How to Reduce PV Module Mismatch Losses | String Sorting, Shading Maps, Inverter Design

Across 12 commercial and industrial rooftop projects we reviewed, the median annual system mismatch loss reached 7.8%, with the worst case reaching 14.3%.

In this article, "system mismatch loss" means the recoverable or partly recoverable gap between the energy that individual modules or strings could produce at their own optimum operating points and the energy they actually deliver after being connected to the same string or MPPT channel.

This should not be confused with direct irradiance loss, inverter conversion loss, cable loss, curtailment, availability loss, or soiling loss. These categories need a clear monitoring boundary and a consistent calculation method, because PV performance evaluation depends on measured irradiance, module temperature, system configuration, and actual operating data[1].

Mismatch loss builds up from current inconsistency inside strings, voltage differences between parallel strings, non-uniform irradiance, different module temperatures, shading differences, and inverter MPPT configuration.

A shaded module first loses the sunlight that no longer reaches its cells. It can then create extra electrical mismatch if the affected module, sub-string, or string is forced away from its own maximum power point.

Module sorting cannot recover missing sunlight, but string layout and MPPT allocation can reduce part of the additional electrical mismatch.

For arrays with relatively consistent orientation, limited permanent shading, suitable MPPT allocation, and reliable commissioning data, a controllable mismatch-loss target below 3% is a practical engineering goal. Complex multi-orientation rooftops, irregular roofs, bifacial arrays with uneven rear irradiance, and sites with unavoidable permanent shading may have a different technical and economic optimum.

The 12 rooftop projects are a subset of the 38 commissioned projects reviewed for this article. The records include rooftop, ground-mount, and agrivoltaic systems, so the case percentages below should be read as project observations rather than universal guarantees.

String Sorting

Match Current Levels

The first principle of string sorting is to place electrically similar modules into the same series string. Modules from the same model, compatible electrical revision, same power bin, and similar current bin should be grouped together.

In our projects, we use a 2% Isc deviation as an initial screening threshold. For series-connected modules, Impp is more important than Isc because every module in the same string must carry essentially the same operating current.

Isc is still useful for detecting obvious outliers, damage, contamination, or measurement problems. However, module voltage and current both change with irradiance and temperature, so field measurements should be corrected or normalised before comparison[2].

In a 50MW ground-mount plant in Ningxia, re-binning modules by measured electrical characteristics reduced the measured string-level mismatch loss from 5.6% to 2.1%. Daily array output rose by about 3.2% under comparable operating conditions.

l Review factory flash-test data.

l Bin modules by measured Impp and Pmax.

l Verify short-circuit current deviation within approximately ±1.5% under comparable irradiance and temperature.

l Assemble strings using modules with similar Impp, Vmp, Pmax, and temperature behaviour.

l Correct field measurements before comparing strings measured at different times.

Modules from the same power bin usually have better consistency than random combinations, but the same power rating does not guarantee the same current. Two modules can have similar Pmax while one reaches it through higher current and the other through higher voltage.

Measured electrical characteristics should therefore take priority over batch identity or power-bin labels alone.

Where mixed bins already exist on site, use string-level monitoring to flag strings whose current remains materially below comparable strings after correcting for orientation, irradiance, temperature, and shading.

An Isc or operating-current deviation above 3% can be used as a diagnostic trigger, but it should not automatically lead to module reshuffling.

l Exclude soiling.

l Check blown fuses.

l Check connector resistance.

l Check damaged bypass diodes.

l Check module cracks.

l Check sensor errors.

l Check temporary shading.

If a persistent electrical difference remains, the affected modules can be reallocated to compatible strings or moved onto a more suitable inverter channel.

For long procurement cycles with multiple delivery batches, review 100% of available factory flash-test records on arrival and then carry out statistically meaningful on-site verification.

On one 4.2MW rooftop project, the second delivery batch measured 4.3% higher in Isc under corrected comparison conditions. Without re-binning, the modelled string mismatch loss would have reached approximately 6.8%.

After batch-segregated wiring and reallocation by measured electrical characteristics, the mismatch loss returned to 2.4%. The weather-normalised annual-output difference was about 4.1% compared with the original mixed-batch arrangement.

