Help to cool down a 70c PS?

[toddw]

I have the Sparkle 550w 24 pin connector PS. It's running an Abit IS-7 (not version E, or 1, plain IS-7) w/3.2E @ 3.6 (can do 3.8+ but PS heat) and a 6800 GT, 2 HDs, 2 fans, nothing else crazy.

Abit EQ says this thing is idling at 46-50c and after a few hours of gaming at 70c+. Now I know the software is known to not be 100% accurate, but I have latest bios (Nov04) and it seems to display everthing else in the ballpark. Can it truly be 30+ centigrade off in this one area? Seemed to display right on my last PSU. Here's what I'm looking for:
1) Either my temp software is way off don't go there, or
2) I got a bad one and this is a suggestion to cool it. (Please note I already added a fan to the other side in a push/pull configuration)

At $100 I was hoping this would be the end all for my PS needs. I'm about to rip it apart and experiment, Thanks!

 

[nst6563]

are you sure you're reading the temp of the PSU, or the CPU. That sounds normal for an overclocked Prescott CPU temp. The temp probes on your motherboard don't read the PSU temp. They MAY read the PWM temp, which is the onboard 3-phase power circuit that controls the power to the cpu...and if that's what it's reading...that's also normal. The PWM is spec'd to run a lot hotter than the 70c you are seeing. But, put some tiny sinks and some airflow over them to cool them down. If it's your CPU temp, get a better hsf or water cooling for that Prescott.

 

[Ice Czar]

well that push pull mod should have helped
you can physically guess what the exhaust temperature is with your fingers
or you can use most any thermometer at the exhaust or actually place a sensor in the supply like say what comes with a DigitalDoc5

offhand Id say you need to confirm the software, and yes it can be that far off if it thinking its reading a different thermoresistor

A resistive device used for temperature sensing that is composed of metal oxides formed into a bead and encapsulated in epoxy or glass. A typical thermistor has a positive temperature coefficient; that is, resistance increases dramatically and non-linearly with temperature. Though less common, there are negative temperature coefficient thermistors.

so which metal oxides impact its "scale"

a little cut and paste

--------------------------------------------------------------------------------------------------------------------------------
here is a great thread about ducting for a PSU
> http://forums.silentpcreview.com/viewtopic.php?t=8450
and some thermodynamics\aerodynamics infor

Lee Garbutt did a push>pull fan mod on a Turbo Cool 425
http://web.archive.org/web/20040212035712/http://www.leesspace.com/quiet_psu.htm
(archived so photos are now gone)
and this thread at spcr can give you lots of ideas about ducting (again)

but do to the design of the heatsinks and the density of the component package
as mentioned above, I wouldnt do the top fan mod on a Turbo Cool

boning up on Fans might be helpful as well
its not just the airflow that is important, but the static pressure
a supply with a greater resistance will benefit from a higher static pressure



Air Flow Vs. Pressure Characteristics
Parallel & Series Operation
Stall of Axial Flow Fans

Basic Fan Laws
How to measure Airflow vs Pressure
How to Achieve Low Noise
Accoustic Noise
EMI
Introduction: Forced Convection Cooling
How to select the right fan or blower
Step 1: The Total Cooling Requirements
Step 2: Total System Resistance / System Characteristic Curve
Step 3: System Operating Point

here are three basic components to any thermal solution:
the heat transfer interface
(in this case the heatsinks total area)

the heat transfer medium's volume\pressure to time
(in this case airflow, with additional variables like airflow resistance)

and the temperature differential
(where using cooler air has a huge impact)
while the efficiency of a given supply at a given load will determine the high side of a temperature differential, by employing a cooler airstream you have the option to lower the lower side of the differential

a real good computer oriented guide to thermodynamics
General Heat Transfer Guide

Heat transfer is all about temperature differentials. Conduction through materials and convection away from surfaces is proportional to the temperature differential that exists. Basically the thermal resistance of a given solution changes with the temperature differential, which is why thermal solutions are rated in °C/W which leaves that variable ambient open but gives you a formula

The heat transfer through the wall follows a simple equation:
Q=k/L(T1-T2)

We can draw some interesting conclusions from this equation. First, heat transfer is proportional to the temperature difference on the object. If the temperature differential doubles, the heat transferred doubles. Second, the conduction coefficient "k" is proportional to heat transfer. If the conduction coefficient doubles, the heat transfer doubles Alternatively, for the same differential temperature, twice as much heat may be transferred. The final observation is "L". As thickness increases, heat transfer decreases. Alternatively, to maintain the same heat transfer through a material twice as thick requires twice the temperature differential.

while that addresses conductance, there is a corrallary for convection (transfer from the heatsink to a fluid (air\water\ect)
its known as Newton's Law of Cooling

your pretty much stuck with the heatsink a supply has, but altering the airflow isnt the only tool left to you
a few simple mods and that supply will increase its range, edit > or you can use less airflow and thus quiter fans for the same capacity\temperature
the main objective being not to draw the hot air from the heatsink through the supply and getting it its own cool airstream, you do of course still need to exhaust the case


I would mention that its only under truely adverse conmditions youd likely have a problem, and that, any alterations you do to the case could be likely used for the next upgrade and are infrastructure investments of both time and $$$
-------------------------------------------------------------------------------------------------------------------------------