How much power does a hard drive use?

[half_pint]

I am just wondering as I have 2 hard drive (and a CDrom a cd/rw)
and I wonder is I am pulling too much power (which might explain
by beeps at boot up (three) sometimes.

I know I only have a 90watt power supply.
I think my CPU uses 15-20watts also.

Thanks. (2 & 3 gig drives)

 

[DaveW]

Your power supply should be about 300 Watts for that load.

 

[kony]

I am just wondering as I have 2 hard drive (and a CDrom a cd/rw)
and I wonder is I am pulling too much power (which might explain
by beeps at boot up (three) sometimes.

I know I only have a 90watt power supply.
I think my CPU uses 15-20watts also.

Thanks. (2 & 3 gig drives)

Most drives have the power requirement listed right on the label. Hard
drives may consume more briefly while they spin up though. 90W is pretty
borderline except for an old Pentium I or lower speed Celeron with an
everything-integrated motherboard.

It may not be that you're exceeding the power supply capacity, but that
it's near the end of it's life, or a combination of the two.

A bit more detail of the situation, system, etc, might be helpful.

 

[half_pint]

In the manual it says the PS is ATX 90W.

I comes with one hard drive (obviously) and a cdrom with a spare
5.25 inch drive bay. So it is only designed for 3 drives of any type.
I could well be pushing it a bit with my 4.
It has onboard sound and video though.

I just looked in a PC mag and the smallest I could find was 300W!!
so my 90W does seem inadaquate not even a 1/3 of the minimum available!!
I would guess extra memory uses more power too.

The power supply theory does fit in pretty welll with the symptoms.

If I can run memtest OK for 10 hours I can hardly blame the memory.
Certaintly not enough to claim a refund.

Maybe switching the speakers off will help - lol.

 

[half_pint]

Most drives have the power requirement listed right on the label. Hard
drives may consume more briefly while they spin up though. 90W is pretty
borderline except for an old Pentium I or lower speed Celeron with an
everything-integrated motherboard.

It may not be that you're exceeding the power supply capacity, but that
it's near the end of it's life, or a combination of the two.

A bit more detail of the situation, system, etc, might be helpful.

A lot of the somewhat confusin details are posted in this group under
my id ( also in microsoft.win98.gen_discussion a little while back)

My system is a Patriot (never heard of it ) PC.
It has a Packard Bell 850 mobo (99% sure)
Cyrix MII -300
128 meg edo ram (originally 32)
Onboard sound and video.
3 gig master 2 gig slave.
CD/RW master CDROM
Floppy, keyboard and mouse (bog standard).

Also I often found only one CD drive would work rarely
both.
Basically I used the last spade bay with the cd/rw so the
2 hard drive is not fixed to anything!!
(Only designed for three 5.25 drives).

When you consider the smallest PS I could find on sale
was 300W my 90W PS does look a triffle inadaquate!!!

I sometimes get reboots and beeeps at startup, also
OE often freezes (see below)

Also I put my news files on my slave drive for OE so this maybe
a clue. As OE will have to use both drives to run I guess, thus
potentially causing power to dip.

I also put my tempory internet file on there too but I don't seem
to have may probs using IE so I could be wrong.

Just looked at the 2 gig drive 12V 0.4Amp or 5Volt 0.5Amp
so thats 2.5 - 5 watts. (not a lot really?).
I can't see the other drive (3 gig untill I unscrew it)
(3.2gig Western Digital) -
Just found the spec for that 5.35 watt (average),

 

[w_tom]

If the power supply was undersized, then a 3.5 digit
multimeter would demonstrate the problem. Critical voltages
are monitored. If supply is at upper power limit, then
voltage drops. Computer would then complain or crash during
this voltage drop. IOW don't guess or wildly speculate.
Watch voltages with a multimeter to learn if power supply
cannot supply necessary power.

Many current and professionally configured systems run as
little as 200 watt supplies without problem. Unfortunately
hype has got some assemblers promoting 500+ power supplies for
systems that never consume half that power. Don't let hype or
speculation make your decision. Simply monitor red, yellow,
and orange wires from power supply. If one of those voltage
outputs is overloaded, then you will see voltage suddenly
drop. Less peripherals (less load on power supply), then
voltage does not drop. If no voltage drop, then power supply
is sufficient for the load.

 

[half_pint]

Also I think you would expect a power surge at start-up.

