn't an abstract score — it's the difference between a part you can touch and one that needs a heatsink the size of the rest of the gadget. It's why a 65 W laptop charger today is the size of a matchbox and runs warm, while the same wattage as a linear supply would be a brick that needs a fan. Multiply by a billion phone chargers, by every data centre, by every electric car, and that efficiency gap becomes power plants' worth of electricity.
Why a transistor makes such a good switch
The whole trick rests on one component behaving in two completely different ways. The same transistor that the linear regulator parks half-on, the switcher slams fully on or fully off, thousands of times a second. Why is full-on so much better? Think of it as a tap. Half-open, water roars past a partly-closed valve and the valve heats up from the friction. Fully open, water flows freely with almost no resistance — no friction, no heat. A transistor switched hard-on has a tiny resistance (often a few thousandths of an ohm); hard-off, it's essentially an open circuit.
Run the numbers and you see why it wins. Suppose a switch carrying 1 A has an on-resistance of 10 milliohms. By Ohm's law the voltage across it is only 0.01 V, and the heat is just 0.01 V × 1 A = 0.01 W — a hundredth of a watt, versus the linear regulator's 7 watts. The transistor of choice for this is the MOSFET, whose on-resistance keeps falling year after year, which is exactly why switchers keep getting smaller and cooler.
- Switch ON. The transistor connects the 12 V source to the inductor. Current ramps up; the inductor stores energy in its magnetic field. The transistor drops almost no voltage, so it stays cool.
- Switch OFF. The transistor disconnects. The inductor refuses to let its current stop suddenly, so it keeps pushing current through a diode (or a second switch) into the load.
- Smooth. The capacitor sits across the output, absorbing the up-and-down ripple so the load sees a near-constant 5 V.
- Repeat fast. Do this tens of thousands to millions of times a second. The faster you switch, the smaller the inductor and capacitor can be — which is why modern chargers are tiny.
Where this track is going
You now hold the one idea the whole field is built on: don't dissipate energy, *move* it — switch a transistor hard on and off, and let an inductor and capacitor launder the chopped pulses into a clean output. Everything ahead is variations on this theme. We'll build the buck converter (steps voltage down, the 12 → 5 case you just saw), then its mirror image the boost converter (steps voltage *up*), then converters that do both. We'll meet the switching-versus-linear trade-off again with real noise and real numbers.
From there the track climbs: how to actually drive these fast switches, how to turn DC back into AC with an inverter (the box that lets a battery run your house or spin a motor), and the wide-bandgap devices (SiC and GaN) that switch so fast they're reshaping electric cars and solar farms. But every rung rests on what you learned today — the difference between burning the difference and switching it away.