What Emissivity Does to an Infrared Reading
Why this matters
An infrared instrument does not measure temperature. It measures how much infrared energy arrives at its lens from the direction you pointed it, then converts that to a temperature using two assumptions you supplied, usually without knowing you supplied them. When the target is a dull painted surface a few degrees off ambient, the assumptions barely matter. When it is bare metal running well above the room, they can move the answer by tens of degrees, in either direction, and the instrument will show that answer with a decimal point on it. Techs have condemned good components and cleared bad ones on that decimal point.
Before you take the reading
The temptation with a non-contact instrument is that it feels like it removes the hazard, so people point it into places they would never put a hand. It does not remove anything.
Aiming into an open electrical enclosure to find a hot lug means the enclosure is open with live parts exposed, which is energized electrical work under 29 CFR 1910.333(b)(2). Establish an electrically safe working condition where the task allows it, and where the reading genuinely requires the equipment energized, work to the energized-work provisions with the right protective equipment and boundaries for the task, proving any companion meter live-dead-live per NFPA 70E-2021, 120.5. An IR gun in your hand is not a substitute for either.
Do not aim the instrument's laser pointer at anyone's face, and do not use the laser to line up a shot near a mirror-finish surface, which will send it somewhere you are not looking.
If the reading requires you to apply tape or paint to raise the target's emissivity, that is a contact task on a surface that may be hot, moving, or both. Do it with the equipment at rest and cool, or use an extension applicator. For anything rotating, stop it and isolate it under 29 CFR 1910.147 for the mechanical and stored energy before your hand crosses the plane of a guard. If you spray a coating, solvent vapor is an inhalation hazard: ventilate the area and follow the product's safety data sheet for respiratory protection, since a glove does nothing about a vapor.
What the instrument measures, and what it converts
Every surface above absolute zero radiates. How strongly depends on its temperature and on how good an emitter it is, and that second property is emissivity: a number from 0 to 1 comparing the surface to a perfect emitter at the same temperature. Flat dark paint, most oxidized metals, rubber, wood, brick and fabric run high, often around 0.9 to 0.95. Polished and bright metals run low, sometimes below 0.1.
Two things reach the lens from the direction you pointed:
- What the target emitted, which is emissivity times a fourth-power function of the target's absolute temperature.
- What the target reflected from everything around it, which is one minus emissivity, times the same fourth-power function of the surroundings' temperature.
The instrument adds those together, divides by the emissivity you set, and takes the fourth root. Absolute temperature here means Rankine, degrees F plus 459.67, and the fourth power is why the error is not a fixed offset.
A low-emissivity surface is therefore two problems at once: it emits weakly, so the signal is small, and it reflects strongly, so most of what you are reading is the room.
The two assumptions, and the card that records them
Any infrared reading worth putting in a file carries the reading and the conditions that make it interpretable. Six fields:
| Field | What goes in it |
|---|---|
| Target and exact spot | The surface, and where on it, precisely enough to return to |
| Surface condition | Bare bright metal, oxidized, painted, taped, wet, dusty |
| Emissivity setting used | The number in the instrument at the moment of the reading |
| Reflected background | The temperature of what surrounds the target, and whether the instrument has a setting for it |
| Distance and spot size | Distance, plus the instrument's distance-to-spot ratio |
| Reading, and the check | The number, and whatever independent check you ran on it |
Without the middle four, the reading in the last row is a number with no error bar, and a year from now nobody can tell whether it agrees with a new one.
The filled-in card
A hot line running through a mechanical room. The surroundings are close to room temperature at 75 degrees F.
| Field | Entry |
|---|---|
| Target and exact spot | Horizontal line, 6 in downstream of the elbow, top of pipe |
| Surface condition | Bare copper, lightly oxidized, dry |
| Emissivity setting used | 0.95 as found, instrument default |
| Reflected background | 75 degrees F room; instrument has no background setting |
| Distance and spot size | 24 in at 12:1, so a 2 in spot on a 1 in pipe |
| Reading, and the check | 152 degrees F as found; taped re-read pending |
Every one of the last four rows is a problem, and the arithmetic below says how much of a problem.
The error, computed three ways off the same surface
Take that lightly oxidized copper as having a true emissivity of about 0.6. Treat 0.6 as illustrative rather than as a spec for your pipe: real values for copper span from below 0.1 polished to around 0.7 heavily oxidized, and the only way to know yours is the tape test below or a published table for that exact surface condition. What is not illustrative is the shape of the error, which follows from the radiance balance directly.
Case one, hot surface, wrong emissivity setting. True surface 180 degrees F (639.67 R), surroundings 75 degrees F (534.67 R), true emissivity 0.6, instrument set to 0.95.
- What arrives: 0.6 x (639.67 to the fourth, about 1.674e11) plus 0.4 x (534.67 to the fourth, about 8.172e10), which totals about 1.331e11.
- What the instrument reports: divide by 0.95 to get about 1.402e11, take the fourth root, get about 611.9 R, which is 152.2 degrees F.