I tried to find evidence.

http://lists.freebsd.org/pipermail/freebsd-hackers/2003-December/004709.html


I am think this explains a lot perhaps.
When my system is switched on everything lights up togeather,
both hard drives and both cdroms kick in, this may cause the
power to the ram to dip, (as well as everything else) causing
all sorts of errors.
When I restart various capacitors will already be half charged
so the dip will not be so bad and it usually succeeds second
time for me and then works just find.
So 'cold starts' would be the hardest, as appears to be the
case with me IIRC.
The probs with OE maybe due to it using both drives togeather.
Maybe I put its message store back on the C: drive not D:

Anyway its all pretty complicated but the power supply
looks a prime suspect.

 

[w_tom]

Power supplies - even on 1950s televisions - have inrush
current limiters that actually starve a power supply during
power up. Not a problem since the power supply must stabilize
before computer even starts to boot. This universal circuit
design even in computers long before the PC existed.

Sure a power supply might initially draw 20% more power.
Who cares? Only important consumption is on computer side and
when computer is executing. Again, the solution is found in
basic facts provided by a 3.5 digit multimeter.

Notice what that other cited post did. He did not make a
conclusion based upon other's speculations. He took a Fluke
meter and tested only THAT system. IOW even that
lists.freebsd.org post demonstrates the answer. Act as an
educated technician. Get the meter and measure. No one can
answer the OP without first obtaining those numerical facts.

 

[VWWall]

If the power supply was undersized, then a 3.5 digit
multimeter would demonstrate the problem. Critical voltages
are monitored. If supply is at upper power limit, then
voltage drops. Computer would then complain or crash during
this voltage drop. IOW don't guess or wildly speculate.
Watch voltages with a multimeter to learn if power supply
cannot supply necessary power.

You have to be careful here! If the overload is on a lead being sensed by the
PS, a meter on a different voltage lead may actually show a voltage
increase, due to the supply increasing voltage to the first lead to
compensate for IR drop. In addition, in spite of its increased
accuracy a digital meter may not be the best fot measuring rapid
voltage changes. Most, (all inexpensive ones), use a ramp up comparison
which indicates when the test voltage is encountered. This may
take a second or so. In addition auto-ranging meters take time to
set the range correctly. A plain old D'Arsenval, (moving coil), meter
may show transient fluctuations better than a digital one. A cheap
digital meter may indicate to several (3 1/2) digits but its internal
voltage standard may be accurate to only +/- 2% or so, and may vary
with temperature. As you say about PSUs---read the meter specs, and
don't believe everything you read! Digital meters evolved from
Fluke 175 0.15% accuracy ($170) to the =/- 1% =/- ).5% FS =/- 1 in
last digit for about $25 at Radio Shack.

Virg Wall

 

[half_pint]

Yes it would be nice to have one of those but unfortunately I don't.
Maybe I will buy one if they are fairly cheap, it would come
in handy anyway.
Actually I am sure my brother(s) uses one for work so maybe I
can borrow one.
Or maybe I get one on ebay?
Would it show a brief voltage drop?
A drop of a few milliseconds might cause the problem.

Then I have the prob of what caused the drop, general
overload or faulty memory or a combination of both!!

Actually I have just been thinking ( which is sometimes a dangerous thing
to do for me!!)

I could (if I had a soldering iron and resistor) solder a resistor across
the
power supply to draw off some power. If I drew off say 10-20% of the
power and the booting problem manifest itself on a regular basis that
mighr prove something (or not!).
I am not sure what the PS voltage is though.
Power =V(squared)/R so 10% of my 90W = 9W=VV/R
if V=12 then R=144/9 ~= 150 ohms.
If V=5 R~=2.5 ohms.

I could also permantly damage my computer in the process - lol !!

Could I set up such a meter to record or beeb if the voltage dropped
below a certain level?

Anyway my computer has behaved it self for the last 3-4 hours.
It usually decides to freeze when I have spent 1/2 an hour typing
a reply in OE so perhaps it realy does just hate me!!
Otherwise the problem is a fairly minor irritant.

I used to have to reboot a few times a day anyway due to software
probs (exceptions, out of memory etc..), maybe some of these were
due to the PS? Who knows? Not me anyway!!!!

regards half_pint.

