Searching TG0H: Heat-sink Sensor ATI Radeon HD 2400 XT.
Hello,
since three days the cpu fan of my iMac runs constantly at 3900rpm short after the start. I've done SMC and PRAM-Reset without a result.
Then I made an ASD to precisely check which device is causing this problem. ASD activity log file shows following failures:
[05:26:56] ERROR - -13009 [SMC error TM_TEMP_BELOW_MIN, the temperature read from the sensor is too frequently below some minimum threshold.] -- (1.00)
TEST FAILED
[29 Apr 2010 05:26:57] SMC: SMC PID Sensor check (test #1) - SMC PID Sensor check
- Checks the SMC PID Sensor warning counter that is monitoring sensors of each PID loop for erratic behaviour.
[05:26:57] ERROR -- 1 [The SMC's PID Sensor check has been tripped] --
TEST FAILED
Hardware Info shows
Sensor
type: Temperature
Location: Temp (TG0H) - GPU 0 Heatsync
Low limit: 10 C
high limit: 105 C
reading: 1.00 C
description: GPU 0 heatsink temperature
The GPU heatsink sensor is reading only 1°C. I think this causes the fan ramping to 3900rpm. But I found out (thanks to Richdave) that this GPU sensor is not available seperatly, so the support got to change the whole VGA – but that's too expensive for me. I've bought an thermal adhesive and thermal paste, now I need the correct temp sensor. All I know is that it should be a TO-92 and it must have low limit 10°C and high limit 105°C.
Does anyone which temp sensor is used in the heat-sink of ATI Radeon? Or can s.o. sell me an ATI Radeon? It could be also broken at least, because there are three identical sensors used (TG0D, TG0H, TG0P).
Questa è una buona domanda?
2 Commenti
I have been doing some research and the device is likely something similar to the Dallas Semiconductor DS1822-PAR device. It is a TO92 package and operates as a 2 wire device powered only by the "HIGH" transition on the data-line. The 3rd pin is not internally connected so I suspect it is joined to the DQ (data) pin rather than clipping it off when the sensor assy is made. This device is stocked by mouser.com and probably many other suppliers too.
You will only be able to confirm the exact device if you extract it from the heat-sink assy. Since you have already found a low thermal resistance adhesive, perhaps it is worth doing?
My only concern about this device is the data sheet refers to each device’s unique 64-bit code suggesting that it is an addressable device. If so then you probably need an OEM device.
da Richdave
Another option may be to swap the cables between the GPU and the CPU sensors. The CPU cable looks like it will reach the GPU connector but the GPU cable will not reach the CPU connector.
I expect these are exactly the same; however without testing I cannot confirm this and they are addressable devices; however further reading suggests the system interrogates and learns the address of each device so perhaps there is not going to be a problem swapping them.
If you do this and you get a valid test on the GPU and the CPU failing that confirms the logic board is OK, leaving the GPU sensor as the fault.
If you get the GPU failing and the CPU failing that suggests the logic board is faulty.
If you get the GPU failing and the CPU passing, it most likely means the device is not interchangeable and you need a OEM sensor to replace the faulty one.
da Richdave