[BUG] Plugin 135 - SCD4x freeze after some times

[iz8mbw]

Hello. using a SCD41 sensor on an ESP32 board, yes I use latest firmware of 1st April 2024.

What happens is after some times (some hours) the sensor freeze and from the web interface I see "0" as value and no more MQTT messages are published. See here the values on Home Assistant:

To recover it I need to restart ESP32 board or do a Power-Off / Power-On. On the logs I don't see nothing when the sensor freeze (no EVENT are generated).

Polling interval set to 60 seconds, "i2c scan" still detect the sensor.

I2C Scan result:

From the log, as you can see, no messages/events from SCD41 sensor (don't confuse the events you see that are referred to another sensor, the SGP41):

[tonhuisman]

Is this a LittleFS build (IDF 5.x) or SPIFFS (IDF 4.4)? Does this also occur with older builds?

[iz8mbw]

Hi Ton, it's a LittleFS build. Same issue with older build but I connected the sensor since some weeks so cannot told to you if a build of 2 months ago (for example) was working.

[TD-er]

How is the sensor connected? Length of the cables, for example. What I2C clock frequency?

[iz8mbw]

20 cm, i2c slow as per screenshot

[TD-er]

Have you also tried with the other I2C device disconnected? Just to make sure it is only an issue with the SCD4x sensor and not related to something else.

[iz8mbw]

No. I'll do more test if needed but cannot understand why no EVENTS are in the log when it is frozen.

[TD-er]

I have not yet any clue what is happening here. One of the options is that the sensor gets locked up. Or maybe the plugin is getting out of sync with the sensor.

[iz8mbw]

Just an update, I changed the power supply of the sensor from 3V3 to 5V and now it's still working since 28 hours.

[iz8mbw]

[tonhuisman]

Just an update, I changed the power supply of the sensor from 3V3 to 5V and now it's still working since 28 hours.

That sounds more like a wiring mistake or a faulty voltage regulator on your board (assuming there is one installed) or a broken sensor, as the SCD41 tech specs say the max VDD should be no higher than 3.6V...

[iz8mbw]

According to Sensirion Data Sheet the maximum VDD is 5.5V

[tonhuisman]

Hah! Official Sensirion pdf says different: 🤔 ❓

[TD-er]

Is the digital level of the data signals related to this voltage? If so, then other I2C devices may also be exposed to higher voltages

[iz8mbw]

https://sensirion.com/products/catalog/SCD41/

[tonhuisman]

Ah, too many Sensirion pdfs in the same folder 😊

[iz8mbw]

Hah! Official Sensirion pdf says different: 🤔 ❓

Sensor in scope is SCD41, not SGP41

[tonhuisman]

Yes, mixup, sorry.

Still, smells like a broken sensor, as it should still work OK from 2.4V, so it's strange it won't cooperate at 3.3V Maybe adding a 10uF capacitor near the sensor might stabilize it better?

Edit: The max. VDD ripple of 30mV seems to indicate it might be a tad sensitive to instable power.

[TD-er]

Hmm the peak current is lower on 5V, so this might still be power supply (or wiring) related. And as you can see, the input high level is related to Vdd. So I guess it is perhaps better to look into the power supply to the sensor as running at 5V may not be very stable regarding I2C communications and it can be dangerous to other I2C devices on the same bus which might not be 5V-tolerant.

Keep in mind that using cheap breadboard wiring will cause a significant voltage drop when running "high" currents, even as low as 200 mA.

So you might want to add some 10 uF and 100 nF capacitor close to the Vdd/GND of the sensor and look into the connection of both GND and Vdd of the sensor.

[iz8mbw]

Guys, for now going to close this issue. Thanks for suggestions.

Ciao!