 

[beav AT wn DoT com DoT au]

I am just wondering as I have 2 hard drive (and a CDrom a cd/rw)
and I wonder is I am pulling too much power (which might explain
by beeps at boot up (three) sometimes.

I know I only have a 90watt power supply.
I think my CPU uses 15-20watts also.

Thanks. (2 & 3 gig drives)

I have a very small dos program that tells you what bios error codes
are. I think, however, that I would get caned if I posted it here.
Would you like me to email it to you?

--
-Luke-
If cars had advanced at the same rate as Micr0$oft technology, they'd be
flying by now.
But who wants a car that crashes 8 times a day?
Registered Linux User #345134

 

[VWWall]

Power supplies - even on 1950s televisions - have inrush
current limiters that actually starve a power supply during
power up. Not a problem since the power supply must stabilize
before computer even starts to boot. This universal circuit
design even in computers long before the PC existed.

Among the specs you always site for PC PSUs, is one for inrush
current. This is a moot point for any ATX supply since the input
capacitors are fully charged whenever the supply is connected to
the line.

When starting, a PC draws more current, usually from motors on
drives. This can cause a voltage drop in a under-rated PSU.
On a second start, the bearings are warmed up and may require less
current. The BIOS usually alots time for drives to come up to speed,
so the rest of the computer is drawing less power at this time.

An interesting experiment is to put a 1 ohm resiator in the AC power
line feeding the computer. Put your 3 1/2 digit meter across it at
start-up and actually _measure_ the power consumed. Compare this with
the power used by the same computer under full disk and CPU activity.

Sure a power supply might initially draw 20% more power.
Who cares? Only important consumption is on computer side and
when computer is executing. Again, the solution is found in
basic facts provided by a 3.5 digit multimeter.

Try some measurements yourself. You might be surprised!

Virg Wall

 

[half_pint]

You have to be careful here! If the overload is on a lead being sensed by the
PS, a meter on a different voltage lead may actually show a voltage
increase, due to the supply increasing voltage to the first lead to
compensate for IR drop. In addition, in spite of its increased
accuracy a digital meter may not be the best fot measuring rapid
voltage changes. Most, (all inexpensive ones), use a ramp up comparison
which indicates when the test voltage is encountered. This may
take a second or so. In addition auto-ranging meters take time to
set the range correctly. A plain old D'Arsenval, (moving coil), meter
may show transient fluctuations better than a digital one. A cheap
digital meter may indicate to several (3 1/2) digits but its internal
voltage standard may be accurate to only +/- 2% or so, and may vary
with temperature. As you say about PSUs---read the meter specs, and
don't believe everything you read! Digital meters evolved from
Fluke 175 0.15% accuracy ($170) to the =/- 1% =/- ).5% FS =/- 1 in
last digit for about $25 at Radio Shack.

Virg Wall

I guess you might need some clever (and expensive) storage
oscilloscope?
I suppose you could program a clever (expensive) meter to detect
a transient voltage drop? I don't know as I am unfamiliar with 'em.
What do you think of my idea of soldering a resistor across
the power rails to drain off some power?
It would certainlty be cheap (which is what attracts me!!).
I might end up frying my computer permanently though.(no so cheap - lol).

I also believe it is the capacitor in PSU's which fail over time, maybe
I can get some cheap some where?
My 90W PSU dont seem a lot though, barely a light bulb!

 

[half_pint]

I am not really too familiar with all this anyway and I don't have
a multimeter at the moment so I can't really do much other
than wild speculation at the moment!!

I am unsure what actually happens at start-up regarding powering
things up, my only thought was that residual charge left in capicitors
on the the computer side (not PSU) helped provide enough
power to get over any demand peak, but as I said I am basically
guessing. Nothing seems particularly consistant, which is a big
problem in finding a solution.
Sometimes it's randon and frequent (3-4 time a day) othertimes I have
had a week problem free.
So I could change a component, have two weeks with no problem
only to find the problem returns and I have wasted my money!
Pretty infuriating really. The best solution maybe to upgrade
my computer as it is really pretty old now, but it would be
handy as a back-up.

 

[half_pint]

I have a very small dos program that tells you what bios error codes
are. I think, however, that I would get caned if I posted it here.
Would you like me to email it to you?

Yes please!!!

(e-mail address removed)

 

[half_pint]

Among the specs you always site for PC PSUs, is one for inrush
current. This is a moot point for any ATX supply since the input
capacitors are fully charged whenever the supply is connected to
the line.

