Thermal imaging camera vs spot pyrometer: when one sensor sees all

Every hot process line eventually faces the same instrument choice: one spot pyrometer per measurement point, or one fixed thermal imaging camera that watches the whole field and reports many points at once. The decision is more arithmetic than opinion. It comes down to four numbers — how many points you need, how stable the target's emissivity is, how many millimetres one pixel or one optical spot covers at your working distance, and how the value reaches your PLC. This comparison walks through each number with worked calculations, gives an emissivity error budget you can reuse on your own target, and ends with a recommendation for each common case — including the case where the camera is the wrong tool despite its impressive image.

Two instruments, two architectures

A spot pyrometer is one detector behind fixed optics: it integrates the infrared power arriving from a single cone and reports one temperature, continuously. Its field is defined by a distance-to-spot ratio on the data sheet — a 50:1 unit views a 20 mm circle at 1 m — and its output is usually the 4-20 mA current loop, the standard process signal that lands straight on a PLC analog input.

A fixed thermal camera is an array of microbolometer or photon detectors working in the 8-14 μm band, and every pixel acts as a tiny radiometer. A 640 × 512 sensor holds 327,680 potential measurement points; software regions of interest (ROIs) turn any group of them into live values, alarms and logged trends. Where the detector families differ across the spectral ranges, see how camera sensor technologies compare, from CMOS to microbolometer.

The decision at a glance

Everything below is the reasoning behind those rows.

Emissivity sets the accuracy budget

Both instruments measure radiance, not temperature, and convert using the surface emissivity ε — the share of blackbody power a surface emits (a perfect blackbody has ε = 1). Typical values from thermography tables: matte non-metallic surfaces 0.85-0.95, oxidised steel 0.7-0.9, shiny aluminium 0.05-0.2. A low-ε surface also reflects its surroundings, so the reading error from a wrong ε setting scales roughly with (1 - ε)/ε — the same slip that barely dents a matte reading wrecks a shiny one.

For a part at temperature T in surroundings at T_amb, the standard grey-body estimate gives the reported temperature as:

Thermal cameras of this class typically specify accuracy of about ±2 °C or ±2 % of reading, whichever is greater; industrial spot pyrometers are commonly quoted near ±1 % on a calibrated target. The table shows what such a spec is worth: on the matte patch, a ±0.05 emissivity slip costs about as much as the whole instrument accuracy; on shiny metal it costs roughly ten times more, and no accuracy spec survives it. A hot furnace wall in the background pushes the shiny-metal errors higher still, because the reflected term then sits far above 20 °C. The arithmetic above is the total-radiation estimate; an 8-14 μm band camera lands in the same order at these temperatures.

Three engineered escapes exist when the surface cannot be patched. A two-colour pyrometer divides radiances at two close wavelengths, so a uniform emissivity drop cancels; it stays accurate on partly shiny targets as long as oxidation or water films do not bend the spectral shape. Shorter measurement wavelengths shrink the error per unit of emissivity error, one reason SWIR, NIR and visible bands get chosen by what the material absorbs. And once steel is hot enough to glow, its own emission dominates the reflection term — see imaging hot forged parts when the part itself glows.

Field of view, pixel size and spot size

A camera can only measure what it can resolve. Two lines of geometry settle it:

The LS-IRC-0044 is a 640 × 512 LWIR thermal camera in the 8-14 μm band with an 88.5° × 73.2° field of view, a 50 Hz frame rate, a GigE interface, a C-mount lens mount and a 9-24 V DC supply.

At 0.5 m working distance its field is 2 × 500 mm × tan(44.25°) = 974 mm wide and 2 × 500 mm × tan(36.6°) = 743 mm tall, so one pixel covers 974 mm / 640 = 1.5 mm. Thermography's working rule: a target must cover at least a 3 × 3 pixel block to flag reliably and 5 × 5 or more to read accurately. This set-up therefore sees a 5 mm hot spot and measures one of 8 mm or larger. Dividing field by smallest target checks any combination the same way: 974 mm / 5 mm ≈ 195 pixels needed across, well inside the 640 available. The calculator below runs that check for your own numbers.

A pyrometer answers the same question with its optical spot. At 50:1, the spot at 1 m is 20 mm across; a 10 mm target underneath averages the part with its cool surroundings — a size-of-source error no emissivity setting can repair. Pixel arrays fail identically only when the ROI is drawn larger than the target; the pixels themselves keep the detail. The full distance-and-target-size procedure is in thermal camera working distance and spot size for safe scans.

