Wild and wacky cooling ideas

[moonunit]

Water is a marvellous substance for moving heat around, but it has this nasty short-circuiting habit. This means we have to cool lumps of metal strapped to our processors.

This means the sides of the cpu, and the bottom are uncooled. This must be changed!

Suppose you put guttering around the socket, and flood this area with a non-conductive liquid (silicone oil perhaps? Suggestions?).
This could drastically increase the amount of heat transferred away from the cpu.

To go one step further into wackiness, what about mounting a large heatsink (cooled by more conventional means) in the coolant to keep it from accumulating heat.

What do you think. Could this work?

[The Overclocker]

there are cases that are mineral oil cooled, koolance have one they they take to most demos, there is not much point in cooling the sides of the cpu as makers will make the cpu thinner and thinner as the making process improves

[Ridenow]

I have seen some water setups where a "box", usually made of plexiglass, was epoxied onto the processor so that the processor is the bottom of the box. Water is circulated through the box as if it was a waterblock. This is called "direct die cooling" and can only be done with processors that do not have bridges on them or steps must be taken to protect the bridges.

[supraway]

I thought about using a waterblock, but instead of running water through it, forcing air through the piping, thus not blowing hot air into the case, and always having cool, dense air coming from outside the case.

I thought about that for about 5 minutes...

[CrystalMethod]

Originally posted by supraway
I thought about using a waterblock, but instead of running water through it, forcing air through the piping, thus not blowing hot air into the case, and always having cool, dense air coming from outside the case.

I thought about that for about 5 minutes...
that's pretty close to an idea that I had. Mine was to use a waterblock and make a presureized system with a tank and compressor. The outlet exausting to the outside of the house, but the end of it restricted, so that the system always has pressure. The effect being simmilar to what you get from the cans of air we use to blow dust out of our systems. The reslease of pressure has a cooling effect. I haven't worked up the guts to try it yet, and I'm shortly going to be trying water cooling.

[moonunit]

Thanks for the input. :)

RE pressure release.

I have visions of sticking an air freshener can on the cpu, and releasing lovely fragrance every time the temperature gets high :).

[ButcherUK]

Originally posted by moonunit

Suppose you put guttering around the socket, and flood this area with a non-conductive liquid (silicone oil perhaps? Suggestions?).
This could drastically increase the amount of heat transferred away from the cpu.

Unlikely to give any improvement, nearly all the heat is transferred out through the top of the core. Cooling the sides and back may lower temps slightly but redirecting the effort into better cooling for the top will yield better results and using oil will lower efficiency of the system significantly offsetting any advantages.

[Diggrr]

I direct die watercool, and haven't found that much of difference.
My radiator is bigger than Big Momma and is made the same way.
I think I may return to bong cooling or go geothermal 'cause that would be the only way I see to get temps down further (short of a peltier.

BTW, not much cooling the sides of the die, there's factory applied epoxy on it which fubars cooling there.

[Maximus Nickus]

Originally posted by Diggrr
I direct die watercool, and haven't found that much of difference.
My radiator is bigger than Big Momma and is made the same way.
I think I may return to bong cooling or go geothermal 'cause that would be the only way I see to get temps down further (short of a peltier.

BTW, not much cooling the sides of the die, there's factory applied epoxy on it which fubars cooling there.

Awesome! HOw do u secure it so there aren't leaks? Where did u get it form?

Whats that stuff on the bridges?

[Diggrr]

It's made out the pvc knob on a 4" drain cap (the screw in type).
I did all the carving with my dremel using a carbide mill bit made by dremel.
It's fastened to the cpu with Marine Goop, left to sit and dry overnight. And it's plain old nail polish on the top of the cpu bits (none on the core), two coats, oven baked at 150F for 3 hours each coat. A (perspex) piece goes on top of it and uses the four bolts around the socket to sandwich it down on the cpu with pressure.
It only cost $6.50 american to make, I had the glue and the Mrs. had the polish. Use non-metalic enamel. I do wish she had green though.

Put clear tape over your pencil marks before you polish them. I lost my overclock.

[Maximus Nickus]

You baked your CPU???????:D

[Diggrr]

AMD specs are 90C max.

I just put the 'oven' into the oven. 150F is below the spec.

[Maximus Nickus]

95C I think.