For inter-batch I-V curve testing, use a calibrated solar simulator suitable for the required measurement uncertainty. Solar simulator performance should be evaluated by spectral match, spatial non-uniformity, and temporal instability rather than by the "AAA-class" label alone[3].

A 1% spatial non-uniformity target is a stringent engineering specification. The final acceptance requirement should follow the project specification, applicable test standard, module type, and measurement uncertainty.

For an operating plant, module reshuffling should be justified by recoverable annual energy rather than by electrical variation alone. If the expected gain is smaller than the labour cost, outage cost, access cost, connector replacement risk, and possible warranty impact, continued monitoring may be more economical.

Check Power Tolerance

Power tolerance is the allowed deviation between nameplate output and actual flash-tested output. Common nameplate bands include 0~+3% and 0~+5%.

Power tolerance is important for procurement and quality control, but it is not the same as electrical mismatch. Actual string loss depends on how the power difference is divided between current and voltage, how many modules are connected in series, the shape of each I-V curve, operating temperature, irradiance distribution, and the inverter operating point.

In our internal comparisons, same-band mixing was commonly around 2% mismatch, while poorly controlled cross-band mixing reached 4.5% in some arrays.

A cross-band string can perform well when Impp and Vmp remain compatible. A nominally same-band string can still perform poorly if its modules have materially different current characteristics.

We recommend reviewing 100% of available flash-test reports. Where the sampling plan supports it, sort modules into narrow measured-output bands such as ±2W.

l Sort by measured Pmax together with Impp and Vmp.

l Use on-site verification when delivery batches are mixed.

l Expand the test scope if abnormal variation is detected.

l Distribute samples across pallets, production dates, and delivery lots.

On a 30MW high-altitude project in Tibet, 4.7% of tested modules showed measured power deviation beyond ±3W. After re-binning, weather-normalised annual output exceeded the original nameplate-based model estimate by 1.8%.

This result should not be attributed to re-binning alone. Actual irradiation, operating temperature, availability, bifacial gain, and conservative model assumptions can also cause measured energy to exceed the initial estimate.

Power tolerance also interacts with temperature coefficients. Modules with more negative Pmax temperature coefficients lose more power as cell temperature rises, which can create divergence between modules that appeared similar under standard test conditions.

Relative power divergence ≈ difference in Pmax temperature coefficient × (cell temperature − 25°C).

For example, if two module groups differ by 0.06%/°C and operate at a cell temperature of 65°C, the temperature-related divergence relative to STC is approximately 2.4 percentage points.

The final string mismatch will not necessarily equal that number because the modules still interact through their complete I-V curves.

On an 8MW project in Xishuangbanna, Yunnan, hot-spot imaging and I-V analysis on the same string showed that mixing modules internally classified as Class B, with a -0.38%/°C Pmax temperature coefficient, and Class A, with a -0.32%/°C coefficient, raised midday string mismatch loss from 2% to 5.3%.

"Class A" and "Class B" were project-specific sorting labels, not universal IEC quality grades. The 5.3% result included coefficient difference, module operating temperature, electrical dispersion, and site conditions.

TOPCon and PERC mixing deserves special attention because a representative Pmax temperature-coefficient gap of around 0.04%/°C can create about 1.8 percentage points of relative divergence at a 45°C rise above STC.

A project-specific loss approaching 4% can occur when temperature-coefficient divergence combines with different degradation states, electrical characteristics, irradiance conditions, and operating-voltage behaviour.

Modules from different cell technologies should not be mixed only because their nameplate wattage is similar.

l Compare Voc.

l Compare Vmp.

l Compare Isc.

l Compare Impp.

l Compare temperature coefficients.

l Compare maximum system voltage.

l Compare dimensions.

l Compare bypass-diode layout.

l Compare connector type.

l Compare degradation assumptions.

l Compare warranty conditions.

Group Similar Modules

In a 6MW same-model array comparison, after 4 years the mixed-bin array lagged the sorted array by 6.2% in annual output.

Similar grouping covers current, power, operating voltage, degradation rate, temperature coefficient, bifacial behaviour, and the physical condition of each module.

The 6.2% difference was a site-specific result. It included long-term electrical dispersion and operating-condition differences, so it should not be treated as the expected result for every mixed-bin system.

