Push Pull Output Stage
A push-pull output stage uses two complementary devices — one that actively sources current into the load, one that actively sinks it back out — so the output can be driven hard in both directions instead of relying on a single device plus a passive resistor or current source to do half the job, and this same two-device structure shows up far beyond audio amplifiers: it's the totem-pole output of every CMOS logic gate, the architecture of transformer-coupled switch-mode power supplies, and the active half-bridge inside a motor driver, all solving the identical problem of fast, symmetric drive in both directions.
Replacing a passive pull-up or pull-down with an active device is the whole innovation. A single-ended stage that sources current through a transistor but relies on a resistor to pull the output back down is limited by how fast that resistor can discharge whatever capacitance sits on the output — lower resistance speeds up the fall but wastes more static power, and there's no way to win both; a push-pull stage replaces that resistor with a second active device that sinks current just as aggressively as the first one sources it, which is why push-pull outputs switch an order of magnitude faster than resistor-pulled single-ended ones at the same power budget.
Every CMOS logic gate is a push-pull stage, which is easy to miss because nobody calls it that in a digital context. A CMOS inverter's PMOS transistor sources current to pull the output high and its NMOS transistor sinks current to pull it low, and the two never conduct heavily at the same time except briefly during a transition — that brief overlap is called shoot-through or crowbar current, and controlling it is exactly the same design concern that push-pull audio output stages and half-bridge motor drivers have to manage, just scaled to gigahertz switching speeds instead of audio frequencies.
Transformer-coupled push-pull topology solved a different problem a century before audio push-pull stages existed: driving an AC transformer efficiently from a DC supply. In a push-pull switch-mode power supply, two switches alternately drive opposite ends of a transformer's primary winding, each handling one half of the AC cycle, which lets the design use the full transformer core flux in both directions instead of only one — this is a structurally different application from audio push-pull (sourcing/sinking a load directly) but shares the same core idea of splitting a bidirectional job between two devices that each only have to handle one direction.
Gate driver ICs exist almost entirely to solve push-pull's one real hazard at speed. Dedicated gate driver chips from companies like Texas Instruments and Infineon insert precise, deliberate dead time between turning one device off and the other on, because any overlap where both the high-side and low-side device conduct simultaneously creates a near-direct short from supply to ground — a failure mode serious enough in power MOSFET half-bridges that it can destroy both devices in microseconds, which is why shoot-through protection is treated as a first-class design requirement rather than an edge case.
Read a push-pull output stage through a *shared-labor* lens rather than an *amplifier-topology* lens: the core idea is simply refusing to let one device and a passive component split a bidirectional job unevenly, and once that idea is separated from any one application it explains why the same two-device structure keeps reappearing — in audio power stages, in every digital logic gate, and in transformer-driving power supplies — as the default answer whenever a circuit has to drive a load symmetrically in both directions.