Skip to content
SLSolar Ledger

Diagnostics

Photovoltaic troubleshooting guide

Faults are listed by what you actually observe, not by what the component is called. Each entry gives the causes worth considering, the measurement that separates them, and what to do once you know. Work down the tests in order — they are arranged so the cheapest discriminating measurement comes first.

12 of 12 symptoms

ArrayIEC 62446-3

One string produces noticeably less than its identical neighbour

Likely causes

  • Partial shading on one module in the string
  • Failed or conducting bypass diode
  • High-resistance connector in the string
  • Soiling concentrated on one row

Tests, in order

  1. 1.Compare open-circuit voltage string to string: a deficit near one module's Voc points to a shorted bypass diode.
  2. 2.Compare operating current at the same irradiance: equal current with low voltage is a diode or cell problem, low current is shading or soiling.
  3. 3.Thermal scan under load; a hot connector or a hot cell block localises the fault immediately.
  4. 4.Curve-trace the string if a tracer is available and compare shape to the healthy string.

Resolution Replace the affected module or connector. A shorted bypass diode is not self-healing and costs roughly one third of that string's output permanently.

ArrayIEC 62446-1

Inverter reports an insulation resistance or earth fault, usually in the morning

Likely causes

  • Moisture ingress into a junction box or connector
  • Cable damaged against a sharp edge or roof tile
  • Water in a DC isolator enclosure
  • Degraded module backsheet

Tests, in order

  1. 1.Note whether the fault clears as the array dries: a morning-only fault is almost always moisture.
  2. 2.Isolate and test insulation resistance string by string at the specified test voltage.
  3. 3.Halve the suspect string repeatedly, testing each half, to bisect down to the faulty section.
  4. 4.Inspect every connector for water staining and every cable entry for missing glands.

Resolution Repair or replace the wet component and re-test. Do not raise the inverter's fault threshold to silence the alarm.

GridEN 50549-1

Inverter shuts down around midday and restarts a few minutes later

Likely causes

  • Grid over-voltage trip caused by voltage rise on the supply cable
  • Undersized AC cable between inverter and board
  • Weak or long distribution feeder with several PV systems on it
  • Over-temperature derating in a hot inverter location

Tests, in order

  1. 1.Read the inverter event log: an over-voltage code with a recorded value near the trip threshold confirms it.
  2. 2.Log AC voltage at the inverter terminals and at the main board simultaneously for 24 h.
  3. 3.Calculate the voltage rise from inverter current and AC cable impedance; compare with the measured difference.
  4. 4.Check heatsink temperature and ambient in the same window to rule out thermal derating.

Resolution Shorten or upsize the AC cable to cut voltage rise first; only then raise the matter with the network operator, with logged data.

Monitoring

Annual output is 10–20% below the design estimate with no faults logged

Likely causes

  • Optimistic design assumptions rather than a hardware fault
  • Shading not modelled at winter sun angles
  • Soiling in a dry or agricultural location
  • Clipping from an inverter sized too small for the array
  • Degradation faster than the warranted rate

Tests, in order

  1. 1.Calculate performance ratio from measured yield and on-site or satellite irradiance, rather than comparing to the sales estimate.
  2. 2.Plot a clear-day power curve: a flat top means clipping, a shoulder means shading, a low but correctly shaped curve means soiling or degradation.
  3. 3.Clean a representative row and compare its string against the others for a week.

Resolution Separate design error from fault before spending money. Only a shaped-curve anomaly justifies hardware intervention.

Storage

Battery sits at full charge and never discharges into the evening load

Likely causes

  • Charge and discharge windows configured for the wrong tariff periods
  • Reserve state of charge set too high
  • CT clamp fitted the wrong way round or on the wrong conductor
  • Inverter in a grid-charge-only or backup-priority mode

Tests, in order

  1. 1.Watch live import, export and battery power for five minutes while a kettle is switched on: correct sensing shows import drop and battery discharge.
  2. 2.Check the CT direction indicator in the inverter's live data; reversed sign is the single most common cause.
  3. 3.Read back the configured schedule and compare with the tariff clock, including any daylight-saving offset.

Resolution Correct the CT orientation or schedule. A reversed CT can also cause the battery to charge from the grid at peak price.

ArrayIEC 63027

Arc-fault detector trips repeatedly with no visible damage

Likely causes

  • Loose or partially mated connector
  • Mixed-brand connectors mated together
  • Electrical noise from a nearby inverter or motor load causing false trips
  • Cracked cell producing intermittent arcing under flex

Tests, in order

  1. 1.Record the time of each trip and correlate with irradiance and wind: wind-correlated trips point to mechanical movement.
  2. 2.Inspect and re-mate every connector on the affected string with the correct tool, checking for brand mismatch.
  3. 3.Thermal scan immediately after a trip while the fault site is still warm.

Resolution Replace mismatched or damaged connectors with a single matched brand. Never disable arc-fault protection to keep a system running.