Modules with similar degradation curves are more likely to keep their electrical output aligned over the long run. Individual module degradation can create string-level mismatch because modules in the same series string do not necessarily age in the same way[4].

Representative project assumptions sometimes use approximately 1% first-year degradation for selected TOPCon products and approximately 2% for selected PERC products, but these values are not universal characteristics of the technologies.

Actual degradation depends on the specific product, wafer type, encapsulation, LID or LeTID susceptibility, installation environment, manufacturing quality, and warranty profile.

In our five-year comparison model, mixing module groups with the assumed 1% and 2% first-year degradation profiles increased mismatch loss by 3.8 percentage points compared with same-curve combinations.

The operational approach is to register modules by serial number, batch, model, electrical revision, and location before string assembly. The traceability record should show which modules were assigned to each string and MPPT.

Modules can then be scanned with an EL tester before string assembly where project risk, logistics, and quality requirements justify it. The acceptance decision should consider whether a crack interrupts current-carrying paths, creates an electrically isolated region, affects a large proportion of the cell, or appears together with abnormal I-V or infrared behaviour.

EL test current, exposure time, camera sensitivity, ambient-light control, image processing, and acceptance criteria should follow the module and EL equipment procedure rather than a universal multiple of Isc.

EL images should be interpreted together with flash-test, I-V, visual, and thermal data.

When we mixed high-efficiency n-type modules with PERC modules, we ran a dedicated 6-month pilot array. We tracked the I-V curve, module temperature, irradiance, and normalised yield to confirm degradation and operating-voltage slopes before scaling up.

The pilot cost about 0.8% of the project budget but brought the modelled lifecycle mismatch loss from 4.5% back to 1.8%.

A pilot array should use a control group with the same orientation, tilt, cleaning schedule, inverter configuration, cable length, and monitoring quality. Otherwise, an apparent technology difference may actually come from weather exposure, rear-side irradiance, soiling, or inverter behaviour.

Shading Maps

Find Shading Sources

Across the 38 projects we commissioned, rooftop parapets, distribution boxes, nearby trees, and utility poles were the four most common shading sources. Their typical heights ranged from approximately 0.3m to 3.5m.

Shadow length should not be estimated by a fixed multiplier alone. It is governed by obstacle height and solar elevation angle, so the same obstacle can cast a short shadow at high sun angles and a much longer shadow in winter mornings and afternoons.

Shadow length = obstacle height ÷ tan(solar elevation angle).

The observed range of around 1.5-2.5× obstacle height is useful only as a quick screening range under specific solar-elevation conditions. The design calculation should use actual coordinates and time-step solar geometry.

A 10% shaded area does not necessarily cause only a 10% power loss. Partial shading can reduce current in a cell group, activate bypass diodes, introduce non-linear electrical loss, and move the affected string away from the operating voltage of neighbouring strings[2].

If shading falls across a bypass-diode coverage zone, the electrical loss can extend beyond the directly shaded cell area. In some of our cases, total shading-related generation loss reached 25-40% in the affected zones.

The number of cells protected by each bypass diode cannot be generalised as 18-24 cells for all modern modules. Traditional full-cell modules, half-cut modules, multi-busbar products, shingled modules, and different junction-box designs can have different internal circuits.

Before modelling partial shading, the designer should confirm the specific module's cell layout, half-cell parallel branches, bypass-diode count, and diode-protection zones.

Shadow direction is as important as shaded area. A narrow shadow crossing many electrical sub-strings can cause more loss than a larger shadow confined to one electrically isolated section.

The fastest way to identify shading sources is to combine aerial imagery, a three-dimensional site model, solar-trajectory modelling, and operating data.

l Capture roof geometry and surrounding obstacles.

l Project the array plane into a solar-design model.

l Calculate annual geometric shadow hours and irradiance-weighted shading loss.

l Connect the model to the actual module, string, bypass-diode, and MPPT layout.

l Separate direct irradiance loss from additional electrical mismatch.

On a 2.8MW rooftop project in Shanghai, parapet walls caused up to 4.2 hours of local shadow on the winter solstice and contributed to approximately 8% total annual energy loss in the affected zones.

After reorienting selected module rows and changing the electrical grouping, the critical shadow time fell to 1.5 hours and the measured total loss in those zones dropped back to approximately 2.5%.

