Hello all,

I have (or had..!) a server running unRAID with 8 conventional HDDs of various size and 1 SSD. Recently it "randomly" powered off and wouldn't power back on - investigation showed a blackened spot on the motherboard and a popped component (SuperMicro X8DTE-F) and the PSU would no longer power up using the paper clip test. So I presume either the PSU took the motherboard out or vice versa.

I got the PSU (EVGA 650 GQ) replaced under warranty and swapped for an EVGA SuperNOVA 750 G+. I stupidly used the old modular PSU SATA cables with the new PSU thinking they would be compatible. It appears they are NOT! From the pinouts I have found, I think there's a chance the 5V and 3.3V pins are swapped - on top of my stupidity and laziness, this seems like terrible engineering from EVGA's point of view - as the plastic is exactly the same shape and you can happily swap the cables around.

I can't find a conclusive answer as to the pinouts anywhere but this picture shows the physical difference.

https://i.imgur.com/yxXwWDS.jpg

The new PSU refused to power up when these SATA cables were plugged in, with trial and error I removed them all and noticed the difference. Upon replacing them with the new modular cables the PSU powered up but lo and behold 2 of my HDDs were now missing. Trying to power them up in a different PC results in no spinning up, seems like they are completely dead.

unRAID allows recovery of data using parity if only one drive has failed, so if I can get just one of these drives going again I should be able to "recover" the other onto a brand new disk.

The drives that have failed are:
Western Digital 8TB WD80EZAZ-11TDBA0 (shucked from a 'MyBook')
Seagate Barracuda ST4000DM004

I'm hopeful I can perhaps do a PCB swap, probably on the Seagate drive as that appears to have an 8-pin BIOS chip, the WD one doesn't seem to have one.

I just wanted to check with the experts here as to whether there's another option here - there don't appear to be any blackened or popped components on either PCB so wonder if there's a fuse/diode or something that could be removed/replaced to get them going. Once running I'd probably run the discs for about 24 hours to recover the data then throw them away "just in case".

Thanks in advance!
Jim

Can't seem to edit my post for some reaosn, but wanted to add the PCB board numbers:

WD: 006-0A90561 (or the same with 001- prefix, on the other side of the board - not sure which is the correct number)
Seagate: 100815595 REV D

you could search for posts on TVS diodes on the forum, there are plenty of topics about it.

On the other hand, if your data is critical, i recommend contacting a pro, coz without experience you can cause more harm than good...

pepe

_________________
Adatmentés - Data recovery

These drives have good protection, so the solution would most likely involve removing a shorted 5V TVS diode and replacing or bridging a fuse, if there is no other damage.

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

If you can upload detailed photos of your PCBs, I can show you what to do. No need for professional intervention.

_________________
A backup a day keeps DR away.

Thanks so much for the replies guys.

Images of the boards here:

Seagate: https://i.imgur.com/PLG9der.jpg

WD: https://i.imgur.com/s7OHL6o.jpg

Check these components.

_________________
A backup a day keeps DR away.

Thank you!

The resistance readings from my multimeter - with the black wire to the cathode end (otherwise "1")

WD. 5v/12v/Schottky 180ohm

Seagate 12v 660ohm
5v 575ohm

I'm afraid I don't know how to test the fuses / 0ohm resistors - can't pick up anything with the multimeter. It doesnt beep when set to continuity over the fuses, not sure if this means they're both blown

A good zero ohm resistor or fuse should measure close to 0.0 ohms on the 200 ohm range. A "1" would indicate that the component was open circuit.

_________________
A backup a day keeps DR away.

OK. Well it would seem in that case that every single 0ohm resistor on the Seagate board (4 of them), and both fuses on the WD board have gone...

As a control I've put the multimeter across two other 0ohm resistors on the WD board and they both read close to 0.

So.. bits of wire soldered in to replace the 0ohm resistors / fuses?

Thanks again for your help

You can flow a blob of solder over each fuse and resistor, but that would be at your risk. That said, the overvoltage didn't damage any diode, so it would have been at the low end of the Richter scale.

These drives have two levels of protection. The first is the TVS diode and fuse, the second is a load switch. All the load switches appear to have survived.

The safest approach would be to use a donor PCB and transfer the "ROM" chip from patient to donor. However, if you damage the ROM, the cost of data recovery will become extremely high.

I would use picofuses to test each PCB, but I realise that this is not practical for most users. That said, other people have bridged the fuses and resistors with 100% positive results.

_________________
A backup a day keeps DR away.

I think I'll give it a go with the WD board at least to start off with, only two fuses to do, it doesn't matter if it gets trashed as I only need one drive to recover all data.

Job for tomorrow though as it is now past my bedtime thanks very much indeed for your insight and help