Noise and Interference
Learning Objectives
- Distinguish noise from interference and give an example of each
- Identify the major noise sources: thermal, shot, flicker (1/f), and quantization noise
- Explain electromagnetic interference (EMI) and how it couples into a circuit
- Calculate thermal noise power/voltage using standard formulas
- Describe practical mitigation techniques: differential amplification, filtering, shielding, and grounding
- Apply noise-reduction reasoning to a realistic circuit design scenario
Quick Answer
Noise is any unwanted, typically random electrical disturbance that corrupts a signal, while interference is an unwanted signal from an identifiable external source (another circuit, a nearby transmitter, or a power line) that couples into a circuit. Both degrade signal quality, reduce measurement accuracy, and can cause outright circuit malfunction if not managed. Every real analog circuit — no matter how carefully designed — has some inherent noise floor from thermal and semiconductor effects, and every real system operates in an electromagnetically "noisy" environment, so understanding noise sources and mitigation techniques is essential for building circuits that actually work reliably outside a textbook.
Noise vs. Interference: Two Different Problems
Noise arises from fundamental physical processes inside components — the random thermal motion of electrons, the discrete (quantized) nature of electric charge crossing a semiconductor junction — and is present even in a perfectly shielded, isolated circuit. It's usually characterized statistically (average power, spectral density) rather than as a specific waveform, since it's inherently unpredictable moment to moment.
Interference comes from an external, often identifiable source: a nearby motor, a Wi-Fi router, a fluorescent light ballast, or a neighboring signal wire coupling energy into your circuit. Unlike fundamental noise, interference can often be traced to a specific cause and eliminated or reduced at its source (better shielding, physical separation, filtering at a specific frequency).
Major Noise Sources
Thermal noise (Johnson-Nyquist noise): Caused by the random thermal motion of electrons in any resistive material — it exists in every resistor at any temperature above absolute zero, is "white" (roughly equal power at all frequencies within a relevant bandwidth), and its RMS voltage is given by:
V_n = √(4kTRB)
where k is Boltzmann's constant (1.38 × 10⁻²³ J/K), T is absolute temperature in kelvin, R is resistance in ohms, and B is bandwidth in hertz.
Worked example: A 10 kΩ resistor at room temperature (T = 300 K) in a 10 kHz bandwidth produces V_n = √(4 × 1.38×10⁻²³ × 300 × 10,000 × 10,000) ≈ √(1.656×10⁻¹¹) ≈ 4.07 µV RMS. This tiny but unavoidable voltage sets a fundamental noise floor for any circuit built with that resistor — no amount of clever circuit design can eliminate it, only manage its impact relative to the wanted signal.
Shot noise: Arises from the discrete, random arrival of individual charge carriers crossing a potential barrier (like a p-n junction in a diode or transistor) — more current doesn't average out perfectly smoothly at a fundamental level, and this graininess shows up as noise, more prominent in devices with lower current levels.
Flicker noise (1/f noise): Noise power that increases as frequency decreases (hence "1/f"), common in transistors and other semiconductor devices, dominant at low frequencies and often the limiting noise source in DC and very-low-frequency precision measurements.
Quantization noise: Introduced whenever a continuous analog signal is converted to a finite number of digital levels (as in an ADC) — covered in depth in the ADC chapter — and can be thought of as the "error" between the true analog value and its rounded digital representation.
Interference Sources
Electromagnetic interference (EMI): Energy radiated or conducted from an external source that couples into a circuit through the air (radiated EMI) or through shared wiring/power connections (conducted EMI). Motors, switching power supplies, and digital clock circuits are common EMI sources.
Radio-frequency interference (RFI): A specific case of EMI originating from radio-frequency sources (transmitters, RF equipment) that can disrupt sensitive analog circuits, especially those with high-gain amplifier stages that can inadvertently demodulate stray RF energy.
Crosstalk: Interference between adjacent signal paths caused by unwanted capacitive or inductive coupling — common in multi-channel systems (like ribbon cables or PCB traces running close together), where a signal on one line induces an unwanted echo on a neighboring line.
Power supply noise: Distortions or ripple introduced by imperfect power supply regulation, which can couple into sensitive analog stages through shared power rails if not adequately filtered/decoupled.
