What this page solves

PV DC combiner boxes aggregate many 600–1500 V strings into a high-current bus that feeds a DC/DC stage, central inverter or PCS. In this junction, string fuses, terminals and busbars carry large currents in confined enclosures that are often exposed to heat, dust and humidity, while remaining physically remote from operators.

In many plants, these components are still supervised mainly by periodic visual inspection and a few high-level alarms from the inverter. Blown fuses, half-melted links, loose terminals and corroded contacts can remain hidden for long periods, only showing up as slow yield loss, local overheating or, in the worst case, DC arcing and fire. A single degraded string may drag down a whole array while staying below simple overcurrent and insulation thresholds.

Relying only on total bus current or inverter power masks many of these issues. Subtle contact degradation increases voltage drop at fuses and terminals long before a fuse opens or a breaker trips. Without continuous measurements at the combiner box, there is little hard data to distinguish genuine electrical ageing from weather effects or module soiling, and it becomes difficult to justify preventive maintenance, warranty claims or insurance discussions.

This page focuses on turning the combiner box into a monitored asset by combining bus-current sensing with voltage-drop (ΔV) monitoring across fuses and critical contacts. The goal is to extract simple, robust signals that indicate fuse health, contact resistance and string presence, then convert those into latched events and reports suitable for local RTUs, SCADA and EMS. Arc-fault signatures, DC surge and lightning counters, and hotspot temperature maps are treated in dedicated pages; here the emphasis stays on fuses, contacts and the DC bus current path.

Combiner box topology & protection elements

A DC combiner box collects many PV strings and routes their current into a shared high-current bus. Each string normally includes a dedicated gPV fuse or equivalent protection link, so that reverse currents from the array do not feed into a faulted string. The fused string outputs join copper busbars or heavy cables that form the main DC bus toward the inverter or DC/DC stage, often passing through a DC disconnect or load-break switch and one or more surge protective devices.

Typical elements include string fuses in plug-in or bolt-on holders, DC breakers or switches at the combiner outlet, SPD blocks referenced to PE or ground, and multiple terminal blocks and busbar joints. Each of these components introduces a small series resistance and contact interface. Under normal conditions the resulting voltage drops are only a few tens of millivolts at rated current, but any mechanical loosening, corrosion or partial melting can increase these drops and concentrate heat in a very small volume.

From a monitoring perspective, not every node needs explicit measurement. The most useful locations are the main DC bus, where a shunt or current sensor captures total combiner current, and the fuse or terminal points where a small, well-defined resistance allows voltage-drop (ΔV) sensing. Measuring ΔV across selected fuses or contacts highlights both open and high-resistance faults, while the bus current reference provides context about irradiance, array loading and the number of strings that are still contributing.

In many designs it is practical to monitor ΔV across a subset of fuses or shared terminals rather than every single string. Additional points, such as the DC disconnect and key busbar joints, can be instrumented when channel count and wiring budget allow. SPD blocks are noted in the topology because they influence the current path and ageing of the box, but detailed surge-waveform capture and lightning event counting are covered in the dedicated DC surge and lightning monitoring page.

The simplified topology in this section therefore distinguishes between elements that define string-level protection (fuses), outlet isolation (DC breaker or switch), surge limiting (SPD) and the passive copper network of terminals and busbars. Later sections build directly on this layout, mapping each physical node to bus-current AFEs, ΔV channels, comparators and digital interfaces that can expose combiner health to plant analytics and SCADA systems.

What needs to be monitored and why (fuse & contact failure modes)

Once the DC combiner topology is clear, the next step is to identify which electrical quantities carry the most information about fuse health, contact integrity and string presence. In practice, most relevant failure modes fall into a few families: fuse behaviour, terminal and busbar contact degradation, outlet switch issues and single-string loss. For each family, bus current and voltage-drop (ΔV) signals expose different patterns that are suitable for automated monitoring.

For fuses, the simplest case is a clean, intentional blow under short-circuit or reverse-current conditions: string current collapses, the fuse opens and the associated ΔV channel returns close to zero. More problematic are nuisance trips and partial melting. A half-melted fuse can continue to conduct with elevated resistance, generating higher ΔV and heat while staying below simple overcurrent thresholds. Without ΔV or temperature monitoring, this state may persist until the body or holder is seriously damaged. In arrays with many strings, the resulting loss in total power can be small compared with normal irradiance variation, making the fault easy to overlook.

