a filtered-air case cooling design

[Gary W. Swearingen]

It seems to me (based only on my reading) that many cases use fans
inefficiently (ie, they're unnecessarily noisy) by having too few
fanless air holes (ie, too much fanning). Consider this extreme
example: The box is an empty cube with one fan blowing in the front
and one equal fan blowing out the back. The air pressure inside is
about the same as outside. Each fan is moving about the same amount
of air as it would if the opposite side of the box (and its fan) were
simply removed or filled with holes. Of course, it's impractical to
remove a whole side of a case, and the airflow pattern would be poor
in such a case, but I suspect that the idea holds partially even in
more practical cases.

Too many cases also fail to filter the air well, allowing boards and
disks to be coated with dust and grunge, leading to poor cooling and
maybe even upsetting delicate circuit capacitances. I suspect that
dust is the cause of many complaints about floppy disk unreliability.


Here's my ideas for good cooling of a typical tower case:

-- Mount all fans (other than those in the PSU) in the side of the
case (near the CPU for ultimate CPU air-cooling, but usually near the
front and near the floor, maybe with some ducting towards the CPU).
(I used a dremel tool. One fan fixed my 500 MHz CPU's problem.)

-- Cover all or much of the side of the case with a furnace filter,
either nicely enclosed (maybe with input near the back) or just taped
on. (I tape on two inches of a firm but airy fiber kind which allows
good airflow even when the fan area is thickly covered with dust.)

-- If desired, add holes on the front to improve airflow around the
HDDs. (I popped out the two 1"-high 3" covers in front of my HDD
cage.)

-- If desired, cover holes with grounded wire mesh to sheild EMI.
(I used 1/4 inch mesh.)

(I didn't use the following ideas.)

-- If needed, add fan-less holes below the PSU. This has the side-
effect of setting up airflow directly from the CPU to the outside.
Outflow ducting will improve this effect (even without a fan there).

-- If needed, add holes above the PSU.

-- If desired, put a noise reduction duct over the PSU and nearby
holes, directing air and remaining noise away from the user. (Not me.)

-- Beware than many front-mounted fans get more air from inside the
case than from outside and so that fan and associated fan holes are
better not used for a better airflow pattern and less noise. (I left
my fan in and it seems to be doing nothing but making noise, though I
suppose it might act as a backup fan should my main fan die while I'm
not listening.)

 

[Matt]

1) Noise comes from fans, not from holes. Reduce noise by using big
slow fans.

2) You get into complications when you start putting holes in the sides.

Putting an intake at the bottom front keeps your hard drives cool.

See pp. 16-20 of:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/23794.pdf

Start with a case that 1) has room for a 120mm intake fan in front of
the hard drive cage, 2) has vent(s) for a 120mm exhaust or two 80mm's
behind the CPU, and 3) is pretty tight otherwise. Cut out the front
wall in front of the hard drive cage and replace it with a 120mm wire
grille. Cut out the bottom of the plastic front panel so you have a big
hidden vent there. Put a piece of furnace filter over your new front
grille, under the front panel. You won't need a fan in front unless you
have a full HD cage or a very hot CPU or GPU. Or you can duct air from
the CPU or the video card. If the rear fan vent(s) are mostly
decorative, cut hole(s) and install wire grille(s) there. Put a speed
control on the rear fan(s) if you are running cool and want to keep the
noise down.

The Antec SLK3700AMB is a very classy case, supports the above mods,
comes with a good dual-fan 350W PSU, and is about $70 shipped from
accupc.com. Of course you don't need to modify it if you don't put a
lot of equipment in it. Its front vent is slightly obstructed, but way
above average. The case is about worth the money even without the PSU.

I don't know whether there are other conservatively-styled cases that
support the above design. There is one Antec case whose fatal flaw is
that the HD cage is turned sideways, which obstructs air flow through it.

 

[Spajky]

It seems to me (based only on my reading) that many cases use fans
inefficiently (ie, they're unnecessarily noisy) by having too few
fanless air holes (ie, too much fanning). ....