When starting, a PC draws more current, usually from motors on
drives. This can cause a voltage drop in a under-rated PSU.
On a second start, the bearings are warmed up and may require less
current. The BIOS usually alots time for drives to come up to speed,
so the rest of the computer is drawing less power at this time.

Hmmmm... thats interesting, maybe the drives are actually still spinning
on the second boot up?
After all even at 5400 rpm on my old disk drives must take a while
to stop spinning? Maybe a minute or two?
I mean 5400 is about 90 revelotions per second which would I expect
take a while to slow down.

Indeed it fits in very neatly with another manifestion of the problem,
that is, why I so often get freezes when using Outlook Express.

You see, I have my power management turned on, and the hard drives
turn off after ten minutes of inactivity.

Now often, when using OE, as I am doing at the moment, typing a message,
the hardrives become idle because everything is in memory.
However sometimes a certain action requires that a drive be accessed and
the drive would have to do a 'cold start' requiring max power, which could
drag down the voltage below a critical level.

I do notice that when the drives turn them selves on again, I hear a loudish
'click' and a whooshing sound (although I might not hear this if the
drive drew too much 'juice' first and caused a freeze/lock-up )

Actually the click may be when the drives goes off as I just heard one now!!

Actually I made it (the drive) start and its a whoosh then a click!

It does sound quite plausible though.

So that one more variable thrown into a rather confusing equation!!!!!!

 

[w_tom]

If doing any computer work, then you need a screwdriver and
a multimeter. Blunt to the point. Both are also powerful
tools that any 'king of the castle' should both own and
understand. If you cannot afford a multimeter, then why do
you have a computer and more than one jacket. Nothing more
need be said. Either you have the meter or you don't really
want to fix your machine today.

If voltage drop happens, then it is long enough for even for
meter to see. Nothing more need be said. That earlier post
said it all. Get the meter. Take those numbers. Additional
postings are only wild speculation until numbers are provided.

It does not matter whether your computer worked now or not.
Fact remains either you have an overload condition constantly
- which only manifests itself as an intermittent failure; or
you don't have a power problem. Only a meter can see existing
computer failure that just did not crash today. Problem can
still exist even if computer did not crash this minute. Meter
reports numbers. No numbers - then only speculation. Get the
meter. Best to also take readings while multitasking many
programs so that multiple peripherals are being accessed
simultaneously.

Your other questions are only confusing you, me, and the
lurkers. Get numbers. Then we can talk.

There are other procedures for other possible reasons for
failure. But until we have established power supply as either
good or bad, then trying to test those other 'usual suspects'
is a waste of time. Power supply is a foundation on which
everything else works. If the foundation is bad, will you
then buy and install new doors to fix the old doors? Get the
meter and report back those numbers.

 

[w_tom]

Power supplies in the days of 90 watts did not have voltage
sense lead. Sense leads on ATX power supplies only evolved
sometime around Pentium 4 time period when currents into
motherboard only began getting high.

 

[CBFalconer]

Yes it would be nice to have one of those but unfortunately I
don't. Maybe I will buy one if they are fairly cheap, it would
come in handy anyway.
Actually I am sure my brother(s) uses one for work so maybe I
can borrow one.
Or maybe I get one on ebay?
Would it show a brief voltage drop?
A drop of a few milliseconds might cause the problem.

Here is where an old fashioned analog meter with a needle is
better. You are likely to get some indication of that hole. With
a DVM you probably won't, depends on the sampling rate and
mechanism etc.

 

[VWWall]

Power supplies in the days of 90 watts did not have voltage
sense lead. Sense leads on ATX power supplies only evolved
sometime around Pentium 4 time period when currents into
motherboard only began getting high.

As you note a sense wire, usually brown and connected to the orange 3.3 V
power lead at pin 11 terminal, was added to the ATX12V spec. This is known
as the default sense lead.

It doesn't require a sense lead per se. The PSU uses the voltage at
one output point in a feedback loop to control the voltage at that
output. The resistance in that output's transformer windings contribute
to the voltage drop. ( E=IR) Raising the voltage at the sense point to
overcome that drop requires an increase in the excitation to the transformer
core. This can cause a lightly loaded output, (smaller I, hence higher E), to
show a voltage increase. If it's a transient, your 3 1/2 digit meter may not
respond fast enough to show it.

Virg Wall