How Accurate Are Thermal Cameras? A Field Guide for Electrical Inspectors

What Does ±2 °C or ±2 % Accuracy Actually Mean?

Open any thermal camera datasheet and you will find the same line: Accuracy: ±2 °C or ±2 % of reading, whichever is greater. For a handheld radiometric camera, that is the manufacturer’s guarantee under defined laboratory conditions — emissivity set correctly, camera stabilized, target filling the field of view, and no reflections. In practice it means:

• At a 50 °C connection, the camera can legitimately read between 48 and 52 °C.
• At a 500 °C furnace surface, the tolerance grows to ±10 °C (2 % of 500).
• The guarantee assumes you have set emissivity correctly — the single largest error source in the field.

Accuracy is not the same as resolution or sensitivity. Resolution (384×288, 640×480) defines how many measurement points the image contains; NETD (thermal sensitivity) defines the smallest temperature difference the detector can resolve; accuracy defines how close the reported temperature is to the true temperature. A camera can be extremely sensitive (15 mK NETD) and still read 5 °C off if emissivity is wrong. Understanding the difference is the difference between a useful inspection program and a false sense of security.

The EasIR handheld thermal camera series is a radiometric tool with the standard ±2 °C / ±2 % accuracy — but like every radiometric camera, its field accuracy is only as good as the settings behind each measurement.

Seven Factors That Change Real-World Accuracy

Even a calibrated camera drifts away from true temperature when field conditions are not the ones in the laboratory. The seven factors that change real-world accuracy:

1. Emissivity — the surface property that determines how much infrared the material radiates. The number one error source; covered in the next section.
2. Reflected apparent temperature — shiny surfaces act like mirrors. A polished copper busbar in front of a hot wall reflects the wall’s temperature, not its own.
3. Distance and atmosphere — humidity, dust, and smoke absorb infrared between you and the target. Beyond roughly 10 m in tropical humidity, readings begin to drift.
4. Focus — an out-of-focus image blends hot and cold pixels together, flattening the true temperature.
5. Measurement angle — reading a surface at a steep angle (past about 60° from perpendicular) lowers the apparent emissivity and reads low.
6. Thermal stabilization — moving from a 35 °C outdoor yard into an air-conditioned switchgear room changes the camera’s internal temperature; allow 5–10 minutes.
7. Target filling — if the hot spot is smaller than the measurement spot (IFOV), the camera averages it with the cooler background and reads low.

In hot, humid LatAm environments — coastal Colombia, the Caribbean, the Brazilian northeast — factors 3 and 6 are more aggressive than in temperate plants, which is why disciplined survey procedures matter even more.

Emissivity: The Number One Source of Error

Emissivity is the ratio of infrared a surface emits compared with a perfect blackbody, from 0 (a perfect mirror) to 1 (a blackbody). Most organic and painted surfaces sit at 0.85–0.95. Polished metals sit very low: shiny copper around 0.05–0.15, aluminum 0.05–0.10, galvanized steel 0.20–0.30.

Here is the danger: if a surface has a true emissivity of 0.2 and the camera is left at 0.95 (the factory default), the camera behaves as if the surface radiated like a blackbody — the reading will be far below the true temperature, sometimes 30–80 °C low on a hot busbar.

Three fixes: paint a small matte black patch, apply electrical tape (emissivity ≈0.95), or measure on a flat surface at an angle that avoids reflections. The Hammer II series lets you set emissivity per measurement spot and stores it with the radiometric image, so the reading stays traceable when you revisit the asset next month.

How to Verify Accuracy in the Field

Manufacturer calibration is valid for a year (or as stated), but a camera can be damaged in transit, after a drop, or after years of tropical heat. A 15-minute field check:

1. Ice bath: mix crushed ice and water, let it stabilize for 5 minutes, and point the camera at the water surface with emissivity 0.95 and reflected temperature about 20 °C. Reading should be 0 ±2 °C.
2. Boiling water (at sea level): should read 100 ±2 °C. In high-altitude LatAm cities — Bogotá (2,640 m), Quito (2,850 m), Mexico City (2,240 m) — water boils at 91–95 °C, so use a thermometer to get the local reference.
3. Body temperature: your own skin reads 32–34 °C with emissivity 0.98.
4. Known reference: if you own a calibrated contact thermometer, compare both on the same painted surface at 40–60 °C.

Rule of thumb: if the camera reads within ±2 °C of the reference at 50 °C, it is healthy for electrical inspection. If it drifts beyond ±5 °C, send it for calibration before it produces a report you cannot defend.

Thermal Camera vs Contact Thermometer Accuracy

Contact probes (thermocouples, RTDs) claim ±0.5–1 °C — better than any thermal camera. But they only measure what they touch. On a live panel you cannot touch a 400 V busbar; on a cable tray you cannot reach the conductor; on a loaded connector you may not even have access. The thermal camera trades some absolute accuracy for three things a probe cannot do: non-contact measurement, full-scene coverage, and a permanent radiometric record.

For trending — comparing last month’s reading with today’s — what matters is repeatability, and a thermal camera with fixed emissivity, distance, and span settings is very repeatable. That is why month-over-month thermography catches faults long before absolute accuracy matters.

Unitech Tools Thermal Cameras Compared

All three Unitech Tools handheld series are radiometric cameras with the standard ±2 °C or ±2 % accuracy, and each adds features that protect accuracy in the field:

Focus is the feature that quietly protects accuracy more than any other: an out-of-focus thermal image is an inaccurate thermal image. The Hammer II’s laser-assisted focus removes that variable, and its dual field-of-view lets you frame a single lug in wide angle, then zoom with the tele lens to fill the measurement spot. The PT II adds 1280×1024 detail for forensic-grade analysis of small components at distance — a small hot spot measured with one pixel is a guess; measured with a filled spot is data.

Best Practices for Reliable Readings

1. Set emissivity before you measure — for every spot, or use saved material profiles.
2. Avoid reflections: change the angle until the hot background disappears from the surface.
3. Fill the measurement spot: get closer or switch to the telephoto lens.
4. Focus on the target before freezing the image.
5. Stabilize the camera 5–10 minutes when moving between environments.
6. Record the conditions: load, emissivity, distance, ambient temperature.
7. Compare like with like: same phase, same load, same angle — the ΔT between phases is more reliable than any absolute number.
8. Trend over time: keep the same settings month to month.

“A thermal camera is a measurement instrument, not a toy. Set the emissivity, fill the spot, and document the conditions — then the ±2 °C spec means something.” — Maintenance Manager, Power Distribution Utility, Peru

Accuracy is a combination of instrument and discipline. Choose a radiometric camera with the right resolution and NETD for your program — EasIR for daily rounds, Hammer II for utility surveys, PT II for precision work — and then protect that accuracy with the habits above.