Sensitivity analysis of inertial test equipment is the core process for verifying and calibrating the measurement accuracy of accelerometers, gyroscopes, and inertial measurement units (IMUs) used across aerospace, automotive, and industrial applications. This technical article explains the key sensitivity parameters, internationally recognized calibration standards such as the ISO 16063 series and IEEE 1554, laboratory-grade test methods, and practical answers to frequently asked questions.
Technical Overview
Inertial test equipment refers to the precision instrumentation used to characterize sensors that measure motion without external references, including linear accelerometers, angular rate gyroscopes, and multi-axis inertial measurement units. Typical equipment in a modern inertial laboratory includes precision rate tables (two-axis and three-axis turntables), precision dividing heads and indexing heads, linear vibration exciters, angular vibration exciters, precision centrifuges, thermal chambers, and integrated multi-axis motion simulators.
Sensitivity, often called the scale factor, is defined as the ratio of the change in sensor output to the corresponding change in the physical input along the sensitive axis. Depending on the sensor type, sensitivity is expressed in units such as pC/(m/s²) or mV/g for accelerometers, and mV/(°/s) or V/(rad/s) for gyroscopes. Sensitivity analysis is the systematic evaluation of this transfer coefficient together with its deviation, nonlinearity, temperature dependence, cross-axis coupling, and long-term drift, all quantified within a traceable uncertainty budget.
Because inertial navigation and guidance systems accumulate errors over time, even a small sensitivity error propagates directly into position and attitude error. Sensitivity analysis therefore serves three purposes: it establishes metrological traceability to national standards, it feeds deterministic error models used for in-system compensation, and it validates whether a device meets its datasheet specification before deployment. Accredited third-party laboratories performing this work typically follow the ISO 16063 series for vibration calibration, IEEE 1554 for inertial sensor test equipment and data acquisition practices, and ISO/IEC Guide 98-3 (GUM) for uncertainty evaluation.
Testing Samples
Accelerometers
Linear accelerometer samples subjected to sensitivity analysis include MEMS capacitive accelerometers, piezoelectric (charge and IEPE voltage-output) accelerometers, piezoresistive accelerometers, and servo force-balance (closed-loop) accelerometers covering ranges from ±2 g consumer-grade parts to ±100,000 g shock sensors. Reference-grade standard accelerometers used as transfer standards in comparison calibration are also characterized.
Gyroscopes and Angular Rate Sensors
Angular rate samples include MEMS vibrating-structure gyroscopes, fiber optic gyroscopes (FOG), ring laser gyroscopes (RLG), and hemispherical resonator gyroscopes (HRG). These devices are analyzed for angular rate sensitivity, in-run bias stability, and angle random walk using rate tables with angular position accuracy at the arc-second level.
Inertial Measurement Units and Integrated Systems
Complete 6-axis, 9-axis, or tactical- and navigation-grade IMUs, attitude and heading reference systems (AHRS), inclinometers, tilt sensors, and inertial navigation systems (INS) are tested as integrated assemblies so that the sensitivity of each channel, axis alignment, and inter-axis coupling are evaluated in the final system configuration.
Testing Items
A comprehensive sensitivity analysis program typically covers the following measurands:
- Scale factor (sensitivity) calibration at reference conditions, including sensitivity deviation from nominal value
- Scale factor linearity over the full dynamic range, and nonlinearity expressed as a percentage of full-scale or best-fit-line residual
- Scale factor asymmetry, quantifying differences between positive and negative input directions
- Scale factor repeatability over successive power cycles and test sessions
- Bias (zero offset), bias stability, and bias repeatability for both accelerometers and gyroscopes
- Cross-axis sensitivity, quantifying unwanted output response to inputs perpendicular to the sensitive axis
- Temperature sensitivity coefficient and thermal hysteresis across the rated operating range, for example −40 °C to +85 °C
- Frequency response, bandwidth, resonance behavior, and dynamic sensitivity flatness for vibration-sensing channels
- Resolution, threshold, noise density, velocity random walk, and angle random walk derived from Allan variance analysis
- Axis misalignment and tri-orthogonality errors within multi-axis IMUs, as well as gyroscope g-sensitivity and g²-sensitivity coefficients
Testing Methods
Static Gravity Reference Methods
Accelerometer sensitivity and bias at low acceleration levels are determined using Earth's gravity as a reference. The device is mounted on a precision dividing head or indexing table and rotated through precise angular increments, typically in 10° or 30° steps over a full 360° rotation. The projection of gravity onto the sensitive axis follows a sinusoidal model, and least-squares fitting of the output versus sine and cosine terms yields sensitivity, bias, and axis misalignment simultaneously. This method delivers expanded uncertainties on the order of 0.1% or better for high-precision setups.
Primary Vibration Calibration by Laser Interferometry
For dynamic sensitivity of accelerometers, primary calibration is performed per ISO 16063-11 using a laser interferometer (Michelson or Mach-Zehnder type with quadrature demodulation or fringe counting) to measure the sinusoidal displacement amplitude of the vibration exciter. The ratio of the accelerometer electrical output to the derived acceleration gives magnitude and phase of sensitivity over a defined frequency and amplitude grid, with achievable expanded uncertainties below 0.5%. Absolute angular vibration calibration of rotational sensors follows the corresponding methods of the ISO 16063 series for angular excitation.
