Equipment Specifications & Inspection Methodology
Complete technical documentation for our thermal inspection platform: camera specifications, ground sample distance at operating altitudes, calibration procedures, and how each specification maps to IEC 62446-3 requirements for solar PV thermographic inspection.
This page is intended for solar O&M engineers, asset managers, and procurement teams evaluating inspection providers. All specifications are from manufacturer datasheets unless otherwise noted. We distinguish between manufacturer specs, calculated values, and our operational parameters.
Primary Equipment
Thermal Camera: Autel EVO II Dual 640T RTK V3
| Parameter | Specification | Source |
|---|---|---|
| Sensor Type | Uncooled VOx microbolometer | Manufacturer |
| Resolution | 640×512 radiometric | Manufacturer |
| Pixel Pitch | 12 µm | Manufacturer |
| Focal Length | 13mm | Manufacturer |
| Spectral Range | 8-14 µm (LWIR) | Manufacturer |
| NETD | ≤50 mK @ 30°C (typical) | Manufacturer |
| Temperature Accuracy | ±3°C or ±3% of reading (whichever greater) | Manufacturer |
| Accuracy Environment | −20°C to +60°C environment | Manufacturer |
| Accurate Measurement Distance | 2–20 m accurate measurement distance | Manufacturer |
| Radiometric Output | R-JPEG with full per-pixel temperature matrix (14-bit) | Manufacturer |
| Frame Rate | 30 Hz | Manufacturer |
What NETD Means in Practice
NETD (Noise Equivalent Temperature Difference) of ≤50 mK @ 30°C (typical) means the sensor can distinguish temperature differences as small as 0.05°C under ideal conditions. For solar PV inspection, this matters because early-stage cell degradation may produce temperature differences of only 2-5°C above reference. A sensor with poor NETD (e.g., >100 mK) would show these subtle anomalies as noise rather than actionable findings. Our 50 mK NETD provides a signal-to-noise ratio of 40:1 for a typical 2°C anomaly, ensuring reliable detection of incipient defects.
Important Caveats
- Accuracy valid only within stated environmental range; performance outside range is unspecified
- Quantitative temperature readings require 2–20 m standoff; beyond 20 m treat values as qualitative unless validated
- Thermal sensitivity (NETD) degrades at elevated ambient temperatures
RGB Visual Camera
Resolution: 50 MP (4K video)
Sensor: 1" CMOS with mechanical shutter
Output: DNG (RAW) + JPEG with EXIF GPS metadata
The RGB camera provides visual documentation of physical damage, soiling patterns, vegetation encroachment, and installation issues. Thermal and RGB images are captured simultaneously with aligned fields of view for defect correlation.
RTK GPS Positioning
| Parameter | Specification |
|---|---|
| Horizontal Accuracy (RTK Fix) | 1 cm + 1 ppm |
| Vertical Accuracy (RTK Fix) | 1.5 cm + 1 ppm |
| RTK Network | Texas VRS network via NTRIP |
| Fallback Mode | ±1–1.5 m GPS fallback if RTK signal lost |
Aircraft Operating Limits
Operating Temperature
−10°C to +40°C
Wind Resistance
27 mph (12 m/s) maximum
Flight Time
~38 min (varies with payload and conditions)
Aircraft operating temperature range is DIFFERENT from thermal camera accuracy range Wind resistance affects radiometric measurement accuracy (wind cools targets) Battery performance degrades in extreme cold
Ground Sample Distance (GSD) at Operating Altitudes
Thermal GSD vs. Altitude
GSD Formula
GSD = (H x pixel_pitch) / focal_length
Where H = altitude AGL, pixel_pitch = 12 µm, focal_length = 13 mm
| Altitude | GSD | px/cell (156mm) | px/cell (78mm half-cut) | IEC 62446-3 Mode |
|---|---|---|---|---|
| 15m (49ft) | 1.4 cm/px | 11.3 | 5.7 | Detailed+ |
| 20m (66ft) | 1.8 cm/px | 8.5 | 4.2 | Detailed |
| 25m (82ft) | 2.3 cm/px | 6.8 | 3.4 | Detailed |
| 30m (98ft) | 2.8 cm/px | 5.6 | 2.8 | Detailed |
| 40m (131ft) | 3.7 cm/px | 4.2 | 2.1 | Simplified |
| 50m (164ft) | 4.6 cm/px | 3.4 | 1.7 | Simplified |
| 60m (197ft) | 5.5 cm/px | 2.8 | 1.4 | Overview |
| 75m (246ft) | 6.9 cm/px | 2.3 | 1.1 | Overview |
IEC 62446-3 Detailed Mode
Requires a minimum of 5 x 5 pixels per cell. With standard 156mm cells, this means flying at or below approximately 30m (98ft) AGL. This mode enables detection and classification of individual cell-level defects: hotspots, PID, microcracks, and busbar anomalies.
