Noise in electronics is any unwanted electrical signal that interferes with the signal you actually care about. It shows up as hiss in audio equipment, snow on older TV screens, or errors in digital data. Some noise comes from outside sources like power lines and radio transmitters, but much of it is generated inside electronic components themselves, produced by the fundamental behavior of electrons. Understanding where noise comes from and how it’s measured helps explain why engineers spend so much effort trying to minimize it.
Thermal Noise: The Unavoidable Baseline
Every resistor and conductor in a circuit generates noise simply because its atoms are vibrating. At any temperature above absolute zero, electrons jiggle randomly due to thermal energy, creating tiny voltage fluctuations across the component’s terminals. This is called thermal noise, or Johnson-Nyquist noise after the physicists who first measured and explained it in the late 1920s.
The amount of thermal noise depends on three things: the component’s resistance, its temperature, and the bandwidth (range of frequencies) you’re measuring. Higher resistance, higher temperature, and wider bandwidth all mean more noise. This is why cooling sensitive electronics improves performance. Radio telescopes and satellite receivers, for example, sometimes chill their front-end amplifiers to cryogenic temperatures to push thermal noise as low as possible. In everyday circuits operating at room temperature, thermal noise sets a fundamental floor that no amount of clever design can eliminate entirely.
Shot Noise: The Graininess of Current
Electrical current feels smooth, but it’s actually made of individual electrons arriving one at a time. Think of it like rain hitting a tin roof. From a distance, rain sounds like a steady patter, but each drop lands at a slightly random moment. In the same way, electrons crossing a barrier in a semiconductor (like a diode or transistor junction) arrive at irregular intervals, creating tiny fluctuations in the current. This is shot noise.
Shot noise is most prominent in devices where electrons must cross a gap or energy barrier, such as tunnel junctions, Schottky diodes, and standard p-n junctions found in nearly every semiconductor chip. Its strength is proportional to the amount of current flowing through the device. Unlike thermal noise, shot noise doesn’t depend on temperature. It depends on the fact that charge comes in discrete packets rather than as a continuous fluid.
Flicker Noise: Dominant at Low Frequencies
Flicker noise, often called 1/f noise (pronounced “one-over-f”), gets louder at lower frequencies. If you plot its power against frequency, it rises steadily as frequency drops, following an inverse relationship. At very low frequencies, flicker noise can dwarf all other noise sources in a circuit.
The physical causes aren’t fully understood in every case, but flicker noise is generally tied to imperfections in materials and manufacturing. Crystallographic defects in semiconductors, contamination at material surfaces, and inconsistencies in how charge carriers move through a device all contribute. It also creeps in through amplifier stages. Even in high-quality oscillators, the amplifiers needed to boost the signal to a usable level introduce flicker noise. This makes it a persistent challenge in precision instruments, audio electronics, and any system that needs to measure slow-changing signals accurately.
Burst Noise: Random Voltage Jumps
Burst noise, sometimes called popcorn noise because of the sound it makes in audio circuits, appears as sudden, random jumps between two or more voltage levels. If you watched it on an oscilloscope, you’d see the signal snap between discrete steps at unpredictable intervals.
This type of noise is linked to defects in semiconductor crystal structures, both in the bulk material and at surfaces. A single flaw in the silicon can trap and release charge carriers randomly, causing the current through that region to hop between states. Burst noise was a bigger problem in earlier generations of integrated circuits, and improvements in manufacturing have reduced it significantly, though it still appears in some components.
External Noise Sources
Not all noise originates inside the circuit. Electromagnetic interference (EMI) from power lines, motors, radio transmitters, and even nearby digital circuits can couple into sensitive analog signals. This external noise enters through several paths: it can radiate through the air and get picked up by wires acting as antennas, it can travel along shared power supply lines, or it can sneak in through ground connections that aren’t truly at the same voltage.
