Cooling CPU with passive heatsink?

J

John Fryatt

Hi,

I'm currently getting some parts together to build a second PC, and I have a
question regarding cooling.

A bit of background.... I also have a Dell Dimension 8300, which doesn't use
the 'normal' cooling setup for the CPU of a heatsink with fan mounted on it,
but instead has a large heatsink on the CPU, and a 120mm case fan with a
duct to draw air across and away from the CPU heatsink.
It seems to work well, at least the PC hasn't gone into meltdown yet, and is
pretty quiet.

I am using an Antec case for the second PC, which has a large 120mm case
mounted fan at the back. I am considering setting up the same config. as my
Dell to cool the CPU. I would have to make a duct to attach to the case fan
to make it draw air across the CPU heatsink, but that shouldn't be too
difficult.

I've read up on this a bit and the ducted fan arrangement seems, in theory
to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub is,
leading to inefficient cooling.

So, what do you think?

Any recommendations re. heatsinks and places to get them? (in the UK)

Oh yes, the system in question will be running an Athlon XP2500 or XP2800,
probably not overclocked.

Regards, John
 
D

Dave C.

John Fryatt said:
Hi,

I'm currently getting some parts together to build a second PC, and I have a
question regarding cooling.

A bit of background.... I also have a Dell Dimension 8300, which doesn't use
the 'normal' cooling setup for the CPU of a heatsink with fan mounted on it,
but instead has a large heatsink on the CPU, and a 120mm case fan with a
duct to draw air across and away from the CPU heatsink.
It seems to work well, at least the PC hasn't gone into meltdown yet, and is
pretty quiet.

I am using an Antec case for the second PC, which has a large 120mm case
mounted fan at the back. I am considering setting up the same config. as my
Dell to cool the CPU. I would have to make a duct to attach to the case fan
to make it draw air across the CPU heatsink, but that shouldn't be too
difficult.

I've read up on this a bit and the ducted fan arrangement seems, in theory
to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub is,
leading to inefficient cooling.

So, what do you think?

Any recommendations re. heatsinks and places to get them? (in the UK)

Oh yes, the system in question will be running an Athlon XP2500 or XP2800,
probably not overclocked.

Regards, John

For the time, effort, and (possibly money) involved, it might be better to
look into watercooling. -Dave
 
T

Timo Geusch

["Followup-To:" header set to uk.comp.homebuilt.]
John Fryatt was seen penning the following ode to ... whatever:
Hi,

I'm currently getting some parts together to build a second PC, and I have a
question regarding cooling.

A bit of background.... I also have a Dell Dimension 8300, which doesn't use
the 'normal' cooling setup for the CPU of a heatsink with fan mounted on it,
but instead has a large heatsink on the CPU, and a 120mm case fan with a
duct to draw air across and away from the CPU heatsink.
It seems to work well, at least the PC hasn't gone into meltdown yet, and is
pretty quiet.

Compaq at least used to use similar arrangements in their SFF (small
form factor) machines, and they tend to be fairly quiet.
I am using an Antec case for the second PC, which has a large 120mm case
mounted fan at the back. I am considering setting up the same config. as my
Dell to cool the CPU. I would have to make a duct to attach to the case fan
to make it draw air across the CPU heatsink, but that shouldn't be too
difficult.

I've read up on this a bit and the ducted fan arrangement seems, in theory
to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter

(a) and (b) of course are somewhat exclusive. How far away is the fan
from the CPU?
c) A heatsink fan has a deadspot in the middle, where the motor hub is,
leading to inefficient cooling.

Yes, but that's mostly becoming an issue if you absolutely need to
squeeze the last drop of cooling ability out of your HSF.
So, what do you think?

Any recommendations re. heatsinks and places to get them? (in the UK)

Depends on the orientation of the heatsink re the fan - you'd want
something with as much surface area as possible. Something like a
Zalman flower cooler maybe, but you'd need to make sure that air gets
drawn through the fins, not over the cooler.
 
J

Johannes H Andersen

John said:
Hi,

I'm currently getting some parts together to build a second PC, and I have a
question regarding cooling.

A bit of background.... I also have a Dell Dimension 8300, which doesn't use
the 'normal' cooling setup for the CPU of a heatsink with fan mounted on it,
but instead has a large heatsink on the CPU, and a 120mm case fan with a
duct to draw air across and away from the CPU heatsink.
It seems to work well, at least the PC hasn't gone into meltdown yet, and is
pretty quiet.