Here's my ideas for good cooling of a typical tower case: ....

see how I did it on my site !

-- Regards, SPAJKY
& visit - http://www.spajky.iscyber.com
Celly-III OC-ed,"Tualatin on BX-Slot1-MoBo!"
E-mail AntiSpam: remove ##

 

[Timothy Daniels]

Start with a case that 1) has room for a 120mm intake fan
in front of the hard drive cage, 2) has vent(s) for a 120mm
exhaust or two 80mm's behind the CPU, and 3) is pretty tight
otherwise. Cut out the front wall in front of the hard drive cage
and replace it with a 120mm wire grille. Cut out the bottom
of the plastic front panel so you have a big hidden vent there.
Put a piece of furnace filter over your new front grille, under
the front panel. You won't need a fan in front unless you
have a full HD cage or a very hot CPU or GPU....

You *may* actually reduce the cooling effect on the hard
drives by removing the front wall and installing a filter. Both
have the effect of reducing the turbulence of the air that enters
the case at the front, turbulence that acts to cut through the
boundary layer of air that adheres to the walls of the hard drives
and which gets down to the "bare" warm metal to cool it. That
turbulence is increased by the many sharp-edged holes normally
found in a case's front walls, and if you remove the wall, you'll
increase the flow rate a bit, but you'll also cut way down on the
turbulence. You may *need* a front fan to just replace the
turbulence. Actual temperature measurements should be the
last word here, but when you cut away the front wall, having
to replace it later could be a problem.


*TimDaniels*

 

[Gary W. Swearingen]

You *may* actually reduce the cooling effect on the hard
drives by removing the front wall and installing a filter. Both
have the effect of reducing the turbulence of the air that enters
the case at the front, turbulence that acts to cut through the
boundary layer of air that adheres to the walls of the hard drives
and which gets down to the "bare" warm metal to cool it.

1) In my limited experience with cases, they don't have holes in front
of or near the HDD cage, nor room to put a fan there (which is a noisy
place to put a fan).

2) I read a couple of articles by guys who soft-mounted their HDDs who
learned that the bulk of HDD heat flowed through the sides and
normally into the cage. (One guy added heat sinks to the sides, the
other guy transferred the side heat to extra top and bottom plates.)

That's an interesting turbulence theory, but I'd like to see some
evidence of how important it is.

 

[Matt]

I offer my views:

You get into complications when you start putting holes in the sides.

Putting an intake at the bottom front keeps your hard drives cool.

See pp. 16-20 of:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/23794.pdf

Start with a case that 1) has room for a 120mm intake fan in front of
the hard drive cage, 2) has vent(s) for a 120mm exhaust or two 80mm's
behind the CPU, and 3) is pretty tight otherwise. Cut out the front
wall in front of the hard drive cage and replace it with a 120mm wire
grille. Cut out the bottom of the plastic front panel so you have a big
hidden vent there. Put a piece of furnace filter over your new front
grille, under the front panel. You won't need a fan in front unless you
have a full HD cage or a very hot CPU or GPU. Or you can duct air from
the CPU or the video card. If the rear fan vent(s) are mostly
decorative, cut hole(s) and install wire grille(s) there. Put a speed
control on the rear fan(s) if you are running cool and want to keep the
noise down.

The Antec SLK3700AMB is a very classy case, supports the above mods,
comes with a good dual-fan 350W PSU, and is about $70 shipped from
accupc.com. Of course you don't need to modify it if you don't put a
lot of equipment in it. Its front vent is slightly obstructed, but way
above average. The case is about worth the money even without the PSU.

I don't know whether there are other conservatively-styled cases that
support the above design. There is one Antec case whose fatal flaw is
that the HD cage is turned sideways, which obstructs air flow through it.

 

[Timothy Daniels]

"Gary W. Swearingen" commented:

1) In my limited experience with cases, they don't have holes in front
of or near the HDD cage, nor room to put a fan there (which is a noisy
place to put a fan).

