Fried hard disk chip - recovery strategies
Geico Caveman
This is a Maxtor Diamondmax 16, 300GB ATA/133 HDD (3.5"). I bought it 3
years ago in a pre-integrated USB enclosure to serve as my backup disk. I
had an encrypted (luks) filesystem on the disk. A few weeks ago, the disk
stopped working. I tried using a different external HD enclosure. One of
the chips on the disk melted and smoked. I tried it again later with a
different enclosure, same story. Probably the original enclosure had some
kind of a safety interlock that tripped when the power characteristics of
the disk changed.
dmesg indicates :
[ 218.627976] usb 4-6: new high speed USB device using ehci_hcd and
address 6
[ 218.761247] usb 4-6: configuration #1 chosen from 1 choice
[ 218.761657] scsi7 : SCSI emulation for USB Mass Storage devices
[ 218.761973] usb-storage: device found at 6
[ 218.761979] usb-storage: waiting for device to settle before scanning
[ 223.750878] usb-storage: device scan complete
[ 229.352472] usb 4-6: USB disconnect, address 6
[ 229.352904] scsi 7:0:0:0: Device offlined - not ready after error
recovery
[ 269.395761] usb 4-6: new high speed USB device using ehci_hcd and address
7
[ 269.528865] usb 4-6: configuration #1 chosen from 1 choice
[ 269.529257] scsi8 : SCSI emulation for USB Mass Storage devices
[ 269.529666] usb-storage: device found at 7
[ 269.529671] usb-storage: waiting for device to settle before scanning
[ 274.518486] usb-storage: device scan complete
[ 280.120104] usb 4-6: USB disconnect, address 7
[ 280.120292] scsi 8:0:0:0: Device offlined - not ready after error
recovery
Over the years, many different computers have backed up to this disk. Some
of those do not exist anymore (the usual story - old machine, cleaned up,
sold, bought a new machine). There is a lot of data on this disk. The
critical data has secondary backups, so I am not worried about that. But
there are a bunch of photographs from various hiking trips etc. that I
would like to recover from this hard disk if possible.
A colleague told me that if I can get the same chip as I currently have on
this hard disk, I can possibly recover my data. The assumption obviously is
that it is the chip that is dead, and not the actual magnetic disk.
Is this true, and if it is, how does one even do such a thing ? This
product has been discontinued, so my only hope appears to be to watch for
ebay auctions.
Is there an alternative way in which I can access this hard disk and mirror
off the data ? FWIW, I bought a new hard disk to replace this which is
ready to go.
Thanks.
Whoever
It may be easier to look for an identical disk and swap the circuit boards
out, rather than attempting to swap individual chips. As you suggest
below, eBay may be the best source. However, a 300GB drive can't be that
old, there may be some old stock out there.
Try googling the exact part number.
The problem with this approach is that you really do need to find an exact
match.
You could also get some quotes from disk recovery experts. If the data is
worth that much to you, it may be worth your while to pay the cost.
Finally, I suggest that you check the definition of the word "backup". If
you don't have another copy of the data, it's not really a backup.
Kryton
"Geico Caveman" <spammer...@spam.invalid> wrote in message
news:gclkpu$9mn$1...@aioe.org...
Geico Caveman,
I have read before that it may be possible to change the PCB of the
dead HDD with a PCB from an identical working HDD.
If you do a bit of googleing on 'replacing hdd circuit boards' you
will find that people have managed it in the past with verying degrees
of sucess.
Try this google results page: http://tinyurl.com/3ek4df
HTH
--
Kryton
Colin Wilson
> years ago in a pre-integrated USB enclosure to serve as my backup disk. I
> had an encrypted (luks) filesystem on the disk. A few weeks ago, the disk
> stopped working. I tried using a different external HD enclosure. One of
> the chips on the disk melted and smoked
Disk recovery companies now charge far less than they ever used to,
and hopefully they would have a drop-in-place replacement board they
can use to recover your data. You're probably not looking at more than
a couple of hundred bucks nowadays.
DanS
> Hello,
>
> This is a Maxtor Diamondmax 16, 300GB ATA/133 HDD (3.5"). I
> bought it 3
> years ago in a pre-integrated USB enclosure to serve as my backup
> disk. I had an encrypted (luks) filesystem on the disk. A few weeks
> ago, the disk stopped working. I tried using a different external HD
<SNIP>
.