IEC 62446-3 Simplified Mode
Requires a minimum of 3 x 3 pixels per cell. Achievable at 40-50m (131-164ft) AGL. Sufficient for detecting bypass diode anomalies, string outages, and module-level temperature deviations. Faster site coverage for large arrays where a full cell-level survey is not required.
Practical Note on Half-Cut Cells
Modern panels increasingly use half-cut cells (~78mm). These require lower altitudes for the same pixel-per-cell count. At 30m AGL, a half-cut cell gets only ~2.8 pixels across—below the detailed mode threshold. For detailed inspection of half-cut-cell panels, we fly at 15-20m AGL to maintain 5+ pixels per cell. We confirm the module type with the client before selecting flight altitude.
Measurement Distance Constraint
The Autel 640T specifies accurate temperature measurement at 2-20m standoff distance. At flight altitudes above 20m, absolute temperature readings should be treated as qualitative unless field-validated. For IEC 62446-3 inspections, we use relative temperature differences (Tdefect - Tref) which remain valid at higher altitudes because the atmospheric attenuation affects both measurements equally. Anomalies flagged at altitude are confirmed with low-altitude verification passes when quantitative temperature values are required.
Calibration & Maintenance Schedule
| Activity | Frequency | Method |
|---|---|---|
| Factory Calibration | Annual | Return to manufacturer. Blackbody reference sources, NIST-traceable. Certificate of calibration provided. |
| Field Verification | Quarterly | Check against portable blackbody reference target at known temperatures. Verify reading is within manufacturer's stated accuracy (±3°C or ±3% of reading (whichever greater)). |
| Pre-Flight Verification | Every flight day | NUC (Non-Uniformity Correction) performed at power-on. Lens cover on/off cycle. Visual confirmation of stable thermal image (no banding, no dead pixels). Ambient temperature noted in flight log. |
| Lens Cleaning | Before each flight | Germanium lens cleaned with lens-safe solution. Dust and fingerprints on the Ge window degrade thermal transmission and introduce measurement error. |
| RTK Base Station Check | Before each flight | Verify RTK Fix status before survey begins. Log RTK Fix percentage in flight metadata. Survey paused if RTK drops below 95% fix rate. |
Calibration records are available on request. We maintain documentation of all calibration activities, including factory calibration certificates, quarterly verification logs, and per-flight NUC records. These are included in Enterprise-tier deliverable packages and available upon request for all service tiers.
IEC 62446-3 Alignment: How We Meet Each Requirement
Environmental Conditions
IEC 62446-3 specifies environmental conditions under which thermographic inspection data is valid. We enforce these as go/no-go gates before and during every solar inspection flight.