Monitoring

Monitoring shows offline but the system is clearly generating

Likely causes

  • Router replaced or Wi-Fi password changed
  • Dongle powered from the inverter and asleep overnight, failing to reconnect
  • Portal account migrated by the installer and never handed over
  • 2.4 GHz band disabled on a new mesh router

Tests, in order

  1. 1.Check the inverter's local display or access point for its own record of yield.
  2. 2.Confirm the 2.4 GHz SSID is broadcasting and not band-steered.
  3. 3.Re-run the commissioning app's network setup rather than resetting the inverter.

Resolution Restore the network link and confirm historical data backfills. Missing monitoring is not lost generation, but it hides the faults above.

ArrayIEC 62446-3

One area of a module runs hot in a thermal image

Likely causes

  • Cell crack or interconnect break
  • Persistent shading from a fixed object such as a vent
  • Soiling deposit or bird droppings on a single cell
  • Failed bypass diode leaving a substring reverse-biased

Tests, in order

  1. 1.Distinguish patterns: a single hot cell suggests a crack or soiling, a whole hot substring suggests a diode.
  2. 2.Clean the area and re-image after an hour of similar irradiance.
  3. 3.Measure module Voc and compare with the nameplate, corrected for cell temperature.

Resolution Replace the module if the hot spot persists after cleaning. Hot spots accelerate encapsulant degradation and are a fire risk, not just a yield loss.

GridIEC 60364-7-712

The RCD protecting the PV circuit trips intermittently

Likely causes

  • DC leakage from a transformerless inverter into an AC-only RCD type
  • Accumulated standing leakage from filters across several circuits
  • Genuine insulation fault on the AC side
  • Moisture in an outdoor socket sharing the circuit

Tests, in order

  1. 1.Confirm the fitted RCD type matches the inverter manufacturer's requirement (commonly type B or type A with a stated DC limit).
  2. 2.Measure standing leakage current with a clamp meter on the circuit under normal operation.
  3. 3.Disconnect the inverter and see whether the trip continues.

Resolution Fit the RCD type the inverter documentation specifies. Sharing an RCD across many circuits is the second most common cause.

Inverter

Inverter output plateaus below its rating on hot clear days

Likely causes

  • Thermal derating from poor ventilation or direct sun on the enclosure
  • Deliberate power limitation configured for the grid connection
  • DC/AC ratio high enough to clip regularly by design

Tests, in order

  1. 1.Read heatsink temperature against ambient in the log at the moment of the plateau.
  2. 2.Check the configured active power limit before assuming a fault.
  3. 3.Compare the plateau level with the nameplate: a clean flat line at exactly the AC rating is clipping, a drifting ceiling is thermal.

Resolution Shade or ventilate the inverter, or accept clipping if the DC/AC ratio was an economic choice. Clipping a few per cent of annual yield is often the cheaper design.

ArrayIEC 62548

Green or white corrosion at frames, clamps or earthing points

Likely causes

  • Dissimilar metals in contact, typically copper against aluminium
  • Coastal salt exposure with unsuitable fixings
  • Damaged anodising at a clamp point

Tests, in order

  1. 1.Identify the metal pair at every corroded joint.
  2. 2.Check earthing continuity across the corroded joint; resistance rises as corrosion advances.
  3. 3.Inspect fixings for the specified stainless grade rather than plated steel.

Resolution Separate dissimilar metals with the correct washer or bonding component and re-test continuity. Corrosion at an earthing point is a safety fault, not cosmetic.

Array

Output stays at zero for days after snowfall while neighbours recover

Likely causes

  • Shallow tilt preventing shedding
  • Frame lip retaining a snow bank along the lower edge
  • Portrait/landscape orientation with the string wired across the lower row

Tests, in order

  1. 1.Observe the shedding pattern; a persistent lower band means one series row is shading the whole string.
  2. 2.Compare with a differently oriented array nearby before assuming a fault.

Resolution Where snow is routine, plan string wiring so a lower-edge band affects one string rather than all of them, and never clear snow with tools that scratch the glass.

Decision trees

Where a fault is really a design question, work the tree instead of the symptom list.

Decision tree

Choosing a DC cable cross-section

Work down the list in order. Ampacity is a safety limit and comes first; voltage drop is an economic limit and is decided afterwards.

  1. Is the cable rated for the design current after every derating factor?

    Design current for a PV string is 1.25 × Isc(STC). Derate for ambient temperature, grouping, conduit and solar gain on the roof surface.

    No
    Increase the cross-section until the derated ampacity exceeds design current. Nothing below this is negotiable.
    Yes
    → go to step 2

  2. Is the DC voltage drop at maximum power below 1 percent?

    Measure the true routed length, not the plan distance, and double it for the two-wire circuit.

    Above 3%
    Go up at least two sizes. A drop this large costs real yield every sunny hour of the system's life.
    1–3%
    Acceptable on long ground-mount runs where the copper cost is material; otherwise step up one size.
    Below 1%
    → go to step 3

  3. Does the AC side stay within the total drop allowed by the wiring rules?

    Most national rules derived from IEC 60364 allow 3 percent for the final circuit and 5 percent overall from the origin of the installation.

