Hacker News new | ask | show | jobs
by reddozen 25 days ago
CO2 and all other air quality indexers have to be very carefully calibrated regularly. It's not some slop you can just throw into a consoomer cheap iot device.

Article author completely ignores this for the obvious reasons.

2 comments

For the purposes of indoor ventilation monitoring you can calibrate by occasionally exposing the sensor to fresh air. Either taking it outside or just the room not having people in it. The sensor will treat the lowest reading it gets as 400ppm since this is what outdoor air is.

A sensor mounted in the office will get calibrated every night when the office is empty.

The outdoor CO2 is rising every year. It is not fixed at 400 ppm. The calibrations you speak of are fake. A good sensor can be expected to remain within 10% of reference for ten years.
The outdoor CO2 is rising every year. It is not fixed at 400 ppm.

But close enough for most purposes. We aren’t doing laboratory measurements here, I just want to know whether or not to open a window.

You don't need to routinely calibrate a good device at all for that purpose. It already will maintain a reading within 10% of the real value. If it's not doing that, it's a bad device. It's very possible your calibration process will actually miscalibrate it and increase the error. Already the outdoor CO2 level is 430, not 400, so you'd be introducing at least a 7% error by calibrating it to 400.

What I do at home is I have multiple meters bought over the years, not all at once. If one of them is too deviated, I can replace it, but this deviation has never happened in the last five years. It did happen once about ten years ago with an old model.

If you want to make measuring CO2 your obsessive compulsion then sure, buy a lab grade instrument. If you want to get a practical reading of the air quality that is more than good enough, affordable, and looks nice on your bedside, the ikea one is a great option.
Not really. For ventilation purposes, a good sensor remain within 10% variation for nearly ten years. We are not running a controlled science experiment here.