My Dell Dimension has a front plastic bezel which is basically just
a cover to hide all the ugly holes on the bottom half of the front of the
metal case. These holes provide cool air directly to the primary and
secondary hard drives and provide the paths for all the air entering
the case. I've *read* that some cases do have places for fans to suck
air in at the front of the case, and sucking in at the front and blowing
out the back helps to overcome the effect of air leaking in or out through
gaps in the case by keeping the case air near ambient pressure, but you're
right about the noise, and it doesn't provide a way to distribute the cool
air over a broad area.

2) I read a couple of articles by guys who soft-mounted their HDDs
who learned that the bulk of HDD heat flowed through the sides and
normally into the cage. (One guy added heat sinks to the sides, the
other guy transferred the side heat to extra top and bottom plates.)

I've *thought* of heatsinking the sides of the hard drives, and I
even found an aluminum extrusion seller on the web with suitable
extrusions for heatsinking the sides. (See their website at
http://www.accelthermal.com/index.html ) But I didn't think that the
sides provided a major path of heatflow. Could you supply URLs
to those articles?

That's an interesting turbulence theory, but I'd like to see some
evidence of how important it is.

Maintenance of turbulence is applied in light aircraft engine cooling
and it was used by factory teams back when motorcycles were air
cooled. Unfortunately, turbulence is a hard thing to measure and to
observe in detail. The classic evidence for boundary layers and the
effect of laminar flow has been water droplets sitting or just creeping
back on the wing of a light airplane while in flight. Right next to the
wing, the airflow (if the wing is an efficient laminar flow design) is
very slow, and the water (or snow) doesn't get swept off the wing.
Add turbulence, though, in the form of vortex generators, and away
they go.

If you take a look at what's on the market for hard drive cooling,
you'll see that some products take the form of fans mounted to blow
right against the face of the hard drive - a way to increase both the
air velocity and air turbulence at the surface of the metal. How noisy
these are, I don't know, but they do get sold.


*TimDaniels*

 

[dorothy.bradbury]

Filters need to be of a large cross-sectional area to ensure
they don't impact on their airflow. DC Tubeaxial fans in a PC
are very poor at overcoming such resistance - they can suck
about 0.1" of water, humans easily do 12" of water up a straw
(as my bank keeps reminding me, must find a nicer bank).

For a good example of case design see:
o Dell - their later cases show it quite nicely
---- few fans, big fans, not particularly high-flow fans
---- blow-thro design, bit economical on airflow but good
o Apple Dual G5 - this shows blow-thro taken to an extreme
---- bit bonkers on the number of fans
-------- 3-blade = lower noise = less pressure = more fans
---- vast open intake of perforated aluminium mesh
---- straight blow-thro design from front to rear
---- ducting by compartmentalisation vs Picasso ducting

The BTX form-factor due in early 2004 will move towards the
Apple solution - altho not quite as efficient a design. Creditable
design from Apple, some engineer smiled as he left to go home.

Most PC cases have restrictive intakes in c/sectional area:
o Punched case metal grills - only 45-55% free air
---- half the area is not able to pass air, huge waste
o Not great cooling design - PSU fan overloaded
---- insufficient case fans, or rear holes unfilled
---- insufficient case fan directly behind the CPU
------- get hot CPU air out of the case vs recirculate it

Side-mounted fans are best done to blow cool air over a
very hot graphics card - or bottom-front purely as grills to
provide a low airflow-resistance inlet.


I know it sounds twee or heath robinson but consider
making a cardboard template of your case side and using
masking tape to fix it on. Take motherboard etc temps,
then cut a lower-bottom side-hole of 80mm in it, fit a grill.
Take motherboard etc temps again. You may find that is
all you need - assuming the exhaust airflow is managed ok.

