Maker Pro
Maker Pro

Hard drive repair (longish)

A

Asimov

Jan 1, 1970
0
"PlainBill" bravely wrote to "All" (26 Mar 05 09:05:55)
--- on the heady topic of "Re: Hard drive repair (longish)"

Pl> From: PlainBill <[email protected]>
Pl> Xref: aeinews sci.electronics.repair:44174

Pl> The new drive (and the old hda with the new
Pl> electronics board) spins up as soon as power is applied. The old
Pl> drive, (and the new hda with the old electronics board) won't spin up
Pl> under any circumstances.

I'd bet the motor controller/driver IC is the culprit and not the data
detection circuits! Perhaps you could try to swap a new motor IC onto
the old board if you are equiped to desolder surface mount IC's?

A*s*i*m*o*v

.... Integrated Circuit (n): a device used to protect fuses.
 
M

Michael A. Terrell

Jan 1, 1970
0
James said:
I've got a couple dead DiamondMax drives right here, they both go click
click click and won't read.


I didn't write this, JURB did.
 
M

Me

Jan 1, 1970
0
Hi PlainBill,
There have been many excellent posts in this thread,
and after a few days of thought, I would like to toss out a so called
"Hail Mary" pass... You did say the HDD spins up with the new, albiet
incorrect control board. Perhaps, just perhaps, this problem can be
solved with a software solution. What I am talking about is to install
the HDD in a machine with a CDR Burner on the second IDE Channel, and
with the HDD on the first IDE channel. I then purpose booting from a
Symantec Ghost 2003 (or later) boot disk, which should recognize the
HDD (hopefully) and if it does, choose to dump the HDD to the
destination CDR Burner. If you are sucessfull with this, then
presuming you have the full version of Ghost, install Ghost with Ghost
Explorer onto a good machine. Then, you could take the CD Ghost image,
and with Ghost Explorer open up the Ghost Image, and, with the
Explorer View, choose all of the folders you wish to save, and dump
them to a folder of your choice on your Hard Disk.. Anyway, I have
saved a few peoples disk information with this process, however, I can
only in theory guess this might work for you.. In years past, I have
tried many of the things the above posters have tried, with basically
the same results. The only thing I may take exception to is last year,
just for fun, I took apart a 2.1 GIG WD HDD, and left the top off, and
not only put fingerprints all over it, gently tweaking the pickup arm,
then using windex and a soft cloth then to clean the disk surface, the
darn thing kept running for days with the top off, sitting on the
bench, with no data loss or corruption. I certainly don't live in a
clean room, but the darn thing still works fine to this day.. I tried
to break it just for fun, and found it just wasn't as "fragile" as
most people claim HDDs are.. Anyway, My two cents worth, hopefully,
Ghost will work for ya! Best Luck! Anthony, Cleveland, OH.
 
R

Ray

Jan 1, 1970
0
Hi there.

What does the bios report for the name of the drive on boot. The drive may
have gone into "safe mode". You can check this by googling the the drive
info reported by the bios. Check any articles that include the word "Ace".
Apparently, only part of the data needed to access the drive information is
in the firmware - the other part is on the storage platters. If there is a
discrepancy, the drive kicks into safe mode and your data appears gone. The
only way to recover the data is to edit and rematch the data in the two
portions.

Ray
 
P

PlainBill

Jan 1, 1970
0
Well, I really blew it this time. About a month ago I replaced my 120
gig hard drive with a new 200 Gig Maxtor, model # 6B200P. In
violation of my common sense, I did NOT keep the old drive as a
backup. Well, the new drive went belly up last week - it wouldn't
even spin up. Some of the stuff on the drive is easily replaceable,
but many of the pictures of my Granddaughter cannot be replaced.

I requested an advance replacement from Maxtor, and when it arrived I
tried to repair the bad drive by swapping the electronics boards. I
verified these had identical part numbers. This had a limited sucess:
The new hda does not spin up with the old electronics board; the old
hda DOES spin up with the new board. However, the drive does not
properly report it's size. The new drive reports it is 203.9 Gig; the
new electronics board with the old hda reports it's size as 250 Gig,
then generates POST errors.

At this point I can restore the electronics boards to the proper hdas
and return the old drive to satisfy the terms of the advance
replacement, since I have not altered anything. The last option I am
considering is a part which appears to be a SST Serial flash chip. It
should be possible to swap these between boards if I unsolder with
chip-quik, but I'm not looking forward to it.

