Custom loop

G-sus

New member
I have access to a CNC machine and materials, so I wanted to experiment a little. I can't really afford a proper watercooling setup, so I'll have some custom water blocks made and see how they perform. I'm not looking for massive overclocks, maybe just a little, but my primary goal is to make my rig silent and keep the temps down. I've been searching the web for some time now, researching all sorts of stuff, but I still can't find answers to some of my questions, so maybe someone with more experience can help.

The CPU water block would be pretty much straight forward, a square piece of either copper or aluminium, with channels for water and an acrylic top, which would also have the mounts and barb fittings. Obviously copper would be better, but by how much?

With the GPU it's a little more complicated. I have 2 XFX 4890s, I've taken the stock cooler off of the card before and noticed one thing, the only part the cooler is touching is the GPU itself, there aren't even any passive heatsinks on the VRMs or any other components, so the question here would be, do I make a wb for the GPU only and add passive heatsinks on all the other components, or try designing a full cover block? Just to be sure what I'm talking about, this is what my card and stock cooler looks like: http://images.ncix.com/forumimages/3827BDA7-27C6-4AFE-AD35971A9680ADE6.jpg

Since this is only an experiment, I will be using an automotive radiator, either a full blown rad on the side panel, or a heater core. Would the full radiator of about 470mmx430mm be enough to cool my phenom 965 and 2 4890s passively and would I even see any benefits of a full radiator vs a heater core? Also I now know that mixing copper with aluminium in a loop is a no no, at least that's what I read, so how much truth is in it? I've seen plenty of people running copper blocks with aluminium radiators, but I just want to be sure it won't clog up my system in a month. And I've read that running antifreeze is also bad, why?

All in all, this is an experiment, not a professional loop, but if it works, I will keep it permanently, maybe redo my entire case to fit the theme :) Give me some opinions on this.
 
i too am a CNC machinist and programmer. the ally would make for good prototyping
before the final out of copper. the mixing of metals can set off a chemical reaction
due to the coolant and crud killers. you can use an inhibitor like formulated glycol
for automotive (constant metal mixing) and should be monitored for reaction.

the channel pass is the delicate balance to achive. efficient thermal transfer without
restriction. and that all can be tested outside the case/system before use.

i'd made a tool "T" shaped to channel in a CPU block (for more thermal surface
area) and ended up "buying" a production piece as the size was .005" and no
where for the chips to discharge and blew-up the tool. might revive it, but have
other things to do...

would like to see a project log something if you do proceed.

airdeano
 
Solidworks would help save you alot of time. I use it quite often and I am designing some new stuff for my computer as well but to do a waterblock etc will take time and money to get right.
 
I'm not sure how Autocad is with generating 3D coordinates.

Typically I use Solidworks, design it and it generates the X,Y,Z axis. Then you put that with MasterCam to setup the tool paths and you're done basically.

I have only seen Autocad work in 2D. I just messed with it(I have Autocad 2013) and it does indeed do 3D but the interface is alittle clunky.
 
It does 3d very well actually, you can first make a 2d base and then expand it up, I've found it easier than making it in 3d from the start. If the price of materials won't be too high, I might make 2 blocks of each, aluminium and copper and compare them.
 
Back
Top