current feedback amplifier
A current-feedback amplifier replaces the high-impedance differential input stage of a conventional op-amp with a low-impedance buffer at its inverting input, so the feedback network injects a current into that input rather than forcing a voltage across it — the result is an op-amp whose closed-loop bandwidth barely shrinks as you increase gain, unlike a standard voltage-feedback op-amp whose gain-bandwidth product is a fixed constant, which is exactly the property that makes current-feedback amplifiers the default choice for high-speed video, RF, and wideband instrumentation circuits.
Gain-bandwidth independence is the headline feature, and it comes directly from feeding back a current instead of a voltage. In a conventional voltage-feedback op-amp, closed-loop bandwidth equals the open-loop gain-bandwidth product divided by the closed-loop gain, so doubling the gain exactly halves the bandwidth — a fixed tradeoff baked into the topology; a current-feedback amplifier's bandwidth is instead set largely by its internal feedback resistor and transimpedance stage, so raising closed-loop gain by changing the external feedback resistor ratio costs far less bandwidth, letting designers dial in a specific gain without paying the same bandwidth tax a VFA would charge.
The low-impedance inverting input is a genuinely different circuit, not just a different spec, and it changes how you're allowed to build feedback networks. A voltage-feedback op-amp's inverting input is high impedance, so designers freely use large feedback resistor values without affecting stability or bandwidth; a current-feedback amplifier's inverting input impedance is low and part of the active signal path, so its feedback resistor value directly sets internal operating currents and therefore bandwidth and stability — datasheets for current-feedback parts specify a recommended feedback resistor range for a reason, and ignoring it produces peaking, instability, or reduced bandwidth that a VFA design habit wouldn't predict.
Slew rate stops being a bottleneck because there's no internal node that has to slew a large voltage. A voltage-feedback op-amp's output stage has to charge internal compensation capacitance through a limited current, which caps how fast its output can change for a large input step; a current-feedback amplifier's input stage responds to current rather than charging a voltage node internally, so it can deliver dramatically higher slew rates — commercial current-feedback parts routinely exceed 1,000 volts per microsecond, which is why they dominate applications like video distribution amplifiers and high-speed ADC driver stages where fast, full-swing transitions matter as much as bandwidth.
It cannot be used as a true integrator or in several classic VFA feedback configurations, which is the real limitation behind the speed. Because the feedback path carries a real operating current rather than just sensing a voltage, current-feedback amplifiers can't tolerate a capacitor directly in the feedback path the way a VFA integrator does — doing so removes the DC feedback current path and the circuit behaves unpredictably — so designs that need true integration, certain active filter topologies, or very high closed-loop gain with light feedback resistors still default to voltage-feedback parts; the choice between CFA and VFA is a genuine topology decision, not just a speed-versus-everything-else tradeoff.
Read a current-feedback amplifier through a *current-in-the-loop* lens rather than a *faster-op-amp* lens: the entire set of CFA properties — flat bandwidth across gain settings, enormous slew rate, and the restriction against certain feedback-capacitor topologies — all trace back to the single decision to let the feedback network set a current at a low-impedance node rather than force a voltage across a high-impedance one, which is a different circuit discipline, not just a faster version of the same one.