Skip to content
SLSolar Ledger
Installation

DC arc faults and rapid shutdown: physics, detection, and safety

How DC arc faults form, why AFCI detection differs from AC, and how module-level rapid shutdown architectures protect firefighters and equipment.

By Solar Ledger Editorial13 min read
Last reviewed

A DC arc fault in a PV array can burn at temperatures exceeding 3,000 degC, ignite roofing materials, and - because PV strings keep generating current as long as light hits the modules - it can persist even after the AC breaker trips. Rapid shutdown and arc-fault circuit interrupters (AFCI) exist because turning off the inverter does not turn off the array.

This article is for designers, installers, and inspectors who need to understand not just that AFCI and rapid shutdown are code requirements in many jurisdictions, but why the physics of DC arcing makes them necessary, how the detection algorithms actually work, and where module-level shutdown genuinely improves firefighter safety versus where it is theater.

We cover arc fault physics, series versus parallel arcs, AFCI detection principles and their limitations, module-level rapid shutdown architectures, and the practical trade-offs between compliance, cost, and reliability.

Why DC arcs are dangerous and hard to detect

An electric arc is a sustained plasma discharge across a gap in a conductor, maintaining current flow via ionized gas rather than solid metal contact. AC arcs benefit from a built-in extinguishing mechanism: the current crosses zero twice per cycle, and if the gap is wide enough relative to the voltage, the arc self-extinguishes at each zero crossing. DC has no zero crossing. Once a DC arc establishes, it can sustain itself indefinitely as long as the source (the PV string) can supply the required current and voltage, which for a string of 20-30 modules is easily several hundred volts.

Series versus parallel arcs

A series arc occurs in the current path itself - a loose connector, a corroded MC4 termination, a nicked cable that has partially severed - and reduces current flow while creating a high-energy discharge at the gap. A parallel arc occurs between two conductors at different potentials, such as damaged insulation allowing a positive and negative conductor to arc across a gap; parallel arcs are generally more energetic because they are not current-limited by the string's own resistance in the same way.

Series arc detection is the harder engineering problem and is the primary target of AFCI algorithms, since parallel arcs on the DC side are less common (typically requiring simultaneous insulation failure of both polarities in close proximity) but series arc precursors - degrading connectors, rodent damage, thermal cycling fatigue - are widespread across aging PV fleets.

How AFCI detection works

Listed AFCI devices, per requirements aligned with UL 1699B, monitor the current waveform for the high-frequency noise signature characteristic of arcing - typically broadband noise in the tens of kHz to low MHz range superimposed on the DC current, distinguishable from the switching noise of the inverter's own power electronics through pattern recognition and adaptive filtering.

The detection challenge is discriminating a genuine arc fault from other noise sources: inverter switching transients, module-level power electronics (optimizers, microinverters) switching noise, and normal connector micro-arcing that does not escalate. False positive nuisance tripping was a significant early complaint with first-generation AFCI inverters, particularly in systems with power optimizers whose own switching signatures could resemble arc noise; algorithm maturity has improved this substantially, but installers should still expect occasional nuisance trips that require investigation rather than blind reset.

Detection response time

Listed devices are required to interrupt the circuit within a bounded time after confirmed arc detection - commonly on the order of 2.5 seconds channel-clearing time for the fastest classified arc events. This matters because arc energy accumulates with duration; a fault cleared in under a second causes negligible thermal damage, while the same arc left for tens of seconds can carbonize surrounding material and ignite it.

Rapid shutdown: the module-level safety case

Rapid shutdown requirements exist primarily for firefighter safety: a firefighter cutting into a roof or approaching a rooftop array during a fire needs assurance that conductors more than a short distance from the array (commonly within 1 m of the array boundary in many rapid shutdown schemes) are de-energized to a low voltage within a bounded time after initiation - commonly 30 seconds to reach 30 V or less in schemes following this pattern, though exact figures depend on the applicable code.

Simply opening the inverter's AC-side breaker does not achieve this: the DC string between modules and the inverter remains fully energized at open-circuit voltage as long as sunlight reaches the panels. Rapid shutdown solves this by requiring a controlled means of dropping the DC-side voltage at the module or string level, initiated automatically on loss of AC power or manually via a clearly labeled switch.

Architectures

Two dominant architectures exist. Module-level power electronics (MLPE) - power optimizers or microinverters - inherently limit output to a safe voltage per module when they lose their communication or control signal, satisfying rapid shutdown without additional hardware. Add-on rapid shutdown transmitter/receiver systems retrofit a shutdown signal onto conventional string inverter systems, using a transmitter at the inverter/combiner and receivers at or near each module that open the circuit on loss of signal.

MLPE-based rapid shutdown tends to be more reliable in the field because the safety function is inherent to hardware already required for other purposes (per-module MPPT, monitoring), whereas add-on transmitter/receiver systems introduce an additional communication link that can itself fail - a documented cause of rapid shutdown non-compliance found during inspections and audits of installed systems.

