0.9 MΩ at Handover

The system is not ready for handover.

• Typical inverter manufacturer minimums are 1 MΩ (some allow 0.5 MΩ). 0.9 MΩ on String A is below the threshold and roughly 20× worse than String B on the same array. That is a real leakage path, not instrument noise.
• The inverter may still run because some units only fault below a lower trip point - running is not passing.

Diagnostic sequence:

1. Isolate and lock out AC and DC; short the string conductors through a resistor to discharge module capacitance before and after each test.
2. Re-test each string conductor to ground separately (positive-to-ground, negative-to-ground) to see whether the fault is on one polarity.
3. Bisect String A at the mid-module jumper - reconnect half and test. That converts a 12-module search into a 6-module one, then a 3-module one.
4. Most likely causes, in order: a conductor nicked by a rail clamp or a sharp edge on the 120 ft run, water in a connector, a pinched home-run jumper, or a damaged module junction box.

Safety: Class 0 gloves, insulated tools rated for the array voltage, no IR testing in rain or with wet connectors, and never with the inverter connected. Do not test on a wet roof.

Criterion (NEC 690.12): within 30 seconds of the initiator actuating, the controlled conductors outside the array boundary must be limited to 30 V or less.

Method:

1. With the system producing, operate the initiator at its normal location.
2. Wait 30 s (time it - do not estimate).
3. Measure DC voltage at the rapid-shutdown device output terminals, i.e., on the conductors that leave the array, using a meter rated for the maximum system voltage. Also confirm the inverter has stopped exporting.
4. Record the voltage, the elapsed time, the instrument, and the initiator location on the commissioning form.

Also show the inspector:

• The initiator is at a readily accessible location, not behind a locked door or in the attic, and is labeled as the PV rapid-shutdown device.
• The service panel and the disconnect carry the required placards and the system diagram.
• The RSD test is repeated at handover and again at any service call, because a failed RSD is invisible until someone needs it.

A passing RSD test does not clear the IR failure - both must be closed out before sign-off.

Do the arithmetic before you blame the meter.

• Array nameplate: 24 × 450 W = 10.8 kW STC.
• Irradiance correction: 980 ÷ 1000 = 0.98 → 10.6 kW.
• Temperature correction: cell temperature well above ambient at 31 °C with full sun; at ~0.35%/°C and roughly +30 °C above STC, derate ≈ 0.90 → ~9.5 kW DC.
• Inverter efficiency ~0.97 → expected ≈ 9.2 kW AC, against a 10 kW inverter rating, so it should not be clipping.

Measured 7.1 kW is roughly 23% below expectation with both strings reporting equal Voc and equal Isc. Equal voltage and equal current at the array against a depressed AC output points to a DC-side leakage or a measurement error, not a single bad module.

Sequence: verify irradiance and AC voltage with independent instruments, confirm the monitoring CT is not mis-scaled, then return to the 0.9 MΩ ground path on String A - a resistive path to ground bleeds current and can pull the operating point down without tripping a fault code.

Decision: do not sign the commissioning sheet. Equal string currents plus a depressed output plus one failing IR reading is a single coherent story, and none of it is closed out.