Sign in with
Sign up | Sign in
Your question

Physical hard drive failure.

Last response: in Storage
Share
Anonymous
a b G Storage
July 29, 2005 8:38:14 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

I have a no-name (Platinum) 160GB 7200RPM hard drive that has had some
sort of physical failure. The bios sees it, but any reads from the
drive result in errors or time out. I have tried many hard drive tools
to try to recover the data, but all point to a physical failure. The
drive is not making the normal 'clicking' sound, but I can hear a
repetitive sound as if it's trying, and failing, to access the same
spot. In the past, I've always just RMA'd drives that had physical
failures if they were under warranty, or trashed them otherwise. This
time, I would like to try and see if I can fix the drive myself. The
data on there is not super important, but I would like to recover it if
possible. I'm willing to try anything to fix the drive, including
opening it, since it is no longer under warranty.

So, I guess my questions are:
1) Has anyone here ever fixed a drive with a physical problem?
2) Any guesses as to what component on the drive has most likely
failed?
3) Any websites detailing the steps to fix a physical hard drive
problem? (Google has been no help so far)
4) I know I could probably take this to some place and pay them
thousands of dollars to recover the data on it. So I'm wondering what
approach they take and if any of their techniques can be done at home
without expensive equipment.
5) Any other suggested groups to crosspost to?

Thanks.

--Phillip
Anonymous
a b G Storage
July 30, 2005 4:25:20 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Previously phillip.thurmond@gmail.com wrote:
> I have a no-name (Platinum) 160GB 7200RPM hard drive that has had some
> sort of physical failure. The bios sees it, but any reads from the
> drive result in errors or time out. I have tried many hard drive tools
> to try to recover the data, but all point to a physical failure. The
> drive is not making the normal 'clicking' sound, but I can hear a
> repetitive sound as if it's trying, and failing, to access the same
> spot. In the past, I've always just RMA'd drives that had physical
> failures if they were under warranty, or trashed them otherwise. This
> time, I would like to try and see if I can fix the drive myself. The
> data on there is not super important, but I would like to recover it if
> possible. I'm willing to try anything to fix the drive, including
> opening it, since it is no longer under warranty.

> So, I guess my questions are:
> 1) Has anyone here ever fixed a drive with a physical problem?
> 2) Any guesses as to what component on the drive has most likely
> failed?
> 3) Any websites detailing the steps to fix a physical hard drive
> problem? (Google has been no help so far)
> 4) I know I could probably take this to some place and pay them
> thousands of dollars to recover the data on it. So I'm wondering what
> approach they take and if any of their techniques can be done at home
> without expensive equipment.
> 5) Any other suggested groups to crosspost to?

Well, good luck. This is hi-tech you are trying to repair. In some
(few) cases this might be feasible without much experience and
without special tools, but in most cases you will not even be able
to even diagnose the problem. You are trying something akin to brain
surgery at home. True, the Azteks did brain surgery without
modern medical knowledge or equipment and true, some of their
patients survived, but care to guess how many did not?

One word of advice though: As soon as you open the drive
you have very limited time to repair and get the data off,
unless you have access to a clean room. Depending on what
particles settle on the disk's surface you may also have
no time at all.

Arno
Anonymous
a b G Storage
July 30, 2005 2:31:50 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

phillip.thurmond@gmail.com wrote

> I have a no-name (Platinum)

Not clear what you mean by that. The name should show
up on the black bios screen at boot time, and with Everest.
http://www.lavalys.com/products/overview.php?pid=1&lang...

> 160GB 7200RPM hard drive that has had some sort of physical failure.

Yeah, looks like it.

> The bios sees it, but any reads from the drive result in
> errors or time out. I have tried many hard drive tools to
> try to recover the data, but all point to a physical failure.

> The drive is not making the normal 'clicking' sound, but I can hear
> repetitive sound as if it's trying, and failing, to access the same spot.

Not all drive click when they recalibrate.

> In the past, I've always just RMA'd drives that had physical
> failures if they were under warranty, or trashed them otherwise.
> This time, I would like to try and see if I can fix the drive myself.

Not very likely with a drive that recent.

> The data on there is not super important, but I would like
> to recover it if possible. I'm willing to try anything to fix the
> drive, including opening it, since it is no longer under warranty.

> So, I guess my questions are:
> 1) Has anyone here ever fixed a drive with a physical problem?

