Almost every fixed carbon dioxide detector sold into industrial safety is built around an NDIR CO2 sensor – a non-dispersive infrared measuring cell. That follows from the gas itself: carbon dioxide is not flammable, so a catalytic bead has nothing to burn, and the electrochemical cells that serve carbon monoxide or hydrogen sulphide well offer no practical, long-lived response to CO2. Infrared absorption is the one principle both specific to carbon dioxide and stable enough for continuous, unattended operation.
Why NDIR Is the Default for Fixed CO2 Detection
The physics is simple. Carbon dioxide absorbs infrared energy strongly in a narrow band near 4.26 micrometres. An NDIR cell passes infrared light through a chamber of known path length and measures how much energy survives at that wavelength: less energy means more CO2. Because the measurement is optical, nothing in the sensing path is consumed by measuring. A catalytic or electrochemical element is chemically involved in its own measurement, so it has a finite service life by design and can be poisoned by silicones, sulphur compounds or solvents. An infrared bench is far less troubled by these, and it is indifferent to oxygen concentration – which matters in inerted vessels and controlled-atmosphere storage.
Most industrial cells are dual-channel: one channel filtered to the CO2 absorption band, a second reference channel where CO2 does not absorb. Comparing the two lets the electronics separate a real concentration change from lamp ageing, window fouling or gain shift. Single-channel cells are cheaper and common in comfort-ventilation products, but cannot distinguish a dirty window from real gas – a real limitation in a dusty plant.
Range Selection: ppm or Percent by Volume
Measuring range is the most consequential specification decision, and the one most often made wrong. Fixed CO2 detectors divide into two families.
ppm-level instruments resolve concentrations from hundreds to a few thousand parts per million. This is the band that matters when the hazard is gradual accumulation in an occupied space and the reference points are occupational exposure limits – in many European jurisdictions an eight-hour limit around 5000 ppm, or 0.5 percent by volume, with a short-term limit several times higher. Clean outdoor air sits around 400 to 430 ppm, so the instrument must distinguish a modest rise above that background with confidence. Laboratories, plant rooms and occupied basements where a slow leak is the realistic scenario are covered this way.
Percent-volume instruments cover ranges such as 0 to 5 percent or higher, and exist for a different hazard: bulk release. A ruptured storage line, a carbonation room, dry-ice handling or discharge from a CO2 suppression system does not produce a gentle ppm-level rise – it displaces breathable air in seconds, and a ppm-range detector saturates immediately.
Specifying one wide-range instrument to cover both costs resolution where it is most needed, because a cell calibrated across a large span cannot resolve small changes near the bottom of it as confidently. Where both hazards coexist, two detectors with different ranges reporting to a shared controller is the more defensible arrangement, and a site’s fixed CO2 detection points are worth reviewing against this split before a tender closes. In confined and inerted spaces, CO2 measurement is commonly paired with an oxygen deficiency monitor, since displacement is what actually harms people.
Accuracy and Drift Over Time
Accuracy on a CO2 datasheet is usually a fixed term plus a proportional term – so many ppm, plus a percentage of reading. Near the bottom of a range the fixed term decides whether a small rise above ambient is real; near the top the proportional term dominates.
Three environmental influences matter in a fixed installation. Temperature affects source output and detector response, which is why a stated accuracy figure is valid only across a stated temperature band. Pressure affects the reading directly, because absorption depends on the number of molecules in the optical path rather than the ratio between them, so the same concentration reads differently at altitude. Third is condensation: water on an optical window scatters light and can look like gas.
Drift has two components. Span drift comes mainly from source ageing and contamination of the optics, and a reference channel compensates for much of it. Zero drift deserves more respect: it is slow, undramatic, and shifts the whole curve rather than breaking it, so a drifting instrument still looks healthy on the display. Only comparison against a known gas reveals it, which is why scheduled calibration with certified CO2 mixtures remains the basis of a credible maintenance programme.
Why Automatic Baseline Correction Belongs in Offices, Not Plants
Consumer and building-ventilation sensors commonly use automatic baseline correction, often labelled ABC. The algorithm tracks the lowest reading seen over a rolling window of days or weeks, assumes that minimum is fresh outdoor air at roughly 400 ppm, and quietly re-zeros to it. In an office that empties overnight and is ventilated each morning, the assumption is broadly true and the feature saves service visits.
In an industrial setting it frequently fails, and it fails in the dangerous direction. A fermentation cellar, brewery, greenhouse under CO2 enrichment, cold store using dry ice or carbonation room may never return to outdoor background – not overnight, not over a fortnight. If the true minimum is 2000 ppm and the algorithm insists it must be 400, the sensor subtracts the difference and every later reading is understated by roughly 1600 ppm. The detector reports normal conditions while the space is loaded, and the error grows with each cycle.
For continuous industrial monitoring the approach should be the opposite: leave the baseline alone and verify it deliberately. That means a defined calibration interval, zeroing with nitrogen or verified CO2-free air rather than assumed room air, span checking with a certified mixture near the alarm threshold, and any automatic baseline feature confirmed as disabled on delivery. A gas alarm controller that logs readings and alarm events makes that history straightforward to produce when an auditor asks.
Why ASA
ASA supplies fixed gas detectors, dust alarm detectors and gas alarm controllers as one product line, so a site needing CO2 alongside oxygen, combustible or toxic gas monitoring is covered from a single source with one technical contact. Our AN-CO2-D fixed carbon dioxide detector holds China Ex Certificate EETI24.0463X (Ex db IIC T6 Gb, IP65), valid to Jan 2030, for installations where the surrounding area is classified on account of other flammable gases present. Because we manufacture and stock our core detector models, a late range change does not become a lead-time problem, and our engineers review point lists and controller architecture before an order is placed.
If you are specifying CO2 detection and are unsure whether a ppm or percent-volume range fits the hazard, send the ASA team your site details – process, space volume and ventilation – and we will come back with a recommendation.
Disclaimer: Product availability, specifications, measuring ranges, alarm configurations and certification coverage vary by model and region. The information in this article is provided for general reference only and does not constitute a performance guarantee, safety assurance or compliance statement. Always refer to the official product documentation and confirm suitability, certification status and local regulatory requirements with our team before selection, installation or use.
