Air Quality Monitor Accuracy: Calibration, Drift, and Verification

Volume I  ·  May 2026  ·  753 words

An air quality monitor that is consistently wrong is worse than no monitor at all — it provides false reassurance or false alarm. Sensor accuracy degrades over time through multiple mechanisms, and the calibration methods available to consumers are limited. This article explains what accuracy to expect, how to verify it, and when readings can inform health decisions.

Accuracy Specifications: What the Numbers Mean

The "± 10 µg/m³ or ± 15%" specification for PM2.5 means that at a true concentration of 10 µg/m³, the monitor may read anywhere from 0 to 20 µg/m³. This is a wide range — wide enough that "good" (0–12) and "moderate" (12.1–35.4) air quality categories overlap in the measurement uncertainty. Consumer PM monitors are useful for detecting trends and events (smoke intrusion, cooking emissions) but not for regulatory compliance or health-effect studies requiring accurate absolute concentrations.

Calibration Methods

CO₂: Automatic Baseline Calibration (ABC)

Most NDIR CO₂ sensors implement ABC: the sensor assumes the lowest CO₂ reading over a rolling 7–14 day window represents outdoor ambient (~420 ppm) and adjusts its calibration to match. This works if the monitor is in a room that regularly achieves ambient CO₂ — a room with windows opened daily or a room that is unoccupied for part of each day. It fails in continuously occupied spaces where CO₂ never drops to ambient, causing the calibration to drift upward over weeks.

For spaces that never reach ambient (sealed buildings, continuously occupied bedrooms), manual calibration is required: take the monitor outside for 20–30 minutes, allow it to stabilize, and trigger manual calibration to 420 ppm if the monitor supports it. Repeat every 3–6 months.

PM: Factory Calibration Only

Consumer PM sensors are factory-calibrated against a reference instrument using a standardized test aerosol (typically Arizona Road Dust or polystyrene latex spheres of known size). There is no consumer-accessible recalibration procedure. When the sensor drifts — from dust accumulation in the optical chamber, laser diode aging, or fan degradation — the readings become progressively less accurate. Cleaning the sensor chamber with compressed air (per manufacturer instructions) restores accuracy if the drift is caused by dust accumulation. If the laser diode has aged, the sensor must be replaced.

VOC: The Uncalibratable Sensor

MOS VOC sensors cannot be calibrated by the consumer. The baseline drifts, the sensitivity to different VOC species varies by an order of magnitude, and the sensor output is a dimensionless index, not a concentration. A TVOC reading of "500 ppb" on a consumer monitor means the sensor's algorithm estimates 500 ppb — but the true concentration could be 200 ppb or 1,500 ppb depending on which VOCs are present and how long it has been since the sensor's last baseline reset. VOC readings from consumer monitors should be treated as qualitative (high, medium, low), not quantitative.

Verification Without a Reference Instrument

You can verify monitor behavior without access to a $15,000 reference instrument:

• CO₂: the exhaled breath test. Exhale gently onto the monitor from 30 cm. Human breath contains ~40,000 ppm CO₂. The reading should spike to 2,000–5,000 ppm and recover to ambient within 1–2 minutes. If it does not spike, the CO₂ sensor is not functioning.
• PM: the match test. Light a match, blow it out, and hold the smoking match 30 cm from the monitor. PM2.5 should spike to 100–500 µg/m³. If it does not, the PM sensor's fan or laser may have failed.
• VOC: the alcohol wipe test. Open an isopropyl alcohol wipe near the monitor. The VOC reading should spike within seconds. If it does not, the MOS sensor may be poisoned or failed.

These are qualitative tests — they confirm the sensor responds, not that it is accurate. A sensor that passes all three tests may still be off by 50% in absolute terms.