What the specification covers
IEC TS 62446-3 is the International Electrotechnical Commission's technical specification for outdoor infrared thermography of operating photovoltaic plants. It covers the modules and the balance of system, including cables, contacts, fuses, switches, inverters and batteries, imaged passively under natural sunlight.
It sets requirements for the measurement equipment, the ambient conditions, the inspection procedure, the report and the qualifications of the people doing the work, and it includes a matrix of typical thermal abnormalities. Edition 1, published in June 2017, is the current edition; a second edition is in development.
The document predates routine drone use, so some of its requirements are demanding to meet from the air. That is where most of the difference between vendors shows up.
Two levels of inspection
The specification separates a quick functional check from a full inspection. A report should say which one it is.
| Simplified inspection | Detailed inspection | |
|---|---|---|
| Purpose | A limited check that modules and balance-of-system components are working, for example at basic commissioning | Analysis of thermal signatures beyond the specification's examples, which may include absolute temperature measurement |
| People | Reduced qualification requirements | An authorized PV plant expert working with thermography experts |
| What it supports | Finding obvious faults. It cannot support authoritative conclusions about module quality | Classified findings and recommended actions an owner, insurer or engineer can act on |
Conditions at the time of capture
Thermal contrast depends on current flowing through the modules, so the specification ties a valid inspection to the conditions on site while imaging.
| Condition | What the specification expects |
|---|---|
| Irradiance | At least 600 W/m² in the plane of the array, measured on site with a sensor calibrated to ±5% |
| Wind | No more than about 28 km/h (17 mph) at the modules; wind cools modules and flattens differences |
| Cloud cover | No more than 2/8 of the sky; moving cloud changes the thermal pattern mid-flight |
| Plant state | Operating, in thermal steady state and, where possible, free of partial shading |
| Soiling | Low and even, under about 10% loss of maximum-power current; clean first if soiling or shading is heavy |
| Air temperature | Recorded with a sensor shielded from sun and wind, accurate to ±2 K |
In practice this means clear days, near midday, from spring through early fall in Southern California. Every value should be logged with timestamps so a reviewer can confirm the survey was flown inside the window.
Camera and resolution
| Item | Requirement |
|---|---|
| Spectral band | 8–14 µm for modules (2–5 µm only for components such as fuses) |
| Thermal sensitivity (NETD) | 0.1 K or better at 30 °C |
| Measurement accuracy | Absolute error under ±2 K |
| Geometric resolution | No more than 3 cm of module edge per pixel, about 5 × 5 pixels on a 6-inch cell |
| Settings | Adjustable emissivity and reflected temperature; radiometric images stored so temperatures can be read later |
| Calibration | Traceable, documented, at least every two years |
| Visual camera | A photo of every safety-relevant abnormality, at roughly 30 times the thermal resolution (9 megapixels or more for a 640 × 480 thermal camera) |
Resolution is the requirement drone surveys most often relax. Flying low enough for 3 cm per pixel takes considerably longer than flying for 5 cm per pixel; one software provider estimates about two and a half times the flight time. A survey flown at 5 cm per pixel can still find faulty strings and hot cells, but it shouldn't claim detailed-inspection resolution.
Viewing angle matters too. The emissivity of module glass changes with angle, and low sun angles add reflections, so capture angles should be planned for the array and tracker position.
Classes of abnormality
Findings are allocated to three classes, each with a recommended response:
| Class | Meaning | Recommended action |
|---|---|---|
| CoA 1 | No abnormality | None |
| CoA 2 | Thermal abnormality | Check the cause and, if necessary, fix it within a reasonable period |
| CoA 3 | Safety-relevant thermal abnormality | Prompt interruption of operation, then check the cause and fix it |
Temperature-difference bands such as "10 °C above neighbors" are common industry practice for grading severity, but they are a vendor's convention layered on top of the specification. A good report states the bands it used and maps them to these classes.
What the report should record
- Which level of inspection was performed, and any known deviations or limitations of the method
- The conditions log: irradiance, air temperature, wind, cloud cover and soiling, with times
- Camera, settings, calibration date and the resolution actually flown
- Each finding located to block, string and module, with its thermal and visual image, temperature difference and class
- Recommended actions based on each finding's class
- The radiometric images themselves, so another thermographer can re-read them
For periodic inspections the specification suggests an interval of four years unless the owner agrees otherwise or national rules set one. Owners who want to track change from season to season, or who are heading into a sale or refinancing, survey more often.
What thermography won't tell you
A thermal survey shows where heat is abnormal. It is not a power measurement, and it doesn't see everything.
- Detection depends on irradiance. A 2025 study found disconnected interconnects stayed detectable at 600 W/m², while cell cracks were hard to tell from healthy modules and early potential-induced degradation and glass cracks showed no pronounced thermal signature
- Components inside closed enclosures and cabinets can't be imaged from the air
- It doesn't replace the electrical commissioning tests in IEC 62446-1 or I-V curve tracing, which confirm what a thermal pattern suggests
Ten questions to ask any vendor
- Which level? Will the report claim a simplified or a detailed inspection?
- What resolution? What centimeters per pixel will be flown on the modules?
- Which camera? Is it radiometric, what is its NETD, and when was it last calibrated?
- How is irradiance measured? On site, in the plane of the array, logged with timestamps, not estimated afterward?
- What happens out of window? Are rows flown under cloud or wind re-flown, and does the report say so?
- Who reviews? Who analyzes and signs the findings, and what are their thermography and PV qualifications?
- How are findings classed? Are classes of abnormality used, and how do severity bands map to them?
- How are findings located? Down to block, string and module, matched to your asset IDs?
- What raw data do I get? Radiometric images, visual pairs, orthomosaics and the conditions log?
- Who owns it? Can you share it with your O&M provider, insurer, engineer or a buyer?
Tessera's solar thermal inspections are planned against this checklist, and our sample solar report shows the format.
Questions
Is IEC TS 62446-3 mandatory?
No. It is a technical specification, not a law. It becomes binding when a contract, an O&M agreement or a lender's requirements call for it, which is why the report should say exactly how the survey met it.
Can a drone survey meet it?
For the modules, yes, if it is flown at the required resolution, inside the condition window, with conditions logged and qualified people reviewing the results. Components inside enclosures still need ground inspection.
How often should a plant be surveyed?
The specification suggests every four years for periodic inspections unless otherwise agreed. Owners who track condition over time, or who face a sale, refinancing or warranty claim, usually survey more often.