| IEC 62446-3 Requirement | Our Protocol | Verification Method |
|---|---|---|
| Plane-of-array irradiance ≥600 W/m² | Minimum 600 W/m²; preferred ≥700 W/m². Flight aborted if irradiance drops below 600 W/m² for more than 2 minutes. | Ground-deployed pyranometer with data logger. Irradiance values timestamped and correlated with each thermal image. |
| Irradiance stability (<10% variation/min) | Flights scheduled during cloud-free windows. Data from intermittent cloud periods flagged and reviewed before inclusion. | Continuous pyranometer logging at 1-second intervals. Post-flight QA rejects frames captured during irradiance dips. |
| Wind speed ≤10 m/s (22 mph) | We target ≤5 m/s (11 mph) for optimal thermal contrast. Flights conducted up to 10 m/s per IEC allowance. Wind speed logged. | On-site anemometer. Wind speeds above 5 m/s are noted as a condition in the report because convective cooling reduces apparent ΔT. |
| Modules under normal operating load | Confirmed with site operator that all strings and inverters are energized before survey begins. Any known offline strings are documented. | Pre-flight coordination with site O&M team. Inverter status logged at survey start. |
Camera & Resolution Requirements
| IEC 62446-3 Requirement | Our Capability |
|---|---|
| Radiometric thermal camera | 640x512 radiometric sensor; full per-pixel temperature data stored in R-JPEG format. 14-bit thermal data, not visual-only palette overlay. |
| NETD ≤0.1 K (recommended) | ≤50 mK (0.05 K) — exceeds recommendation by 2x |
| ≥5x5 px/cell (detailed mode) | Achieved at ≤30m AGL for 156mm cells. See GSD table for altitude/resolution tradeoffs. We confirm cell dimensions with the client and select flight altitude accordingly. |
| ≥3x3 px/cell (simplified mode) | Achieved at ≤50m AGL for 156mm cells. Used for high-altitude screening passes on large arrays before targeted low-altitude detail passes. |
| View angle ≤60° from perpendicular | We target ≤30° from perpendicular (nadir or near-nadir) to minimize emissivity variation and reflected apparent temperature. Flight path planned to maintain consistent viewing geometry across the array. Module tilt angle factored into flight planning. |
Flight Pattern & Coverage
Systematic Grid Survey
- Pattern: Parallel flight lines aligned with tracker rows or fixed-tilt row orientation
- Forward Overlap: ≥70% (thermal), ≥80% (RGB) — ensures every module appears in at least 3 frames
- Side Overlap: ≥60% (thermal), ≥70% (RGB) — ensures full coverage between adjacent flight lines
- Speed: 3-5 m/s depending on altitude and GSD target. Slower speed = less motion blur in thermal frames.
Image Capture
- Trigger: Distance-based (not time-based) to maintain consistent overlap regardless of wind-induced speed variations
- Orientation: Camera nadir (pointing straight down) or tilted to match module plane-of-array angle for perpendicular viewing
- Simultaneous capture: Thermal and RGB images paired at each trigger point
- Metadata: RTK GPS position, altitude, yaw/pitch/roll, timestamp (UTC) embedded in every frame
Why High Overlap Matters
70%+ forward overlap means every module appears in at least 3 sequential frames. If one frame has a transient artifact (passing bird shadow, brief cloud), the anomaly can still be confirmed or rejected using adjacent frames. This redundancy also enables orthomosaic generation — stitching all frames into a single georeferenced site map where every pixel has a GPS coordinate.
Data Processing Pipeline
Ingest & Quality Gate
Raw R-JPEG thermal frames, RGB images, flight logs, and pyranometer data are ingested. Automated QA rejects frames captured during irradiance dips (<600 W/m²), excessive motion blur, or RTK signal loss. Typically 5-15% of frames are rejected during QA depending on conditions.
Radiometric Orthomosaic Generation
Thermal frames are stitched into a georeferenced orthomosaic using photogrammetric processing. Crucially, this is a radiometric mosaic — the per-pixel temperature values are preserved through the stitching process, not averaged into visual-only palettes. The result is a single GeoTIFF where every pixel has both a GPS coordinate and a temperature value.
Anomaly Detection & Classification
Temperature differences (ΔT) between suspected anomalies and adjacent healthy reference modules are calculated and normalized to 1000 W/m² standard irradiance using the formula:
Anomalies are classified by type (hotspot, bypass diode, string outage, wiring) and severity using normalized thresholds. See our IEC 62446-3 methodology page for the full classification system and severity bins.
Peer Review & Validation
Every anomaly is manually reviewed against the paired RGB image to confirm physical plausibility. Suspected false positives (bird droppings, temporary shading, reflections) are flagged and downgraded or removed. For suspected transient anomalies, the repeat-frame redundancy from our 70%+ overlap is used to confirm persistence.
