On Sep 11, 8:40 am, kony <s...@spam.com> wrote:
> On Tue, 11 Sep 2007 04:09:35 -0700, Randy Brick MacKenna
>
> <randymacke...@yahoo.com> wrote:
> >Thanks! Excellent points. You are correct about the case -- this is
> >an old Celeron 2.2Ghz eMachines computer that I rescued from someone
> >who spilled a Coke into it, then decided to wash out with a garden
> >hose. After thorough drying, the machine runs just fine. I put the
> >new WD drives in it because I'm setting it up as a Linux server in my
> >home.
>
> Many OEM cases are unfortunately, only set up to cool one
> drive well.
>
> >So, the case is probably not designed for these newer drives. Late
> >last night I added a high-quality fan that I had, by wire-tying it
> >directly to the drive cage. Now, with the cover on the machine the
> >drives are 97 F (36 C) -- which is fine. The ambient temp of this
> >room (my basement) is always between 60 and 75 F, so that's not a
> >problem.
>
> >I suppose I can live with this -- or upgrade to a better case -- but
> >even so, I am still very surprised that a hard drive that is marketed
> >as "runningvery cool" can only achieve that distinction when it is
> >actively cooled.
>
> If we could get rid of marketing departments, in some ways
> it would help. They may make a claim justified by merely
> beating some other products in a particular area, and IIRC,
> some WD drives, likely yours also, do run a bit cooler than
> others but not so much that it's a good idea to ignore the
> same cooling measures.
>
>
>
> >Maybe they run cooler than other drives with similar active
> >cooling...I'm not sure. Maybe I was hypnotized by some marketing hype
> >(although Tom's hardware guide was quoted, which I thought was fairly
> >unbiased).
>
> Tom's Hardware is a fairly biased site in some instances,
> but in this case they may have made a reasonable comparison,
> if they have measured drive temp with a thermometer instead
> of just an onboard sensor reading (which is only a spot-temp
> of one component on the drive in some cases).
>
>
>
> >If the consensus is that these temps are perfectly normal, and
> >"coolness" can only be achieved through active cooling (whether by
> >direct impinging on the drives, or good case design), then I will just
> >live with it.
>
> Yes I would call them normal, thoguh a bit higher than I'd
> have expected when the ambient is as low as the 60F figure
> you mentioned.
Ah, no -- last night it was a 75F night in the basement.
I guess I'm okay now with my active cooling, until I get a better case
for this system. Gosh, I have four fans now -- PS fan, CPU fan, HD
fan, and case induction fan.
This thing is now really more of a computerized set of fans, than a
fanned computer ;-)
> I suppose I can live with this -- or upgrade to a better case -- but
> even so, I am still very surprised that a hard drive that is marketed
> as "running very cool" can only achieve that distinction when it is
> actively cooled.
Seriously? They have motors in them that spin platters up to, and maintain
at, at seven thousand two hundred revolutions per minute. Do you think that
the heat produced from this motor (and the read/write heads actuator
mechanism) is simply going to slip through the fabric of the universe into
another dimension? They use around 20 watts. Imagine a 20 watt bulb in a
closed box, (or a corner of a box without sufficient air movement). How hot
do you think it would get? (Ultimately, for the purposes of this analogy at
least, all energy is given off as heat.) It doesn't matter much if a bulb
gets hot. Not so with a HDD though.
> Maybe they run cooler than other drives with similar active
> cooling...I'm not sure. Maybe I was hypnotized by some marketing hype
> (although Tom's hardware guide was quoted, which I thought was fairly
> unbiased).
>
> If the consensus is that these temps are perfectly normal, and
> "coolness" can only be achieved through active cooling (whether by
> direct impinging on the drives, or good case design), then I will just
> live with it.
Live with it. Modern drives require active cooling of some sort. Samsung
Spinpoint drives are supposed to be cool too but I have one here as a
paper-weight. It was mounted in a "dead zone" in a client's case in a warm
room and failed very quickly. No good trying for a warranty claim on it as
the S.M.A.R.T. system records the highest temp it's been operated at and, if
it's not within manufacturer specs, no warranty.
