Why Does Shunt Resistance Look Normal at STC but Collapse in Low Light

When a module looks fine at 1,000 W/m² but starts losing fill factor or efficiency at 200 W/m², it is easy to blame a falling Rsh. That conclusion can be wrong. At 200 W/m², photocurrent is roughly 20% of its STC level when temperature and spectrum are similar, while module voltage stays well above 20%. Even if the physical shunt path has not changed, leakage can take a much larger share of the power the module has left.

If fitted Rsh does fall sharply as irradiance drops, do not treat the number as proof of a physical change. Check the short-circuit-side I–V slope, repeat the measurement at several irradiance levels, verify temperature and illumination accuracy, and compare dark I–V leakage. Common PV models can actually predict higher Rsh at lower irradiance, so a measured “collapse” needs to be verified before it is accepted as a real material change.^[2]

Separate Physical Rsh from Fitted Rsh

Rsh is often discussed as if it were one resistor inside the module. In practice, the number can mean different things.

In the standard single-diode model, the shunt term is:

Ish = (V + I × Rs) / Rsh

Total current is:

I = IL - ID - Ish

IL is light-generated current, ID is diode current, and Ish is shunt current.^[1]

• Physical leakage: a real unwanted current path inside the cell.
• Model Rsh: the resistance term used by a mathematical model.
• Extracted Rsh: the value calculated from a measured I–V curve.

These three values are related, but they are not interchangeable. A fitted Rsh is not the same as putting an ohmmeter across a physical resistor inside the module.

The difference matters in low-light testing. The De Soto model makes Rsh increase as absorbed irradiance decreases.^[2] The PVsyst model also changes Rsh with irradiance instead of using one fixed value at every light level.^[3]

So if software reports 2,000 Ω at STC and 300 Ω at 100 W/m², the material has not necessarily become nearly seven times more conductive. The result may come from voltage-dependent leakage, a local junction defect, low-current measurement noise, temperature error, uneven light, or unstable parameter fitting.

Compare 1,000, 200, and 100 W/m²

The first thing to look at is how much photocurrent is left as irradiance drops. When temperature and spectrum are similar, photocurrent changes approximately in proportion to irradiance:

Iph ∝ G

Voltage does not fall at the same rate. Solar-cell voltage has an approximately logarithmic relationship with light-generated current:

Voc ≈ nVT × ln(IL / I0 + 1)

So reducing irradiance from 1,000 W/m² to 200 W/m² can cut photocurrent to roughly 20%, while voltage stays far above 20% of its STC value. That difference is what makes a finite shunt much harder to ignore in low light.

Take a simplified module-level Rsh of 500 Ω:

These are illustrative calculations, not measured module data or a full single-diode simulation. Rsh stays fixed at 500 Ω in all three cases, yet the estimated loss rises from 0.75% to 5.1% of available power. The reason is straightforward: useful power falls much faster than voltage.

Read the Short-Circuit Side of the I–V Curve

For a suspected shunt problem, the I–V curve is more useful than a single Rsh number. IEC 60904-1 defines procedures for measuring photovoltaic I–V characteristics.^[5]

• Near Isc: a larger slope suggests stronger effective shunt leakage.
• At the knee: a rounded knee reduces fill factor and Pmax.
• Near Voc: diode and recombination behavior becomes more important.

A local estimate is:

Rsh ≈ -(dV / dI) near V = 0

Do not expect two test systems to return exactly the same Rsh if they use different voltage ranges, point counts, smoothing methods, or instrument resolution. For production control or supplier comparison, consistency of the test method matters as much as the number itself.

Calculate the Fill-Factor Loss

Voc and Isc can still look normal even when useful output has already dropped. Fill factor shows the change more clearly:

FF = (Vmp × Imp) / (Voc × Isc)

Assume:

• Voc = 50 V;
• Isc = 10 A;
• Voc × Isc = 500 W.

With Voc and Isc unchanged, a drop in FF from 80% to 70% cuts Pmax from 400 W to 350 W. That is a 12.5% loss in maximum power. For anyone evaluating low-light quality, this is why FF deserves more attention than Voc or Isc alone.

