Overclock.net › Forums › Cooling › Cooling Experiments › And another Heatpipe + Watercooling idea...
New Posts  All Forums:Forum Nav:

And another Heatpipe + Watercooling idea...

post #1 of 8
Thread Starter 
To explain where the idea is coming from: When including my CPU into the watercooling loop (previously GPUs only), in fact the cooling performance decreased compare to my Scythe Mugen 2. The temperatures on my Phenom II X6 1090T@4GHz while benching are in fact about 3°C higher (>50°C vs ~47 for the Mugen - so it's not terribly bad per se, but certainly not very impressive). That is with basically an overkill array of radiators (60mm thick 360, a 240 and two 120s) installed in an external case sucking in cold air from the bottom of the room, and with all seven 120mm fans set to eardeafening speeds.

So, since I figure that both the airflow and the radiator area of my watercooling loop are vastly superior to the single-fan Mugen, I guess the heatpipes in that puppy might actually be the reason why the conventional air cooler performs better. My guess: since the coolant liquid in them vaporizes and condenses, it absorbs and emits comparatively huge amounts of energy during those phase shifts, and therefore literally soaks the heat away from the source quicker than water, which just heats up by a few degrees but doesn't shift it's phase.

If this is the case, then clearly it would be best to combine the heat transfer speed and vaporization cooling effect of heatpipes with the radiator area of watercooling.

Now, the idea I had for a testbed in the simplest fashion imaginable is this:
- use a stock CPU cooler with heatpipes, such as the boxed AMD Phenom II X6 cooler I have sitting around
- cut through the existing aluminum fins, remove most of them, bend and cut the remaining bits around the pipes in shape for axial watercooling
- use rectangular plastic tubes over each of the the heatpipes in a fashion like this:


- connect the resulting heatpipe coolers and test the cooling performance.

Of course, it's a simple, fun and cheap project, so I'll probably do it anyway just for **** and giggles, but I'd certainly like to know if anybody did this or a very similar thing already (couldn't find it searching for "heatpipes watercooling" - they're all following some other idea, as far as I can tell). Any suggestions and comments on the feasability of this idea are appreciated, especially when they are based on personal experience and/or calculations.

Chris
post #2 of 8
I don't think it would give an increase in cooling, but it would be fun to see tongue.gif

You have to remember that, when you add a CPU to a loop with what I asume is min. 2 gpus then you have a limited cooling ability, when in the same loop. You have to remember that the GPUs (usually) runs quite alot hotter than a CPU, or atleast put out more heat since they do more computing. When you put them together you are lowering the delta Temp between the CPU and the water, and also increasing the temp of the water, meaning the radiators have to get rid of even more heat.
Use the rig builder in the top right corner to make your rig show in your sig, so we can see it smile.gif

If you were to have the CPU in one loop and the GPUs in another, you should get an increase in your CPU temp. Since you would be taking away the heat from the GPU in the CPUs water.

I find it very believeable that the air cooler can cool it better since it is only cooling the CPU, and the waterloop is cooling all your components.
Gaming Rig
(14 items)
 
  
CPUMotherboardGraphicsGraphics
Core i7-3930K ASUS Sabertooth X79 MSI GTX 580 Lightning Extreme Edition MSI GTX 580 Lightning Extreme Edition 
RAMHard DriveCoolingCooling
Corsair Vengeance Corsair SSD Koolance CPU-370 Koolance Asus Sabertooth X79 Waterblocks 
CoolingCoolingCoolingOS
2xBitspower GPU Waterblocks RX480 RS120 Windows 7 Ultimate 
PowerCase
Corsair AX1200W Silverstone TJ11 
  hide details  
Reply
Gaming Rig
(14 items)
 
  
CPUMotherboardGraphicsGraphics
Core i7-3930K ASUS Sabertooth X79 MSI GTX 580 Lightning Extreme Edition MSI GTX 580 Lightning Extreme Edition 
RAMHard DriveCoolingCooling
Corsair Vengeance Corsair SSD Koolance CPU-370 Koolance Asus Sabertooth X79 Waterblocks 
CoolingCoolingCoolingOS
2xBitspower GPU Waterblocks RX480 RS120 Windows 7 Ultimate 
PowerCase
Corsair AX1200W Silverstone TJ11 
  hide details  
Reply
post #3 of 8
Thread Starter 
Hi, thanks for the rig-builder tip - did just that right now. Actually I was planning to do the whole dual-loop thing as you describe it, but I wasn't quite content with the noise level of the pump I got as part of my last abandoned-watercooling-project-off-ebay package; so maybe if I can score another decent pump for €20, then I'll give it a try. Nevertheless, I'm talking GPUs in idle for my quoted CPU temperatures - for those two 6950s I'm running, their energy consumption in idle should come to around 20-30W I suppose, so not exactly overwhelming, compared to what the CPU is consuming at full speed and load (125W TDP quoted for stock frequencies and voltages). I measured the water temperature once at idle with a rather accurate sensor, and it came to pretty much exactly 1°C above ambient with the CPU and the GPUs on idle. So I don't think there is huge room for improvement on that side by getting the GPUs out of the CPU loop, especially since right now the heat from that CPU is getting dissipated through all four radiatorswhile the GPUS are idling - I'd have to split up the radiators between those two systems, or buy even more of them - which can probably never hurt.
But seriously, the problem has to be related to the transfer of heat from the CPU to the water - I've manually ground and polished that CPU cooler of mine before I installed it (since it was rather scratched), and I've installed my usual amount of AC MX4 (same as on the air cooler before). Had a look at the paste application and looking for unevenness on the surface - nope, looks just right, with a nice and evenly thin layer of paste all over - again, just like I remember seeing it on that Mugen. Only other difference I could think of is that right now I'm not using a backplate - which I guess I'll just fashion together right now, as I'm having a stock backplate sitting around and just need to enlarge and thread the screw holes.

