Staircase I-V Curve at Commissioning Sign-Off
During commissioning of a 486 kWDC rooftop system, you are reviewing field test results before the utility witness test. The array has three 125 kW string inverters, each with 12 strings of 13 bifacial 450 W modules at 25° tilt and 180° azimuth. Ambient temperature is 34°C and plane-of-array irradiance is 890 W/m². An I-V curve tracer on Inverter 3, String 7 shows Voc within 1% of expected, Isc 8.4% below the sibling strings, and a pronounced staircase step in the curve near the knee. Insulation resistance measured 2.8 MΩ at 1000 V test voltage, above the 1 MΩ pass threshold. Visual inspection from the roof edge shows no obvious shading; a parapet wall sits 4 m south of the array, but the sun is high. The EPC wants to sign off and return the following week. A torque audit found three loose MC4 connectors on other strings.
1. How do you interpret the staircase I-V curve and the low Isc, and what should happen to sign-off?
Read the worked answer
Do not sign off. Voc within 1% and Isc 8.4% low with a step near the knee points to a current-limiting condition inside the string, not uniform soiling. One module or substring is producing less current, and its bypass diode begins conducting as the tracer sweeps voltage, producing the staircase.
Possible causes include a cracked cell, a failed bypass diode, a mismatched or wrong-bin module, a high-resistance connector, or localized shading. At 890 W/m² with a high sun angle, the parapet is unlikely to shade String 7, but confirm with a shade/time check. Compare String 7's Isc and operating current with sibling strings, thermally scan each module, check bypass diodes, inspect junction boxes and connectors, and isolate modules electrically to locate the weak one.
The three loose connectors on other strings make the entire connector population suspect, so expand the torque audit rather than treating those as isolated findings.
2. Does the 2.8 MΩ insulation resistance result clear the string?
Read the worked answer
No. 2.8 MΩ is above the common 1 MΩ minimum, but it is low for a dry, newly installed string; healthy strings often read in the tens or hundreds of megohms.
A pass at 1000 V proves only that no low-resistance path exists at that moment. It does not rule out moisture-dependent faults, damaged cable insulation, contaminated connectors, or defects that appear only under wet conditions. Re-test with the same instrument and voltage after confirming test leads and guard connections, compare against sibling strings, and investigate anything below the population median. Record actual values and the test conditions, not just pass/fail.
3. What acceptance documentation and retest sequence should be required before energization?
Read the worked answer
Hold String 7 as a punch item. The EPC should locate and repair or replace the affected module or connector, then repeat the I-V curve, insulation resistance test, and visual inspection. Require a torque record for every re-terminated connection, not only the three originally found.
Acceptance should be against IEC 62446-1 test forms and the contract's stated tolerances - for example, every string within an agreed percentage of the expected curve at measured irradiance and temperature. Schedule the utility witness test only after the punch list is closed and retests pass. A single anomalous string is enough to delay acceptance when the specification requires all strings to be within tolerance.