>
> A colleague told me that if I can get the same chip as I
> currently have on
> this hard disk, I can possibly recover my data. The assumption
> obviously is that it is the chip that is dead, and not the actual
> magnetic disk.
>
> Is this true, and if it is, how does one even do such a thing
> ? This
> product has been discontinued, so my only hope appears to be to watch
> for ebay auctions.
>
> Is there an alternative way in which I can access this hard
> disk and mirror
> off the data ? FWIW, I bought a new hard disk to replace this which is
> ready to go.
Some info here. IIRC, the older/smaller the HD, the better chances of a
board swap working.
st...@worldbadminton.com
> Hello,
>
> This is a Maxtor Diamondmax 16, 300GB ATA/133 HDD (3.5"). I bought it 3
> years ago in a pre-integrated USB enclosure to serve as my backup disk. I
> had an encrypted (luks) filesystem on the disk. A few weeks ago, the disk
> stopped working. I tried using a different external HD enclosure. One of
> the chips on the disk melted and smoked. I tried it again later with a
> different enclosure, same story. Probably the original enclosure had some
> kind of a safety interlock that tripped when the power characteristics of
> the disk changed.
>
Really depends upon how valuable this data is to you.
You might be successful finding another driver board and
moving the actual mechanism to that board. If the drive itself
is OK that should be successful. As you alluded to, EBAY
is likely the way to go.
If you want a few less "mights" and "maybe" but are willing
to spend some $$, send it to one of the data recovery
services. They are very good at what they do.
Stan
Thad.Floryan
> [...]
> If you want a few less "mights" and "maybe" but are willing
> to spend some $$, send it to one of the data recovery
> services. They are very good at what they do.
A friendly tip: be SURE you get a firm price quote in writing
(e.g., FAX or at least email) before sending your disk to any
recovery company. What I have seen happen even with the top-
rated recovery companies is they'll phone-quote a price, accept
the disk, then hold it and your data hostage for a price often
5x to 10x the phone quote; it's takes a threat of a lawsuit to
get them to back down. I've seen this happen at 7 clients with
3 different recovery companies. Be forewarned.
Unruh
>Hello,
> dmesg indicates :
SEnd it to a data recovery company and let them try. You do not have the
means to recover that disk. They will probably have to open it in a clean
room, remover the platters and put them into another drive, and recover the
data. Of course it will cost you.
>Thanks.
TBerk
Your Hard Drive is suffering (likely) more than a single chip failure.
Just focusing on your one burnt chip may not be enough.
Traditionally the DIY version was to obtain a same model HD and
carefully swap the whole electronics package over to the bad unit, just
long enough to transfer the data off.
If you try that you can do so as your own risk taking behavior dictates.
How to do it? - Google is your friend. Be Careful. (Better yet -->)
Another option is to send it out to a data recovery service; they
usually will charge a two tier schedule: an initial, (usually
non-refundable) minimum to take it on, and another fee to actually get
the data onto another HD or DVD, etc.
Good Luck with that and I'll end off with the following recommendation:
Single Point of Failure can exist in your backup plan as well. CD-Rs and
DVD-Rs are really cheap these days. Your time is your own but back up
your back-up; store a quarterly version offsite somewhere.
TBerk
TBerk
TBerk
(your follow up only pointed to a single NG.) Hmmm.
The Natural Philosopher
Years ago we swapped the whole drive board in a hard disk and it did
subsequently work.
Try and get an identical disk from somewhere and swap..
If the drive motor has siezed up though (had that too) sometimes a drop
of oil on the bearings (not a good normal practice) will allow the thing
to spin up enough to get the data off.
Otherwise you need to invoke the expensive process of sending teh disk
away to somewone who can remount the pllatters and get the data off that
way. Dont expect that to be cheap, but companies do esxist that can.
> Is this true, and if it is, how does one even do such a thing ? This
> product has been discontinued, so my only hope appears to be to watch for
> ebay auctions.
>
Yes, or ask here. Or on more localised groups.
Someone must have a identical working unit.
> Is there an alternative way in which I can access this hard disk and mirror
> off the data ? FWIW, I bought a new hard disk to replace this which is
> ready to go.
