PCB repair questions for WD5000AAJS

[nhgoey]

Hello. I am attempting to repair (or replace if necessary) the circuit board for a Western Digital 500 GB 3.5" SATA HDD, WD5000AAJS.

Background: I am not too concerned with the data. I am more interested in the challenge of the task. Therefore I am not looking to send it off somewhere. Failure will not be a disaster so much as a disappointment.

Part numbers and such:
* HDD: WD5000AAJS
* PCB: 2060-701477-002, Rev. A
* MCU: Marvell 88i6745-TFJ1
* Note: Position U12 for flash ROM is not populated. I assume the flash is integrated into the MCU.

Symptoms: The drive spins up fine. It will click several times. The drive will flash up for just an instant in "My Computer" in Windows and then disappear, at which point it spins down, pauses, and then the cycle repeats. I cannot get the drive to be recognized long enough to read the SMART data. If I boot directly into WD's DLDIAG diagnostic program it does not even see the drive.

HOWEVER, there is a very interesting phenomenon which convinces me that the board is bad:

I noticed that the board gets rather hot. If I put an ice cube on the back of the board near the position of the MCU the drive will actually work perfectly for a few minutes. I can read the SMART data and all values are good. I can even read some files and folders. It only works for a few minutes though. This behavior is consistent and repeatable.

So it seems that freezing the back side of the board will make it work temporarily.

Now some questions:

1. What do you think is the most likely problem?
2. Does the MCU (Marvell 88i6745) need to be totally replaced?
3. Do you believe bad solder joints or corroded traces could be responsible? I see some badly corroded pads near the Hynix SDRAM chip--they are covered in black spot corrosion that seems like it could interrupt the continuity between the IC pins and the nearby components and vias to which they connect.

Anything else you can think of?

If at all possible of course I would like to avoid having to replace the whole board or MCU due to the difficulty of transferring the flash contents.

[pruggero]

Hello nhgoey, and welcome.
I'm not a pro and I cannot help you doing any diagnose, but IF the problem it's the board, and you can make it work for few minutes, you would be able to read the flash from inside the MCU using WDR Demo (http://malthus.zapto.org/viewforum.php?f=32). Once got the flash content you could buy a new board and write the patient flash data on it!

Have you ever tried to plug the board when detached from the disk?

Best Regards
Paolo

[fzabkar]

If you don't want to replace the board, try gently cleaning any oxidisation from the preamp connections on the PCB with a soft pencil eraser.

Then check the voltage test points for Vcore and Vio to confirm that neither is too high. You can do this with the board removed from the drive.

See http://malthus.zapto.org/viewtopic.php?f=59&t=231#p632

Your Marvell 88i6745-TFJ1 MCU is known for "head mimic" faults, but I don't know if the fault that you are observing is related in any way.

[dobrevjetser]

I've seen this with drives that had a failing spindle. Could also be the pcb.
First try to rule out the pcb like spildit and pruggero suggest.

Dobre

[ShaneWard]

Might be your VCM controller is nackered, of you can find a way to replace that, or keep it cool, you might be able to get data off, otherwise you might have to replace the board and ROM chip.

I had some old drives do something simuler. I just stuck a heat sink on the VCM controller and that normally does the trick. Sometimes heat will damage it.