Yes, and there is a difference between good enough
to get most of the the data back and fixed properly.

> 2) Any guesses as to what component on the drive has most likely failed?

The only thing its possible to say is that it doesnt appear to be
able to read the data off the platters properly. That can be due
to a number of failures, everything from the read amp inside the
sealed enclosure failing to a poor joint in the connection to that,
to something more basic on the logic card. Not possible to say
which it is without doing stuff like looking at the signals from the
heads with a CRO etc.

> 3) Any websites detailing the steps to fix a physical
> hard drive problem? (Google has been no help so far)

There's a few around, but they're all mostly rather silly.

You could try putting the drive in a plastic bag and putting it
in the freezer for a few hours and then checking to see if you
can see any data on the drive as quickly as possible before it
warms up. And repeat the process until you get the data you need.

That can work if the fault is a dry joint or a crack
in the flexible connection to the heads etc.

> 4) I know I could probably take this to some place and
> pay them thousands of dollars to recover the data on it.
> So I'm wondering what approach they take

Everything from checking those basics with a CRO and
fixing what is found to be wrong to opening the drive in
a clean room and fixing a physical problem in there, to
replacing the head amp etc in a clean room.

> and if any of their techniques can be
> done at home without expensive equipment.

Only really the freezing technique.

With older drives you can often just replace the logic
card with one off another drive of the same model,
but that isnt very likely to work with a drive that
modern, even with two brand new fault free drives.

> 5) Any other suggested groups to crosspost to?

This is the best for those questions.
Related resources
Anonymous
a b G Storage
July 31, 2005 2:39:41 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

phillip.thurmond@gmail.com wrote:

> I have a no-name (Platinum) 160GB 7200RPM hard drive that has had some
> sort of physical failure. The bios sees it, but any reads from the
> drive result in errors or time out. I have tried many hard drive tools
> to try to recover the data, but all point to a physical failure. The
> drive is not making the normal 'clicking' sound, but I can hear a
> repetitive sound as if it's trying, and failing, to access the same
> spot.

That's recalibration--it's either dealing with a marginal sector that is not
quite to the point of being marked bad or something has gone wrong in the
read circuitry so that all attempts at reads give the appearance to the
firmware of being reads of bad sectors.

> In the past, I've always just RMA'd drives that had physical
> failures if they were under warranty, or trashed them otherwise. This
> time, I would like to try and see if I can fix the drive myself. The
> data on there is not super important, but I would like to recover it if
> possible. I'm willing to try anything to fix the drive, including
> opening it, since it is no longer under warranty.
>
> So, I guess my questions are:
> 1) Has anyone here ever fixed a drive with a physical problem?

It's been done.

> 2) Any guesses as to what component on the drive has most likely
> failed?

The platter is most likely. If so you don't have a prayer of fixing it--the
only possible repair is to read everything _else_ off of it, and if the
sector the drive is trying to read is one that the drive needs in order to
start then you're going to have to rewrite the firmware to not need that
sector, which is for an individual whose understanding of disks is such
that he has to ask questions here going to be a huge undertaking.

It _may_ be that you have lucked out and it's an electronics problem, in
which case replacing the circuit board with one from an identical drive
might fix the problem.

> 3) Any websites detailing the steps to fix a physical hard drive
> problem? (Google has been no help so far)

Not really.

> 4) I know I could probably take this to some place and pay them
> thousands of dollars to recover the data on it. So I'm wondering what
> approach they take and if any of their techniques can be done at home
> without expensive equipment.

The only thing you can try that's not going to be expensive would be to swap
out the controller, and don't bet on that working.

Note that some of the recovery services now have reasonably low cost
recovery available if you're willing to sit at the bottom of the priority
queue for however long it takes.