Stripping the heatpipes for that other project so far hasn't gone as smooth as I hoped, aluminum rather annoyingly has a tendency to smear and stick to my saw blades, and therefore progress is rather slow (no power saw/dremel available at my current location).

Chris
Edited by decnet - 12/3/12 at 11:25am
post #4 of 8
Thread Starter 
Alright, backplate is installed - and maybe it's just the cooler room temperature (am in Europe, so it's evening right now and snowing outside - not particularly chilling in my room, but probably a bit cooler than usual), or maybe it helped, but at least I'm seeing around 48 °C running my usual CPU test. I prefer to think that's three degrees (Celsius )improvement out of drilling four 4mm holes so that the backplate fits the CPU cooler block screws. Not bad actually.
post #5 of 8
Sounds interesting. Would love to see pics of this and see it working. Sounds like a fun project!
Good Luck!!
SixthElement
(17 items)
 
   
CPUMotherboardGraphicsRAM
i7-4770k Z87-GD65 MSI R7950 TF3 RipjawsX 8G 1600 CL9 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 3 Western Digital Caviar Black LG DVD Multi / Lightscribe H100 
OSMonitorKeyboardPower
Win 7 Ult x64 Samsung 2494SW CM Storm Quickfire Reds Corsair HX750 
CaseMouseMouse PadAudio
550D G700 XTrac Ripper XXL Onkyo TX-SR804 
Audio
Bowers & Wilkins 686 BookShelfs 
CPUMotherboardGraphicsRAM
AMD 1090t 990FXA-UD7 MSI R7950 TF3 TBD 
Hard DriveOSPowerCase
Samsung 20G Win 7 Ultimate Corsair CM650 Fractal Design XL 
  hide details  
Reply
SixthElement
(17 items)
 
   
CPUMotherboardGraphicsRAM
i7-4770k Z87-GD65 MSI R7950 TF3 RipjawsX 8G 1600 CL9 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 3 Western Digital Caviar Black LG DVD Multi / Lightscribe H100 
OSMonitorKeyboardPower
Win 7 Ult x64 Samsung 2494SW CM Storm Quickfire Reds Corsair HX750 
CaseMouseMouse PadAudio
550D G700 XTrac Ripper XXL Onkyo TX-SR804 
Audio
Bowers & Wilkins 686 BookShelfs 
CPUMotherboardGraphicsRAM
AMD 1090t 990FXA-UD7 MSI R7950 TF3 TBD 
Hard DriveOSPowerCase
Samsung 20G Win 7 Ultimate Corsair CM650 Fractal Design XL 
  hide details  
Reply
post #6 of 8
Thread Starter 
First pics of having actually done something in the real world - and no, it's not pretty. Not meant to be anyway. However, if anything I have learned one important lesson already: You can actually solder on heatpipes using a plain 60W iron. Had my doubts for obvious reasons, but at any rate, that allows me to attach cooling fins to the pipes, and not worry about leaving some remains of the original cooling fins in place (it all started to look rather unappetizing fairly soon).

I figure I'll use this as a testbed or proof of concept to see how it performs against an identical (actually, slightly larger) AMD stock cooler I have sitting around. If it can beat that, then I figure something like this based on a serious multi-heatpipe design could definitely take those high-end air coolers to battle.


Chris
Edited by decnet - 12/4/12 at 3:28pm
post #7 of 8
@Chris - What really opened my eyes and got me thinking is this - phase change. The ΔT of water or any liquid or solid continues to rise until it hits the melting or vaporization point. When that happens as long as the pressure of the system doesn't increase, ΔT=0 until all the liquid is either gone or the temperature or energy output of the system no longer can sustain the phase change. It's actually how refrigerators and air conditioners work. That would make for a cooler that couldn't get hotter - theoretically - it would depend on the cooling medium used, water boils at 100°C but alcohol is much lower as are a slew of other volatile dangerously explosive solvents....sounds like a blast!
post #8 of 8
If your looking for a BLAST keep an eye on my post here:-

http://www.overclock.net/t/1346823/subambient-2-phase-submersion-cooled-pc

I'm going to be submerging the whole mobo in liquified refrigerant hfc 227ea boiling point of -17c all cooled with a 3.6kw air con unit.....don't think there will be any middle road, it will either be fantastic or go incredibly, spectacularly and perhaps horribly wrong!!
Edited by technogiant - 3/10/13 at 2:28pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Cooling Experiments
Overclock.net › Forums › Cooling › Cooling Experiments › And another Heatpipe + Watercooling idea...