Spontaneous Hard Drive Combustion

[shawty]

Been a while since I posted anything, but I thought I'd drop this amusing(ish) story into the running...

We here about hard drives failing, and usually it's the dreaded unreadable platter, or sector errors or a drive just reaching the end of it's life and smart giving you a 4 minutes to midnight warning of impending doom.

I like any I.T. professional have had my fair share of those over the years, so much so they really don't phase me any more....

I have to say though, I have NEVER had one spontaneously combust on me before.

To be fair, the drive in question however was a very old drive, it was a 250Gb Western Digital V1 IDE that I originally had back in the late 1999's in one of my 386 PC's, it's survived this long without so much as a burb on a bad sector or a single read error, I was essentially just using it as a temp file dump in my second PC it had been working fine all day and for the week before that.

Without warning, there was a strong smell of dying electronic components, and on peering through the vents on the PC in question noticed a little yellow flame...

Needless to say, the power cable was yanked out the back with so much force I almost pulled the PSU out the case with it...

Anyway.... here's the post mortem pictures :-)







At first I suspected the PSU, but after completely disassembling it, examining it and testing the voltages on it, that seems to the best condition part in the PC.

I thought perhaps the spindle motor jammed and the controller ramped up the amps trying to get it spinning again, but opening it and spinning the motor by hand proved otherwise.

So I'm guessing that after 22 years of operation it just decided it had, had enough.

I've packed the actual drive away minus it's controller board in storage now, who knows I might come across a compatible board one day and recover the data,

Shawty

[james_s]

I guess the seal finally broke and all the magic smoke leaked out 

[shawty]

Yup... pretty much.  :-D

4 days later and my office still stinks of burned electronics...     

[rsjsouza]

Yup... pretty much.  :-D

4 days later and my office still stinks of burned electronics...   

It's the Ampère smell... We've all been through that.   

[Gyro]

At first glance, that just looked like a burnt spindle motor driver IC, but the burn mark isn't centred over the die, it's more like a smoke trail. It was probably the foam that actually flamed. If you get rid of all that crud and burnt foam, you might just find some blown passive component either on the motor tracks, close in to the IC or maybe even an smd supply decoupling cap.

It's a long shot, but it might just be salvageable. It might be a fun exercise if you get bored - you'd probably get a nice sense of achievement if you were able to resurrecting it from a fiery death!

[Nominal Animal]

While I've had several drives die the typical ways, I've only managed to let the magic smoke out from a single IDE drive.  I had a MOLEX power Y-splitter I used that had the +5V and +12V lines swapped on one of the forks.  Oopsie.

In 1998, I bought a Maxtor IDE HDD that was so imbalanced, running it on a flat horizontal surface it would travel about half an inch a second.  Made an awful noise, too.  The seller was so impressed they brought me a replacement the same day; I think they sent it back to the manufacturer for a looksee.

The only one combustion/explosion I can remember now, is when I had a voltage regulator on a Gigabyte GA-MA78G-S2H literally explode.  Cratered the IC and all, with an audible POP!.

[RoGeorge]

I know why it happened, look:



 

[MrMobodies]

Just noticed from the color of some of the contacts (squared in read) what looks to me like it got very hot there.



Is that burn marks on some of the components further down the board (squared in orange) ?

[AaronLee]

To be fair, the drive in question however was a very old drive, it was a 250Gb Western Digital V1 IDE that I originally had back in the late 1999's in one of my 386 PC's, it's survived this long without so much as a burb on a bad sector or a single read error, I was essentially just using it as a temp file dump in my second PC it had been working fine all day and for the week before that.

250Gb? Generally Gb would mean gigabits, but drives are normally spec'd in GB (gigabytes), and a 250GB drive in the late 1990's, based on my fuzzy recollection, must have been a real top-of-the-line drive.

I remember in the early 90's when typical drives were just beginning to push the gigabyte barrier, and the drive manufacturers would market them as 1600MB drives, etc. I guess they thought that sounded cooler to the customers than 1.6GB, but to me it was sort of annoying.

Edit: I just went back and looked at the photo again, and noticed the copyright silk screening showing 2004. So likely it was a 250GB drive, but not from the late 1990's.

[amyk]

Not too uncommon failure mode...

http://forum.hddguru.com/viewtopic.php?f=1&t=7370
http://forum.hddguru.com/viewtopic.php?f=1&t=17833
https://forum.hddguru.com/viewtopic.php?f=1&t=32272
https://forum.hddguru.com/viewtopic.php?f=1&t=35098

http://www.kerrywong.com/2012/12/12/bring-a-dead-wd1001fals-back-to-life/

Maybe fzabkar or one of the other HDD recovery people on this forum may add some more info; but I remember the SMOOTH motor drivers do have a habit of burning out and haven't found anyone with a definitive analysis of the cause; perhaps some of the members here who also like to decap chips could help take a look at what died on the die itself - power MOSFETs?