Terminal blocks, crimp lugs and busbar joints introduce similar risks. Loosening, corrosion, contamination or poor torqueing reduce the real contact area and cause contact resistance to drift upward. Under full load, the affected point experiences disproportionate voltage drop and heating even though current is still flowing. The most revealing observable quantity in these cases is the small DC voltage drop across the contact at a known current, not just the absolute current level. Monitoring ΔV at selected fuses and terminals highlights these high-resistance states long before catastrophic failure or visible discoloration occurs. Local temperature sensing can support this diagnosis but is covered in detail in the J-box hotspot monitoring page.

Outlet DC switches and breakers introduce a different class of problems. A welded or stuck contact may refuse to open even when a disconnect command is issued, leaving downstream equipment energized. In this case, bus current remains high while the mechanical position indicator reports an open state. Conversely, a degraded outlet contact behaves more like a high-resistance terminal, adding ΔV and heat on the way to the inverter or DC/DC stage. Single-string loss, whether due to an open fuse, a disconnected cable or a failed connector, appears as a drop in current for that path and a corresponding decrease in total combiner current at a given irradiance. In large arrays, string-level ΔV and current information often provide the only practical way to distinguish between normal environmental effects and silent string outages.

Typical failure modes and observables

With this mapping in place, the monitoring problem becomes a matter of selecting which ΔV and current points to instrument and how to route them into AFEs, ADCs and comparators. The next section builds on this table and focuses on practical bus-current and multi-point ΔV sensing architectures that fit inside a DC combiner box.

Bus current & ΔV sensing architectures

The failure patterns identified in the previous section require a sensing architecture that captures both total DC combiner current and small voltage drops across selected fuses and contacts. A single analogue output current transducer is no longer sufficient. Instead, the combiner box benefits from a small multi-channel AFE that measures bus current, multiplexes ΔV channels, and feeds comparators and a local controller with signals suitable for thresholds, time-over-threshold logic and logging.

For bus current, shunt-based sensing remains a common choice. A low-value shunt in the main DC path provides a well-defined relationship between current and voltage, at the cost of a small power loss. The associated AFE must tolerate the high common-mode voltage of 600–1500 V DC and offer sufficient accuracy across a range from low irradiance to full load. High-side current-sense amplifiers, sigma-delta current-sense modulators and metering-grade ADCs are typical options. In a DC combiner, bandwidth requirements are moderate: it is important to follow irradiance ramps, MPPT dynamics and fault transients on the order of tens of milliseconds, but there is no need to reproduce switching waveforms from downstream converters.

Voltage-drop sensing across fuses and contacts is more sensitive to layout and front-end design. Kelvin taps placed as close as possible to each side of a fuse, terminal or switch contact isolate the resistance of the element of interest from the rest of the copper network. The resulting ΔV is typically in the tens to hundreds of millivolts at rated current. An instrumentation amplifier or low-drift op amp with high common-mode rejection and modest gain then scales this signal up to a range suitable for ADC or comparator inputs. Zero-drift devices help distinguish slow, real changes in contact resistance from op-amp offset drift over the service life of the combiner box.

Filtering and response time represent a key trade-off. Strong switching transients, EMI and MPPT steps can introduce high-frequency content on both bus current and ΔV signals. Simple RC low-pass networks placed close to the AFEs reduce these disturbances and protect the measurement chain, but excessive filtering can mask the onset of rapid faults. For fuse and contact monitoring, a bandwidth in the low-hundreds-of-hertz range often provides a good balance, revealing abrupt step changes and slow drifts while ignoring fast switching edges. Short-circuit and crowbar protection on the DC bus typically rely on separate, faster-acting protection elements.

Multi-point ΔV sensing can be implemented with dedicated multi-channel instrumentation amplifiers, with per-path gain and filtering, or with an analogue multiplexer feeding a smaller number of precision amplifiers and ADC channels. The best choice depends on the number of strings, wiring complexity and required sampling cadence. When a central MCU or RTU already exists inside the combiner box, multiplexed ΔV channels can be scanned at a rate of a few samples per second per channel, which is usually sufficient for detecting half-melted fuses, high-resistance contacts and string loss. Selected analogue outputs can also feed comparators directly for local, hardware-based thresholding and latched event flags.

