1) Cquirke describes a problem directly traceable to the
defective power supply - not due to unexpected power loss or
'destructive power restoration. All power supplies must work
just fine and must never be damaged by a brownout.
All utility companies must never brownout, surge or cause power
outages. Everything always works as expected. Riiight.
Slowly applied power is not bad for power supplies.
It's not so much a matter of "slowly applied", it's stutter, spikes
and sags. Mind you, the only definition of "slowly applied" would be
a slow fade from blackout through brown to full power.
Some PSUs are "universal", i.e. auto-detect and handle 115V or 220V AC
input (where I live, power is 220V AC). Others are set for the local
mains voltage. Either way, the expect power to be of a certain level,
and don't like exceptions - but having said that, I find even "normal"
PSUs are surprisingly tolerant of bad power, e.g. visible dipping of
light brightness, etc.
Then again, there's a limit to what they can take; one day here, the
lights in the room suddenly got brighter and my PC immediately reset.
In the next 2 days, I had three PCs in for fixing, that dated their
problems from this moment; one with a popped PSU.
supplies include an internal circuit to slow the power up.
"Power Good"; yep. Really cheap PSUs have been known to simply wire
that to the +5V DC output ;-)
Will a failed power supply cause disk drive or other
hardware failures? Maybe if that same defective human bought
the power supply. But power supply failures must never damage
any hardware in any computer as was standard even 30 years ago
Having said that, I've seen a common and nasty failure pattern with
some ATX PSUs. Typically, they fail with a loud BANG when plugged in
or switched on (ATX "power" switch). The PC won't power up, and if
you shake it, you will hear a rattle. Open up the PSU, and you will
see a loose empty shell of a blown electrolytic cap, as well as loose
bits of dull aluminium foil from inside.
You will then typically find that mobo/CPU, RAM and HD are stone dead,
and that the CD-ROM will fail within a week.
What has happened, is that the PSU spiked the +12V line when it died,
and this fried most things that were connected to it.
Blowing up when ATX "on" is an interesting failure pattern in itself.
I suspect what is happening here, is that the standby current when the
ATX is "off" warms up the circuitry (given that the fan is off during
this time). Then when turned "on", the extra load goes pop!
I've seen PSUs that get warm (as in, the back of the case gets warm)
when plugged in but "switched (ATX) off". When running, they cool
down. These are often the same PSUs that go postal on the +12V
I'm seeing a LOT more PSU failures post-ATX than pre-ATX. In fact,
before Intel's bulky Pentium II necessitated case redesign dressed up
as a new (ATX) standard, the issue of bad PSUs killing other hardware
was theoretical only, in my own experience (building PCs pro since
1995, and for fun before that going back to 1988).
That assumes a computer guy (who first demands numerical
specifications) - not a bean counter - purchased that supply.
Most of us builders get whatever PSU came with the case, and most end
users get whatever PSU came with the PC. Guess the rest
The bad series of PSUs I referred to didn't come in the cheapest
available cases, BTW.
Damage from power restoration is mostly urban myth. In
fact damage often blamed on power restoration may have
actually occured just before power is lost.
Sure; and that's the only "urban myth" mileage I'm going to let you
get away with <g> Hard to tell whether a spike at the time the power
failed (likely causing the power to fail) is what did the damage, or
rough power on re-entry.
Would you strobe your mains power to your own PC (i.e. switch the
mains on and off really quickly, or plug in the mains lead gradually
so you can enjoy the sputtering noises and sparky smell)?
2) Welcome to the redundancy provided by NTFS. Unlike FAT,
an NTFS filesystem does not have "any damaged files are
arbitrarily broken to some assumed size or deleted". Any file
arbitrarily broken (not properly stored on hard drive) is deleted.
Oh, that's a BIG improvament! So instead of having a truncated file,
I have no file at all. Gee, thanks.
Filesystem resorts back to the last known good copy
of that file. FAT has no such abilities with is why FAT
filesystems might put data at risk during a power loss.
Power loss, by interrupting file operations, is inevitably going to
lose or corrupt data. Even if your file system was utterly brilliant,
there still might be unsaved or unflushed data in RAM that will be
gone forever. So power outages are to be avoided.
Don't confuse a sane file system (which reduces OS support calls) with
preserved data. Even MS doesn't pretend NTFS preserves data.
3) Any drive that does not include this 'safe shutdown'
feature belongs with those power supplies that are damaged by
a power restoration. Defective hardware in a pile labeled
with the name of the person responsible for buying crap.
Oh, give it up, will ya? Don't ever think that because something's
designed to work in some way, that it actually will.
There aren't many bad HD vendors around these days, in fact it's an
industry that's had quite a shake-out, and we seldom see new entrants
anymore. Over the years, there have been some sucky vendors (JTS,
anyone?) but what one does find, is that if you sell more than 100 of
any vendor's HDs, and wait a few years, you will see some fail.
I have piles of failed HDs, and they aren't all JTS. There are
Fujitsus, WDs, IBMs, Seagates, Quantums, Maxtors - the lot. Many have
"just one bad sector"; many are stone dead, and several are between
these extremes. Some are full of bad sectors and latency, being the
same pattern as "just one bad sector" a bit further on, while others
work fine but take 3 or 4 power-ons before they spin up, etc.
From what I've seen, I stand by my advice to set CMOS so that the PC
stays "off" after a power outage. It's SOP on the PCs I build.
--------------- ----- ---- --- -- - - -
I *am* a power user!
I have electricity bills to prove it!
and Happy New Year