Edit: the contact oxidation is a common issue too, and probably unrelated to the motor controller burning out.

[james_s]

250Gb? Generally Gb would mean gigabits, but drives are normally spec'd in GB (gigabytes), and a 250GB drive in the late 1990's, based on my fuzzy recollection, must have been a real top-of-the-line drive.

I remember in the early 90's when typical drives were just beginning to push the gigabyte barrier, and the drive manufacturers would market them as 1600MB drives, etc. I guess they thought that sounded cooler to the customers than 1.6GB, but to me it was sort of annoying.

Edit: I just went back and looked at the photo again, and noticed the copyright silk screening showing 2004. So likely it was a 250GB drive, but not from the late 1990's.

That's being a bit pedantic, he obviously meant gigabytes. The era is obviously off though, and additionally I'm not aware of any 386 PC that could support a 250GB drive. Even in the 486 era it was common to need overlay software to support anything over 512MB. In the late 90s I had a Pentium-166 that was pretty hot stuff at the time. There were still a few 486 PCs in being used but a 386 was ancient by then.

[Halcyon]

I've seen similar failures in numerous drives, it's not exactly unknown. That level of damage is more than you usually see however. Most times it just stops working with maybe a small crack in the chip or a fuse blows.

The discolouration of pads and vias is quite common.

[james_s]

It looks like the spindle motor driver that blew up. Those are 3 phase BLDC motors so the driver will have several half bridges. Any sort of glitch or failure that causes a high and low side transistor to turn on at the same time will result in a shoot-through and if the designer of the drive didn't think about that and include a fuse then this sort of failure is not all that surprising.

[shawty]

At first glance, that just looked like a burnt spindle motor driver IC, but the burn mark isn't centred over the die, it's more like a smoke trail. It was probably the foam that actually flamed. If you get rid of all that crud and burnt foam, you might just find some blown passive component either on the motor tracks, close in to the IC or maybe even an smd supply decoupling cap.

It's a long shot, but it might just be salvageable. It might be a fun exercise if you get bored - you'd probably get a nice sense of achievement if you were able to resurrecting it from a fiery death!

Nah, too late now.... it's been donated to the dustbin's favourite charity :-)   I was actually tempted to remove the IC's that where still in one piece to "play with" but then I thought, nah got a big enough box of junk to start with.

[shawty]

While I've had several drives die the typical ways, I've only managed to let the magic smoke out from a single IDE drive.  I had a MOLEX power Y-splitter I used that had the +5V and +12V lines swapped on one of the forks.  Oopsie.

In 1998, I bought a Maxtor IDE HDD that was so imbalanced, running it on a flat horizontal surface it would travel about half an inch a second.  Made an awful noise, too.  The seller was so impressed they brought me a replacement the same day; I think they sent it back to the manufacturer for a looksee.

The only one combustion/explosion I can remember now, is when I had a voltage regulator on a Gigabyte GA-MA78G-S2H literally explode.  Cratered the IC and all, with an audible POP!.

Oh yea, I remember those things, back in my younger days as a trainee Home PC tech working in a local workshop.  (These where the days where you had those little plug in power regulator cards on the main board)

386/486 era, made quite a few of those go bang.

I remember once, launching an IC off of a board with so much force it embedded itself in the plastic bezel above my workspace!

[shawty]

I know why it happened, look:



 

Please tell me that's a real film...  I'm a sucker for cheesy B-Movie style sci-fi like that :-D

[shawty]

Just noticed from the color of some of the contacts (squared in read) what looks to me like it got very hot there.



Is that burn marks on some of the components further down the board (squared in orange) ?

Oh...

Yea, I noticed a lot of the Red ones, but I didn't notice the orange.   I wonder if that Voltage Regulator (Looks like Q8) shorted and that was the root cause.

Interesting.

[shawty]

To be fair, the drive in question however was a very old drive, it was a 250Gb Western Digital V1 IDE that I originally had back in the late 1999's in one of my 386 PC's, it's survived this long without so much as a burb on a bad sector or a single read error, I was essentially just using it as a temp file dump in my second PC it had been working fine all day and for the week before that.

250Gb? Generally Gb would mean gigabits, but drives are normally spec'd in GB (gigabytes), and a 250GB drive in the late 1990's, based on my fuzzy recollection, must have been a real top-of-the-line drive.

