Overclock.net › Forums › Cooling › Cooling Experiments › Subambient full submersion phase change cooled pc
New Posts  All Forums:Forum Nav:

Subambient full submersion phase change cooled pc - Page 33

post #321 of 333
Thread Starter 
Quote:
Originally Posted by ivanlabrie View Post

....it's amazing that you came up with something like this and eventually got it to work biggrin.gif +rep

All power to the inventors of the world.....the unseen army of little men working in their sheds....lol
post #322 of 333
Good read. Sorry that the temp delta turned out so high on such a design. Does anyone know what the temperature delta is on Ln2 runs? You would think they would be similar if there is a limitation within the CPU and not with the cooling itself as speculated.
Zybane
(20 items)
 
  
CPUMotherboardGraphicsRAM
8700K @ 5.23 GHz ASUS ROG Strix Z370-I Titan Xp - 2038 MHz G.Skill TridentZ 16GB 3514 MHz CL14 1T 
Hard DriveOSMonitorKeyboard
Samsung 960 Pro - 1TB Windows 10 Pro Asus PG27VQ 27" Curved 165Hz G-Sync Razer BlackWidow TE Stealth Chroma 
PowerMouseAudioAudio
Corsair AX1500i 1500W Titanium Logitech G403 Denon AVR-X7200WA Elac BS 312 Speakers 
AudioAudioAudioOther
Sennheiser HD-700 Headphones Velodyne MiniVee Subwoofer Buttkicker LFE Razer Orbweaver Stealth Chroma 
  hide details  
Reply
Zybane
(20 items)
 
  
CPUMotherboardGraphicsRAM
8700K @ 5.23 GHz ASUS ROG Strix Z370-I Titan Xp - 2038 MHz G.Skill TridentZ 16GB 3514 MHz CL14 1T 
Hard DriveOSMonitorKeyboard
Samsung 960 Pro - 1TB Windows 10 Pro Asus PG27VQ 27" Curved 165Hz G-Sync Razer BlackWidow TE Stealth Chroma 
PowerMouseAudioAudio
Corsair AX1500i 1500W Titanium Logitech G403 Denon AVR-X7200WA Elac BS 312 Speakers 
AudioAudioAudioOther
Sennheiser HD-700 Headphones Velodyne MiniVee Subwoofer Buttkicker LFE Razer Orbweaver Stealth Chroma 
  hide details  
Reply
post #323 of 333
Quote:
Originally Posted by CallsignVega View Post

Good read. Sorry that the temp delta turned out so high on such a design. Does anyone know what the temperature delta is on Ln2 runs? You would think they would be similar if there is a limitation within the CPU and not with the cooling itself as speculated.

Liquid Nitrogen boils at 77Kelvin, about -195 Celsius so it boils about 100celcius lower than his build. Since his chamber is under negative pressure his butane boils at a higher temp.

Techno, you should consider trying to do a direct die type application and see what kind of numbers you get. Maybe the heatsink is the weak link. If it was direct die the CPU boils the butane, and that boils off. If you use a heatsink, the cpu heats the heatsink, and that boils the butane. Since you are relying solely on evaporation you might notice better results with direct die. It's worth a shot, since you already plan on taking the heatsink off to reseat it.
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
post #324 of 333
Thread Starter 
I would like to see what kind of silicon temps an LN2 system would produce running IBT on max settings at 5.5Ghz.....if the limitation is internal then they should like wise have a large heat delta...but LN2 users don't generally do that sort of thing they just go for max OC frequency and run a non intensive benchmark. But it would be interesting as it would test out the internal limitation theory.

Another possibility is the TIM ...I've disassembled and found it was washing away despite my plans that it shouldn't.

Zythe I'm not actually pursuing the project further atm as it's too risky......but to use direct cooling of the die wouldn't work as the heat density of the surface area is too great...thats why you have to use a heat sink....above 6-10 watts/cm^2 the bubbles generated start to insulate the component....the only way to get direct die cooling would be to have a forced flow over the die surface.....perhaps by means of a pump to actively drive off the bubbles.
post #325 of 333
Quote:
Originally Posted by technogiant View Post


*snip*

Zythe I'm not actually pursuing the project further atm as it's too risky......but to use direct cooling of the die wouldn't work as the heat density of the surface area is too great...thats why you have to use a heat sink....above 6-10 watts/cm^2 the bubbles generated start to insulate the component....the only way to get direct die cooling would be to have a forced flow over the die surface.....perhaps by means of a pump to actively drive off the bubbles.

