Switched-Mode Power Supplies

A linear supply steps down with a 50 Hz transformer, whose size is fixed by that frequency, then burns the difference in a regulator. A 5 V 2 A output from a 12 V rail wastes 14 W to deliver 10 W.

An SMPS rectifies the mains directly to about 325 V DC, then chops it at 100 kHz into a small high-frequency transformer. Core size falls roughly as 1/f, so the magnetics shrink by three orders of magnitude.

That single reordering — rectify first, switch fast — is why a modern charger fits inside a plug pin. Everything else about an SMPS follows from it, including the awkward parts.

The high-frequency transformer provides galvanic isolation between the mains and anything the user can touch — a safety requirement with legally specified creepage and clearance distances.

The regulation feedback must cross the same barrier without breaching it, which is why an optocoupler sits in almost every SMPS control loop — and why it is a reliability item worth specifying carefully.

The savings compound: less heat means a smaller heatsink, a smaller enclosure and less material throughout. No-load consumption fell too, which is what standby power regulations were written to capture.

The front-end capacitor charges only while the mains exceeds its voltage, so the input current is a train of narrow spikes rich in harmonics. This is a distortion power-factor problem, not a displacement one — so no capacitor bank corrects it.

The third harmonic adds rather than cancels across three phases, overloading neutral conductors. One supply is negligible; a building full of them is a genuine network problem, which is why active PFC is mandatory above 75 W in the EU.

An active PFC stage is a boost converter whose control loop forces the input current to follow the rectified mains waveform, giving a power factor above 0.99. It also delivers a stable 400 V bus, which simplifies the stage after it.

Fast switching generates interference that leaves conducted along the mains lead and radiated from the current loops. Input filtering handles the first; layout that keeps high-di/dt loops small handles the second.

Conformance testing is a legal requirement and a failure means redesign rather than a tweak, which is why EMI is planned from the first schematic. Spread-spectrum modulation dithers the switching frequency to smear the peaks below the limit.