Posts Tagged ‘Hard Disk Structure’

Hard Disk Details(11)

November 28th, 2008 Comments off

Random Notes and Ideas For Data Recovery

    1. Drive goes to sleep, replace the board live
    2. Partitions start on Cylinder Boundaries
    3. Hard Drives have a Safe Mode
    4. You can fix LaCie problems with a Mac mounting them in the system
    5. Drives that you plug in that cause windows to Crash – Use Ubuntu to Read Files
    6. When problems with MFT then retry smaller blocks
    7. If drive parts are good then rewriting the SA area is the part that needs repairing
    8. SA Code can be replaced to do data destruction or encryption
    9. If you are thinking of a hard drive as 0’s 1’s then you are wrong. The equipment interprets signals to make the representation of 0s or 1s. Designers have taken into account the signal distortion and interface problems to make the work
    10. Remove a chip from the PCB and re-solder the chip onto a good board to fix specific problems with chips that are burned, cracked, etc
    11. Soft resets on SATA also need to do a hard reset the controller as it cannot be reset any other way like the bus is reset in a PCI or ATA
    12. ATA-3 Spec – hard drive read without retry was disabled and now is internal on the drive
    13. Seagate Drives use a serial interface of which you can find online. It will show you stats on the drive. If you see FFFF mask FFFF mask it is a head error
    14. If a drive is read with a standard read then it does not need to be read again but it might be good to use ECC to compare in a later pass
    15. Force the drive to use PIO mode instead of DMA/UDMA modes. Some hard drive failures cause the drive to fail reading UDMA but might still work in PIO
    16. Powers on good drive, while board is still in use move it to a new drive. Wrong defect tables and can be cleared
    17. If the platters are misaligned you can write data over the servo wedge and thereby destroying any chance that you can ever read the data
    18. As the thermal heat increases stability of the bits drop rapidly and with the addition of Areal density – degradation is much higher. There are fewer atoms in each bit to retain the bit orientation. Currently the drive will test for decay and if detected will automatically rewrite the data it detects
    19. Hard drives stored in heat for long term storage is extremely bad
    20. Adaptec ATA Raid 1200A Controller in combination with MHDD is great for recovery software.
    21. To determine if there is an HPA – Look at the LBA Maximum and if it is equal to Maximum Native LBA then there is no HPA
    22. Partitions created using standard disk partitioning tools, fdisk, Windows Disk Management, Partition Magic, will all be cylinder aligned. You only have to scan cylinder boundaries for partitions. Dynamic disks do not use partition tables, they use LDM which is at the end of the disk and needs to be done backwards. It uses one single partition occupying the entire disk minus one cylinder. When volumes are added or deleted the partition table is not updated. There are only 4 partitions possible with the standard Windows tools
    23. All partition table signatures end in 55 AA – if this is gone the OS will regard this as not existing. 80 is active 0B fat32 0F extended
    24. Everything in NTFS is a file – $boot
    25. Sector is the smallest addressable unit on the disk. You can read more than one sector but you cannot read less
    26. If doing a head replacement try straws for head stack replacements around the heads to keep them protected. Cut off a small piece of a drinking straw and place it over the head area of each and every head
    27. Even when the lower part of a head stack does not have heads they are still numbered.
    28. Increasing numbers of drive have no chance for parts replacement due to changes in the hardware
    29. Some drives store the lists in the NV Ram on the PCB. The table on one drive will not match the table on another drive and are unique. That might cause the same logical blocks to be mapped to different physical blocks on different hard drives. It is possible to have a swapped board cause a space on the hard drive to be overwritten due to the mapping problem.

Hard Disk Details(10)

November 28th, 2008 Comments off

Matching Serial Numbers on Hard Drives
This link is where I keep track of documentation on how each hard drive needs to be matched for a working donor drive.  I get this any where I can, use it if you can, and if you happen to find something out please let me know so I can add it to the collection!

Drives with the same model number can still have different numbers of heads, therefore the board is different. It is possible to identify the number of heads in a drive: Maxtor,Quantum, Seagate from the serial numbers:

REFIRBUSHIED drives cannot be used as a donor drive. Head 0 is the bottom head and could be bad. And substandard parts are often installed. It is very difficult to match a refirb drive to a good drive with the same problems. This also makes it difficult to make repair a refirb drive.

Quantum – the third number in the serial number shows the headsQuantum = HA code must match

Seagate – the third SYMBOL in serial number represents the heads. Seagate’s sometimes have extra heads and when one is refurbished it is possible to turn off a bad head and turn on an alternate one and then the firmware number revision might change.