Separate Shunt Loss from Series Resistance

Series resistance behaves differently because its power loss rises with the square of current:

PRs ≈ I² × Rs

For a hypothetical effective Rs of 0.20 Ω:

High Rs is easier to expose when current is high. Finite Rsh is often easier to expose when useful current is low.

For incoming inspection or failure analysis, use these patterns to decide what to test next. They are not universal pass/fail rules, and more than one loss mechanism can exist in the same module.

Rule Out Recombination

A module can lose low-light performance without having a shunt problem. Recombination is one common reason.

With a shunt, current takes an unwanted electrical path. With recombination, light-generated electrons and holes disappear before they are collected.

In the single-diode model, the diode-current term represents voltage-dependent diode and recombination behavior, while Ish represents the shunt path.^[1]

A larger Isc-side slope points more directly toward a shunt. If Voc and FF both fall strongly, diode or recombination behavior should also be checked before the module is classified as a low-Rsh failure.

Rule Out Partial Shading

Uniform low irradiance and partial shading should not be mixed together during diagnosis.

At uniform 200 W/m², every cell receives roughly the same reduced light. Under partial shading, some cells generate less current than others, while a series string still forces them to carry the same string current.

Strong partial shading can produce:

• steps in the I–V curve;
• multiple knees;
• bypass-diode operation;
• large mismatch losses.

If those features are present, lowering Rsh in a model is not a useful explanation.

The same principle applies in the field. If module-level I–V looks normal but cloudy-day AC output is weak, check inverter startup voltage, MPPT behavior, low-load efficiency, DC wiring, soiling, and shading before blaming the module.

Identify Defects That Can Create Real Leakage

Once the electrical pattern points toward leakage, the next step is finding the physical cause.

IEA PVPS treats cracking, PID, corrosion, and electrical degradation as separate failure modes because a similar power-loss symptom can come from different electrical causes.^[8]

Test at Six Irradiance Levels

For QA, supplier comparison, or failure analysis, STC plus one low-light point is too little information. A six-point sweep shows whether the change is gradual, sudden, or inconsistent.

IEC 61853-1 covers module power measurements across multiple irradiance and temperature conditions instead of limiting evaluation to STC.^[9]

For repeatability, use the same test point twice rather than trusting the first result:

1. measure at 1,000 W/m²;
2. reduce irradiance;
3. repeat the low-light measurement;
4. return to 1,000 W/m²;
5. measure again.

If the original curve does not return, investigate temperature drift, simulator drift, measurement instability, or device instability before assigning the difference to Rsh.

Correct Temperature Before Comparing Rsh

A module under 1,000 W/m² can run much hotter than the same module at 200 W/m². If those two curves are compared directly, part of the difference may simply come from temperature.

IEC 60891 provides procedures for correcting measured PV I–V characteristics for differences in irradiance and temperature.^[10]

For low-light efficiency, use temperature-controlled or temperature-corrected values:

Relative efficiency = Efficiency at G / Efficiency at reference condition

These values are calculation examples, not industry acceptance limits.

NREL-associated low-irradiance work used temperature-adjusted electrical data when evaluating normalized module efficiency because temperature must be separated from irradiance effects.^[6]

Check Light Uniformity Before Blaming Rsh

A low-light test is only useful if the whole module receives comparable irradiance.

If one section receives 120 W/m² and another receives 90 W/m², the difference is:

(120 - 90) / 120 = 25%

Those cells naturally generate different photocurrents. Because series-connected cells must operate at one string current, the mismatch can lower FF and distort the I–V curve even when no internal shunt defect exists.

IEC 60904-1 includes spatial nonuniformity among the factors that must be considered in accurate PV I–V measurement.^[5]

Check Spectrum Before Comparing Low-Light Curves

Two tests can both read 200 W/m² and still expose the module to different wavelength distributions.

Changing lamp power, adding filters, or switching simulator modes can alter both irradiance and spectrum. If the reference sensor and the PV module respond differently to those wavelengths, part of the measured change may come from spectral mismatch rather than Rsh.