All my machines HDDs have fans (usually within 10mm of them) forcing air
directly over both top and bottom surfaces. New drives in old cases (which
were designed with 5,400rpm drives in mind) is the biggest cause of drive
failure in my personal experience. The problem is magnified considerably
when more than one drive is used in such a case as most cases from that era
only had room for a second drive (if indeed they had it) as an
after-thought.
It might not seem like a big step, 5,400rpm to 7,200rpm. However, IME, the
heat output more than doubles. Often the heat from a 5,400rpm drive could be
removed simply by making sure there is good contact with the case and
secured by four tight screws. The case can then 'sink' the drive heat. That
doesn't work with 7,200rpm drives by a long shot.
--
TTFN.
On Tue, 11 Sep 2007 10:50:35 -0700, Randy Brick MacKenna
<randymackenna@yahoo.com> wrote:
>I guess I'm okay now with my active cooling, until I get a better case
>for this system. Gosh, I have four fans now -- PS fan, CPU fan, HD
>fan, and case induction fan.
>
>This thing is now really more of a computerized set of fans, than a
>fanned computer ;-)
The most common case fan arrangment now follows AMD &
Intel's recommendations (for ATX opposed to BTX), including
a rear case exhaust fan. When the OEMs have this
arrangement they're often able to do away with the CPU fan
by using a duct to cause the air to flow through the CPU
heatsink just before that exhaust fan. By having a
reasonably sized passive front intake only in front of the
drive rack, they can cool at least one hard drive passively.
It is possible to cool more than one passively but often
they rotate the drive by 90' to allow the case to be
shorter.
On Sep 11, 9:27 pm, "~misfit~" <misfit6...@hayoo.com.au> wrote:
> Randy Brick MacKenna wrote:
> > I suppose I can live with this -- or upgrade to a better case -- but
> > even so, I am still very surprised that a hard drive that is marketed
> > as "runningvery cool" can only achieve that distinction when it is
> > actively cooled.
>
> Seriously? They have motors in them that spin platters up to, and maintain
> at, at seven thousand two hundred revolutions per minute. Do you think that
> the heat produced from this motor (and the read/write heads actuator
> mechanism) is simply going to slip through the fabric of the universe into
> another dimension? They use around 20 watts. Imagine a 20 watt bulb in a
> closed box, (or a corner of a box without sufficient air movement). Howhot
> do you think it would get? (Ultimately, for the purposes of this analogy at
> least, all energy is given off as heat.) It doesn't matter much if a bulb
> getshot. Not so with a HDD though.
>
Yup, I agree now that active cooling is the norm for these
drives...thanks to everyone here for the education on that point.
But, are we sure the motor is the primary source of heat in these
drives? The amount of energy required to keep the platters spinning
at constant rotational velocity, once the target velocity has been
reached, should be minimal. Changes in speed (e.g. spin up or spin
down) take energy and produce heat -- but physics tells you that once
a platter is at speed, the amount of energy to keep it at that speed
is hardly anything. It's just got to overcome gravity and friction.
So, I'd bet that keeping a platter spinning at 7200 RPM consumes
barely measurably more energy than one at 5400 RPM.
I thought the main source of heat in these drives was the (ever
denser) packaging of the electronics within them...maybe I'm wrong and
it is the motor...but I can't picture it...
On Wed, 12 Sep 2007 05:30:25 -0700, Randy Brick MacKenna
<randymackenna@yahoo.com> wrote:
>Yup, I agree now that active cooling is the norm for these
>drives...thanks to everyone here for the education on that point.
>
>But, are we sure the motor is the primary source of heat in these
>drives?
It is definitely one source, though others include the
friction of platters spinning at thousands of RPM, the
actuator coil, and the active electronics on the circuit
board.
>The amount of energy required to keep the platters spinning
>at constant rotational velocity, once the target velocity has been
>reached, should be minimal.
