Accelerometer linearity testing verifies that a transducer's output remains proportional to the applied acceleration across its entire dynamic range, which is essential for accurate vibration and shock measurement. Accredited third-party laboratories perform these evaluations in accordance with ISO 16063, GB/T 20485, and related international standards, issuing traceable calibration certificates that support quality assurance, product qualification, and regulatory compliance. This article explains the technical principles, applicable sample types, test items, measurement methods, and typical applications of accelerometer linearity verification.
Technical Overview
Linearity is one of the most important static and dynamic performance characteristics of an accelerometer. It describes the degree to which the transducer's input-output calibration curve approximates an ideal straight line over its specified measurement range. In practical terms, a highly linear accelerometer maintains a constant charge or voltage sensitivity (output per unit of acceleration) whether it is measuring a small fraction of its range or its full-scale value.
Linearity deviation is typically quantified as the maximum relative deviation of the sensitivity measured at various acceleration amplitudes from a reference sensitivity, expressed as a percentage. Common acceptance criteria range from ±1% for precision laboratory-grade sensors to ±5% for general industrial-grade devices. Excessive nonlinearity distorts measured waveforms, generates spurious harmonic content in frequency spectra, and introduces amplitude-dependent measurement errors that compromise shock characterization, modal analysis, and machine condition diagnostics.
Nonlinear behavior in accelerometers originates from several physical and electronic sources, including nonlinear stiffness of the seismic mass suspension, insufficient or uneven mechanical preload in piezoelectric elements, saturation and slew-rate limitations in built-in electronics, capacitance nonlinearity in MEMS sensing structures, and demodulation errors in variable-capacitance or servo-type designs. Linearity testing is therefore a key element of type evaluation, incoming inspection, periodic recalibration, and post-stress verification following exposure to temperature extremes, mechanical shock, or sustained vibration.
Testing Samples
Piezoelectric Accelerometers
Both charge-output and low-impedance voltage-output (IEPE/ICP-type) piezoelectric accelerometers are routinely submitted for linearity verification. These sensors dominate industrial vibration monitoring and laboratory modal testing, and their linearity is commonly evaluated over acceleration ranges from a few m/s² up to several thousand m/s² at reference frequencies defined by ISO 16063-21.
MEMS Capacitive Accelerometers
Micro-electro-mechanical systems (MEMS) accelerometers, including consumer-grade, automotive-grade, and tactical/industrial-grade devices, require linearity verification because of the inherent nonlinearity of their comb-finger capacitive sensing structures. Testing typically covers DC-response characteristics, amplitude linearity over the full range, and zero-offset stability.
Piezoresistive and Strain Gauge Accelerometers
These DC-responsive sensors are widely used in crash testing and long-duration transient measurement. Linearity verification confirms that bridge output remains proportional to acceleration up to full scale, and also evaluates hysteresis and zero-measurand output, which are closely related error sources.
Servo (Force-Balance) Accelerometers
High-precision servo accelerometers used in inertial navigation, seismology, and platform stabilization demand very tight linearity tolerances, often below ±0.1% of full scale. Verification is performed using static gravimetric methods and precision centrifuge systems.
Testing Items
Depending on the sensor type, specification, and applicable standard, a typical linearity examination program may include the following test items:
- Amplitude linearity deviation of sensitivity over the specified acceleration range
- Reference sensitivity determination at standard conditions (e.g., 160 Hz, 100 m/s² or 10 m/s²)
- Frequency response linearity over the operating bandwidth
- Transverse sensitivity ratio to assess off-axis response contributions
- Hysteresis and zero-measurand output (zero offset) for DC-response sensors
- Full-scale output deviation and endpoint nonlinearity
- Linearity verification before and after environmental or mechanical stress exposure
- Temperature-related sensitivity shift interaction with amplitude linearity
Testing Methods
Comparison Calibration Method
The comparison method per ISO 16063-21 (equivalently GB/T 20485.21) is the most widely used technique. The accelerometer under test is mounted back-to-back with a reference standard transducer on an electrodynamic vibration exciter driven at a fixed sinusoidal frequency. The acceleration amplitude is increased in discrete steps from the lower end of the range toward full scale, and the sensitivity is calculated at each level from the ratio of the test transducer output to the reference transducer output. The linearity deviation is then determined as the maximum departure of these sensitivities from the reference sensitivity, typically obtained by least-squares regression or endpoint fitting.
Primary Laser Interferometry Method
For reference-grade and high-accuracy requirements, primary calibration by laser interferometry in accordance with ISO 16063-11 provides the lowest measurement uncertainty. Fringe-counting, minimum-point, or sine-approximation (phase demodulation) techniques directly relate displacement amplitude, and hence acceleration, to the transducer output, eliminating dependence on a reference sensor. ISO 16063-13 extends primary interferometric methods to shock calibration for high-amplitude linearity evaluation.