We recommend a 4K survey at an altitude that provides safe and sufficient ground resolution. On many large sites this may be around 80-120m, subject to local flight rules, roof geometry, safety constraints, and image-detail requirements.

Flights between 10:00 and 14:00 local apparent solar time can document roof geometry and high-irradiance shading conditions, but they should not be the only survey window.

Low-sun morning and afternoon observations, especially in winter, are essential for identifying the longest and most economically significant shadows.

In our 38 projects, 76% of shading issues were identified through combined aerial review, field observation, and operating-data comparison. The drone image alone did not establish annual loss; it provided geometry that was evaluated through solar-path modelling and checked against string-level power data.

Fixed cameras with solar-trajectory software can also record shadow boundaries throughout the year. Continuous observation is useful where tree growth, new construction, rooftop equipment, temporary storage, safety rails, or advertising structures may introduce new shading after commissioning.

The operations team should link image timestamps to inverter and irradiance data. This helps confirm whether a power reduction is caused by shading, an electrical fault, or a weather transient.

Track Sunlight Hours

Sunlight hours is a useful first-stage indicator of shading exposure, but it is not sufficient on its own.

One hour of shading near sunrise usually removes less energy than one hour of shading during a high-irradiance period.

The more useful metric is irradiance-weighted shading loss.

This means the amount of plane-of-array solar energy blocked during each interval, together with the additional electrical mismatch caused by the array topology. Plane-of-array irradiance and module temperature are central inputs in PV system performance monitoring[5].

Use a sunshine recorder, calibrated irradiance sensors, or solar-path modelling software to measure or model annual sunlight exposure and irradiance for each array zone.

The analysis should identify "low sunlight zones", but it should also compare the timing of shade with potential unshaded production.

Zones that fall more than 15% below the field average are useful investigation triggers, not automatic instructions to rebuild the array.

On a 6MW commercial rooftop project in Tangshan, Hebei, we identified 18% of array zones with modelled sunlight hours 25% below the field mean.

Without intervention, those modules were expected to reduce the affected MPPT yield by approximately 4-7 percentage points. The effect depended on whether the shaded modules were concentrated in a separate electrical group or distributed across multiple otherwise unshaded strings.

Before selecting module-level power optimisers, compare four options:

Change the physical layout.

l Reallocate strings to independent MPPTs.

l Shorten or separate affected strings where allowed.

l Install module-level electronics.

Optimisers may recover energy under non-uniform conditions, but they add equipment cost, rooftop components, communication requirements, conversion loss, and additional potential failure points.

They should be selected through lifecycle cost-benefit analysis rather than used as an automatic response to every low-sunlight zone.

Sunlight tracking also uncovers seasonal shading: zones that are shade-free in summer but heavily shaded in winter.

For fixed-tilt systems, inter-row spacing should be checked against low winter solar elevation rather than assessed only under summer conditions.

Designing for zero winter-solstice shading is not always the economic optimum on a space-constrained rooftop because wider spacing may reduce installed capacity.

In parts of North China, a representative winter-solstice solar elevation around the design period may fall near 22°, although the exact value depends on latitude, date, and time. At this angle, an obstacle can cast a shadow roughly 2.5 times its height.

At still lower morning or afternoon elevations, the shadow becomes much longer. The design calculation should therefore use site-specific coordinates and time-step solar geometry, such as the approach used in established solar-resource tools[6].

On one 5.4MW project, spacing designed mainly around summer conditions cost an extra 5.7% of winter output in the affected area.

Switching to a winter-aware arrangement recovered much of the loss. The final decision balanced winter shading, annual irradiance, installed capacity, maintenance access, roof loading, fire routes, cable length, and project economics.

We recommend using at least 30-minute annual shadow-factor sampling for broad shading studies. Where narrow pole or cable shadows matter, a finer time step may be required because a 30-minute average can smooth out fast-moving shadows.

The model should be calibrated against at least one period of measured site data before it is used to support an annual-loss claim.

Plan String Layout

On a 12MW agrivoltaic project, grouping the 24 modules affected by a similar tree-shadow pattern into one controlled electrical group cut total annual shading-related generation loss from 7.3% to 3.9%, a 3.4 percentage-point improvement.