In some designs, Hall-effect or fluxgate sensors replace the shunt for bus-current measurement to avoid additional power loss and simplify insulation coordination. These sensors naturally provide galvanic isolation and can offer wide dynamic range. Even in these cases, the role of ΔV sensing does not disappear: bus current establishes overall loading, while ΔV across fuses and contacts remains the primary indicator of contact resistance and localized ageing. Shunt-based and Hall-based approaches therefore differ mainly in bus-current implementation; the need for well-designed ΔV AFEs across selected elements is common to both.

The block-level architecture that emerges is a combination of one bus-current channel and a set of ΔV channels sharing an AFE, comparator bank and digital interface. Later sections focus on how thresholds, debounce timing and latching convert these analogue signals into robust events and alarms that can be integrated into SCADA and plant analytics.

Comparator thresholds, debounce & event latching

Turning bus-current and ΔV measurements into actionable alarms requires more than raw ADC values. Comparator thresholds and references define which patterns count as faults, while debounce and time-over-threshold logic prevent cloud transients and MPPT steps from generating spurious events. Latching ensures that once a serious condition has been detected, such as a blown fuse or a high-resistance contact, the event remains visible until a deliberate reset is issued from local controls or SCADA.

For ΔV channels, one common strategy is to define at least two thresholds. A warning threshold flags ΔV values that are higher than normal at a given current, indicating early contact degradation. A higher trip threshold marks clearly abnormal ΔV that is consistent with a half-blown fuse or severely damaged terminal. Precision voltage references and resistor networks generate these limits; multi-channel comparators then evaluate each amplified ΔV signal against the reference. In addition, a combination of ΔV close to zero and near-zero string current indicates an open fuse or disconnected string, particularly when neighbouring strings are still carrying current at similar irradiance.

Bus-current comparators play a complementary role. A continuous-current limit protects against long-duration overloads that cause thermal stress, while a higher, short-duration threshold can flag transient overcurrent conditions. In a combiner box environment, these comparators are typically used to generate monitoring events rather than to drive primary trip coils directly. They provide fast, hardware-level detection of abnormal loading that can be correlated with inverter status, grid conditions and other plant signals at higher levels.

Debounce and blanking are essential to avoid false trips. PV arrays are inherently dynamic; passing clouds, inverter MPPT sweeps and load changes can cause short-lived excursions in both current and ΔV. Simple debounce requires a signal to exceed its threshold for a minimum time window before an alarm is accepted. More advanced schemes use time-over-threshold metrics, accumulating the duration that a signal remains above a warning limit. If the accumulated time exceeds a configurable budget, a degraded condition is declared even if each individual excursion was short. Startup and switching operations can be masked with blanking windows to prevent expected transients from generating alarms.

Once a fuse opens or a high-resistance contact is confirmed, the associated alarm should not disappear automatically when conditions fluctuate. Latched events provide the required persistence. Comparators with built-in latches, external RS flip-flops or MCU-managed status bits capture the first occurrence of a fault and hold it until a reset command is issued locally or via SCADA. This approach supports traceability, encourages proper root-cause analysis and avoids situations where intermittent faults keep clearing themselves before maintenance staff can review the evidence.

MCU / SCADA interface, interrupts & reporting

The final step in a combiner monitoring chain is to hand clean, meaningful information to the plant control system. A local MCU or RTU inside the combiner box aggregates latched comparator outputs and AFE measurements, timestamps events and exposes them through a fieldbus or Ethernet interface. This intermediate controller layer allows per-plant configuration of thresholds, debounce times and alarm grouping without changing the underlying analogue hardware.

Comparator outputs and latch signals connect to MCU or RTU GPIOs, often as interrupt inputs. Edge-triggered interrupts ensure that fuse openings, overcurrent conditions and high-resistance contacts are captured immediately, even if they occur between regular polling cycles. In parallel, I²C or SPI interfaces link the controller to current-sense amplifiers, sigma-delta modulators or metering ADCs. Through these links the controller reads bus-current values, ΔV measurements and device diagnostics, and can program thresholds, gains or sampling rates for adaptable behaviour across different combiner designs.