> 5) Any other suggested groups to crosspost to?
>
> Thanks.
>
> --Phillip

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
Anonymous
a b G Storage
August 1, 2005 4:19:10 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"J. Clarke" <jclarke.usenet@snet.net.invalid> wrote in message news:D cip1t019fl@news3.newsguy.com
> phillip.thurmond@gmail.com wrote:
>
> > I have a no-name (Platinum) 160GB 7200RPM hard drive that has had some
> > sort of physical failure. The bios sees it, but any reads from the
> > drive result in errors or time out. I have tried many hard drive tools
> > to try to recover the data, but all point to a physical failure. The
> > drive is not making the normal 'clicking' sound, but I can hear a
> > repetitive sound as if it's trying, and failing, to access the same spot.
>
> That's recalibration--it's either dealing with a marginal sector that is not
> quite to the point of being marked bad or something has gone wrong in the
> read circuitry so that all attempts at reads give the appearance to the
> firmware of being reads of bad sectors.
>
> > In the past, I've always just RMA'd drives that had physical
> > failures if they were under warranty, or trashed them otherwise. This
> > time, I would like to try and see if I can fix the drive myself. The
> > data on there is not super important, but I would like to recover it if
> > possible. I'm willing to try anything to fix the drive, including
> > opening it, since it is no longer under warranty.
> >
> > So, I guess my questions are:
> > 1) Has anyone here ever fixed a drive with a physical problem?
>
> It's been done.
>
> > 2) Any guesses as to what component on the drive has most likely failed?
>
> The platter is most likely. If so you don't have a prayer of fixing it--the
> only possible repair is to read everything _else_ off of it,

> and if the sector the drive is trying to read is one that the drive needs in
> order to start

It's recognized by BIOS so obviously it passed that point.

> then you're going to have to rewrite the firmware to not need that sector,

In this case, just overwriting it.

> which is for an individual whose understanding of disks is such
> that he has to ask questions here going to be a huge undertaking.
>
> It _may_ be that you have lucked out and it's an electronics problem, in
> which case replacing the circuit board with one from an identical drive
> might fix the problem.
>
> > 3) Any websites detailing the steps to fix a physical hard drive
> > problem? (Google has been no help so far)
>
> Not really.

No 'details' but real people to help guide:
http://forums.actionfront.com/

>
> > 4) I know I could probably take this to some place and pay them
> > thousands of dollars to recover the data on it. So I'm wondering what
> > approach they take and if any of their techniques can be done at home
> > without expensive equipment.
>
> The only thing you can try that's not going to be expensive would be to swap
> out the controller, and don't bet on that working.
>
> Note that some of the recovery services now have reasonably low cost
> recovery available if you're willing to sit at the bottom of the priority
> queue for however long it takes.
>
> > 5) Any other suggested groups to crosspost to?
> >
> > Thanks.
> >
> > --Phillip
Anonymous
a b G Storage
August 1, 2005 6:31:10 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Generally data recovery companies replace the drive's board first, then
the preamp board inside the HDA. If that doesn't help, they may check
if a head needs to and can be replaced. The last step is taking the
drive apart, putting each platter on a spin stand (like a turntable)
and reading the analog data with a high performance head and converting
the flux changes into user data with software. This can generally
recover anything that's not physically erased or damaged but takes a
lot of time.

Your drive clearly has problems reading data since you hear it
retrying, but you won't be able to do anything about it (except maybe
trying various temperatures.) If it was easy data recovery wouldn't be
so expensive. Don't waste your time. An open drive makes a nice paper
weight or bookend.
Anonymous
a b G Storage
August 2, 2005 5:06:02 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"RPR" <rohbeck@yahoo.com> wrote in message news:1122931870.433597.66940@o13g2000cwo.googlegroups.com
> Generally data recovery companies replace the drive's board first, then
> the preamp board inside the HDA. If that doesn't help, they may check
> if a head needs to and can be replaced. The last step is taking the
> drive apart, putting each platter on a spin stand (like a turntable)
> and reading the analog data with a high performance head and converting
> the flux changes into user data with software. This can generally
> recover anything that's not physically erased or damaged but takes a
> lot of time.
>
> Your drive clearly has problems reading data since you hear it
> retrying, but you won't be able to do anything about it

What exactly did you not understand in "It's recognized by BIOS".

> (except maybe trying various temperatures.)
> If it was easy data recovery wouldn't be so expensive.
> Don't waste your time. An open drive makes a nice paper
> weight or bookend.
Anonymous
a b G Storage
August 4, 2005 5:24:38 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

RPR wrote:
> Generally data recovery companies replace the drive's board first, then
> the preamp board inside the HDA. If that doesn't help, they may check
> if a head needs to and can be replaced. The last step is taking the
> drive apart, putting each platter on a spin stand (like a turntable)
> and reading the analog data with a high performance head and converting
> the flux changes into user data with software. This can generally
> recover anything that's not physically erased or damaged but takes a
> lot of time.

