3. Measurement system architecture

Trigger synchronization is not a property of a single device; it is an emergent property of the entire measurement chain. A crash test may involve 60–200 channels distributed across the vehicle, the barrier or impactor, the sled system, and the facility. Each channel passes through a sensor, a cable, a data acquisition unit (DAU), and a recording or telemetry path before it becomes a timestamped sample. If any link in that chain introduces an unknown or uncompensated delay, the resulting dataset will show events at the wrong relative times — even if every individual device is functioning correctly. This chapter maps the architecture of a typical crash measurement system and identifies where synchronization is established, where it can be lost, and what design choices keep the whole chain coherent.

The four functional tiers

A crash measurement system can be decomposed into four functional tiers, each with distinct timing responsibilities:

  • Sensors and transducers — accelerometers, load cells, string potentiometers, pressure transducers, strain gauges, and angular rate sensors. Their timing contribution is the mechanical response delay of the sensing element plus any internal signal conditioning (e.g., anti-alias filtering in a smart sensor).
  • Data acquisition units (DAUs) — the on-vehicle or on-sled modules that excite, amplify, filter, and digitize sensor signals. DAUs own the sample clock and are the primary place where a common time base is established.
  • Onboard recorders — the logging device that stores DAU output locally, often with an internal battery and a pre-trigger ring buffer. The recorder defines the trigger instant and the recording window.
  • Offboard instrumentation — facility-side systems: the central trigger generator, high-speed cameras, lighting controllers, the test controller, and the data reduction workstation. These must be aligned to the same event as the onboard chain.

The critical architectural principle is that one tier must be the time master. In most facilities the central trigger generator or the test controller issues a hardware trigger pulse that is distributed to every tier simultaneously; each device then timestamps that pulse edge against its own clock. Synchronization quality is therefore determined by trigger distribution skew and by each device's clock stability over the recording window.

Inside a DAU: sample clock, trigger latch, and pre-trigger buffer

A representative DAU used in vehicle crash testing digitizes at 10–20 kS/s per channel with 16-bit resolution. Internally, three timing elements matter:

  • The sample clock, typically derived from a crystal oscillator with stability on the order of ±25 ppm. Over a 500 ms recording window this contributes at most about 12.5 µs of drift — negligible compared with mechanical phenomena, but relevant when comparing channels across different DAUs whose clocks are not shared.
  • The trigger latch, a hardware register that captures the sample counter value at the moment the external trigger edge arrives. This is what allows post-test alignment: the trigger's position within the record is known to one sample period.
  • The pre-trigger ring buffer, which continuously overwrites the oldest samples so that data before the trigger is retained. A typical configuration stores 100–200 ms of pre-trigger data, which is essential for capturing the approach phase and for verifying that the trigger fired at the intended event.

When multiple DAUs are used — common when channel count exceeds a single unit's capacity, or when modules are placed close to sensors to shorten cable runs — they must share either a common sample clock or a common trigger with individually latched counters. Sharing the clock is preferable because it eliminates inter-module drift entirely; sharing only the trigger leaves each module's oscillator free-running, and the resulting relative drift must be corrected in post-processing.

Place DAUs physically close to their sensor clusters to minimize cable length, but route the trigger and clock distribution lines separately from high-current firing lines and camera strobe cables. Inductive coupling from a strobe discharge into a trigger line is a classic cause of spurious or jittered triggers.

Onboard recorder behavior and the trigger instant

The onboard recorder is often conflated with the DAU, but they play different roles. The recorder's job is to define when the recording window opens and closes, and to preserve data through the crash event (battery-backed memory, shock-rated enclosure). Many recorders support multiple trigger sources: a hardware trigger input from the facility, an internal threshold trigger on a selected channel (e.g., acceleration exceeding 2 g), and a manual arm/fire command. In a synchronized architecture, the hardware trigger from the facility is the authoritative event; threshold triggers are used as backups or for standalone tests.

A practical detail: the recorder's trigger timestamp and the DAU's latched sample counter must be reconciled. If the recorder timestamps the trigger using its own real-time clock while the DAU latches a sample index, the mapping between the two is established by a single calibration measurement — injecting a known pulse and observing where it appears in both records. This mapping should be verified at the start of each test day, not assumed from the previous campaign.