Accurate timestamps are essential for root-cause analysis and correlation with grid or inverter events. A local real-time clock maintained by the MCU or synchronized from a higher-level gateway provides this time base. When an interrupt arrives, the controller records an event entry containing the time, string or combiner identifier, event type and a snapshot of relevant analogue quantities such as ΔV, bus current and internal temperature. Short circular buffers hold recent history so that transient issues can be reviewed even if communication to the control room is temporarily interrupted.

Upstream communication typically uses familiar industrial buses. At the serial level, Modbus-RTU over RS-485 is widely used to integrate multiple combiners into a central data concentrator. Ethernet-based systems may use Modbus-TCP, lightweight vendor-specific protocols or, in larger substations, IEC 61850 or similar station-bus schemes. In all cases, the combiner controller does not attempt to implement full microgrid control logic; it exports a compact view of combiner health that higher layers can consume alongside inverter, meter and protection data.

For SCADA and analytics, it is useful to classify events into a small number of categories. Trip-level events cover definitive faults such as fuse openings and severe overcurrent that require immediate attention. Warning events flag borderline conditions that may clear on their own but still merit investigation. Degraded events indicate confirmed long-term issues such as persistent high contact resistance. Each event record carries a timestamped snapshot and a code that identifies the affected string or combiner, enabling fleet-wide dashboards and maintenance planning. On the hardware side, this interface layer highlights the roles of current-sense and AFE ICs, comparator and reference devices, GPIO expanders, digital isolators, fieldbus transceivers and microcontrollers with the security features and robustness appropriate for long-life PV infrastructure.

Diagnosable events & analytics (string-level visibility)

String-level visibility turns basic electrical measurements into actionable maintenance intelligence. Instead of only observing total combiner current, the system can track how each string contributes, how fuse and contact ΔV evolve over time and how events correlate across multiple combiners. The result is a set of diagnosable patterns that point to soiling, shading, contact degradation or upstream faults, and a clear way to prioritise which strings and combiners deserve attention first.

A common scenario is a single string delivering 10–20% less current than its neighbours for many hours or days, even though irradiance and module orientation are comparable. With per-string current and ΔV monitoring, this pattern appears as a persistent negative deviation from the group average while fuse and contact ΔV remain near normal. Likely causes include heavy soiling, partial shading or gradual module degradation. This situation does not warrant an immediate trip but clearly qualifies as a degraded condition that should be scheduled for cleaning or inspection before energy yield is unnecessarily lost over an entire season.

Another class of patterns involves correlated trips. If several strings in the same combiner, or even across multiple combiners, report fuse-open or string-loss events within a very narrow time window, the root cause is unlikely to be independent string faults. Combined with a sudden step change in bus current and, in some plants, matching entries in surge or protection logs, this behaviour points towards upstream issues such as main DC breaker operations, surge events or inverter-side protection actions. Recognising this pattern avoids wasting time on string-by-string inspections when the focus should be on system-level protection and grounding.

Slow trends in contact ΔV form a third, highly valuable pattern. When the ΔV across a particular fuse holder, terminal block or busbar joint steadily rises over weeks or months at similar load, the implied resistance is increasing. Causes include corrosion, insufficient tightening torque, thermal cycling and loss of plating. In daily operation this may not be obvious: string current can remain within a few percent of normal. However, the higher local I·R loss elevates temperature and accelerates ageing. Detecting this trend early allows preventive maintenance crews to rework or replace the affected hardware during planned outages instead of reacting to discoloured, overheated terminals or nuisance trips.

To support systematic maintenance planning, these raw patterns can be converted into simple health scores. Per-string scores combine current deviation from peers, ΔV deviation from design values and recent event history. Per-combiner scores aggregate the worst string health, the number of degraded strings and the frequency of trips. SCADA or asset-management tools can then present a ranked list of combiners and strings by severity, allowing crews to tackle the most critical issues first while still having visibility into lower-priority anomalies that may become more serious over time.

