search

Sensirion / arduino-i2c-scd4x

Arduino library for Sensirion SCD4x sensors
BSD 3-Clause "New" or "Revised" License
48 stars 19 forks source link

Receiving I2C errors occasionally from readMeasurement() #6

Closed marcelstoer closed 2 years ago

marcelstoer commented 2 years ago

https://github.com/Sensirion/arduino-i2c-scd4x/blob/f055511173928e380a2d2b722f93ac7ce8db544a/examples/exampleUsage/exampleUsage.ino#L110

Occasionally readMeasurement() prints "Received NACK on transmit of address" or "Wrong CRC found" on one of my test devices. It's a SEK-SCD41 kit connected to an ESP32 using the excellent wires that come with the kit.

These errors are accompanied by ESD-IDF I2C messages when it's a NACK issue, and only then:

[I][esp32-hal-i2c.c:1138] i2cProcQueue(): Bus busy, reinit
[W][esp32-hal-i2c.c:1419] i2cCheckLineState(): invalid state sda(21)=0, scl(23)=1

Do you happen to have any indication as to what might be the root cause of this with your hardware? Have you seen this before?

When I2C communication fails, the CO2 reading I get out of readMeasurement() is 16300-something in https://github.com/Sensirion/arduino-i2c-scd4x/blob/master/examples/exampleUsage/exampleUsage.ino#L108. I think it should be internally reset to zero/0 so that your sample code would skip it.

rnestler commented 2 years ago

Hi @marcelstoer

Sorry for the long delay. Do you have access to a logic analyzer or oscilloscope to see whats happening on the bus?

Do you happen to have any indication as to what might be the root cause of this with your hardware? Have you seen this before?

Bus busy may indicate clock stretching? I'm not sure if the SCD4x does clock stretching in certain cases.

When I2C communication fails, the CO2 reading I get out of readMeasurement() is 16300-something in https://github.com/Sensirion/arduino-i2c-scd4x/blob/master/examples/exampleUsage/exampleUsage.ino#L108. I think it should be internally reset to zero/0 so that your sample code would skip it.

This is a good point. But if readMeasurement fails you actually shouldn't access co2, temperature or humidity as shown in the example. But we should initialize the values in the example anyways.

marcelstoer commented 2 years ago

Thanks for your feedback Raphael. I don't have access to the device anymore as I've since given it away, sorry. Unfortunately, I don't have access to a logic analyzer either yet (currently being shipped to me). Hence, the point of raising this issue was simply to see if it's an issue you had come across before.

if readMeasurement fails you actually shouldn't access co2, temperature or humidity as shown in the example

Absolutely. I didn't analyze why the uninitialized co2 value becomes 16300-something in case of error. To be on the safe side I am now skipping CO2 readings outside the [1,10'000] range.