I am using an Antec case for the second PC, which has a large 120mm case
mounted fan at the back. I am considering setting up the same config. as my
Dell to cool the CPU. I would have to make a duct to attach to the case fan
to make it draw air across the CPU heatsink, but that shouldn't be too
difficult.

I've read up on this a bit and the ducted fan arrangement seems, in theory
to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub is,
leading to inefficient cooling.

So, what do you think?

Im' using a ducted fan for a P4 Northwood build. You could use a huge
gigantic heatsink and it could possibly be passive. The problem is
attaching a large weight to the CPU and make the installation stable.
Intel specifies a weight limit on HS+fan of 450 gr. The Zalman duct
fan that I use is 500 gr with fan, so it's just about the limit. I know
that you'll be running an AMD processor, but similar considerations may
apply.
 
P

Paul Hopwood

John Fryatt said:
A bit of background.... I also have a Dell Dimension 8300, which doesn't use
the 'normal' cooling setup for the CPU of a heatsink with fan mounted on it,
but instead has a large heatsink on the CPU, and a 120mm case fan with a
duct to draw air across and away from the CPU heatsink.
It seems to work well, at least the PC hasn't gone into meltdown yet, and is
pretty quiet.

I've read up on this a bit and the ducted fan arrangement seems, in theory
to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub is,
leading to inefficient cooling.
So, what do you think?

Very common in machines built by larger OEMs, such as Dell, HP/Compaq
etc. Generally very effective and much quieter than the "let's stick
this bloody aircraft turbine on this cheap nasty metal block" approach
favoured by retail and after-market coolers.
Any recommendations re. heatsinks and places to get them? (in the UK)

I believe Dell use Coolermaster heat sinks in some of their machines
coupled with a custom duct designs. AFAIK the same solutions aren't
available retail as they need to be integrated with the case to work
properly. You could probably pick suitable OEM parts from spares
stockists or eBay, although they tend to be for P4 chips as they're
dominant in the "branded" market.
Oh yes, the system in question will be running an Athlon XP2500 or XP2800,
probably not overclocked.

Only ever seen these solutions used on Intel chips which, as a general
rule, don't require quite as much cooling. That said, now Prescott is
with us the OEMs will almost certainly of had to re-design their
cooling solution to cope. I've not taken a recent machine apart to
see if they're still using the same fan/duct approach.

The only OEM machine I came across using an AMD chip used the more
"conventional" cooler which might imply they couldn't get sufficient
cooling using the kind of solution you're looking to use but that's
not to say it can't be done.


--
 
G

Graham W

John said:
Hi,

I'm currently getting some parts together to build a second PC, and I
have a question regarding cooling.
I am using an Antec case for the second PC, which has a large 120mm
case mounted fan at the back. I am considering setting up the same
config. as my Dell to cool the CPU. I would have to make a duct to
attach to the case fan to make it draw air across the CPU heatsink,
but that shouldn't be too difficult.

I've read up on this a bit and the ducted fan arrangement seems, in
theory to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub
is, leading to inefficient cooling.

So, what do you think?

I am a duct enthusiast (I couldn't say fan, could I?) and I wrote up an
article on my website in the 'Miscellanea' section about 'Recent
xp1800+ build' as it was at the time. [It's now a xp3200+ !]
Oh yes, the system in question will be running an Athlon XP2500 or
XP2800, probably not overclocked.

Get the mobile XP-M2500+ which runs cooler and is unlocked.

HTH
 
A

Alex Fraser

Paul Hopwood said:
John Fryatt said:
A bit of background.... I also have a Dell Dimension 8300, which doesn't
use the 'normal' cooling setup for the CPU of a heatsink with fan mounted
on it, but instead has a large heatsink on the CPU, and a 120mm case fan
with a duct to draw air across and away from the CPU heatsink.
[snip]
Only ever seen these solutions used on Intel chips which, as a general
rule, don't require quite as much cooling.

Now come on Paul, you know better than that. The difference between AMD and
Intel's mainstream chips (in processing power vs dissipated power) became
pretty small when the P4 first appeared, and has certainly been small enough
to be ignored for a couple of years.

Alex
 
J

John Fryatt

Only ever seen these solutions used on Intel chips which, as a general
rule, don't require quite as much cooling. That said, now Prescott is
with us the OEMs will almost certainly of had to re-design their
cooling solution to cope. I've not taken a recent machine apart to
see if they're still using the same fan/duct approach.

My Dell machine, which is the ducted cooling machine referred to earlier,
uses a Prescott.
 