Does anyone have any experience with these drives, or any advice to
give?

PlainBill
I'd like to thank everyone who made a suggestion, and give a status
update on this. First of all, if you made a suggestion; thanks. For
the record, setting a drive limit jumper does not help.

Through a little chicanery I obtained a second DiamondMax 10 200 Gig
hard drive. Unfortunately, that also was the same firmware as the
replacement. So I started to do some research.

Surprisingly, there are only 5 ICs on this drive. There is a L7250E
motor / voice coil controller, which is quite well documemted, and
does not have any memory. There is a 40003 IC, which I have not been
able to find any solid data, but did find a siggestion that it might
be a pwm encoder / decoder (possibly r/w data?). There is a
K4S641632H 8 Mbyte memory chip - obviously buffer memory. There is
the 25VF010 1 Mbit serial flash memory. And lastly, we have an Agere
040116600; which would have to be the controller.

As the next experiment, I swapped boards between the two working
drives; a quick test seems to indicate there is no HDA specific
information on the board.

I compared all the numbers on each IC with it's equivalent part on the
other boards. In all cases, each number was either identical on all
three ICs (part number, revision level, etc) or different on all three
ICs (date of manufacture, lot number, etc).

The conclusion: The flash memory does NOT contain a map of bad
sectors. The firmware is either in the flash menory, or in the main
controller IC. For my next experiment I will try swapping this
between two boards with DIFFERENT firmware.

PlainBill
 
M

Michael Black

Jan 1, 1970
0
PlainBill said:
The conclusion: The flash memory does NOT contain a map of bad
sectors. The firmware is either in the flash menory, or in the main
controller IC. For my next experiment I will try swapping this
between two boards with DIFFERENT firmware.
This makes no sense. Flash memory is memory, and is bound to be
where the map of sectors is. Because this is drive specific, it is
in changeable memory. There would be no difference visually between
the boards because the part numbers would not change for the
differing bad sectors.

Michael
 
P

PlainBill

Jan 1, 1970
0
This makes no sense. Flash memory is memory, and is bound to be
where the map of sectors is. Because this is drive specific, it is
in changeable memory. There would be no difference visually between
the boards because the part numbers would not change for the
differing bad sectors.

Michael
Think about it: A 200 Gig drive has 280 MILLION sectors of 512 bytes
each. If only .1 % were bad, that is still 280,000 bad sectors (most
drives had far more than .1% bad sectors). Please describe the
algorithm that would allow placing such a map of bad sectors into a
128 Kbyte memory. It is far, far, more likely that the bad sector map
is on the platters themselves.

On the other hand, 128 Kbytes of firmware would seem much more
reasonable especially if the processor itself has a block of bootstrap
code.

PlainBill
 
K

Ken Weitzel

Jan 1, 1970
0
PlainBill said:
I'd like to thank everyone who made a suggestion, and give a status
update on this. First of all, if you made a suggestion; thanks. For
the record, setting a drive limit jumper does not help.

Through a little chicanery I obtained a second DiamondMax 10 200 Gig
hard drive. Unfortunately, that also was the same firmware as the
replacement. So I started to do some research.

Surprisingly, there are only 5 ICs on this drive. There is a L7250E
motor / voice coil controller, which is quite well documemted, and
does not have any memory. There is a 40003 IC, which I have not been
able to find any solid data, but did find a siggestion that it might
be a pwm encoder / decoder (possibly r/w data?). There is a
K4S641632H 8 Mbyte memory chip - obviously buffer memory. There is
the 25VF010 1 Mbit serial flash memory. And lastly, we have an Agere
040116600; which would have to be the controller.

As the next experiment, I swapped boards between the two working
drives; a quick test seems to indicate there is no HDA specific
information on the board.

I compared all the numbers on each IC with it's equivalent part on the
other boards. In all cases, each number was either identical on all
three ICs (part number, revision level, etc) or different on all three
ICs (date of manufacture, lot number, etc).

The conclusion: The flash memory does NOT contain a map of bad
sectors. The firmware is either in the flash menory, or in the main
controller IC. For my next experiment I will try swapping this
between two boards with DIFFERENT firmware.

PlainBill

Hi Plain (may I call you that? :) :)

Sorry to hear you haven't quite got your pics back; glad to
hear that you're still trying and learning...

