buck-converter

A buck converter is a switching voltage regulator that steps a DC input voltage down to a lower DC output voltage, doing it by rapidly switching a transistor on and off and letting an inductor and capacitor average that chopped waveform into a clean, lower steady voltage — the same switching-regulator principle underlying every high-efficiency power supply, but specifically arranged to step voltage down rather than up. Because a buck converter dissipates very little power in the switch itself, it routinely achieves efficiencies well above 90%, which is exactly why it's the dominant topology for converting a higher battery or supply rail voltage down to whatever lower voltage a processor core, memory chip, or other low-voltage block actually needs. How a buck converter actually chops and averages its input into a lower, stable output looks like this:

A few things are worth knowing about why a buck converter's output voltage comes directly from how long the switch stays on, and why that relationship is both simple and surprisingly powerful:

A buck converter's output voltage, in the ideal case, is simply the input voltage multiplied by the fraction of each switching cycle the transistor spends closed — a quantity called duty cycle — which is why regulating the output comes down almost entirely to continuously adjusting that one timing parameter. Keeping the switch closed for 30% of each cycle and open for the remaining 70%, on average, delivers roughly 30% of the input voltage to the output, assuming ideal lossless components; a feedback loop constantly measures the actual output voltage and nudges the duty cycle up or down to hold that output exactly on target despite changes in input voltage or load current. This duty-cycle control is the entire regulation mechanism — there's no separate voltage-dropping element burning power the way a linear regulator's pass transistor does, which is the direct source of a buck converter's efficiency advantage.

An inductor's current can't change instantly, and that single physical fact is exactly what lets a buck converter keep delivering current to the load even during the portion of each cycle when the switch is open and disconnected from the input. When the switch closes, current through the inductor ramps up, storing energy in its magnetic field while drawing power from the input source. When the switch opens, that stored magnetic energy doesn't vanish — the inductor's current keeps flowing in the same direction, now circulating through a separate path, commonly a diode or a second switch, called the freewheeling or synchronous rectification path, continuing to supply the load even though the input is momentarily disconnected. This is the core trick that makes the whole topology work: the inductor acts as a temporary energy reservoir bridging the gaps when the switch isn't directly connecting input to output, which is exactly why a buck converter can't be built with a capacitor alone — only an inductor's resistance to instantaneous current change provides that bridging behavior.

Synchronous rectification — replacing that freewheeling diode with a second, actively controlled switch — meaningfully improves efficiency at the cost of needing more complex switch-timing control, and it's become the standard approach in most modern buck converter designs. A simple diode in the freewheeling path always drops some forward voltage while conducting, wasting power proportional to that drop times the current flowing through it, same as the loss mechanism in a linear regulator, just confined to a smaller fraction of the cycle. Replacing that diode with a second transistor, switched on exactly when the diode would otherwise conduct and timed carefully to avoid both switches being on simultaneously, cuts that conduction loss dramatically, which is why essentially every modern high-efficiency buck converter for processor and memory power delivery uses synchronous rectification rather than a simple diode.

Multiple buck converter phases, interleaved and switching slightly out of sync with each other, are how high-current processor power delivery handles enormous, fast-changing current demands that a single-phase buck converter alone couldn't supply cleanly. A modern processor core can demand well over a hundred amps with current swings that happen in nanoseconds as workloads change, a demand far beyond what one inductor and one switch pair can deliver without excessive ripple and stress on the components. Splitting that total current across several buck converter phases, each handling a fraction of the load and switching at staggered times relative to the others, spreads the current demand across multiple sets of components while the interleaving itself partially cancels out current ripple at the shared output — which is why high-end processor voltage regulator modules are built from many parallel buck phases rather than one oversized single-phase converter.

Read a buck converter as duty-cycle-controlled energy transfer through an inductor, not voltage dropped and wasted: the output voltage comes directly from the fraction of each cycle the switch spends closed, the inductor's resistance to instantaneous current change is what bridges the gaps when the switch is open, and real high-current designs — synchronous rectification, multi-phase interleaving — exist specifically to push that already-efficient core mechanism even further toward the near-lossless power delivery a processor's power-hungry, fast-switching workload actually demands.