Typical diagnosable patterns and recommended actions

With these patterns and scores in place, combiner monitoring evolves from simple data collection to a structured diagnostic tool. Maintenance teams gain a ranked list of strings and combiners by health, while planners obtain long-term trends that support decisions on cleaning intervals, hardware upgrades and design improvements for future plants.

Recommended IC roles & design checklist

IC roles mapping (function-focused)

A combiner fuse and contact monitor can be built from a small set of analogue and mixed-signal IC categories. The exact choice depends on DC voltage level, number of strings, required accuracy and the preferred interface to the plant control system. The following mapping highlights which roles each class of device typically plays.

• Bus current sensing: high-side current-sense amplifiers and current-sense or metering ADCs measure total combiner current across a shunt or current transducer. Key parameters include common-mode voltage capability, gain accuracy, offset, temperature drift and bandwidth sufficient to capture irradiance ramps and fault transients without reproducing converter switching noise.
• ΔV monitoring around fuses and contacts: instrumentation amplifiers and zero-drift op amps amplify small voltage drops in the tens to hundreds of millivolts across fuse holders, terminal blocks and busbar joints. High CMRR, low offset and low drift help distinguish real resistance changes from front-end errors over the plant lifetime.
• Threshold comparison: window comparators and multi-channel comparators with reference ladders convert analogue ΔV and current signals into discrete warning and trip flags. These devices often include internal hysteresis and can share a precision reference to ensure consistent thresholds across many channels.
• Event latching and GPIO expansion: GPIO expanders with interrupt outputs or small companion MCUs capture comparator outputs, latch fault conditions and expose status bits to the main controller. This approach scales to high string counts while keeping MCU pin counts and board routing manageable.
• Isolation and communication: digital isolators and isolated RS-485 or Ethernet transceivers separate high-voltage measurement domains from low-voltage control logic and connect the combiner monitor to plant SCADA or local gateways. Isolation ratings and surge performance must match the installation’s insulation and lightning-protection strategy.
• Smart protection options: in some designs, intelligent high-side switches or eFuses replace traditional fuses. These devices integrate programmable current limits, fast shutdown and diagnostic outputs that report fault type and duration. In a combiner context they can augment or replace passive fuses, but the detailed eFuse behaviour and system trade-offs belong to dedicated protection design discussions.

Design checklist before selecting ICs

A short design checklist helps frame the sensing and monitoring requirements before specific parts are chosen. Reviewing these items early avoids running into resolution, loss or interface limitations later in the project.

• Maximum string and bus currents, minimum useful ΔV: determine nominal and worst-case currents per string and at the combiner bus, then compute the smallest ΔV that is acceptable across shunts and contact points without harming efficiency. These values set amplifier gain, ADC resolution and shunt or conductor sizing.
• Allowed voltage-drop budget: sum the expected voltage drops of fuses, connectors, shunts and busbar segments to ensure that total I·R losses align with system-efficiency targets. If the budget is tight, ΔV monitoring may need to focus on a subset of strategic points rather than every possible junction.
• Overload and short-circuit timing requirements: clarify whether the combiner monitor is purely diagnostic or also part of the primary protection chain. This decision influences comparator speed, time-over-threshold settings and whether comparator outputs need to interact directly with trip coils or only inform higher-level controllers.
• Environmental, corrosion and insulation ratings: define ambient temperature range, pollution degree and required insulation levels. These factors drive IC operating-temperature grades, package options, creepage/clearance layouts, coating choices and the isolation voltage of digital isolators and communication transceivers.
• SCADA / EMS interface and timestamp strategy: decide which fieldbus or Ethernet protocol will carry combiner health data and where the authoritative time reference resides. This determines whether the combiner MCU needs its own RTC, how events are timestamped and how many historical entries should be stored locally to bridge communication interruptions.
• String count and channel scaling: confirm the number of strings per combiner and identify which strings or contacts need individual ΔV channels. This drives the required number of amplifier, ADC, comparator and GPIO-expander channels and helps choose between fully parallel and multiplexed architectures.

Example IC options (non-exhaustive)

The following part numbers illustrate typical devices used in combiner fuse and contact monitoring. They come from multiple vendors and represent classes of components rather than specific endorsements. Equivalent parts from other manufacturers can be substituted as long as the key parameters align with the design checklist.

