Archived from groups: alt.comp.periphs.mainboard.asus (
More info?)
Paul wrote:
> In article <ul52d.92707$yh.91431@fed1read05>, Frank <fb@nospam.com> wrote:
>
>
>>I've got a P4C800-E Deluxe with a Prescott 2.8 overclocked to 3.4. I
>>have an sp-94 heat sink and need a better fan. My cpu temps are running
>>46/116 idle 54/129 under full load. I'm looking for the best compromise
>>noise to cool fan.
>>Please all give opinions and experience.
>>Thanks,
>>Frank
>
>
> First off, if you had included a room temperature reading, and
> a case temperature reading during the idle and full load tests,
> it would have been possible to determine, roughly, what thermal
> resistance you are getting currently. Case to room temperature
> delta should be 7C or less, according to AMD. Some people have
> too high a case temperature, in which case beefing up the
> CPU HSF doesn't help. The case temp has to be low, for the
> CPU HSF to do its job.
>
> Thermal_resistance in C/W = (CPU_temp - Case_temp)/processor_stated_power
>
> Check the review sites for some curves:
>
> http://www.overclockers.com/articles798/
>
> For raw power, Delta fans get the nod. They make some
> pretty powerful fans, but you'll need ear protection if
> the computer is next to you.
>
> You do need to use a little common sense, though. You can
> use a rheostat to vary the voltage to the fan (or even Q-fan
> if it is working on your board), but if you select a really
> powerful fan, it may not be possible to operate it at a low
> enough voltage, to get the RPMs down. Many fans have an
> operating range of 7-12V, while a few will still run from
> 5-12V. It is better to get a fan that is in the right ballpark
> in terms of CFMs, so you don't need to adjust the RPMs too much.
> (Based on the curves in the above article, I'd probably start
> with a fan at about 1/2 the CFM of that Delta. So, maybe a
> 92mm 34CFM fan and rheostat would allow an adjustment range.
> The author of that article does get a useful range from the
> Delta, but my thinking is, you could select a fan that is a
> bit quieter.)
>
> Go to Digikey.com and try FBA09A12L1BX or FBA09A12H1BX in the
> search box. The BX on the end means they have a tachometer
> output. They are 92mm by 25mm deep fans. They use a sealed
> sleeve bearing, unlike normal sleeve bearings which are open
> to the air. The CFMs are 43 and 57 respectively. The letters
> L,M,H,U stand for low, medium, high, ultra, representing how
> powerful the fan is. This page is how I got the info:
>
>
http://dkc3.digikey.com/PDF/T043/1297.pdf
>
> As I see it, one of the problems with the SP-94 design,
> is the dimensions of the fin area of the heatsink are not
> square. This means that when the fan is mounted on top of the
> heatsink, air spills over the side of the heatsink. While this
> may gave a beneficial effect on the components right next
> to the processor socket, it represents a waste of air flow.
> It means you end up using a much higher CFM fan than would
> otherwise be needed, and that means more noise.
>
> Ideally, you want a plenum (air box) that runs from the fan
> to the heatsink fins, and doesn't allow the air to escape out
> the sides. You may be able to construct something from cardboard,
> for example, but for the plenum to work well, the fan should
> be lifted a short distance from the heatsink. Building a
> framework isn't really as much fun as you'd think (I tried
> it once, using sheet metal, and my result wasn't nearly
> rigid enough).
>
> I don't really see your temps being too out of line, especially
> considering the overclock. If you want better numbers, another
> option is water cooling, and that is especially useful if you
> are having trouble keeping the case temperature under control.
> I've read at least one posting, where somebody "remoted" the
> radiator of their water cooling system in another room, so the
> noise of the pump and fan wasn't next to them.
>
> HTH,
> Paul
Thx's Paul, I'll check it out!
Frank
Facebook
Twitter
Instagram