How Air Quality Sensors Are Calibrated

Factory calibration takes place under controlled conditions before an instrument is shipped, and again after every annual return. The typical sequence for a gas sensor is: stabilise the instrument in a temperature- and humidity-controlled chamber; expose to certified zero gas (typically N₂ or scrubbed air); record the zero response and adjust; expose to one or more certified span gases at concentrations covering the working range; record span responses and derive a calibration curve; verify against an additional check-point not used in derivation.

For PM sensors, factory calibration uses a reference aerosol (often a polydisperse salt or Arizona Road Dust) generated at known mass concentration, with a co-located reference gravimetric or beta-attenuation sampler providing the truth value.

Factory calibration certificates should state the standards used, their uncertainty, the temperature and humidity conditions, and the resulting instrument uncertainty as a function of concentration. A certificate that lists "calibrated" with no numbers is not useful.

Between factory calibrations, field checks confirm an instrument has not drifted out of tolerance. The standard procedure for gas sensors is the zero and span check.

Zero check. Expose the sensor to a known zero gas (or scrubbed ambient air) and confirm the reading is within the expected zero offset.

Span check. Expose the sensor to a known concentration of the target gas and confirm the reading is within tolerance — typically ±10 % of the certified value.

If either check fails the sensor is either field-adjusted (if the manufacturer's procedure allows it) or pulled for factory recalibration. The check itself, the result and the action taken go into the maintenance record.

Co-location is the gold standard for characterising indicative-grade sensors. The procedure: install the indicative sensor immediately adjacent to a reference-grade instrument (ideally MCERTS-certified for the parameter of interest); operate both for a minimum of four to six weeks covering a representative range of conditions; pair the time-aligned data and fit a regression model.

The regression — usually a linear model with humidity and temperature correction terms — becomes the correction function applied to the indicative sensor's data. Properly co-located low-cost PM sensors can achieve agreement within 15–25 % of a reference monitor across normal ambient ranges, which is sufficient for many indicative applications and qualifies for MCERTS Indicative status. See sensor calibration for the service-level methodology.

Co-location is also the only way to validate a sensor against the specific aerosol mix in a particular location — generic factory calibration cannot predict the response to wood-smoke versus road-dust dominated environments.

Every sensor drifts. The mechanisms differ: NDIR sensors drift through optical contamination and source ageing; electrochemical cells drift through electrolyte depletion; MOS sensors drift through poisoning of the sensing surface; optical PM sensors drift through laser ageing and inlet fouling.

Drift is managed by: monitoring sensor self-diagnostics where available (most NDIR sensors report a quality flag); periodic field verification against a known reference; running correction factors derived from co-location; and scheduling factory recalibration before drift exceeds the tolerance the application requires.

Verification is distinct from calibration. A verification check tells you whether the instrument is currently within tolerance, without adjusting it. A successful verification means you can trust today's data; an unsuccessful verification means yesterday's data and today's data are both suspect until the instrument is re-calibrated.

Validation is the umbrella exercise that demonstrates an instrument is fit for purpose across its working envelope. It draws on calibration certificates, verification records, drift histories and any uncertainty analysis specific to the deployment. For regulated work (MCERTS, BREEAM evidence, expert reports) validation documentation is required, not optional.

A working maintenance record for every monitor in a fleet should include: factory calibration certificate and date; field check schedule and results; firmware version history; site location and any environmental anomalies; co-location data and derived correction functions; and a clear ownership trail. Without this, the data the fleet produces is hard to defend if challenged.

For the deployment, calibration and validation service we provide to UK clients, see sensor calibration, air quality sensors, air quality monitors and air pollution sensors.

Calibration cannot rescue a sensor used outside its design range, a sensor in an environment where its cross-sensitivities dominate, a sensor whose inlet is obstructed, or a sensor whose siting is wrong. Bad data from a calibrated sensor is still bad data. The metrology programme is necessary but not sufficient — it has to sit alongside a deployment plan that respects the physics of the sensors involved.

• What is the difference between calibration, verification and validation?

  Calibration adjusts an instrument's response so its output matches a reference. Verification confirms an instrument's output is within a stated tolerance of a reference, without adjustment. Validation is the broader exercise of demonstrating an instrument is fit for purpose across its operating range. The three are commonly confused, and the resulting paperwork conflates them too.

• How often does a CO₂ sensor need recalibration?

  Dual-wavelength NDIR sensors typically hold calibration to within ±50 ppm for 12 months. Single-wavelength sensors with automatic baseline correction maintain themselves provided they see outdoor air weekly, but should be field-checked quarterly. Heavily contaminated environments accelerate drift.

• Can sensors be calibrated on-site or do they need to go back to the manufacturer?

  Both have a role. Factory calibration uses certified gas mixtures and reference instruments in temperature- and humidity-controlled environments — the gold standard for traceability. Field calibration uses portable reference gas and co-located reference monitors — faster, cheaper, and the right approach for a fleet between annual returns.

• What is co-location and why does it matter?

  Co-location is the side-by-side operation of an indicative sensor and a reference instrument for a period long enough to characterise the indicative sensor's response across environmental conditions. The resulting correction function is what turns indicative data into something defensible. It is the methodology behind MCERTS Indicative and most credible low-cost networks.