Analog Circuit
An analog circuit is an electronic circuit that processes signals which vary continuously, so that the voltage or current at any moment can take any value within a range and carries meaning in its exact size. A microphone's output, a radio wave picked up by an antenna, the voltage across a temperature sensor and the swing of a loudspeaker cone are all analog quantities. Where a digital circuit asks only whether a signal is above or below a threshold, an analog circuit cares about how much, so small errors are not erased but pass through as errors in the result.
The classic building blocks are amplifiers, filters, oscillators and references. An amplifier increases the amplitude of a signal, and a transistor biased in its active region is the elementary amplifying element; the gain is the ratio of output to input. A filter, built from resistors, capacitors, inductors or active devices, passes some frequencies and blocks others, as in the bass and treble controls of a stereo or the channel selection of a radio. An oscillator uses amplification plus feedback to generate a repeating waveform, which is the heart of every clock, radio transmitter and synthesizer; crystal oscillators fix the frequency with a quartz resonator. Voltage references and regulators supply the stable levels that everything else depends on.
The operational amplifier turned feedback into a general-purpose tool. An op-amp is a high-gain differential amplifier that, in combination with a few resistors, can add, subtract, integrate, differentiate, filter and buffer signals, because negative feedback makes the circuit's behavior depend on the resistor ratios rather than on the amplifier's imprecise gain. The principle of negative-feedback amplification was developed by Harold Black at Bell Labs in 1927 to stabilize telephone repeaters. Integrated op-amps arrived with Robert Widlar's μA709 at Fairchild in 1965, and the μA741, designed by David Fullagar and released in 1968 with internal compensation, became an industry standard that students still meet in the laboratory. Analog computers once used the same blocks to solve differential equations electrically, long before digital machines took over.
Noise is the fundamental limit, and it cannot be fully removed. Every resistor produces thermal noise, known as Johnson-Nyquist noise from measurements by John Johnson and a theory by Harry Nyquist in 1928, with a voltage of about the square root of 4 × k × T × R × B, where k is Boltzmann's constant, T the temperature, R the resistance and B the bandwidth. Transistors add shot noise and flicker noise, and components drift with temperature and age. Because there is no restoring threshold, an analog signal degrades at every stage, and the first stage of a chain matters most: a low-noise amplifier at the front of a receiver sets the sensitivity of the whole system. That is why analog designers talk about signal-to-noise ratio, dynamic range and distortion, and why precise analog parts are laid out and trimmed with great care.
Analog persists because the world is analog and some jobs are cheaper or faster in analog form. Every sensor, microphone and radio antenna delivers an analog signal, and before it can be digitized it must be amplified and filtered, and the converters themselves are mixed-signal circuits. Radio-frequency front ends in phones, Wi-Fi and radar amplify and mix signals at gigahertz frequencies. Power electronics, such as voltage regulators and battery chargers, switch and shape large currents, and audio amplifiers drive loudspeakers. A modern system-on-chip is mostly digital by area, but its phase-locked loops, voltage regulators, memory sense amplifiers and radios are analog, and they often demand the most specialized design effort.
Analog computing is now returning in AI hardware. Neural-network inference is dominated by multiply-accumulate operations, and an array of resistive or memory elements can perform them in place: voltages applied to rows, conductances storing weights, and the currents summed on columns by Kirchhoff's current law, so Ohm's law does the multiplication. Several startups and research labs have built analog or mixed-signal in-memory chips that use far less energy per operation than moving data to a digital processor, at the price of limited precision, noise, and device variation that the neural network must tolerate. A common misconception is that analog is obsolete or less advanced than digital; it is more accurately the unavoidable physical layer beneath every digital system.
Read analog circuits as electronics that works directly with quantity rather than with a yes-or-no decision: the continuous, noise-limited layer where sensors, radios and power meet the physical world, and the contrast against which digital logic's discipline of thresholds, regeneration and bits is best understood.