Deliverable Generation
Final outputs are generated in multiple formats for different stakeholders. See Deliverable Formats below.
Deliverable Formats
GeoTIFF Thermal Orthomosaic
Georeferenced thermal map of the full site with preserved radiometric data. Compatible with QGIS, ArcGIS, and most asset management GIS systems.
CRS: UTM Zone 14N (EPSG:32614) for orthomosaics; WGS84 (EPSG:4326) for point data
CSV Anomaly Log
Row-level defect records with GPS coordinates, normalized ΔT, severity classification, anomaly type, and image references. Designed for direct import into CMMS platforms.
See CSV schema documentation for field definitions.
KML/KMZ Field Maps
Pin-drop anomaly locations color-coded by severity. Designed for field crew tablets. Tap a pin to see anomaly type, ΔT, severity, and a thermal thumbnail.
Compatible with Google Earth, Avenza Maps, and standard GPS navigation apps
PDF Inspection Report
Human-readable report with executive summary, findings by severity, thermal+RGB image pairs for each anomaly, and inspection conditions metadata. Format varies by service tier.
Enterprise tier includes full IEC 62446-3 methodology appendix with calibration records
Data Interoperability
All deliverables use open, standard formats. No proprietary viewer required. For integration with specific asset management platforms (PowerFactors, Raptor Maps, etc.), we can adjust field mappings and export schemas. See our data interoperability page for details.
Quality Assurance
Every inspection passes through these QA gates before deliverables are released to the client.
Environmental compliance: Pyranometer data confirms ≥600 W/m² for every included frame. Wind speed, ambient temperature, and cloud conditions logged.
Spatial coverage: Orthomosaic checked against site plan for gaps. Any uncovered areas noted in the report with reason (access restriction, tracker in stow, etc.).
RTK positioning: RTK Fix percentage for the survey is reported. Frames captured during RTK Float or standalone GPS mode are flagged with degraded position accuracy.
False positive review: Every flagged anomaly verified against RGB imagery and repeat frames. Transient artifacts removed. Borderline cases are included with a reduced confidence notation rather than silently dropped.
Normalization check: ΔT values are spot-checked against raw radiometric data to confirm normalization formula was applied correctly. Reference temperature (Tref) selection validated against 6-10 adjacent healthy modules per the IEC methodology.
Accuracy Disclosure
Temperature measurement accuracy: ±3°C or ±3% of reading (whichever greater), valid within a −20°C to +60°C environment; accurate measurement distance 2–20 m.
IEC 62446-3 anomaly classifications assume ≥600 W/m² plane-of-array irradiance with stable conditions
Temperature measurement accuracy: ±3°C or ±3% of reading (whichever greater), valid within −20°C to +60°C environment
Quantitative readings assume 2–20 m measurement distance; beyond this, treat values as qualitative
What This Means for Your Reports
Our defect classification relies on relative temperature differences (ΔT between anomaly and reference), not absolute temperature readings. Relative measurements are more robust than absolute values because atmospheric and environmental effects largely cancel out. The High Thermal Sensitivity absolute accuracy specification is relevant for specific engineering analyses (e.g., comparing measured module temperature to modeled predictions) but does not directly affect ΔT-based anomaly classification.
Related Technical Resources
- IEC 62446-3 Methodology — Anomaly classification, severity bins, ΔT normalization, and CSV schema
- Solar Inspection Services — Service tiers, deliverables, and pricing for utility-scale solar
- Equipment Overview — General equipment capabilities and professional certifications
- Thermal Drone Inspection Guide — How infrared cameras work, radiometric vs. non-radiometric, and cost estimator
- Solar Panel Inspection Guide — Defect types, inspection methodology, and energy loss calculator
- Data Interoperability — Deliverable formats, GIS compatibility, and CMMS integration
Questions About Our Equipment or Methodology?
We welcome technical questions from solar engineers, asset managers, and procurement teams. If you want to see sample deliverables or discuss how our methodology applies to your specific site configuration, reach out.
Based in Hale Center, TX. Serving the Texas Panhandle, South Plains, and ERCOT footprint. No travel fees within 40 miles.