Solving Intel's Heat Problems..

[schmism]

So I have an electrical related question.   I had never really processed what 400w ment to the CPU.  Until Alex said it pulled 260a.  scuzeme WHAT!    Having built my DIY solar setup last year dealing with high current DC is no joke.  but 260a.   HOW!?!?  In the "normal world" thats well in to the 4/0 wire capacity or over 100 sqmm of cross sectional area to carry that.   how is that even remotely possible on motherboard traces?!?!

[Kilrah]

Well mobo traces ARE big even if you can't see them because they're buried in inner layers, VRMs are as close to the chip as possible for a reason, and some heating is tolerable...

[Quackers101]

On 12/5/2021 at 7:47 PM, J-from-Nucleon said:

Huh. Maybe the next gen won't be a Vapourised lake

it will be liquid cooled CPU, that produces steam. you only have the tap on, to supply it with fresh water for cooling.

[ZapTech7]

wtf????? if we could build a server with this w/ dual 12900Ks we would have the biggest, coolest powerhouse in canada but i wanna try it :|

[raitodemon]

I scratched my head, and I have tried to imagine a vacuum  chamber for the desing of sub zero LTT. With the help of 3D Printing, all is possible. 

Why vacuum ? Because frost become liquid really fast and become vapor. The vapor can be exfilled outside of the PC.

 

I don't know if this is feasible. But I want to share it the "design" I had imagine. Just one side is draw of the  vacuum chamber  (and this is a subjective cut view not at Scale).

 

[Quackers101]

wait you want vacuum around the whole module? not going to happen.

from space, production and how everything would be able to handle that or get it to become in a vacuum to whatever else.

[raitodemon]

Yeah around the whole module, it's a dream of insulation. 

For the production, 3D printing can produce all the parts. Just the vacuum joints will be troublesome.No materials are choosen (it can be plastic, aluminum, or resin). But Hard plastic could take the little vacuum that we need to exfill the moisture.

 

The vacuum chamber is hold in place by the four screws of the module. And the top vacuum chamber can be hold by plastic latches when not woking. (in work the vacuum should seal the top cover)

[raitodemon]

I have another simplier option to draw ^^ i'll come back tomorrow 

[Bitter]

On 12/7/2021 at 1:51 PM, schmism said:

So I have an electrical related question.   I had never really processed what 400w ment to the CPU.  Until Alex said it pulled 260a.  scuzeme WHAT!    Having built my DIY solar setup last year dealing with high current DC is no joke.  but 260a.   HOW!?!?  In the "normal world" thats well in to the 4/0 wire capacity or over 100 sqmm of cross sectional area to carry that.   how is that even remotely possible on motherboard traces?!?!

Depends on the voltages at play too, but yeah that's a lot of power but remember the distance is pretty short! I've had a socket 115x board with a Xeon doing 300W at 1.3V which is 230A, there used to be issues with some boards burning CPU power or ground pins from all the amps, you think board traces are small look at the socket pins and the contact area they make to the CPU!!!

[raitodemon]

Okay, I tried to redisign a vacuum chamber for sub zero. Simpler this time. 

Not a lot of motherboard have free space but the gigabyte aorus x570 have free space on it. If you want to look. 

 

I think it can work, but it will be surely a pain to fit it... 

[Bitter]

I don't think the PCB is vacuum tight, pretty sure you'd be sucking air through the layers of fiberglass and the solder hole vias on the board.

[raitodemon]

3 minutes ago, Bitter said:

I don't think the PCB is vacuum tight, pretty sure you'd be sucking air through the layers of fiberglass and the solder hole vias on the board.

yeah that's why the protection square must be present. 

[Bitter]

6 minutes ago, raitodemon said:

yeah that's why the protection square must be present. 

You've never seen a board with the socket removed, right? There's bare PCB under the socket too.

[Alder Lake]

On 12/5/2021 at 1:47 PM, J-from-Nucleon said:

Huh. Maybe the next gen won't be a Vapourised lake

/s

Raptor Lake? more like Vaptor Lake, more like Vapor Lake

[Alder Lake]

I've also been planning a cryo-cooled PC, maybe I can get some advice from this

[Lightwreather]

2 minutes ago, Alder Lake said:

Raptor Lake? more like Vaptor Lake, more like Vapor Lake

I was think of that, but that can also imply it being not available in stock.........

