4. Instrumentation and data acquisition

Far-side occupant excursion scoring depends entirely on the quality of the measurement chain that feeds it. A score is only as trustworthy as the sensors that produced it, the sampling regime that captured it, and the filtering that conditioned it before analysis. This section treats the instrumentation suite as a single coherent system: accelerometers that establish the kinematic reference, load cells that quantify occupant–restraint interaction, high-speed cameras that provide spatial ground truth, and the digital filtering that reconciles these heterogeneous signals into a common, defensible dataset.

Accelerometers: Establishing the Kinematic Reference

Accelerometers serve two distinct roles in far-side testing. Sled- or vehicle-mounted accelerometers record the input pulse that drives the event, while occupant-mounted accelerometers capture the response of the head, chest, and pelvis. The input pulse is the independent variable against which every response is correlated; it must be measured with a sensor whose range and bandwidth comfortably exceed the expected event. A typical frontal or oblique sled pulse may peak between 30 g and 60 g with significant content up to 300–500 Hz, so a ±500 g piezoresistive or capacitive accelerometer with a flat response to at least 1 kHz is a reasonable starting specification.

Occupant instrumentation follows the same logic but with tighter constraints. Head accelerometers are conventionally mounted as a triaxial cluster at the head center of gravity, often via a bite bar or a rigid head-cap adapter, so that measured acceleration corresponds to the rigid-body motion used in HIC and excursion calculations. Chest accelerometers are mounted on a rigid chest plate or directly to the sternum region of the dummy. The critical practice is to document the exact mounting location and orientation, because a triaxial sensor rotated even 10° relative to the anatomical axes introduces cross-axis coupling that contaminates the resultant. Establish the sign convention explicitly before the test: for example, +X forward along the vehicle longitudinal axis, +Y to the occupant's left, +Z upward, with all sensors aligned to this frame or with a documented rotation matrix.

Signal conditioning for accelerometers is usually handled inside the sensor or in a nearby charge amplifier. Piezoelectric units require low-noise coaxial cabling and careful grounding to avoid triboelectric artifacts; piezoresistive units need stable bridge excitation. In either case, the cable route must be strain-relieved so that dummy motion during the event does not tug on connectors and generate spurious spikes.

Load Cells: Quantifying Restraint Interaction

Load cells measure the forces that restraints impose on the occupant and the forces the occupant imposes on the seat and interior. Common locations include seat-belt webbing tension load cells, buckle load cells, seat-pan and backrest load cells, and, in far-side configurations, lateral support or door-trim contact load cells. Each measurement answers a specific question: belt tension reveals how the restraint loads up during excursion, buckle force indicates the direction and magnitude of the restraint reaction, and seat load cells show how the occupant's mass redistributes as the torso translates laterally.

Selection hinges on expected force range and dynamics. Belt tension during a far-side event may reach several kilonewtons, so a 20 kN tension load cell with a 5–10 kHz natural frequency is appropriate; seat-pan cells may see higher quasi-static loads but lower frequency content. Mounting stiffness matters as much as sensor range: a load cell installed with compliant brackets or loose fasteners will measure the bracket dynamics rather than the occupant–restraint force. Torque fasteners to specification, use spherical seating washers where the design permits, and verify that the load path passes through the sensing element rather than bypassing it through adjacent structure.

High-Speed Video: Spatial Ground Truth

High-speed cameras provide the spatial context that point sensors cannot: they show where the head actually traveled, how the torso rotated, and whether the occupant contacted the far-side interior. For excursion scoring, camera placement is a measurement decision, not a documentation afterthought. A minimum of two synchronized views is standard—one lateral view perpendicular to the occupant's excursion plane and one oblique or overhead view to resolve out-of-plane motion. Frame rates of 1000 fps are typical for occupant kinematics; 2000 fps or higher is warranted when tracking small head excursions or rapid belt engagement.

Camera calibration is essential. Place a calibration target with known dimensions in the occupant's plane of motion before the test, record it, and use it to convert pixel coordinates to millimeters. Synchronize all cameras to a common trigger, ideally the same trigger that arms the data acquisition system, so that video frames and sensor samples share a time base. Without a shared trigger, correlating the moment of peak head excursion in video with the peak in the accelerometer trace becomes guesswork.

Data Acquisition and Filtering

The DAQ system must sample every channel at a rate consistent with the highest-frequency content of interest. A practical rule is to sample at least ten times the highest frequency you intend to analyze; for occupant accelerometry with content to 1 kHz, 10–20 kHz per channel is common. Anti-alias filtering must be applied in hardware before sampling, with the cutoff set below the Nyquist frequency, otherwise high-frequency energy folds back into the passband and corrupts the data irreversibly.

Post-acquisition digital filtering follows standards such as SAE J211/1, which specifies channel filter classes. For occupant accelerations, CFC 1000 (a fourth-order Butterworth low-pass at 1650 Hz) is the conventional class; for belt forces, CFC 60 or CFC 180 is often used depending on the analysis. Apply the filter in both forward and reverse directions (zero-phase filtering) when the analysis requires no phase distortion, and document the filter class, cutoff, and order alongside every reported result. A common failure is reporting a peak value from unfiltered data and comparing it against a filtered threshold—an apples-to-oranges comparison that can shift a score by a meaningful margin.

Practical Verification Before the Test

A short pre-test verification sequence catches most instrumentation failures before they become unrecoverable. First, perform a shunt calibration or known-input check on every load cell channel and confirm the DAQ reads the expected engineering value. Second, tap-test accelerometers lightly and confirm each channel responds with a plausible transient and no clipping. Third, record a few seconds of quiescent data and inspect the noise floor; a channel with elevated noise or a slow drift usually indicates a grounding or cable problem. Fourth, fire the trigger and confirm that all cameras and the DAQ arm simultaneously and that the calibration target is visible in every view. Finally, verify that the filter class and sampling rate are recorded in the test log, because these settings are as much a part of the result as the measured values themselves.

When these elements are in place, the instrumentation suite produces a dataset in which accelerometer peaks, belt forces, and video-derived excursions can be compared on a common time base with known filtering. That coherence is what allows a far-side excursion score to be defended—not merely reported.