• Bus current sensing:
  • INA240 (TI) – high-side current-sense amplifier with enhanced PWM rejection, suitable for shunt-based bus-current sensing on low to medium-voltage rails.
  • AD8210 (Analog Devices) – high common-mode current-sense amplifier for industrial environments with good accuracy over temperature.
  • ADS131M04 (TI) – 4-channel, 24-bit metering ADC that can digitise bus current and selected ΔV channels for high-resolution analytics.
• ΔV sensing and instrumentation:
  • INA333 (TI) – low-power instrumentation amplifier with programmable gain for Kelvin ΔV measurements across fuses and terminals.
  • ADA4528-2 (Analog Devices) – dual zero-drift op amp with extremely low offset and drift, suitable for precise contact-resistance trend monitoring.
• Comparators and window detection:
  • LM339 / LM2903 families – multi-channel comparators with open-collector outputs, often used for basic over/under-threshold detection.
  • TLV1704 (TI) – rail-to-rail, 4-channel comparator suitable for low-voltage logic interfacing with modest propagation delay.
  • MAX9140 (Analog Devices/Maxim) – low-power comparator with rail-to-rail inputs, useful where many channels are required with tight power budgets.
• GPIO expansion and local control:
  • MCP23017 (Microchip) – 16-bit I²C GPIO expander with interrupt pins, ideal for aggregating comparator outputs from many strings.
  • STM32G0 / STM32L0 series (STMicroelectronics) – low-power MCUs frequently used as local combiner controllers for event logging and Modbus/Ethernet gateways.
• Isolation and fieldbus transceivers:
  • ISO7741 (TI) – digital isolator for SPI, I²C or GPIO lines between measurement front-ends and low-voltage controllers.
  • ADM2582E (Analog Devices) – integrated isolated RS-485 transceiver with DC-DC converter, suitable for Modbus-RTU links from remote combiners.
  • SN65HVD72 (TI) – robust RS-485 transceiver for non-isolated segments or where isolation is provided separately.
• Intelligent high-side switches / eFuses (optional):
  • TPS25982 (TI) – eFuse with adjustable current limit, fault reporting and latch/auto-retry modes, applicable to low-voltage combiner segments and auxiliary rails.
  • ITS4xxx families (Infineon) – automotive high-side switches with integrated protection and diagnostic feedback, usable where DC string currents fall within their range.

These devices illustrate the types of ICs that support reliable combiner fuse and contact monitoring. Final selection should align with plant voltage levels, derating policies, qualification standards and vendor strategies, using the design checklist as a common reference between design, reliability and procurement teams.

Application mini-stories (field issues & retrofits)

Real-world field experience shows how bus-current and ΔV monitoring in combiner boxes changes the way faults are detected and maintained. The following mini-stories illustrate how adding per-string ΔV sensing, intelligent thresholds, latching and communication can expose problems that were previously invisible, and how a small set of analogue front-ends, comparators and controllers enables scalable retrofits in both new and existing plants.

Ground-mount plant: slow contact corrosion caught before failure

A large ground-mounted PV plant had been operating for several years with only bus-current metering in each combiner. Production data showed occasional small deviations between parallel fields, but nothing that clearly pointed to individual strings or connection issues. Over time, several fuse holders and terminal blocks began to suffer from corrosion and thermal cycling, slowly raising contact resistance. Because the affected strings still delivered current, their impact on total combiner current was modest and easily lost in normal irradiance and inverter behaviour.

A retrofit added per-string ΔV monitoring at critical fuse and terminal locations using instrumentation amplifiers and low-drift op amps, sharing a precision reference. Multi-channel comparators with warning and trip thresholds converted the amplified ΔV signals into discrete events, while GPIO expanders aggregated the outputs into a compact interrupt interface for a local MCU. The MCU handled debouncing and time-over-threshold logic, timestamped events and exposed a simple set of health registers and logs over the existing RS-485 network to the plant SCADA.

Within a few months, several strings showed a clear upward trend in ΔV at specific fuse holders despite normal string current. The diagnostic analytics flagged these as degraded conditions with elevated contact resistance. Maintenance crews scheduled targeted inspections during the next planned outage, finding discoloured metal, loose hardware and early signs of overheating at exactly the flagged locations. Cleaning and replacement were performed before any nuisance trips, hot spots or safety incidents occurred, avoiding unplanned downtime during peak summer production.

