> it will return a Total Volatile Organic Compound (TVOC) reading and an equivalent carbon dioxide reading (eCO2) over I2C.

It's an estimated reading of CO2, so it's not a direct reading, the sensor has to do some math to estimate the CO2

It's not very precise and/or dependent on temperature, etc

I'm using an NDIR-based USB-CO2-meter (~40 €, allegedly USB only for power, but it actually spits out measurement data over USB) and I think it's pretty accurate. When I put it outside it generally reflects ambient CO2 pretty accurately (440-450 ppm) and I found through blind-testing that "air staleness" correlates very well with CO2 concentration. For me, air feels noticeably stale at around ~700-800 ppm, which seems largely inline with academic results.

You can actually see when the device acquires a reading, because the tube flashes.

I found this on what eCO2 is:

https://ec.europa.eu/eurostat/statistics-explained/index.php...

Re your other questions; yeah I gather there is some experimentation involved, even with out of the box commercial products.

If others have suggestions on chips or products to buy for the household, I would greatly appreciate it! The MH-Z19 looks like an option for me. Commercial products seem too unreliable for the price.

I think that is CO2e, which should be a different thing to my understanding. (CO2e is used to estimate the greenhouse gas potency of other gases like methane. Whereas I think eCO2 is some kind of estimate of amount of CO2 in the air, but what, exactly, I do not know)

Mine from Amazon is accurate enough to remind me of opening the window

Why not use a real one? MH-Z19 or Z14.

because my google-fu was not good enough to find those.... Regardless, I would be curious on the same questions about real world accuracy on those as well.

I have two of them, one inside and one outside. They both go down to roughly 400 ppm and the indoor one goes very high when it's all cooped up inside. The outdoor one fluctuates since I'm in an urban area. I believe they are accurate to within about 10%.

During the smoke event, they combined with my indoor/outdoor pm2.5 sensors have been extremely useful.

They should go down to 400ppm because the periodic autocalibration which is on by default assumes the lowest level seen in 24 hours is 400ppm.

You should be using the newer version, MH-Z19B, not MH-Z19.

The protocol is compatible though.

Also note they come in up-to-3000ppm and up-to-5000ppm variants. Be sure to get the latter.

In case it helps anyone: a very fast way to get it working and get graphs is to connect it to a Raspberry Pi running HomeAssistant. Takes a few clicks and 3 lines of config.

For portability, you can also hook up to an ESP8266/ESP32 and configure to talk to HomeAssistant via ESPhome: https://esphome.io/components/sensor/mhz19.html