Static Gravimetric (Tilt or Turn-Over) Method
For DC-response accelerometers such as MEMS, servo, and piezoresistive types, linearity can be evaluated using Earth's gravity as a reference. The sensor is progressively tilted on a precision index table or dividing head, applying acceleration components between −1 g and +1 g. The deviation of the output from the best-fit line yields the static linearity error and also reveals asymmetry between positive and negative acceleration responses.
Centrifuge Method
To extend static linearity verification beyond ±1 g, the test article is mounted on a calibrated centrifuge. By varying rotational speed, sustained accelerations of tens or hundreds of g are applied along the sensitive axis, enabling full-scale linearity determination for inertial-grade and automotive sensors.
Shock Excitation Method
For high-g linearity assessment, shock calibration using drop-ball impactors, shock machines, or Hopkinson pressure bars is performed in accordance with ISO 16063-22. Sensitivity is determined at peak acceleration levels spanning the dynamic range, and amplitude linearity is evaluated by comparing shock sensitivities with the sinusoidal reference sensitivity, which also quantifies shock-to-vibration sensitivity differences.
Data Evaluation and Acceptance Criteria
In all methods, the calibration data set is fitted with a straight line using the least-squares or endpoint convention, and the linearity error is reported as the maximum deviation expressed as a percentage of reading or a percentage of full scale. The expanded measurement uncertainty of the calibration system, traceable to national metrology institutes through an unbroken chain, is stated on the certificate together with ambient conditions, mounting details, excitation frequencies, and the applied acceleration levels. Results are compared against manufacturer specifications, verification regulations such as JJG 233 for piezoelectric accelerometers, or project-specific acceptance limits.
Applications
Accelerometer linearity testing services support a broad spectrum of industries and use cases:
- Aerospace and aviation: flight-test instrumentation qualification, flutter testing, store certification, and satellite and launch-vehicle payload verification
- Automotive: crash-test data channel qualification, vehicle NVH development, braking and suspension analysis, and safety-system sensor validation
- Machinery condition monitoring: ensuring that vibration severity values used for predictive maintenance alarms remain amplitude-accurate
- Structural health monitoring: bridges, dams, tunnels, high-rise buildings, and wind turbines
- Seismology and geotechnical engineering: strong-motion instrumentation and ground-vibration measurement
- Inertial navigation and guidance: linearity verification of servo and MEMS inertial sensors
- Consumer electronics and IoT: qualification of embedded motion sensors in wearables, smartphones, and controller units
- Calibration and testing laboratories: reference-standard maintenance, interlaboratory comparisons, and accreditation scopes
FAQ
What is accelerometer linearity?
Accelerometer linearity is the degree to which the sensor output remains proportional to the applied acceleration over its specified range. It is quantified as the maximum deviation of measured sensitivity from a reference or best-fit sensitivity, expressed as a percentage of reading or full scale.
Which standards govern accelerometer linearity testing?
The primary references are the ISO 16063 series, particularly ISO 16063-11 for primary interferometric calibration, ISO 16063-21 for comparison calibration, ISO 16063-13 and ISO 16063-22 for shock calibration, together with the corresponding GB/T 20485 series and metrological verification regulations such as JJG 233 for piezoelectric accelerometers.
What acceleration and frequency ranges can be tested?
Typical capabilities cover sinusoidal vibration from below 1 Hz to several tens of kHz and acceleration levels from fractions of 1 m/s² up to tens of thousands of m/s² using shock methods. Static gravimetric and centrifuge methods cover DC-response sensors from ±1 g to several hundred g.
What are typical linearity acceptance criteria?
Acceptance limits depend on sensor class: laboratory reference accelerometers may be specified at ±0.5% to ±1%, industrial piezoelectric sensors commonly at ±1% to ±2%, MEMS devices at ±0.5% to ±1% of full scale, and general-purpose sensors at up to ±5%. Contractual or project specifications always take precedence.
How often should linearity verification be performed?
Most quality systems and calibration policies recommend recalibration at intervals of 12 to 24 months, or sooner following mechanical overload, exposure to extreme temperature, high-level shock, or whenever measurement integrity is in doubt.
How is linearity different from hysteresis?
Linearity describes deviation from a straight-line input-output relationship regardless of the direction of change, whereas hysteresis describes the difference in output for the same acceleration approached from increasing versus decreasing levels. Both are evaluated during DC-response sensor testing, and both contribute to the overall accuracy budget.