CPU Cooler designs?

On Tue, 26 Sep 2006 20:55:53 +0100, Gerry_uk
wrote:

Hi kony,

A properly
set up system has no need for the side intake and is usually
as well off having it blocked completely.

Perhaps, but here's something to try.

Run your ordinary ATX fan cooled PC for two hours in the summer
including some time under load. Measure the CPU and MB temps, then
remove the side panel near the CPU, watch the temps FALL like a brick!

If your ambient (room) temp is much higher in summer, it
might be better to make the room hospitible to the user, and
have the system benefit as a side-effect.

If the system is intended to be used in hot environment,
this should be taken into account when designing or
modifying the system cooling.

If you take the side panel off and get lower temp, the real
question is whether the temp was too high before you took it
off. The secondary issue (still quite important) is whether
the OTHER parts are as cool. In general, taking the side
panel off will tend to make the HDDs and other lower-front
quadrant parts run hotter than with it on. Never forget
that the CPU is not the only part that needs to run cool
enough, it is merely the one most people are overly fixated
on- and yet it has a large heatsink.

A balanced system, properly set up, will have a CPU heatsink
of acceptible size. It should never be necessary to remove
the side panel and enough airflow to keep the other
components cool enough, is likewise enough airflow to keep
CPU cool enough. If it is not cool enough and the other
components are, the solution is a heatsink upgrade or to
increase chassis airflow (but the latter increases noise
more than the former, assuming "heatsink upgrade" doesn't
mean merely choosing one with a higher RPM fan on it).

On Tue, 26 Sep 2006 22:19:22 -0500, "Vanguard"
wrote:

"kony" wrote in message
news
On Mon, 25 Sep 2006 22:38:03 -0500, "Vanguard"
wrote:


Blowing air out the side panel means you are sucking in pre-warmed
air
over the CPU before exhausting it. Blowing air into the case
directly
at the CPU means you get the greatest temperature differential.
Since
the PSU is nearby, the air warmed by the CPU gets drawn out and
exhausted through the PSU rather than over the memory sticks or
chipset.

Blowing air in (or having no side fan at all) results in the
most pre-heated air flowing over the memory or chipset.
Granted, the air might be slightly cooler.

Slightly cooler? Try 15 C cooler.

Only if the case wasn't designed properly. CPU temp rise is
in direct correlation to chassis ambient temp and the
chassis should not be 15C hotter. If the chassis is that
hot, the other parts were suffering too and putting in the
side duct may make them even hotter.

Remember the system is not only comprised of a CPU, if the
CPU as hot, so are other parts that you may not be
monitoring actively- and if you aren't monitoring them
actively, it can be even more important to have a larger
thermal margin.

While my room temperature is 20 C,
the inside temperature is 35 C (no, I haven't gotten to modding my
home PC's case yet).

What are you using to determine the case temp? If you rely
on the "system" or "case" temp resport via software, I
presume you realize that is often not the actual chassis air
temp but rather, of a different chip that produces it's own
heat.


Actually I prefer a quiet PC so I let Speedfan
slow down the fans which lets the inside get hotter unless
temperatures exceed the configured thresholds whereupon the fans speed
up and are more noisy. Cooler internal temperatures would mean cooler
air over the CPU (and GPU). Having to push warmer air over the CPU
means less efficient cooling. If I can get MORE cooler outside air
inside the case then there would be fewer times when Speedfan would
have to speed up the fans.

There is no need to speed up the fans at 35C, so there is
either more significant information lacking in your
description, or another issue (which is likely chassis
airflow, if the actual chassis air temp is 15C above room
ambient). Even if the latter is true, we can ignore the CPU
temp and not throttle up fans, rather focusing on the other
parts. Remember that the CPU is more heat resistant than
many of them, and has a longer expected lifespan too.




The key to reducing warm air flowing in circular or
counterproductive directions is to not interfer with the
time-tested and proven chassis airflow pattern from bottom
front to mid-top rear.

