A transformerless inverter DC ground fault rarely starts as a dramatic failure. It often appears as an isolation alarm at startup, after rain, or during a damp morning. The tempting response is to clear the event and try again. That can hide the evidence and expose people to energized metal.
Direct answer. Treat a transformerless inverter DC ground fault as an insulation and shock-hazard investigation. Preserve the exact alarm and identify the inverter model and grounding method. Isolate the system under the manufacturer’s procedure. Test only within approved boundaries, repair the defect, and reconcile the drawings before restart. Never use a universal resistance threshold or generic megger sequence.
This guide is for PV designers, commissioning engineers, and qualified service teams. It explains the engineering records and decision points. It does not replace the product manual, lockout procedure, adopted code, or qualified electrical work.
The ground-fault glossary entry owns the short definition. This article owns transformerless topology, isolation monitoring, safe fault triage, drawing evidence, and repair closeout.
What is a transformerless inverter DC ground fault?
A DC ground fault is an unintended conductive path between an energized PV circuit and earth or bonded metal. In a transformerless system, the inverter has no internal isolation transformer separating the PV DC side from its AC conversion stage.
Transformerless does not mean ungrounded equipment. Module frames, racking, enclosures, raceways, and other exposed metal still need the applicable bonding and equipment-grounding path. The difference concerns the current-carrying DC circuit and inverter topology.
The fault path can form through damaged insulation, moisture, a pinched cable, a failed connector, a module junction box, or contaminated equipment. The resulting current can be too small for ordinary overcurrent protection. The inverter therefore uses dedicated monitoring and protection functions.
The IEC 62109-2:2011 scope covers particular safety requirements for PV inverters. It applies to grid-interactive, stand-alone, and multimode products. The standard works with IEC 62109-1 rather than replacing it.
Why does transformerless topology change fault behavior?
Without galvanic isolation, the PV array’s voltage to ground depends on the inverter topology, operating state, parasitic capacitance, and control method. Those relationships can change between startup, power conversion, shutdown, and fault conditions.
That is why a single conductor-to-ground voltage is not a universal pass or fail test. A reading must be interpreted through the exact manufacturer’s procedure. The product may measure insulation resistance before connection, monitor residual current during operation, or use both methods.
Use four separate terms correctly:
| Term | What it describes | What it does not prove |
|---|---|---|
| System grounding | The intentional relationship between a current-carrying conductor and ground | That exposed metal is bonded correctly |
| Equipment grounding | The fault-current path for exposed conductive parts | That DC insulation is healthy |
| Functional grounding | A connection used for a stated equipment or system function | Permission to bond a conductor arbitrarily |
| Isolation monitoring | Supervision of insulation between active DC circuits and ground | The physical location of a defect |
The grounding versus bonding guide explains the drawing distinction. Do not use floating, grounded, and bonded as interchangeable words.
What does the inverter actually monitor?
A transformerless inverter can use more than one ground-fault supervision method. The exact functions, thresholds, timing, and response belong to the listed product and its current manual.
Common functions include:
- Insulation-resistance monitoring. The inverter evaluates resistance between PV conductors and ground before or during connection.
- Residual-current monitoring. The inverter watches for current imbalance that indicates an unintended path outside the normal circuit.
- Ground-fault indication. The product records an event, changes state, or prevents operation when its criteria are met.
- Protective interruption. The equipment opens or inhibits the applicable circuit under its evaluated design.
These functions should not be collapsed into one label on a drawing. Ground-fault protection, arc-fault protection, rapid shutdown, overcurrent protection, and surge protection address different events.
The current UL 1741 scope covers inverters, converters, controllers, and interconnection system equipment. It also states that covered products are intended for installation under NFPA 70. A listing establishes the evaluated product scope, not a site-specific diagnosis.
The UL inverter compliance guide explains model-level evidence. Save the exact certification and manual for the installed model and suffix.
Does every isolation alarm prove a ground fault?
No. An isolation or ground-fault event is a protective signal that requires investigation. It does not identify the damaged component by itself.
SMA’s current Sunny Tripower X manual says specific events can indicate defective or insufficient insulation from the PV system to ground. It uses the phrase may be a ground fault. The same manual directs the work to a qualified person.
Possible categories include:
- A real conductor-to-frame or conductor-to-ground insulation failure.
- Moisture lowering insulation resistance in a connector, box, cable, or module.
- A damaged surge protective device or DC switch component.