    No
    Enlarge the AC cable or move the inverter closer to the board. Do not solve an AC problem by enlarging DC cable.
    Yes
    → go to step 4

  4. Is the conductor mechanically and environmentally suited to the route?

    Rooftop DC cable needs UV-stable, double-insulated, halogen-free construction to EN 50618 with a 90 °C or higher conductor rating.

    No
    Change cable type before changing size. A correctly sized cable in the wrong insulation still fails early.
    Yes
    The size is settled. Record the calculation in the commissioning file.

Decision tree

String, optimiser or microinverter

Architecture is decided by the roof, not by preference. Answer in order and stop at the first branch that terminates.

  1. Does any part of the array suffer unavoidable partial shade during generating hours?

    Yes, on some modules
    → go to step 2
    No
    → go to step 3

  2. Is the shade confined to a minority of modules?

    Shade on a handful of modules is a module-level problem. Shade across the whole array is a siting problem no electronics can fix.

    A few modules
    Optimisers on the affected modules, or microinverters if the roof is complex as well as shaded.
    Most of the array
    Reconsider the location. Distributed electronics recover a fraction of the loss at full hardware cost.

  3. Does the array span more than two roof planes or orientations?

    Yes
    Microinverters, or a multi-MPPT string inverter with one plane per tracker.
    No
    → go to step 4

  4. Do local rules require module-level rapid shutdown?

    Common in North America under NEC 690.12; increasingly requested by insurers elsewhere.

    Yes
    Optimisers or microinverters, which satisfy the requirement inherently.
    No
    A string inverter. Fewest components on the roof, highest conversion efficiency, lowest cost per watt.

Decision tree

Is a battery justified on this installation

Storage pays on the spread between what you avoid and what you would have been paid for exporting. Test that spread before anything else.

  1. What is the gap between the import price and the export price per kilowatt-hour?

    Under 5c
    No. Under net metering or a generous export tariff, arbitrage cannot repay the hardware. Buy storage only for backup.
    5–15c
    → go to step 2
    Over 15c
    → go to step 2

  2. Is there routine exported surplus to charge from?

    Compare annual generation against the share of load that falls in daylight. A battery cannot store energy that was already consumed.

    No surplus
    No. Enlarge the array first; storage without surplus simply cycles grid electricity at a loss.
    Regular surplus
    → go to step 3

  3. Is the evening and overnight load large enough to discharge the battery most days?

    Under 3 kWh
    Size to that load only. A larger unit will sit part-charged and never earn its capacity.
    3–10 kWh
    → go to step 4
    Over 10 kWh
    → go to step 4

  4. Does the annual saving exceed the installed cost divided by the warranted cycle life?

    Divide installed cost by warranted throughput in kilowatt-hours to get a cost per stored unit, then compare it to the import–export spread.

    No
    Not yet. Free load shifting captures a third of the same benefit; revisit when prices or tariffs move.
    Yes
    Proceed, sizing to the evening load rather than to the array.

Decision tree

Responding to a shading problem

Diagnose the geometry before buying hardware. Most shading losses are cheaper to design out than to electronically compensate for.

  1. What casts the shadow?

    Own structure
    Move or re-orient the array. A flue, dormer or parapet you control is a layout problem, not an electronics problem.
    Vegetation
    → go to step 2
    Neighbouring building
    → go to step 3

  2. Can the vegetation be managed for the life of the system?

    Yes
    Trim, and record the clearance in the maintenance schedule. Re-model the shade at mature tree height, not today's.
    No
    → go to step 3

  3. When does the shadow fall across the array?

    Shading between 10:00 and 15:00 in the shoulder seasons drives most of the annual loss; early-morning shade in December costs almost nothing.

    Peak hours
    → go to step 4
    Edges of the day
    Accept it. Model the loss honestly in the yield estimate and leave the array as designed.

  4. Can the affected modules be grouped onto their own string?

    Yes
    Separate the string onto its own MPPT input. Cheapest effective fix, no extra roof-mounted electronics.
    No
    Fit optimisers on the affected modules only, or microinverters if the layout is fragmented as well.

Decision tree

Fixing the array size

Size from the number you can measure, then test the answer against the three limits that actually cap it.

  1. Do you have twelve consecutive months of consumption data?

    No
    Get it before sizing. A single quarter over- or under-states annual demand by a wide margin in either direction.
    Yes
    → go to step 2

  2. Will a heat pump, electric vehicle or extension land within five years?

    Likely
    Add the expected load now. Retrofitting modules later means a second scaffolding visit and often a new grid application.
    Speculative
    Size for today but reserve roof space and a spare MPPT input. → go to step 3
    No
    → go to step 3

  3. Does the unshaded roof area accommodate the calculated capacity?

    Plan on 5.5 to 6.5 m² of gross pitched roof per kilowatt after setbacks, and 10 to 14 m² per kilowatt on a flat roof with row spacing.

    No
    Take the capacity the roof allows and accept lower coverage. Do not fill setbacks or access routes to hit a target.
    Yes
    → go to step 4

  4. Does the network operator's export limit or connection band cap the system below that size?

    Yes
    Design to the limit, or price the upgrade before committing. This is the most common late-stage redesign.
    No
    Size is confirmed. Round up to whole modules and check the inverter DC/AC ratio next.