Sampling Rate (Frequency of Measurement) in Measurement Systems

Sampling rate, also called sampling frequency, is a foundational concept in any measurement or data acquisition system. It refers to how many times per second a continuous-time (analog) signal is measured and converted into a digital value. This parameter, measured in hertz (Hz), defines how finely the system can resolve changes in the measured phenomenon over time. A higher sampling rate provides finer granularity, crucial for capturing rapid events, while a lower rate may suffice for slow or static signals.

Why Sampling Rate Matters

Sampling rate is crucial because it determines how well a digital system can represent the original analog signal. In aviation, for example, black box recorders must sample fast enough to capture sudden control movements or transient vibrations. In biomechanics, force plates for jump analysis need high rates to detect brief, high-magnitude forces. In industrial monitoring, vibration sensors must capture high-frequency oscillations to detect early signs of machinery faults.

Too low a sampling rate leads to “undersampling,” missing critical events or distorting the signal—a phenomenon called aliasing. Excessively high rates, in contrast, burden storage and processing resources without improving useful information.

The Process of Sampling

Sampling is a two-step process:

1. Measurement: The analog signal is measured (sampled) at regular intervals.
2. Digitization: Each measured value is converted into a digital number via an analog-to-digital converter (ADC).

The time between samples is the sampling interval (inverse of the sampling rate). For example, a 1 kHz sampling rate means one sample every 1 millisecond.

Units: Hertz (Hz)

Sampling rate is expressed in hertz (Hz), or samples per second. In some applications, kilohertz (kHz, thousands of samples per second) or megahertz (MHz, millions) are used.

Typical examples:

• Audio (music): 44.1 kHz (CD quality)
• Voice recorders: 8–16 kHz
• Aviation black boxes: 1–4 Hz for slow parameters (altitude), >1 kHz for fast parameters (vibration)
• Biomechanics (jump analysis): 1000 Hz or higher

Theoretical Foundations

Nyquist-Shannon Sampling Theorem

The Nyquist theorem is the mathematical foundation for sampling. It states:

To perfectly capture all information in a signal, the sampling rate must be at least twice the highest frequency present in the signal.

This threshold is called the Nyquist rate. If the signal contains frequencies up to 500 Hz, you must sample at least at 1000 Hz.

Aliasing

Aliasing occurs when a signal is sampled below the Nyquist rate. Higher-frequency content is “folded” into lower frequencies, distorting the digitized signal. In safety-critical systems, aliasing can hide or misrepresent important events.

Example:
If a 600 Hz vibration is sampled at 800 Hz, it appears as a 200 Hz vibration in the data—potentially masking a fault.

Anti-Aliasing Filters

To prevent aliasing, analog anti-aliasing filters are used before the ADC. These filters block frequencies above half the sampling rate, ensuring only valid signal components are digitized. Since filters are not perfect, engineers often choose a sampling rate higher than twice the highest frequency of interest, allowing a “transition band” where the filter can roll off.

Practical Considerations

Under-Sampling

Sampling below the required rate causes:

• Missed rapid events (e.g., aircraft control movements)
• Distorted frequency content
• False conclusions in analysis

Example:
Vibration in an aircraft engine at 800 Hz, sampled at 1 kHz, is at risk of aliasing if the anti-aliasing filter is not effective.

Over-Sampling

Sampling far above the needed rate:

• Improves time resolution for transient analysis
• Increases storage and processing needs
• May amplify noise
• Offers little benefit if the signal lacks high-frequency content

Best practice: Sample at 2.5–10 times the highest frequency of interest, then down-sample or average if needed.

Signal Content: Know Your Frequencies

Every measured process has characteristic frequencies:

• Human gait: <20 Hz (sample at 50–100 Hz)
• Explosive sports/jumps: up to 300 Hz (sample at 500–1000 Hz)
• Audio (music): up to 20 kHz (sample at 44.1 kHz)
• Vibration monitoring: up to 10 kHz (sample at 25–30 kHz)

Tip: Review literature, perform spectral analysis (FFT), and consult manufacturer guidelines to choose the right rate.

Sensor and Hardware Limitations

• Sensor bandwidth: Don’t sample above what the sensor can measure accurately.
• ADC limitations: High-resolution ADCs often sample more slowly than low-resolution, high-speed ones.
• Anti-aliasing filter quality: Poor filters require higher oversampling to compensate for gradual roll-off.

Time Resolution vs. Frequency Content

• Time resolution: Shorter intervals allow detection of rapid changes/events.
• Frequency content: The highest frequency you can analyze is half the sampling rate (Nyquist frequency).

Application Examples

Aviation

• Flight Data Recorders (FDR): Sample slow parameters (altitude) at 1–4 Hz; fast parameters (accelerations, control positions) at 8 Hz to several kHz.
• Cockpit Voice Recorders (CVR): Audio sampled at 8–16 kHz for intelligibility.
• Vibration monitoring: Sensors sample at 25–30 kHz for bearing fault detection or engine health monitoring.

Biomechanics

• Jump/force analysis: Force plates sample at 1000–2000 Hz to capture rapid force changes.
• Gait analysis: Motion capture operates at 100–200 Hz for walking/running.

Industrial Monitoring

• Rotating machinery: Vibration sensors sample at 2.5–3x the highest machine frequency.
• Transient detection: Impact events require tens of kHz sampling rates.

Audio and Communications

• CD-quality audio: 44.1 kHz to capture all audible frequencies.
• Professional audio: 48 kHz or higher for studio use.

Environmental Monitoring

• Temperature/pressure: Slow-changing, sampled at 1 Hz or less.

Recommended Sampling Rates Table

Selecting the Right Sampling Rate

1. Identify the highest frequency of interest.
2. Check sensor and DAQ bandwidth.
3. Choose a sampling rate at least 2.5–10x that frequency.
4. Apply anti-aliasing filtering matched to the chosen rate.
5. Balance time resolution, data volume, and analysis needs.
6. Consult regulatory or industry standards (e.g., ICAO, EUROCAE, ISO).

Common Misconceptions

• Higher sampling rates always mean better data: Not true—beyond a certain point, you only increase data volume and noise.
• Nyquist rate is always sufficient: In practice, sample well above the Nyquist rate to compensate for real-world filters and unexpected transients.
• Anti-aliasing filters are optional: Without filtering, aliasing can corrupt your data, no matter how high the sampling rate.

Conclusion

Sampling rate is the backbone of digital measurement systems, dictating how accurately you can capture, analyze, and interpret dynamic phenomena. Whether you are designing an aircraft data acquisition system, configuring a biomechanics lab, or setting up industrial monitoring, understanding and applying the correct sampling rate is essential for reliable, actionable data.

For guidance on optimizing your measurement systems, or to discuss your specific needs, contact our experts or schedule a demo .