3. Diagnostic data acquisition

When an airbag fails to deploy, deploys late, or deploys when it should not, the investigation almost never begins with a physical teardown. It begins with data. Modern restraint systems are instrumented by design: the Airbag Control Unit (ACU) continuously records sensor inputs, arming decisions, and deployment commands, and the inflator modules themselves carry manufacturing lineage that can be traced back to a specific production lot. The discipline of diagnostic data acquisition is about knowing which data sources exist, what each one can and cannot prove, and how to correlate them into a defensible root cause.

This chapter covers four primary evidence streams — Event Data Recorder (EDR) output, crash pulse reconstruction, Diagnostic Trouble Code (DTC) logs, and inflator lot traceability — and shows how they are combined during a restraint system failure investigation.

EDR Data: What the ACU Actually Records

The EDR is not a separate black box in most vehicles; it is a firmware function inside the ACU (or, in some architectures, inside the powertrain or safety domain controller). It captures a time-stamped snapshot around a triggering event. Typical recorded channels include:

• Pre-crash and crash kinematics: vehicle longitudinal delta-v, lateral delta-v, yaw rate, and speed at discrete sample points (commonly 10 ms or 2 ms intervals depending on the OEM and regulatory regime).
• Driver inputs: brake switch state, accelerator position, steering angle.
• Restraint status: seat belt buckle status per seating position, occupant classification sensor (OCS) state, and whether each squib circuit was commanded to fire.
• Timing markers: time from algorithm wake-up to arming, and time from arming to deployment command.

Two practical cautions apply. First, EDR data is not a full bus log — it is a curated subset, and the selection logic is OEM-specific. Second, EDR records are written once and locked; if the ACU loses power mid-event, the record may be truncated. Always image the EDR through the OEM-approved tool (for example, a Bosch CDR kit with the correct vehicle interface) before any other diagnostic action, because clearing DTCs or reflashing can overwrite volatile buffers.

Crash Pulse Reconstruction

The crash pulse is the acceleration-versus-time signature experienced by the vehicle body. It is the single most useful artifact for answering "should the restraint have fired?" because deployment algorithms are calibrated against pulse characteristics — principally peak acceleration, pulse duration, and the integral of acceleration (delta-v).

Reconstruction typically proceeds from two directions. The forward path uses ACU accelerometer channels recorded in the EDR, integrated once to velocity and twice to displacement. The independent path uses a dedicated crash test accelerometer or a vehicle-mounted data logger to validate the ACU's own measurement. Discrepancies between the two are themselves diagnostic: a large offset suggests accelerometer drift or a mounting fault; a phase lag suggests filtering differences.

Consider a worked example. An ACU reports a longitudinal pulse with a peak of approximately 28 g at t = 42 ms and a computed delta-v of 19 km/h over a 120 ms window. The restraint calibration for this platform arms the frontal loop above a delta-v threshold of roughly 25 km/h within the same window. The recorded delta-v sits below threshold, so a no-deploy outcome is consistent with the calibration — provided the pulse data is trustworthy. The investigator must then verify the accelerometer channel against an independent source before concluding the algorithm behaved correctly.

# Illustrative pulse integration (example values, not measured data)
t_ms   = [0, 10, 20, 30, 40, 50, 60, 80, 100, 120]
a_g    = [0,  4, 12, 22, 28, 24, 16,  8,   3,   0]
# trapezoidal integration to delta-v (km/h)
dv = 0
for i in range(1, len(t_ms)):
    dt = (t_ms[i] - t_ms[i-1]) / 1000.0
    dv += 0.5 * (a_g[i] + a_g[i-1]) * 9.81 * dt
dv_kmh = dv * 3.6   # ~19 km/h for the values above

The result is only as good as the sample rate and the anti-alias filtering. A 10 ms sample interval can miss a short, high-amplitude spike that a 2 ms interval would capture — a real limitation when investigating low-speed or pole-type impacts.

DTC Logs and Freeze-Frame Data

DTCs are the ACU's own fault history. For restraint investigations, the relevant codes cluster into a few families: squib circuit resistance out of range (open or short), sensor plausibility faults, communication faults on the safety CAN, and internal ECU faults. What matters more than the code itself is the freeze-frame or extended data attached to it — the operating conditions at the moment the fault was set.

Read DTCs with the OEM scan tool in the correct diagnostic session, and record the order and timestamps of codes, not just their presence. A squib open-circuit code that appears only after a crash event tells a very different story from one present for thousands of ignition cycles.

Inflator Lot Traceability

When the failure mode is inflator-related — no deployment, under-inflation, or over-aggressive deployment — the physical module must be tied back to its manufacturing record. Every inflator carries a lot or serial marking, and the OEM maintains a traceability database linking that marking to propellant batch, assembly line, test data, and date of manufacture.

The traceability workflow is straightforward but must be disciplined:

1. Photograph the inflator marking in situ before removal, with a scale reference.
2. Record the full alphanumeric lot/serial string exactly, including any laser-etched or dot-peened characters that are easily misread.
3. Query the OEM traceability system to retrieve the production record for that lot.
4. Compare the retrieved record against the failure symptom — for example, a lot with a known propellant moisture excursion versus a symptom of weak gas output.

Correlating the Four Streams

No single stream proves root cause. The EDR establishes what the ACU decided and when; the crash pulse establishes whether that decision was physically justified; the DTC log establishes whether the system was healthy beforehand; and lot traceability establishes whether the hardware itself was suspect. The diagram below shows how these streams converge on a root cause determination.

The correlation step is where most investigations succeed or fail. Align every stream on a common event time zero — typically the first accelerometer threshold crossing — and build a single timeline. If the EDR shows a deployment command but the squib circuit DTC shows an open, the fault is in the loop, not the algorithm. If the pulse is below threshold and no command was issued, the algorithm is exonerated and attention shifts to whether the threshold itself was appropriate for the impact type. If the pulse is above threshold, a command was issued, and the inflator still did not fire, lot traceability becomes the primary lead.

Practical Acquisition Checklist

The following sequence reflects sound field practice and can be adapted to most OEM workflows:

1. Secure the vehicle and preserve power to the ACU if safe to do so.
2. Image the EDR with the OEM tool; hash and archive the raw file.
3. Read and photograph all DTCs, including freeze-frame and timestamps, before clearing anything.
4. Record inflator lot/serial markings photographically before removal.
5. Collect independent accelerometer data if a logger was present.
6. Query the traceability database and retrieve the production record.
7. Build the unified timeline and document each conclusion with its supporting stream.

Each stream answers a different question, and the strength of the final root cause rests on how well they agree. When they conflict, the conflict itself is the finding — and it usually points to the next test to run.