I remember in the early 90's when typical drives were just beginning to push the gigabyte barrier, and the drive manufacturers would market them as 1600MB drives, etc. I guess they thought that sounded cooler to the customers than 1.6GB, but to me it was sort of annoying.

Edit: I just went back and looked at the photo again, and noticed the copyright silk screening showing 2004. So likely it was a 250GB drive, but not from the late 1990's.

I've definitely had this thing a long time, if I recall at the time it did cost me an awful lot of money, 1998/2004 it's still rocked on for close to 20 years however... an impressive feat for any platter of spinning rust!

[shawty]

Not too uncommon failure mode...

http://forum.hddguru.com/viewtopic.php?f=1&t=7370
http://forum.hddguru.com/viewtopic.php?f=1&t=17833
https://forum.hddguru.com/viewtopic.php?f=1&t=32272
https://forum.hddguru.com/viewtopic.php?f=1&t=35098

http://www.kerrywong.com/2012/12/12/bring-a-dead-wd1001fals-back-to-life/

Maybe fzabkar or one of the other HDD recovery people on this forum may add some more info; but I remember the SMOOTH motor drivers do have a habit of burning out and haven't found anyone with a definitive analysis of the cause; perhaps some of the members here who also like to decap chips could help take a look at what died on the die itself - power MOSFETs?

Edit: the contact oxidation is a common issue too, and probably unrelated to the motor controller burning out.

"But I remember the SMOOTH Motor Drivers do have a habit....."

I was reading the links you posted in thinking to myself "Cripes, these are all WD and Smooth IC's....   says something for the quality....  :-D :-D

[shawty]

250Gb? Generally Gb would mean gigabits, but drives are normally spec'd in GB (gigabytes), and a 250GB drive in the late 1990's, based on my fuzzy recollection, must have been a real top-of-the-line drive.

I remember in the early 90's when typical drives were just beginning to push the gigabyte barrier, and the drive manufacturers would market them as 1600MB drives, etc. I guess they thought that sounded cooler to the customers than 1.6GB, but to me it was sort of annoying.

Edit: I just went back and looked at the photo again, and noticed the copyright silk screening showing 2004. So likely it was a 250GB drive, but not from the late 1990's.

That's being a bit pedantic, he obviously meant gigabytes. The era is obviously off though, and additionally I'm not aware of any 386 PC that could support a 250GB drive. Even in the 486 era it was common to need overlay software to support anything over 512MB. In the late 90s I had a Pentium-166 that was pretty hot stuff at the time. There were still a few 486 PCs in being used but a 386 was ancient by then.

I worked in a computer repair lab way back then, so I had access to all sorts of software and other interesting stuff that most folks didn't.   

[shawty]

It looks like the spindle motor driver that blew up. Those are 3 phase BLDC motors so the driver will have several half bridges. Any sort of glitch or failure that causes a high and low side transistor to turn on at the same time will result in a shoot-through and if the designer of the drive didn't think about that and include a fuse then this sort of failure is not all that surprising.

So If I do eventually find a donor drive with the same board, there's a chance I might make that board go bang too?

[wraper]

Just noticed from the color of some of the contacts (squared in read) what looks to me like it got very hot there.

It didn't get hot. Larger vias just were not gold plated. The same as pads which are connected to the heads. Not 100% sure but IMHO if you see tarnished pads on HDD PCB, it's silver plating.
It's very common. Smaller vias are not tarnished because they are covered by solder resist.

[james_s]

Oh yea, I remember those things, back in my younger days as a trainee Home PC tech working in a local workshop.  (These where the days where you had those little plug in power regulator cards on the main board)

386/486 era, made quite a few of those go bang.

I remember once, launching an IC off of a board with so much force it embedded itself in the plastic bezel above my workspace!

Several years ago I pulled my old IBM XT out of the shed at my mom's place and took it home. When I turned it on the fan in the PSU twitched briefly but then nothing happened except I could hear a faint tick from somewhere. I put my ear down close to it and flipped the power switch a few more times trying to locate the sound and then BANG! A piece of shrapnel whizzed by my nose followed by a puff of smoke. Turns out a tantalum bypass capacitor on the motherboard was shorted and after a handful of attempts of the upgraded power supply to start up it finally cleared the short.

[Halcyon]

...after a handful of attempts of the upgraded power supply to start up it finally cleared the short.

 

Written in such an eloquent way.

[ejeffrey]

One of the few hard drives I have had fail in the middle of the bathtub curve was an exploding chip that erupted out of the package.  It wasn't as dramatic as this probably because it shorted out the PSU enough to cause it to shut down immediately so it didn't lead to extensive charring or damage to the surroundings.

I know the drive mechanism was fine because I swapped the controller PCB from an almost identical hard drive and it worked fine.