- Contamination or conductive debris inside equipment.
- A wiring error introduced during installation or later service.
- A module population or product combination outside the stated equipment limits.
- A measurement, firmware, or inverter issue requiring manufacturer support.
Do not replace diagnosis with a label. Record the event code, timestamp, inverter state, affected input, weather, recent work, and restart history. That evidence narrows the investigation without energizing anything.
Why do wet-weather faults disappear later?
Moisture can create a temporary leakage path through damaged insulation or contaminated surfaces. Sun and heat can dry the path, raising measured insulation resistance. The inverter may then start normally.
That recovery does not prove the defect is gone. It proves the condition changed.
Create an environmental event log:
| Field | Record |
|---|---|
| Alarm time | Exact timestamp and event code |
| Weather | Rain, fog, dew, washdown, humidity, or dry conditions |
| Array state | Pre-start, producing, shutting down, or off |
| Affected equipment | Inverter, maximum power point tracker, input, or string if identified |
| Recent work | Connector, module, cable, roof, pest-control, or cleaning activity |
| Recovery | Manual reset, automatic restart, drying period, or no recovery |
Intermittent faults often need inspection under the conditions that trigger them. That work still requires an approved safety plan. Do not create wet conditions around energized PV equipment to reproduce an event.
The rooftop cable-routing guide covers abrasion, support, routing, and water-entry controls that prevent many insulation failures.
What is the safe first response to a DC ground-fault alarm?
Stop, preserve evidence, and apply the product-specific shutdown procedure. Do not touch module frames, racking, connectors, or DC equipment until a qualified person establishes the safe work condition.
SMA’s Sunny Tripower X ground-fault procedure warns that system parts may remain live during a fault. It directs qualified personnel to disconnect voltage sources and prevent reconnection before work.
The same procedure warns against touching array substructure or frames during the unsafe condition. It also prohibits reconnecting strings that still have ground faults.
Use this first-response record:
- Capture the event code, time, input, operating state, and portal data.
- Stop remote reset attempts and informal troubleshooting.
- Identify the exact inverter model, manual revision, and installed topology.
- Establish the approved shutdown, isolation, and lockout boundary.
- Restrict access to the affected array and equipment.
- Assign the investigation to a qualified person.
- Preserve drawings, prior test values, and recent change records.
Do not improvise. PV conductors can remain energized whenever modules receive light. Opening an AC breaker does not establish a safe DC work condition.
How should a qualified team diagnose the fault?
Use a bounded sequence that moves from evidence to smaller test sections. The exact electrical measurements must come from the product, module, and test-equipment instructions.
Stage 1: Lock the fault context
Create one event record before anyone clears the alarm. Include inverter model, serial number, firmware, event code, timestamp, DC input, weather, array state, and recent work.
Then freeze the current drawings. A field team needs the string map, one-line, module layout, cable routes, combiner records, and installed equipment schedule. Mark any known difference between drawings and the field.
Stage 2: Confirm the architecture
Identify the array grounding method, equipment-grounding path, inverter topology, maximum system voltage, input arrangement, optimizers, storage, and surge protection. Do not import a test from another model.
The string inverter glossary covers the broad architecture. The product manual decides the service procedure.
Stage 3: Establish a safe work boundary
Follow the manufacturer’s shutdown and wait requirements. Apply site lockout and verify absence of voltage where the procedure requires it. Use instruments rated for the available DC voltage and environment.
SMA publishes different instrument-voltage statements for different products. Its Sunny Boy manual names one rating, while the Sunny Tripower X manual names another. That alone defeats a universal tool rule.
Stage 4: Divide the system at approved points
Separate the investigation into documented boundaries such as inverter input, combiner output, string, homerun, or module segment. Only use connection and disconnection steps allowed by the equipment instructions.
The purpose is controlled isolation, not repeated trial energization. Record each boundary, condition, result, and technician.
Stage 5: Test under the exact procedure
Use the manufacturer’s stated measurement method and limits. Some manuals describe conductor-to-ground voltage checks. Others use insulation-resistance testing with approved isolation and shorting equipment.
The SMA Sunny Tripower X procedure says the insulation test voltage should approach, but not exceed, the module’s maximum system voltage. It also requires suitable disconnection and shorting equipment. If that equipment is unavailable, SMA says not to perform the insulation measurement.
