experimental WC setup

[v8envy]

So I've been mulling the idea of building a high performance WC setup using a car radiator. The idea right now is:

1. replace side panel of case with a car radiator (probably a 90s jeep model, but I'll linger around a junkyard and see if I can find something with a nicer form factor in copper). I'd modify my side panel to mount said radiator to the case.

2. use a 20" square box fan to pull air through the case and radiator. I'm estimating about 200-300 cfm of airflow at low speed, 1000-ish on high. Alternatively I could mount a furnace air filter on the box fan and blow air into the case, cooling components and making the PSU happier.

This will be used to cool a Q6700 or QX9??? and a possible pair of 48XX on a DFI UT-whatever board. My target OC is 4 ghz-ish.

So now, the questions. Have any of you read anyone else attempting this? If not, my initial gut feel is I'd need high flow water blocks for CPU, GPU and chipset combined with high flow pumpage. There would be no reservoir. I'd have the pump (or pumps) take coolant from the bottom of the radiator, possibly split into 3, with coolant returning to the top. Spillover would be managed just like in a car -- by going to an external container.

I'm guestimating something like this should be able to take quite a few thousand watts of heat out of the system, assuming I can find pump(s) to flow 10+ gpm and water blocks designed for that kind of flow. I can further reduce temperatures by replacing the box fan with a room A/C unit.

Comments? Ideas? Pointing and laughing?

 

[Tim]

Honestly, even though I have no experience doing something like that, I'd say go for it. I think we'd all appreciate seeing your hard work and the outcome of a project like this. Besides, you're not doing it for the feasability of doing it, you're doing it for the passion. People who point and laugh are jealous, I say go for it!

As far as the radiator, I've seen a couple of side panel mounted ones in the past here and there. How about getting one just smaller than the panel and building a shroud for it, kind of like an extension of the panel. I think that would be cool, and would definately look better than just a rad popped onto a panel. Maybe if you can find a wide enough case, you can mount the rad flush with the outside of the panel.

 

[aigomorla]

V8 your gona run into a lot of problems

Car radiators unless your talking about heatercore is made of alu.

The waterblocks are usually made of copper, your first problem is mixing metals.

Also 10+ GPM in a h20 system with blocks is asking for it. Your PSI is gonna be off the chart.

Go simple and customary, if not do a custom extreme on a fridge compressor or something like derwen did. You'll have more fun this way.

 

[v8envy]

aigo, just the guy I wanted feedback from.

Car rads may be aluminum (not all though), BUT! They still have to be brass coated or at the very least anodized -- or you'd get the exact same type of corrosion you see in a PC water cooling setup mixing aluminum and brass/copper. Antifreeze is lots more corrosive than a water&methanol mix. Engine blocks have iron and brass (freeze plugs), and as you know heater cores have copper as well.

You have an excellent point re: PSI being insane. That'll bust loose from a fitting sooner or later spraying coolant where coolant is best not sprayed. I think I see why people use multiple pumps instead of one big pump and a splitter. Any idea of what a 'high flow' pump and block is considered? I only need the high flow to make sure enough coolant is circulating through the radiator, that problem can be solved a different way. With a smaller compact car radiator I may even get away with as little as 1-2 GPM.

I'm going to test this setup on my E2180/DS3L, I won't be going extreme right off the bat. Anyone can buy high $ parts and slap them together, this is more of a 'cool it works' or 'damn, it sucked ass' sort of project. I think I can fabricate well enough to make the end results attractive, and it'll have a hell of a lot more 'wow' factor lugged to a lan party than just assembling a kit. Phase change is interesting, but I'm not getting into the whole solving condensation (and worse) problems right off the bat. Water is about the right level of complexity for my fabrication skills. I can always add a peltier or ten to the radiator as a chiller later if I feel the need for sub-ambient cooling. And yes, I know it'll weigh about 80 pounds -- I'm a meaty enough guy to carry that somewhere to show it off, at least a couple of times.

I may also prototype ripping apart my existing Koolance case for the pump and hoses. Hmm...

 

[Gillbot]

I say toss it on a junker system and torcher test it first.

