Why Does Module Power Drift During Flash Testing | Stabilization, Heat, Light Soaking

Power drift during flash testing usually comes from temperature, a changing module state, simulator error, scan timing or electrical contact. On a 600 W module, a 5°C rise can change Pmax by about 7.5–10.5 W when the temperature coefficient is between -0.25%/°C and -0.35%/°C. An extra 10 mΩ of resistance at 14 A removes 1.96 W. Check the full I-V curve, stabilize the module temperature and repeat the same test sequence before reporting permanent degradation.

Use the I-V Curve to Find the Cause

IEC 60904-1 defines the measurement of photovoltaic I-V curves under natural or simulated sunlight.^[1] Standard Test Conditions use 1,000 W/m² irradiance, 25°C cell temperature and the IEC reference terrestrial spectrum.^[2]

Correct Temperature Before Comparing Power

The table below shows how a 5°C temperature rise affects a 600 W module. The coefficients are calculation examples; use the coefficient for the exact module model.

IEC 60891 defines the procedures used to translate measured I-V curves between different temperature and irradiance conditions.^[3]

Do not use room temperature as cell temperature. After transport or light soaking, the centre cells may remain warmer than the rear surface, frame and laboratory air. Use the same sensor type, position and attachment method for every comparison. Record both the temperature and its rate of change.

For Tongwei products, confirm the module family on the module product page and use the coefficient in the matching file from the technical document library. A coefficient from another wattage, glass structure or cell technology can produce an incorrect STC correction.

Separate Random Scatter from a Moving Trend

The values below are diagnostic examples, not commercial module test records.

The random series has a range of only 0.4 W, or about 0.07% of 600 W. It moves above and below the average without a clear direction. The second series falls by 4.1 W, or about 0.68%, across five flashes. Do not hide a steady decline inside an average. Check temperature, connection resistance, scan timing and module stabilization first.

Set the Module to a Known State

Two flash results are not comparable when one module was recently illuminated and the other was stored in darkness. Record the following before each test:

IEC 61215-2 defines the test procedures used in terrestrial module design qualification.^[4] IEC 61215-1 states that qualification results are not a numerical prediction of module lifetime.^[5]

Do not stop a stabilization process because the result has moved above a pass limit. Stop only when the stated procedure is complete and Pmax, Voc, Isc and fill factor no longer show an unacceptable trend.

Match Stabilization to the Cell Technology

Tongwei's cell technology overview identifies current TOPCon and HJT routes. The test history must match the actual cell technology, not only the module size or rated wattage.

IEC 61215-1-1 provides technology-specific requirements for crystalline-silicon modules, including control of light-induced effects.^[6]

IEC TS 63342 uses current injection at elevated temperature to detect LETID sensitivity. It does not predict the exact loss that will occur outdoors.^[7] In an international laboratory study, deliberately LETID-sensitive modules lost about 6% of Pmax on average, while other module types showed much smaller or negligible losses. The reported 6% is not a normal loss value for every module.^[8]

A Fraunhofer study found that light soaking at a 25°C module temperature provided repeatable stabilization for tested TOPCon modules after UV exposure and dark storage.^[9]

A silicon heterojunction study reported efficiency gains of up to 0.3 absolute percentage points after light soaking, mainly through higher Voc and fill factor.^[10] An increase from 23.0% to 23.3% is 0.3 absolute percentage points, or about 1.3% relative.

A five-laboratory CIGS study found that light exposure followed by forward bias while the module cooled to 25°C produced more repeatable results than testing immediately after hot exposure and relying on a temperature correction.^[11]

CdTe and CIGS studies also found temporary electrical changes after dark heat, forward bias and light exposure.^[12] For amorphous-silicon modules, controlled tests found that operating temperature and previous exposure history were the main factors affecting stabilized output.^[13]

Record Every Light-Soak Condition

Light dose is irradiance multiplied by time. At 1,000 W/m², one hour equals 1 kWh/m².

Equal dose does not produce an equal electrical state when temperature or bias is different.

Module B may change faster because of the higher temperature, while open-circuit operation creates a different carrier condition from MPP operation. Record irradiance, dose, temperature, operating point, cooling method and time to the next flash.

Control Dark Storage Before Retesting

The following values are a diagnostic example, not commercial module data.