Effects of Noise on Circuit Performance
Noise and interference degrade circuits in several concrete, measurable ways: they reduce signal amplitude relative to the noise floor (lowering signal-to-noise ratio), they can push a system's effective dynamic range down (making it harder to distinguish a weak wanted signal from the noise background), and in digital or data-transmission contexts, sufficiently severe noise increases bit error rate. In feedback systems, injected noise can even trigger instability or unwanted oscillation if it couples into a sensitive feedback path.
Mitigation Techniques
Differential amplification: Using an instrumentation or differential amplifier (covered in Signal Conditioning) to amplify only the difference between a signal and its return path rejects common-mode noise — interference that appears equally on both wires of a twisted pair, such as most externally-coupled EMI — while still amplifying the wanted differential signal.
Filtering: Low-pass, high-pass, band-pass, or notch filters (covered in the Filters and Amplifiers chapter) remove noise and interference outside the frequency range of the wanted signal — a notch filter tuned to 50/60 Hz is a classic technique for removing mains-hum interference from sensitive measurements.
Shielding: Enclosing sensitive circuitry or cables in a grounded conductive enclosure (a Faraday cage) blocks radiated EMI from reaching the circuit; twisted-pair cabling reduces inductive pickup by having each conductor's induced noise largely cancel against its twisted partner.
Grounding: Careful, deliberate grounding practices (a single-point "star" ground, avoiding shared ground paths between noisy and sensitive circuits) prevent ground loops — one of the most common real-world sources of injected noise in measurement systems.
Component selection: Choosing low-noise op-amps and transistors (specified by datasheet noise figures like input voltage noise density in nV/√Hz), and choosing resistor values as low as practical (since thermal noise scales with √R) in noise-sensitive front-end stages.
Digital signal processing (DSP): After an ADC, digital filtering and averaging techniques can further remove residual noise that made it through the analog front end, though this cannot recover information lost to noise/aliasing before digitization.
Real-World Example: A Differential Amplifier for Noise Reduction
Consider an ECG (electrocardiogram) sensor: the actual heart signal on the skin is only a few millivolts, while 50/60 Hz mains interference picked up by the body (acting as an antenna near power lines and equipment) can be far larger. By using electrodes at two locations and feeding both signals into a differential (instrumentation) amplifier, the mains interference — which appears nearly identically on both electrodes (common-mode) — is largely subtracted out, while the genuine ECG signal (which differs between the two electrode locations) is amplified. This common-mode rejection is why virtually every biomedical signal acquisition system uses differential sensing rather than a single-ended measurement against a distant ground.
Why It Matters
No real circuit operates in a perfectly quiet, noise-free, interference-free environment — thermal noise is a fundamental physical limit, and interference from motors, switching supplies, and RF sources is everywhere in the modern world. Recognizing which noise or interference mechanism is limiting a design (and choosing the matching mitigation: differential sensing for common-mode interference, filtering for out-of-band noise, shielding for radiated EMI, better grounding for ground loops) is what separates a circuit that works only on the lab bench from one that works reliably in the field.
Common Mistakes
Misconception 1: "Noise and interference are the same thing and can be fixed with the same techniques." Why it's wrong: Noise is an inherent, random physical phenomenon (thermal, shot, flicker) present even in a perfectly isolated circuit, while interference comes from an identifiable external source and can often be eliminated entirely by addressing that source (shielding, physical separation, filtering at a specific frequency). Correct understanding: Fundamental noise can only be reduced (lower temperature, lower resistance, narrower bandwidth, better component choice), never eliminated; interference can potentially be eliminated entirely by removing or blocking its specific external source.
Misconception 2: "Since thermal noise is 'white' (present at all frequencies equally), filtering can't help reduce its impact." Why it's wrong: While thermal noise power is spread evenly across frequency, the total thermal noise power in a circuit is proportional to the bandwidth over which it's measured (V_n = √(4kTRB)) — restricting bandwidth with a filter directly reduces the total noise power that reaches later stages. Correct understanding: A filter that only passes the frequency range of the wanted signal reduces the amount of thermal (and other broadband) noise power admitted into the system, even though it can't change the noise's per-hertz spectral density.