Narrow fields beat wide ones at distance

At 3 m, the wide lens above spreads its 640 pixels across 5.8 m of conveyor and pixel size climbs past 9 mm — wider than most hot spots. Narrow the field instead. The MV-CI003-GL-N25 pairs a 0.3 MP LWIR detector with a 25 mm focal length, a GigE interface and an M34x0.75 mount, built for exactly that trade: fewer pixels, each covering a smaller angle.

When a camera accepts interchangeable optics, dedicated LWIR glass protects the resolution you chose; the thermal, SWIR and UV lens range carries nearly 200 LWIR models, and the lens finder filters them by mount and image circle. The LS-SWL-0006 is a 17 mm F1.2 thermal lens sized for a 1/1.2" sensor with an 11-16 mm image circle. In hot areas, also check infrared lens athermalization: keeping focus as temperature changes before assuming focus holds through the shift.

Getting the value into your control system

A pyrometer ends at a wire: 4-20 mA into the PLC analog input, often with an alarm relay beside it. Integration is an afternoon.

A camera ends at Ethernet. Its 50 Hz frame stream rides over GigE or USB3; ROI software turns pixels into values and hands them to the PLC over the industrial protocol it already speaks, or through the camera SDK when the value feeds a data-acquisition bench — see integrating a thermal camera into a DAQ application via its SDK. That software needs a home on the line, and a fanless industrial mini PC in the control panel is the usual host. The payoff is the record: a radiometric JPEG keeps a 16-bit temperature value in every pixel, and you can export high-bit-depth grayscale frames for offline analysis — an audit trail a single-point sensor cannot offer.

For OEM instruments that embed temperature sensing inside their own housing, the module route removes the enclosure entirely: the LS-IRC-0016 is a 640 × 512 LWIR camera module with a USB3 interface, made to live inside your machine rather than beside it.

Which one, case by case

One critical number, one control loop, small and stable target — the spot pyrometer. A wire, millisecond response, minutes of set-up, and two-colour types cover the shiny cases.

Many points on one part, temperature profiles, trend maps, operator diagnostics — the fixed thermal camera. One sensor reports every ROI; the array costs the same whether you draw 3 regions or 300. Sensitivity and optics trade-offs behind that choice are in thermal camera NETD and lens choice for process monitoring.

Mixed lines — both. The camera watches the field and finds the hot spot; the pyrometer or a tight camera ROI holds the control value. Wide surveillance plus one calibrated point is exactly how cameras for continuous steel casting monitoring are set up.

Shiny metal with drifting emissivity — patch the surface first. If patching is impossible, a two-colour pyrometer carries the control number while the camera covers the rest of the field.

Dusty, wet or hot installations — both instruments need protection: purged optics, stable mounting temperature, and a window where the process would coat the lens, chosen with choosing an infrared window material for a thermal sensor enclosure.

When the camera is the wrong upgrade

Three cases belong to the pyrometer. Resolution: if no safe mounting distance gives your smallest target a 3 × 3 pixel block — run the two-line geometry above before believing any demonstration image — the attractive picture carries wrong numbers. Determinism: the pyrometer's analog value reaches the PLC with no software in the loop; if the process cannot tolerate an application restarting mid-shift, keep the wire. Proportion: for a single point with no image requirement, a camera plus ROI software plus a host computer is more hardware than the measurement needs. Browse the infrared cameras category or the camera finder once the arithmetic says the array earns its place.

Checks before you order

1. List every measurement point with its size in millimetres — the point count decides the architecture before any data sheet does.
2. For each camera candidate, compute FOV = 2 × WD × tan(angle/2) and IFOV = FOV / pixels; require a 3 × 3 pixel block on the smallest target.
3. For each pyrometer candidate, compute spot size = distance ÷ distance-to-spot ratio and confirm the target fills it.
4. Write down the emissivity of every surface; flag anything below 0.5 for a patch, a two-colour sensor or a glow check.
5. Fix the signal path — analog into the PLC, or Ethernet plus a software host — because it decides the wiring before the sensor does.
6. Check the background: a hot wall or flame in view breaks the cool-background assumption behind every error budget above.
7. Plan the environment: purge air against lens fouling, stable mounting temperature, and a window where the process coats the optics.

When the arithmetic lands on the camera side, the configurator walks through sensor, interface and optics step by step, and the cameras topic page collects the related guides. For a second opinion on a specific line, reach the team through the contact page.