The successful design did not rely only on putting shaded modules into a single series string. It also considered MPPT allocation, string voltage, bypass-diode behaviour, and the irradiance pattern of parallel strings.

The goal of string layout planning is to place modules with similar orientation, irradiance, temperature, and electrical behaviour in the same controllable electrical group.

In some projects, concentrating modules under the same shading source into one string or one MPPT limits the number of affected electrical groups.

In other projects, placing a heavily shaded string in parallel with unshaded strings on the same MPPT can create a large operating-voltage mismatch.

The correct arrangement must therefore be verified through electrical shading simulation.

Concentrated shading still costs direct irradiance. If bypass diodes activate, the affected string's optimum operating voltage can fall.

When that string remains in parallel with unshaded strings, all strings on the MPPT must operate at a common voltage, which can create string-to-string voltage mismatch.

In our 38 projects, a shading-aware concentrated layout brought additional electrical mismatch loss on selected shaded electrical groups down from an average 6.8% to within 2.5%.

These results were achieved where the affected modules shared similar shading schedules and were allocated to suitable MPPT channels.

String layout planning must also account for module orientation differences. Mixing east-facing and west-facing modules within the same series string should be avoided because the modules can receive substantially different irradiance at the same time and are forced to carry the same current.

Different east-facing and west-facing strings should preferably use independent MPPT channels. If they must share an MPPT, the designer should verify that the strings have the same module count, compatible Vmp behaviour, comparable temperature exposure, and a combined I-V curve that the inverter can track effectively.

It is also important to avoid mixing persistently low-irradiance zones with normal-irradiance modules in the same series string.

Where possible, concentrate affected modules at controlled array edges or allocate them to a separate MPPT.

For bifacial modules, string layout should also reflect rear-side irradiance. Modules above bright roof zones, open ground, dark membranes, cable trays, vegetation, or structural beams can receive different rear irradiance even when front-side sunlight is uniform.

A front-side shading map alone may therefore underestimate bifacial mismatch.

Inverter Design

Use More MPPTs

Across 22 projects we tracked, increasing MPPT count from 2 to 4 lifted average system output by 3.8%, reaching 7.2% on heavily shaded or multi-orientation sites.

The benefit came from separating strings with different orientations, shading patterns, temperatures, or optimum operating voltages. Doubling MPPT count does not automatically improve output if all strings already behave similarly.

Each MPPT channel independently controls the operating voltage of its assigned string group. Series-connected modules within one string operate at essentially the same current, while parallel strings connected to the same MPPT operate at essentially the same voltage.

This distinction matters. A weak-current module can constrain a series string, while parallel-string mismatch is mainly created when strings prefer materially different operating voltages.

When strings with different orientations or shading conditions share one MPPT, their combined I-V curve may contain multiple local power peaks or a broad unstable optimum. The inverter must then select a common operating voltage for all parallel strings.

Partial shading can create non-linear array behaviour, so global maximum power point tracking becomes important under non-uniform conditions[7].

Depending on topology, severe orientation or shading mismatch can create an additional MPPT and mismatch loss of 5-15%.

l Uniform ground-mount arrays may gain little from extra MPPTs.

l Irregular rooftops with several orientations may gain more.

l Heavily shaded zones should not be paralleled with clean zones unless the combined I-V curve is verified.

l Physical DC inputs should not be mistaken for independent MPPT channels.

MPPT selection should account for tracking accuracy, global maximum power point search capability, and dynamic response. MPPT algorithms differ in speed, complexity, and tracking behaviour under changing irradiance[8].

In our tests, response times around 2-3 seconds were associated with temporary 2-4% power deviation during selected transients. Faster systems responded in under 0.5 second and kept temporary deviation below 0.5%.

These results depend on the test definition, measurement interval, irradiance ramp, firmware version, inverter model, and array configuration.

For cloudy and multi-orientation regions, at least 4 MPPT channels may be a useful starting specification when the roof geometry supports four distinct electrical zones.

A response target below 1 second can also be included where dynamic performance is important. In our field comparison, this configuration reduced measured cloud-transition-related loss from 3.2% to 0.9%.

The result should be understood as a comparison of complete inverter and array configurations, not as proof that MPPT count alone caused the full difference.

Select the Right Strings

The pairing of inverter and string is not simply a sum of rated powers.