Time-tested?

Yes

Time has shown that the ATX case was NOT designed for
best cooling.

"Best" is arbitrary if you only focus on CPU temp,
particularly when you describe above a CPU that is not even
remotely close to overheating!

It provides absolutely no zoning of airflow.

Wrong, it does far better than having a side duct. WIthout
the side duct there is a clear airflow path from bottom
front to mid-top rear. Add the side duct and the bottom
front will necessarily have lower airflow rate.

In fact,
it stupidly enforces turbulence because of the twisting required for
the airflow,

Turbulence ON the parts being cooled is exactly what you
want. Turbulence in the rest of the case is minimal, it is
proven so and WAS TESTED when ATX was spec'd. They didn't
ignore cooling of everything but the CPU though, which seems
to be what you are doing even when CPU is far cooler than it
needs be.

and turbulence is resistance to airflow.

Inside a chassis that is much larger than the intake or
exhaust areas, the effect of turbulence is minimal, beyond
consideration.

Do you see any
ATX-style cases that have channels built in to keep flat cables out of
the way so they don't block the airflow?

It is a marketing feature unless the system integrator has
no idea how to route cables properly. Plenty of system
builders have demonstrated, millions of times, that a case
with ribbon cables and no channels built in will run fine IF
the primary considerations are considered- and they don't
require a side duct, rather as mentioned previously, the
side duct interferes with airflow around cables by reducing
the amount of airflow in this region. IF your side-duct
was blowing out of the case instead of in, it would increase
the airflow around drives, cables, but this then decreases
airflow around (many if not all) CPU VRM circuits, and out
the PSU, and possibly through the CPU heatsink too.

Why did they design it so
hard drives could be shoved against each other (so there is no airflow
between them)?

Your proposition of the side duct does not change this, only
further reducing airflow there.

Further, you only assume they designed to have drives
"shoved up", when in actuality it was case manufacturers
that only tried to put the most drives in the least space
possible that resulted in the "shoved" scenario.

Low front intake is okay for drives in the lower cage
but not in the upper drive cage. ATX (without modification) is a poor
design for cooling.

Drives in the upper cage are also cooled worse if you
introduce a side duct, unless you had the duct opposite the
rack (which is almost never done, it's far lower).

A better question is why you feel they are not cooled well?
Optical drives do not need a lot of airflow, those that are
particularly hot running tend to have a fan in their casing
but once designs that weren't so hot were produced, the fans
were eliminated. If you have a hard drive up there, you are
choosing to not adhere to the normal component placement and
as such, are taking it upon your self to provide a front
intake (perforated if not actively fan cooled) front
faceplate to accomodate this change in drive location.

ATX does accomodate drives fine, if you don't assume some
poorly made case defines what ATX is, rather than the truth-
that ATX is the definition and the poorly designed case made
other design mistakes.



By reducing the bottom front intake rate, by use of a side
intake if not another method, there are lower velocity flow
everywhere except into the 'sink. The exhaust fans will
exhaust at same rate (providing front intake was sufficient)
but a short-loop is created, any air not exhausted into the
air most immedate to the exhaust fan will take a longer path
till exhausted, and slower.

Short loops are exactly what are needed for zoning the airflows to
minimize them from intermingling with each other.

Short loops are robbing the lower front area of the case,
unless you then add a front mounted fan which increases
noise level and then reduces the effect of the side duct.

Intermingling is exactly what your side duct causes, because
you prevent the airflow from going in the proven effective
path from bottom front of case towards the rear, by reducing
the flow rate significantly. This is proven fact and was
taken into account and measured when ATX was spec'd.

Cool the hot stuff
first with direct outside air and expel it immediately.