Fujtsu needs the first xx-Xxxx to match

IBM and HITACHI DRIVES – Usually the same driveIBM MLC codes have to match

Hitachi ATMR 80gigs fails mostHitachi 3.5 – Firmware code needs to matchHitachi 2.5 – PCB rev has to match

DCM codes for the (5th??? And) 6th numbers must match.No Western Digital drives with the letter R in the code. EB and BB models.Western Digital Drives EB and BB have the head stack affixed from the lid. Western Digital the sixth char in the model is the cache. U = 2meg V=8meg

Samsung the 4th Char in the alpha code on the label on the rear side needs to matchSamsung the 7th char in the model is the size of the buffer H=8megs

The second number of the serial number represents the number of heads Maxtor needs the 2nd and 3rd char to match:

Hi you all, this is the answer I received directly from Maxtor
Dear Mr. Robert,… here is the paragraph that deals with your model type (DiamondMax Plus 9):
For the following Maxtor hard drive models: Fireball 3, DiamondMax 16, DiamondMax Plus 8, DiamondMax Plus 9, Diamond Max 10 and all MaxLine products there is also a GTLA Number on the model (next to barcode on the bottom of the drive). Format 1Y222J2223322. 1, 2 and 3 stand for numbers, Y and J for letters. The numbers 1 and 3 as well as the letter Y need to be identical to be able to replace the PCB on these drives.  This number can be found on the large sticker on the top of the drive.

Unfortunately we cannot give you any more information than this. Any of your DiamondMax Plus 9 drives could possibly have a matching PCB, however it is most likely to be an older one as the drive in question is almost 3 years old.

Kind regards,
Gisela Schubert Technical Support
Maxtor Ireland Ltd.

copied from:

Serial Number on Hard Drive
The boot sector in the FAT32 partition

The boot sector in the FAT partition
The data contained in the boot sector after the OEM name string is referred to as the BIOS parameter block or BPB

Hard Disk Details(9)

November 28th, 2008 Comments off

Slide 4500:  Doing a Platter Swap for a Single Platter

List of items needed:
The first step is to get a hard drive as close to identical as the bad drive you have that is a working drive. At the bottom of this paper you will find help about matching hard drives and serial numbers.
You need a clean area to work on with as little dust floating around as possible.
You will need about 1 hour to do this carefully
A screwdriver set with T3-T8.  These are my favorite
Post-it Notes
Other tools depending on the drive
Anti-Static Gloves ($5 at the local store)

Just move the head as careful as you can to get it out of the way

This is a fairly simple task compared to a head swap. The hardest part is again getting the heads aligned and back on the platter correctly.
If you have a ramp on your drive it is fairly simple to get the head moved out of the way enough to get the platter in position.

Remove the platter from the good drive.

I usually will try to put a screwdriver in the shaft just to the edge of the center of the platter and turn the drive just enough to get the platter to slide on to the screw driver. I will do the same for the bad drive to move the platter to the good drive.

The platter will most likely never be used again so just get it out however you can without affecting the rest of the drive.

Again I use the Post-it notes in the shape of a V to get the heads back on the platter as I did in the head replacement.

Be very careful to keep the orientation in the same direction to so that the platter will be in the correct location when you put the platter back on the new drive.

Slide TBD: Doing a Platter Swap for a Multi-Platter

In order to do a Multi-Platter replacement you will need a special tool. If you have more than one platter and you take out the platters and any one of them turns at all, you will never get them aligned again or be able to read the data. This is because the data is written in a cylinder. Since the data is in a cylinder you must have the exact same alignment of the platters in order to move them to a new hard drive.

There is a special tool called a Platter Replacement Stand. You can get one at for around $250 plus postage. It is a really heavy stand and weighs about 10 pounds.  The platter replacement tool is what you really need and it looks a lot like a coffee can with a slit in the side.  Once you have moved your heads out of the way, this can sits down around all the platters and you can push down on a piece of metal mounted in the slit to tighten it around the platters.  It also has a lid inside that sits on the top ring of the platters that will hold the screws and keep them from rolling around all over the platters.

The pressure from the “coffee can” will hold all the platters together; however you still have to be really careful about taking it out and turning it. It should go straight from one hard drive to the other as quickly as possible with as little movement as possible.

This is the best possible way to keep the drive platters lined up.

You will still reassemble the drive just like you do in a head stack replacement or a single platter replacement. The only difference is using this device to move the platters.

The plate inside the tool holds the screws so that they do not scratch the platter.

Realign the heads.

Hard Disk Details(8)

November 28th, 2008 Comments off

Slide 4259: Head Replacement Section
This is the only section from last year I kept and it is because this is directly related to fixing this click of death problem.

NOTE: If there is only one platter it might be easier to move the platter than to move the assembly. You have to make that choice.