Another idea to perhaps rescue your pics if I may?

If you can now easily get the drive to spin up, but bios
still won't recognize it, how about opening it via api calls
as a device? I think a few hex editors are capable of doing
this for you, I know that WinHex will. (Use the tools, disk
editor option rather than Open; then choose the phyical device
rather than the logical device option [the logical device
won't even appear])

At this point it should be relatively easy to search for
pic headers (and footers); then simply write them to some
imaginary filename on a good drive.

Course it could be a *lot* of work going through 120 gigs,
but if your grand daugher and her pics are as important to
you and her family as mine are to us, then....

Hope that helps, and as an aside... thank you. Kinda like
taunting you about locking the barn door, but your plight
has caused us to start burning and snail mailing pics on CD's
between family members far more often. (my kids live in
different cities) Darn, CD's are virtually free; so is postage,
so well worth it for the duplicity and off-site advantages.
Thanks again! Now praying for your success! :)

Take care.

Ken
 
P

PlainBill

Jan 1, 1970
0
I'd like to thank everyone who made a suggestion, and give a status
update on this. First of all, if you made a suggestion; thanks. For
the record, setting a drive limit jumper does not help.

Through a little chicanery I obtained a second DiamondMax 10 200 Gig
hard drive. Unfortunately, that also was the same firmware as the
replacement. So I started to do some research.

Surprisingly, there are only 5 ICs on this drive. There is a L7250E
motor / voice coil controller, which is quite well documemted, and
does not have any memory. There is a 40003 IC, which I have not been
able to find any solid data, but did find a siggestion that it might
be a pwm encoder / decoder (possibly r/w data?). There is a
K4S641632H 8 Mbyte memory chip - obviously buffer memory. There is
the 25VF010 1 Mbit serial flash memory. And lastly, we have an Agere
040116600; which would have to be the controller.

As the next experiment, I swapped boards between the two working
drives; a quick test seems to indicate there is no HDA specific
information on the board.

I compared all the numbers on each IC with it's equivalent part on the
other boards. In all cases, each number was either identical on all
three ICs (part number, revision level, etc) or different on all three
ICs (date of manufacture, lot number, etc).

The conclusion: The flash memory does NOT contain a map of bad
sectors. The firmware is either in the flash menory, or in the main
controller IC. For my next experiment I will try swapping this
between two boards with DIFFERENT firmware.

PlainBill

Hi Plain (may I call you that? :) :)

Sorry to hear you haven't quite got your pics back; glad to
hear that you're still trying and learning...

Another idea to perhaps rescue your pics if I may?

If you can now easily get the drive to spin up, but bios
still won't recognize it, how about opening it via api calls
as a device? I think a few hex editors are capable of doing
this for you, I know that WinHex will. (Use the tools, disk
editor option rather than Open; then choose the phyical device
rather than the logical device option [the logical device
won't even appear])

At this point it should be relatively easy to search for
pic headers (and footers); then simply write them to some
imaginary filename on a good drive.

Course it could be a *lot* of work going through 120 gigs,
but if your grand daugher and her pics are as important to
you and her family as mine are to us, then....

Hope that helps, and as an aside... thank you. Kinda like
taunting you about locking the barn door, but your plight
has caused us to start burning and snail mailing pics on CD's
between family members far more often. (my kids live in
different cities) Darn, CD's are virtually free; so is postage,
so well worth it for the duplicity and off-site advantages.
Thanks again! Now praying for your success! :)

Take care.

Ken
Ken,

Thanks for the advice; it might yet come to that. I believe Linux
might be an ideal OS for this - and give me the incentive to learn a
LOT about it.

At this point I have several alternatives. I've been practicing my
rework technique on a spare card, and chipquik makes removing 8 pin
SMT chips a snap; once I demonstrate I can swap a pair of chips twice
without damaging anything, I will be ready to do it for real.

Two other alternatives also present themselves: Several vendors
claim to have the 6B200R0 drives available. Locally, several retailers
carry the drive in the internal kit, I believe I've 'cracked' the
labling and may be able to identify if the drive in the kit is a
6B200R0 (B4GBA firmware), or a 6B200P0 (B2GBA firmware). Lastly, ads
for the Maxtor drives appear to indicate the 6B200R0 drive is
optimized for multimedia; I just might try to get Maxtor to send me
the correct replacement.

