Overclock.net › Forums › Cooling › Phase Change › Need help with a phase changing system
New Posts  All Forums:Forum Nav:

Need help with a phase changing system

post #1 of 16
8/30/12 at 8:33am
Thread Starter 
I'm slightly new to phase changing simply on building them. I understand how they work on both a build, and a semi-deep scientific level. I'm a Math / CS major so I can do understand almost everything behind the superheat and etc concepts.

I'm going to be making a phase changer unit soon. The basic ideas of it are going to be direct die on both CPU and GPU, but then trying to find a way to do a phase changer system on the RAM.
The CPU and GPU parts are easy, but the RAM is where I'm having trouble.
The RAM slot are

| | | |
| | | |
| | | |

Thats the basic layout of them. I had an idea of doing the system for the RAM by doing a tubbed exaporator that just cools the air around it. VERY near air as in around 3 -5 MM away at all points maximum. I was going to make it into a box that essentially just fit over it, and leave some room above the motherboard about 5MM of this evaporator unit. Essentially it would work like a box with the bottom cut off that is suspended above the RAM on the motherboard to cool the air to a point that it will cool the RAM. I can calculate the superheat that will be needed but how do I make sure no condensation problems would arrise on this kind of a system while still insuring it does its job?
Reply
Reply
post #2 of 16
8/30/12 at 9:43am
  • NoL
  • Extreme Cooler
Not how I'd do it. Consider this instead.

Build an evaporator as normal that is the size of the footprint of the ram slots on the motherboard. Then mount copper L angle to the evaporator that clamps onto the side of the ram. No matter what you will need to insulate.

There are not serious or even real gains from subzero ram though in most cases, and on your first system just try a single evaporator or you'll be in for a bad time. thumb.gif
Reply
Reply
post #3 of 16
8/30/12 at 11:38am
Thread Starter 
To me I'm just going for an overclocking hell. I've got 8-8-8 stock timing ram. I just built the thing so I'm looking at how to make the cooling system optimal. The other part of this was that I'm just using heavy air cooling for the short term while this is a multi-month project. The RAM is just because I want to squeeze the most out of those damned things. Otherwise I'm thinking of doing one for the CPU and one for the GPU. But I'm build the unit separate and then putting it in the computer after I've let it run for a few days and I can make sure its completely safe. I figured asking the pros on phase changers about how a with ram would work. I understand RAM overclocking is overkill but overkill is just way to fun if its possible has some slight benefit. Also any tips on insulation in general for the entire system? I've spent the entire day researching different thermodynamics equations with my Engineering Physics professor about the properties more in depth, granted he doesn't understand the computer aspects of it.
Reply
Reply
post #4 of 16
8/30/12 at 12:31pm
  • NoL
  • Extreme Cooler
Just use something like liquid electrical tape to coat sensitive components, then add closed cell foam over it. Easy enough.

Don't bite off too much at first, build the CPU system first, get down how to braze, vacuum, test, charge, etc. Then go big.
Reply
Reply
post #5 of 16
8/30/12 at 2:35pm
Thread Starter 
Ok so I assume a just going with the CPU / GPU die evaps would be best and just allowing the natural cooling thats going to occur from that being around it / the fans that are in the case should keep the ram cool enough.

Other than that for the rest of the rig I was thinking of building the phase changer system in which the normal diagram goes something like this:


but I was thinking of trying to make something like this:


The hard part from what I understand would be to make sure a the point of splitting it doesn't cause to much friction and cause it to become gaseous inside the pipes. So I know it would have to be an a smooth split and not a sharp edge that does it and also try to make as dragged out as i could and not an immediate curve in it. That would be annoying to find a splitter for the suction tubing that would be able to do that. Also geting the capilary tubing to do this properly would be rather annoying. I can do the math behind the variables behind it once I figure out whether it would be even design wise possible.
Reply
Reply
post #6 of 16
8/30/12 at 2:53pm
  • NoL
  • Extreme Cooler
It's possible to work with multiple evaporators in just that way, however, turbulence at the capillary entry is not the issue. Balancing the suction lines to the resistance/mass flow of the metering devices is nearly impossible in real life. In calculation you can tweak a line here, and change a length there, but in reality, you'd have to recover your refrigerant, cut the line, rebraze, repressure test, revacuum, recharge, retune, repeat.

So for now? One compressor, one condenser, one filter, one metering device, and one evaporator/suction.

wink.gif
Reply
Reply
post #7 of 16
8/30/12 at 3:40pm
Thread Starter 
That's my fallback plan, I guess my best idea was to turn minimize that was to make both of the tubes in the end an equal length so that way they meet at the splitting point at same point obvously. Then the distance they have to travel to their respective evaps is the same by the shorter using redundant coiling. Then since the suction tubes would have to follow the same principle of equal length they have to travel to get back to the convergence point I would just have them meet up there into a common one and flow back to the compressor. so essentially there is no differance in tubing length that could affect the rate that its going at. Also i would use the same sized evap on both. That meaning that the CPU evap would be slightly big and the GPU evap might be slightly smaller.

Does this just sound like a retard speaking or something that i could run with as a starting point?
Reply
Reply
post #8 of 16
8/30/12 at 4:46pm
  • NoL
  • Extreme Cooler
Start with simple, the difference between hypothetical and actuality is a tough mistress to please.
1) How does the capillary get to the evaporator? Do you wrap it around the suction?

If so, you've added turbulence, good luck matching that on the other one.

2) Are the loads the same?
If so, great, try and match the subcooling.
Oh wait, you can't. Or the loads won't be the same. Now the subcooling is definitely not equal.

3) Same everything?
Don't be so sure, even turbulence in the pipes can cause preference and dead zones.

4) Slight position difference of evap from CPU to GPU?
Gravity! Womp womp womp!

The amount of problems make it impossible to have two matching evaporators. So you will have to have different capillary lengths, and now you are at trial and error. It's a tough life.
That's why in the HVAC industry we use zone controls and thermostatic expansion valves fed by recievers. But you can't do that for small CPU or GPU evaporators. That requires large coils and evaporators that are about the size of your computer.
Reply
Reply
post #9 of 16
8/30/12 at 6:15pm
Thread Starter 
Well damn, you just got a ton of rep good sir.thumb.gif

rolleyes.gif Few random questions then.rolleyes.gif On the idea of sub cooling. When the tubes combine again would it really take that long before the temps where mixed enough to not be or just as a whole bad idea. Also any ideas to get around the gravity problem? potentially calculate out the delta displacement?

Also any solutions around not causing any turbulence?

I understand it will end up as trial and error. But i figured as such i just wanted a lower range of value i would have to do trial and error with
Reply
Reply
post #10 of 16
8/30/12 at 6:59pm
  • NoL
  • Extreme Cooler
The tubes will combine, but while you are normally taught that a given volume has the same pressure and such, it doesn't apply to mixed gas/liquids or when restrictions are present. And they're is quite a bit of restriction here. It's just not something you can calculate, much less per system. Most of this is empirical.

Gravity problem requires new ideas in evaporator development, but other than that, you'll have a bad time. Your video cards for instance, the liquid will sit on the non-GPU-core face by gravity, and they're isn't much you can do about that.

And you can't not cause turbulence. You want turbulence. Otherwise you end up with laminar flow, and that's horrible for heat exchange.
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Phase Change
Overclock.net › Forums › Cooling › Phase Change › Need help with a phase changing system