Guess what Plato...

All that talk about hard drives failing and today I start getting physical
errors from one of my WDSE drives. Run the diagnostics, almost immediately
fails the quick test, also shows 84KB of bad sectors.

I think I should just plug my machines directly into the power line and run
them in a lightning storm. I go through painstaking effort backing up
drives, using UPSs to control power outages and spikes/brownouts, and the
drives still fail. Other people use $5 surge suppressors and nothing else
and their drives last for years. Doesn't make sense.

--
website: http://www.demonlag.com
AIM: FrznFoodClerk

DeMoN LaG

DeMoN LaG wrote:
>
> All that talk about hard drives failing and today I start getting physical
> errors from one of my WDSE drives. Run the diagnostics, almost immediately
> fails the quick test, also shows 84KB of bad sectors.

One thing I do before I install a new drive for somebody is test the PS
voltages. I'm certainly not going to guarantee a new drive for a year
[which includes redoing windows and all their apps] if the PS shows 5.3
or 5.35 volts on the 5 volt line, which, in fact, I _very_ often see on
PCs with failed hard drives. I also insist on that ten buck surge
protector. Interestingly, in 9 years of doing this my partner and I have
never had to do a warranty hard drive replacement following the above.
Of course, this excludes those times when hard drive makers produce a
defective model for a bit which fail no matter what. Sometimes I also
wonder if surges via the phone line can phark hard drives as well.


> I think I should just plug my machines directly into the power line and run
> them in a lightning storm. I go through painstaking effort backing up
> drives, using UPSs to control power outages and spikes/brownouts, and the
> drives still fail. Other people use $5 surge suppressors and nothing else
> and their drives last for years. Doesn't make sense.

Plato <|@|.|> wrote in news:407cc7a4$0$182$:

> protector. Interestingly, in 9 years of doing this my partner and I have
> never had to do a warranty hard drive replacement following the above.
> Of course, this excludes those times when hard drive makers produce a
> defective model for a bit which fail no matter what. Sometimes I also
> wonder if surges via the phone line can phark hard drives as well.

There is no phone line hooked to any of my machines, and these are drives
that died on stuff like $90 Antec power supplies, not just on generic
garbage. The supply that drive is fed from right now reads 4.85v on the +
5, which though not perfect, is only like 3% or 4% off of spec, which
should be fine. I've got alarms set at 5% and none of them ever trip. I
dunno. I think the failure rate of large capacity drives is higher than
lower drives. The same machine that has gone through 2 of these drives has
had the same 6.4GB quantum through it all. That quantum gets way more work
than any of the large drives do, as they are like, streaming a 5MB MP3 file
over a network vs the quantum running a mail server and doing constant
document backups and the like. I'd say utilization is in the 80/20 range
of 80 quantum 20 other.

--
website: http://www.demonlag.com
AIM: FrznFoodClerk

DeMoN LaG wrote:
> Plato <|@|.|> wrote in news:407cc7a4$0$182$:
>
>> protector. Interestingly, in 9 years of doing this my partner and I
>> have never had to do a warranty hard drive replacement following the
>> above. Of course, this excludes those times when hard drive makers
>> produce a defective model for a bit which fail no matter what.
>> Sometimes I also wonder if surges via the phone line can phark hard
>> drives as well.
>
> There is no phone line hooked to any of my machines, and these are
> drives that died on stuff like $90 Antec power supplies, not just on
> generic garbage. The supply that drive is fed from right now reads
> 4.85v on the + 5, which though not perfect, is only like 3% or 4% off
> of spec, which should be fine. I've got alarms set at 5% and none of
> them ever trip. I dunno. I think the failure rate of large capacity
> drives is higher than lower drives. The same machine that has gone
> through 2 of these drives has had the same 6.4GB quantum through it
> all. That quantum gets way more work than any of the large drives
> do, as they are like, streaming a 5MB MP3 file over a network vs the
> quantum running a mail server and doing constant document backups and
> the like. I'd say utilization is in the 80/20 range of 80 quantum 20
> other.

What do you guys use to test the PS?

--
Night_Seer

DeMoN LaG wrote:
> All that talk about hard drives failing and today I start getting physical
> errors from one of my WDSE drives. Run the diagnostics, almost immediately
> fails the quick test, also shows 84KB of bad sectors.
>
> I think I should just plug my machines directly into the power line and run
> them in a lightning storm. I go through painstaking effort backing up
> drives, using UPSs to control power outages and spikes/brownouts, and the
> drives still fail. Other people use $5 surge suppressors and nothing else
> and their drives last for years. Doesn't make sense.
>
I used to bug the reliability engineers that worked for me by insisting that
there's no such thing as an electronic failure--they're all mechanical.

Most mechanical devices like hard drives fail for mechanical reasons. Even
electronic failures are due to a physical, (mechanical), happenings. A
power supply is not likely to cause a failure unless it's very much over
voltage, or contains enough ripple to cause capacitator failures.

Of course a high enough surge, not suppressed by the PSU, will damage
electronic circuits. A cheap surge supressor won't help much. w_tom's
correct about this! A lightning induced surge is on the power lines
in parallel, and will seek ground right through any surge supressor.
A UPS can help if it has sufficient input/output isolation.

The highest "electronic" failure rates are due to connectors, including
solder joints, improperly plated through circuit traces and broken CB
traces.

Next are bearing and moving component failures. The newer higher speed
HDs put greater stress on the bearings which also operate at higher
temperatures. They also use much higher density of data on the platters.
This means smaller surface defects can cause failures. In addition the
servo accuracy required is higher. Keeping a drive cool is more important
than the perfect power supply.

Finally, the last is "luck of the draw". They're lemons in every
production batch--we all get one once in a while.

Virg Wall
--

It is vain to do with more
what can be done with fewer.
William of Occam.

Night_Seer wrote:
>
> What do you guys use to test the PS?

A simple $20 voltmeter. Fits in your shirt pocket. Has a black and a red
prong thing to put in the connectors that go to your drives. Generally,
the 12 volt line is not a problem. It always seems to be the 5 volt line
that increases volts with age. If I see 5.3 volts or above the PS has to
be changed.

Plato <|@|.|> wrote in news:407e0748$0$40638$:

> A simple $20 voltmeter. Fits in your shirt pocket. Has a black and a red
> prong thing to put in the connectors that go to your drives. Generally,
> the 12 volt line is not a problem. It always seems to be the 5 volt line
> that increases volts with age. If I see 5.3 volts or above the PS has to
> be changed.

What do you do when you see:
1.73 (1.75)
3.13 (3.3)
4.80 (5)
11.89 (12)
?

--
website: http://www.demonlag.com
AIM: FrznFoodClerk

I suspect slowly failing electrolytic capacitors. Resulting
high ripple voltage would create intermittent failures
elsewhere in the system - such as a disk drive. Your meter
readings are typical of excessive ripple voltage.

Where is the failing capacitor? Maybe in the power supply.
An oscilloscope (or one of those new meters that acts like an
oscilloscope) is required. Meter has reported something is
wrong requiring further analysis. Ripple voltage may be
excessive. If so, then you now have a solid reason why
failures occurred. A problem that others might have instead
speculated to erroneously blame a hard drive. But first,
excessive ripple voltage must be observed with appropriate
test equipment.

DeMoN LaG wrote:
> What do you do when you see:
> 1.73 (1.75)
> 3.13 (3.3)
> 4.80 (5)
> 11.89 (12)
> ?

DeMoN LaG wrote:
>
> What do you do when you see:
> 1.73 (1.75)
> 3.13 (3.3)
> 4.80 (5)
> 11.89 (12)

nothing.