Wrong, you have to consider the pressure gradients, to have
suction sufficient enough to intake air into the front of
the chassis unless you add another fan which has drawbacks
as mentioned above. Go ahead and add several addt'l fans if
you like, but if you do that you again eliminate the need
for the side duct by more directly addressing the problem
with your case (if it actually has 15C rise in internal air
temp). Actually, if you case has that high a rise, it would
possibly indicate the problem in airflow that I've already
described.




Side intake is mostly Intel's attempt to ship cheaper
heatsinks with P4, particularly Prescott CPUs.

Wrong. Side intake was to circumvent the ATX spec's poor airflow
design.

You have supplied zero evidence of this, thus far only
describing a chassis that is overheating, then that the CPU
is overheating as a result unless you add a side duct. You
have still ignored the REST OF THE OVERHEATING PARTS.

A CPU can run at 60C for years, longer than parts like HDDs,
fans, motherboard can.



The CPU is not the only part that needs cooled

Exactly, so why use the same airflow to cool EVERYTHING?

"Same airflow" means a pseudo-linear path, because this
maximizes flow rate per fan over everything.

You are ignoring a primary detail in system cooling, that
your CPU was not described as overheating and you are not
reporting temps and temp changes of any other parts. This
is a HUGE MISTAKE and quite often why hard drives die.

ATX is a
poor design for cooling. Modders knew that for a long time before
case makers starting adding the extra intake/exhaust ports.

Actually, the vast majority of systems sold and running
today are OEM systems, and they do not have all these
changes the modders made. It may be true that if a
so-called modded had a poor generic case, they needed to
make "some" kind of change to it, and they may not make the
best choice from a airflow:noise ratio perspective, OR they
might be overclocking without a suitable heatink, or as
mentioned several times already, they may not be focusing on
any parts temps except the CPU and system (actually another
chip temp).

BTW, I mod plenty of cases, even early ATX that had no aim
of cooling today's higher heat parts, even overclocking
parts quite a bit as a hobby and do not need to do these
things you imply are important- because I actually meaure
temps of ALL parts that heat up and observe the temp
changes. Most modders don't do this, they just throw a lot
of fans at a situation and end up with a loud system...
certainly louder than Dell et al. OEMs produce.

It is possible to add enough very large, low RPM fans to
offset this and still have low noise:airflow ratio, but then
the expense goes up for quality fans or the maintenance of
these cheap fans increases, and fans mounted on exterior
walls of the chassis always let more noise escape, because
there's nothing to really "escape", they're already facing
the external room.




In fact, I've seen CPU and case temperatures
drop in some cases by reversing the backpanel fan so it is an intake
fan rather than an exhaust fan (its normal position) but you need to
test in your own case.

This should never be done, it necessarily increases temps of
other parts (unless the case was otherwise unusual in it's
airflow before this fan was flipped over).

Not if you add MORE venting, like adding a top grill

LOL.

A top grill is almost never a good idea. If the case were
incredibly poorly designed to it was starving for air, or
had a very very poor PSU that was overheating more than
anything else, having the top vent increase airflow might
seem to help, but still robs the lower front quadrant of
airflow, it is a poor choice in resolving the prior choice
of a poor case in general. Such a modder shows they have
made two mistakes, choosing a bad case then choosing a
lesser effective mod and one that may actually reduce
airflow in other parts of the system.

There are plenty of examples of ATX cases that run fine
without these kind of misguided hacks. You are attempting
to suggest ATX is a problem when ATX was not the variable,
rather the other aspects of particularly poor cases were the
variable.

It doesn't really matter if you agree, there are plenty of
examples of ATX cases cooling fine. The most significant
problems with ATX are these two:

1) CPU manufacturer that provides a cheap heatsink that
needs help. The word Prescott comes to mind and the entire
industry has already acknowledged it and praised subsequent
CPUs for not having this known problem.

2) Video cards with very high heat and small heatsinks so
as to not block one or more PCI slots.

