DHT11 temperature stops updating periodically

[chrispontiga]

Thanks for sharing this project. I recently set this up for my new network cabinet, but found that DHT11 temperature sensor stops updating its temp values periodically for 30-45 minutes at a time:

Without the temperature input, the fan speed also stays steady, then kicks up to catch up once the temperature updates again.

The logs during these states display the DHT11 returns the exact same value repeatedly, without error.

[00:05:14][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:14][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.48193 % with 1 decimals of accuracy
[00:05:14][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:15][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:15][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.47862 % with 1 decimals of accuracy
[00:05:15][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:17][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:17][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.47533 % with 1 decimals of accuracy
[00:05:17][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:18][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:18][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.47204 % with 1 decimals of accuracy
[00:05:18][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:19][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:19][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.46874 % with 1 decimals of accuracy
[00:05:19][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:21][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:21][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.46545 % with 1 decimals of accuracy
[00:05:21][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:22][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:22][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.46216 % with 1 decimals of accuracy
[00:05:22][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:23][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy
[00:05:23][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 85.45885 % with 1 decimals of accuracy
[00:05:23][D][sensor:094]: 'Humidity': Sending state 32.00000 % with 0 decimals of accuracy
[00:05:24][D][sensor:094]: 'Temperature': Sending state 31.80000 °C with 3 decimals of accuracy

I have yet to look into the signal integrity on the data line from the DHT11. But I have tried to swap sensors and got similar behavior. The only other things I could think of is connecting directly to the header on the ESP rather than the 3-pin JST header, or trying shorter wires to the sensor. Is a pullup resistor required on the data line?

Any other insights would be appreciated.

[patrickcollins12]

A few theories.

I notice that the update frequency in your log is more frequent than 1 second. Note the update_frequency below. Double check your update frequency is >1.3s, since the DHT11 doesn't have a high refresh rate.

esphome seems to be operating correctly.

Other likely issue is a faulty sensor. Since they're so inexpensive, I'd just try a new sensor.

Maybe wiring issues. You could try adding the recommended resistor to the sensor.

  - platform: dht
    pin: GPIO33
    temperature:
      name: "Temperature"
      id: console_fan_temperature
      accuracy_decimals: 3
      ...
    # the DHT11 can only be read every 1s. Use 1.3s to be safe.
    update_interval: 1.3s

[chrispontiga]

Thanks for the reply. The update interval is set at 1.3s from the original code. My feeling is that it is a signal integrity issue. I will occasionally experience warnings as follows:

[18:23:14][W][component:214]: Component dht.sensor took a long time for an operation (0.06 s).
[18:23:14][W][component:215]: Components should block for at most 20-30ms.
[18:23:15][D][sensor:094]: 'Temperature': Sending state 31.79999 °C with 3 decimals of accuracy
[18:23:15][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 20.55629 % with 1 decimals of accuracy
[18:23:15][D][sensor:094]: 'Humidity': Sending state 33.00000 % with 0 decimals of accuracy
[18:23:16][W][component:214]: Component dht.sensor took a long time for an operation (0.05 s).
[18:23:16][W][component:215]: Components should block for at most 20-30ms.
[18:23:17][D][sensor:094]: 'Temperature': Sending state 31.79999 °C with 3 decimals of accuracy
[18:23:17][D][sensor:094]: 'Fan Speed (PWM Voltage)': Sending state 20.55299 % with 1 decimals of accuracy
[18:23:17][D][sensor:094]: 'Humidity': Sending state 33.00000 % with 0 decimals of accuracy
[18:23:17][W][component:214]: Component dht.sensor took a long time for an operation (0.05 s).
[18:23:17][W][component:215]: Components should block for at most 20-30ms.

I have tried two sensors already with similar results. I do have one more from the package I can try. I will also try a 10k pull up since that's minimal effort with some clip leads. I'll provide an update after some tinkering.

[chrispontiga]

Reporting back with my troubleshooting from this weekend. I have tried three different DHT11 sensor modules, but all appear to perform similarly where the temperature reading stops updating for some number of minutes. The 10k pull-up resistor made no noticeable effect, because I think the pull-up may already be included on the PCB of the module I am using.

The last DHT11 module I tried came from a different set ordered from Amazon. It appeared the HiLetGo-brand appeared to have the highest reviews. Using this module combined with shorter 3-inch jumpers wires (rather than the longer 6-inch wires I previous used) appeared to have an improved performance. I'm not sure if it s a better module or the shorter wires was the factor. But visually, the board looks identical to the no-brand-name modules I first used.

At this point I am unsure if the non-updating value is a defect of my board, or if it's actually standard operating function of the sensor. I would expect that the sensor to continue to provide new values since it reports 3-decimal point accuracy and it's highly unlikely that my networking cabinet is remaining to a thousandth-of-a-degree temperature stable. But I was thinking that perhaps the DHT11 has a minimum sensitivity and requires a change of several hundredths of a degree before reporting. I couldn't find any information to confirm with its very limited documentation. I ordered a BME280 module for delivery tomorrow to see if performs any better.

[patrickcollins12]

that graph shows it reducing to 88.2 not to 0.0?

My DHT-11

Perhaps your ESP32 board?
Is it really logging during that period or is the ESP32 locked up with memory issues during that period?

[patrickcollins12]

Try reducing the update frequency to every 10 seconds. Try a better sensor like the BME280

https://randomnerdtutorials.com/dht11-vs-dht22-vs-lm35-vs-ds18b20-vs-bme280-vs-bmp180/

[Tntdruid]

Use a https://esphome.io/components/sensor/sht3xd.html for better data the dh11 is not that great.