Then you can cut the case up (automotive nibbler) knowing
that you are adding value to it vs creating a mess on the floor

 

[Gary W. Swearingen]

"Gary W. Swearingen" commented:

I didn't quite grok the third poster when I wrote that. He basically
just wanted to put holes in front of the HDD cage (and in the lower part
of the front as needed?) and have all fans on the back blowing out.
He'd put filter material behind the front plastic. That's a fine
design, with the advantage of having all fans blowing either in or out
(that's the ideal which is not met in a positive-pressure system with a
standard fan-out PSU), but with the disadvantage of sucking unfiltered
air through floppy/tape/etc drives and having a rather small filter.
I prefer the positive pressure concept, even if it is less efficient
(ie, noisier).

http://www.accelthermal.com/index.html ) But I didn't think that the
sides provided a major path of heatflow. Could you supply URLs
to those articles?

Sorry. I didn't save them and several minutes with
http://www.google.com/advanced_search didn't find them.

The one guy started off by sandwiching (there's a good keyword)
his HDD between two slabs of aluminum, thinking they'd make good
heatsinks for the top and bottom of the HDD. He had stuck a
temperature probe under a HDD circuit board or somewhere like that.
He got little cooling, until he added (IIRC) bars between the plates
and the HDD sides (which were left hanging without the bars).

The other guy mounted the HDD to (aluminum?) "U" channel. IIRC, he
tried a single channel but wound up using two together so it was a
kind of "W" channel. He let them stick out a couple (?) inches behind
the HDD to catch more air.

Maintenance of turbulence is applied in light aircraft engine cooling
and it was used by factory teams back when motorcycles were air ....
back on the wing of a light airplane while in flight. Right next to the

Thanks. There should be lots of creative ways to generate turbulence
in useful places for those going for the ultimate efficiency. Of
course, the turbulence itself generates some heat (reducing the
airflow through the case in the same manner as cables, etc.), but it's
probably worth it sometimes.

If you take a look at what's on the market for hard drive cooling,
you'll see that some products take the form of fans mounted to blow
right against the face of the hard drive - a way to increase both the
air velocity and air turbulence at the surface of the metal. How noisy
these are, I don't know, but they do get sold.

And the worst kinds of junk often get glowing reviews, but I surmise
from my reading that those tiny fans make way too much noise, and in
the worst location, for the amount of air they move. (And which helps
less than one might think if the HDD-side-heat flow theory is
correct.) One could probably do better with other methods, but they
are an easy fix to meet some needs.

 

[Timothy Daniels]

[......]
For a good example of case design see:
o Dell - their later cases show it quite nicely
---- few fans, big fans, not particularly high-flow fans
---- blow-thro design, bit economical on airflow but good
o Apple Dual G5 - this shows blow-thro taken to an extreme
---- bit bonkers on the number of fans
-------- 3-blade = lower noise = less pressure = more fans
---- vast open intake of perforated aluminium mesh
---- straight blow-thro design from front to rear
---- ducting by compartmentalisation vs Picasso ducting

Would you explain that last point about "ducting by
comparmentalisation vs. Picasso ducting"? I can't picture
what you're describing.


*TimDaniels*

 

[Gary W. Swearingen]

Filters need to be of a large cross-sectional area to ensure
they don't impact on their airflow. ....
Most PC cases have restrictive intakes in c/sectional area:
o Punched case metal grills - only 45-55% free air
---- half the area is not able to pass air, huge waste

Which makes it tough to do good filtering with a design that only
blows air out the back, sucking in through the front. It seems to me
that one big filter should cover much of one side. (On the solder-side
of the MB with fans next to HDDs might be best with hot drives, but
the CPU-side near lower front gives otherwise-better airflow.)

I guess some of this depends upon your air quality, but if you live on
a dusty road with with dogs in baseboard-heated rooms and you hate to
vacuum, small filters can get very dirty very fast.