[Alder Lake]

Just now, J-from-Nucleon said:

I was think of that, but that can also imply it being not available in stock.........

let's just hope that's not how it turns out

[raitodemon]

5 hours ago, Bitter said:

You've never seen a board with the socket removed, right? There's bare PCB under the socket too.

Ah another problem to solve ^^ Maybe a liquid insulator that harden to fill the gaps ? 

[raitodemon]

3 minutes ago, raitodemon said:

Ah another problem to solve ^^ Maybe a liquid insulator that harden to fill the gaps ? 

Maybe insulate the border socket with a liquid insulator and make a joint that envellop the border of the socket and the cpu ?

[raitodemon]

17 minutes ago, raitodemon said:

Maybe insulate the border socket with a liquid insulator and make a joint that envellop the border of the socket and the cpu ?

"Liquid insulator" think obviously like super glue or like a silicon sub-zero resistant. or use a flexible material like rubber 

[Bitter]

Also fiberglass PCB isnt non gas permeable from it's edges, it'll wick air through between the layers it's made from. You'd need to either vacuum chamber the whole board which would result in a lot of parts overheating or build a special PCB just for this approach with some kind of non permeable PCB or special CPU socket. And you can't just -20 the whole board because some components change characteristics or go out of spec enough to cause stability problems with very low temps.

 

Pumping in dry inert gas, submerging in non conductive hydrophobic fluid, or sticking with traditional conformal coating and insulation are the feasible options here sadly. Dry gasses can be problematic because of not only their asphyxiation hazards but also because components are made to be air cooled often passively and different gasses cool differently than air does. Nitrogen is a good choice because it's only about 30% off from what air is and it's pretty cheap, "air cooling" should work similarly enough. Also using a compressed gas into a chamber would let you bleed gasses outside and use the decompression of the gas to add some minor cooling to the whole chamber and board. Alternatively you could recapture, compress, dry, and reuse the gasses in the chamber and just use dried compressed air and avoid the dangers, but also have to run essentially a liquid air plant at Wich point you'd just use the liquid air as a refrigerant probably and then bleed off a little into the chamber to dry it up I guess. 

 

 

It's never as simple as it seems and the simple fixes are often the best ones. Unless we're getting LTT liquid gasses in LTT branded dewars.

[raitodemon]

2 hours ago, Bitter said:

Also fiberglass PCB isnt non gas permeable from it's edges, it'll wick air through between the layers it's made from. You'd need to either vacuum chamber the whole board which would result in a lot of parts overheating or build a special PCB just for this approach with some kind of non permeable PCB or special CPU socket. And you can't just -20 the whole board because some components change characteristics or go out of spec enough to cause stability problems with very low temps.

 

Pumping in dry inert gas, submerging in non conductive hydrophobic fluid, or sticking with traditional conformal coating and insulation are the feasible options here sadly. Dry gasses can be problematic because of not only their asphyxiation hazards but also because components are made to be air cooled often passively and different gasses cool differently than air does. Nitrogen is a good choice because it's only about 30% off from what air is and it's pretty cheap, "air cooling" should work similarly enough. Also using a compressed gas into a chamber would let you bleed gasses outside and use the decompression of the gas to add some minor cooling to the whole chamber and board. Alternatively you could recapture, compress, dry, and reuse the gasses in the chamber and just use dried compressed air and avoid the dangers, but also have to run essentially a liquid air plant at Wich point you'd just use the liquid air as a refrigerant probably and then bleed off a little into the chamber to dry it up I guess. 

 

 

It's never as simple as it seems and the simple fixes are often the best ones. Unless we're getting LTT liquid gasses in LTT branded dewars.

That's sad... But i'm gonna to try anyway to do it, for science !

[Bitter]

2 hours ago, raitodemon said:

That's sad... But i'm gonna to try anyway to do it, for science !

I'd love to see the results and be wrong!

[raitodemon]

9 hours ago, Bitter said:

I'd love to see the results and be wrong!

I wonder... If I manufacture a PCB vacuum tight with only the CPU socket, and an adaptator to the socket of the motherboard does that create latency ? If I can separe the CPU from actual motherboard, i can vacuum it more easily. 

 

[raitodemon]

45 minutes ago, raitodemon said:

I wonder... If I manufacture a PCB vacuum tight with only the CPU socket, and an adaptator to the socket of the motherboard does that create latency ? If I can separe the CPU from actual motherboard, i can vacuum it more easily. 

 

No, it will bust the timings ...