Mac Pro 2008 PSU fan stuck at full speed

Question

Level 1

1 points

Mac Pro 2008 PSU fan stuck at full speed

Hi all,

I've got my old office Mac Pro 2008 (3.1?) with 2 quad core Xeons sat around at home, and I've got a problem where the PSU fan is always on full tilt.

Using TG Pro, I can manually control the fans, and can throttle it back to various speeds, but when left to OSX's own whims, it's always flat out. I've verified it's this fan by slowing all the other fans down/speeding them up individually. Plus, if I unplug the PSU fan, the noise/speed stops altogether.

Any ideas what this would be? I don't think it's the fan itself, else TG Pro wouldn't be able to alter the PWM to change the speed, so I'm guessing it's an OSX problem, or more likely (given it happens during the boot process) a BIOS issue?

To that end, when I was refitting the Apple RAID card, I hit the 'reset' button on the motherboard up near the PCI slots. I had this machine running at work on Friday, before refitting the RAID card/hitting the reset button, and it was running at normal speed then... (Although it didn't have OSX installed, and the HDD I was trying to install to was dead.)

Any help appreciated!

Cheers,

Joe

Mac Pro, OS X El Capitan (10.11), null

Posted on Nov 1, 2015 3:52 AM

Reply

Answer 1

p0077050 Author

Level 1

1 points

Nov 1, 2015 3:59 AM in response to p0077050

Just to add, I tried to boot the PC from USB to see if it would happen then, and it started the fan at full speed while still loading the startup manager, so it's not OS related, but a 'problem' at the BIOS/EFI system level.

Reply

Answer 2

Nov 1, 2015 7:29 AM in response to p0077050

Modern Macs use measured temperatures to set the fan speeds in a feedback loop. The SMC handles most of the details.

If a sensor shows a value that is out of bounds, the fan or fans run at high speed. Also, these fans are generally three-wire (two voltage input plus one RPM-output) so a defective RPM indication in the fan can also cause the fan to run at maximum.

In some Macs, the fans inside the power supply itself are controlled from inside the power supply with Analog circuitry, and are not speed-controllable.

Running the diagnostic may provide more information.

Reply

Answer 3

p0077050 Author

Level 1

1 points

Nov 1, 2015 9:42 AM in response to Grant Bennet-Alder

Thanks. None of the sensors looked to be reading erroneously from what I could see. This fan is a four wire affair. The I'll check the RPM being reported by TG Pro next time it's running, but I think it looked about right (i.e., went up and down as I moved the fan speed slider).

It's currently installing 10.7, as I put a new SSD into it, and installed El Capitan, and wouldn't boot into Apple Hardware Tools (incredibly unhelpfully) as the tools aren't in the CoreServices folder, and I can't copy them on (even as root, and after chmod -R 777 the whole CoreServices folder). I believe that reinstalling 10.7 (onto a another spare SSD), it should allow me to boot into the AHT and see what's going on. (Not the most user friendly of experiences, I must say!).

I'll update once I've finally run AHT

Reply

Answer 4

p0077050 Author

Level 1

1 points

Nov 1, 2015 11:38 AM in response to p0077050

Extended test run. Error detected: 4SNS/1/40000000:Vp0C-128.000

Nothing showing up on an initial Google... Hmm

Edit - it's a V, not a Y. The photo I took of the screen was a little blurry!

Based on that, code and this site, http://www.cnet.com/uk/news/how-to-invoke-and-interpret-the-apple-hardware-tests /, it looks like the PSU is partialy unhappy.

https://discussions.apple.com/thread/5645880?tstart=0Suggests that it's power down the 12v. It seems odd to me that a power output discrepancy would cause a fan to max out...

Reply

Answer 5

Nov 1, 2015 12:25 PM in response to p0077050

Since that reading appears to be a power of two (128-something), it is also possible that sensor is disconnected and not providing a meaningful reading.

Reply

Answer 6

p0077050 Author

Level 1

1 points

Nov 1, 2015 12:28 PM in response to Grant Bennet-Alder

I'd say that you're correct! (I'd have expected a 255 value for something out of range, but maybe that's too much time spent with Windows/Solaris/Linux machines).

Using this http://tim.id.au/laptops/apple/macpro/macpro_early2008.pdf (page 150) I tracked down the PSU pins, and had a look at the one in K6, which was the one providing values to the mother board. I removed this yesterday (don't ask, long story), and clearly hadn't put it in again correctly. After removing the RAM raiser support and reconnecting the connectors in K6, it's now booted with no noisey fans. Success!

Reply