The plant adopted the same pattern across additional fields, treating ΔV trend monitoring as part of a standard preventive maintenance strategy. The combination of instrumentation amplifiers, comparators, GPIO expansion, MCUs and RS-485 transceivers created a reusable blueprint for corrosion and contact-health monitoring in similar ground-mount installations without redesigning the entire protection scheme.

Rooftop system: half-damaged fuse revealed by per-string ΔV and event logging

A commercial rooftop system with multiple small combiners reported intermittent underperformance on one inverter input. Array-level monitoring suggested that one section was consistently a few percent low, but rooftop access was limited and manual inspections rarely coincided with the anomaly. The fuses protecting individual strings were suspected, yet no complete fuse opens or obvious hot spots were observed during spot checks.

The site operator implemented a lightweight upgrade: per-string ΔV sensing across each fuse using small shunts and zero-drift amplifiers, feeding window comparators that distinguished between normal, elevated and near-zero voltage drops. Comparator outputs were latched and connected through an I²C GPIO expander to a compact MCU inside the combiner. When any string crossed its warning or degraded threshold for long enough, the MCU logged the event with a timestamp and asserted an interrupt towards an existing data logger via a simple dry-contact interface.

Over several weeks, one string began to accumulate repeated degraded events. Its ΔV pattern alternated between elevated and nearly normal as temperature and loading changed, consistent with a fuse element that had partially melted or a weakened contact in the holder. Guided by the event history, the maintenance team inspected that specific fuse position and found clear evidence of a half-damaged fuse and heat-discoloured plastic. Replacement restored string current to match its peers, and the periodic warnings disappeared. The case demonstrated that per-string ΔV plus event logging can expose intermittent fuse problems that would otherwise remain hidden until a complete open or more serious failure occurred.

Brownfield retrofit: from bus-only metering to ΔV sensing and remote maintenance

A fleet of older PV plants relied solely on bus shunts at each combiner to provide current information to a central data concentrator. The operators wanted better visibility into string-level issues but faced tight retrofit budgets and limited downtime. Complete replacement of combiner hardware was not acceptable, so the upgrade strategy had to reuse existing enclosures, power supplies and RS-485 wiring wherever possible.

The retrofit design added ΔV sensing on a subset of critical strings in each combiner using instrumentation amplifiers and compact multi-channel comparators. A small low-power MCU collected ΔV measurements and existing bus-current data, applied the same threshold and time-over-threshold logic used in newer plants, and generated simple trip, warning and degraded status bits. Digital isolators and isolated RS-485 transceivers interfaced this new controller with the legacy data concentrator, which only required minor firmware updates to interpret a few additional registers and event codes.

After deployment, the brownfield sites gained the ability to flag suspect strings and contact points remotely, without opening every combiner for inspection. Maintenance teams received periodic reports highlighting which combiners had strings with low health scores or rising ΔV trends, and could plan visits accordingly. The same IC building blocks as in new designs—current-sense AFEs, precision ΔV amplifiers, comparators, GPIO expanders, MCUs and isolated communication transceivers—were reused in a cost-aware way, extending the life and maintainability of the existing assets without a full redesign.

FAQs about combiner fuse and contact monitoring

1. When is per-string fuse/ΔV monitoring needed instead of just a single bus shunt?

Per-string fuse and ΔV monitoring adds value when string count is high, when soiling or shading varies across the field, or when long cable runs and ageing connectors increase contact-resistance risk. A single bus shunt reports only aggregate current and often hides partial fuse damage or localized connection issues that degrade yield or create hot spots over time.

2. How small a voltage drop across a fuse or terminal is still worth measuring in a PV combiner?

In many PV combiners, tens of millivolts across a fuse or terminal already indicate meaningful power loss and temperature rise. Monitoring ΔV in roughly the 20–200 mV range allows detection of abnormal resistance growth while keeping conduction losses acceptable. The exact threshold depends on string current, efficiency targets, amplifier gain, ADC resolution and expected noise levels.