The recovery from 597.6 W to 599.4 W shows that much of the apparent loss was reversible. It does not identify the exact material mechanism. Keep dark-storage time, temperature, light level, connection state and measurement delay the same for every sample.

Check Spectrum, Reference Device and Light Uniformity

IEC 60904-7 defines spectral mismatch correction for the specific combination of simulator spectrum, reference device and module under test.^[14]

If current reads 1% low while voltage and fill factor remain unchanged, a 600 W module can appear close to 594 W:

This is a simplified calculation, not a claim that every simulator has a 1% error. Recalculate spectral mismatch when the cell technology, glass, encapsulant, lamp or reference device changes.

IEC 60904-9 classifies solar simulators separately for spectral match, irradiance non-uniformity and temporal instability.^[15] IEC 60904-2 defines the selection, calibration and care of photovoltaic reference devices.^[16]

A full module covers a much larger area than a reference cell. If rotating the module 180° changes Pmax from 600 W to 597 W while temperature and cables stay unchanged, inspect the test-plane irradiance map. The 3 W difference is evidence of a position effect, not proof by itself that the simulator fails its IEC class.

Compare Both Sweep Directions

NREL's module measurement facilities include a pulsed system using 30–100 ms flashes with sweeps in both directions, as well as a continuous-light system.^[17] The range describes that laboratory equipment; it is not a minimum pulse requirement for every tester.

A 50 ms pulse lasts 0.05 seconds. During that time, the lamp must remain stable, the load must scan the I-V curve, and the current and voltage channels must stay synchronized.

The difference is 5.4 W, or 0.9%. When temperature and irradiance are stable, check pulse length, sweep speed, capacitance and load synchronization. Do not average the two curves before finding the cause.

Calculate Contact and Cable Loss

Contact loss follows I²R. At 14 A:

An added resistance of 10 mΩ removes almost 2 W at 14 A, or about 0.33% of a 600 W result. Voc can remain normal because almost no current flows at open circuit.

Check connector pins, clamp pressure, current cables and voltage-sense positions. A four-wire connection uses separate current and voltage leads so cable voltage drop is not counted as module loss.

Control Rear Irradiance on Bifacial Modules

IEC TS 60904-1-2 defines the additional I-V measurement requirements for bifacial photovoltaic devices.^[18]

A simplified equivalent-irradiance example uses:

• Front irradiance: 1,000 W/m²
• Rear irradiance: 100 W/m²
• Bifaciality factor: 70%

Equivalent irradiance = 1,000 + (100 × 0.70) = 1,070 W/m²

If rear irradiance falls to 80 W/m²:

Equivalent irradiance = 1,000 + (80 × 0.70) = 1,056 W/m²

The two setups differ by 14 W/m², or about 1.3% of the first result. Keep the background, support rails, cable position, rear-light level and module-to-background distance unchanged. Use the model-specific bifaciality value from the applicable Tongwei document rather than a general 70% assumption.

Use One Test Sequence

If a current step, inactive area or unexplained fill-factor loss remains, use EL or infrared testing. Tongwei's module manufacturing process overview shows why flash testing and EL imaging are separate checks: flash testing measures electrical output, while EL helps locate damaged cells and current paths.

Set a Defensible Pass Limit

An NREL module self-reference calibration procedure reported ±1.1% expanded uncertainty for Pmax at a coverage factor of two. This is the result of a specialised calibration procedure, not a universal production-line limit.^[19]

Consider a result that falls from 600.0 W to 596.4 W:

• Power difference: 3.6 W
• Relative difference: 0.6%

If the measurement uncertainty is close to ±1%, the 0.6% difference is not enough by itself to prove permanent degradation. Retest at stable temperature, check the control module and use another diagnostic method before making a pass-or-fail decision.

Finally

A 5°C temperature rise can move a 600 W result by 7.5–10.5 W, 10 mΩ of extra resistance can remove 1.96 W at 14 A, and a 1% current error can appear as roughly 6 W of lost power. Check the I-V pattern before changing the test method: lower Voc usually sends you to temperature first, lower Isc to irradiance or spectrum, and lower fill factor to contacts or scan timing. Keep the module state, temperature, position, reference device and sweep settings fixed. Report degradation only when the loss remains under controlled retesting and is supported by EL, infrared or another independent test.