As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Point the camera at a surface, let it settle to the background temperature, and read the apparent temperature after setting emissivity and reflected temperature. A thermal camera measures infrared radiance from the surface, which it converts to temperature. It does not see through walls, it does not measure air temperature, and it cannot confirm what is behind a panel. The skill is in controlling the surface and the conditions, not in the camera. Models such as the MiLESEEY TR20 publish a 320 x 240 detector and a stated accuracy of plus or minus 2 percent, which is enough for building and electrical work if you understand what that figure covers.
What the sensor measures, and what the image is not
A thermal camera’s detector responds to long-wave infrared radiance arriving from the scene, roughly in the 8 to 14 micrometre band for the uncooled microbolometer sensors used in handheld and phone-attached models. The camera converts that radiance to an apparent surface temperature using a model that assumes a value for emissivity, the efficiency with which a surface radiates compared with a perfect black body. Every temperature you read is the output of that calculation, not a direct measurement of the object’s internal temperature.
This matters because a shiny metal surface at 60 degrees Celsius radiates very little of its own energy. What the camera sees is mostly infrared reflected from the surroundings. A polished copper pipe can read close to room temperature while carrying hot water. The camera is not broken; the surface is behaving as physics dictates. The practical fix is to change the surface, not the camera: a strip of matte electrical tape, a dab of flat black paint, or a patch of masking tape gives a known high-emissivity target.
The image is also not a photograph. The colour palette is an arbitrary mapping of temperature to hue. Two different palettes on the same scene will make different features look alarming. Use a fixed span with known limits rather than auto-scaling when you compare one part of an inspection with another, and say in your notes which palette and span you used.
Setting emissivity and reflected temperature before you trust a number
Emissivity is the single setting that changes readings most. Most cameras default to 0.95, which is reasonable for painted surfaces, plaster, brick, wood, and human skin. Bare metals sit far lower, often below 0.2, and the exact value depends on the alloy and the surface finish. If the listing does not publish a table of emissivity values, use the camera’s built-in material table if it has one, or apply tape. Do not guess a number for a metal you cannot identify.
Reflected temperature is the second setting. It tells the camera what infrared is arriving from the surroundings and bouncing off the target. In a room at 20 degrees Celsius with plain walls, the default is often close enough. Next to a furnace, a sunny window, or a hot roof, it is not. The standard method is the crinkled aluminium foil test: place a sheet of crumpled foil near the target, set its emissivity to about 0.95 in the camera, and read its apparent temperature. That value is your reflected temperature. Enter it, then re-read the target.
Distance and spot size come next. Every camera has an instantaneous field of view, the angular size of one detector element. The further you stand, the larger the patch of surface that feeds one pixel. If that patch is bigger than the component you care about, the reading is an average of the component and its surroundings. This is why a small terminal in a crowded panel reads cooler than it is: the pixel is mostly filled with air and neighbouring metal. Move closer, or use a camera with a narrower field of view. The emissivity setting and the spot-size calculation are the two adjustments that separate a usable survey from a decorative image.
| Setting | What it controls | Typical default | When to change it |
|---|---|---|---|
| Emissivity | How much of the received energy is treated as emitted by the surface | 0.95 | Bare metal, glass, polished surfaces |
| Reflected temperature | Infrared bouncing off the target from surroundings | Around 20 C in many cameras | Near hot or cold sources, outdoors, near windows |
| Palette and span | How temperature is mapped to colour | Auto span | Any comparison between images |
| Distance and focus | Which patch of surface feeds each pixel | Fixed focus on some phone models | Small targets, crowded panels |
Focus, stability, and the conditions that decide whether the survey works
Focus is not cosmetic. An out-of-focus thermal image blurs temperature across neighbouring pixels, so a small hot terminal smears into a warm patch and a cold leak spreads into a broad cool area. Manual-focus cameras let you sharpen the edge of the target; fixed-focus phone attachments rely on you holding the right distance. If the listing does not state a minimum focus distance, treat the figure as unknown and check the edge sharpness on a known object before you rely on small-target readings.
Thermal contrast needs a temperature difference. A wall with a missing insulation patch only shows up when the inside and outside temperatures differ enough to drive heat through the gap. In mild weather, the same wall looks uniform. The standard advice is to run the heating or cooling for a period before the survey, and to inspect when the difference across the building envelope is greatest. The camera cannot create contrast that the building is not producing.
Solar loading is the most common outdoor error. A wall that has been in direct sun for hours holds heat in its outer skin, which masks the pattern underneath. Survey shaded elevations, or wait until the sun has been off the surface long enough for the stored heat to dissipate. Wind has the opposite effect, stripping heat from the surface and flattening the pattern. Note the weather, the time since sunset, and the indoor-outdoor difference in your report, because a thermal image without those conditions cannot be interpreted later.
Reading a panel or a wall: the sequence that avoids false alarms
For electrical work, the sequence is: establish the load, image the panel, then measure. A loose connection only heats under current. An empty circuit reads ambient and tells you nothing. Record the load at the time of the image, because a thermal anomaly on a lightly loaded circuit may be harmless and the same joint under full load may be a fault. Compare like with like: the same phase, the same load, the same ambient. A single warm breaker is a question, not a conclusion.
