Seemingly Random Temperature Sensor Failures, schematics + board view
Intermittent Temperature Sensor Faults, possible SMC issue?
I previously posted about Possible faulty cpu temperature sensor spikes but no errors from ASD resulting in kernel task spikes. I thought I had potentially remedied the problem by replacing the HDD data cable as I was getting bad read/write speeds but the issues seems to have comeback.
Running the onboard ASD I get three sensor errors:
- TC0D - SMC IO Failed
- TC0P - SMC IO Failed
- TM0P - SMC IO Failed
On other times the reading are all at the normal level
Which relates to the CPU 0 Die Analog, CPU Proximity and Battery Charger Proximity (which I think the last is the same as the 5/3.3V proximity) respectively.
This is backed up by abnormal readings in TGPro and iStats menu - readings over 100 and non-existent intermittently, and for prolonged periods of time - in a seemingly non-deterministic way - whilst all the other temperature sensors give normal readings.
It is always the ‘cpu die’ and ‘cpu proximity’ sensor that go bad at the exact same time.
Looking at the schematic for my logic board I think that it could potentially relate to U5510 “CPU Proximity/CPU Die/5V-3.3V Proximity “.
The trouble is is that it is hard to diagnose even if I go about reading the voltages etc, as its intermittent - usually during higher cpu load (but not always). I also find it odd that when its ‘working’ the readings are fairly normal, so the sensor seemingly ‘works’ at times. I don’t know if there are any other explanations - such as other parts of the system pulling down the voltage?
I notice that if any sensor data is unavailable the CPU is immediately throttled at 800 MHz. So for example if I unplug the trackpad (removing the palm rest sensor readings). The computer is stuck at 800 MHz. I cannot seem to replicate the fault with the other CPU/proximity when the CPU is throttled like this, so I can’t rule other culprits by unplugging everything off the board.