Offboard instrumentation and cross-domain alignment

Offboard systems introduce the largest synchronization challenges because they operate in different domains. High-speed cameras run on their own frame clocks at 1000–2000 fps; a camera at 1000 fps has a 1 ms frame interval, so its timing resolution is inherently coarser than the DAU's 50–100 µs sample interval. The camera's trigger input starts the recording, and the frame in which the trigger pulse appears (or a dedicated sync LED visible in-frame) provides the alignment reference. Lighting controllers must fire in phase with camera exposure; a strobe that fires 200 µs late relative to the exposure window produces a dim or missing frame rather than a timing error in the data, which is a useful diagnostic distinction.

The test controller ties the facility together: it arms the DAUs and recorders, starts the cameras, and issues the fire command to the propulsion system. The interval between the fire command and the actual release of the propulsion system (the "fire-to-release" latency) is a known, measurable offset that must be characterized and applied when comparing the trigger event to the physical impact. Facilities typically measure this latency with a contact sensor on the propulsion release mechanism and record it as a per-test metadata field.

Measurement chain tiers and trigger distribution Sensors Accelerometers vehicle, sled, barrier Load cells barrier, seat, belt String pots / IR-TRACC displacement Pressure / strain transducers Angular rate sensors rotation DAUs DAU-A (front) 16 ch, 20 kS/s sample counter latch DAU-B (rear) 16 ch, 20 kS/s shared clock input DAU-C (barrier) 8 ch, 20 kS/s isolated ground Pre-trigger buffer 100–200 ms ring Onboard recorder Logging unit battery-backed shock-rated trigger timestamp Trigger sources HW / threshold / manual authoritative = HW Recording window pre + post trigger e.g. 200 ms + 500 ms Offboard Trigger generator time master HW pulse out High-speed cameras 1000–2000 fps sync LED in frame Lighting controller strobe phase vs exposure Test controller arm / fire fire-to-release latency shared clock Analog / digital signal path Hardware trigger distribution (time master → all tiers) Shared sample clock

Where synchronization is lost: a diagnostic walk-through

Consider a test in which the vehicle accelerometers and the barrier load cells appear to disagree by roughly 3 ms at the onset of impact. The DAUs are separate units, each with its own oscillator, and only the trigger is shared. Over a 500 ms window, a 25 ppm oscillator difference produces about 12.5 µs of drift — far too small to explain 3 ms. The more likely causes, in order of probability, are: (1) the barrier DAU's trigger input was configured with a digital filter or debounce that delayed the latch by several milliseconds; (2) the load cell amplifier's low-pass filter (e.g., 1 kHz, 2-pole) introduced a group delay of roughly 0.3–0.5 ms, not 3 ms, so this alone is insufficient; (3) the trigger cable to the barrier DAU was routed alongside a strobe cable and picked up a spurious edge that latched the counter early, making the true impact appear late in the record. The diagnostic procedure is to inject a common electrical pulse into all DAU trigger inputs simultaneously and compare the latched sample indices; any channel whose latch differs by more than one sample period indicates a distribution or configuration problem rather than a sensor problem.

Do not "fix" apparent timing offsets by shifting channels in post-processing until the physical cause is identified. A channel shifted to match another may be correct for one test and wrong for the next, and the shift can mask a genuine sensor or cabling fault.

Architecture choices that preserve synchronization

Three design decisions have outsized influence on timing coherence. First, distribute a common sample clock to all DAUs rather than relying on trigger-only synchronization; this removes oscillator drift as a variable. Second, keep the trigger distribution star-topology from a single generator, with equal-length or documented-length cables, so that propagation skew is known and bounded. Third, record a synchronization reference in every domain — a sync LED in camera frames, a trigger marker channel in the DAU record, and a fire-to-release contact channel — so that cross-domain alignment can be verified after the test rather than assumed. Together these choices reduce the synchronization problem from an open-ended uncertainty to a small set of measurable, documented offsets that can be applied consistently across a test campaign.

← Previous 📖 Contents Next →