Ground loops are a particularly common culprit. When two pieces of equipment share a ground connection through multiple paths, small voltage differences between those paths drive currents that appear as noise, typically a hum at the power line frequency. Anyone who has heard a 60 Hz buzz from a guitar amplifier has experienced a ground loop firsthand.
Quantization Noise in Digital Systems
When an analog signal is converted to digital form, it passes through an analog-to-digital converter (ADC) that maps a continuous range of voltages to a finite set of discrete values. This rounding process introduces quantization noise. Each sample carries a small error equal to the difference between the true analog voltage and the nearest digital step the converter can represent.
This error is effectively random and behaves like noise added to the signal. The size of each step depends on the converter’s bit depth. A 16-bit ADC divides its input range into 65,536 levels, producing much smaller rounding errors than an 8-bit converter with only 256 levels. The root-mean-square value of quantization noise equals the step size divided by the square root of 12, which means higher-resolution converters produce proportionally less noise. This is why high-fidelity audio uses 24-bit converters and why scientific instruments push for the highest bit depth their speed requirements allow.
Measuring Noise: SNR and Noise Floor
The most common way to quantify noise is the signal-to-noise ratio, or SNR. It compares the power of the desired signal to the power of the noise, expressed in decibels (dB). A higher number means a cleaner signal. An SNR of 20 dB means the signal is 100 times more powerful than the noise. At 60 dB, the signal is a million times stronger.
In practice, audio equipment with an SNR around 90 to 100 dB sounds very clean to most listeners. Professional recording gear often targets above 110 dB. For RF receivers and data communication systems, the required SNR depends on the modulation scheme and acceptable error rate, but the principle is the same: more distance between signal and noise means better performance.
The noise floor is the level below which signals simply can’t be detected. It represents the total noise power present in a system under a given set of measurement conditions. Every amplifier, cable, and connector in a signal chain adds some noise, and the noise floor is the cumulative result. A system’s dynamic range, the span between the loudest signal it can handle and the quietest it can detect, is directly determined by the noise floor at the bottom and distortion limits at the top.
Engineers also use a metric called noise figure, which measures how much noise a particular component (like an amplifier) adds to a signal passing through it. A perfect, noiseless amplifier would have a noise figure of 0 dB. Real amplifiers always add some noise, so their noise figure is always above zero. In receiver chains, the noise figure of the first amplifier matters most, because all subsequent stages amplify whatever noise that first stage introduces.
Reducing Noise in Practice
Since noise can never be fully eliminated, engineers use a combination of strategies to keep it manageable. The approach depends on whether the noise is generated internally or picked up from the environment.
For externally coupled noise, shielding is the first line of defense. Enclosing sensitive circuits in a grounded metal enclosure blocks electric field interference. Twisted-pair wiring reduces magnetic field pickup by ensuring that any interference induced in one twist is canceled by the opposite twist in the next. A properly implemented twisted pair can provide 55 dB or more of noise rejection, though this benefit disappears if ground loops are present.
Breaking ground loops is often critical. This can be done with transformers, optical couplers, or differential amplifiers, all of which pass the desired signal while rejecting the common-mode voltage that drives ground loop currents. Balanced circuits, where the signal travels on two conductors with equal and opposite voltages, are inherently resistant to interference because any noise picked up equally on both conductors cancels out at the receiving end. This is why professional audio and long-distance communication systems use balanced connections.
For internally generated noise, component selection matters. Low-noise amplifiers use transistor types and circuit topologies optimized to minimize thermal and flicker noise contributions. Reducing bandwidth to only what the application requires is another powerful tool, since both thermal and shot noise scale directly with bandwidth. A system that only needs to pass frequencies up to 10 kHz will have far less noise than one open to 1 MHz. Decoupling capacitors placed near integrated circuits filter noise on power supply lines, preventing it from coupling between stages. And when the noise floor of a single measurement is too high, averaging multiple readings can pull weak signals out of the noise, a technique used everywhere from medical imaging to deep-space communication.