When we started building the AirGradient Go, we had a clear idea of what we wanted: a portable, open-source air-quality monitor that is transparent in its design decisions and genuinely useful outside a lab. The Go is meant to be carried - from room to room, into a restaurant, onto the street, into a car, back home. Portability is the entire point.
Great investigation, and quite striking results (I’d have expected much closer behaviours). I really hope the S12 is the device chosen in the end, as it appears to be the clear better option for a mobile product per this analysys.
But talking about calibration with “outside air” – isn’t it about time we updated the “reference” to 430 or even 440 ppm instead of 400 ppm? The last time average world levels were at that point was in 2016, and as you know there’s this thing going on where it seems to keep growing year after year… latest measurements being consistently over 430 ppm:
I realise it doesn’t matter much for measuring air quality indoors, where we’re looking at thousands of ppm, but still… accuracy matters 
Great article. Love seeing team show test results and building in public. Also hope S12 is chosen. I’d be willing to wait extra time for a better sensor
Yes, we will be going with the S12 as it’s clearly the best sensor from this comparison
Great point! Ambient CO2 levels can also vary slightly depending on location (this map is very helpful for this: https://earth.nullschool.net) so perhaps an adjustable baseline would be a nice feature to have?
In the event the S12 isn’t available in time, will the sensor module be user upgradable?
Re: user adjustable baseline, I personally don’t think it matters for consumers in typical application. The measurement error is already 1-2x any difference in “actual” baseline to the 400ppm calibration baseline. Also as you note, how can we tell what the “actual” baseline should be in our specific location, and when on the move?
Love this kind of post too!
Great to see the community is interested in CO2 sensing. Keep up the good work.
Small addition to the report. AS per the specs Sensirions sensor used was the SCD41, there are 4 versions of SCD4x.
On the picture of the board I could not make out a humidity and temp sensor for compensation in the STCC4 measurements. In the datasheet this is clearly requested. Was this implemented somehow?
Hi @ebneterc, the AirGradient Go has an SHT40 for measuring temperature and relative humidity. You can see more detail in the technical specification table here: https://www.airgradient.com/portable/
Implementation was done with direct connection of the SHT40 to the STCC4 as per specification. We double triple checked that after we got these weird results from the STCC4 but both hardware and firmware as implemented correctly.