J B wrote:
> Hello all,
> As some of you may know there was a lightning storm in England
> recently. The power went out for a second and my PC which was
> connected to a Cyberpower BR-650E surge protector and UPS lost power.
> Somehow the surge protector and UPS was not able to protect the PC.
> I rebooted and the BIOS said the motherboard detected a power surge.
> On entering Windows everything seemed to be fine so I went to bed.
> The following morning I could not enter Windows 7 (blue screen) or
> SuSE Linux 11.3 (something about a recursive error). When I tried
> using a Windows 7 installation DVD, it blue screens on "Loading
> Clearly some hardware has been damaged by the power surge. Should I
> buy a replacement PSU, motherboard, CPU, graphics card or RAM?
> I use the computer as my main PC in a SOHO and really need it up and
> running ASAP.
The good news (if there is any), is that a lot of stuff is
working on your computer. If you're seeing a BIOS screen,
and not badly scrambled, lots of stuff had to work
for that to show up.
For the sake of completeness, I'd run memtest86+.
You can get that from www.memtest.org
, half way down the page.
Disconnect all storage devices, except whatever storage device
will be used to boot memtest86+. This is to prevent broken storage
devices from upsetting the BIOS.
If you have one stick of RAM, the bottom 640K can't be tested.
If you have two sticks of RAM, pull them, then reinsert them
in "single channel mode". One stick will be the "high memory"
stick, and fully tested. The second stick will be the "low memory"
stick, and it's bottom 640K won't be tested. Run one pass of memtest86+
error free, before continuing.
Then, swap the sticks in their two slots. This makes the low address
stick the high one. And then it will be completely tested.
If you owned four sticks, repeat the procedure using only the two
If that passes, my real suspicion, is something happened to the disks.
Do you have backups ?
Was the backup drive disconnected during the lightning storm ?
Backups should be disconnected when not in use. This reduces the chance
they'll be burned by lightning.
If you have access to a second computer, connect the drive(s)
from the damaged computer, one at a time. You could also bring
over the optical drive, and test it.
Your first test, would be a read-verify, to prove the sectors
are all accessible. While a natural reaction would be to reach
for CHKDSK to do this, the thing I don't like about CHKDSK, is
it is a "repair-in-place" tools. If you had an IDE cable on the
drive, and the cable was bad, CHKDSK has been known to trash
a disk, because each write attempt fails due to the bad cable.
CHKDSK works best, if the disk is actually healthy, and all
that was needed, was rewriting some structures.
If you don't have backups, you can even consider backing up the
damaged drives. And in the process, the backup tool may complain
about bad sectors. The backup procedure (sector by sector) then
functions as the "read-verify test".
So my first purchase, might be some replacement drives, which will
function quite nicely in the first ten minutes of your experiments,
as your backup destination when you copy the questionable disks.
With backups in hand (sector-by-sector in case the file system(s)
are damaged), now it's "safe" to use CHKDSK. If CHKDSK makes
noodle soup out of the drive, you have your backup to restore with.
And if the backup attempt failed, you also get evidence of
a hard drive failure. (Then the disks purchased, become your
new blank disks for OS reinstallation.)
For home repair work, a second computer is very convenient. It's
pretty hard to (cheaply) repair a single PC, while commuting to
the public library to use their machine to Google stuff :-)
And in this case, a second machine, with disk interfaces
suitable for testing your storage devices, would be
an excellent tool to have access to.
In terms of resilience to power surges:
1) A cheap ATX supply ($20) can have virtually no protection features.
A transient could come right though it. The surge protection on
a UPS, may still let dangerous transients through.
2) The motherboard has switching converters on it. These provide
a measure of buffering on transients. The processor has the
Vcore switcher (and that protects the processor). System RAM has its
own little one or two phase supply (providing a measure of protection).
This is a potential reason the CPU and RAM will survive, even if the
motherboard is blown. (And we don't have any symptoms yet, which
positively verify a bad motherboard.)
3) Hard drives, have something like an MOV across +5V and +12V to ground.
They're on the disk controller board, near the power connector.
This suppresses transients. A continuous power surge (for many seconds)
will cause the protection devices to burn. The MOV like device is not
intended to protect against a continuous surge. Only a pulse of too
much voltage gets stopped. The +12V one, may clamp at around 15V or so.
The disk cabling is susceptible to induced currents. So the interface
on a drive could be ruined.
I haven't read of enough surge cases, to rate all of these possibilities.
The CPU and RAM do seem to hold up very well. The reason I'm having you
test the RAM above, is because RAM naturally goes bad on its own, and
if you haven't tested it within the last year or so, it's time to test it
now and find out if it is still good. I don't think the transient event
got it. But it could still be bad. Then the next step, is testing
disks and optical drive, one at a time, on another computer. Making
If you get any more new interesting symptoms, post back.