List of items needed:
The first step is to get a hard drive as close to identical as the bad drive you have that is a working drive. At the bottom of this paper you will find help about matching hard drives and serial numbers.
You need a clean area to work on with as little dust floating around as possible.
You will need about 3 hours to do this carefully
A screwdriver set with T3-T8.  These are my favorite
Post-it Notes
Other tools depending on the drive

Process for Head Replacement:
1.  You will need to disassemble the heads and other components from the drive to clear the room for the head and components.

2.    Disassemble the new hard drive, and carefully use folded paper to move the heads apart and to keep them apart as much as possible.

NOTE: If you are going to move the heads off of a drive platter you should always spin the motor in the direction away from the heads and the arm while you are moving the actuator arm to get the heads off. Move with care.

If you are storing the heads or going to put them down, you can try cutting sections of a drinking straw around the head itself. If the drive has a ramp it is very useful to help line up the heads to take them off and to put them back on.

*** There is often a screw under the assembly of the actuator arm that needs to be removed to move the heads.

3.    Carefully lift the assembly out of the drive and move it to the bad drive and reassemble. It will take about two hours to assemble correctly if you take your time. Do everything you can to get the heads lined up again.

NOTE: It is helpful to fold a piece of post-it notes in a V shape and to make the V towards the platters and the heads on each side of the V.  You can get the paper to slide onto the platter and turn the platters with a screwdriver while you gently move the heads back into place.

You must get them lined up and review it before you turn the drive back on or the heads may slide into place and hit the edge of the platter ripping them off and scratching the platter.  It is good to practice with another drive you do not care about before doing this.

Hard Disk Details(7)

November 28th, 2008 Comments off

Slide 3791: The cause of the click is from four possible areas, all resulting in the SA not being able to be read.

1.    System Area of the drive cannot be read because the platter is scratched.

2.    The head itself has a problem and cannot read the SA area.

3.    Preamp on Actuator to the Head has gone bad and is not passing the correct signal to the electronics

4.   The firmware on the board is damaged and does not initialize. This is sometimes caused by static electricity walking across the carpet to install the drives, or there is a short on the board, and additionally I see where someone has allowed the board on the bottom of the drive to touch metal cause it to burn.

All will result in the same problem and will sound like the Click of Death. Recovery Software will not help you correct any of these until after you have repaired the drive and it is running again.

Correcting Problems
Now we move on to some of the things you can do about it on your own.  The click of death is a very difficult problem to solve and in some cases will not be able to be solved especially without some very high end and expensive equipment. But I will tell you what I have been able to fix without that equipment.

Slide 4009: Swapping the PCB (printed circuit board) Live to get around a SA area that cannot be read.

I have done this process several times successfully. It is not perfect but it is a possible chance you will have to recover your data. The first step is to get a hard drive as close to identical as the bad drive you have that is a working drive. At the bottom of this paper you will find help about matching hard drives and serial numbers. If the System Area is badly damaged or corrupt and for some reason the drive will not read the System Area you can attempt to do a live swap. What this means is that you can hook up the good drive, then you use software or windows and tell the drive to go to sleep.  This will cause the drive to spin down but will still be live and powered up and mounted.  Once the drive goes to sleep and the drive stops spinning you can unscrew the board, carefully so as not to let the screws roll around on the board, and disconnect the board and connect it to the bad drive. I suggest that once you do this, you go after the files you need very quickly. It’s possibly you will be able to make an image of the drive.  Keep in mind, that whatever bad blocks that the drive had assigned to the other drive will be bad here as well.  You could try to use some software to clear bad blocks before attempting this, however I don’t suggest it in most cases. That is because it is one more possible item that might cause failure. I would prefer to use the drive that was working and lose a few blocks. After you get what you can then you can attempt to make changes and go back for more data. This is a concept that works about 25% of the time.

Slide 4199: Imaging in Reverse

In dealing with damaged hard drives, I have run into many problems with cache memory on the drive. The problems will often show up as timeouts or ECC failures as well. For example, I try to read from a drive with16 megs of ram for cache and receive errors but the drive is otherwise appears ok. If there is an error 16 megs away from the sector I am reading my drive will die. As of now there is no way to turn off this cache.  However, if you can image your drive backwards there is no cache. Memory on a drive only caches data forward. There are only three ways I know of to image a drive backwards. The first is free, and it is to use dd_rescue. dd_rescue has a special setting for imaging a drive backwards. There is also a special script for dd_rhelp to control dd_rescue for the purpose of data recovery. You can use this on Linux and it works on drives regardless of the operating system on the drive you are recovering from. Typically you will start at the MaxLBA number and work backwards down to 0 LBA. It works quite well and will work on a surprising number of drives that cannot be read any other way. Your other two choices are Media Tools Pro from RecoverSoft ( for Windows, which is about $400, or a piece of hardware which is extremely efficient at doing this type of recovery called Deepspar Disk Imager (, which will cost between $3000 and $4000 depending on configuration. But you should contact each of these vendors for pricing, or use the free option!