3. How can a half-blown fuse be distinguished from normal irradiance changes using ΔV and current?

A half-blown fuse typically shows elevated ΔV at similar or only slightly reduced current compared with peer strings, and the ΔV/I ratio drifts upward over time. Normal irradiance changes move all strings together. Comparing each string with the combiner median, analysing trends over days and applying time-over-threshold filters helps separate fuse degradation from routine cloud and MPPT behaviour.

4. What response time do combiner overcurrent comparators need compared with DC breakers and arc-fault protection?

Comparator response time should complement, not compete with, primary DC breakers and dedicated arc-fault devices. Microsecond to low-millisecond comparators are usually fast enough to detect sustained overcurrent while allowing breakers and AFCI units to handle sub-millisecond faults. Time-over-threshold and coordination studies ensure combiner alarms support protection selectivity rather than causing premature trips or conflicting actions.

5. How many channels of ΔV sensing make sense for a 12 or 24 string combiner, and where should they be placed?

For 12–24 string combiners, full per-string ΔV sensing gives the best diagnostic coverage, especially in new builds. In retrofits or cost-sensitive designs, monitoring high-risk strings and representative positions near long cable runs or exposed terminations still reveals many issues. Channels should target fuse holders and terminal blocks where corrosion, loosening and heating are most likely to occur.

6. Should combiner events be latched locally or only reported via SCADA?

Local latching preserves short-lived or intermittent faults that might be missed by slow SCADA polling. Events such as fuse opens, severe ΔV excursions and repeated degraded conditions benefit from latches and clear reset procedures. Higher-level SCADA still aggregates events and trends, but local latching ensures that evidence of dangerous or rare conditions is not silently cleared or overwritten.

7. How can nuisance trips from clouds and MPPT dynamics be avoided when setting thresholds and filters?

Avoiding nuisance trips requires combining amplitude thresholds with hysteresis and time-over-threshold filters tuned to plant behaviour. Thresholds should be set from data across many irradiance and MPPT conditions, comparing strings within the same combiner. Events are raised only when deviations persist beyond a defined window and exceed peer differences, not whenever short cloud transients or tracker movements occur.

8. Can the same shunt be reused for both metering and protection thresholds, and what are the pitfalls?

A single shunt can often serve both metering and comparator thresholds, provided bandwidth, accuracy and layout meet both needs. The main pitfalls are excessive filtering that slows protection response, or shared traces that inject switching noise into metering ADCs. Separate sense traces, careful filtering placement and clear requirements for each function reduce conflicts and preserve selectivity.

9. What isolation strategy is recommended between ΔV AFEs and the combiner communication or SCADA interface?

ΔV AFEs typically sit close to high-voltage conductors, while communication ports connect to low-voltage control networks. Digital isolators or isolated RS-485 and Ethernet transceivers create a clear boundary between these domains. Isolation ratings, creepage and surge performance should match system voltage, insulation coordination and lightning-protection design to prevent ground-reference shifts from damaging sensitive interface and controller circuitry.

10. How do combiner fuse monitoring requirements change for 1500 V PV systems compared with 1000 V?

At 1500 V, higher insulation requirements, larger fault energy and greater string lengths increase the importance of robust fuse selection and contact monitoring. ΔV sensing must respect larger creepage distances and layout constraints, and isolators and transceivers must support higher surge levels. The basic principles remain the same, but voltage ratings, spacing and qualification margins become more demanding.

11. When is it worth retrofitting ΔV monitoring into existing combiner boxes?

Retrofitting ΔV monitoring makes sense when a fleet shows frequent connection issues, unexplained production losses or high manual inspection costs, and when existing combiners have space, power and communication to support extra electronics. Plants with long remaining lifetime, strong O&M obligations and available RS-485 or Ethernet links benefit most, especially if upgrades can be standardised across multiple sites.

12. How should combiner events be logged and time-stamped to support warranty claims and O&M contracts?

Combiner logs should record event type, string ID, ΔV and current snapshots, plus a reliable timestamp. Time can come from a local RTC periodically synchronised to SCADA or from upstream time-stamped commands. Storing a rolling buffer of significant events and exporting them in a consistent format gives asset owners traceable evidence for warranty discussions and O&M performance reviews.