Incorrect Molex Install has Killed Four Drives!

[ekim]

Hoping you guys might be able to help with a predicament I have found myself in. I have a Lian Li D8000 case which I have installed Lian Li's backplane boards to, which allows me to easily install/remove drives.

After completing the upgrade and booting in to UnRaid and noticed that two of the drives couldn't be found. I had assumed something hadn't been seated right so shutdown checked connections were firm and rebooted. The drives were still unable to be found so I shut down swapped some drives around and rebooted. At this point four drives couldn't be found!

What later transpired was that I had managed to install a molex upside down to one of the backplane boards! You can see some images of it here:

https://goo.gl/photos/PxSzn9Ljy8jPZRGk7

Whilst it shouldn't be possible to install a molex connector upside down the connector on the board of the BP2SATA appears to allow me to install the molex upside down without any more resistance that it would take to install it correctly!

Any how! On to the problem at hand, and what I would like some advice on. I now have four 3tb WD Red drives which no longer spin. I have tested (with a working drives) swapping the PCBs and I can get the dead drives to spin. So I suspect mechanically everything is fine and I have just fried the boards. From my research I understand that the PCBs have a ROM chip that will need to be swapped to a donor PCB to get the drives working again.

Someone on Reddit also mentioned that it might just be the TVS diodes that have been fried and that the drives could be rescued by shorting, or removing it?

I have some images of both sides of all of the PCBs here:

On one of the PCBs there is a chip that is melted so I am assuming that would definitely need a donor PCB to get up and running again. I have a multimeter so I can test the PCBs, but I don't know how to use it! So can give measurements if needed, just need guidance!

I'd be keen to get some opinions on the following:

1) If I were to get donor PCBs and swap the ROM chips - assuming the drives work what level of trust could I put on that drive moving forward? Would it be a permanent fix?

2) Is replacing the TVS diodes an option?

3) What would you do?

The data on the drives isn't critical - I can live without it, but I'd be willing to pay a little money (£50-75 per drive) to get them up and running again. After all if it is an easy thing to do to bring them back to life it will be cheaper than buying new drives!

Sorry for the long post - thanks for bearing with me.

[ekim]

I can't edit my post so here is the gallery of PCBs. It's board 1 (I commented on each of the photos numbering the boards) that has the fried chip.

As an extra bonus here is a video showing how easy it is to install the Molex the wrong way round:

https://youtu.be/s_yuqhJCgmE

[data-medics]

First off, I'd post the pictures of the PCB's so we can see if you even have TVS diodes. Many newer WD's don't so that might not even be a factor.

A HDD will work with the diode removed if you have a clean power supply, however, it'll be more susceptible to internal damage from power surges. As to replacement PCB's, they are usually quite dependable if they are programmed rather than having physical soldering work done to them. If you physically solder the U12 chips over to the replacement boards, it's possible a component could get overheated and end up failing prematurely.

It is also possible that the surge could have gone right through the boards already and killed the heads. Just having the drive spin up only proves the motor isn't cooked. I'd be much more concerned about the read/write heads which you really won't know if they are good until you replace the PCB.

[fzabkar]

The following articles should enable you to identify and test the TVS diodes and zero-ohm resistors.

TVS FAQ :
http://www.hddoracle.com/viewtopic.php?f=100&t=86

Catastrophic failures in Western Digital PCBs:
http://www.hddoracle.com/viewtopic.php?f=59&t=1119

[ekim]

Nightmare I didn't include the link in my post and I can't edit!

This should work gallery of PCBs:

https://goo.gl/photos/DVUREuG5eEtdLMUy9

[ekim]

Ok, tested the PCBs (hopefully by the time this post goes up my gallery link will have also gone up!)

I think I have identified the TVS diodes as D3 and D4 on each board

PCB 4 - 2060-771824-003 REV A
https://goo.gl/photos/BGiek7q3S73hK9L17
https://goo.gl/photos/mhaAchPCgtU7bCnN8
D3: ~05.3 in both directions
D4: 1 in both directions

PCB 3 - 2060-771945-003 REV A
https://goo.gl/photos/ANXXqLnmyQPmHxVg8
https://goo.gl/photos/MiDf9zm7n8XMDmb58
D3: ~0.02 in both directions
D4: 1 in both directions

PCB 2 - 2060-771945-003 REV A
https://goo.gl/photos/LJbUeAKdqHGAT6sz8
https://goo.gl/photos/oSw8oKQFp9dx33dW7
D3: ~0.05 in both directions
D4: ~84 one direction and 1 in the other

PCB 1 - 2060-77195-001 REV P1
https://goo.gl/photos/jmyPVu1wXEengMiL7
https://goo.gl/photos/umb9QQAnFGoxDoKT6
D3: ~0.02 in both directions
D4: ~129 in one direction and ~58 in the other

[colanco]

Check R67, if the value is equal to "1" (open circuit), most likely the preamplifier / heads are damaged.

[ekim]

In case anyone has issues with Google Photos I have attached the images to this post.

I have taken the readings from the resistors, but I think the post is awaiting approval. If it hasn't shown up in a couple of hours I'll post it again as I appear to be able to post freely now!