Depends on your definition of minimal. Certainly it takes
less energy that spinning up from a cold start, but consider
that the motor is 'sunk to the drive frame and it doesn't
take much heat to warm a hunk of metal by 20C.
>Changes in speed (e.g. spin up or spin
>down) take energy and produce heat -- but physics tells you that once
>a platter is at speed, the amount of energy to keep it at that speed
>is hardly anything.
You can google for drive power consumptions, which may be
lower than that stated on a label or spec sheet (as those
may be peak current or at least in the drives most active
seeking & I/O state) but are still going to be a few hundred
mA on the 5V & 12V rails. Suppose only 5W total and this is
still enough to heat a drive fame by 20C.
>It's just got to overcome gravity and friction.
>So, I'd bet that keeping a platter spinning at 7200 RPM consumes
>barely measurably more energy than one at 5400 RPM.
Perhaps, but there are too many variables to so easily
conclude this as any given 5K4 RPM drive doesn't necessarily
have all parts in common with a different 7K2 RPM drive.
Some 5K4 RPM drives also warm up a fair amount and it is
prudent to ensure they receive *enough* airflow too.
>
>I thought the main source of heat in these drives was the (ever
>denser) packaging of the electronics within them...maybe I'm wrong and
>it is the motor...but I can't picture it...
Why would it have to be one or the other instead of both
(producing significant heat)? A drive frame just doesn't
have a large surface area and is smooth like a metal
container for the parts instead of a maximal heatsink area
as on a purpose-designed heatsink. They could make drive
frames with larger heatsink area, but then it would require
a larger mounting bay or smaller platters.
If you took a CPU producing 5W of heat and put a heatsink on
with similar surface area and minimal airflow through it,
that 'sink would also typically rise by more than 10C over
the ambient temp. Power has to go somewhere, you have a
(typically) 300W or larger PSU and though the system might
only use 50-75% of that potential, it's going to turn into
heat among all consumers. We might as well say it would be
nice of a CPU or PSU didn't produce enough heat to need
active or induced-passive airflow, but performance drives
the industry.
As for motor vs electronics, why do we need to know this?
It's not as though you can effectively separate one from the
other and use the drive. Wishing that computer parts used
far less power than they do is a bit futile, unless you're
willing to either pay significantly more for parts optimized
towards power conservation, and/or lower performance. The
industry is typically focused on performance more than power
and cooling requirements, and of course to have a reasonable
reliability while still staying cost-competitive with other
similar products.
Since a typical computer target application includes a case
with rear exhaust fans, and the ideal is to target OEM
sales, the drive manufacturer needs not design all drives
towards least heat, only towards a manageable level of heat
in the target environment. If the system OEM had to choose
between a drive that costs 40% more but produces less heat,
when the system was designed with enough cooling margin to
accomdate the lower priced drive, then when all else is
equal they'd choose the lower cost drive and so these sales
determine which parameter is more important in selling
drives.
Randy Brick MacKenna wrote:
> On Sep 11, 9:27 pm, "~misfit~" <misfit6...@hayoo.com.au> wrote:
> > Randy Brick MacKenna wrote:
> > > I suppose I can live with this -- or upgrade to a better case --
> > > but even so, I am still very surprised that a hard drive that is
> > > marketed as "runningvery cool" can only achieve that distinction
> > > when it is actively cooled.
> >
> > Seriously? They have motors in them that spin platters up to, and
> > maintain at, at seven thousand two hundred revolutions per minute.
> > Do you think that the heat produced from this motor (and the
> > read/write heads actuator mechanism) is simply going to slip
> > through the fabric of the universe into another dimension? They use
> > around 20 watts. Imagine a 20 watt bulb in a closed box, (or a
> > corner of a box without sufficient air movement). Howhot do you
> > think it would get? (Ultimately, for the purposes of this analogy
> > at least, all energy is given off as heat.) It doesn't matter much
> > if a bulb getshot. Not so with a HDD though.