In summary, I will repeat what I wrote above as it is key in
your misunderstanding:

You are attempting to suggest ATX is a problem when ATX was
not the variable, rather the other aspects of particularly
poor cases (and system setup) were the variable. When
someone takes on the role of system designer, selects parts
that don't work very well, only then do they have to find a
solution and what you have described is an attempt to
band-aid the problem instead of addressing it directly, then
ignoring the effects on anything but CPU that was not
overheating at all per your description.

I have demonstrated exactly what i have described far too
many times to count, as have the majority of systems running
today made by most OEMs.

"kony" wrote:
Turbulence ON the parts being cooled is exactly what you
want. Turbulence in the rest of the case is minimal...


You've done a 180º on THAT, Kone-head.


Inside a chassis that is much larger than the intake or
exhaust areas, the effect of turbulence is minimal, beyond
consideration.


You've done a 180º on THAT, Kone-head.


I think I'll call you "One-Eighty Kone-Head".

*TimDaniels*

Keep in mind that the manufacturer's design of the case
must conform to FCC spurious emission requirements
which preclude opening large holes in the chassis. I
really don't know how the "game machine" manufacturers
get away with their plexiglass sides - unless those sides
have a metal film or the manufacturers' low volume production
exempts them from FCC requirements.

*TimDaniels*

On Wed, 27 Sep 2006 11:28:51 -0700, "Timothy Daniels"
wrote:

"kony" wrote:
Turbulence ON the parts being cooled is exactly what you
want. Turbulence in the rest of the case is minimal...


You've done a 180º on THAT, Kone-head.

Nope, read more carefully.




Inside a chassis that is much larger than the intake or
exhaust areas, the effect of turbulence is minimal, beyond
consideration.


You've done a 180º on THAT, Kone-head.


I think I'll call you "One-Eighty Kone-Head".


I think I'll call you miserable.
You torment yourself and falsely assume trolling will help.
Obviously not Tim, or it would've done so already.

On Wed, 27 Sep 2006 11:34:13 -0700, "Timothy Daniels"
wrote:

Keep in mind that the manufacturer's design of the case
must conform to FCC spurious emission requirements
which preclude opening large holes in the chassis.

No they do not have to conform!

An assembled, whole system must pass such tests but a case
alone does not have to.


I
really don't know how the "game machine" manufacturers
get away with their plexiglass sides - unless those sides
have a metal film or the manufacturers' low volume production
exempts them from FCC requirements.

See above, it doesn't have to meet a spec because it has no
power plug and produces zero emissions AS-SOLD.

Gerry_uk wrote:
Hi,

As I understand it, most ATX style PC setups have a fan above the CPU
that blows air down onto the CPU to keep it cool, and the hot air
bounces around inside the PC case until it can find an exit (if it's lucky).

As I see it, there are two problems with this,



darn right!!! but no how you think.

The fans have been installed backwards!!!!!!

The (CASE FAN) fan above the CPU is an exhaust fan. It blows the hot
air out of the computer.

There is also an intake fan on the front at the bottom. Usually hidden
"thanks" to the cases with snazzy fronts and some tiny hole somewhere
stopping you seeing how it can get any air in. But if you open the
computer you'll see a space there to fit a fan.

this also makes sense because hot air rises.

I'm sure I once saw a picture in a document called AMD cooling
requiremetns or something, that showed an arrow indicating the
directino of the air. Anyhow, it is like that.


a) unless you have a vent in the side of the PC case, the air being
taken in by the CPU cooler will not be cool, because it's air from
inside the PC case?

Interesting point. I think I saw in a scott mueller video that CPU
heatsinks do **Suck** air downwards and blow it out the bottom. I
almost forgot about that.
Where did you read that?


b) the hot air from the bottom of the heat sink ends up warming up the
Motherboard?

I was looking at the Dell CPU coolers of the GX280, GX620 workstaions
(Intel P4 / Pentium D) and the PE2400, PE 2600, PE2800 servers (Intel
Xeon). The design is completely different. The air is sucked in from the
front of the case, straight over the CPU and out of the back - how cool
is that? Pretty cool, and there's hardly any noise either.


the front of the case is nowhere near the CPU. And generally the fan
is at the bottom front.