Attachments

Board 4

Board 4

Board 3

Board 3

Board 2

Board 2

Board 1

Board 1

[ekim]

Retook the readings. I have included the original TVS readings too. Bit of a mixed bag I think, but would be interested to hear thoughts.

PCB 4 - 2060-771824-003 REV A
https://goo.gl/photos/BGiek7q3S73hK9L17
https://goo.gl/photos/mhaAchPCgtU7bCnN8
D3: ~05.3 in both directions
D4: 1 in both directions
R67: 1 in one direction and ~190 in the other
R64: ~0.03 in both directions

PCB 3 - 2060-771945-003 REV A
https://goo.gl/photos/ANXXqLnmyQPmHxVg8
https://goo.gl/photos/MiDf9zm7n8XMDmb58
D3: ~0.02 in both directions
D4: 1 in both directions
R43: 1 in both directions
R60: ~0.01 in both directions

PCB 2 - 2060-771945-003 REV A
https://goo.gl/photos/LJbUeAKdqHGAT6sz8
https://goo.gl/photos/oSw8oKQFp9dx33dW7
D3: ~0.05 in both directions
D4: ~84 one direction and 1 in the other
R43: ~27 in both directions
R60: ~0.01 in both directions

PCB 1 - 2060-77195-001 REV P1
https://goo.gl/photos/jmyPVu1wXEengMiL7
https://goo.gl/photos/umb9QQAnFGoxDoKT6
D3: ~0.02 in both directions
D4: ~129 in one direction and ~58 in the other
R67: ~0.03 in both directions
R64: ~0.02 in both directions

[fzabkar]

Boards #2 and #3 are probably salvageable by removing diode D3 and flowing a blob of solder over R43. We can measure the onboard supply voltages before reinstalling the boards on the drives, if you are concerned.

Board #1 has a dead motor controller. The nature of the damage suggests a high likelihood of preamp failure. If you are willing to gamble US$50, then you could purchase a replacement board, including firmware transfer (U12), and hope for the best.

Board #4 also appears to have a damaged motor controler, although the damage isn't visible. The preamp would also be suspect. Once again you could take a gamble on a PCB swap plus firmware transfer.

[fzabkar]

This is my best guess as to the onboard supply voltages.

Attachments

Vcore.jpg (154.96 KiB) Viewed 20137 times

Preamp_Volts.jpg (106.92 KiB) Viewed 20137 times

[ekim]

Thanks. If I can bring two drives back to life I can use the redundancy I have on Unraid to recover the data that was on the other two drives.

If I were to remove the diodes and add the solder how reliable will the drives be? It'll likely take around 8hrs to get the other drives data recovered. I have a reliable PSU (Seasonic X660) so I'm confident it should be able to supply clean power.

In terms of removing the diodes and adding the solder is this fairly easily done? I have a soldering iron and solder here but I am not the best soldering in the world. I do have a local electronics repair shop that might be able to help.

Thanks again for the help!

[fzabkar]

Practice on scrap boards until you are comfortable with soldering. Otherwise your local shop shouldn't charge you too much for such a simple task.

As for reliability, I wouldn't trust a drive that has been stressed as yours have, at least not for critical data storage. That said, I don't recall any user that was unable to recover their data after repairing a drive with a shorted TVS diode. Assuming that your drives have no history of bad sectors, then I can't imagine that a professional hardware imager would perform measurably better than a DIY recovery.

Best of luck.

[ekim]

Thanks, in terms of the measuring the onboard supply voltages what will I be looking for?

I've reached out to my electronics repair shop - so hopefully they'll be able to help.

Thanks again

Mike

[fzabkar]

Vcore will probably be in the range of 0.9 - 1.0V.

Vio1 and Vio2 will probably be around 1.8V and 2.5V, possibly 3.3V. These voltages are required by the SDRAM, serial flash memory (U12), and the MCU. One of these voltages may even be a second Vcore supply (in the case of dual-core MCUs).

Vw - Vz are a mystery to me, but I suspect they may be involved with the piezo transducer on the headstack. This is for fine servo tracking.

The -5V preamp supply voltage may appear briefly after power-on and then disappear. This is because the Marvell MCU is testing for the presence of the preamp at this time.

This article may help your technician understand the power supplies:
http://www.hddoracle.com/viewtopic.php?f=59&t=231

[ekim]

Had a look at the article and it is all well above my head but will pass it on. Would the board need to be powered to test all of those voltages? How would you go about powering it, connect it up to a PSU and fire it up?

[fzabkar]

Would the board need to be powered to test all of those voltages? How would you go about powering it, connect it up to a PSU and fire it up?

Yes and yes. Use a screw hole as the ground reference for your multimeter.

[colanco]

Follow the advice of fzabkar. Is the best way.

Anyway, I would like to place my bet ....

#2 #3 and #4 --> preamp damaged

#1 -->I would bet by damaged preamp, although I think it is the one that has more chances to be saved, (R67 is ok)

[colanco]

(R67 is ok)

Or may not, and the measure corresponds to an indirect value ...

[fzabkar]

In the case of PCB #1, the OP's measurements appear to be inconsistent with the visual evidence.