J

John Fryatt

Hi Graham,

Took a look at your Web page. Excellent, and it reinforced my plan to go the
duct route.
I am not obsessive about it but the idea of lots of roaring fans is not a
good thing, for me.

I saw a PC for sale on eBay recently which purported to have 12 fans! The
thing must sound like a Bell Jetranger when it starts up! :-(

Regards, John


Graham W said:
John said:
Hi,

I'm currently getting some parts together to build a second PC, and I
have a question regarding cooling.
I am using an Antec case for the second PC, which has a large 120mm
case mounted fan at the back. I am considering setting up the same
config. as my Dell to cool the CPU. I would have to make a duct to
attach to the case fan to make it draw air across the CPU heatsink,
but that shouldn't be too difficult.

I've read up on this a bit and the ducted fan arrangement seems, in
theory to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub
is, leading to inefficient cooling.

So, what do you think?

I am a duct enthusiast (I couldn't say fan, could I?) and I wrote up an
article on my website in the 'Miscellanea' section about 'Recent
xp1800+ build' as it was at the time. [It's now a xp3200+ !]
Oh yes, the system in question will be running an Athlon XP2500 or
XP2800, probably not overclocked.

Get the mobile XP-M2500+ which runs cooler and is unlocked.

HTH
--
Graham W http://www.gcw.org.uk/ PGM-FI page updated, Graphics Tutorial
WIMBORNE http://www.wessex-astro-society.freeserve.co.uk/ Wessex
Dorset UK Astro Society's Web pages, Info, Meeting Dates, Sites & Maps
Change 'news' to 'sewn' in my Reply address to avoid my spam filter.
 
J

John Fryatt

Thanks all for the advice. I kind of expected to get replies saying "don't
be stupid, use the fan/heatsink", so I am quite gratified to see some
positive remarks.
I think I'll give the duct and passive heatsink a try. I like my new Dell
for its relative quietness, and I'll try to get similar levels from the
second machine. Unfortunatley I've ended up with a Radeon graphics card*
which does have a small fan. Can't win 'em all I suppose. ;-)
Maybe I'll try taking off the fan and attaching a larger heatsink, but
that's one for the future, once I get the box up and working.

John

* - not an issue with the Dell as it isn't a gaming box so I'm using a
Matrox card which doesn't need a fan.
 
G

Graham W

John Fryatt wrote:
[...]
Unfortunatley I've ended up with a Radeon
graphics card which does have a small fan. Can't win 'em all I
suppose. ;-)
Maybe I'll try taking off the fan and attaching a larger heatsink, but
that's one for the future, once I get the box up and working.

If the Radeon has a thin plate (with a large hole in the middle to let air
in) covering the heatsink fins secured with some tiny cross-headed
screws, take it off!

The edge of the hole is probably too close to the tips of the fan's blades
and is generating high-frequency zizzing from the tip vortices. You'll
likely
find that the plate is mainly decorative rather than a necessary element
in the air flow design. I did this with my GF4 - Ti and it made a very
acceptable improvement.
 
D

David Maynard

John said:
Thanks all for the advice. I kind of expected to get replies saying "don't
be stupid, use the fan/heatsink", so I am quite gratified to see some
positive remarks.
I think I'll give the duct and passive heatsink a try. I like my new Dell
for its relative quietness, and I'll try to get similar levels from the
second machine.

The secret to the thing is getting the air to flow where it needs to, which
sounds simpler than it is.

For example, blowing air down onto a heatsink forces the air in that
direction (an obvious observation) but pulling air over it means the air
will follow the path of least resistance which, left to it's own
predilection, is almost never through the heatsink; hence the duct and it's
particular configuration. That also impacts the physical design of the heat
sink.

In a nutshell, a typical ducting arrangement will have some 'inlet' to the
heatsink with the heatsink shrouded by the duct to force the air to flow
along it's surfaces.

Problems include flow resistance from the duct itself (walls), and the
distance the air must flow through this 'pipe'; plus, axial fans are better
at 'blowing' than 'pulling' air. Those two often offset any 'improvement'
one envisions may result from not having fan 'dead spots' and the like.

Something people don't always think of, off the top of their head, is that
the shape of the duct also impacts noise. A round hose, typical for home
builders because it's easy, lends an amplified hollow 'rushing' sound
that's quite noticeable, and familiar if you think about it: turn on your
car heater blower. A rectangular duct helps break up internal resonances
and dampen sound reflections, as do bends and internal baffles. They also
add more flow resistance.