The CPU is at the back of the case near the top, just below the PSU.
It's impossible for a front fan to suck air in over the CPU. It's far
away. It just sucks air into the case.
Unless this server is somehow unlike any computer i've ever seen. But
if it uses an ATX MBRD then I find that hard to contemplate. You
should have included a picture - if you could find one.

"kony" wrote:
"Timothy Daniels" wrote:

Keep in mind that the manufacturer's design of the case
must conform to FCC spurious emission requirements
which preclude opening large holes in the chassis.

No they do not have to conform!

An assembled, whole system must pass such tests but a case
alone does not have to.


So now you're pointing out that an empty case does not
have to conform to FCC requirements. What an insightful
comment! Wot an idiot you are. Even if you have nothing
valuable to add, ya gotta keep spewing.

*TimDaniels*

"kony" wrote:
"Timothy Daniels" wrote:

"kony" wrote:
Turbulence ON the parts being cooled is exactly what you
want. Turbulence in the rest of the case is minimal...


You've done a 180º on THAT, Kone-head.

Nope, read more carefully.


Riiiiight - as you've said before, "turbulence ON the parts is
what aids their cooling, NOT turbulence hitting the parts".
Please differentiate between the two for us Kone-Head,
you'll say ANYTHING FOREVER to be the last one posting.
Start spewing....

*TimDaniels*

On Thu, 28 Sep 2006 19:12:01 -0700, "Timothy Daniels"
wrote:

"kony" wrote:
"Timothy Daniels" wrote:

"kony" wrote:
Turbulence ON the parts being cooled is exactly what you
want. Turbulence in the rest of the case is minimal...


You've done a 180º on THAT, Kone-head.

Nope, read more carefully.


Riiiiight - as you've said before, "turbulence ON the parts is
what aids their cooling, NOT turbulence hitting the parts".
Please differentiate between the two for us Kone-Head,
you'll say ANYTHING FOREVER to be the last one posting.
Start spewing....


Turbulence before the air gets to the part, decreases
chassis intake and exhaust rates to whatever extent this
turbulence persists.

Turbulence on the part is necessary for best effectiveness
of the sink.

The goal is to keep the air moving at good rate, only slowed
down when meeting it's purpose.

Let's look at another example-

You have a lizard that eats insects. It's using camouflage
to sit still and wait for the insects to come near it. It
wants, even needs if to be able to eat enough, for the
insects to approach at a certain rate, then for those
insects to sit still at least long enough to be captured by
it's tongue.

If the insects each in turn, sat still for a long long time
at each interval prior to moving close enough for the lizard
to eat them, which is a lower average rate of travel, the
total insect:time ratio is lower. So the lizard wants them
coming as fast as possible, but the shorter the time spend
in front of the lizard, the lower the insect:capture ratio
is.

The lizard may survive if there are enough insects that
their average rate of travel doesn't matter, and likewise a
computer if someone stuffs extra fans in it to combat the
excessive turbulence extraneous to hot parts contact.

The optimal solution is not to add those fans, not to
interfere with their flow rate either- it's to retain
maximum flow to the parts by the cooling system, and the
rate of exhausting that hot air out of the system. The
(average) ambient temp of the semi-enclosed system depends
on this intake:exhaust rate, so if the total I/O airflow
rate is reduced, it heats up less for each subsequent time
interval because it is closer to the temp of the part
surface it had just been in contact with.

Naturally I don't expect you to agree at all Tim, you were
still trolling as usual but maybe other people won't be as
confused as you are about these OBSERVABLE phenomena.

Again Tim- the reason why cases have the holes in the front
is to keep the ratio of opening to opening length minimized
for reduction of EMI. It would be a trivial thing to set a
metal stamp to a more complex pattern to increaes turbulence
if that were their goal instead.