2008 Alu iMac - CPU fan runs high and noisy

So it seems a common thread with iMacs... the fans get all screwy.

I have tried a SMC reset, PRAM reset, removing the PRAM battery and shorting the power switch connector for a while all to no avail.

One observation that has me curious is that if there are no apps running the system goes to sleep and fan shuts off, however if I have an app open like FF and walk away, the screen will eventually turn off but the fan never does. In all other respects it seems to be in sleep mode but without the old slow pulsing light it is hard to be sure??

I was brave and powered up the system with the CPU fan disconnected The CPU was quite stable for a while and then rose reasonably quickly to around 50C while the CPU HS was at around 37C. That at least proves the sensors for the CPU are monitoring temp changes...

The wiring for the 4 wire fans have the following colours and there is a little ^ on the board near the connector. I assume this is pin 1 so that would make the following

Pin 1 Black - Control input - PWM input, wider pulse width = higher RMS voltage = higher fan speed
Pin 2 White - Tacho output - increased frequency = faster fan speed
Pin 3 Black - 0VDC (Ground)
Pin 4 Brown - +12VDC

So measuring the CPU tacho I get 126Hz = 7560RPM and iStat shows 3770RPM

The ODD tacho gives 33Hz = 1980RPM and iStat shows 980RPM

The ODD control input has 1.6V on it. I am trying to make the ODD temp rise to confirm that the control signal increases to drive the fan faster...

I have no explanation for why the actual fan speed Vs the measured speed is a ratio of about 2:1. I think that perhaps my cheap $20 meter is counting transitions (+ve and -Ve changes) rather than cycles.. Pity I don't have a CRO

With the ODD fan connected ad running at 980RMP (iStat) the control voltage is 1.6V
With the ODD fan disconnected, the control voltage is 12V

With the CPU fan connected ad running at 3850RMP (iStat) control the voltage is 12V
With the CPU fan disconnected, the control voltage is 12V

I have isolated the control wire on the fan assy and the fan shuts off as expected, this suggests that the internal FET is working OK.

The down side is that it also suggests I have a logic board fault.

With the fan off, I can monitor the tacho signal and manually spin the fan, iStat shows the fan speed but the control signal from the board remains at 12V, again suggesting that the logic board is faulty

I have confirmed that the control signal is not a hard short to the 12V supply and I have confirmed the CPU fan connector is OK by measuring all signals direct on the board.

I linked the control signal from the ODD fan to the control wire on the CPU fan and it ran at about 1000RPM proving the fan can be controlled.

I removed the logic board to look at it and attempt to trace the fan circuits... I was not able to find any obvious problems but as I don't have a circuit diagram I was unable to fully trace the circuits to check conclusively.

I did however make one discovery... I forgot to put the RAM in and when I booted so I got no video and a repeating beep. It took me a few minutes to realise my mistake but I did notice that in this state the CPU fan ran at normal speed/noise level.

That suggests the fan control circuit can indeed control the fan properly...

So now I am at square 1.... CPU fan runs high for no apparent reason and no apparent fault

Anyone know where I can get a full circuit diagram from??

I have booted the iMac without RAM again, to retest my observations

with LCD temp sensor connected fan runs at normal speed - quiet
without LCD temp sensor connected fan runs at full speed - noisy

So I have worked out that the 4 wire sensors in the iMac work on a bus system and are interrogated to check their readings.

The chip on the Ambient sensor is a LM75A device and I'm pretty sure the same is true of the LCD sensor. Other 4 wire sensors in other model iMacs use the TMP75 which is similar but not sure if its interchangeable

From the Texas datasheet I see that the data lines are pulled high via pull up resistors so that makes sense. Checking the connector on the LCD sensor I get same results...

I ran iStat and pulled the ambient sensor out, and iStat changed the ambient reading to a -. plugging it into the LCD port I get the ambient reading displayed again proving that there is a bus system and addressing of the sensor.

