a buck-boost converter
A buck-boost converter is the switching regulator that copes when your input voltage sometimes sits above and sometimes below the output you need — it can step down or step up as required. The classic case is a battery: a lithium cell starts near 4.2 V full and sags to about 3.0 V flat. If you need a steady 3.3 V, you must buck (step down) at the start of the discharge and boost (step up) near the end. A plain buck or a plain boost cannot do both; a buck-boost can.
There are a few topologies. The simplest inverting buck-boost gives a negative output with Vout = negative Vin times D / (1 - D). Modern parts use a four-switch buck-boost that keeps the output the same polarity and smoothly hands over between buck mode (input well above output), a blended region (input near output), and boost mode (input below output), all under one feedback loop. The single shared inductor stores energy in one phase and releases it in the next, just as in the simpler converters, but the switch arrangement lets the ratio go either above or below one.
Why it matters: buck-boost is the natural choice for anything battery-powered that must hold a fixed rail across the whole discharge — wearables, handheld instruments, single-cell USB outputs. The price is extra switches and control complexity, slightly lower peak efficiency than a dedicated buck, and a tricky transition region where input and output are close. But for the job of riding a battery from above to below the target voltage without the rail blinking, nothing else does it as cleanly.
A handheld GPS needs a steady 3.3 V from one lithium cell. Fresh at 4.0 V the buck-boost works in buck mode; after hours of use at 3.1 V it has slid into boost mode — and the 3.3 V rail never wavered, so the device stays alive to the last drop instead of dying when the cell crosses 3.3 V.
Steps down when input is high, up when it is low — one rail across a whole battery.
The hand-off region where input is near output is the hard part: efficiency dips and ripple can rise there. If your input never crosses your output, a plain buck or boost is simpler, cheaper, and usually more efficient — only use buck-boost when the crossover really happens.