A solar module can pass a visual inspection, match its datasheet at the flash test, and still carry damage that nobody on site can see. Cracked cells, broken connections and inactive cell areas sit beneath the glass and leave no visible trace. Electroluminescence (EL) imaging is the routine test built to find them. The IEA’s PVPS programme notes that more EPC and O&M companies now offer regular EL and infrared inspections of operating plants, alongside I-V curve measurements and monitoring. That growth has made EL images a common part of procurement and handover. It has not made them any easier to read.
How EL Sees What the Eye Can’t
A solar cell is designed to absorb light and turn it into electricity, and the process also works in reverse. IEA PVPS describes EL imaging as passing a direct current through the module and recording the light it gives off with an infrared-sensitive camera. For crystalline silicon, that emission peaks at around 1,150 nm, beyond what the human eye can detect.
Healthy cell areas glow evenly in the resulting image. Areas where current cannot flow properly appear darker, or fully black where a section of cell has been cut off electrically. Because the signal is faint, high-quality EL images are taken in a darkened room in the factory. In the field, they are taken at night, at sunset, or under curtains that block stray light.
What an EL Image Reveals, and What It Leaves Open
The defect EL is best known for is the microcrack. IEA PVPS notes that EL gives insight into microcracks and other defects in the cell material that are developing inside the module. A crack may cross a cell without disconnecting anything, or it may isolate part of the cell entirely, leaving that area dark in the image. Broken interconnects, soldering faults and inactive cell regions show up in a similar way. Cells affected by shunting-type potential-induced degradation (covered in our PID blog) also appear darker than their neighbours.
The harder question is what a visible defect will cost in output, and that answer has become less straightforward. A 2026 study involving the Fraunhofer Center for Silicon Photovoltaics tested 230 mini-modules across full-cell PERC and half-cut TOPCon designs. It found that crack length and crack count were not reliable predictors of power loss in modern modules, and that TOPCon modules showed no meaningful correlation at all. The researchers concluded that crack sensitivity depends mainly on irradiance, busbar count and cell architecture, more than on how long a crack is.
IEA PVPS also cautions that a single EL image, taken at one current and one temperature, cannot always pin a dark area to a specific physical cause. Pairing it with images at a second current, or with infrared thermography, narrows the diagnosis. In practice, EL is very good at showing that something is wrong and where. Judging how much it matters usually takes more than one image.
Why the Timing of the Test Matters
A crack can form at almost any point between stringing and commissioning, and when it is found shapes who is responsible for it. IEA PVPS notes that EL can link defects to different stages of a module’s life. They may come from production problems missed by quality control, from careless handling during transport or installation, or from environmental stress after installation. Each points to a different warranty or maintenance response.
Inside the factory, EL timing decides what can still be fixed. An image taken before lamination catches faulty cells while they can still be replaced. Once the module is laminated, a defect usually means rejecting the whole module. A final check after framing and junction box fitting catches damage introduced late in the line.
Factory testing, however thorough, ends at the loading dock. Cracks caused by vibration in transit or by rough handling on site can only be found by imaging the modules again after they arrive. IEA PVPS points out that EL images taken before installation are especially valuable if a dispute with the producer arises later, because production defects can still be clearly told apart from transport and installation damage at that point. The report also observes that EL is often requested much later, only after a plant starts underperforming, by which time those causes are much harder to separate.
Field EL takes time, which is why it is usually done on a sample. IEA PVPS estimates that EL inspection of a 4 MWp plant takes roughly 20 to 40 hours by drone and 100 to 150 hours with a handheld camera, under ideal conditions. For most projects, imaging a representative share of delivered modules before installation is the practical compromise.
Reading EL Results as a Buyer
EL images are easy to collect and harder to judge against a clear threshold. The IEC technical specification for EL defines how images should be captured, processed and interpreted. IEA PVPS notes that IEC technical specifications in this area describe test procedures and do not set pass/fail criteria. In most projects, what counts as an acceptable image is decided by the contract.
That makes the procurement conversation more important. When comparing a solar panel manufacturer in Gujarat, or anywhere else, the useful questions go beyond whether EL testing is done at all. It helps to know at which production stages EL runs, whether the images are stored against each module’s serial number, and which defect classification the manufacturer applies when deciding to repair, downgrade or reject a module.
On the buyer’s side, writing EL acceptance criteria into the supply contract before delivery gives both sides an agreed reference. So does setting a sample size for EL on arrival, which turns potential disputes into routine checks. Quotes from solar panel suppliers in Gujarat that include stored EL images by serial number also make warranty claims easier to support years later.
GREW Solar’s G12R datasheet specifies 100% string-level EL and triple-stage EL testing. Its production flowchart places those stages before lamination, where a faulty cell can still be repaired, after lamination, and again at final inspection once the module is framed and its junction box fitted. These are manufacturer-stated process details.
As cells get thinner, busbar counts rise and cell designs change, the link between what an EL image shows and what a module will lose is being reworked by researchers in real time. The defects themselves are easy to see. What still has to be agreed between buyer and supplier is how much each one matters.
