cooling advice please

G

Guest

Guest
Archived from groups: alt.comp.hardware.homebuilt (More info?)

Hello,

My system:

Asus A8N-SLI deluxe mobo
Athlon 64 3500 winchester
2x512MB corsair pc3200 twinx-xl
Audigy 2 zs
2x inno3d geforce 6800GT 256MB PCI-E cards
3x SATA HDD
2x dvd drive
floppy/7in1 card reader combo
600W PSU

The case is a full sized tower with plenty of room in it. Each HDD has a
spare bay below it and they are all seated directly in front of the
front intake fan. The case is equipped with 2 120mm fans (one on the
front, one on the back). It also has a grill on the side which will seat
directly above the gfx cards.

Are the fans supplied enough to cool this beast? I'm a little worried
about the overheating issue. I don't intend to overclock at all. My
cooling options would be to mount a couple of 120mm outtake fans over
the grill on the side and/or fit some water coolers to the gfx cards.

Advice please!

Cheers,

Phil.
 
G

Guest

Guest
Archived from groups: alt.comp.hardware.homebuilt (More info?)

You have more cooling than you actually require:
o Your SATA drives would benefit from direct cooling
---- you have an intake fan
---- 3x SATA drives dissipates ~45W, so need ~10-15cfm
---- clearly your HD is more than adequately cooled
o Your CPU & graphics card are high heat dissipation
---- you have a large 120mm exhaust fan
---- so the case-airchanges-per-hour will be fine

In general you choose a reasonable cooling baseline:
o The PSU fan is assumed to cool the PSU (not the PC)
---- particularly if you want the PSU fan to be quiet
o The CPU & GPU fans merely removes heat from heatsinks
---- that heat is dumped into the case
---- whereupon case exhaust fans must remove it from the case
o If CPU & GPU dissipate 150W @ 100% 24/7, you need 30cfm+
---- 30cfm is /before/ adjustment upwards for case airflow resistance
---- case resistance = hi for punched-case-grills (45-55% free air)
---- case resistance = lo for round-wire-grills (89-91% free air)
---- so you may want to adjust 50% higher to ensure cooling
o So 45cfm from a case fan, or more than 1 case fan
---- a single 120mm or 92mm can move 45cfm relatively quietly
---- a single 80mm will be move 45cfm noisily, twin 80mm quietly

In your case you have twin graphics cards - so CPU & GPU will
be probably capable of dissipatinng 250W at 100% maximum.
That would require 50cfm before adjustment of say 50% - 75cfm.

The aim is to get case ambient close to room ambient:
o Very often it is not due to insufficient case air changes per hour
---- that may not be due to insufficient fans
---- it may simply be due to a high resistance case - change the grills
o An example is removing the case-side to results in lower case temps
---- in such cases reduce case airflow resistance (use the cfm you have)
---- only then consider fans re airflow specs or size etc

Once the case is built, you verify the baseline:
o Run the machine at idle & verify temps are acceptable
---- case temp should be reported as <40oC
---- case temp for Intel low-noise should be <32oC
o Run the machine under load
---- good idea is a CPU burn-in program, but also a memory-test program
---- case-temp should remain within the above limits
o Verify the HD temps
---- a cooler HD generally lasts longer, don't exceed maker's max temp spec

Ideally you should do the test under hot summer conditions, but in reality
it is easy to repeat the test in hot weather and see if any action is needed.

With a 120mm exhaust & intake you should have no problems.
If you have punched-case grills, before build I would change them to a
round-wire grill - it has benefits in terms of noise, not just in airflow terms.

You also do not want to short-circuit airflow too much, you want in at the
front bottom, and out at the rear top (especially behind the CPU if possible).

So don't add extra grills unnecessarily - change based on proving need and
then change just one variable at a time re cause & effect. Keep cables away
from the CPU cooler, care with airflow around twin graphics cards.

Your PSU is a bit overkill, go for a standard Sparkle/FSP-Group/Fortron as
they are well designed without being excessively expensive. Indeed running
a very high wattage PSU at a low level can create more heat: PSUs generate
their highest efficient at high load, not at low load. At low load the higher
inefficiency results in more energy being dissipated as heat (to be removed).

Since the machine is new, do perform a memory burn-in. Usual note.
--
Dorothy Bradbury
www.dorothybradbury.co.uk for quiet Panaflo fans