Do not connect an insulation tester through an inverter, optimizer, battery, surge protector, monitoring input, or other electronics unless the manufacturer permits it. Test voltage can damage connected equipment.
Stage 6: Inspect the likely physical points
Electrical localization narrows the area. It does not replace physical inspection.
Check the isolated section for:
- Pinched or abraded DC cable.
- Unsupported cable touching sharp metal or roofing.
- Incompatible, loose, contaminated, or poorly assembled connectors.
- Water entry in junction boxes, combiners, switches, or raceways.
- Cracked module backsheets or damaged junction-box leads.
- Pest damage and conductive debris.
- Failed surge protective devices.
- Cable damage at expansion joints and roof transitions.
- Work performed after the last accepted test.
Do not unplug a loaded DC connector. Follow the switching and de-energization sequence for the installed equipment.
Stage 7: Repair, retest, and close out
Replace or repair the defect under the applicable instructions. Repeat the approved test at the repaired section and required upstream boundaries. Confirm the protection event clears without bypassing it.
Update the string map, one-line, equipment schedule, photos, test record, and as-built package. The solar as-built drawing guide covers that final record.
How should insulation-resistance results be interpreted?
Interpret resistance against the designed array and the exact manufacturer method. A single value without string length, module type, temperature, humidity, test voltage, and equipment boundary has little diagnostic value.
Build a result record with these fields:
| Field | Why it matters |
|---|---|
| String identifier | Connects the measurement to the layout and inverter input |
| Module quantity and model | Establishes the tested insulation population |
| Connected equipment | Shows whether electronics or surge devices affected the boundary |
| Weather and module condition | Explains moisture and temperature differences |
| Test voltage and instrument | Makes the result reproducible and checks rating limits |
| Positive and negative results | Preserves polarity-specific evidence when the method uses it |
| Expected basis | Names the manufacturer data or engineered comparison |
| Technician and time | Establishes who performed the controlled test |
Compare similar strings under similar conditions. A marked difference can identify the affected branch even when a generic threshold would mislead. The comparison still does not replace the product’s acceptance rule.
SMA’s current Tripower manual says the expected value should use exact module data when available. It then compares each string result with that expectation. The manual treats a significant deviation as fault evidence within its product procedure.
Do not normalize a poor result by averaging it with healthy strings. Do not combine measurements from different weather, test voltages, or connection states without marking the difference.
An upward result after drying is useful evidence. It is not an automatic release. Inspect and repair the water path, damaged insulation, or contaminated component that created the change.
What commissioning baseline should be preserved?
A commissioning baseline makes later fault triage faster and safer. It shows how the healthy installed system behaved before weathering, construction damage, or service work changed it.
Preserve these baseline records when the applicable procedure allows them:
- Final one-line, string map, and module layout.
- Exact inverter, module, optimizer, battery, and surge-device models.
- DC input allocation and string polarity checks.
- Manufacturer-approved insulation or isolation results.
- Equipment-grounding and bonding verification.
- Connector assembly and cable-routing inspection records.
- Alarm-free startup and current firmware record.
- Photographs of combiner, disconnect, homerun, and transition points.
- Roof penetration and cable-support conditions.
- Date, weather, instrument, calibration, and technician.
Do not add a test merely to populate the baseline. Perform only tests allowed by the installed equipment and project safety plan.
The baseline must match the final installed array. Results collected before an equipment substitution or string reconfiguration belong to the superseded revision. Keep them for history, but do not present them as the current acceptance record.
Use consistent string identifiers across the layout, combiner labels, inverter portal, test sheet, and maintenance system. A result named String 4 is weak if the field label says MPPT B2 and the drawing says S-07.
When should the investigation be escalated?
Escalate when the procedure, evidence, or system condition does not support a safe conclusion. Production pressure is not a basis for broadening the test.
Contact the inverter or equipment manufacturer when:
- The event persists after all verified faulted strings are removed.
- The manual does not cover the installed equipment combination.
- Readings conflict with the product’s diagnostic sequence.
- Firmware, sensing, or inverter hardware may be involved.
- The stated test would expose connected electronics to an unclear voltage.
- The product identifies an internal service condition.
Escalate to the engineer, authority having jurisdiction, or utility when a repair changes approved equipment, grounding method, protection, disconnects, conductor routing, system rating, or operating mode.
The solar permit package checklist helps identify shared permit records. Utility approval and permit approval remain separate when the modification affects both.
