Cathar, *another* waterblock idea needs insight...

hey Cathar, JFettig, Thorilan, nikhsub1, and Fushy, (oh yeah, and you too NeoMoses)... or anyone else who's got some experience with this stuff....

i need some feedback and hopefully some insight again on another waterblock idea...

and this one is directed at the "t.i.m. goop" issue brought up at the last thread here: http://forum.oc-forums.com/vb/showthread.php?s=&threadid=225813


so we're all losing 11 degrees celcius right?

--------------------------------------------------------------------------

and since water IS the ultimate tim, why don't we just use water right? cause it'll just evaporate right? and direct to die is losing effect surface area also right?

so, combine the cascade and the direct to die but not what most of you are thinking of...

I DO NOT MEAN TO SIMPLY USE A CASCADE TOP ON THE DIE....

but....

but....

redo an entire cascade block that has a small hole going through the entire waterblock that has only the job of "lubricating" the die, as in just feeding it water that acts like a "tim goop" joint...

now seal the entire cpu again but only with seal string or something right around the die, and behold, the tim joint problem is now solved, right?


in other words, take a cascade, drill THROUGH one on the impingement cups, use no "goop" what so ever, and then seal the sides so it doesn't leak water


so, that *should* be the best of both worlds, the efficency of the waterblock world combined with the temp drop that people get from the direct to die world also...

right?

and now, the same questions again....


would that work?

and if it would, how well?


thanks for the time everyone

and also if someone actually gets what i'm trying to say, would they also be so kind as to use their 1337 mspaint skillz and draw up a pic and upload it? cause i still don't get how to upload pics...

sorry, and also thanks...

i understand it but the sealing of the cpu makes it permenant which is what makes it un economical but it could work. it might work best for a pentium with a larger core area .
i thought about how to seal a cpu without being permenant and there is a way but you will need a pliable rubber or type of silicon that is PERFECT in shape that will fir on the top and sides of the cpu very very tightly ( maybe even have to be cast onto it.)

i have an idea that wouldnt cost really all that much money for the sealing of the cpu but thatn it would pretty much be direct die anyways which if dont correctly will work great but so far peoples implimentations havent been great because to do it well requires vast more resources than most of us have.

oh and i forgot you wouldnt need any metal at all as if the ruber/sylicone sleeve can be mounted to a complete polycarb / other material for direct die since it dont require a metal layerfor the heat to transfer through

it would almost work the way a promy looks for its insulation of nearly the whole cpu

ooooh im getting more ideas about how it can be feasably done if we can find a material suitable

Mount the block, then use a syringe and needle to inject high quality silicone between the block and proccessor right up against the core (they make high temp for wood stoves). Let it cure for a few days and it'll be good. This would be permanent untill you didn't want it to be. Just give it a twist. cappillary action would draw the water into tim.

The only problem, as far as I can tell from reading earlier posts on direct-die cooling, is that the CPU core is NOT waterproof. In truly direct-die setups, water will seep into the core of a CPU and destroy it within a matter of months. Not good.

Ken

is the core itself waterproof? I thought i read somewhere that water will eventually seep into the core and destroy the chip? Anyone know anything about this?

not to mention water conducts electricity.. so i think thats a bad idea =d

As I understood it, the core itself is neither electrified on it's surface nor water permeable...

However, the package around the die is both and that, I think, is what Fafnir is trying to overcome.

Water is not the ultimate tim. Water absorbs heat rather well but does not give it up that easily, hence the need for the much larger surface area of a rad to cool it back down. Putting stagnant water between the block and the die might send the block into orbit though.

stagnant water!?!? that would just get hotter and hotter and would eventually fry the chip wouldnt it?

Water is not the ultimate tim. Water absorbs heat rather well but does not give it up that easily, hence the need for the much larger surface area of a rad to cool it back down. Putting stagnant water between the block and the die might send the block into orbit though.

Click to expand...

