Vibration is measured by tracking how far, how fast, or how forcefully something moves back and forth, captured as three core parameters: displacement, velocity, and acceleration. Sensors convert that physical motion into electrical signals, which are then analyzed in either the time domain (raw waveform) or the frequency domain (broken into individual frequency components). The choice of parameter, sensor, and analysis method depends on what you’re trying to learn, whether that’s diagnosing a failing bearing, testing a product prototype, or assessing whether a bus driver’s daily exposure exceeds safety limits.
The Three Core Parameters
Every vibration can be described by three interrelated quantities. Displacement is the simplest: how far the object moves from its resting position, measured in micrometers or thousandths of an inch. Velocity is the rate of that movement, or how fast the object is traveling at any instant. Acceleration is the rate of change of velocity, capturing how quickly the motion speeds up or slows down, and it’s the parameter most directly tied to the forces acting on a structure.
These three aren’t separate measurements of different things. They’re mathematically linked. Velocity is the first derivative of displacement over time, and acceleration is the second derivative. In practice, this means you can measure one and calculate the others, though each parameter naturally emphasizes a different part of the frequency range. Displacement readings are largest at low frequencies, velocity dominates in the mid-frequency range, and acceleration signals are strongest at high frequencies. This is why an engineer diagnosing a slow-rotating imbalance will focus on velocity, while someone hunting for a defective ball bearing spinning at thousands of RPM will look at acceleration.
Sensors That Capture Vibration
Piezoelectric Accelerometers
The most widely used vibration sensor is the piezoelectric accelerometer. Inside it, a small “proof mass” sits on top of a crystal made from a piezoelectric material. When the sensor experiences vibration, the proof mass resists the motion due to its own inertia and squeezes the crystal slightly. That mechanical stress separates positive and negative charges within the crystal’s molecular structure, producing a tiny electrical charge on its surface. The charge is proportional to the force applied, which in turn is proportional to acceleration. Remove the force, and the crystal snaps back, generating a brief current in the opposite direction.
Piezoelectric accelerometers cover a wide frequency range, from near zero up to above 40 kHz, making them suitable for everything from structural testing to high-speed machinery monitoring. They’re rugged, relatively small, and need no external power in their simplest form, though modern versions with built-in signal conditioning (called IEPE sensors) do require a constant current supply from the data acquisition system.
MEMS Accelerometers
Micro-electro-mechanical systems (MEMS) accelerometers use a tiny silicon structure, fabricated on a chip, that flexes in response to motion. They’re far smaller and lighter than piezoelectric sensors, which gives them a real advantage when the object being tested is lightweight or delicate. Attaching a heavy traditional sensor to a thin panel, for example, can change the panel’s vibration behavior and introduce measurement error. MEMS sensors largely avoid this problem.
The tradeoff is noise. MEMS accelerometers have a higher background noise floor than piezoelectric sensors, which has historically limited their use in precision applications like modal analysis. However, their cables are much smaller, which reduces a subtle source of error: the cable itself can absorb energy from the vibrating structure, artificially inflating the damping values you measure. For lightweight structures where cable mass matters, MEMS sensors can actually deliver more accurate results.
Eddy Current Proximity Probes
For rotating shafts inside turbines, compressors, and pumps, the standard approach is the eddy current proximity probe. This sensor never touches the shaft. Instead, it generates a small magnetic field from its tip, and as the shaft surface moves closer or farther away, it changes the electrical characteristics of that field. The result is a voltage signal proportional to the gap distance, giving a direct measurement of shaft displacement.
These probes are permanently installed inside the machine housing, typically in pairs mounted 90 degrees apart (called an XY configuration) so they can track the shaft’s orbit in two dimensions. They’re calibrated to a specific shaft material, usually steel, and require a minimum shaft diameter to work accurately. If the shaft is too small, the sensor underreports vibration levels. Installation details matter: the probes must be far enough from shaft collars or other features to avoid interference, and when two probes are mounted close together, their magnetic fields can interact and corrupt the readings.
Laser Doppler Vibrometry
When you can’t attach a sensor at all, laser Doppler vibrometry (LDV) offers a completely non-contact alternative. A laser beam is aimed at the vibrating surface, and the reflected light experiences a Doppler frequency shift proportional to the surface’s velocity, the same principle that causes an ambulance siren to change pitch as it passes you. Inside the instrument, this reflected beam is combined with a reference beam, and the interference pattern between them is decoded to extract surface velocity or displacement.