Might just take a closer look at the LCD sensor and see what it's address is and either make a new sensor or change the address on the ambient sensor and see where that gets me...

I have not yet located anyone that sells the LCD sensor

So I have worked out the following...

The LCD sensor uses a TMP275 chip which is functionally compatible with the LM75A used on the Ambient Sensor.

The small board that holds these sensor chip has 4 locations for a SMD 1K resistor to set the addressing so the sensor will work as the required sensor on the bus.

The positions are designated as follows

RA0H - Address line 0 HIGH
RA0L - Address line 0 LOW

RA1H - Address line 1 HIGH
RA1L - Address line 1 LOW

The chip has 3 address lines but on both sensors address line 2 is tied low so it seems the bus is using a 2 bit address with the sensors addresses as follows.

Ambient sensor - 11, Binary 3, resistors set on RA0H and RA1H

LCD sensor - 01, Binary 1, resistors set on RA0L and RA1H

So now either the LCD sensor chip is faulty or the incorrect addressing is set (that is someone has installed a wrong sensor)

I have been reading my newly acquired service manual and CPU fan run on can be caused by a faulty PSU temperature sensor too!

The plot thickens just when I have ordered a new LCD sensor....

ARGH...

I ran AHT and confirmed the LCD sensor was faulty and replaced it, re ran AHT and all passed, ran extended AHT and still passes so I boot and CPU fan still runs on HIGH, the strange thing is that during the extended AHT the fan is reduced to normal speed for a while...

So I try SMC and PRAM resets again to no avail...

Phantom Gremlin said:

It looks like you're talking to yourself. But don't worry, someone's out here, reading. Too bad I don't have an answer to your problem.

Click to expand...

yeah it does feel like that! but thanks for reading anyway...

Still no joy with this.. I wonder if the ASD will do any better??

So I have run both the EFI and OSX versions of the ASD 3S123 and neither report any errors and both are able to drop the CPU fan speed down to the normal idle speed during tests so that confirms (IMO) the h/w is working fine!!

* screams I'm a teapot while running around in circles*

InteractionDsgn said:

Hi Richdave,

how did you exactly fixed your problem?
I have the same issue:
My CPU fan runs shortly after start witj 3900rpm, while there are no disturbing processes or high temperatures. I checked it with istatpro and smcfancontrol.
I run altough the Apple hardware test (aht), it displays the following error:

4SNS/1/40000000:TGOH-131.000

I found out that it is a faulty sensor, but which one exactly?

Click to expand...

I have not yet fixed my issue... I think it is a PSU sensor incorrectly reporting hight temp but still working thru it.

However,

From the manual

4SNS/1/40000000:TG0H

GPU temperature sensor (located on back side of MLB/video card) is disconnected or faulty.

Verify that the two pin sensor cable is securely connected to the MLB. The connector is located on the back side of the MLB and requires removal of the MLB to access it. If it continues to fail the diagnostic replace the video card.

If the sensor is disconnected the CPU fan will run at full speed.

Click to expand...

Check that the cable is not crimped/damaged causing short. Generally speaking, A high reading is typically symptomatic of a shorted sensor but an unplugged or open circuit sensor usually causes a fail-safe full fan speed in most applications.

I'd expect you should be able to get a replacement sensor without having to trash the entire VGA card, the other possibility is that there will be identifying numbers on the sensor housing that will allow you to buy another device from somewhere. I think it will be a fairly standard 3 wire device that is glued into the heatsink.

InteractionDsgn said:

4MOT/1/40000003: HDD-13## (->## this two numbers vary)

Click to expand...

This error suggests your HDD fan motor is failing or faulty. The numbers after the HDD are likely the expected fan speed but I am not certain about that.

From the 2008 iMac 20in Service manual

4MOT/1/40000002:HDD Hard drive blower/fan

Verify that the hard drive blower/fan cable is securely connected to the MLB and that there are no obstructions that would prevent the blower/fan from rotating. Replace the hard drive blower/fan if the error continues.

Click to expand...