Comparison Calibration Against a Reference Transducer
In the workhorse laboratory method defined by ISO 16063-21, the device under test is mounted back-to-back with a calibrated reference accelerometer on an air-bearing vibration exciter. Both sensors experience the same motion, and the sensitivity of the test device is derived from the output ratio. Sine sweep, stepped sine, and broadband random excitation are used to characterize magnitude and phase of sensitivity versus frequency. This method is widely applied by accredited third-party laboratories for routine verification and recalibration.
Rate Table Methods for Gyroscopes
Gyroscope sensitivity, linearity, and asymmetry are measured on precision rate tables by commanding a sequence of constant angular rates in both rotation directions, for example ±0.01, ±1, ±10, ±100, and ±500 °/s. Least-squares fitting of output versus input rate yields the scale factor and its nonlinearity. The Earth rate method and multiposition (static) method are used to separate bias from Earth's rotation component, while symmetrical position averaging cancels thermal and magnetic drift influences. IEEE 952 and IEEE 1554 provide the recommended test procedures, data acquisition formats, and error model terminology.
Precision Centrifuge Testing
For high-g accelerometer ranges beyond the capability of shakers, a precision centrifuge generates sustained centripetal acceleration derived from radius and angular velocity measurements. As described in IEEE 1293 practice, a series of acceleration points up to tens of thousands of g permits extraction of the full nonlinear calibration polynomial, including second-order and third-order coefficients, cross terms, and misalignment errors.
Environmental and Temperature Sensitivity Evaluation
A temperature-controlled chamber, either standalone or integrated with a rate table, cycles the device across its operating range while periodic sensitivity checks are performed at dwelling temperatures. The slope of sensitivity versus temperature defines the temperature coefficient, and the separation between heating and cooling curves quantifies thermal hysteresis. Combined temperature-vibration or temperature-rotation profiles assess sensitivity stability under realistic mission environments per IEC 60068 environmental test methodology.
Stochastic Characterization by Allan Variance
Long static data records acquired under stable thermal conditions are processed using Allan variance techniques to identify quantization noise, angle or velocity random walk, bias instability, and rate random walk. These stochastic parameters complement deterministic sensitivity coefficients and are essential inputs for Kalman filter tuning in navigation systems.
Applications
- Aerospace and defense: acceptance testing and periodic recalibration of gyroscopes, accelerometers, and IMUs for flight control, guidance and navigation, and missile seeker stabilization platforms
- Automotive: validation of MEMS inertial sensors used in electronic stability control, airbag crash sensing, rollover detection, and dead-reckoning functions supporting ADAS and autonomous driving
- Industrial condition monitoring: sensitivity verification of accelerometers deployed for machinery vibration monitoring, predictive maintenance, and structural health monitoring programs
- Energy and exploration: qualification of high-temperature, high-shock inertial sensors for downhole drilling guidance and pipeline inspection tools
- Surveying and geodesy: calibration of inertial units for mobile mapping systems, airborne gravimetry support hardware, and precision tilt references
- Robotics and unmanned systems: characterization of IMUs for drones, autonomous vehicles, legged robots, and underwater vehicles where attitude accuracy depends directly on scale factor quality
- Metrology and quality assurance: accredited third-party laboratories providing traceable calibration certificates supporting ISO 9001 quality systems, laboratory accreditation, and product certification
FAQ
What is the difference between sensitivity and scale factor?
The two terms describe the same transfer characteristic: the ratio of change in output to change in input. In vibration metrology the term sensitivity is standard, expressed for example in pC/(m/s²) or mV/g, while in inertial navigation literature the term scale factor is more common, often expressed as a dimensionless ratio relative to a nominal value or in pulse-equivalent units for digital output sensors.
How often should inertial sensor sensitivity be verified?
Typical recalibration intervals are 12 months for industrial accelerometers and reference standards, and 6 to 24 months for tactical- and navigation-grade inertial units depending on the manufacturer specification, accumulated operating hours, and exposure to shock or thermal stress. Shortened intervals are recommended after any suspected overload event.
What accuracy level can be achieved for sensitivity calibration?
Primary laser interferometric calibration can achieve expanded uncertainties (k = 2) below 0.5% for magnitude and better than 1° for phase. Comparison calibration typically achieves 1% to 3%, while static gravity methods and precision centrifuge methods can reach 0.1% to 0.5% under well-controlled conditions. The exact figures depend on the device type, range, and the uncertainty budget documented in the calibration certificate.
Which standards govern inertial test equipment sensitivity analysis?
The principal references are the ISO 16063 series for vibration and shock transducer calibration (including primary laser interferometric methods and comparison methods), IEEE 1554 for inertial sensor test equipment and data acquisition, IEEE 952 for gyroscope specification and test procedures, IEEE 1293 for precision centrifuge testing of accelerometers, and applicable national standards such as the GB, EN, and JIS families. Environmental conditioning commonly references IEC 60068 test methods.
How is cross-axis sensitivity measured and why does it matter?
Cross-axis sensitivity is measured by applying a known excitation along each axis perpendicular to the sensitive axis, using either a precision rotary stage with gravity, a shaker with a specialized cross-axis fixture, or a rate table for gyroscopes. The output produced by the perpendicular input, expressed as a percentage of the primary sensitivity, must be compensated in multi-axis systems because uncompensated transverse coupling directly degrades attitude and navigation accuracy.
Can sensitivity be verified in the field without a laboratory?
Limited field verification is possible using built-in self-test functions, gravity reference checks at 0 g, ±1 g positions, and Earth rate observations for gyroscopes. However, these checks only confirm gross functionality and bias trends. Full sensitivity analysis with traceable uncertainty statements requires calibrated laboratory equipment and controlled environmental conditions provided by accredited third-party laboratories.