7.1 Equipment Calibration and Accuracy Specs

Quantitative Level II thermography rests on a simple chain: radiation leaves the target, the camera’s optics and detector convert in-band radiance into a signal, and factory calibration maps that signal to an apparent blackbody temperature. If calibration is wrong, every emissivity correction, NETA ΔT, ISO zone, and trend chart inherits the error. This section covers how cameras are calibrated, how accuracy specs are written, how you check performance in the field, and when Level II judgment says “send it to the lab.”

What “Calibration” Means for an IR Camera

A thermal imager does not contain a thermocouple pressed against the scene. It measures infrared radiation in a spectral band (often LWIR ~7.5–14 µm for industrial work) and reports temperature only after applying radiometric calibration curves. Those curves are built by presenting the camera with known blackbody temperatures under controlled conditions and recording detector response (counts, voltages, or digital levels) at many set points across each range.

Blackbody reference sources

A blackbody used for calibration is a high-emissivity cavity or flat-plate source whose temperature is controlled and measured with contact standards traceable to a national metrology institute (for example NIST in the United States). Key ideas:

Factory calibration typically covers each selectable temperature range, sometimes with and without spectral filters or high-temperature windows. The result is stored as calibration tables or polynomial coefficients in camera non-volatile memory. When you later set emissivity and reflected apparent temperature (RAT), the camera still starts from that blackbody-equivalent radiance solution and then corrects for non-ideal targets.

Internal vs external references

Many modern cameras also use an internal shutter or flag that periodically presents a known reference temperature (or temperature-stable surface) to the detector. That process is often called NUC (non-uniformity correction) or internal calibration. It reduces drift and fixed-pattern noise between frames. Internal NUC is not a substitute for laboratory blackbody calibration. It keeps the array uniform relative to a local reference; it does not re-establish SI-traceable accuracy across the full measurement range.

Accuracy Specs: Reading ±2 °C or ±2%

Manufacturers commonly publish accuracy in forms such as:

• ±2 °C or ±2% of reading, whichever is greater
• ±1 °C or ±1% of reading (higher-end or lab-oriented instruments)
• Separate figures for different ranges or with/without filters

How “whichever is greater” works

At low temperatures, the fixed degree error usually dominates. At high temperatures, the percentage term dominates.

Exam implication: A camera “accurate to ±2 °C or ±2%” is not always within ±2 °C. At 400 °C the allowed band is about ±8 °C before you even add emissivity and atmospheric errors. Severity classifications near a hard absolute limit (for example an OEM winding limit) must respect that uncertainty.

Accuracy is not resolution, NETD, or repeatability

Level II candidates must separate four related but different concepts:

A camera can have excellent NETD (pretty, low-noise images) and still be out of calibration by several degrees. Conversely, a well-calibrated camera with coarse NETD may still be acceptable for coarse absolute limits but poor for subtle building-envelope ΔT.

Stated conditions on the datasheet

Accuracy claims assume conditions such as:

• Target is a blackbody (or specified high ε) filling the measurement area
• Ambient temperature within a stated band (often ~15–35 °C for handheld industrial units)
• Correct range selected, optics clean, focus correct
• No intervening window unless transmittance is set
• After warm-up and recommended NUC interval

Field measurements on low-ε metals, through dirty IR windows, or with wrong RAT are outside the accuracy claim. Calibration does not rescue bad setup.

Field Checks: Verification, Not Full Calibration

Level II practice includes field verification so you catch gross drift, wrong range tables, fogged optics, or user error before you issue a quantitative report.

Practical field-check methods

1. High-ε reference patch — Apply known high-emissivity tape or flat black paint (ε ≈ 0.95) to a stable target, set camera ε to match, set RAT correctly, and compare to a contact probe on the same patch after thermal equilibrium.
2. Portable blackbody or calibrated hot plate — If your program owns a field blackbody, measure at one or more set points within the range you will use that day.
3. Ice-point / known process reference (when available) — A well-stirred ice bath is a classic 0 °C contact check for probes; for cameras you still need a high-ε surface at a known temperature (the ice itself is not a perfect IR target geometry for every FOV).
4. Side-by-side camera comparison — Two recently calibrated cameras on the same high-ε target should agree within combined uncertainty; large disagreement flags a problem.

Document the check: date, camera serial number, range, ambient, reference temperature, IR reading, difference, and pass/fail against your program tolerance (often “within manufacturer accuracy” or a tighter internal limit such as ±1 °C near ambient).

What a field check can and cannot do

When to Send the Camera for Laboratory Calibration

Send (or schedule) laboratory recalibration when:

• The manufacturer’s recommended interval is due (commonly annual for many industrial programs; always follow OEM and client QA requirements)
• A field verification fails or shows progressive drift across checks
• The camera suffered drop, shock, moisture intrusion, or lens damage
• You changed or repaired optics, detectors, or boards that affect radiometric paths
• Contract, NETA-style testing program, or ISO quality system requires a current certificate for quantitative work
• Results near a go/no-go absolute limit must be defensible in an audit

After lab calibration, retain the certificate (serial number, date, ranges calibrated, uncertainty, as-found/as-left data if provided). Level II reports that claim absolute temperatures should be able to answer: “Was the instrument within calibration when this image was taken?”

Level II Responsibility

Level I may operate a camera under procedure. Level II is expected to own measurement quality:

1. Know the accuracy budget — Manufacturer accuracy + emissivity uncertainty + atmosphere/window + size-of-source + operator setup. Do not report three-decimal temperatures that the accuracy band cannot support.
2. Select range and focus deliberately — Wrong range can put the target on a poorly characterized part of the curve or near saturation.
3. Refuse quantitative claims when the camera is out of cal, the target is unresolved (IFOV/spot-size violation), or ε/RAT are unknown.
4. Document calibration due date or certificate ID in the equipment section of quantitative reports when client QA requires it.
5. Train and QA Level I work so pretty images with expired calibration certificates do not become “official” temperatures.

Exam-style scenarios

Scenario A — Spec interpretation. A camera is specified ±2 °C or ±2% of reading. Measuring a 300 °C process surface (high ε, correct setup), the accuracy band is dominated by 2% → about ±6 °C. Choosing “always ±2 °C” is wrong.

Scenario B — NUC vs lab cal. A technician says “I did NUC this morning, so we are calibrated.” Correct Level II answer: NUC/internal shutter corrects uniformity and short-term drift; laboratory blackbody calibration establishes absolute accuracy.

Scenario C — Failed field check. High-ε patch at 45 °C contact; camera reads 58 °C after correct ε and RAT. Action: stop quantitative reporting, check range/focus/optics, re-verify; if still out, remove from service for lab investigation—do not “adjust the report by 13 °C” without understanding the fault.

Scenario D — Contract requirement. Client requires annual NIST-traceable calibration. Your camera’s last cert is 18 months old but field checks look fine. For that client’s quantitative work, lab recert first; field checks alone do not satisfy the contract language.

Common Traps

Summary for Recall

Cameras are calibrated against blackbody sources so in-band radiance maps to temperature. Accuracy is often ±X °C or ±Y% of reading, whichever is greater—percentage grows important at high temperature. Field checks with high-ε references or portable blackbodies verify performance; laboratory calibration establishes traceable accuracy. Level II owns the decision to measure, document, or ground a camera, and never confuses pretty thermal contrast with certified absolute temperature.