Table: rapid shutdown response comparison

ArchitectureVoltage reduction mechanismTypical time to safe voltageAdded component count
MicroinverterAC-side isolation per module, no DC stringImmediate (no DC string exists)None beyond microinverter itself
Power optimizer + string inverterOptimizer drops output to ~1 V per module on signal lossWithin 30 s per typical schemeOptimizer per module
Add-on transmitter/receiver retrofitReceiver opens circuit at module on signal lossWithin 30 s per typical schemeTransmitter + receiver per module
Conventional string inverter, no rapid shutdownNone (not compliant in jurisdictions requiring it)Not applicableNone

Worked example: arc energy dissipation

A series arc forms across a 3 mm gap at a degraded connector carrying 8 A at a local voltage drop of 30 V across the arc. Power dissipated at the arc equals 8 A x 30 V = 240 W, concentrated in a few cubic millimeters of ionized gas and adjacent material. If the AFCI takes 2.5 seconds to detect and clear the fault, total energy delivered to that point is 240 W x 2.5 s = 600 J - comparable to the energy of a small soldering iron held in place for the same duration, sufficient to char adjacent polymer insulation or ignite dry debris. This illustrates why detection speed, not just detection occurring at all, is a meaningful safety parameter.

Interaction with insulation monitoring

Arc fault detection and insulation resistance monitoring address related but distinct failure modes: an insulation fault indicates an unintended path to ground or between conductors, while an arc fault indicates an active discharge event, which may or may not stem from an insulation breakdown. Systems that report both parameters to a monitoring platform give O&M teams a much clearer picture of a developing connector or cable fault than either signal alone, since a rising insulation resistance trend combined with an isolated arc trip is a strong indicator of a specific failing termination rather than a systemic wiring issue.

Common mistakes

  • Assuming any inverter with an "AC disconnect" satisfies rapid shutdown requirements without verifying DC-side voltage actually drops within the required time and distance.
  • Ignoring nuisance AFCI trips as a "known issue" and disabling the function rather than investigating a genuine intermittent connection.
  • Mixing rapid shutdown transmitter/receiver hardware from different vendors or firmware revisions, which can break the communication protocol silently.
  • Failing to label rapid shutdown initiation switches and array boundaries per the applicable code, leaving first responders without clear guidance.
  • Retrofitting rapid shutdown receivers without verifying RF or powerline communication signal integrity across the full array, especially on large or metal-roofed installations where signal attenuation is significant.
  • Treating rapid shutdown compliance as a one-time commissioning check rather than verifying function after any module or optimizer replacement.

Frequently asked questions

Does rapid shutdown eliminate all DC voltage on the roof?

It reduces conductor voltage outside a defined boundary and, within that boundary, drops each module's output to a low voltage (commonly around 1 V per module in optimizer-based systems) within the required time; it does not eliminate DC entirely inside the array boundary but does within it once initiated.

Can microinverter systems skip AFCI entirely?

Many microinverter systems have very short DC conductor runs (within the module or a short whip), which reduces series arc risk substantially, but applicable code and listing requirements should be checked directly since requirements vary by jurisdiction and system configuration.

Why do AFCI devices sometimes trip with no apparent fault?

Nuisance trips can result from switching noise generated by module-level power electronics, loose but non-arcing connections, or electromagnetic interference from nearby equipment; each trip should be logged and investigated rather than dismissed, since repeated nuisance trips can mask a genuine developing fault.

How does arc fault risk change as a PV array ages?

Connector and cable insulation degrade with UV exposure, thermal cycling, and rodent activity, increasing series arc precursor risk over a 20-25 year system life; periodic thermal imaging and connector inspection during O&M visits catches many precursors before they arc.

Is rapid shutdown required for ground-mounted arrays?

Requirements vary by jurisdiction and are generally focused on rooftop arrays because of firefighter access concerns; ground-mount systems are often exempt or subject to different rules, so the applicable local code should be consulted directly.

Standards and references

UL 1699B: Outline of Investigation for Photovoltaic (PV) DC Arc-Fault Circuit Protection. IEC 63027 and related work on PV arc fault detection. NFPA 70 (National Electrical Code) rapid shutdown provisions, referenced for the general architecture pattern even outside jurisdictions using NFPA 70 directly. Sources: NREL, "Photovoltaic System Reliability" and arc-fault detection studies; Fraunhofer ISE, module-level electronics reliability reports; IEA PVPS Task 13, "Review of Failures of Photovoltaic Modules and Systems."

See also /blog/mppt-explained, /glossary, /calculators/system-sizing, /calculators/inverter-sizing, and /blog/cable-sizing-and-voltage-drop for related sizing considerations.

arc-faultrapid-shutdownafcisafetyinstallation

Standards referenced

  • IEC 63027

Standards are cited for orientation. Always work from the current published edition and the national annex that applies in your jurisdiction.

Sources and further reading

  1. [1]IEC 60364-7-712 — Low-voltage installations: photovoltaic power systems — IEC
  2. [2]IEC 62446-1 — Grid connected PV systems: documentation, commissioning tests and inspection — IEC
  3. [3]Best Practices for Operation and Maintenance of Photovoltaic and Energy Storage Systems — NREL
  4. [4]Task 13: Performance, Operation and Reliability of PV Systems — IEA PVPS

Update history

  • First published.

  • Technical review: worked examples re-derived, standards references checked against current editions, sources verified.

The weekly brief

One email a week: new articles, tariff changes and hardware worth knowing about. No spam.