Argon One V2 fan runs at 100% all the time

ARGON ONE (V2 and M.2) DISCUSSION
1

I have only had this little case for a couple of weeks, but now the fan runs continuously at 100% as far as my ear can tell.

Furthermore, there doesn’t appear to be a control point anymore via i2c: It used to be located at 0x01a.

tc@PCP1:~$ i2cdetect -y 1
0 1 2 3 4 5 6 7 8 9 a b c d e f
00: – – – – – – – –
10: – – – – – – – – – – – – – – – –
20: – – – – – – – – – – – – – – – –
30: – – – – – – – – – – – – – – – –
40: – – – – – – – – – – – – – – – –
50: – – – – – – – – – – – – – – – –
60: – – – – – – – – – – – – – – – –
70: – – – – – – – –

Sounds like the fan controller has failed… Suggestions are welcome.

2

Do you have already made the case powerless (unplug the power supply)? This is important to reset the MCU to a defined state if that MCU stucks after a power off scenario or program mode. The I2C usually works again afterwards.

3

I see, so powering off isn’t enough because it truly isn’t actually powered off. Yes, that worked this time. Out of the three V2 cases I have, only one of them is exhibiting this random behavior. Any way to determine how it got “stuck” because this feels like a hardware issue versus software…

4

A known way to me to safely trigger this issue, is calling i2cdump. Another possible way is an unordered shutdown, for example if the shutdown was initiated via remote control / power button at the case. The MCU disconnects from I2C after the process was initiated, but waits internally for level change of specific GPIO pins to get the information that the RPi was shut down finally. This can sporadic fail, especially if the current running process needs usually more than 10 seconds to graceful shutdown the process after the term signal.

In the past one user could isolate a iobroker service as the trouble maker, that communication to the MCU was disturbed after every boot and the fan control was failing afterwards.

Perhaps the firmware version in your Argon ONE cases are different. The currently used Linux kernel / libraries can also make a difference, in addition to the possibility of defective hardware.