>
No. Apatrt from contacting a good professional data recovery company and
paying serious money.
> Thanks.
The Natural Philosopher
All irreplaceable data should be in at least two places. What that means
and how you arrange it, is the subject of endless discussion.
For me its simply a second hard drive and rdiff-backup.
YMMV.
Dirk T. Verbeek
>>
>
> All irreplaceable data should be in at least two places. What that means
> and how you arrange it, is the subject of endless discussion.
Indeed.
My GF and me each have, at our own far apart places, our photo's backed
up on each other's computers plus we each have a large backup disk.
Quadruple it might be but in case of say a fire or flood we have a very
good chance of recovery.
Rikishi42
> Traditionally the DIY version was to obtain a same model HD and
> carefully swap the whole electronics package over to the bad unit, just
> long enough to transfer the data off.
Did that, works fine. But you_do_ need an exactly identical drive. Model and
revision.
--
Elevators smell different to midgets
Theo Markettos
> Some info here. IIRC, the older/smaller the HD, the better chances of a
> board swap working.
>
> http://www.deadharddrive.com/
Some drives have a kamikaze function. If they detect something wrong (like
me having taken the lid off and poking around the board or head flexi with a
scope probe) they stop working and refuse to spin up the platters ever
again. So there's a chance that fiddling like this will make it worse. The
kamikaze bit is probably stored on the logic board, though, so a board swap
might still work.
(I'm pretty sure I didn't fry the drive, incidentally, as a high impedance
scope probe wouldn't do much more than load a particular signal. I
definitely didn't short anything)
As you say, older drives are more robust. I had an 80MB drive happily
booting Windows 3.1 with the lid off for several weeks.
Theo
Peter Chant
> Some drives have a kamikaze function. If they detect something wrong
> (like me having taken the lid off and poking around the board or head
> flexi with a scope probe) they stop working and refuse to spin up the
> platters ever
> again. So there's a chance that fiddling like this will make it worse.
> The kamikaze bit is probably stored on the logic board, though, so a board
> swap might still work.
Have you got any links referencing that, I tried looking up "hard drive
kamilkaze" and "hard drive anti-tamper" on google and nothing showed up on
the first pages of either. I'm just curious.
I wonder why anyone would build such a feature in, or how? What purpose
would it serve? Only a tiny proportion of people would try this to recover
data. As a security feature I can't see what good it would do. The only
logical idea is that some how the manufacturers are in league with the data
recovery companies to prevent amateur attempts at recovery, I don't think
that is plausible as it would probally cause more trouble than it saved and
is dubious ethically.
Pete
Theo Markettos
> > Some drives have a kamikaze function. If they detect something wrong
> > (like me having taken the lid off and poking around the board or head
> > flexi with a scope probe) they stop working and refuse to spin up the
> > platters ever again.
>
> kamilkaze" and "hard drive anti-tamper" on google and nothing showed up on
> the first pages of either. I'm just curious.
It happened to me - I was poking around a drive head flexi with a scope
probe and all of a sudden the drive shut down. And no matter what I did it
would never spin up again. Up till then it would quite happily spin up with
the lid off. I'm quite sure I didn't short anything. Even if I did short a
signal, the most likely eventuality would be that it'd work again after
powercycling, but it never did. It never even tried to spinup, fail and
give up.
Similar sorts of things happen if I remove a platter... it might spin up
once to spot there's a missing platter, but doesn't a second time.
Even an early 1980s ST506 drive refused to spinup when there was a platter
missing. I can't remember exactly, but I don't /think/ the step servo
responded to ST506 step requests either. Which I was quite disappointed by,
as I'd hoped to use it as a stepper motor controller.
Of all the drives I tried (full height ST506 to modern-ish 4GB) I couldn't
make them spin up once I'd removed platters.
> I wonder why anyone would build such a feature in, or how? What purpose
> would it serve? Only a tiny proportion of people would try this to recover
> data. As a security feature I can't see what good it would do. The only
> logical idea is that some how the manufacturers are in league with the data
> recovery companies to prevent amateur attempts at recovery, I don't think
> that is plausible as it would probally cause more trouble than it saved and
> is dubious ethically.
