MLX90614 and it's weird readings

ESPHome
1

Hello! I have some problems with MLX90614. It was working perfectly fine on it’s own being an only device on the I2C. That MLX90614 breakout board has internal 10k pull ups

i2c:
  - scan: True
    frequency: 100kHz

sensor:
  - platform: mlx90614
    update_interval: 20s
    object:
      name: Object

[17:30:53][I][i2c.arduino:096]: Results from i2c bus scan:
[17:30:53][I][i2c.arduino:102]: Found i2c device at address 0x5A

[17:30:56][VV][scheduler:226]: Running interval 'update' with interval=20000 last_execution=74490 (now=94491)
[17:30:56][VV][i2c.arduino:176]: 0x5A TX 24
[17:30:56][VV][i2c.arduino:148]: 0x5A RX FFFFD6
[17:30:56][VV][i2c.arduino:176]: 0x5A TX 07
[17:30:56][VV][i2c.arduino:148]: 0x5A RX 5739CE
[17:30:56][VV][i2c.arduino:176]: 0x5A TX 06
[17:30:56][VV][i2c.arduino:148]: 0x5A RX 3B39D1
[17:30:56][D][mlx90614:112]: Got Temperature=20.4°C Ambient=19.9°C
[17:30:56][V][sensor:043]: 'Object': Received new state 20.430000
[17:30:56][D][sensor:094]: 'Object': Sending state 20.43000 °C with 1 decimals of accuracy
[17:30:56][VV][api.service:140]: send_sensor_state_response: SensorStateResponse {
  key: 3055412916
  state: 20.43
  missing_state: NO
}

I added VL53L0X and AHT25 on another separate I2C just to be safe since everything has gone wrong with my I2C bus projects in the past. Things started happening. When I added VL53L0X to the captive portal the MLX90614 started give out errors on reading and VL53L0X gave out component failed. AHT25 worked just fine

captive_portal:
  
i2c:
  - id: bus_a
    sda: GPIO21
    scl: GPIO22
    scan: True
    frequency: 100kHz
  - id: bus_b
    sda: GPIO25
    scl: GPIO26
    scan: True
    frequency: 100kHz

sensor:
  - platform: mlx90614
    update_interval: 20s
    i2c_id: bus_a
    object:
      name: Object
  - platform: aht10
    i2c_id: bus_b
    variant: aht20
    temperature: 
      name: temperature
    humidity: 
      name: humidity
    update_interval: 25s
  - platform: vl53l0x
    name: VL53L0X
    i2c_id: bus_b
    update_interval: 30s

[20:10:40][C][i2c.arduino:071]: I2C Bus:
[20:10:40][C][i2c.arduino:072]:   SDA Pin: GPIO21
[20:10:40][C][i2c.arduino:073]:   SCL Pin: GPIO22
[20:10:40][C][i2c.arduino:074]:   Frequency: 100000 Hz
[20:10:40][C][i2c.arduino:086]:   Recovery: bus successfully recovered
[20:10:40][I][i2c.arduino:096]: Results from i2c bus scan:
[20:10:40][I][i2c.arduino:102]: Found i2c device at address 0x5A
[20:10:40][C][i2c.arduino:071]: I2C Bus:
[20:10:40][C][i2c.arduino:072]:   SDA Pin: GPIO25
[20:10:40][C][i2c.arduino:073]:   SCL Pin: GPIO26
[20:10:40][C][i2c.arduino:074]:   Frequency: 100000 Hz
[20:10:40][C][i2c.arduino:086]:   Recovery: bus successfully recovered
[20:10:40][I][i2c.arduino:096]: Results from i2c bus scan:
[20:10:40][I][i2c.arduino:102]: Found i2c device at address 0x29
[20:10:40][I][i2c.arduino:102]: Found i2c device at address 0x38

[20:54:37][VV][scheduler:225]: Running interval 'update' with interval=20000 last_execution=14188 (now=34188)
[20:54:37][VV][i2c.idf:197]: 0x5A TX 24
[20:54:37][VV][i2c.idf:173]: 0x5A RX FFFFD6
[20:54:37][VV][i2c.idf:197]: 0x5A TX 07
[20:54:37][VV][i2c.idf:173]: 0x5A RX BB2139
[20:54:37][VV][i2c.idf:197]: 0x5A TX 06
[20:54:37][VV][i2c.idf:173]: 0x5A RX A1B93B
[20:54:37][D][mlx90614:112]: Got Temperature=-100.4°C Ambient=nan°C
[20:54:37][V][sensor:043]: 'Object': Received new state -100.449997
[20:54:37][D][sensor:093]: 'Object': Sending state -100.45000 °C with 1 decimals of accuracy
[20:54:37][VV][api.service:140]: send_sensor_state_response: SensorStateResponse {
  key: 3055412916
  state: -100.45
  missing_state: NO

I tried to get the VL53L0X to work and got it to work with a custom component BUT MLX90614 never gets a reading read.

esphome:
  name: snow-depth-meter-tof
  friendly_name: Snow depth meter ToF
  includes:
  - tof_vl53l1x.h
  libraries:
    - "Wire"
    - "VL53L1x"

.
.
.

captive_portal:


i2c:
  sda: GPIO21
  scl: GPIO22
  scan: True
  frequency: 100kHz

sensor:
  - platform: bh1750
    name: "lux1_test"
    address: 0x23
    update_interval: 60s

  - platform: aht10
    variant: AHT20
    address: 0x38
    temperature:
      name: "temp test"
    humidity:
      name: "humid test"
    update_interval: 60s