The technology you described does not exist, no data recovery company
or intelligence/law enforcement agency recovers data by reading analog
data off platters on a spin stand.
Anonymous
a b G Storage
August 4, 2005 8:43:36 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

lazinator wrote:

>
> RPR wrote:
>> Generally data recovery companies replace the drive's board first, then
>> the preamp board inside the HDA. If that doesn't help, they may check
>> if a head needs to and can be replaced. The last step is taking the
>> drive apart, putting each platter on a spin stand (like a turntable)
>> and reading the analog data with a high performance head and converting
>> the flux changes into user data with software. This can generally
>> recover anything that's not physically erased or damaged but takes a
>> lot of time.
>
> The technology you described does not exist, no data recovery company
> or intelligence/law enforcement agency recovers data by reading analog
> data off platters on a spin stand.

And of course you are privy to the internal workings of every law
enforcement and intelligence agency in the world.

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
Anonymous
a b G Storage
August 5, 2005 3:37:29 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Previously lazinator <laslo@actionfront.com> wrote:

> RPR wrote:
>> Generally data recovery companies replace the drive's board first, then
>> the preamp board inside the HDA. If that doesn't help, they may check
>> if a head needs to and can be replaced. The last step is taking the
>> drive apart, putting each platter on a spin stand (like a turntable)
>> and reading the analog data with a high performance head and converting
>> the flux changes into user data with software. This can generally
>> recover anything that's not physically erased or damaged but takes a
>> lot of time.

> The technology you described does not exist, no data recovery company
> or intelligence/law enforcement agency recovers data by reading analog
> data off platters on a spin stand.

HDDs have long since switched to analog reading and decoding, so there
is no advantage left in doing so. With classical digital decoding
doing maximum likelyhood decoding would gain you some signal
improvement (1.5dB better S/N ratio, if I remember my coding theory
course correctly). Not anymore.

Arno
Anonymous
a b G Storage
August 5, 2005 3:37:30 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Arno Wagner wrote:

> Previously lazinator <laslo@actionfront.com> wrote:
>
>> RPR wrote:
>>> Generally data recovery companies replace the drive's board first, then
>>> the preamp board inside the HDA. If that doesn't help, they may check
>>> if a head needs to and can be replaced. The last step is taking the
>>> drive apart, putting each platter on a spin stand (like a turntable)
>>> and reading the analog data with a high performance head and converting
>>> the flux changes into user data with software. This can generally
>>> recover anything that's not physically erased or damaged but takes a
>>> lot of time.
>
>> The technology you described does not exist, no data recovery company
>> or intelligence/law enforcement agency recovers data by reading analog
>> data off platters on a spin stand.
>
> HDDs have long since switched to analog reading and decoding, so there
> is no advantage left in doing so. With classical digital decoding
> doing maximum likelyhood decoding would gain you some signal
> improvement (1.5dB better S/N ratio, if I remember my coding theory
> course correctly). Not anymore.

????? Disks have always used analog signalling for data storage. One
frequency it's a one, the other it's a zero.
>
> Arno

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
Anonymous
a b G Storage
August 5, 2005 3:37:31 AM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"J. Clarke" <jclarke.usenet@snet.net.invalid> wrote in message news:D cudne0fl3@news3.newsguy.com...
> Arno Wagner wrote:
>
> > Previously lazinator <laslo@actionfront.com> wrote:
> >
> >> RPR wrote:
> >>> Generally data recovery companies replace the drive's board first, then
> >>> the preamp board inside the HDA. If that doesn't help, they may check
> >>> if a head needs to and can be replaced. The last step is taking the
> >>> drive apart, putting each platter on a spin stand (like a turntable)
> >>> and reading the analog data with a high performance head and converting
> >>> the flux changes into user data with software. This can generally
> >>> recover anything that's not physically erased or damaged but takes a
> >>> lot of time.
> >
I've heard this before from others.

> >> The technology you described does not exist, no data recovery company
> >> or intelligence/law enforcement agency recovers data by reading analog
> >> data off platters on a spin stand.
> >
How would you know?

> > HDDs have long since switched to analog reading and decoding, so there
> > is no advantage left in doing so. With classical digital decoding
> > doing maximum likelyhood decoding would gain you some signal
> > improvement (1.5dB better S/N ratio, if I remember my coding theory
> > course correctly). Not anymore.
>
I assume you are talking about PRML.