I think drives do have some kind of self-preservation mode. Notice how that
they're only accessible from the IDE bus if they spin up. Good drives spin
up, terminally bad drives don't. I've never had a situation where I've
attached a dead drive to an IDE cable and had 100% of sector read errors:
dead drives just don't show up in BIOS. Of course it /could/ be that these
are all logic board faults, but the ones I induced above weren't.
I don't mean drives with some bad sectors, those do show up and give sector
read errors.
I can't see a particular reason other than to emphasise the 'it's dead, buy
a new drive' whereas sector errors might not be spotted. Perhaps there's
some manufacturer-specific command to ignore the errors and try regardless,
but I don't have access to this documentation if it exists.
If there's something mechanically wrong inside there may be a wish to not
cause further damage: for risk of fire or not cause further damage to other
platters. If you're in a datacentre you just want a 'dead' indication
rather than an 'I'm struggling through but please put me out of my misery'.
For some reason retrieving customers' data doesn't seem to be that
important.
Theo
(University of Cambridge Computer Laboratory)
Peter Chant
> It happened to me - I was poking around a drive head flexi with a scope
> probe and all of a sudden the drive shut down. And no matter what I did
> it
> would never spin up again. Up till then it would quite happily spin up
> with
> the lid off. I'm quite sure I didn't short anything. Even if I did short
> a signal, the most likely eventuality would be that it'd work again after
> powercycling, but it never did. It never even tried to spinup, fail and
> give up.
Static discharge frying something?
>
> Similar sorts of things happen if I remove a platter... it might spin up
> once to spot there's a missing platter, but doesn't a second time.
>
That is strange.
> Even an early 1980s ST506 drive refused to spinup when there was a platter
> missing. I can't remember exactly, but I don't /think/ the step servo
> responded to ST506 step requests either. Which I was quite disappointed
> by, as I'd hoped to use it as a stepper motor controller.
>
> Of all the drives I tried (full height ST506 to modern-ish 4GB) I couldn't
> make them spin up once I'd removed platters.
>
I wonder, could the platters you removed act as part of the servo loop for
speed control. However, in that case did the drive ought not to run up to
maximum speed?
>> I wonder why anyone would build such a feature in, or how? What purpose
>> would it serve? Only a tiny proportion of people would try this to
>> recover
>> data. As a security feature I can't see what good it would do. The only
>> logical idea is that some how the manufacturers are in league with the
>> data recovery companies to prevent amateur attempts at recovery, I don't
>> think that is plausible as it would probally cause more trouble than it
>> saved and is dubious ethically.
>
> I think drives do have some kind of self-preservation mode. Notice how
> that
> they're only accessible from the IDE bus if they spin up. Good drives
> spin
> up, terminally bad drives don't. I've never had a situation where I've
> attached a dead drive to an IDE cable and had 100% of sector read errors:
> dead drives just don't show up in BIOS. Of course it /could/ be that
> these are all logic board faults, but the ones I induced above weren't.
>
Hmm, self preservation mode of committing hari kiri? Trouble is, without
disassembling its hard to tell why a drive has died. I've had drives die,
the just started getting errounous quickly then disappeared. I've had
others that just made that dead drive ticking noise, presumably the heads
hitting the end stops.
> I don't mean drives with some bad sectors, those do show up and give
> sector read errors.
>
> I can't see a particular reason other than to emphasise the 'it's dead,
> buy
> a new drive' whereas sector errors might not be spotted. Perhaps there's
> some manufacturer-specific command to ignore the errors and try
> regardless, but I don't have access to this documentation if it exists.
>
I would have thought that that would have become common knowledge by now,
along with last ditch data recovery tools to get the data off a dying
drive.
> If there's something mechanically wrong inside there may be a wish to not
> cause further damage: for risk of fire or not cause further damage to
> other
> platters. If you're in a datacentre you just want a 'dead' indication
> rather than an 'I'm struggling through but please put me out of my
> misery'. For some reason retrieving customers' data doesn't seem to be
> that important.
>
Not totally convinced you could put enough power into a stalled disk drive
motor to cause a fire risk. Have heard impressive grinding noises off of a
dying drive - for which a temporary fix (not mine) was to pick up the whole
PC and drop it about 6 inches back onto the desk. It worked, it would
then boot! BTW - no critical data and it would need a new drive anyway
when this failed as a plan.