PlainBill
 
J

James Sweet

Jan 1, 1970
0
Thanks for the advice; it might yet come to that. I believe Linux
might be an ideal OS for this - and give me the incentive to learn a
LOT about it.

At this point I have several alternatives. I've been practicing my
rework technique on a spare card, and chipquik makes removing 8 pin
SMT chips a snap; once I demonstrate I can swap a pair of chips twice
without damaging anything, I will be ready to do it for real.

Two other alternatives also present themselves: Several vendors
claim to have the 6B200R0 drives available. Locally, several retailers
carry the drive in the internal kit, I believe I've 'cracked' the
labling and may be able to identify if the drive in the kit is a
6B200R0 (B4GBA firmware), or a 6B200P0 (B2GBA firmware). Lastly, ads
for the Maxtor drives appear to indicate the 6B200R0 drive is
optimized for multimedia; I just might try to get Maxtor to send me
the correct replacement.


Have you tried any of the off the shelf data recovery programs yet? There's
a few of them that even let you download a somewhat functional demo to see
if it sees anything off your drive.
 
PlainBill said:
Thanks for the advice; it might yet come to that. I believe Linux
might be an ideal OS for this - and give me the incentive to learn a
LOT about it.

I have done this before and it can work slowly, but well. If the drive
is recognized by your BIOS, this may work. This is the same idea as
using Ghost to take an image of the drive and then look at the image on
another computer.

Under Linux (and most Unixes), almost all the devices in the system are
represented by special files in the /dev directory. Typically, the
first primary partition on your first IDE hard drive is /dev/hda1 , the
second primary partition is /dev/hda2 , etc. The entire disk as a
string of bytes can be accessed through /dev/hda , which works even if
you don't have a partition table on it.

If the data of interest has a published file format (which is true for
JPEG, TIFF, PNG, and GIF - probably the types of the pictures of interest),
you can write a program to open the file that represents the whole hard
drive and scan through it looking for the headers of files you are
interested in. Most of these file types have headers of a defined
length and either store the overall length data in the file or indicate
the end of data with a unique marker, so you can also find the end of
each file. You then copy each file to another drive. As Ken said, you
will lose the original filename doing this, but that is easy to fix. If
the pictures are straight from a digital camera in JPEG/EXIF format, and
the date and time on the camera were set correctly, the date, time, and
other info like flash and shutter speed are stored inside the JPEG file
by the camera. Thse can be viewed with standard tools and may help you
to decide when the pictures were taken, so you can decide what to rename
the files to.

I would suggest Knoppix ( http://www.knoppix.org ) as a good way to try
this with Linux. Knoppix is a complete Linux system that runs from a
bootable CD; it doesn't require any installation on your hard disk at
all. You can download it for free and burn it to CD. Here is how I
would proceed:

- Decide what PC you are going to use. Make the Knoppix CD, then boot
it on this PC, making sure it works reasonably well. It deals well
with most newer PCs (less than 3 or 4 years old); it might not have
the right drivers for all the hardware in older PCs. If the sound
card doesn't work, who cares, but if it can't see the CD or the hard
disk, you might have trouble.

- Power down and attach your broken disk to this PC. Make sure the
master/slave jumpers are set correctly.

- I would recommend having a place to put a copy of the broken drive.
This isn't strictly necessary - you can work directly from the broken
drive - but it's better to work from a copy. The obvious answer is
the original hard drive for that PC, but I have had mixed luck with
the support in Knoppix for writing to NTFS drives. FAT32 drives work
better. If you have a working disk at least as big as your broken
disk, the simplest thing is probably to format the working disk as
FAT32 and attach it to the PC as well.

- Boot up Knoppix. Figure out where your hard drives are. There will
probably be an icon on the desktop for your non-broken drives, but
your broken drive may not show up.

- Assuming your broken drive came up as hdb, do something like this to
figure out if you're going to have any luck at all:

grep 'JFIF' /dev/hdb # for regular JPEG files, or
grep 'Exif' /dev/hdb # for digital-camera JPEG files, or
grep 'GIF' /dev/hdb # for GIF files

It may take a little while, depending on where the first picture is
on the disk. If it comes back and says 'Binary file /dev/hdb matches',
then there is hope.

