Energized DC at Dawn: Opening a Connector Under Load
At 7:10 a.m. on a clear January morning the ambient temperature is -6°C and plane-of-array irradiance is about 90 W/m². Your crew is at a residential 9.6 kWdc array: 24 modules in two strings of 12, each module rated Voc 41.6 V at 25°C with a Voc temperature coefficient of -0.30%/°C. The inverter has a 600 Vdc maximum input. The homeowner reports intermittent production. The inverter's AC breaker is off and its display is dark. A helper proposes to pull apart an MC4 connector at the array to "kill the string" so he can inspect a suspected loose home-run connection. He is wearing nitrile gloves and carrying a 12-year-old meter marked CAT II 600 V with a cracked probe boot. Your own meter, rated CAT III 1000 Vdc, reads 545 Vdc across the string at the roof. You also notice a nick in the DC conductor insulation where it passes a rail.
1. Explain why the string reads 545 Vdc and why the inverter being off does not make the array safe to work on.
Read the worked answer
Why 545 V. STC string Voc is 12 × 41.6 V = 499 V. The cells are cold: around -6°C ambient, with too little irradiance to warm them, so ΔT ≈ -31°C from 25°C. Applying -0.30%/°C gives about +9.3%: 499 × 1.093 ≈ 545 V, matching the measurement. A PV array reaches its highest voltage on the coldest, clearest morning - exactly when someone is most likely to be on the roof.
Why the dark inverter means nothing. A module is a current-limited source. At 90 W/m² the current is small, but Voc barely falls with irradiance, so the array is fully energized whenever there is light. The inverter is a load, not a switch. Opening its AC breaker removes the grid connection and stops conversion, but the DC source circuits stay live. No control on the system can de-energize a PV array; only physical measures can - opaque covers over the modules, working in verified darkness, or disconnecting and proving dead at the exact point of work. The 545 V reading is proof that the array is live right now.
2. State your immediate actions, including PPE and test-equipment requirements, and address the nicked conductor.
Read the worked answer
Stop the connector plan. MC4 connectors are not load-break devices. Separating them under current draws an arc that will damage the connector, can start a fire, and can arc-flash the worker. A 545 Vdc string can source 10.9 A Isc and sustain an arc that does not self-extinguish at a current zero the way AC does.
Sequence: cover both strings with opaque material (or defer to a planned covered/night work window), then test-before-touch with a CAT III 1000 Vdc meter, proving the meter on a known live source first. Retire the CAT II meter - CAT II is not rated for PV source circuits or the transient environment at the array.
PPE and tools: voltage-rated gloves with leather protectors (Class 00 is rated 750 Vdc; many sites specify Class 0), insulated tools, arc-rated clothing per the site risk assessment, and a second person.
Nicked conductor: treat it as a live exposed part. De-energize the string by covering, repair or replace the home run, and add abrasion protection at the rail. Taping over a nick is not a repair - the conductor is a shock and ground-fault path for the life of the system.
3. The helper suggests returning at 5 a.m. when it is still dark, so the array will be "dead." Evaluate that plan.
Read the worked answer
Darkness is a legitimate way to get a de-energized array, but the plan as stated is unsafe for three reasons.
- It is not verifiable by assumption. First light arrives quickly, and a 12-module string can go from a few volts to more than 500 V while the worker is still on the roof. Testing at the point of work, every time, is what makes the array safe - not the clock.
- It ignores stored energy and the rest of the system. Inverter DC-link capacitors hold charge after the array is dark; the AC side and any battery or rapid-shutdown equipment must also be locked out and proven.
- It trades an electrical hazard for a fall hazard. A 5 a.m. start on a frost-covered roof, with cold hands and poor light, is a worse risk profile than the same job done in daylight with the array covered.
A workable plan: cover the strings, verify 0 V at the point of work with a proven meter, apply lockout/tagout, and complete the repair in daylight. If night work is chosen, it needs its own lighting, fall-protection, and verification plan - not just an early start.