Cooling Self-contained video card watercooling idea

I was in class today during english (boring) and was dozing off like always and then i started to think about random computer stuff then Ymkfx's, forum member at XFN, video card watercooling idea came into mind. So i started doing sketches of another version of it that could work a little better. So i came home and drew it up on the comp. This is for a 6800 series card, btw.



So basically this unit will be above the card when its placed in the case. It'll be atleast 2in. tall and about the size of the card but extended out more so there will be a big gap between the unit and the mobo so it won't interfere with big heatsinks and ramslots.

Now, most of the parts will have to be custom made like the radiator and possibly pump although i've seen a pump about that size. Reservoir also. Maybe even the fan, depending on how everything will fit and the sizing. The GPU block is basically based off the danger den block they have out now for the 6800 series.

I really have no clue if this will even work lol but just an idea. I'm surprised noone has done anything like this yet for video cards. Or maybe someone has and i just totally missed it =/ I know some companies are planning to make self-contained watercooling for there CPU's (manufacturer to remain anonymous)

I guess the biggest issue in this right now is the sizing and fitting. It's possible to put the unit under the card, taking up atleast 2 PCI slots. Also, cooling wise, i doubt it will be THAT much better than air and current aftermarket cooling parts, but it's a start. If someone would be able to design an awesome radiator at a small size (smaller than 80mm) and still perform well, this might do OK.

What are you're thoughts? Any flaws? Or will this idea 'flow?'

P.S. Here's an un-resized version http://dreality.net/6800watercooling.jpg

 

Would this get attached to the card itself? Because you'll definately need some sort of support bracket if you're planning that.

 

Yea definately. It'll be attached to the card, but that i haven't figured out how yet. If the unit was under the card then pci brackets and a support bracket would do it, but with it being like this then i'm not sure how the unit will be attached. But yea, a support bracket will definately be in place.

 

2 questions...

1. How does the rad get any air? It needs space on either side.
2. Why? I cant see any benefits to having an entire loop just positioned ontop of the GFX card.

 

i see potential in the area of people that can't afford/put together a whole system. Nice design, are you going to make it or will this die in concept stage?
The reservor could be eliminated and you could place a T-line or just do a completely sealed system.

 

Yea, i noticed the res. can be totally optional and can cut down the cost.

I was thinking about trying to put this together, but being only 16 and lacking tools + money, it would be hard.

How the rad would get air. Well, i was thinking the unit will have a few centimeteres of a gap between the card and the unit so you can have a little 'vent' under the rad. I can't think of the right wording right bow since its early in the morning and i have school in a few minutes. But i think some of you know what i mean.

 

Did someone say: "support bracket"?



Anyway, it is an interesting idea but you'd struggle to get a small enough pump. But the idea is not new: lookie :here:...

 

ohh nice rig man

ok i did some thinking and where are you going to get a rad that small?

 

Thanks for the link. I knew the idea had to have been thought of already. The nanocooler one is much smaller and compact.

The radiator for my idea would have to be custom made. I have yet to see a rad smaller than 80mm.

A pump that is about 2x2in is at frozencpu.com. But ofcourse if someone really wanted to use this idea they could custom make a pump and other parts to make the unit smaller.

 

How about this piezoelectric micro-pump?

Not sure the flow rate is all that, but I'm currently wondering how I might use one to watercool my P-IV notebook - which at the moment burns my legs if I surf eBay in my jim-jams for too long.

 

Oh, and another thought - maybe you could combine the radiator and support bracket idea into one, moving it away from the card and letting air flow through it.

 

7 ml/min. eh? I'd measure it out in a cup to get an idea of just how little that is...

 

1cc (1ml) of water heats at a rate of 1°c per watt per second.

At 70w you would get a 600°c dT (10°c per second) of water temp from before/after the waterblock. This is ofcourse assuming there is no radiator/disipation of the heat to the air. This idea would require a very good rad but tbh I really dont think the flowrate could even possibly be near enough.

In terms of specific heat movement abilty, a passive aircooler would probably be able to move a lot more than 7ml worth of water.

With a proper w/c pump it is more like 0.5°c dT. I hope that explains why it would not work atall.