I designed a case for my ESP32 and BME680. Unfortunately, the ESP32 radiates heat that influences the readings of the temperature sensor.
I thought that my design would have a good enough air flow and I added a little wall to keep the heat away from the sensor.
I also have a dfrobot 0610 presence sensor in there facing the living room. I visited the link you sent and I am not experiencing erratic temperature swings using sample_rate: LP and it seems accurate after I managed to fine-tune the temperature_offset.
I use some Roof insulation to stop heat getting to the sensor. Don’t rub your eyes after handling it. I tend to use bmp280. Heat correction usually less than 1 degree C.
I note that the latest ESPHome update includes the GitHub update for the BME’s and it is supposed to specifically address temperature over-statement. After the update and calibration I still need to use temperature_offset: 5 vs 5.2. Not much of a change.
I note that the humidity is wildly lower than 2 other humidity sensors I have deployed so I am doubting the accuracy of the sensor itself.
I face the same issue. If you put in a temperature offset for the BME-component, there is a ‘correction’ of the humidity, as they assume that absolut humidity is still the same.
What version did you have before the latest update? Which of the 3 BME components have you implemented? I wonder because I would provide that info on GitHub, so that they can check the code for that version.
In my case, temperature is very accurate with the new update - as long as I don’t put the sensor in my enclosure. Heat from the ESP32 is causing misreadings. I will make another version of my enclosure an the set an offset for humidity only.