Sensors & Measurement · Study deck

Sensor Noise and Averaging Limits

Picture a light sensor whose value jumps beside a motor.

Physics Phoebe is your guide for this deck.

sensorbestpractices

After studying this chapter

Learning objectives

You will be able to:

Identify thermal, flicker, and ADC quantization noise in sensor front-ends.
Estimate whether front-end noise, ADC step size, sensor tolerance, or environment dominates the error budget.
Explain why averaging improves white noise but not slow drift or fixed bias.
Choose between averaging, bandwidth limiting, shielding, grounding, reference checks, and calibration.

Major section

Start With the Measurement Story

Averaging may calm the display while hiding the real cause.

The team must learn whether the change is random noise, drift, interference, or a fixed error.
Bandwidth means the range of change a measurement path can follow.
Signal-to-noise ratio means the useful signal level compared with the noise level.

Major section

After Sensor Lab Best Practices

Sensor Lab Best Practices teaches the defensive workflow: validate readings, filter noise, fuse sensors, monitor health, and use hysteresis.

This page explains when those safeguards work and when they only hide the wrong failure mode.
If variation is random and fast, bandwidth limiting and averaging can help.
If the value drifts, moves with supply voltage, changes with placement, or disagrees by a fixed offset, the answer is evidence, calibration, or hardware cleanup rather than more samples.

Try it: After Sensor Lab Best Practices in the chapter

Major section

Overview: Every Reading Comes With Noise

The lab log must keep the test conditions, raw values, and uncertainty needed to show why the result deserves trust.

Noise management begins by deciding what independent information is available, not by choosing an averaging window.
Same Sensor means more samples over time: Different Sensors add another kind of evidence.
That pair is the key lesson.

Sensor fusion pipeline combining noisy same-sensor readings and different sensor measurements into a state estimate with uncertainty.

Major section

Overview: Every Reading Comes With Noise (continued)

These inputs reduce doubt in different ways.

A smooth number can still be wrong.
The mathematical gist: A 12-bit ADC over 3.3 V has a 0.8057 mV code step and a uniformly distributed rounding floor of $q/\sqrt{12}=232.6\ \mu$ V RMS.
Bits reduce quantisation; they do not remove reference noise, sensor tolerance, drift, or aliasing.

Major section

Quantify Noise and Averaging

A front end contains several noise sources, and their units must agree before they can be combined.

The next workbench makes the front-end crossover visible instead of leaving it as separate equations.
Shot noise is a current-noise process associated with discrete charge crossing a junction.
At low frequency, junctions and FETs can add flicker noise.

Input-referred noise budget matrix locating shot, flicker, resistor thermal, amplifier voltage and current, and mechanical thermal noise.

Major section

Quantify Noise and Averaging (continued)

Change only the upper band edge from 1010 Hz to 110 Hz: the shaded band contracts, SNR improves, and quantization becomes the larger remaining floor—but the converter itself remains exactly 16 bit.

so the amplitude density is $i_{n,\text{shot}}=\sqrt{2qI_{DC}}\ \text{A}/\sqrt{\text{Hz}}$ .
The square root matters: $2qI_{DC}$ is a power density, not an amplitude density.
A common empirical form is the Hooge relation.

Major section

Quantify Noise and Averaging (continued)

An amplifier voltage-noise density $e_n(f)$ is already in volts per square-root hertz.

Its current-noise density $i_n(f)$ becomes a voltage through source impedance $Z_s(f)$ , while a source resistance contributes Johnson noise $4kTR$ .
Mechanical thermal noise belongs at the transducer input and is converted through the sensor sensitivity before it joins this voltage budget.
SNR describes random-noise separation over a stated bandwidth; it is not the whole accuracy claim.

Major section

Quantify Noise and Averaging (continued)

Aliased out-of-band noise must be limited by an analog filter; a digital average cannot undo noise already folded into baseband.