Misconception 3: "A differential amplifier eliminates all noise from a measurement." Why it's wrong: Differential amplification only rejects common-mode noise — interference that appears identically on both signal paths. Noise that differs between the two paths (differential-mode noise, or noise generated internally within the amplifier itself) is not rejected and is amplified right along with the wanted signal. Correct understanding: Differential sensing is a powerful but partial solution, specifically effective against common-mode interference (like mains hum picked up equally by both electrodes); it must be combined with filtering, shielding, and low-noise component choices to address other noise sources.
Comparison and Connections
| Feature | Noise | Interference |
|---|---|---|
| Origin | Inherent physical processes (thermal, shot, flicker) | External identifiable source (motor, transmitter, adjacent circuit) |
| Predictability | Random, statistical description only | Often traceable to a specific frequency/source |
| Can be eliminated? | No — only reduced | Often yes, by removing/blocking the source |
| Typical mitigation | Lower bandwidth, lower resistance/temperature, low-noise components | Shielding, filtering at the interfering frequency, physical separation |
| Noise Type | Frequency Behavior | Typical Cause |
|---|---|---|
| Thermal (Johnson-Nyquist) | Flat (white) across frequency | Random electron motion in resistive materials |
| Shot | Flat (white) across frequency | Discrete charge carriers crossing a junction |
| Flicker (1/f) | Increases at low frequency | Semiconductor surface/trap effects |
| Quantization | Depends on signal, roughly uniform in ADC's Nyquist band | Rounding in analog-to-digital conversion |
Practice Questions
Recall 1: State the formula for thermal (Johnson-Nyquist) noise voltage and define each symbol. Answer guidance: V_n = √(4kTRB); k = Boltzmann's constant, T = absolute temperature (K), R = resistance (Ω), B = bandwidth (Hz).
Recall 2: List four major sources of noise discussed in this chapter. Answer guidance: Thermal noise, shot noise, flicker (1/f) noise, quantization noise (crosstalk, power supply noise, and EMI/RFI are interference sources, though sometimes grouped alongside noise topics).
Understanding 1: Explain why reducing a circuit's bandwidth reduces the impact of thermal noise, even though thermal noise is "white" (equal power density at every frequency). Answer guidance: Total thermal noise power/voltage depends on the bandwidth over which it's integrated (V_n ∝ √B); a narrower bandwidth (via filtering) admits a smaller total amount of the otherwise evenly-spread noise power, even though the noise's per-hertz density is unchanged.
Understanding 2: Explain why a differential amplifier is especially effective at removing 50/60 Hz mains hum picked up by long sensor wiring. Answer guidance: Mains hum tends to couple nearly identically (in phase and magnitude) onto both conductors of a twisted-pair or closely-routed signal path, making it a common-mode signal; a differential amplifier subtracts the two inputs, canceling the common-mode hum while preserving and amplifying the genuine differential signal.
Application 1: Calculate the RMS thermal noise voltage of a 1 kΩ resistor at room temperature (300 K) over a 1 MHz bandwidth. Answer guidance: V_n = √(4 × 1.38×10⁻²³ × 300 × 1000 × 1,000,000) = √(1.656×10⁻¹¹) ≈ 4.07 µV RMS. (Note the same numeric result as the 10kΩ/10kHz example in the text, since R×B is the same product — this illustrates the √(RB) dependence.)
Application 2: A precision measurement circuit is picking up 60 Hz interference from nearby mains wiring. Suggest two specific, distinct mitigation techniques and explain how each addresses the problem. Answer guidance: (1) A 60 Hz notch filter removes the specific interfering frequency from the measured signal while passing other frequencies largely undisturbed. (2) Shielding the sensitive wiring (e.g., using shielded, twisted-pair, or coaxial cable with proper grounding) reduces the amount of 60 Hz interference that couples into the circuit in the first place, addressing the problem at its source rather than after the fact.