My Fan fell off for 5mins hit 95C!!! THen it fell of for 5hours!!!!!!!!!! The board (old thing luckily) is dead but Cpu's fine!!!!

[Diggrr]

It sounds like you're lucky there's a little headroom in the spec.

I downloaded AMD's pdf file to get the dimensions for the keep out zone around 462 sockets. It does say 90C. I translated their drawings to make a full size printout on which to model my waterblocks.
I found however, that Abit doesn't exactly follow AMD's spec. They're off by a millimeter or two at different places. Even the holes are .5mm off between each other.

[Maximus Nickus]

I was very lucky!

I mean theres a ragged edge all along the top of the core where I've put a HSF on wrong!!! (Once wasn't on right at all!!! backwards:rolleyes: ) Amazingly it still oc's well!!!

I found metal shavings covering the chip where the core had come off!!!!!!

Not only that i've modded it with a knife :eek: so with my old board (which didn't have multiplier adjustment) I could set the default to 11 and oc to 1200 with 109FSB (Board didn't go any higher!!!)

One hell of a chip, I got it off a guy who works for the MOD though so maybe they get good chips??!!!!:burn:

[Diggrr]

I'd say!

I can't get my 1G t-bird above 1.2G. Stupid strange core codes anyway...AQGA week 4. I've only seen one other in different databases and it was at 1.2 also. That's what I get for shopping local, huh?

[Maximus Nickus]

Mines got AVIA on it, 1999, oh well i'm lucky eh?

Anyways its one hell of a chip as its survived hitting 95C then about 150C!!!!!
Also about half the top of the Core is gone!!!! (About a 10th....) where my WBK38 woundn't come off.....

[ButcherUK]

My axia hit 110C on the in socket probe once when I mismounted the hsf... still works :)

[Maximus Nickus]

There really strong chips aren't they!!!

And who says Intels are better just because they have a little shim with em! There cooler because they don't work hard enough!!!!!!:D

[PsYko420]

Mineral oil isn't a very good idea as it is a very very bad conducter of heat, that should be enough of a reason why it won't work.
And water isn't conductive at all, it's actually a good insulator, you just need absolutely PURE water which can only be acheived for about one second.

For more info pm me.

[Diggrr]

Deionised water is supposed to be a great dielectric. Good enough to use in homemade capacitors.
Just try asking for it at the grocery or hardware stores though....you get a vacant look like the brain sucking aliens have just left them to wander in the isles.:D

[ButcherUK]

try an auto store - it's used for topping up car batteries.

[CrystalMethod]

Originally posted by nick_cw
There really strong chips aren't they!!!

And who says Intels are better just because they have a little shim with em! There cooler because they don't work hard enough!!!!!!:D
I'll pull my HSF, running at full load against someone's AMD, who's wiling to do the same, any day. Point being, Intel's run cooler and more stable at high temps, and don't fry 90% of the time when they overheat, they just shut down, whereas AMD's will fry.

[Diggrr]

Thanks ButcherUK, I tried that and they suggested I get distilled water from the grocery.

[Silversinksam]

You can spend several hours building a SLORB (http://forums.overclockers.ws/vb/showthread.php?s=&threadid=51271) and then even though it comes out perfect you decide to not use it.




:p

[Neco]

"" I'll pull my HSF, running at full load against someone's AMD, who's wiling to do the same, any day. Point being, Intel's run cooler and more stable at high temps, and don't fry 90% of the time when they overheat, they just shut down, whereas AMD's will fry. ""


Well ummm...... you can't run against anything when your CPU shuts down ?? heh...

Besides, comparing a chip that has throttling, to a chpi that doesn't, is just a really really moot point.


Hey my car is faster than yours ! ....ok so what if your doesn't have an engine ??



:beer:

-Neco.

[Maximus Nickus]

Actually AMD Xp's have a in-built diode so if the HSF fell of it would shut down, its only because no motherboard manufacturers haven't supported it yet. (Actually some Phillips oe does.)

Also P4's don't get hot as there a lot less efficient and because Intel can hit up to 3ghz with that design, whereas AMD need to use a smaller Micron Process, so there at there limit. Don't forget AMD's are faster so they use more voltage = more heat!

Just wait till the Hammer is out, then AMD will be ahead with the speed race and with efficency.