Advice on my Active Cooling design...

[karl]

The other day whilst doing nothing important at school, I did a few quick doodles of a new water block design I had in mind, it probably won't work, but nonetheless I did a few quick sketches on my computer when I got home to see what you guys thought...

<A HREF="http://www.btinternet.com/~karl28/design1pic1.jpg" target="_new">http://www.btinternet.com/~karl28/design1pic1.jpg</A>

I haven't had time to add the dimensions onto the drawing, but I hope you can see the general idea. The block is 50mm x 90mm, and should in theory fit most Socket A boards, as it only requires 15mm either side of the processor. The block should be long enough to incorporate two 40mm pelts, which can just be seen.

The reason I designed a block with 2 pelts is because of the increasing amount of heat that Athlon-T processors are giving out, and as a result, most pelts are no longer able to cool Thunderbirds to any significant level, if at all.

Here's a quick pre-rendered view of the water block, in which you should be able to see the 4 cross-drilled channels, and the drilled hole for the thermistor...

<A HREF="http://www.btinternet.com/~karl28/enhancedview1.jpg" target="_new">http://www.btinternet.com/~karl28/enhancedview1.jpg</A>

Time for some quick Math;

Heat Production = CPU Spec Watts * (New speed/Spec speed) * (New Volts/Spec Volts)²
Athlon 1200 = 72 * (1733/1200) * (1.85/1.75)²
Athlon @ 1733MHz = 116 watts

dTload = (1 - (heat load/max cooling power)) * max temp difference
dTload = (1-(116/(172/2))*69 = 46°C

Th = Tamb + (C/W)(Pin + Qc)
Th = 20°C + (0.0.25 C/W)(172 + 172 + 116 Watts)
Th = 20C + 11.5°C= 31.5°C

Tc = Th - dT
Tc = 31.5°C - 46°C = <b><font color=red>-14.5°C</font color=red></b>

Under full load, the maximum heat the processor can dissipate is 116W, and in theory, this limit will never be reached in its full extent, however if it was to be reached, the processor should still be kept at a constant <b><font color=red>-14.5°C</font color=red></b>, as long as there is no heat loss from the cold-plate into the case, which will of course happen to a small degree. I don't know how much heat a processor dissipates at idle, nor how to work it out, but at a guess I expect to see temperatures as low as <b><font color=red>-40°C</font color=red></b>.

So, what do you people think of my design, and it's possible merits and drawback? Any opinions or comments would be greatly appreciated.

Kind regards in advance,
Karl Brown

<A HREF="mailto: karlbrown1@hotmail.com "> karlbrown1@hotmail.com </A>

 

[peteb]

well, with the amount of work you have put in to date - I sure hope it does the trick and that it cools your tbrid like anything.

My only concern - and I cannot follow your math, so I'll just check your logic , have you factored the heat <i>generated</i> by the pelts into your equasions? Prabably I guess.

Only other question is how the heck are you going to dissipate 200-300W of heat back out of your water again? That's a bit more than the standard RAD - which means you are talking 2nd stage pelts again, or a dual stage radiator and industrial pumping.

I say go for it, looks like you got the time, skills and enthusiasm to make it work. If it works and you get a finished product, let me know - I might even take one off your hands!!!!

Pete.

-* This Space For Rent *-
email for application details

 

[killall]

not at all a bad idea... just remember that you will not get full efficiency, you will not get above 50% i dont think if youre just building this thing at home... and anyway by the time youre finished this cooler will be able to do a lot more than just get rid of that amount of heat... your first few attempts wont be too successful, possibly leaks and condensation problems... if you can get it to work... then market it...

you do not strengthen the weak by weakening the strong

 

[Crashman]

How are you planning to stop condensation?

Cast not thine pearls before the swine

 

[karl]

I did calculate the heat generated by the pelts in addition to the processor, which equates to something near 400W.

I have a home-made aluminum waterblock on my computer at the moment, which I am using only for experimentation. My Athlon-T at 1100MHz and 2.5 volts is giving off close to 145W, and this initial setup of mine is able to keep the waterblock at about 5°C above ambient, and the processor die at about 6° above. In light of this, a more efficient setup should allow me to remove near on 400W of heat, with a rise of no more than 10°C, shouldn't it?

I haven't really given the topic of condensation much thought, as my pelt is currently used as a pop drink cooler But I don't see condensation as being a problem, I will simply have to insulate the processor and board in the same way everyone else does. The cold plate is pretty big, and has a much larger surface area than normal, due to its length, I think totally insulating this will help efficiency greatly.

I read the Vapochill article again, and noticed that they use a small heating element on top of the processor, around its edge, which keeps the pins warm, and stops the coldness traveling down into the motherboard, do you think this is a viable option for me to take??

Many thanks for your advice and opinions in advance,
Karl Brown