Flicker noise is different: in many semiconductor and resistive devices its measured voltage noise grows with the DC voltage across the device and rises as frequency falls.
This dimensional form is safer than memorising a transcription with a stray resistance term.
Across unrelated parts, however, resistance alone cannot predict $N$ or $\alpha_H$ ; the datasheet or a measured spectrum must supply the device-specific evidence.

Deck summary

Key takeaways

Averaging may calm the display while hiding the real cause.

Sensor Lab Best Practices teaches the defensive workflow: validate readings, filter noise, fuse sensors, monitor health, and use hysteresis.
The lab log must keep the test conditions, raw values, and uncertainty needed to show why the result deserves trust.
These inputs reduce doubt in different ways.
A front end contains several noise sources, and their units must agree before they can be combined.

Retrieval practice

Recall check 1 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q1A stable reference shows fast random jitter; another sensor has a fixed offset; a third changes when a motor starts. Which remedy plan fits?

AAverage every sensor longer as if fixed offset and motor noise were random.

BAverage random jitter, calibrate fixed offset and inspect motor coupling.

CCalibrate each sensor because independent sensor adjustment resolves the motor-linked reading disturbance.

DChange display rounding alone without correcting any of the three errors.

Show answer

Answer: B Correct: the three symptoms have different causes and need different remedies.

Retrieval practice

Recall check 2 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q2A sensor report shows only a smooth final reading. What should the lab record add before using it for an alert?

ARaw readings, test conditions, reference checks and uncertainty estimate

BMore decimal places as supposed proof of raw noise and reference accuracy

CThe filter name and screenshot alone without raw or reference evidence

DOne passing value alone without setup or uncertainty information

Show answer

Answer: A Correct: these show the signal, its conditions, and how much trust the result deserves.

Retrieval practice

Recall check 3 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q3Which three noise sources fundamentally limit a typical sensor front-end?

AWi-Fi interference, cosmic rays, and cable colour in the front-end.

BQuantization alone, with no thermal or flicker noise in the circuit.

CI2C bit errors, packet retries, and network collision noise.

DThermal noise, flicker noise, and ADC quantization noise.

Show answer

Answer: D Thermal noise, flicker noise, and quantization noise set the practical front-end noise floor, and each responds to different fixes.

Retrieval practice

Recall check 4 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q4Averaging 4 independent samples of a white-noise-limited reading improves SNR by how much, and by what mechanism?

ABy 4 dB, because SNR improves by N in dB.

BAbout 6 dB; averaging cuts white noise RMS by 2x.

CBy 12 dB, the same as averaging 16 samples.

DNot at all, because averaging cannot reduce noise.

Show answer

Answer: B sqrt(4) = 2, which is 20·log10(2) ≈ 6 dB, matching 3 dB for each of the two doublings.

Retrieval practice

Recall check 5 of 5

Physics Phoebe says: answer from memory, then check your reasoning.

Q5A near-DC measurement is limited by 1/f (flicker) noise and slow thermal drift. Why does averaging thousands of samples give disappointing improvement?

A1/f noise and drift are correlated between samples and concentrated at low frequency.

BAveraging always improves any noise by sqrt(N), so the result must be a coding bug.

C1/f noise is actually quantization noise, which cannot be averaged.

DThe signal is too large for averaging to have any effect.

Show answer

Answer: A Averaging needs independent samples; low-frequency noise and drift behave like a wandering baseline rather than independent hiss.

Print reference

Answers 1 of 2

Answer key.

B · Correct: the three symptoms have different causes and need different remedies.
A · Correct: these show the signal, its conditions, and how much trust the result deserves.
D · Thermal noise, flicker noise, and quantization noise set the practical front-end noise floor, and each responds to different fixes.
B · sqrt(4) = 2, which is 20·log10(2) ≈ 6 dB, matching 3 dB for each of the two doublings.

Print reference

Answers 2 of 2

Answer key.

A · Averaging needs independent samples; low-frequency noise and drift behave like a wandering baseline rather than independent hiss.