The instrument reads 28 degrees F low on a surface that is 105 degrees F above its surroundings.
Case two, same surface near ambient. Same emissivity, same setting, same room, but a true surface of 100 degrees F (559.67 R).
- What arrives: 0.6 x about 9.811e10 plus 0.4 x about 8.172e10, totalling about 9.156e10.
- Reported: divide by 0.95, fourth root, about 557.2 R, which is 97.5 degrees F.
Now the error is 2.5 degrees F low. Same surface, same wrong setting, same room, and the error went from 28 degrees F to 2.5 degrees F because the target moved closer to its surroundings.
That is the rule worth carrying: the emissivity error scales with how far the target is from the things around it. Near ambient, a wrong setting is almost harmless. On anything genuinely hot or genuinely cold, it dominates.
Case three, the correction that makes it worse. Now set the emissivity correctly to 0.6 on the 180 degrees F surface, on an instrument that has no reflected-background setting, so it is doing nothing but dividing by the number you gave it.
- What arrives is unchanged at about 1.331e11.
- Reported: divide by 0.6 to get about 2.219e11, fourth root, about 686.3 R, which is 226.6 degrees F.
The reading is now 47 degrees F high. Setting emissivity correctly, without also accounting for the reflected background, overshot by more than the original wrong setting undershot, because the reflected component from a cooler room got divided by a small number as though the target had emitted it.
Direction check on those three results against the printed values: 152.2 is below the true 180, 97.5 is below the true 100, and 226.6 is above the true 180. Two low, one high, and the high one is the case where a tech thought they had fixed it.
An instrument that does carry a reflected-temperature or background setting, given both 0.6 and 75 degrees F, subtracts the reflected term first and lands on 180. If yours has only an emissivity setting, it is doing case three.
The tape test beats the arithmetic
Nobody should be doing fourth-power algebra on a service call. The field move is to stop guessing the target's emissivity and change it instead.
Apply a strip of matte tape with a known high emissivity, commonly around 0.95, or a spot of flat dark paint, to the target. Let it come to the surface's temperature, which takes appreciably longer than people allow, at least a minute on a light surface and longer on anything with a stagnant film of air over it. Set the instrument to the tape's value, aim at the tape, and read.
Then you have a defensible number on that spot, and one more thing: aim at the bare surface immediately beside the tape and note what it reads. The difference between the two is your instrument's error on that surface condition, at that temperature, in that room. That is a calibration you can reuse on the rest of the same piping for the length of that visit, and it is a far better use of five minutes than a table lookup.
The same fix in reverse explains the classic false alarm: a bright bus bar or a shiny cabinet wall that reflects a genuinely hot component nearby will read hot even though it is not, because most of what left it toward your lens was reflection. Move your angle. A real hot surface reads the same from several angles; a reflection moves or vanishes when you do.
Distance to spot, the other way the number goes wrong
The reading is an average over whatever the optics saw, so a target smaller than the spot gets averaged with its background, and the answer slides toward the background.
Distance-to-spot ratio is the instrument's own specification and varies enormously between tools, so read it from the documentation rather than assuming. On a 12:1 instrument at 24 in, the spot is about 2 in across. On a 1 in pipe that means roughly a quarter of the spot area is on the pipe and the rest is on the wall behind it, and the reported number is nowhere near the pipe's temperature no matter what emissivity you set.
The working rule is that the target should be at least twice the spot diameter, which usually means walking closer rather than adjusting anything. Be aware that the spot size quoted at a given ratio is typically defined to contain most of the energy, not all of it, so treat the stated spot as a floor and give yourself margin.
The field check that belongs on the card
Fill in the last field with an independent check, not with confidence.
The cheapest one that cannot lie: a stirred bath of ice and water, aimed at from a few inches, with the instrument set to about 0.95. Water at the surface of a well-stirred ice bath is at its freezing point and is a strong emitter, so a healthy instrument reads close to 32 degrees F. If it does not, nothing further in this article helps you, because the tool itself is out.
The second check is a contact probe on the same spot immediately after the infrared reading, with the probe given time to settle and with enough contact to actually couple. Where the two disagree by more than a couple of degrees on a taped, filled, close-range target, believe the contact probe and go looking for what the optics were seeing.
Record whichever check you ran in that last field. A reading with a check beside it is evidence. A reading alone is an anecdote with a decimal point.
References
- ASHRAE Handbook - Fundamentals, heat transfer chapter, for the radiant exchange relationship the conversion is built on
- Instrument manufacturer documentation for distance-to-spot ratio, emissivity range, and whether the tool provides reflected-background compensation
- 29 CFR 1910.333(b)(2) for energized electrical work; 29 CFR 1910.147 for mechanical isolation and stored energy; NFPA 70E-2021, 120.5 for live-dead-live proving
- See related: How to Read a Surface Temperature Honestly; Using an Infrared Thermometer for Diagnosis