2021-09-02 16:20:19 temperature.c:695:get_temp_info: read temp sensor failed: chain = 0, sensor = 0, chip = 55, reg = 0
2021-09-02 16:20:09 temperature.c:695:get_temp_info: read temp sensor failed: chain = 1, sensor = 3, chip = 195, reg = 1
2021-09-02 16:20:10 temperature.c:695:get_temp_info: read temp sensor failed: chain = 2, sensor = 0, chip = 55, reg = 0
As the kernel log, it's not being able to read the temps out of these temp sensors, but the problem isn't exactly that, it's very unlikely for a temp sensor to break, in most cases it's actually a bad chip that tricks the firmware into thinking that issue with one of the sensors.
However, the above is valid when the error only shows for 1 hash board, but then you said the 3 hash boards were working fine and then stopped together, and since it's highly unlikely that ALL there boards will fail at the same time, this confirms Zeusbtc's tech's theory which suggests that in most cases when you get a temp error on all three boards it means your PSU is bad.