One String Down 30% and a Ground Fault After Rain
A 15 kWdc array has three strings of 12 modules (400 W, Imp 9.5 A, Voc 42.5 V) on a string inverter with two MPPTs. Strings A and B share MPPT1; string C is alone on MPPT2. After a spring hailstorm, the owner reports production down about 28%. At 1:00 p.m., with irradiance 900 W/m² and cell temperature 32°C, the technician measures at the DC disconnect: string A 8.9 A, string B 5.3 A, string C 9.0 A; open-circuit voltages are 508, 507, and 509 V. No fuses are blown and there is no visible damage from the ground. The inverter log shows a "DC ground fault" alarm on three rainy mornings; it clears by mid-afternoon each time. An insulation resistance test at 1000 V on string B, with both ends disconnected, returns 0.6 MΩ; strings A and C read above 20 MΩ. An IR scan from the roof edge shows a hot spot on module B7, which has a hairline crack in its glass.
1. Interpret the current and voltage measurements. What failure modes do the voltages rule out, and what is the likely cause in string B?
Read the worked answer
Voltage is continuity; current is resistance. All three strings read 507-509 V. Expected STC Voc is 12 × 42.5 = 510 V, and at 32°C cell temperature it should sag slightly, so ~508 V is normal. That rules out two candidates:
- No open string or open substring - a broken string would read 0 A, not 5.3 A.
- No shorted or conducting bypass diode - a bypassed module drops Voc by about 42 V (roughly one-twelfth), and a whole substring would drop it much further. All three voltages are essentially equal.
Therefore string B has a current-limiting series resistance: a poor crimp or MC4 contact, a corroded connector, or a cracked cell with high internal resistance. The ratio 5.3 / 8.9 ≈ 0.60 is about a 40% current loss on that string. The IR hot spot on cracked module B7 supports a localized high-resistance path, not a failed diode.
Note the array-level number. One string at 60% should cost roughly 13% of array output, but the owner reports 28%. That gap is consistent with the rainy-morning outages and with MPPT1 being pulled off its true maximum power point by the mismatch between A and B - both worth checking after the repair.
2. Interpret the 0.6 MΩ reading and state your procedure before string B is re-energized.
Read the worked answer
The reading fails. For a system in this voltage class, insulation resistance should be at least 1 MΩ, and most commissioning standards and manufacturers expect far higher (a common criterion is a resistance-area product such as 40 MΩ·m², or the module datasheet value). 0.6 MΩ means current is leaking to ground through a damaged insulation path.
Why it appears only when wet: the fault disappears as the array dries, so it is a moisture-dependent path - a cracked module, a damaged backsheet, a pinched conductor at a rail or clamp, or a junction-box gasket - not a bolted fault. The wet-morning alarm and the dry-afternoon clearing are the signature.
Procedure: isolate and lock out string B; do not re-energize it. Confirm the test method (both ends disconnected, inverter isolated, 1000 V test voltage, stable reading, temperature noted). Dry and inspect B7 and the connectors on both sides; look for water tracking at rail clamps and for abrasion where the conductor crosses the roof edge. Replace the cracked module and any damaged connector, then repeat insulation resistance, polarity, and continuity tests and record the values before reconnecting. If the reading stays low, sectionalize the string to isolate the affected module or home run.
3. The site supervisor suggests simply clearing the ground-fault alarm each morning since it resets on its own. Respond, and list what else must be verified before the system is returned to service.
Read the worked answer
Clearing the alarm is not a repair. The fault has not healed; it dried out. Resetting leaves an energized array with a leakage path to the grounded rack - a shock and fire hazard, and a condition that can raise voltage on surfaces assumed to be at ground potential. It also erases the evidence for the next technician and masks the defect through the dry season until the first heavy rain.
Before return to service:
- Repair or replace cracked module B7 and any damaged connector, with photos and serial numbers recorded.
- Re-run insulation resistance, polarity, and open-circuit voltage on string B and compare with A and C; record all three.
- Torque-check the connectors and terminations in the affected home run, and verify no conductor is pinched at a clamp or abrading on the roof surface.
- Confirm string B's current matches A and C at similar irradiance, and confirm MPPT1 is tracking properly with both strings connected.
- Reconcile production against a modeled or neighboring baseline to confirm the 28% loss is fully explained by the string fault, the rainy-morning outages, and the MPPT mismatch - otherwise the system will be back in the next rainy season with the same alarm.