The design must consider the inverter's absolute maximum DC input voltage, MPPT operating-voltage range, startup voltage, full-power MPPT range, maximum operating current per MPPT, maximum short-circuit current, number of physical inputs, and the module's temperature-dependent voltage and current.

String open-circuit voltage rises as module temperature falls. The increase at -10°C may be about 11% for a particular module with a corresponding Voc temperature coefficient, but it is not a universal value.

Voc at minimum temperature = Voc at STC × [1 + βVoc × (minimum cell temperature − 25°C)].

The calculation should use the selected module's actual datasheet coefficient and an appropriate design minimum temperature for the site. It should not automatically use -10°C.

If calculated cold-condition string Voc exceeds the inverter's absolute maximum DC input voltage, the design is unacceptable. This can create equipment damage, safety, and warranty risks.

If string voltage remains below the absolute maximum but outside the normal MPPT range, the inverter may fail to track correctly during part of the operating period.

On a 20MW project in Xinjiang, winter string Voc approached or exceeded the intended inverter design ceiling and contributed to alarms, unavailable periods, and approximately 4.8% annual output loss.

String count should also match the inverter's independent MPPT channels and current capability.

For complex rooftops, one string per MPPT provides the highest level of independent control when inverter architecture and project economics allow it.

Two parallel strings per MPPT are also common when they have the same module count, orientation, tilt, technology, and shading condition.

More than two parallel strings may be technically acceptable on some inverter models, but only when the input count, maximum operating current, maximum short-circuit current, connector rating, cable protection, and reverse-current design permit it.

A large-format bifacial module may, for example, be configured at approximately 28 modules per string in a suitable large ground-mount design, but only after cold-condition Voc and hot-condition Vmp are checked against the selected inverter and local design rules.

This must not be confused with 28 strings per MPPT. The number of modules in series depends on voltage limits, while the number of parallel strings per MPPT depends on inverter current and input limits.

On distributed rooftops, fewer parallel strings per MPPT can provide finer mismatch control than a heavily paralleled arrangement.

However, the design should not be expressed as a fixed rule such as "drop from 28 strings to 14 strings per MPPT" without a specific inverter. The correct arrangement must be calculated from inverter architecture and string electrical conditions.

String selection should also factor in maximum operating current and maximum short-circuit current.

A selected 100kW inverter may accept approximately 26A operating current per MPPT, but other 100kW models can have materially different limits. The exact datasheet must control the design.

Module operating current, including expected bifacial gain and high-irradiance conditions, should remain below the inverter's maximum usable MPPT current.

Separately, design short-circuit current, including the applicable safety or code factor such as 1.25 where required, should remain below the inverter's maximum short-circuit-current limit and the ratings of cables, connectors, fuses, and switchgear.

Operating-current limits and short-circuit-current limits are different checks and should not be merged into one comparison.

For bifacial projects, rear-side gain can increase current above the front-side STC value. The design should therefore use an appropriate bifacial current-gain assumption instead of checking only nameplate front-side Isc.

High albedo, cold clear conditions, cloud-edge enhancement, or reflective roof surfaces can temporarily increase current and cause clipping or MPPT current limiting even when voltage design is correct.

Limit Voltage Difference

Inter-string voltage difference is an important diagnostic signature of mismatch, but it must be interpreted correctly.

Parallel strings connected to one MPPT operate at a common terminal voltage during normal operation.

The relevant mismatch is the difference between the voltage each string would prefer at its own maximum power point and the common voltage imposed by the MPPT.

A measured or modelled preferred-voltage difference above 5% should trigger investigation. In our internal cases, differences above 10% were associated with approximately 3-6% additional loss.

These values are diagnostic project thresholds, not universal inverter derating triggers. Most inverters do not automatically derate simply because an operator records a 5% difference between separately tested string voltages.

During commissioning, measure each string's Voc and compare only strings with the same module count, module type, irradiance exposure, and module temperature.

Holding corrected Voc difference within approximately ±2%, with an investigation cap around ±3%, can help identify wrong module counts, mixed module types, abnormal temperature differences, wiring errors, damaged modules, open bypass paths, or measurement problems.

Voltage difference is driven mainly by inconsistent module counts, module Voc or Vmp spread, module temperature, activated bypass diodes, and different electrical conditions.