If you are going to make a copy of the broken drive:

- Assuming that your recovery drive came up as hdc1 , do something like
this to take a complete copy of your broken drive. This is similar to
what programs like Norton Ghost do. This step may take a few hours:

mount /dev/hdc1 /mnt/hdc1 # make recovery drive available to system
# copy entire contents of hdb to "whole-drive.bin" in the root of hdc1,
# ignoring any read errors:
dd if=/dev/hdb of=/mnt/hdc1/whole-drive.bin conv=noerror

- Power down and disconnect your broken drive. You'll use the copy you
made on the recovery drive from now on. Go on to the next step,
writing a program.

If you are NOT going to make a copy of the broken drive:

- You need to write a program to scan through the broken drive (or the
copy of it) and extract the pictures of interest. This can be in
anything you know and have a compiler for; if you want to do it on
the Knoppix system, I know it comes with C and Perl at least. If you
have a compiler for Windows, and you made a copy of the broken drive,
you can attach the recovery drive to a Windows system and recover the
files there.

- When I have done this, I have it name the output files with a serial
number: 00000000.jpg, 00000001.jpg, etc. These numbers will be
assigned in the order that the pictures are found on the disk. You'll
also need to turn on whatever support your compiler has for files bigger
than 2 or 4 GB, since the copy of the broken drive will be bigger than
that. The program will look something like this:

open whole-drive.bin or /dev/hdb

while not end-of-file:
advance one byte
found image file header?
yes: read file length if it's in the header
while not file length or file footer
copy byte to output file
advance one byte
end while
close output file
no: proceed
end while

close input file

- If you're going to recover files on the Knoppix system, you'll need a
place to put the recovered files. If the recovery drive still has
lots of space free (or at least as much space as you think the
pictures will take), it can be a good spot. Another option would be
to recover some files to the RAM disk that Knoppix sets up, then FTP
them to another computer. I believe Knoppix includes CD-recording
software, so you could also burn the recovered files to CD, if you
have at least one CD-ROM and one CD-R drive. (The Knoppix CD has to
stay in the CD drive while you're running the system.) Or, if you
have a USB drive, you could recover files to that, possibly dumping
them on another computer if the USB drive is small.

- Anyway, write your program, fire it up, and let it extract a few
images. Look at them in an image viewer program (several to choose
from on Knoppix or Windows) and debug as required. When you're
happy, fire it up again and let it sit. It'll take on the order of a
few hours to chug through the whole drive.

- When you're done, power down. If you recovered files to a drive,
attach it to your daily driver and copy the files to its hard disk, or
CD, or whatever. Use your favorite image viewer to look at the files
and rename them to something sensible.
Lastly, ads for the Maxtor drives appear to indicate the 6B200R0 drive
is optimized for multimedia; I just might try to get Maxtor to send me
the correct replacement.

Yeah, they put plastic over the platters so your pictures and movies
don't get thumb prints on them. :) All this probably means is that the
on-drive buffer is bigger, so it's not a very big deal if you can't get
the "correct" drive.

Matt Roberds
 
P

PlainBill

Jan 1, 1970
0
Well, it's time to bring this long, sad story to an end. I've given
up and returned the original drive to Maxtor. For anyone who wonders,
Maxtor will NOT attempt to repair a drive.

I obtained a fourth drive, with firmware identical to the original.
Much to my dismay, ALL controllers reported the original hda as a 250
Gig (not 200). All working controllers reported all hdas correctly,
and would read a partition table and data correctly on them.

The only possible conclusion was that when the original controller
died, it altered something on theplatters, rendering parts of it
unreadable. An attempt to do a byte by byte read of the original
drive failed, several recovery utilities failed to find a usable
partition table.

PlainBill
 
J

James Sweet

Jan 1, 1970
0
PlainBill said:
Well, it's time to bring this long, sad story to an end. I've given
up and returned the original drive to Maxtor. For anyone who wonders,
Maxtor will NOT attempt to repair a drive.

I obtained a fourth drive, with firmware identical to the original.
Much to my dismay, ALL controllers reported the original hda as a 250
Gig (not 200). All working controllers reported all hdas correctly,
and would read a partition table and data correctly on them.

The only possible conclusion was that when the original controller
died, it altered something on theplatters, rendering parts of it
unreadable. An attempt to do a byte by byte read of the original
drive failed, several recovery utilities failed to find a usable
partition table.


Bummer. Well it's possible that the motor bearings were binding and not
allowing the drive to fully spin up, might be why the original motor driver
died.
 
Top