From a noise aspect the best place for the fan would be in the 'middle'
somewhere so it's inlet and outlet are both baffled from the outside world,
but there just isn't the room in a normal ATX case for that. Apple does it
that way in at least one of their machines though, and for that reason:
dern near silent.
 
J

John Fryatt

David Maynard said:
The secret to the thing is getting the air to flow where it needs to, which
sounds simpler than it is.

For example, blowing air down onto a heatsink forces the air in that
direction (an obvious observation) but pulling air over it means the air
will follow the path of least resistance which, left to it's own
predilection, is almost never through the heatsink; hence the duct and it's
particular configuration. That also impacts the physical design of the heat
sink.

In a nutshell, a typical ducting arrangement will have some 'inlet' to the
heatsink with the heatsink shrouded by the duct to force the air to flow
along it's surfaces.

Yes, my Dell is like that. The duct goes right over the heatsink.
Problems include flow resistance from the duct itself (walls), and the
distance the air must flow through this 'pipe'; plus, axial fans are better
at 'blowing' than 'pulling' air. Those two often offset any 'improvement'
one envisions may result from not having fan 'dead spots' and the like.

Something people don't always think of, off the top of their head, is that
the shape of the duct also impacts noise. A round hose, typical for home
builders because it's easy, lends an amplified hollow 'rushing' sound
that's quite noticeable, and familiar if you think about it: turn on your
car heater blower. A rectangular duct helps break up internal resonances
and dampen sound reflections, as do bends and internal baffles. They also
add more flow resistance.

Again, the Dell has a rectangular duct. I am planning to use that as a
guide.
Thinking off the top of my head (not finalised yet) I'd probably make the
duct from plastic sheet, folded appropriately and joined.
That's a good point though re. sound effects of different shapes.
 
D

David Maynard

John said:
Yes, my Dell is like that. The duct goes right over the heatsink.

Yeah, I've seen a few of the Dell internals. Nice and simple.

The Compaq AP550 dual P-III is an interesting whole system ducting setup
(completely custom) from front PSU inlet, across the dual CPUs, through the
drives, into the Mobo area, and then out the back. Can barely tell it's on
even in a dead silent room.

Again, the Dell has a rectangular duct. I am planning to use that as a
guide.
Thinking off the top of my head (not finalised yet) I'd probably make the
duct from plastic sheet, folded appropriately and joined.
That's a good point though re. sound effects of different shapes.

Why not plain old paperboard? Probably doesn't reflect sound as well as the
plastic and even easier to work with.
 
W

w_tom

Passive cooling is not difficult. But firsto learn the
'simple multiplication' formulas and relationships. Learn the
"degrees C per watt" rating for a heatsink. Serious heatsink
manufacturers should provide a chart that related that "degree
C per watt" number to an LFM number - the speed that air moves
through that CPU. More LFM lowers the "degree C per watt"
parameter. An exponential relationship exists between
heatsink thermal resistance and airflow speed.

Yes, a ducted fan with more CFMs can increase the LFM across
the heatsink. But as charts will demonstrate, the "degree C
per watt" parameter will decrease less as more air is moved.

This is the point: first do simple theoretical (arithmetic)
analysis. Then confirm that design with a real world example
and measurement devices (ie thermometer). When done, the room
can be at 38 degrees C and still processor does not exceed
manufacturer temperature limits. Better with a 5 degree
safety margin.

Manufacturers who want professional customers and therefore
provide those numbers, the charts, and tutorials on heatsink
design:
http://www.wakefield.com
http://www.aavidthermalloy.com

Manufacturer of fans who provides numbers and charts so help
select a fan or determine what currently exists:
http://www.comairrotron.com

Lastly - I applaud you for doing what every computer builder
should do - learn the whys and hows of computer design and
assembly. Design using the numbers rather than just echo
traditional beliefs.
 
S

spodosaurus

w_tom said:
Passive cooling is not difficult. But firsto learn the
'simple multiplication' formulas and relationships. Learn the
"degrees C per watt" rating for a heatsink. Serious heatsink
manufacturers should provide a chart that related that "degree
C per watt" number to an LFM number - the speed that air moves
through that CPU.

Ari doesn't move through a CPU, electricity does. Did you mean heatsink?
 
S

Spajky

I'm currently getting some parts together to build a second PC, and I have a
question regarding cooling. ...
Dell Dimension 8300, which doesn't use
the 'normal' cooling setup for the CPU of a heatsink with fan mounted on it,
but instead has a large heatsink on the CPU, and a 120mm case fan with a
duct to draw air across and away from the CPU heatsink.
..... is pretty quiet....

make a duct to attach to the case fan
to make it draw air across the CPU heatsink, but that shouldn't be too
difficult.....