IEC 60904-3 defines the reference spectral irradiance used for terrestrial PV measurement.^[4]

Check Measurement Error at Low Current

Instrument error that looks insignificant at STC can become large in percentage terms at low current.

For a 10 mA error:

At 10 A, 10 mA is only 0.1%. At 0.2 A, it is already 5%. This is one practical reason low-light Rsh values can jump around even when the module itself has not changed.

Before accepting a large Rsh shift, check current resolution, voltage resolution, irradiance accuracy, temperature, sweep timing, and repeatability.

Confirm Leakage with Dark I–V and Imaging

Once the light I–V data points toward leakage, dark I–V can provide a second electrical check because it removes photocurrent from the measurement.

A stronger shunt diagnosis has at least two independent electrical signs, for example:

• low-light FF drops repeatedly;
• the Isc-side slope becomes larger;
• dark I–V shows higher leakage.

Imaging can then help locate the affected area:

• EL: cracks, inactive areas, and broken current paths;
• IR: unusual local heating;
• PL: material and recombination nonuniformity.

A dark EL region does not automatically prove a shunt, and a hot IR spot does not automatically prove low Rsh. For a failure report, the electrical result and the image should point to the same area and the same mechanism.^[8]

Dark I–V and externally biased module tests can involve hazardous DC voltage and current and should be performed with suitable equipment and qualified electrical procedures.

Do Not Compare Cell Rsh with Module Rsh

Cell-level and module-level Rsh values are not directly interchangeable.

In a series-connected module, current is common to the cells while cell voltages add. Module-level resistance terms therefore depend on the cell connection and model definition.^[1]

For the same reason, cell-level electrical information should not be mixed directly with module-level electrical information.

A localized bad cell can also be masked within an otherwise normal-looking module result. NREL developed a method for measuring shunt resistance in individual series-connected cells because module-level measurements may not expose every local shunt clearly.^[7]

Use Relative Efficiency, Not Raw Low-Light Watts

A 450 W module is not supposed to produce 450 W at 200 W/m². What matters is whether its conversion efficiency falls more than expected after temperature correction.

Consider two hypothetical 450 W modules:

The STC difference is only:

(450 - 449) / 450 = 0.22%

At 200 W/m², the relative-efficiency gap is 8 percentage points. Module B also has lower FF, a larger Isc-side slope, and higher dark leakage. For a buyer, QA team, or failure-analysis engineer, that combination is much more meaningful than the 1 W STC difference.

These values are illustrative and are not presented as product data or universal acceptance limits.

Use Multi-Irradiance Data for Factory and Supplier Checks

For production control, one average STC Pmax can hide a growing number of weak low-light outliers.

Useful factory data include:

• FF at several irradiance levels;
• temperature-corrected relative efficiency;
• Isc-side slope;
• Rsh distribution, not only average Rsh;
• dark I–V for suspect samples;
• EL defect distribution;
• lot-to-lot changes.

The practical question is not only whether the average stays stable. A production line can keep the same average STC power while the share of weak low-light units increases. The lower tail of the distribution may show the problem first.

Tongwei describes its broader process and quality-control framework in the TW Solar Quality White Paper. Exact voltage, current, cell configuration, and rated power should be checked against the relevant module product data before extracted resistance values are compared between products.

For supplier or lot comparison, request:

• multi-irradiance I–V curves;
• FF versus irradiance;
• relative efficiency versus irradiance;
• sample size;
• measurement uncertainty;
• production samples rather than only selected engineering samples;
• independent laboratory results where available.

IEC 61853-1 provides a standardized basis for measuring module performance across different irradiance and temperature conditions.^[9]

Finally

At 200 W/m², photocurrent is roughly 20% of its STC level, while module voltage remains much higher than 20%. That alone can make a finite shunt take a much larger share of available power. A falling fitted Rsh is therefore a result to verify, not proof that the physical resistance collapsed. Use several irradiance points, temperature-corrected data, the Isc-side slope, FF, repeatability, and dark I–V together. When those measurements point in the same direction, the diagnosis is much more reliable than any single Rsh value.