DC cable voltage drop is a separate loaded-operation issue. During a true open-circuit Voc measurement, string current is close to zero, so normal cable voltage drop is negligible.

Cable length, conductor cross-section, connector resistance, and terminal quality should instead be assessed under operating current.

Module counts connected in parallel to the same MPPT should normally be equal unless a specifically validated inverter topology permits otherwise.

Loaded cable drop should be addressed through suitable conductor sizing, controlled route length, correctly assembled connectors, tightened terminals, and thermal inspection.

A comparable 50MW project held measured corrected voltage difference to approximately ±1.5%, beating the project's ±2% commissioning baseline. The result depended on consistent string length, module allocation, installation quality, and controlled measurement conditions.

The last lever for limiting voltage-related mismatch is matching the inverter's MPPT voltage range to the string design.

If string Vmp falls below the minimum MPPT voltage during hot or low-irradiance conditions, the inverter may leave its optimal tracking region. If string voltage approaches the upper range in cold conditions, operating margin becomes insufficient.

These effects can contribute to losses of 2-4% in poorly matched systems.

Evaluate inverters with a wide and usable MPPT voltage range, such as 200-1000V or 150-800V where appropriate. A wider range is not automatically better unless it overlaps the string's real operating-voltage envelope.

The design calculation should confirm:

l Cold-condition Voc remains below the inverter's absolute maximum DC voltage.

l Cold-condition operating voltage remains within the upper MPPT range.

l Hot-condition Vmp remains above the minimum MPPT voltage.

l The string remains within the full-power MPPT voltage range at the required AC grid voltage.

l Startup voltage can be reached early enough without excessive morning and evening loss.

l Maximum operating current and maximum short-circuit current remain within inverter limits.

The objective is to keep the inverter in effective MPPT tracking mode for more than 95% of meaningful energy-producing periods rather than 95% of clock hours.

Low-light periods with almost no available solar energy should not be weighted the same as high-irradiance production periods.

In our comparison of 8 mainstream inverters under low-irradiance conditions, wide-voltage-range models achieved 96.8% average daily tracking success, while narrow-voltage-range models managed 88.2%.

In that project dataset, the difference translated to roughly 3% higher annual output.

"Tracking success" means the percentage of qualifying energy-producing intervals during which the inverter remained within the project's accepted deviation from the reference maximum power point.

The figure depends on irradiance threshold, sampling interval, inverter firmware, array configuration, and reference method. It should not be treated as a standardised performance rating unless the same test procedure is used.

When the MPPT voltage range is matched with string count, confirm that the inverter's maximum DC input current per MPPT is not exceeded under realistic high-current conditions.

STC current, applicable safety factor, bifacial gain, high-albedo exposure, cloud-edge enhancement, cable rating, and inverter short-circuit limit should all be checked separately.

Cold clear conditions mainly increase voltage, while high irradiance and rear-side gain increase current. The two design cases should not be merged into one worst-case statement.

A practical commissioning workflow should start with monitoring data rather than immediate module movement.

l Compare normalised string current, MPPT voltage, irradiance, module temperature, and power at the same timestamp.

l Check whether an abnormal string follows a daily shadow pattern or remains low under uniform sunlight.

l Use I-V curve tracing to distinguish current limitation, voltage loss, bypass-diode activation, and series-resistance problems.

l Use infrared imaging to identify hot cells, connectors, fuses, diodes, and junction boxes.

l Use EL inspection when cracks or inactive cell areas are suspected.

Before deciding on corrective action, exclude non-mismatch causes.

l Soiling

l Vegetation

l Snow

l Grid curtailment

l Inverter clipping

l Outages

l Communication gaps

l Sensor drift

l Tracker faults

l Cable losses

l Unavailable modules

Only after the loss mechanism is identified should the project team choose between cleaning, repair, module replacement, string reshuffling, MPPT reassignment, physical layout changes, optimiser installation, inverter replacement, or continued monitoring.

The final design and any retrofit must remain within inverter and module manufacturers' instructions, project warranty conditions, local electrical code, connector ratings, conductor ampacity, overcurrent-protection design, isolation requirements, and safe work procedures.

Strings should not be disconnected, reconfigured, or tested under unsafe live-DC conditions.