eCO₂ is an estimate from a metal-oxide VOC sensor such as SGP30. AirGradient writes that a deodorant can raise it. Adafruit writes that the SCD-30 is NDIR true carbon dioxide and “isn’t approximating it from VOC”. A title that says CO₂ does not tell you which of those two you have.
Two different instruments
Adafruit’s SCD-30 tutorial states the module is non-dispersive infrared true carbon dioxide, unlike SGP30, which approximates the number from VOC. The SCD-40 / SCD-41 page states photoacoustic true carbon dioxide and again separates it from SGP30 estimation. Those pages repeat manufacturer specifications. They are not Adafruit co-location tables, and they are not AirMeasure UK measurements.
RevK’s 2022 hardware note makes the same split in plainer language: cheap “CO₂” sensors measure something else; the true sensors he named were SCD30 and SCD41.
Why eCO₂ fails as a ventilation number
AirGradient’s 2024 photoacoustic article, a manufacturer test, writes that eCO₂ from SGP30 is highly unreliable and that a deodorant can lift the reading. That is a VOC event, not a closed-window occupancy event. HSE’s workplace page is written around NDIR. An eCO₂ number cannot be dropped into that page’s 1500 ppm ventilation line.
The same AirGradient article reports that indoor two-week tracks of SenseAir S8 and SCD40/41 stayed close, while outdoor scatter was larger for the photoacoustic pair. Those R² figures stay attached to that manufacturer post. This publication has no outdoor or indoor co-location of its own, so it does not rewrite them as a rule that “SCD41 cannot be used outdoors” or “S8 is always closer”.
What these pages do not say
These pages do not say that a named retail unit has true carbon dioxide. That claim would need the NDIR or photoacoustic model in a handbook or teardown, and a second true-CO₂ instrument beside it. Those files are not here. On the pages observed, the title printed CO₂; the body did not name NDIR or photoacoustic. On the related temtopair.co.uk pages, the body denied a gas sensor altogether.
Sources and scope
Sources checked on 23 September 2026. Numbered references point to the original publishers. Their sample measurements are not AirMeasure UK measurements. Manufacturer, tutorial and official material is identified separately.
- Adafruit Learn, Adafruit SCD-30 NDIR CO₂ ↗Tutorial stating the SCD-30 is NDIR true carbon dioxide: “Unlike the SGP30, this sensor isn’t approximating it from VOC.” Field recalibration needs a known concentration. Automatic self-calibration is written as needing regular exposure near 400 ppm, continuous power, and about seven qualifying intervals of at least 18 hours (about five days). Component retailer, repeating a manufacturer specification.
- Adafruit Learn, Adafruit SCD-40 and SCD-41 ↗Photoacoustic true carbon dioxide, again distinguished from SGP30 estimation. Range and stated accuracy are the manufacturer’s figures for SCD-40 / 41 / 43. Not an Adafruit co-location table.
- AirGradient, Photo-Acoustic vs NDIR CO₂ sensors ↗Achim Haug, 22 September 2024. Manufacturer science post. Indoor two-week tracks of SenseAir S8 and SCD40/41 were close. Outdoors, two S8 units correlated at R² 0.90 and two SCD units at 0.36; versus a reference, S8 about 0.64–0.69 and SCD about 0.16–0.24. eCO₂ from SGP30 is written as highly unreliable; a deodorant can raise it. These are AirGradient’s tests, not an independent laboratory and not AirMeasure UK numbers.
- RevK, Environmental sensors — SCD41 CO₂ ↗8 March 2022. Cheap “CO₂” sensors are written as measuring something else. True sensors named are SCD30 and SCD41. Board heat affects on-board temperature; bedroom temperature used an external DS18B20. Hardware notes, not a medical series.
- HSE, Using CO2 monitors ↗UK workplace ventilation page. Prefers NDIR on a portable device; head height; more than 50 cm from people; away from doors, windows and supply vents. A single snapshot can mislead. Persistently above 1500 ppm is written as a sign of poor ventilation. In a room with an air cleaner, carbon dioxide does not stand in for ventilation. Workplace guidance, not a home health standard and not an instrument certificate.
No instrument in this note has been tested or certified by this publication. Evidence policy · Corrections
Write the conditions beside the reading.
The CSV has column headings and one empty row. It contains no fabricated sample readings.
Download air-quality event record ↓