> ????? Disks have always used analog signalling for data storage. One
> frequency it's a one, the other it's a zero.
> >
Long ago FM/MFM worked like that. RLL uses single flux changes.
Anonymous
a b G Storage
August 5, 2005 1:59:37 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Previously J. Clarke <jclarke.usenet@snet.net.invalid> wrote:
> Arno Wagner wrote:

>> Previously lazinator <laslo@actionfront.com> wrote:
>>
>>> RPR wrote:
>>>> Generally data recovery companies replace the drive's board first, then
>>>> the preamp board inside the HDA. If that doesn't help, they may check
>>>> if a head needs to and can be replaced. The last step is taking the
>>>> drive apart, putting each platter on a spin stand (like a turntable)
>>>> and reading the analog data with a high performance head and converting
>>>> the flux changes into user data with software. This can generally
>>>> recover anything that's not physically erased or damaged but takes a
>>>> lot of time.
>>
>>> The technology you described does not exist, no data recovery company
>>> or intelligence/law enforcement agency recovers data by reading analog
>>> data off platters on a spin stand.
>>
>> HDDs have long since switched to analog reading and decoding, so there
>> is no advantage left in doing so. With classical digital decoding
>> doing maximum likelyhood decoding would gain you some signal
>> improvement (1.5dB better S/N ratio, if I remember my coding theory
>> course correctly). Not anymore.

> ????? Disks have always used analog signalling for data storage. One
> frequency it's a one, the other it's a zero.

As you said "one is 1 the other is 0". But that is classical (M)FM.
Modern modulation is far more sophisticated and modern disks read
values 0.0 ... 1.0 for the bits and then decode them into the
real target data or do error correction of these "soft" bits.
Classical disks read the bits directly as 0 or 1 and if that
failed did error correction on the hard 0/1 values.

Arno
Anonymous
a b G Storage
August 5, 2005 1:59:38 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Arno Wagner wrote:

> Previously J. Clarke <jclarke.usenet@snet.net.invalid> wrote:
>> Arno Wagner wrote:
>
>>> Previously lazinator <laslo@actionfront.com> wrote:
>>>
>>>> RPR wrote:
>>>>> Generally data recovery companies replace the drive's board first,
>>>>> then the preamp board inside the HDA. If that doesn't help, they may
>>>>> check if a head needs to and can be replaced. The last step is taking
>>>>> the drive apart, putting each platter on a spin stand (like a
>>>>> turntable) and reading the analog data with a high performance head
>>>>> and converting the flux changes into user data with software. This can
>>>>> generally recover anything that's not physically erased or damaged but
>>>>> takes a lot of time.
>>>
>>>> The technology you described does not exist, no data recovery company
>>>> or intelligence/law enforcement agency recovers data by reading analog
>>>> data off platters on a spin stand.
>>>
>>> HDDs have long since switched to analog reading and decoding, so there
>>> is no advantage left in doing so. With classical digital decoding
>>> doing maximum likelyhood decoding would gain you some signal
>>> improvement (1.5dB better S/N ratio, if I remember my coding theory
>>> course correctly). Not anymore.
>
>> ????? Disks have always used analog signalling for data storage. One
>> frequency it's a one, the other it's a zero.
>
> As you said "one is 1 the other is 0". But that is classical (M)FM.
> Modern modulation is far more sophisticated and modern disks read
> values 0.0 ... 1.0 for the bits and then decode them into the
> real target data or do error correction of these "soft" bits.
> Classical disks read the bits directly as 0 or 1 and if that
> failed did error correction on the hard 0/1 values.

Can you guide me to more information on this?

> Arno

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)
Anonymous
a b G Storage
August 5, 2005 3:49:46 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

"J. Clarke" <jclarke.usenet@snet.net.invalid> wrote in message news:D cvrdq0e4k@news1.newsguy.com...
> Arno Wagner wrote:
>
> >> ????? Disks have always used analog signalling for data storage. One
> >> frequency it's a one, the other it's a zero.
> >
> > As you said "one is 1 the other is 0". But that is classical (M)FM.
> > Modern modulation is far more sophisticated and modern disks read
> > values 0.0 ... 1.0 for the bits and then decode them into the
> > real target data or do error correction of these "soft" bits.
> > Classical disks read the bits directly as 0 or 1 and if that
> > failed did error correction on the hard 0/1 values.
>
> Can you guide me to more information on this?
>
It is all explained at PCguide, including PRML: http://pcguide.com/ref/hdd/geom/data.htm
Anonymous
a b G Storage
August 5, 2005 6:01:52 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

I didnt say that i'm privy to the internal workings of every LEA/intel
agency in the world, but you are so confident in your statement that
you must be privy to it. You must've seen this magic machine with your
own eyes ?

