Hardware Test Methodology — Standard Definitions and Methods

This document is the shared rulebook for OpenVVVF hardware power tests. It exists so that a "300 A steady-state hold" run in September means exactly the same thing as one run next year on a 450 V chassis: same definitions of steady state, same calculated quantities, same evidence package, same naming. Test plans state what to run; this document states how.

Scope: power-stage characterization on the Gen7 control module with the C2 chassis assembly (dynamometer-coupled machine, bench DC supply), covering the 200 V and 450 V DC-bus classes. Methods follow common endurance-test practice and are aligned with EN 50155-style severity thinking (temperature classes, supply-voltage variation, functional performance categories); no compliance claim to EN 50155 is made here — it is methodological alignment only. Ceiling/floor variant tests (voltage, temperature, switching-frequency extremes) are reserved for a future revision of this document and are out of scope for now.

1. Standard bench configuration

2. Standard definitions

• Steady state (thermal): |dT/dt| ≤ 0.1 °C over any 5-minute window on the baseplate NTC under constant operating point. Confirmation requires the window to complete while the operating point is still held.
• Plateau temperature: mean of the baseplate NTC over the final 10 minutes of a hold that has reached steady state. Always quoted with the ambient and as ΔT over ambient.
• Hold: operating point held within ±5 % of command (or ±10 A, whichever is larger) for the stated window. Average and min/max of the held quantity must be reported.
• Staircase: current steps of 25 A after an orientation jog of −10…−25 A (confirms regen/motoring sign before full power); dwell per step ≥ 5 s and ≥ 2× current-loop settling time.
• Stop criteria (in priority order): any protection trip; machine thermal limit (no machine thermistor → imager-monitored, case temperature recorded at stop); operator stop. An operator stop on a machine limit with the DUT stable is a valid, reportable outcome — record the reason and the DUT state explicitly.
• Regen / motoring signs: negative DC-link power = power leaving the link (motoring draw); regen holds are identified by machine delivering P. Quote directions in words, not just signs.

3. Standard calculated quantities

All calculated from logged FOC quantities with the peak-amplitude convention:

• P = (3/2)·(vd·id + vq·iq) — machine-terminal real power (W); negative = machine generating.
• Q = (3/2)·(vq·id − vd·iq) — machine-terminal reactive power (var). Quote as magnitude (machine magnetizing demand); the sign depends on the logged dq frame orientation and is not comparable across runs.
• |S| = √(P² + Q²); PF = |P| / |S|, quoted over the hold window.
• Inverter efficiency: regen η = mean(P_DC-link) ÷ |mean(P_machine)|; motoring η = |mean(P_machine)| ÷ |mean(P_DC-link)|.

Reconciliation requirement: machine-terminal P and DC-link P must imply a thermally plausible inverter loss. If the implied loss is impossible for the observed temperature rise (e.g. ~3 kW "dissipated" while the baseplate climbs 44 K in 10 min), the session's dq-voltage logging does not reconcile — do not publish P/Q/PF/η for that session; add a data-quality observation with the firmware build (see the 100 A and 200 A Sept-22 reports). Efficiency quoted within measurement uncertainty reads as ≈95–100 %; values at or above 100 % mean the loss is below telemetry resolution, and must be captioned as such.

4. Standard test archetypes

A. Steady-state thermal hold — staircase to target current; hold until steady state per §2 (or the applicable stop criterion); TI250 photo pass once plateau is confirmed; record plateau, ΔT over ambient, all §3 quantities; log ≥ 5 min of cooldown.

B. Burst / capability run — staircase to a short-duration target; hold for a defined time cap or until a protection trip; for trip-bound runs, record the trip flag word, trip temperature, and time-from-command. Machine thermal limit must be known before starting.

C. Endurance — hold at target current ≥ 20 minutes; purpose is sustained-conduction confidence, not plateau acquisition; peak temperature and any drift reported.

D. Protection validation — exercise one protection chain deliberately (e.g. drive baseplate into the OT threshold); record threshold value, fault flags, gate behavior, and post-event cooldown. The DUT must trip; a non-trip is a fail.

E. Electrical quality — oscilloscope capture of phase currents during a hold (correctly rated probes), bus regulation quality, and the §3 calculated block.

Component stress by archetype (evidence transfer)

A result transfers across component revisions, capacitance values, or voltage classes only for components the test does not electrically stress. Every report must state which components were stressed; the transfer decision then follows this table:

Where capacitors are not stressed, one sentence in the report suffices (e.g. "DC-link capacitors not stressed: current-limited bus at ≤60 V, negligible ripple"). Where they are stressed — short-circuit tests and any new voltage class — capacitance and voltage must be recorded and the energy or ripple argument made explicitly.

5. Required evidence package (every power test)

1. Test Report under Docs/Safety-and-Compliance/Testing/Hardware/ with sections: Test setup, Test conditions (table), Procedure, Electrical results, Thermal results, Telemetry overview, Observations, Conclusion, Artifacts.
2. Conditions table including: hardware + firmware hash, machine + speed, bus, current command + window, average held current (avg/min/max), DC-link power, §3 quantities (P, Q, |S|, PF, η where reconciling), ambient (recorded or inferred), thermal results.
3. Telemetry overview plot, standard 5-panel layout: Iq; DC-link power; machine-terminal P (kW) and Q (kvar); baseplate NTCs with ambient reference; bus voltage.
4. Full-rate session log + 1 Hz decimated log; decimated log linked in the Telemetry Viewer.
5. Photos per the archetype (bench, scope, thermal passes with crosshair/max readings, machine thermal image if limit-governing).
6. Naming and doc_id per the Testing naming convention in Docs/Agents.md.

6. Variant parameter matrix

Parameters that change between chassis/bus-class executions. Fill in per campaign; tests that consume this table reference it rather than restating values.

7. EN 50155 alignment notes

• Functional performance: these tests target performance category "characteristic not exceeding specified tolerance / auto-recovery" for protection events — a trip must be clean, flagged, and survivable, and the DUT must be reusable after cooldown without degradation.
• Temperature practice: EN 50155-style temperature-class thinking applies to future ceiling/floor campaigns (ambient extremes, −40 °C cold start, elevated ambient derating); those tests are reserved and out of scope for this revision.
• Supply variation: future ceiling/floor runs cover bus over/undervoltage behavior; until then, all quoted bus values are nominal-class.
• Traceability: the reconciliation requirement in §3 and the firmware-hash logging in §1 are the provenance hooks that make these reports usable as safety-case evidence.