Do not hide a changed inverter model behind an updated cut sheet. A different product can change protection functions, grounding compatibility, input limits, listings, and approved shutdown behavior.
Why is a generic megger test dangerous?
An insulation-resistance tester applies a test voltage. Connected electronics may not tolerate that voltage or polarity. The array can also present hazardous voltage before the tester is connected.
Three limits must agree:
| Limit | Source of truth |
|---|---|
| Test-equipment rating | Current instrument manual and calibration record |
| Permitted test voltage | Inverter, module, and connected-equipment instructions |
| Test boundary | Approved electrical isolation plan and system drawing |
Do not choose a familiar test voltage from memory. Do not use an online threshold from a different inverter family. Do not infer that a high resistance reading proves every conductor and connector is healthy.
SMA’s Sunny Boy ground-fault procedure shows why model control matters. It provides a product-specific sequence and instrument boundary. The Sunny Tripower X procedure is similar in purpose but not identical in every limit.
The tester, inverter, modules, optimizers, and surge devices all belong in the method statement. If those records conflict, stop and obtain manufacturer direction.
Can conductor-to-ground voltage locate the fault?
Some manufacturer procedures use positive-to-ground, negative-to-ground, and conductor-to-conductor voltages to estimate fault location. This method depends on stable readings and the stated topology.
The Sunny Boy procedure says the sum of both conductor-to-ground readings should approximate the conductor-to-conductor voltage under its stated test conditions. It then uses the voltage ratio to estimate where the fault lies.
That is a localization aid, not a universal equation. Parallel paths, moisture, surge devices, module electronics, optimizers, and unstable leakage can distort the result. The identified area still needs isolation, inspection, repair, and an approved verification test.
Do not perform live voltage measurements unless the current product procedure requires them. The technician, instruments, personal protective equipment, and work controls must match the hazard assessment.
How is a ground fault different from an arc fault?
A ground fault is an unintended path from an energized conductor to ground or bonded metal. An arc fault is current flowing across an unintended gap or damaged connection. One event can create conditions for the other, but their detection methods differ.
| Event | Typical electrical signature | Protection record to verify |
|---|---|---|
| Ground fault | Leakage or insulation breakdown to ground | Isolation and ground-fault monitoring |
| Arc fault | Electrical arcing and characteristic signal behavior | Arc-fault circuit protection |
| Overcurrent | Current above the protected circuit’s design range | Fuse or circuit breaker |
| Surge | Short-duration overvoltage | Surge protective device and grounding path |
Do not close a ground-fault ticket because arc-fault protection shows no alarm. Do not replace a ground-fault investigation with continuity testing of the bonding path.
The arc-fault glossary entry covers that separate failure mode. Both functions should appear accurately in the design and commissioning records.
What should the permit and one-line drawings show?
The drawing set should show the grounding method, equipment bonding, protection functions, and test boundaries without inventing device behavior. A generic GFP note is not enough.
Include these items when applicable:
- Exact inverter manufacturer, model, suffix, and listed system context.
- PV source and output circuit arrangement.
- Maximum system voltage and DC input assignments.
- Equipment grounding conductors and bonding path.
- Module-frame and racking bonding method.
- DC disconnects, combiners, and surge protective devices.
- Inverter-integrated ground-fault and isolation functions from the product evidence.
- Arc-fault and rapid-shutdown functions as separate records.
- Optimizers, module-level electronics, batteries, and other connected electronics.
- Labels, shutdown sequence, and applicable service notes.
Do not label the system grounded or ungrounded from habit. Determine the current-carrying conductor relationship from the listed equipment, design, and adopted rules.
The NFPA 70 development page is the official code source. The enforceable edition depends on state and local adoption. Use the NEC edition-by-state guide to start the adoption check, then verify the project AHJ.
Avoid a universal statement such as NEC 690 always requires one named device or topology. The answer can change with edition, system grounding, array location, listed equipment, and local amendments.
What original record prevents repeat faults?
Use a Four-Record Fault Map. It binds the alarm, topology, test boundary, and physical repair into one closeout package.
| Record | Minimum contents | Release question |
|---|---|---|
| Alarm record | Event code, time, input, weather, operating state | What did the inverter actually report? |
| Topology record | Current one-line, string map, equipment, grounding method | What system was present during the event? |
| Test record | Procedure revision, boundary, instruments, results, technician | What was tested safely and under which method? |
| Repair record | Defect, repair, photos, retest, drawing revision, restart approval | What changed, and who accepted restart? |
This map prevents a common failure. The field team replaces a connector, clears the alarm, and leaves no link to the string map. The same fault then returns, but the next technician cannot reconstruct the earlier work.