> >
>
> Yup, I agree now that active cooling is the norm for these
> drives...thanks to everyone here for the education on that point.
>
> But, are we sure the motor is the primary source of heat in these
> drives? The amount of energy required to keep the platters spinning
> at constant rotational velocity, once the target velocity has been
> reached, should be minimal. Changes in speed (e.g. spin up or spin
> down) take energy and produce heat -- but physics tells you that once
> a platter is at speed, the amount of energy to keep it at that speed
> is hardly anything.
I studied physics. We never used the term "hardly anything". If there was
absolutely no friction from the bearings or drag from the air inside the
enclosure then you could be stating the truth. However, that isn't the case.
Perhaps you're familair with why cars, after a certain velocity, have to
increase horsepower by a factor of 10 to increase speed by a factor of 0.1?
The same holds true for HDDs. Even though platters are fairly "slippery" as
far as aerodynamics go, they still require ever-increasing amounts of power
to maintain speed as target velocity increases. They're not running in a
vacuum.
Then there's the friction from the bearings to overcome..
> It's just got to overcome gravity and friction.
> So, I'd bet that keeping a platter spinning at 7200 RPM consumes
> barely measurably more energy than one at 5400 RPM.
I'll take that bet.
> I thought the main source of heat in these drives was the (ever
> denser) packaging of the electronics within them...maybe I'm wrong and
> it is the motor...but I can't picture it...
Picture it. It's the motor overcoming the friction and drag.
The electronics are subject to Moore's law in a peripheral way. The
electronics on-board the drive are being made using a smaller fabrication
process in much the same way as CPUs are. This results in less power
consumption/heat output and better efficiency. The only part of the PCB that
could buck this trend is the on-board cache, that could put a bit more heat
back into the equation. Not so much though.
An anecdote for you: Last week I was pulling down a couple old servers and
testing the parts to sell on my country's equivalent of eBay. I had a bunch
of 9.1 and 18GB drives to "verify". These are 10,000rpm SCSI drives. Here's
a pic of the end of one of them:
The servers in question (Dell PowerEdge and IBM Netfinity) have freakin'
*HUGE* fans and air-flow around the drives. I found out why when I left an
18GB drive verifying, sitting on a book, and wandered off to read for a
while. When I got back to the machine about an hour later the drive was
almost too hot to pick up. (I won't be selling that one, although it still
seems to work fine.) It had finished verifying ages ago, the heat generated
was simply from the motor naimtaining the platter speed. There was no
read/write activity in that time.
--
TTFN.
kony wrote:
> On Wed, 12 Sep 2007 05:30:25 -0700, Randy Brick MacKenna
> <randymackenna@yahoo.com> wrote:
>
>
> > Yup, I agree now that active cooling is the norm for these
> > drives...thanks to everyone here for the education on that point.
> >
> > But, are we sure the motor is the primary source of heat in these
> > drives?
>
> It is definitely one source, though others include the
> friction of platters spinning at thousands of RPM, the
> actuator coil, and the active electronics on the circuit
> board.
>
> > The amount of energy required to keep the platters spinning
> > at constant rotational velocity, once the target velocity has been
> > reached, should be minimal.
>
> Depends on your definition of minimal. Certainly it takes
> less energy that spinning up from a cold start, but consider
> that the motor is 'sunk to the drive frame and it doesn't
> take much heat to warm a hunk of metal by 20C.
>
>
> > Changes in speed (e.g. spin up or spin
> > down) take energy and produce heat -- but physics tells you that
> > once a platter is at speed, the amount of energy to keep it at that
> > speed is hardly anything.
>
> You can google for drive power consumptions, which may be
> lower than that stated on a label or spec sheet (as those
> may be peak current or at least in the drives most active
> seeking & I/O state) but are still going to be a few hundred
> mA on the 5V & 12V rails. Suppose only 5W total and this is
> still enough to heat a drive fame by 20C.
>
>
>
> > It's just got to overcome gravity and friction.