 

[Rubycon]

Why would you need 10gpm in that large of a cooler? You know how much heat that can remove given the number of passes and dwell constant? Easily in the 10kW range! :shocked:

Make it flow low, slow and easy in that setup. Your biggest enemy is going to be air pockets and with that much mass you'll need an oxygen scavenger. You would be much better of sticking with something smaller such as a heater core.

 

[WoodButcher]

HAhahaha (points) hahaha (points again, thinks a minute)
Ya know, I don't think you need high pressure at all, be sure to find a cross flow rad and flush that puppy well. Keep in mind a car operates at less than 15 lbs, I don't develop enough heat to raise the pressure enough to reach the overflow if youstarted a 1/2" down the neck. I say cross flow because the water will take the path of least resistance and you want to use the whole rad not just 2 tubes.
Aigo, your showing your age, (young pup you!) I remember when rads were all copper and brass, shoot! now I'm showing my age,,,,,
V8, I like your idea, like aigo said watch your metals, getting a used alum rad would be tempting fate, anodized, plated, whatever, the chances are good whoever owned the car didn't use the right antifreeze or used a flush. Keep us posted!

 

[aigomorla]

V8 you'll never ever see me recomend mixing metals in any loop. This is one of my biggest peeves in h2o world.

I wont condone you for mixing metals, but i will give you a good stern talking to.

Annodizing does not make the metal into superman, more like super dog with a bone of kryptonite in his mouth.

Additives will slow the process down, but bah, heres what a legend in this hobby has to say about mixing metals.

Cathar and Marci: Both absolute legends in h2o world.

Quote:
Originally Posted by Cathar
...

I wholly agree with Marci's statements. Using aluminium is just a recipe for disaster. Yes, an informed customer will take the steps to ensure that it's not a problem.

Being a bit of a perfectionist though, there is actually a problem with the approach of demanding additives to protect the system. All additives decrease the thermal properties of water, and by the time we're talking about enough additives to protect aluminium in a copper system, we're actually talking about a significant performance reduction.

Aluminium doesn't belong in a copper system for the reasons that Marci listed, and it doesn't belong in any performance system either.

Stew.



Pay attention expecially to that last line. Cathar is the maker of the storm series waterblock, and also little river blocks, the prototype storms.

Cathar was also in the team that developed the PA series radiator. Marci is a rep from Thermochill, makers of the PA.

So yeah, they know what there talking about. :T

 

[dawp]

not all radiators are aluminum. the ones i've seen that are were those that have plastic tanks. i had an aftermarket in my '89 probe that was copper. Think i paid $118 for it about 10 years ago.got tired of replacing the tanks every couple of years. Most older cars, pre 80s, had copper radiators. you'll just have to scout junkyards until you find one that fits your needs.


I just check autozone.com to see what they had and most likely you will need something from a 4-banger as those rads are smaller, like a pinto or vega.

 

[v8envy]

Good stuff. I was planning on using the radiator out of my '08 jeep -- it only has 2000 miles on it, used with factory coolant, still in good shape. I should be able to modify my side panel to contain it. I figure I'll need to upgrade the cooling on the jeep sooner or later anyway (wife demands big knobby tires, which means MOAR POWER to turn em, which means more heat...), might as well get some use out of the discards. Had I known what she really wanted was a Barbie Jeep Adventure I'd have never let her buy that thing for me.

Back on topic: this radiator is aluminum and crossflow -- pretty much every radiator made in the last 10 years is. However, even crossflow radiators need enough velocity in the coolant to work properly. Aluminum ones need less than brass/copper because they typically have wider tubes yet smaller cross-sections, but they still need it. That's why I was looking for high flow -- to make sure plenty of turbulence is there to take care of trapped air and to make sure I'm using the whole thing. Which implies having to use Zalman or other AL water blocks.

IIRC aluminum radiates heat better than copper (but conducts it worse), which makes it ideal for both water blocks and radiator in a high flow system. You bet I'll be using 1/2" tubing (or larger?)

Rubycon, I won't get anything but a tiny fraction of that theoretical 10kw power dissipation with water temperatures only a few degrees above ambient. If I'm wrong and it can dissipate a good 400 watts with water only 5-15C or so above ambient... I may be able to go completely passive, using convection or just the airflow from the PSU fan for cooling. Now that would be interesting.