Controlling power to BME280 sensor with ESPhome

Third party integrations
1

Not sure if I’m close to a solution or a million miles away, but in short BME sensors generate a little bit of heat when they’re powered. I’ve found the most reliable way to get a good measurement from them is to power them up briefly, take a measurement, then disconnect to let them reach ambient again before the next attempt.

Here’s what I have so far:

sensor:
  - platform: bme280
    id: bme_sensor
    temperature:
      name: "BME280-garage Temperature"
      oversampling: 1x
    pressure:
      name: "BME280-garage Pressure"
      oversampling: 1x
    humidity:
      name: "BME280-garage Humidity"
      oversampling: 1x
    address: 0x76
    update_interval: never
    
  - platform: template
    id: dummy_sensor
    lambda: |-
      digitalWrite(5, HIGH);
      delay(3000);
      id(bme_sensor).update();
      digitalWrite(5, LOW);
      return NAN;
    update_interval: 10s

I can confirm this sends power to the sensor for the 3 seconds, but no data is being returned to home assistant (I’m using MQTT for reference). I’m pretty new to HA and ESPHome but not new to IoT/electronics/sensors. Hoping someone can spot some dumb newbie error I’m making or just knows a smarter way to do this…

2

Narrowed it down a little: the i2c bus loses the sensor when it turns off. Any way to restart or reinitialize it?

Some possible useful code for others:

esphome:
  name: garage
  platform: ESP8266
  board: d1_mini
  on_boot:
    - lambda: |-
        pinMode(5, OUTPUT);
        digitalWrite(5, HIGH);

(some other stuff)

sensor:
  - platform: bme280
    id: bme_sensor
    #operation_mode: 2
    temperature:
      name: "BME280-garage Temperature"
      oversampling: 1x
    pressure:
      name: "BME280-garage Pressure"
      oversampling: 1x
    humidity:
      name: "BME280-garage Humidity"
      oversampling: 1x
    address: 0x76
    update_interval: never
    
  - platform: template
    id: dummy_sensor
    lambda: |-
      digitalWrite(5, HIGH);
      delay(5000);
      id(bme_sensor).update();
      delay(1000);
      digitalWrite(5, LOW);
      return {};
    update_interval: 10s

a fun error message that I can get:

[18:15:38][C][mqtt.sensor:028]:   State Topic: 'garage/sensor/bme280-garage_humidity/state'
[18:15:48][W][i2c:070]: Received NACK on transmit of address 0x76
[18:15:48][W][bme280.sensor:190]: Invalid temperature, cannot read pressure & humidity values
3

Are you using a BME sensor breakout board powered by 5V? If so, the heat is the regulator. Use a board without the regulator or bypass the regulator with a blob of solder and power from 3.3V. I ran into the same issue.

4

That would assume that the LDO is the source of the heat, which it does contribute to but is not the real problem. Example:

FLIR0038

This is a breakout board with the LDO removed, 3.3v being fed from a very low noise lab supply. The heat is exactly where the BME sensor is. Here’s that same board before removing the regulator being fed 5v, for reference:

FLIR0036

And showing the board with the modification to test:

FJIMG_20201022_144456
FJIMG_20201022_1444563120×4160 1.72 MB

The reality is these are active components, not like an NTC or something, and that means they generate heat when powered. There needs to be a way to remove that power for the time that the unit is not being measured to minimize the thermal buildup.

5

Interesting. My issue was solved via bypassing the regulator. Had maybe a 1 degree rise, which I consider acceptable.

Unfortunately I can’t help you with the ESPHome aspect as I don’t use it.