SUMMARY: Tips for recovering data from a broken SCSI disk

From: Johan Nyberg (
Date: Sat Feb 27 1999 - 16:15:07 CST

(sorry for a second posting of this message; the Subject was
empty by mistake in the first one)

Hi Sun Managers,

I am very sorry for the late summary. I lost a SCSI disk and wanted,
if possible, to recover some or all of the data on the disk (my
original question is at the end of this letter). I got several replies,
of which the one from Harvey Wamboldt <> was the
most extensive and very useful. I have below included Harvey's list of
things one could try to do to recover the data on a broken disk.
I tried several of his advices, but finally had to give up on the

Thanks to:

Birger Wathne <>
Harvey Wamboldt <>
Rick Lantaigne <>
Wade Stuart <>
Rodney C. Marable <>

Advice by Harvey Wamboldt <> for trying to recover
data from a broken hard disk:

Assuming that you are prepared to write off the disk, there are a
couple of things you could try. It's important to try them in the
correct order, since they increase in risk until you reach the point
where the disk is effectively destroyed. In no particular order they
 o Verify the disk is being cooled ... a failed fan is a common
    problem, and the noise of a bad fan bearing is often mis-diagnosed
    as a bad drive.
 o Change cables and terminators (cables, SCSI IDs and terminations
    are probably the major source of disk problems).
 o (SCSI disks only) verify the SCSI termination voltage (both at the
    computer and the disk). If the voltage is missing, try to
    determine why and fix the problem. If the voltage is too low, try
    enabling the termination voltage at both ends of the cable (this
    is not normally recommended since it can lead to a ground loop
    which increases noise levels). [If you are a hardware hacker and
    the voltage is too low, you can even try replacing the termination
    regulator with a higher drive component].
 o Put the disk on a different power supply (usually by swapping it
    to another enclosure).
 o Try the disk on a different computer.
 o If "mechanical" noises are coming from inside the disk, don't run
    the disk until you've eliminated the source of the noise (or
    verified that the disk can be run without destroying it).
 o Physically remove the disk from all power sources, then observing
    proper anti-static precautions, carefully open up the disk
    enclosure. Determine whether the disk can be run with the case
    open (ie everything is screwed in tight). Some disks can be run
    with the case open. If the disk was making noises, try to
    determine the source.
 o DON'T TOUCH THE PLATTER SURFACES! (Or allow anything to come in
    contact with them).
 o Assuming everything is screwed in tight, turn the disk over and
    shake it to see if anything falls out. If it does, try to
    determine what it is and where it came from. If it's an
    electrical component, chances are you may have to replace the
    electronics (with electronics from an identical drive).
 o Examine the platter surfaces for scratches (You will need a bench
    light or flashlight for this). If a surface is scratched by one
    of the heads (or arms), then that head is probably toast. Data on
    that surface is probably unrecoverable unless you can replace the
    head assembly and the scratching isn't too bad. If the scratching
    is severe enough to affect the rotation of the disk, then that arm
    may have to be bent or cut away to allow the disk to rotate
 o Touching only the edge of the platters rotate the disk to check for
    sticky or rough bearings (use something extremely clean for this,
    hands aren't a particularly good choice, plastic [such as plastic
    wrap] is generally better but is a major source of static so don't
    let it touch the electronics).
 o If the disk appears ok mechanically, then hook it up to power and
    turn it on. If the disks don't start spinning, try giving them a
    push. If that makes them go, then maybe the problem is the motor
    is no longer powerful enough to start spinning the disk.
 o You should be so lucky that the problem is a broken power lead,
    but hey, you should check anyway.
 o If the problem is the disk bearing (it's sticky or rough), then
    you could try an infinitesimally small drop of ultra fine machine
    oil on the bearing. This may smooth it out enough to run the
    drive long enough to get some data off.
 o If the problem is the head stepper motor assembly, then it is
    theoretically possible to replace the head and stepper with an
    identical unit from another drive. At a minimum this will require
    a good assortment of miniature tools. Given the alignment (and
    re-wiring) problems this will introduces, this procedure has a
    pretty low chance of success.
 o If there are no signs of mechanical problems then the problem may
    be with power. Perform a continuity check on the power and
 o If there are no signs of mechanical problems, and you have power
    and ground continuity, then the problem may be with the
    electronics. Do a comprehensive visual examination of the
    electronics. Look for blackened or discoloured components,
    examine the traces on the board, look for signs of heat (melting,
    balled up solder, curled traces, etc). If you have an identical
    drive, then you may be able to swap the electronics (this may
    involve cutting, unsoldering and resoldering wires. Depending on
    the drive this may be fairly easy or virtually impossible). Also,
    note that drives which use EEPROM for various parameters may mean
    difficulty recovering data even if the drive is usable after the
 o Connect the drive to power and check voltages (you will need a
    meter for this). Check the electronics board, the stepper motor,
    and to the disk motor (assuming you can figure out where they are
    and can reach them). Check the fuses (if you can find them). If
    there is no power, you may be able to solder in a jumper wire to
    bring in power (or ground).
 o In all cases, you probably want a low level disk read/write
    diagnostic program to determine the effect of the changes you
I'm certain that I've missed or messed up more than a few of the
suggestions offered here, so take all this advice with a grain a salt
and a pound of caution. And of course, if you aren't a certified
technician you shouldn't ever open up an electrical device anyway ;-)
There is a reason that data recovery companies charge so much,
***** FOR... ;-) *****
Good luck,
Harvey M Wamboldt ^ E-Mail:
MDA Inc 1000 Windmill Rd. Suite 60 ^ Fax: (902)468-2278
Dartmouth NS, B3B 1L7, Canada ^ Phone: (902)481-3531

My original question: Tips for recovering data from a broken SCSI disk

Hi Sun Managers,
I have an array of four 4GB disks, two IBM and two FUJITSU,
setup in a striped/concatenated metadevice using DiskSuite.
Yesterday one of the IBM disks (model IBM-DFHSS4W) started
logging hardware errors and before we could do an emergency
backup the disk stopped working. I wonder if any of you have
some tips of how one could try to get the disk working again so that
we could recover the lost data.

I have dismounted the broken disk from it's box and when I put power
on it I can hear that it starts spinning up for about 3-5 sec, then
there is a click sound followed by a short (1 sec) spinning up
sequence after which it spins down and stops.
The disk goes through this spin-up/spin-down equence twice then it stops
(no spinning sound can be heard).
One of our technicians told me to try to hit the disk (not too hard)
against the table on the side which is opposit to where to power and
the scsi connector is. I tried this but it did not help.
Before I give up and read back an 11 days old backup I thought I
should ask you guys for any possible advice.
I know there are companies in our country who can recover data
from crashed hard disks, but they are too expensive for us.

Johan Nyberg  
The Svedberg Laboratory, Uppsala Univ., BOX 533, S-75121  Uppsala, Sweden
Tel: +46-18-4713047 (office),  +46-18-324314 (home).  Fax: +46-18-4713833

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:16 CDT