> Theo
> (University of Cambridge Computer Laboratory)
Theo Markettos
> Theo Markettos wrote:
>
> > It happened to me - I was poking around a drive head flexi with a scope
> > probe and all of a sudden the drive shut down. And no matter what I did
> > it would never spin up again. Up till then it would quite happily spin
> > up with the lid off. I'm quite sure I didn't short anything. Even if I
> > did short a signal, the most likely eventuality would be that it'd work
> > again after powercycling, but it never did. It never even tried to
> > spinup, fail and give up.
>
> Static discharge frying something?
It's possible, though I was at an antistatic workstation at the time (might
not have been wearing a wriststrap though).
> I wonder, could the platters you removed act as part of the servo loop for
> speed control. However, in that case did the drive ought not to run up to
> maximum speed?
They're not used as servo loop for speed control, but all voice coil (ie not
stepper motor) drives put servo tracks on the platters for head positioning.
Earlier drives devote a platter to it, later drives either put servo tracks
or recover servo information from the data stream. It might be feasible for
the servo data to be read during spinup, but that would require some
variable filtering of the head signal and heads that will float
aerodynamically at low spin speeds.
As you say it would only make sense if the drive tried to spin up the drive
each time before giving up.
> Hmm, self preservation mode of committing hari kiri? Trouble is, without
> disassembling its hard to tell why a drive has died. I've had drives die,
> the just started getting errounous quickly then disappeared. I've had
> others that just made that dead drive ticking noise, presumably the heads
> hitting the end stops.
Can you get SMART data out of these, and does it tell you anything about
what the drive thinks is wrong? I do have something of a problem of
interpreting SMART data as I never know what's normal and what's anomalous.
> I would have thought that that would have become common knowledge by now,
> along with last ditch data recovery tools to get the data off a dying
> drive.
Indeed, unless it's under NDA and the tools that do it aren't widely
released (for someone to reverse engineer). I'd be surprised if drives
didn't have some degree of low level control available (ROM space and
preventing users from breaking their drives are the two reasons why not I can
think of).
I agree, this behaviour is odd.
I should add that I wasn't out to measure this effect - I wanted the heads
for something else and was trying to characterise the head preamplifier chip
(which usually have no data available), but in most cases I caused terminal
damage to the drive before I would have expected. So I wasn't keeping notes
of this behaviour but I did observe it and the above is my memory of it.
Theo
Peter Chant
Sorry, only got two PC's here, and only the current disks are new enough for
SMART. Well I might have had one disk which gave enough warning to boot
_once_ and once only!
>> I would have thought that that would have become common knowledge by now,
>> along with last ditch data recovery tools to get the data off a dying
>> drive.
>
> Indeed, unless it's under NDA and the tools that do it aren't widely
> released (for someone to reverse engineer). I'd be surprised if drives
> didn't have some degree of low level control available (ROM space and
> preventing users from breaking their drives are the two reasons why not I
> can think of).
>
> I agree, this behaviour is odd.
>
What would surprise me is that I've been using PC's since the 8086, there
are a few hard drive manufacturers. That is rather a long time for similar
NDA's from several competing manufacturer's not to get leaked, especially
as a drive "kamikaze" mode would likely be contentious, and thousands of
people by now would be aware.
Also, the board swap scenario you discussed. If the kamikaze mode present
where would it reside? You'd need some way of finger-printing the platters
to the board. Otherwise when a board dies and you do a board swap how does
the old board know there is a problem? The platters will be fine, as far
as it is concerned nothing has changed unless it can remember its previous
state?
> I should add that I wasn't out to measure this effect - I wanted the heads
> for something else and was trying to characterise the head preamplifier
> chip (which usually have no data available), but in most cases I caused
> terminal
> damage to the drive before I would have expected. So I wasn't keeping
> notes of this behaviour but I did observe it and the above is my memory of
> it.
Sounds odd as you say. If your probes were loading the amp (cable
capacitance?) and confusing the drive why did it not come back to life on
removal of the probes? Was the scope grounded? Earth loop between the
drives power supply and the scope? However, again on removal of probes
things should have been back to normal.
Pete
--
http://www.petezilla.co.uk