Analysis 1: A designer lowers the value of a feedback resistor in a sensitive amplifier stage specifically to reduce noise, but this also reduces the stage's voltage gain. Analyze this tradeoff using the thermal noise formula. Answer guidance: Since thermal noise voltage scales with √R, a smaller resistor does directly reduce the thermal noise contributed by that resistor. However, since amplifier gain in configurations like the inverting op-amp amplifier depends on resistor ratios, reducing one resistor without proportionally adjusting others changes the gain too — the designer must balance the noise-vs-resistance tradeoff against the need to maintain adequate signal gain, often by choosing the smallest resistor values that still meet other design constraints (input impedance, power dissipation) rather than reducing resistance without limit.
Analysis 2: Compare and evaluate two competing noise-reduction strategies for a low-level sensor amplifier: (a) increasing gain early in the signal chain, versus (b) filtering aggressively before any amplification. Explain which is generally preferred and why. Answer guidance: Amplifying early (option a) is generally preferred for genuine noise (like the sensor's or first-stage amplifier's own inherent noise) because it boosts the wanted signal above the noise contributed by later stages before that noise has a chance to dominate — this is why the first amplifier stage's own noise figure is so critical to overall system performance. Aggressive filtering before any amplification (option b) is useful for removing known out-of-band interference, but it does nothing to combat the amplifier's own inherent noise and can even reduce the signal margin if the filter also attenuates part of the wanted signal; in practice, systems typically combine minimal early filtering (for gross interference) with early low-noise amplification for best overall performance.
FAQ
Q1: Can noise ever be completely eliminated from a circuit? A: No — fundamental sources like thermal noise exist in any resistive component above absolute zero temperature. Engineers can only reduce noise's impact (narrower bandwidth, lower-noise components, lower operating temperature or resistance), never eliminate it entirely.
Q2: Why does flicker (1/f) noise matter more for precision DC measurements than for high-frequency circuits? A: Flicker noise power increases as frequency decreases, so it's most significant at low frequencies (including DC), making it the dominant noise source in precision, slow, DC-coupled measurements, while at higher frequencies other noise sources (thermal, shot) typically dominate instead.
Q3: What's the practical difference between shielding and filtering as noise/interference mitigation techniques? A: Shielding physically blocks external electromagnetic energy from reaching a circuit or cable in the first place (addressing interference at the coupling path), while filtering removes unwanted frequency content from a signal after it has already entered the circuit — the two are often used together for a robust solution.
Q4: Why do ground loops cause noise, and how is this different from EMI? A: A ground loop occurs when two points assumed to share the same ground potential actually differ slightly, driving an unwanted current through the shared ground path that injects noise into signals referenced to that ground — this is a wiring/grounding topology problem, distinct from EMI, which is externally radiated or conducted energy coupling into a circuit from an unrelated source.
Q5: Is a higher signal-to-noise ratio always necessary, or does "good enough" depend on the application? A: It depends entirely on the application — a voice communication link might tolerate a much lower SNR than a precision scientific instrument or a high-fidelity audio system; the acceptable SNR is set by how sensitive the end use is to residual noise (intelligibility, measurement precision, perceived audio quality).
Quick Revision
- Noise = inherent, random, unavoidable (thermal, shot, flicker, quantization); interference = external, identifiable source (EMI, RFI, crosstalk).
- Thermal noise: V_n = √(4kTRB); exists in every resistor above absolute zero, scales with √(resistance × bandwidth × temperature).
- Shot noise arises from discrete charge carriers crossing a junction; flicker (1/f) noise dominates at low frequencies.
- Quantization noise is introduced specifically during analog-to-digital conversion.
- EMI/RFI couple into circuits via radiated or conducted paths; crosstalk is coupling between adjacent signal lines.
- Differential (instrumentation) amplifiers reject common-mode noise/interference but not differential-mode noise.
- Filtering reduces total noise power admitted into a system by restricting bandwidth, even for "white" noise sources.
- Shielding and twisted-pair wiring block/cancel radiated and inductively-coupled interference.
- Proper (single-point/star) grounding prevents ground loops, a common real-world noise source.
- Amplifying early in a signal chain, using low-noise components, minimizes the impact of later-stage noise contributions.
Related Topics
Prerequisites: Analog Signal Fundamentals, Signal Conditioning, Operational Amplifiers
Related: Analog-to-Digital Converters, Analog Communication Systems, Feedback Systems
Next: Analog Measurement Techniques