2. Pendulum test rig architecture

A pendulum impact rig is deceptively simple in concept: raise a known mass to a known height, release it, and let it strike a test article. In practice, the validity of every result depends on how precisely the rig controls the impact energy, the impact velocity vector, and the timing of the measurement chain. This section walks through the five subsystems that define a pendulum rig — the structural frame, the pivot assembly, the release mechanism, the instrumentation package, and the data acquisition (DAQ) system — and explains how each one constrains what the test can and cannot measure.

Structural Frame and Reaction Path

The frame exists to hold the pivot rigidly while absorbing the reaction impulse of the strike. A compliant frame lets the pivot translate during impact, which corrupts the effective impact velocity and adds low-frequency content to every accelerometer channel. Frames are typically welded steel box sections or bolted I-beam assemblies, anchored to a seismic mass or a reinforced floor slab through grouted base plates.

Two frame parameters matter most. The first is static stiffness at the pivot: a useful design target is that the first natural frequency of the loaded frame be at least five times the highest frequency of interest in the measurement. For an airbag deployment event where accelerometer data is filtered to 1650 Hz per SAE J211 Channel Class 1000, this pushes the frame's first mode above roughly 8 kHz, which in practice means heavy sections and short spans. The second is anchor integrity: preloaded anchor bolts should be torque-checked before each test campaign, because a single loose anchor can shift the frame mode into the measurement band.

Pivot Assembly and Bearing Selection

The pivot defines the pendulum's swing plane and its friction losses. Three configurations are common:

• Rolling-element bearings (deep-groove ball or tapered roller): low friction, but finite radial clearance that allows micro-translation under impact load.
• Crossed-roller or air bearings: near-zero clearance and very low friction, used when impact velocity repeatability must be better than 1%.
• Knife-edge or flexure pivots: used for small-scale rigs where friction must be negligible, but limited to low loads.

Pivot height sets the drop geometry. If the pendulum arm length is L and the release angle from vertical is θ, the theoretical impact velocity at the arm's lowest point is:

v = sqrt(2 * g * L * (1 - cos(theta)))

Example: L = 2.50 m, theta = 60 deg, g = 9.81 m/s^2
v = sqrt(2 * 9.81 * 2.50 * (1 - 0.500))
v = sqrt(24.525) = 4.95 m/s

Real rigs lose energy to bearing friction and air drag, so the measured velocity is typically 1–3% below this value. That gap is why an independent velocity gate is mandatory rather than optional.

Release Mechanism

The release mechanism determines when the pendulum actually starts moving, and any delay or asymmetry here propagates directly into velocity error. Three families dominate:

The release must also produce a clean separation: no residual contact that drags on the arm after the trigger. A common failure mode is a solenoid that releases the arm but leaves the keeper plate brushing the arm face, adding a few milliseconds of drag and reducing impact velocity by 2–4%. The diagnostic is a velocity-gate reading consistently below the calculated value while the release delay looks normal.

Instrumentation Package

A pendulum rig for airbag work typically carries four measurement groups:

1. Velocity gates — two or more optical or magnetic sensors spaced a known distance apart along the swing path. They measure the actual impact velocity and provide the trigger for the DAQ.
2. Accelerometers — mounted on the pendulum face, on the test article, and on the frame near the pivot. Piezoresistive types are preferred for impact because they respond to DC and have no significant zero shift.
3. Load cells — either in the pendulum face or in the reaction fixture, to capture the force-time history of the strike.
4. High-speed video — typically 1000–5000 fps, used to verify the impact vector and to cross-check the accelerometer-derived velocity.

Accelerometer mounting is the most frequent source of bad data. A stud-mounted sensor with a thin layer of grease or mica washer gives the best high-frequency response; adhesive mounting is acceptable below about 5 kHz; magnetic mounting should be avoided entirely for airbag impact work because it resonates in the measurement band.

Data Acquisition and Signal Chain

The DAQ system must satisfy three constraints simultaneously: sufficient sample rate, anti-alias filtering, and channel-to-channel skew control. For airbag impact testing, SAE J211 defines the filter classes — Channel Class 60, 180, 600, and 1000 — where the class number is the cutoff frequency in Hz. A Class 1000 channel requires a sample rate of at least 10 kHz, and 20 kHz is common practice.

The signal chain runs: sensor → charge amplifier or bridge conditioner → anti-alias filter → ADC → storage. Each stage introduces its own error. Charge amplifiers drift with temperature; bridge conditioners need shunt calibration before each campaign; anti-alias filters must be applied before the ADC, not as a post-processing step, or aliased content will fold into the passband.

System Integration: How the Subsystems Interact

The five subsystems are not independent. Frame stiffness sets the noise floor; pivot clearance sets velocity repeatability; release timing sets the trigger reference; instrumentation placement sets what the data actually represents; and the DAQ chain sets the bandwidth and phase fidelity. A rig is only as good as its weakest link, and the weakest link is usually the one that was specified last.

Practical Verification Before a Campaign

Before running a test series, verify the rig as a system rather than as individual components. A short, repeatable checklist catches most integration faults:

• Velocity gate calibration: pass a known-length flag through the gates and confirm the measured velocity matches the calculated value within 0.5%.
• Release delay measurement: trigger the release and record the time from trigger command to first gate crossing. Compare against the manufacturer's specification.
• Shunt calibration: inject a known shunt resistor across each bridge channel and confirm the DAQ reads the expected equivalent input.
• Frame tap test: strike the frame near the pivot with a modal hammer and confirm the first mode is above the measurement band.
• Dry run: swing the pendulum with no test article and compare measured velocity against the calculated value. A deviation greater than 3% indicates friction or release drag that must be resolved before the campaign proceeds.

Understanding how these five subsystems interact is what separates a rig that produces repeatable, defensible data from one that produces numbers. The next section moves from the rig itself to the test article and the specific impact conditions that airbag validation requires.