For building work, the sequence is reversed: establish the temperature difference, then image, then confirm. A cool patch on an interior wall can be missing insulation, a stud, a duct, or a draught. The thermal image tells you where to look; it does not tell you what you have found. Confirm with a moisture meter, a borescope, or by opening the surface. The moisture inspection method depends on evaporative cooling, which means wet patches often read cooler than dry ones, but so do unheated voids and metal studs.
For machinery, the useful comparison is against the same machine in the same state, not against a generic temperature. A bearing that runs 15 degrees warmer than its twin under the same load is a candidate for inspection. A bearing that runs 15 degrees warmer than a different machine may be perfectly normal. Trend the same point over repeated surveys and record the load and ambient each time.
Phone-attached cameras such as the FLIR One for USB-C Android are convenient for this kind of spot checking because the app carries inspection guidance and merges the thermal and visual images with MSX. That merging is useful for locating the anomaly, but it also makes the image look more like a photograph than the data warrants. Read the temperature, not the picture.
Where the camera stops being the right instrument
A thermal camera cannot see through a wall, a panel, or a sealed casing. It reads the outer surface. If the surface is insulated from the fault, the fault is invisible. This is why thermal surveys of buildings work best from inside during a temperature difference, and why a thermal image of a closed electrical enclosure may show nothing even when a terminal inside is loose.
It does not measure air temperature, air leakage directly, or gas concentration. A draught shows up because it cools a surface, not because the camera detects moving air. A thermal camera is not a combustible gas detector, not a carbon monoxide alarm, and not a substitute for a clamp meter. Where a product is a safety device, the camera is an inspection aid, not the safety device.
Accuracy figures need reading carefully. A stated accuracy of plus or minus 2 percent, as published for the MiLESEEY TR20, is typically a percentage of reading plus a fixed offset, but the listing does not publish the offset. At a 20 degree reading, 2 percent is 0.4 degrees before any offset. That is fine for finding a 15 degree anomaly and marginal for judging a 2 degree one. The same caution applies to noise-equivalent temperature difference, or NETD: a figure below 40 millikelvin, as stated for the TR20, describes the smallest detectable difference under controlled conditions, not the accuracy of an absolute reading.
Resolution claims also need translating. The detector resolution is the number of sensing elements. Super-resolution or TISR processing, as used on the FOXWELL RT280 to upscale from 120 x 90 to 240 x 180, improves the appearance and can help locate a feature, but it does not create detector elements. Treat the native array as the honest measure of small-target capability. Where a listing mixes native and upscaled figures, use the smaller number for planning.
Finally, the camera is not a report. The value of a survey is in the conditions recorded alongside each image: load, ambient, emissivity, reflected temperature, distance, and the time since the heating or cooling was applied. An image without those is a picture of a warm thing.
The habit that makes thermal surveys useful
Set the surface, set the conditions, then read the number. A thermal camera rewards the operator who controls emissivity, reflected temperature, focus and temperature difference before pressing the trigger. The instrument is the easy part. The discipline of recording load, ambient and settings with every image is what turns a colourful screen into evidence you can defend later. If you are choosing between models, start with the resolution and field of view that match your smallest target, and read the smartphone thermal camera guide before deciding whether a phone attachment or a handheld suits your work.
Frequently Asked Questions
Can a thermal camera see through walls?
No. It reads the temperature of the surface you point it at. It can show a pattern caused by heat moving through the wall, such as a missing insulation patch or a cold bridge, but it cannot show what is inside the cavity. Confirmation requires a moisture meter, a borescope, or opening the surface.
Why does a shiny metal pipe read cold when it is hot?
Bare metal has low emissivity, so it radiates little of its own heat and reflects the surroundings instead. The camera reports mostly reflected energy. Apply matte tape or flat black paint to a small area, set emissivity to about 0.95, and read the tape.
What accuracy should I expect from a handheld thermal camera?
The listings here publish plus or minus 2 percent for the MiLESEEY TR20. That is a percentage of reading, often with an additional fixed offset that the listing does not publish. It is adequate for finding clear anomalies and not adequate for calibrating a sensor. Check the stated figure and the conditions it applies to.
Does super-resolution make a 120 x 90 camera as good as a 320 x 240 one?
No. Processing such as TISR or Vivid IR improves the displayed image and can help locate a feature, but the detector still has 120 x 90 or 80 x 60 elements. Small-target measurement depends on the native array and the field of view. Use the native figure when planning a survey.
Can I use a thermal camera to find a gas leak?
Not directly. Some gases absorb infrared and can be visualised with specialist cameras tuned to specific bands, but a general-purpose thermal camera does not detect gas concentration and is not a combustible gas detector or a carbon monoxide alarm. Use the appropriate gas instrument.
What conditions give the best building survey?
A steady temperature difference across the building envelope, ideally at least 10 degrees Celsius, with the heating or cooling running for several hours beforehand. Avoid direct sun on the surface and avoid high wind. Record the indoor and outdoor temperatures, the time, and the weather with each image.