1. Hybrid III Head Acceleration Trace Review
Why the Raw Trace Comes First
Every Hybrid III head injury metric — HIC, 3 ms clip, peak resultant — is derived from the same source: the three orthogonal head accelerometer channels (typically Ax, Ay, Az in the SAE J211 head coordinate system, with +Ax forward, +Ay left, and +Az up). Before any calculation, the analyst must confirm that the trace is physically plausible. A HIC computed from a trace that contains a pre-trigger offset, a clipped pulse, or an un-filtered resonance artifact is not a conservative estimate — it is simply wrong. This case establishes the review routine: segment the trace into pre-impact, impact pulse, and post-impact regions, and record the boundaries and quality flags that later metric calculations will depend on.
The Three Regions of a Head Acceleration Trace
A clean Hybrid III head trace from a frontal sled or barrier test has a characteristic shape that can be segmented by inspection:
- Pre-impact baseline — the interval from the start of the recording to the first sustained deviation from the quiescent level. In a well-configured acquisition this is flat within the noise floor (typically a few g peak-to-peak on a 500 g full-scale channel). Any drift, step, or 50/60 Hz pickup here is a data-quality flag, not a physical event.
- Impact pulse — the main acceleration excursion, beginning at the pulse onset (the point where the signal rises decisively above the noise band) and ending when the signal returns to within the noise band and stays there. This is the interval that feeds HIC and peak calculations.
- Post-impact ringing — decaying oscillation after the pulse, caused by structural resonance of the head/neck assembly and the dummy's internal mounting. Ringing is real signal, but it must not be mistaken for a second impact, and it must not be allowed to extend the HIC integration window artificially.
Worked Review of a Frontal Sled Trace
The following example uses a synthetic but representative Hybrid III 50th percentile head Ax trace, sampled at 10 kHz and filtered to SAE J211 Channel Filter Class 1000 (CFC 1000). The numbers below are teaching values chosen to make the segmentation logic explicit; they are not measured test data.
| Region | Time window (ms) | Observed characteristic | Review decision |
|---|---|---|---|
| Pre-impact baseline | 0 – 18 | Mean ≈ 0.4 g, peak-to-peak ≈ 2.1 g, no drift | Accept; record noise floor |
| Pulse onset | 18 – 22 | Rise from 1 g to 15 g within 4 ms | Mark onset at 18.6 ms (first sample > 5× noise σ) |
| Impact pulse | 18.6 – 96 | Peak 148 g at 41 ms; returns below 5 g at 96 ms | Accept; pulse duration ≈ 77 ms |
| Post-impact ringing | 96 – 180 | Decaying oscillation, initial amplitude ≈ 9 g, period ≈ 6.5 ms | Flag as ringing; exclude from pulse metrics |
The onset threshold is the key quantitative decision. A practical rule is to compute the standard deviation σ of the pre-impact baseline and declare onset at the first sample that exceeds 5σ and remains above it for at least 1 ms. With σ ≈ 0.35 g in the example, the threshold is ≈ 1.75 g, and the first qualifying sample falls at 18.6 ms. This is reproducible and independent of the analyst's eye.
Segmenting the Trace Visually
The diagram below shows the same trace with the three regions annotated, the onset threshold band, and the ringing envelope. Use it as the mental template when you open a new trace: locate the flat baseline, find the decisive rise, and confirm that the decay after the pulse is oscillatory rather than a second step.
Quality Flags to Record Before Calculation
The review is not complete until the analyst has written down the segmentation and any anomalies. The following checklist captures the fields that downstream HIC and peak calculations will consume. Fill it in for every trace; a trace that cannot be filled in cleanly should be escalated rather than processed.
| Field | Example value | Why it matters |
|---|---|---|
| Channel / axis | Ax (forward) | Sign convention must match the dummy spec before combining axes |
| Sample rate | 10 kHz | CFC 1000 requires ≥ 10 kHz; lower rates alias the pulse |
| Filter class | SAE J211 CFC 1000 | HIC is defined on CFC 1000 data; other classes change the peak |
| Pre-impact noise σ | 0.35 g | Sets the onset threshold (5σ) |
| Onset time | 18.6 ms | Start of the HIC integration window |
| Pulse end time | 96 ms | End of the pulse; ringing excluded |
| Peak value / time | 148 g @ 41 ms | Sanity check against expected severity |
| Ringing period | ≈ 6.5 ms | Distinguishes structural resonance from a second impact |
| Anomalies | None | Any drift, step, or clip must be logged and resolved |
Common Failure Symptoms and What They Indicate
- Baseline drifts upward before the pulse. Usually a charge amplifier or cabling issue, or thermal drift on a piezoelectric channel. Re-zero or re-condition the channel; do not subtract a linear trend and proceed.
- Pulse appears clipped at a round number (e.g., exactly 500 g). The channel is saturating at full scale. The true peak is unknown; the trace cannot support a HIC calculation.
- Ringing amplitude comparable to the pulse. Suggests a loose accelerometer mount or a damaged cable. Inspect the mounting block torque and cable strain relief before trusting the pulse.
- Onset threshold never crossed cleanly. Noise floor is too high relative to the event, or the trigger fired late. Check gain settings and trigger source.
With the trace segmented and the quality flags recorded, the data is ready for metric calculation. The onset and pulse-end times established here define the integration window; the noise σ defines the threshold; and the ringing period tells the analyst what to expect when the resultant trace is assembled from all three axes.