I've read up on this a bit and the ducted fan arrangement seems, in theory
to be better than a heatsink mounted fan, because..
a) It is bigger and can move more air
b) Being bigger it can run a little slower and therefore be quieter
c) A heatsink fan has a deadspot in the middle, where the motor hub is,
leading to inefficient cooling.

So, what do you think?
Oh yes, the system in question will be running an Athlon XP2500 or XP2800,
probably not overclocked.

Thermal Power dissipation of that CPUs : around 60-70W
/maybe getting a mobile version of that Cpu for having a third less
power consumption would be a good idea even if they are a bit pricier/

Big companies like Dell have an equipe of staff
to engineer & design an efficiently AirFlow thru case /different types
& setups/ and cooling arrangements, because of doing that good is
quite a time consuming job for an individual & a lot of
experimenting!!!

You can believe me, because I have made it by myself thru time for my
setup running 7 fans in total, but all of them running around 5V &
silent! & in my case there is practically no difference with
temperatures if I have the case opened or closed (believe it or not &
it means excelent air flow!); but my setup has only half of Thermal
Power dissipation than your planned! ...

In your place I would mount together the PC using also bigger HS with
bigger & slower fans & run them on reduced voltage + a quality
designed case with a god probably (overrated) 2-fan PSU with regulated
fan RPMs & self thermocontroled too ...

Full "ducting" in really not necessary; some hard cardBoard paper may
be useful to make some "flaps type" air directors to enhance fluid
airFlow thru the case & reducing fan speeds @ 2/3 (7V!) IMHO would be
enough for your setup!

There are a lot of different mods explained to silence down PCs on the
Web; a lot of times there is no need for much extra spending but
instead a lot of using "common sense" with a bit of thinking before
starting a job! Expensive & cheap does not mean always best & worst!

You can also browse thru my site & DL it for later viewing & studying
a bit; IMHO you can get some ideas there ... / I know, I know, I did
not updated it for awhile & is not completed like it should be/...

IMHO with these days "powerfull" CPUs, I do not recommend trying
making a fanless air Cpu cooling at home; running minimum a bigger fan
@ 5V on its bigger HS is a better solution. (see the temps-my
experiment under Comp/benches ...)

good luck!
 
J

John Fryatt

spodosaurus said:
Ari doesn't move through a CPU, electricity does. Did you mean heatsink?

Best to clarify I suppose, but I think the answer is fairly obvious.
 
J

John Fryatt

In your place I would mount together the PC using also bigger HS with
bigger & slower fans & run them on reduced voltage + a quality
designed case with a god probably (overrated) 2-fan PSU with regulated
fan RPMs & self thermocontroled too ...

Full "ducting" in really not necessary; some hard cardBoard paper may
be useful to make some "flaps type" air directors to enhance fluid
airFlow thru the case & reducing fan speeds @ 2/3 (7V!) IMHO would be
enough for your setup!
IMHO with these days "powerfull" CPUs, I do not recommend trying
making a fanless air Cpu cooling at home; running minimum a bigger fan
@ 5V on its bigger HS is a better solution. (see the temps-my
experiment under Comp/benches ...)

Thanks for the advice. I'll certainly look at your site.

As I see it I am not proposing fanless cooling for the CPU, but rather a
setup where the fan is not mounted directly on top of the heat sink. That
way, as you say, it can be bigger and run slower. The duct is to connect the
fan to the heat sink, in effect. Also, the remote fan will dump the warmed
air outside the cas, rahter than just circulate it inside, and also double
up as general ventialtion device for the whole PC.

I think this whole issue is not that simple, and needs a bit of thinking
about, including doing a few calculations, as w_tom suggested.
Plan A is to get the PC built and running using the supplied heatsink fan
from AMD. Plan B will then be to look at the ducting etc. as discussed here.

Interestingly, my Dell started to acclerate it's fan a couple of times the
other day, making me think about that.
I'm tempted to see what I can do there to improve cooling. Initial thoughts
are to...
1) lap the heatsink and put it back on with good thermal compund (e.g.
'Arctic Silver')
2) make an extra vents in the front of the case (this is a Dell Dimension
8300, and the new 8400 has the same case design except for an additional
vent in the front for "improved cooling" - admittedly for faster 3.4GHz
CPUs)

Thanks for all your advice everyone, it's been illuminating.

Regards, John
 
W

w_tom

Sentence should have read "CPU heatsink". My mistake
properly noted and therefore corrected here.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top