Store the map with the commissioning and operations records. If the system changes, create a new revision. Do not overwrite the event history.
How do optimizers and batteries change troubleshooting?
Optimizers, module-level electronics, and batteries add active components to the DC system. They also add limits on isolation testing, disconnection, and acceptable equipment combinations.
A test that is safe for plain module strings may damage connected electronics. A battery can add another energized source. Backup equipment can change the AC and DC isolation boundaries.
Before testing, identify:
- Every energy source.
- Every power-conversion device.
- Every communication or sensing conductor.
- Every surge protective device.
- The approved shutdown state for each component.
- The equipment that must be disconnected before testing.
- The manufacturer-approved restart order.
The solar and storage plan-set guide covers drawing coordination. The fault procedure must still come from the installed product manuals.
Which shortcuts create the most risk?
The worst shortcuts remove evidence or defeat protection.
| Shortcut | Why it fails | Controlled alternative |
|---|---|---|
| Repeatedly resetting the inverter | Clears context and can re-energize the faulted circuit | Freeze event data and investigate |
| Using one resistance threshold for every system | Product, array, weather, and topology differ | Use the exact manual and engineered expectation |
| Testing through connected electronics | Test voltage can damage equipment | Define and verify the isolation boundary |
| Treating dry-weather recovery as repair | Moisture path may return | Correlate weather, inspect, repair, and retest |
| Replacing parts without updating drawings | Future technicians lose the installed truth | Revise the as-built and string map |
| Confusing bonding continuity with insulation health | The tests answer different questions | Keep both records and acceptance criteria separate |
| Bypassing the alarm to restore production | Removes a safety control without resolving the fault | Escalate to manufacturer and qualified review |
The 2013 Sandia ground-fault simulation report documented fire and shock risks from undetected PV ground faults. Its specific blind-spot analysis concerned AC-isolated, DC-grounded systems. Do not misapply that topology to every transformerless inverter.
The lasting lesson is narrower. A protection label does not eliminate fault-detection limits. System topology and device operation must be understood together.
Common transformerless ground-fault questions
Can the inverter be restarted after the alarm disappears?
Not solely because the alarm disappeared. A qualified person should follow the product procedure, identify the cause, verify the repair, and confirm the restart conditions. Moisture-related insulation faults can disappear after drying and return later.
Is a transformerless inverter unsafe?
No. Transformerless inverters are covered by product-safety standards and can be installed safely within their evaluated system. The design must preserve the required grounding, bonding, protection, equipment compatibility, installation, and maintenance controls.
Does opening the AC breaker make the DC side safe?
No. PV modules can energize DC conductors in light. The product-specific shutdown and verification procedure must establish the safe work condition. Backup sources and batteries can add other energized paths.
Can any insulation tester be used?
No. The instrument rating, permitted test voltage, measurement method, and electrical boundary must match the equipment instructions. Connected inverters, optimizers, surge devices, batteries, and monitoring circuits may require isolation first.
Should a plan set show a separate ground-fault device?
Only when the approved design and listed equipment require one. Some protection is integrated into the inverter. Drawings should identify the actual evaluated function without inventing a separate device or implying universal NEC treatment.
Final fault-to-restart checklist
Before restart, verify these records:
- Exact inverter event and timestamp preserved.
- Model, suffix, firmware, and manual revision confirmed.
- Adopted code edition and AHJ requirements checked.
- Grounding method and equipment-bonding path identified.
- Shutdown and lockout completed by qualified personnel.
- Test instruments and boundaries approved.
- Faulted section localized without bypassing protection.
- Physical defect found and repaired.
- Required upstream and repaired-section tests passed.
- Strings with unresolved faults remain disconnected.
- One-line, string map, photos, and as-built records updated.
- Manufacturer escalation completed when the event remains.
- Restart authorization recorded.
A transformerless inverter DC ground fault is not only an inverter event. It is a consistency test across equipment, wiring, weather, drawings, protection, and field records.
Heaven Designs can support the solar permit design and coordinated electrical drawing scope. For a project review, share the current one-line, string map, equipment schedule, event record, and manufacturer documents through the project quote form. Fault diagnosis and energized work must remain with qualified personnel under the applicable safety program.