  - platform: custom
    lambda: |-
      auto my_sensor = new MyCustomSensor();
      App.register_component(my_sensor);
      return {my_sensor};
    sensors:
      - name: "Distance"
        id: distance_raw
        accuracy_decimals: 0
        unit_of_measurement: "mm"

MLX90614

[22:19:44][VV][scheduler:226]: Running interval 'update' with interval=20000 last_execution=37494 (now=57495)
[22:19:44][VV][i2c.arduino:176]: 0x5A TX 24
[22:19:44][VV][i2c.arduino:148]: 0x5A RX 0909FF
[22:19:44][VV][i2c.arduino:176]: 0x5A TX 07
[22:19:45][VV][i2c.arduino:127]: RX 3 from 5A failed with error 0
[22:19:45][W][component:237]: Component mlx90614.sensor took a long time for an operation (1029 ms).
[22:19:45][W][component:238]: Components should block for at most 30 ms.

VL53L0X

[22:19:41][VV][scheduler:226]: Running interval 'update' with interval=15000 last_execution=39500 (now=54501)
[22:19:41][V][sensor:043]: 'Distance': Received new state 23.000000
[22:19:41][D][sensor:094]: 'Distance': Sending state 23.00000 mm with 0 decimals of accuracy
[22:19:41][VV][api.service:140]: send_sensor_state_response: SensorStateResponse {
  key: 1240670792
  state: 23
  missing_state: NO
}

After removing AHT25 ja VL53L0X components from the code, it shows different kind of error readings

[13:53:58][VV][scheduler:226]: Running interval 'update' with interval=20000 last_execution=1615739 (now=1635744)
[13:53:58][VV][i2c.arduino:176]: 0x5A TX 24
[13:53:58][VV][i2c.arduino:148]: 0x5A RX 0000F2
[13:53:58][VV][i2c.arduino:176]: 0x5A TX 07
[13:53:58][VV][i2c.arduino:148]: 0x5A RX A06A0F
[13:53:58][VV][i2c.arduino:176]: 0x5A TX 06
[13:53:58][VV][i2c.arduino:148]: 0x5A RX A06A19
[13:53:58][D][mlx90614:112]: Got Temperature=272.8°C Ambient=272.8°C
[13:53:58][V][sensor:043]: 'Object': Received new state 272.769989
[13:53:58][D][sensor:094]: 'Object': Sending state 272.76999 °C with 1 decimals of accuracy
[13:53:58][VV][api.service:140]: send_sensor_state_response: SensorStateResponse {
  key: 3055412916
  state: 272.77
  missing_state: NO
}

I have tried different cables, short and long. Different ESP32 devboards, different pins for I2C, different I2C devices but everything always goes south. Something might work on it’s own but when adding another device everything gets broken. At least with BH1750 and VL53L0X.

I have this project box which is going to measure the depth of snow coverage outside with VL53L0X, ground/snow surface temperature with MLX90614, lightness with BH1750 and AHT25 is going to measure temperature and humidity inside the device enclosure.

I have been on this project for over a month and I never get multiple devices to work properly. Or they might work for a while and they just randomly stop working or might stop working when I reboot the ESP32.

I am in the verge of forgetting about I2C completely. But the plug and play nature of I2C I get the feeling that there might be some kind of internal conflict with all of those. They might work just fine when all of the devices are connected to the ESP32 but the breaking happens when I add them in to the code. Sometimes it works for 5 minutes and then never. Sometimes it never boots and goes to a bootloop after listing every possible 7bit address on the i2c_scan.

2

It’s definitely not.
How is your wiring? I2C was developed for connecting ICs on same PCB.
Try your setup with max 30cm bus length.
You should use active terminator for longer wiring.
Interferences and crosstalk; use twisted pair wiring, both paired with gnd (SDA/GND and SCL/GND). And all connections have to be rock solid, I2C doesn’t support hot swapping.

3

The MLX is in the end of 100cm long 4 wire shielded cable. VL53L0X is something like 5cm away from the ESP32 and AHT25 is another 5cm from VL53L0X. Both AHT25 and VL53L0X are wired with 4 wire ribbon cable.

The MLX worked well with 10-100 kHz even when those other modules were connected - just when they were not added yet to the captive portal.

I will try that twisted pair thing soon. I gotta take a breather

4

What you mean with this? Captive portal is for fallback wifi…

More than that, make it shorter.
Also how are your pullups?

5

This is something that seems that I’ve known wrong. I meant the configuration after the captive portal. I though that fallback wifi is under wifi → ap

MLX90614 has it’s own 10k ones, and for VL53L0X I’ve added 4k7 ones externally.

I started to test with Arduino IDE and MLX90614 still showed weird over 200 degree temperatures, but after I desoldered VL53L0X and AHT25 from it, MLX90614 started to work normally

I now twisted the lines to VL53L0X (Or it’s actually VL53L0/1XV) and now it works with Arduino IDE. I asked ChatGPT to create me a test to print values which they all give out and it works with Adafruit_MLX90614.h, Adafruit_AHT10.h and Adafruit_VL53L1X.h.

MLX90614 Object Temperature: 23.65 °C, Ambient Temperature: 22.63 °C
AHT20 Temperature: 21.79 °C, Humidity: 64.62 %
VL53L1X Distance: 5.50 cm

With ESPhome VL53L1X custom component library bricked everything else from working but ToF sensor started to work

6

You should not twist the lines. You need to twist line with GND.

7

I mean I did as you mentioned. SDA/GND and SCL/GND paired

1 Like
8

Have you tried power cycling the ESP/MLX? It happened to me and that fixed it.