Somebody asked "How would you know". Well, i've been doing this for a
while and i talked to enough people in various agencies to know what
they can and cant do.
Coincidentally John isnt completely wrong, we developed technology he
was describing and right now it's in a late prototyping stage. It was
first publicly demonstrated at the NASA MSST2004 storage show in Apr
of last year. But again, currently it is not being used in production
recoveries, and nobody has/uses anything like this.

Somebody had questions on how drives work internally, i can write some
stuff up on monday if there's still interest.
Anonymous
a b G Storage
August 5, 2005 8:41:05 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

Previously J. Clarke <jclarke.usenet@snet.net.invalid> wrote:
> Arno Wagner wrote:

>> Previously J. Clarke <jclarke.usenet@snet.net.invalid> wrote:
>>> Arno Wagner wrote:
>>
>>>> Previously lazinator <laslo@actionfront.com> wrote:
>>>>
>>>>> RPR wrote:
>>>>>> Generally data recovery companies replace the drive's board first,
>>>>>> then the preamp board inside the HDA. If that doesn't help, they may
>>>>>> check if a head needs to and can be replaced. The last step is taking
>>>>>> the drive apart, putting each platter on a spin stand (like a
>>>>>> turntable) and reading the analog data with a high performance head
>>>>>> and converting the flux changes into user data with software. This can
>>>>>> generally recover anything that's not physically erased or damaged but
>>>>>> takes a lot of time.
>>>>
>>>>> The technology you described does not exist, no data recovery company
>>>>> or intelligence/law enforcement agency recovers data by reading analog
>>>>> data off platters on a spin stand.
>>>>
>>>> HDDs have long since switched to analog reading and decoding, so there
>>>> is no advantage left in doing so. With classical digital decoding
>>>> doing maximum likelyhood decoding would gain you some signal
>>>> improvement (1.5dB better S/N ratio, if I remember my coding theory
>>>> course correctly). Not anymore.
>>
>>> ????? Disks have always used analog signalling for data storage. One
>>> frequency it's a one, the other it's a zero.
>>
>> As you said "one is 1 the other is 0". But that is classical (M)FM.
>> Modern modulation is far more sophisticated and modern disks read
>> values 0.0 ... 1.0 for the bits and then decode them into the
>> real target data or do error correction of these "soft" bits.
>> Classical disks read the bits directly as 0 or 1 and if that
>> failed did error correction on the hard 0/1 values.

> Can you guide me to more information on this?

It has been a long time since I studied the subject. Google
returns relevant hits for google(soft decision decoding) and
google(maximum likelihood decoding). The actual decoding is
significantly more difficult than the classical "hard" decoding
and may even be prohibitively expensive for some codes.

Unfortunately I did not find a concise introduction. The course
I learned this had hand-drawn slides by the lecturer....

Arno
Anonymous
a b G Storage
August 10, 2005 7:42:57 PM

Archived from groups: comp.sys.ibm.pc.hardware.storage (More info?)

I know we did it many years ago at DEC for customers who lost data on
our drives. Combing in a new head stack was risky because the platters
sometimes got scratched, so they took the disk stack apart. Generally
we used a hand made souped up read channel with a spin stand, but I
know of one guy who worked on A/D sampling the read signal from the
preamp and storing one track worth of data in a high speed buffer that
was interfaced to a VS4000. A few years back I chatted with a guy from
Ontrack about this and he agreed that that's the way to go. If it's
been used with high density disks recently I don't know. The main
advantage I can see is that you don't have to rely on servo info which
may be shot, or invalid if the platters have slipped. At each diameter
you can decode a couple of sectors even if you don't compensate for
runout. If you move the head in small enough increments (using e.g. an
interferometer as the reference), you should be able to catch
everything, a little in the way VXA does the reading. And I can easily
imagine storing an entire analog image of the surface and running image
processing algorithms on it. Hmm, TB sized images. Sounds like fun.

Ralf-Peter
!