> > So, I'd bet that keeping a platter spinning at 7200 RPM consumes
> > barely measurably more energy than one at 5400 RPM.
>
> Perhaps, but there are too many variables to so easily
> conclude this as any given 5K4 RPM drive doesn't necessarily
> have all parts in common with a different 7K2 RPM drive.
> Some 5K4 RPM drives also warm up a fair amount and it is
> prudent to ensure they receive *enough* airflow too.
>
>
>
> >
> > I thought the main source of heat in these drives was the (ever
> > denser) packaging of the electronics within them...maybe I'm wrong
> > and it is the motor...but I can't picture it...
>
> Why would it have to be one or the other instead of both
> (producing significant heat)? A drive frame just doesn't
> have a large surface area and is smooth like a metal
> container for the parts instead of a maximal heatsink area
> as on a purpose-designed heatsink. They could make drive
> frames with larger heatsink area, but then it would require
> a larger mounting bay or smaller platters.
>
> If you took a CPU producing 5W of heat and put a heatsink on
> with similar surface area and minimal airflow through it,
> that 'sink would also typically rise by more than 10C over
> the ambient temp. Power has to go somewhere, you have a
> (typically) 300W or larger PSU and though the system might
> only use 50-75% of that potential, it's going to turn into
> heat among all consumers. We might as well say it would be
> nice of a CPU or PSU didn't produce enough heat to need
> active or induced-passive airflow, but performance drives
> the industry.
>
> As for motor vs electronics, why do we need to know this?
> It's not as though you can effectively separate one from the
> other and use the drive. Wishing that computer parts used
> far less power than they do is a bit futile, unless you're
> willing to either pay significantly more for parts optimized
> towards power conservation, and/or lower performance. The
> industry is typically focused on performance more than power
> and cooling requirements, and of course to have a reasonable
> reliability while still staying cost-competitive with other
> similar products.
>
> Since a typical computer target application includes a case
> with rear exhaust fans, and the ideal is to target OEM
> sales, the drive manufacturer needs not design all drives
> towards least heat, only towards a manageable level of heat
> in the target environment. If the system OEM had to choose
> between a drive that costs 40% more but produces less heat,
> when the system was designed with enough cooling margin to
> accomdate the lower priced drive, then when all else is
> equal they'd choose the lower cost drive and so these sales
> determine which parameter is more important in selling
> drives.
An interesting fact for you Kony. A friend bought an end-of-line Compaq just
over a year ago. A Socket A Sempron. The first thing to do with a machine
like that, as you know, is to add some RAM. I opened it up and noticed that
the HDD was sitting right up under the floppy, not in the airflow at all.
There is a 90mm exhaust fan and vents in the lower front of the case. I
would have expected to see the drive mounted down there.
On closer inspection of the drive I saw that it was a 5,300rpm drive! An
80GB 5,300rpm drive? I didn't know they made them. Every drive I've bought
in the last century, going back to a 20GB Maxtor, has been 7,200rpm.
(Discounting laptop drives of course.)
Just thought it curious. The PC in question, even with extra RAM, is slow as
a wet week. It really needs that drive replacing. However, that would
require fabricating a new drive cage in airflow. I might get around to it
one day.
Randy Brick MacKenna wrote:
> On Sep 11, 9:27 pm, "~misfit~" <misfit6...@hayoo.com.au> wrote:
> > Randy Brick MacKenna wrote:
> > > I suppose I can live with this -- or upgrade to a better case --
> > > but even so, I am still very surprised that a hard drive that is
> > > marketed as "runningvery cool" can only achieve that distinction
> > > when it is actively cooled.
> >
> > Seriously? They have motors in them that spin platters up to, and
> > maintain at, at seven thousand two hundred revolutions per minute.
> > Do you think that the heat produced from this motor (and the
> > read/write heads actuator mechanism) is simply going to slip
> > through the fabric of the universe into another dimension? They use
> > around 20 watts. Imagine a 20 watt bulb in a closed box, (or a
> > corner of a box without sufficient air movement). Howhot do you
> > think it would get? (Ultimately, for the purposes of this analogy
> > at least, all energy is given off as heat.) It doesn't matter much
> > if a bulb getshot. Not so with a HDD though.
