odd…seems like a series of sensor have all shifted the range they read. Granted its a BME680, but why would multiple sensors around the house suddenly all read 3-5x the value of the previous day. not all are doing it. Why would just a single parameter from a sensor, suddenly shift scales, across multiple devices, when not all parameters from a single device are effected? No the environ is NOT what the sensors are saying. CO2 has been hovering around 400-1000 on three devices in particular. Seemly overnight, they now are reading in the 2000-3500 range. Other sensors nearby are not shifted. currently at esphome dev 2025.5.0-dev20250423
This issue seems to have reared its head 4/23///not sure if I updates the core or not . Any clues as to what might have shifted these sensor so fast?
The BME680 performs auto cal, so that might have something to do with it.
As the saying says, “a man with one clock/sensor always knows what time/value it is, a man with more is never sure.”
Like neel-m wrote above, auto calibration expects that in last x days sensor has seen fresh air (~400). But if your sensors show higher than expected values, it means they should have been exposed to LOW/CO2 free atmosphere. The tricky part is that BME doesn’t sense CO2, but others pollutants in air, so you need deeper research.
Where you get those values? Esphome log?
interesting readings… I have multiple CO2, TVOC, particle, and temp/hum sensors etc. I’ve been running for a couple months on the bench just to look at accuracies, programming, long term stability…etc.
A cou8ple devices, sense NDIR CO2 sensors along with the BME (and others) that calc co2 fromany and all VOCS they sense.
ALL the BME680s, regardless of esp device on the bench, jumped from 400-1000 avg to well over 2500-3500. Yet, all the NDIR and other TRUE CO2 sensors are all still reading the proper ranges (sub 500) at the same time.
So… I’d asked the wife if she’d cleaned the house recently (fumes) or sprayed anything. Then the next day we se on the news about the New Jersey Pine barrens fire, and how it’s effecting Maryland’s weather/pollution. So I have to assume these sensors are seeing what ever it is in the air that burned, as normal ‘real’ co2 sensors are still reading just fine
Note the device names. All the 680s are WAY high (except one in the Living room) the SCD30 and MZ-19are reading correctly (well… close enough)
The Soil-Moisture-Sensor (SCD30) is in a greenhouse and doesn’t seem to be effected
Some nasty VOC that bme is particularly sensitive to…
Soil moisture? Not a best pick for that purpose… 
ha… that reference is an SCD30 in the device/assy that houses a couple A/D muxs handling reading a multiple plant soil moisture sensors…hence the name of the orig device…before more and more sensors got added…