EOL testing for EV 3-in-1 drive units: cycle time, judgement and traceability

EOL is not a repeat of the R&D test

The purpose of an EOL station is to catch assembly defects and outliers, not to re-characterise the product. So the test list should target properties that the assembly process can damage: phase sequence, resolver zero offset, gear mesh noise, seal leakage, bearing assembly quality and insulation withstand.

Budgeting the cycle

In a 60-second cycle, a typical split is 10 s load/clamp, 8 s insulation and electrical safety, 6 s resolver zero calibration, 20 s bidirectional performance and spot-check efficiency, 12 s NVH sweep, 4 s upload and judgement. What usually sets the cycle is not the measurements but ramp times and fixture motion.

The most effective way to shorten the cycle is normally a more dynamic dynamometer and a better speed profile, not deleting test items.

Setting NVH thresholds

Order analysis is the core of e-drive NVH judgement. Collect order-amplitude distributions from a few hundred known-good units at the start of production, set initial limits statistically (mean plus a number of standard deviations), then refine them against rework and field-complaint data. Fixed thresholds carried over from a previous programme tend to fail after a supplier or gear-batch change.

Traceability

Raw waveforms, judgement results, equipment status and calibration validity should be linked to each serial number and archived long-term. When a batch issue appears, traceable raw data is the only fast route to root cause. LANDI's LDEV-X in Production mode together with LD-LMS covers the whole path from test execution to archiving and retrieval.