Rapid Shutdown That Passes Only While the Grid Is Present

The numeric result (18 V in 20 s) meets the threshold, but the test was run in only one operating mode, so it does not demonstrate that the function works.

What the rule asks for:

• Upon initiation of rapid shutdown, controlled conductors within the array boundary must be reduced to 30 V or less within 30 seconds (the boundary is 1 ft from the array in current editions; earlier editions used 3 ft and longer conductor allowances).
• Conductors outside the array boundary must also be reduced to 30 V or less within 30 seconds, with limited length permitted inside the building.
• The initiation device must be labeled, readily accessible, and the placard must show the shutdown equipment and device location.

Why backup mode matters:

• Fire crews most often arrive at a building that is already de-energized. That is exactly the condition the shutdown function is meant to survive.
• In island mode the inverter forms its own grid, so the module-level transmitters keep receiving a keep-alive signal and the array is held at operating voltage. Opening the RSD switch therefore does not reduce array voltage.
• The likely root cause is where the shutdown transmitter or initiator is powered: if it sits on the backed-up panel, it never loses power and never initiates on loss of utility.

A shutdown function that works only while the utility is present does not demonstrate the required performance and must be retested in every mode.

Test and document, do not argue.

On site:

• Re-create the condition: battery in backup/island mode, RSD switch open, then measure DC voltage at the array boundary, at the first junction outside the boundary, and at the inverter DC terminals. Repeat in grid-tied mode for comparison.
• Trace the power source for the shutdown transmitter/initiator to its circuit breaker. Note whether that breaker is in the backed-up panel.
• Time the decay from initiation to the 30-second limit, using a stopwatch against the meter.
• Photograph the meter display with the reading, the switch position, the battery/inverter status screen, and the circuit directory showing which panel feeds the transmitter.

Record:

• Inverter operating mode, battery state of charge, grid status, ambient temperature, meter model, and calibration date.
• Every measurement with location and time stamp, not a single summary number.
• The exact wording on the placard versus the installed equipment.

A finding that says "RSD failed" is easy to dispute. A finding that says "with the utility disconnected and the battery at 71% SOC carrying the critical loads panel, 240 V DC was measured at the array boundary 4 minutes and 12 seconds after the RSD switch was opened" is not.

Classify it as a safety nonconformance requiring correction, not as an informational note. The as-installed system permits the array to remain energized with the shutdown device in the off position, and the placard tells responders the opposite.

Acceptable corrections (designer's choice, but it must be engineered):

• Move the shutdown transmitter or initiator supply to a non-backed-up circuit so loss of the utility removes the keep-alive signal.
• Add a listed shunt-trip device or contactor that interrupts the array on initiation of shutdown regardless of inverter mode.
• Use equipment whose listing confirms array shutdown independent of AC presence, with documentation from the manufacturer rather than a verbal assurance.

Also correct:

• The placard and diagram so they reflect the battery and the actual shutdown path.
• Any labeling that implies opening the RSD switch de-energizes the array in all conditions.

Before sign-off, require a repeat test in grid-tied and island modes with the results recorded, then a re-inspection that includes the 30-second timing and the circuit feeding the shutdown transmitter. Note in the report that the original commissioning test was valid only for the condition under which it was performed.