Are you running the AHT continuously? If not you may want to run it in a loop for a few hours and see what happens

The fan error may be indication the fan is starting to fail thus sometimes it's speed does not fall within the 10% allowable limit between actual and target speed. The sensor may be faulty or have a loose connection.

You may have 2 separate faults...

Have you tried resetting the SMC yet? It looks after the temperatures and fans so it is common to both errors.

InteractionDsgn said:

Hi.
When I start after a break there's only the 4SNS/1/40000000:TGOH-131.000 failure.

Click to expand...

THis suggests a sensor fault or connection problem with the GPU sensor.

Have you opened up the system to look inside?

InteractionDsgn said:

Hi Richdave.
I've made an Apple System Diagnostics which tells me that the GPU Heatsink Temperature Sensor is faulty. It's reading 1°C

Click to expand...

reading 1C means the sensor is OPEN circuit so it may be disconnected but if you have not opened your system that seems unlikely. So I would suggest the sensor is probably faulty however it could still be the MLB (main board)

"
SMC:SMC PID Sensor check
Error--1 [The SMC's PID Sensor check has been tripped]
TEST FAILED
"

Doesn't know what that means.

Click to expand...

Just means the sensor failed the SMC diagnostic

FWIW I got my manual HERE

InteractionDsgn said:

Thanks for that. I guess the TG0H is faulty however too. I've asked welovemacs.com for parts and they said following part should be replaced: Ambient Temperature Sensor - Apple Computer 922-7792.

Click to expand...

This diagnosis seems at odds with the symptom you have posted! The TG0H error suggest the GPU heat-sink sensor is faulty. An ambient temp sensor error would be a TA0P error. The part number they have given you IS for the AMBIENT sensor.

The GPU sensor IS NOT listed as a separate part. You would have to remove it and try to source an equivalent part from somewhere. The device appears to be a standard 3 wire device in a TO 92 or similar package (note there are only 2 wires shown as 2 of the device leads appear to be soldered together to 1 wire). I have attached a photo from my unit for reference so you know what it is you need to remove and examine. Be warned though this is glued into the GPU heat-sink with a special glue that has low thermal resistance so you'd need to also get a suitable adhesive to glue the new one in.

Assuming you are still going to go this route, You need to extract the device and look at the flat face of the device and record the numbers/letters on it. These will be the device identifier then Google is your friend!

The other option is to just replace the video card in it's entirety but that will be expensive!

Video Card 661-4672, ATI Radeon HD 2600XT or 661-4673, ATI Radeon HD 2400XT depending on what you have.

FWIW ifixit.com have many good and easy to follow guides to pull down your system.

InteractionDsgn said:

Once again thank you very much, although this are no good news for me.



The TO-92 is the sensor, or what? I really don't get that one, sorry – especially the 3 wire device. Which of the 3 wires have to be soldered together?

Thanks!

PS:
Do you have any clue where I couldn get an equivalant part?
I've found a low thermal resistant glue.

Click to expand...

The TO92 is the package (casing type) that the thermal device appears to come in. I don't know what the device is exactly as I have not pulled mine out and have no plan to. You won't know exactly what it is until you remove it and examine it. The flat side of the casing should have the manufactured part number on it unless Apple have used unmarked devices...

The device will have 3 leads but you will have to solder 2 of them together.

IF you decide to pursue this approach, it will become much clearer when you pull the old one out as you will have to move the cable from the old to the new device.

Another option would be to buy a faulty VGA card on eBay and see if the sensor in that is OK by just plugging in the sensor only from the faulty card. This is a little risky but you can pick them up reasonably cheap from time to time.

You must understand I cannot be certain about the exact fault or details of replacement sensor as I cannot physically see/test your system and I have not had to replace a GPU temp sensor myself yet. I am trying to guide you based on my experiences with iMac and other systems. Fault finding/repair has an element of trial and error as sometimes the obvious cause is not the actual reason.

Have you reset the SMC yet? Link HERE