3. ES-2 Dummy Rib Deflection Sensor Linearity Test

Why Linearity and Hysteresis Matter on ES-2 Rib Deflection Sensors

The ES-2 dummy measures lateral chest compression with three rib deflection sensors (upper, middle, lower rib) that are essentially potentiometer-type displacement transducers. Each sensor converts rib-to-spine displacement into a voltage that the data acquisition system scales to millimetres. Injury criteria such as rib deflection and the derived chest compression rate depend directly on this scaling, so a sensor whose output is non-linear or exhibits hysteresis will bias the injury assessment even when the channel is electrically healthy. A zero and gain check alone cannot detect this: a sensor can read correctly at 0 mm and at full stroke yet deviate in the middle of its range or return to a different voltage after a load-unload cycle.

This case establishes a bench procedure to quantify both effects on the ES-2 rib deflection sensors using a calibration fixture, then applies explicit acceptance limits to decide whether a sensor must be replaced before the dummy is released for a test.

Test Setup and Instrumentation

The sensor is removed from the rib (or the rib is blocked and the sensor actuated through the fixture) and mounted in a calibration fixture that provides a controlled, dial-indicator-referenced displacement input along the sensor's measurement axis. The fixture must hold the sensor body rigidly and move only the sensor's input element, so that fixture compliance does not appear as sensor hysteresis.

• Displacement reference: a calibrated dial indicator or digital displacement gauge with 0.01 mm resolution, zeroed at the sensor's mechanical zero.
• Excitation and readout: the dummy's normal DAQ channel, using the production excitation voltage and the production scaling (mm/V) so the measured values are directly comparable to test data.
• Stroke range: 0 mm to the sensor's rated full stroke (for ES-2 rib deflection sensors this is typically on the order of 0–75 mm; confirm the rated stroke from the sensor datasheet before setting the test range).
• Load points: 0, 25, 50, 75 and 100 % of rated stroke, approached in ascending order, then in descending order, with a short dwell at each point to let the reading settle.

Record the DAQ output in millimetres at each commanded displacement for both the ascending and descending passes. Three complete cycles are recommended; the first cycle is treated as a conditioning pass and the second and third are used for the acceptance decision.

Procedure

1. Mount the sensor in the fixture and connect it to its normal DAQ channel. Confirm the channel reads near zero at the mechanical zero position; if the offset exceeds the channel's zero tolerance, note it but continue — offset is handled separately from linearity.
2. Command the ascending sequence 0 → 25 → 50 → 75 → 100 % stroke. At each point, wait for the reading to stabilise (typically 2–3 s), then record the commanded displacement and the DAQ reading.
3. Command the descending sequence 100 → 75 → 50 → 25 → 0 % stroke, recording in the same way.
4. Repeat steps 2–3 for three cycles. Discard cycle 1 as conditioning.
5. Compute, for each load point, the mean ascending reading and the mean descending reading across cycles 2 and 3.

Computing Linearity and Hysteresis

Two quantities are extracted from the recorded data:

• Linearity error: fit a least-squares straight line to the mean ascending readings versus commanded displacement over the full stroke, then express the maximum deviation of any point from that line as a percentage of full-scale output (FSO).
• Hysteresis: at each load point, take the absolute difference between the mean ascending and mean descending readings, and express the largest such difference as a percentage of FSO.

As a worked example with illustrative numbers (not measured data): suppose the sensor's full-scale output is 75.0 mm at 100 % stroke, and at the 50 % point the mean ascending reading is 37.9 mm while the mean descending reading is 37.1 mm. The hysteresis at that point is 0.8 mm, or 0.8 / 75.0 × 100 ≈ 1.07 % FSO. If the largest deviation from the fitted line across all points is 0.9 mm, the linearity error is 0.9 / 75.0 × 100 = 1.2 % FSO.

Reading the Results

The figure below shows how the ascending and descending traces separate on a healthy sensor versus a failing one, and how the two metrics are derived from the same data set.

Decision and Documentation

Apply the acceptance limits to the computed values. A sensor that passes both linearity and hysteresis limits is released for the test. A sensor that fails either limit is removed from service and replaced; do not attempt to compensate a non-linear or hysteretic sensor by adjusting channel scaling, because the error is not a constant gain and will vary with deflection level. Record the following for each sensor so the decision is traceable:

Common Failure Symptoms and Troubleshooting

• Large hysteresis with good linearity: usually mechanical friction or binding in the sensor or fixture, not an electrical fault. Check that the fixture moves only the input element and that the sensor body is not preloaded.
• Bow-shaped deviation from the fitted line: genuine non-linearity in the sensing element; replace the sensor.
• Noisy or jumping readings at a dwell point: poor connection or damaged lead; inspect the connector and cable before repeating the test.
• Consistent offset at zero: handle as a zero/offset issue separately; it does not by itself indicate non-linearity, but confirm it is within the channel's zero tolerance.

Once every rib deflection sensor on the dummy has passed this check, the channel scaling is trustworthy and the deflection values used in the subsequent injury assessment can be treated as valid measurements rather than instrument artefacts.