I think you missed one of his comments. He said he would take the cascade and drill a hole through one of the cups and over the die and that would continually replace the water so there would not be stagnant water. I think it was about a year ago maybe longer there was an article somewhere on TIMs and I think it was the one where they tested toothpaste as a TIM and many prominent TIMs as well like AS2. Anyway they also tested water which gave I think somewhere near the likes of 6-7Cº better than AS2. Though I may have seen that article the same day as I saw the toothpaste article so they may have not have been in the same article.

stated here is that just about everything beats out as3/ceramique/pcm+

http://www.dansdata.com/goop.htm

example given is the toothpaste, it not only beat out the arctic silver, but is also superior because its CHEAP...

and then when dan tried the barbeque sauce that beat arctic silver too,... so we're talking about overcomming the "tim goop" barrier that Cathar has stated here:

http://forum.oc-forums.com/vb/showthread.php?s=&threadid=225813

(somewhere halfway down the thread)

now, if you actually look hard at the dan's article, just plain WATER beats out everything... 0.41°C/W vs. 0.50°C/W for the as3....

but the problem is that IT WOULD DRY UP.... which means it'll dissapear, which is bad....

it doesn't matter if its stagnent or what have ya, but its just gotta be there...

so,... if you redo a waterblock to *keep* supplying the water, won't that overcome the problem with the water drying up?

and also, I know definitely for a fact that both the cpu and its packaging IS WATERPROOF... people only break CPU's cause they proally sprung a leak some where or water vapour migration and it hit one of the resister besides the die or something which IS waterproof, but not electricity proof.... which can be cured or fixed by placing the cpu in the oven for 100 celcius....

and besides, i always clean my cpu with soap and water, and i still don't see how if water seeps in how my barton could still do 2.5ghz....

BBQ sauce and toothepaste? Hotdam you could apply with tongue. What about chip dip, it seems like a more logical choice.

Hmmm, I'd be more interested in focussing on the correct way for making both the CPU die and the waterblock base extremely flat and smooth.

CPU dies are flat, but they're not "smooth". You can even see on AMD CPU's with your eye (if you have good vision) that they are not smooth, with the surface looking like a stack of tiny little pyramids.

The better waterblocks that we buy have bases that are typically flat down to around 1 micron or so, which really isn't flat enough IMO. Not saying that I can do better, I can't with my present tools, but it's something that could be looked at.

Would you pay $25US extra for a waterblock base that was lapped flat down to say 100nm (0.1 microns) and mirror smooth. Would you pay $25US + shipping for someone to do the same job on your CPU die?

If we achieve these levels of flatness then we could possibly consider a number of non-evaporating liquid pastes.

However, this all assumes that there's no thermal flexing going on in either the CPU or the waterblock, which is not true. Both the CPU and block will bend slightly under different temperatures.

Maybe we should all just stick to some really high quality thin state-change TIM pads?

wow, I almost missed this thread, and I even got an invitation. The basic problem I see with the original idea is the same problem direct die coolers see. A CPU is not truly waterproof.

Cathar has a good idea of having your CPU/WB professionally lapped, but how many people would pay to do that? Probably not even 50% of everyone browsing this forum. With such small numbers, it's not gonna become mainstream anytime soon.

Luckily for us, however, many large companies are throwing a LOT of money at the TIM problem. Do you think OEMs are gonna start putting copper heatsinks and/or watercooling systems on computers any time soon? I'll guarantee that you won't find a $500 compaq with a WC system anytime soon.

How about that:
- idea with cascade was nice, but problem was tht CPU isn't waterproof
- so maybe to use silver amalgamat (or however it's called) to link such block and CPU - it wouldn't let water to CPU and wouldn't give problems to heat removal because after silver aamalgamat dries, silver alloy remains which is far better than any thermal paste
- info about silver amalgamants could be taken from dentists as they used it eariler (it's uncommon now because mercury is used to prepare it)