One-way HS compounds?

[maluckey]

Has anyone though of a way to make a one-way HS compound? Is there a way, while staying air-cooled, and non-TEC, that I can keep/slow down heat from returning to the HS once vacated. I'm not talking Space Shuttle tiles, just something easy, and functional.

I'm having a hell of a time with my OC'ed, air cooled setup. As the load drops to the CPU, the temps are all dropping like a stone. Then when the load increases, the temps climb as rapidly. This just can't be good, and in fact, if I raise the CPU temps, or massively lower them I can OC higher than "normal". I normally run this board for daily work at 173 mhz FSB ( Abit KR7A-133, KT 266a ), and at this level, it's all day strong. Any higher, and without the abovementioned procedures, it goes spooky on me at around 180 mhz FSB.

Specs: Motherboard: Abit KR7A-133 (KT-266a)
CPU: XP 2100 @ 13 x 173-180
HS: Alpha PAL8045 with 60 CFM fan
P/S generic (tweaked) 500wat

 

[Mday]

heat conducts in all directions possible. heat basically kinetic energy being transfered molecule to molecule. if you were to stop it in one direction, you stop it in all directions. heat is never "vacated"

 

[syberscott]

Maybe a heat diode?

 

[Mingon]

It sounds like you have a bad contact between your CPU and your heatsink. The t'bred core are quite difficult to cool due to them having a small surface area. Try lightly lapping your heatsink and see if the bottom is flat.

 

[zsouthboy]

What about that company that had that peltier-type thing that allowed heat in only one direction?... i'll go look for a link....

 

[zsouthboy]

Ahhh here it is:

http://www.coolchips.gi/technology/overview.shtml

Where can I buy one ?

 

[Abzstrak]

try varying the fan speed in relation to the temp... simple circuit to construct. You could do with same with the power input to a TEC if you wanted, just alittle more complicated due to power requirements... You'll just have to experiment some with the fan speed (or TEC power) required for certain temps.


That being said, if your air cooled, I cant see how the temps should vary much, unless your heatsink is too small.... maybe just increase the size of it to "buffer" the temp changes

 

[txxxx]

Originally posted by: Abzstrak
try varying the fan speed in relation to the temp... simple circuit to construct. You could do with same with the power input to a TEC if you wanted, just alittle more complicated due to power requirements... You'll just have to experiment some with the fan speed (or TEC power) required for certain temps.


That being said, if your air cooled, I cant see how the temps should vary much, unless your heatsink is too small.... maybe just increase the size of it to "buffer" the temp changes

A Variable Resistor (POT type) will do. Find out how much power your current fan at constant speed consumes, sometimes its on the FAN itself, otherwise you can use R = V/I (V= Voltage, and I= Current, both found on FAN I think or a multimeter can be used).

This way you can get an Idea of what value resistor needed.

Edit : use V= IxR to find voltage you require, nothing too low ofcourse

 

[maluckey]

I've looked into the TEC, but simplicity is for me, and the Gurus out there say that TEC elements like to be either on, or off, regardless of the voltage. It was explained that varying the current often causes worse problems than I already have.

I switched to cheaper HS compound, and strangely, it works more stable. Then, I put on Arctic Silver III and it went back to the some old act (Hmmmmmnnnnn). I was reading that a 24 hour, high temp burn in might help. This seems to be sensible, as burn ins help with many things maechanical and electronic. I burned it in, and slowly raised the CPU temp to 50 C, by applying load, and lowereing fan CPU. The CPU has never seen higher than 46 C. before, so I set the shutdown for 53 C. It seems to be a little better off this morning, but time will tell.

Lastly, I also checked the surface of the HS. It is very smooth, as are all Alphas, so lapping may not come into play here. I did check to see if it was seated correctly, and it is. When I mean wild fluctuations in temps, I mean 3-4 degrees C. in a rapid time frame.

Update: I'm hoping to continue this on the overclocking forum, because it looks like thermal expansion of the HS, and the CPU die are the culprit, and maybe someone has a guidelinefor rule of thumb to shim the retaining springs for more pressure.