> >
>
> Yup, I agree now that active cooling is the norm for these
> drives...thanks to everyone here for the education on that point.
>
> But, are we sure the motor is the primary source of heat in these
> drives? The amount of energy required to keep the platters spinning
> at constant rotational velocity, once the target velocity has been
> reached, should be minimal. Changes in speed (e.g. spin up or spin
> down) take energy and produce heat -- but physics tells you that once
> a platter is at speed, the amount of energy to keep it at that speed
> is hardly anything. It's just got to overcome gravity and friction.
> So, I'd bet that keeping a platter spinning at 7200 RPM consumes
> barely measurably more energy than one at 5400 RPM.
>
> I thought the main source of heat in these drives was the (ever
> denser) packaging of the electronics within them...maybe I'm wrong and
> it is the motor...but I can't picture it...
Randy, forgot to mention in my other reply to you. If HDDs spinning at
5,300rpm and 7,200rpm used similar amounts of energy (and therefore gave off
similar amounts of heat) why is it that (by far the majority of) laptop
drives are still either 4,200rpm or 5,300rpm? Maybe you know something that
laptop manufacturers don't? The difference in response times between 7,200
and 5,300 drives is vey noticable. If there was a "barely measurable"
difference in energy consumption surely they'd want their laptops to be
significantly more responsive?
On Sep 12, 9:31 pm, "~misfit~" <misfit6...@hayoo.com.au> wrote:
> Perhaps you're familair with why cars, after a certain velocity, have to
> increase horsepower by a factor of 10 to increase speed by a factor of 0.1?
> The same holds true for HDDs. Even though platters are fairly "slippery" as
> far as aerodynamics go, they still require ever-increasing amounts of power
> to maintain speed as target velocity increases. They're not running in a
> vacuum.
Okay, I stand corrected, thanks. I guess I was remembering 'object in
motion tends to stay in motion', and didn't realize there'd be that
kind of magnitude of delta in energy consumption for a platter
spinning at (steady state) 5400 versus 7200rpm. I thought the
maintenance energy to keep them 'at speed' would be similar. The drag
on the platter from the surrounding air, and the friction of the
spindle bearings must be the significant pieces that I was ignoring
(!)
Thanks for the education...For other folks who commented -- I wasn't
trying to solve a design problem here -- just find this fascinating
and wanted to understand the underlying drivers.
On Thu, 13 Sep 2007 13:55:55 +1200, "~misfit~"
<misfit61nz@hayoo.com.au> wrote:
>An interesting fact for you Kony. A friend bought an end-of-line Compaq just
>over a year ago. A Socket A Sempron. The first thing to do with a machine
>like that, as you know, is to add some RAM. I opened it up and noticed that
>the HDD was sitting right up under the floppy, not in the airflow at all.
>There is a 90mm exhaust fan and vents in the lower front of the case. I
>would have expected to see the drive mounted down there.
>
>On closer inspection of the drive I saw that it was a 5,300rpm drive! An
>80GB 5,300rpm drive? I didn't know they made them. Every drive I've bought
>in the last century, going back to a 20GB Maxtor, has been 7,200rpm.
>(Discounting laptop drives of course.)
>
>Just thought it curious. The PC in question, even with extra RAM, is slow as
>a wet week. It really needs that drive replacing. However, that would
>require fabricating a new drive cage in airflow. I might get around to it
>one day.
I can't claim to have seen *all* Compaq models, but of those
I've seen that weren't small form factor or under-monitor
desktop type, they typically had a vertical screw-on
carrier, and a couple slits in the bottom of the case (or
holes tapped for screws).
I'm wondering if the drive below the floppy was something
the owner (or another person, later) did. What kind of case
was it? It's unlikely I have all Compaq cases from that
era, but I do have several.
Re: Caviar drive - running too hot? 
Linear Mode