---- insufficient case fan directly behind the CPU
------- get hot CPU air out of the case vs recirculate it

Too many people (not necessarily Dorothy) seem to think that you have
to suck air out the back, instead of forcing it out with positive
pressure, or that you have to blow it in the front or side, instead of
forcing it in with negative pressure. They are about equivalent and
you can also duct air going in or out near both fans and fanless
holes. (The biggest difference is that unducted fans have a
significantly- more directed blow flow than suck flow, but that's
usually only a significant factor for a side fan directly over a heat
sink. The other differences, of course, are that with negative
pressure, it's about impossible to avoid dust in 5" drive cracks, and
with positive pressure, you're wasting PSU fan efficiency.) In any
case, the "side" opposite the fans should have lots of fanless holes,
for best efficiency (cooling/noise).

 

[dorothy.bradbury]

Would you explain that last point about "ducting by

comparmentalisation vs. Picasso ducting"? I can't picture
what you're describing.

Sorry, very cryptic, doing backups deep into early hours.

The Apple G5-Dual when viewed with the side off shows fans
operating in a front-to-rear blow through design. Look closer
and you will see horizontal dividers running across the chassis.
The dividers act as ducts to create compartments, a very smooth
low-turbulent front to rear airflow path - beneficial for low noise.

"Picasso" ducting is where there are many ducts fighting to grab
their bit of the airflow of their particular hot spot. The Apple uses
a simple motherboard design to create front-to-rear compartments.
The BTX form factor could end up quite a mess - duct from fan over
a huge passive CPU cooler, graphics card on riser card horizontally
behind the CPU extending the duct poorly, PCI-X cards nearby.

I guess the BTX design will look better when released, but the initial
releases look a somewhat messy & cluttered design - focused around
using 1 fan to save cost. That's not a bad thing, fewer fans is less noise.
Apple have perhaps gone for more fans to counter complex ducting.

If you can, view a Dual-G5 Apple from the side, cover off.
Then take a look at the BTX pictures, especially the M-BTX examples.

 

[dorothy.bradbury]

The Apple G5 will be an interesting test here re filter/don't:
o Pure low-resistance open-flow duct-sized-matched-to-fan-size design
o So dust build up may be mitigated compared to the worst PC cases

That said, it's also a case where the broad perforated aluminium sheet
front could actually accomodate a filter in terms of physical area. The
holes look like 3R5, a 3mm hole on 5mm pitch, so only ~43% free air.
So a different more open grill to compensate for a similar sized filter.

The Apple fans are 3 blade which generate low-static pressure, so
cfm may drop below design spec with a filter even with changes.

Using the Apple case a benchmark isn't a bad idea though.
It's the front of PC cases that is the problem - it's too much legacy.

 

[Timothy Daniels]

The Apple G5-Dual when viewed with the side off shows fans
operating in a front-to-rear blow through design. Look closer
and you will see horizontal dividers running across the chassis.
The dividers act as ducts to create compartments, a very smooth
low-turbulent front to rear airflow path - beneficial for low noise.

Does this mean there is a fan just inside the front portion of
each duct blowing air through the duct, followed by passive
holes at the rear of the duct? Or are all the ducts fed in parallel
with the size of the exit hole controlling the airflow through each
duct?


*TimDaniels*

 

[dorothy.bradbury]

Does this mean there is a fan just inside the front portion of

each duct blowing air through the duct
Yes.

followed by passive holes at the rear of the duct?

Fan at the rear too. So a push-pull arrangement.

Several horizontal ducts about the width of a fan - so focused airflow.
o Top = Drive compartment - DVD (front) & Hard drive bay (rear)
---- thin fan in front of the drive bay, looks like a radial-blower sideways
o Upper-Middle = Card slot compartment
---- thin fan in front of the bay
o Lower-Middle = Dual CPU compartment
---- 2 large 3-blade fan in front (quiet, low rpm)
---- 2 thinner 7-blade fan at rear
o Bottom = PSU?? Can't see from the pictures, but likely
---- likely position, stability, cooling

The CPU heatsinks are passive (large) with heatpipes.

That description is clearer I think