A component called a Bragg cell shifts the reference beam’s frequency by about 40 MHz, which allows the system to distinguish motion toward the laser from motion away from it. The result is extraordinary precision: velocities as small as 10 nanometers per second can be detected, and displacements resolved to sub-picometer levels. Detectable frequencies span from near zero to above 1 GHz, far beyond what any contact sensor can achieve. LDV is used in applications ranging from testing microelectronics to medical imaging research.
How Mounting Affects Your Measurement
The way you attach a contact sensor to a surface has a direct impact on the frequency range you can trust. Stud mounting, where the accelerometer is screwed directly into a tapped hole in the test surface, provides the stiffest connection and the highest usable frequency range. It’s the gold standard against which other methods are compared.
Adhesive mounting is more convenient but introduces a compliant layer between the sensor and the surface, which acts like a small spring and creates a resonance that limits high-frequency accuracy. Thin foam tape maintains accuracy up to about 2,000 Hz. Thicker foam tape extends that slightly to around 2,500 Hz. Foam dots and poster strips, however, are essentially unusable for vibration work, with their accuracy collapsing above just 200 Hz. Magnetic bases are another common option for quick measurements on steel surfaces, though their added mass and the stiffness of the magnetic contact both influence the result. The international standard ISO 5348 provides specific guidance on each mounting method and its limitations.
From Raw Signal to Useful Data
A vibration sensor produces a voltage that changes over time, called a time-domain waveform. While you can spot obvious impacts or periodic motion in this raw signal, most diagnostic information is hidden inside it. A machine might vibrate at its running speed, at twice that speed, at the frequency of individual gear teeth meshing, and at the natural frequencies of its bearings, all simultaneously. In the time domain, these all add together into a complex wave that’s difficult to interpret by eye.
The fast Fourier transform (FFT) solves this by breaking the complex waveform apart into its individual frequency components, each with its own amplitude and phase. The output is a frequency spectrum: a graph showing how much vibration energy exists at each frequency. A spike at the shaft’s rotation speed suggests imbalance. A spike at twice that speed often points to misalignment. A cluster of peaks at much higher frequencies might indicate a bearing defect. Patterns that are invisible in the time domain become obvious in the frequency domain.
The FFT works on the mathematical principle that any periodic signal can be represented as a sum of sine and cosine waves. The algorithm calculates the coefficients of those waves, averages multiple samples to reduce noise, and presents the result as a clean spectrum. It’s the foundation of nearly all vibration analysis.
RMS, Peak, and Crest Factor
Once you have a vibration signal, you need a single number to characterize its severity. The three most common are peak, peak-to-peak, and RMS. Peak value captures the maximum excursion in one direction and is most useful for shock events, single sharp impacts where the worst-case force matters. Peak-to-peak measures the full range of motion from maximum positive to maximum negative, which is particularly relevant for displacement measurements where you need to know if a shaft has enough clearance inside its housing.
RMS (root mean square) is the most commonly used metric because it reflects the energy content of the vibration. A high RMS value means the vibration is delivering sustained destructive energy to the structure, not just occasional spikes. It accounts for the entire time history of the waveform, not just the extremes.
The crest factor bridges peak and RMS by dividing the peak value by the RMS value. A perfectly smooth sine wave has a crest factor of 1.414. As impulsive events like bearing spalls or gear tooth cracks develop, they create sharp spikes that push the peak value up without changing the RMS much, driving the crest factor higher. A rising crest factor over time is an early warning sign of developing mechanical damage, often detectable before the RMS value itself begins to climb.
Measuring Vibration Exposure in People
Vibration measurement isn’t limited to machines. Workers who operate heavy vehicles, jackhammers, or other vibrating equipment face health risks from chronic exposure. International standards set specific limits based on an eight-hour daily exposure value, expressed as a weighted acceleration in meters per second squared.
For whole-body vibration, the kind experienced by truck drivers or forklift operators, the action level is 0.43 m/s². Below this, risk is considered low. Above it, employers are expected to implement controls such as better seating, reduced exposure time, or equipment maintenance. The absolute exposure limit is 0.87 m/s², which must not be exceeded over an eight-hour workday. These thresholds are consistent across the ISO 2631 standard, the American Conference of Governmental Industrial Hygienists, and U.S. Department of Defense guidelines.
An alternative metric called the vibration dose value (VDV) is sometimes used instead, measured in m/s^1.75. It’s more sensitive to peaks and shocks than the simple RMS-based method. The action level using VDV is 8.5 m/s^1.75, with an upper limit of 17 m/s^1.75. The choice between the two methods depends on how impulsive the vibration is: steady, predictable vibration is well characterized by the RMS approach, while intermittent shocks are better captured by VDV.