It's unfortunate to hear you aren't going to be continuing this, or at least not in the near future.

If heat density is this prevalent of an issue issue, perhaps a you could decrease the heat density by using a peliter. Since they actively move heat, a coldplate and a nice big 62mm peliter, or perhaps several cascaded smaller peltiers would increase the surface area, while allowing active thermal transfer, simultaneously keeping the reliance on evaporation. I wonder if they would still operate properly at such low temperatures though.
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
The Laboratory
(11 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX-8350 Vishera 8-Core 4.0GHz Asus Crosshair V Formula-Z Radeon R9 290X Generic 1333MHz R.A.M. 
Hard DriveOptical DriveCoolingOS
Wester Digital 5400RPM HP CH20L "The Russian Winter" Windows 8 
KeyboardPowerCase
Steel Series Merc Fatal1ty 1000Watt PSU DIY Rig 
  hide details  
Reply
post #326 of 333
Thread Starter 
Yes I'd be really interested to know if heat density is the issue here.....if it is then I think there would be little that could be done....very high overclocks at high load would have a large inherent heat delta.....but this may just have been an issue with the TIM...just don't know...but I am aware of another extreme chilled water cooled build that has similar large temps deltas on the cpu but not the gpus (which have lower heat densities) so there may be a nugget of truth in this discussion even if it is not the whole story.
post #327 of 333
Very informative read, enjoyed reading this thread from start to finish!
Too bad you can't make a running 24/7 system out of it but looking at the (relatively) high load temps I can totally understand that it doesn't justify the risk or cost.

About the heat transfer limitations of the CPU I think you have very much overestimated the thickness of the solder. Although if the IHS isn't pressurized when soldered the capillarity of the solder might try to raise the IHS, making the 'gap' close to or even greater than 1 mm.
You could possibly ask vaulgar (from the delidded thread) if he could run a prime95 run on a CPU that's pulling around 200 watts on LN2. I would be very much interested in his findings, and I doubt I'm alone in this biggrin.gif
post #328 of 333
Thread Starter 
Hi Jetskyer,....that's some dedication right there reading the entire thread...lol

I think part of the problem with the large temp delta was the cpu TIM that I used......on disassembling it I found that the outer regions of it had gone quite liquidy so that didn't work as I hoped....a better idea would be to use a non gallium metal alloy that would solid at the temps required.

But I also noted that even on the gpu's which had solid heatsinks attached with thermal epoxy that although the temps were a big improvement on my air cooled chamber....about 40c lower the temp delta was still quite large about +10c while surrounded by -30c liquid butane. I really was hoping for better temp deltas than that and can only think that even with the larger gpu die area and so lesser heat densities that some fundamental limits are being reached either within the silicon or its interface when you put so much power through them.

So in all as you said the risk/cost appears not to make it viable if these deltas are fixed.

I have toyed with the idea of using the system at atmospheric pressures as was my original intention as air sucking in would not be such a concern but I've been banned by the wife from anymore such risky projects....lol.
post #329 of 333
Heh yea, I just had this tap open at all times. When I had to wait for something I just went at it and read another page or two. Finished the last posts here in Norway since it's a very efficient thread (data-wise. only got 35MB's a day here on the cellular network)

But you're right, 40 degree delta for the gpu's is quite high. Although one might expect the thermal adhesive isn't designed to work at temperatures around freezing point but that's just speculation from my side. Other than that I recon the temperature sensors are at the far end of the silicon (the PCB-side) which isn't heatsinked. I would say that the distance through the silicon, the thermal paste and into the heatsink all adds up to too much resistance.
Another thing that crosses my mind with the gpu's is that it's heatsink probably is aluminum and of course optimized for air. The distance the heat has to travel through it most likely is so much that the tips of the aluminum are near the butane temparature, whilst the base is closer to -10 to 0 degrees.
This of course can be disproved by a simple calculation but since it's the holiday I'm not going to do that tongue.gif

I wonder a little what would happen with a delidded ivy but that's not really worth the trouble.
It's cool you've come this far though!
post #330 of 333
Im a little late to the party but i wanted to show a video I saw on youtube that show case a similar build using a different chemical.

https://www.youtube.com/watch?v=EN3m1bJvS4Q
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Cooling Experiments
Overclock.net › Forums › Cooling › Cooling Experiments › Subambient full submersion phase change cooled pc