Cheap Project - Home made? help please....
 

Hey everyone, I'm new to the forums, would just like to say hi.

A few weeks ago I started looking into water cooling, but the price of 150 bucks for 3 heatsinks for my CPU, Chipset, and Videocard was a bit above my budget, I almost won a PolarFLO Pro set on ebay but lost by 2 bucks at 126$, I decided to go and try to make my own Water Blocks if some people could help me out.

Now I am somewhat limited to recources, I can get all the copper and connections I need, but I don't have a fancy CNC machine or anything, I can get things soldered very well with some help from my family, and we can drill most things, but we don't have that many professional metal working tools...

Do any of you know some good designs for Chipset, CPU, and 9800 pro water block designs? I am restricted to Soldering and Drilling pretty much as I mentioned earlier...

Any help would be apprieciated! ;)

The pin grid blocks that #rotor is famous for would be a good place for you to start. Fairly easy to make, good performers, and easily adapted for anything from tiny chipset to giant peltier. Just search for #rotor or '#rotor block'. There's even some templates floating around that will make life even simpler.

Good luck, and post pictures of the results!

You should check out some of the #rotor blocks. They perform very well and can be made easily with a drill press and a dremel if you have those. Basically, they are the easiest waterblocks to make that have good performance characteristics.

The easiest way to make them is to arrange a grid pattern on top of a block and drill out holes of equal size and depth in that grid. Then, using the dremel, you connect the holes in the block to form channels and pins. These blocks can be quite effective at cooling pelts too.

That is my .02

I don't know too much about peltier systems, but I think I'm going to try to do without, since i heard you can easily go below ambient temperature with pelts which can cause condensation, or are pelts safe to use really? Anyway I'm a TOTAL noob when it comes to pelts.

I was looking over some of those Rotor Blocks and a lot of customs on this forum...

1 thing that comes out, wow that guy Jaydee is amazing, anyway he uses a mill for this designs, however this one...

http://forums.procooling.com/vbb/att...achmentid=2421

Do you think you can accurately do this with a drill and a good stand for it?

If so i might try to do that...

Since it seems like it would perform a lot better than just a plain Rotor block... Am I right?

Does anyone know the sizes I need to make the bottom of the water blocks for these components? P4 478 chipset, p4p800 deluxe chipset, and the ati radeon 9800 pro main chip?

I found the mounting dimensions for the 478 chipset, and I can't find it for the Mobo or Video card either.

For the mobo and vid card, in the input and output of the water is it more effective to put them on a 45 degree angle or so compared to the chipset due to water flow, so you get a 90 degree angle instead of 180 degree angle, or does it not change too much and only make the creation process more complex?


One more question, is this jet theory really that effective, I mean there is already that much turbulence in the water it shouldn't matter too much...

just a small correction, or two.... :)


my pin-grid design was thought up, primarily for the purpose of chilling Peltiers effectively. The fact that this design is relatively easy to manufacture, is but only a pleasant after-thought, So too is the fact that it is so versatile in it's application to virtually any conceivable shape and size. The Modular nature of the grid in the design has to take the credit for that.

Here are some ideas i thought of...

As you can see with those with translucent stripes, those are representing holes i could drill to through the side and then plug up on the outsides, easy construction for sure. The other ones pretty much speak for themself.

If any of you pro's could help me out... For the CPU block, which would perform the best. I should be able to construct all of these with the tools I have, but I wouldn't mind putting in extra work for better performance?

So what do you think is best?

PS I won't use a peltier

have you considered converting aircooled heatsinks? A 1U skivved heatsink would make an excellent waterblock and would relatively easy to bodge using a dremel or something like that.

But the only heatsinks I have right now is the stock one that came with the CPU... Otherwise I would have to purchuse another heatsink, perferably copper, while I could at the same time create my own out of one copper block for less money, takes a little more effort and im but I'm okay with that.

Plus i might get anywhere from 3 to 10 copper blocks for free, from a hookup.

Okay never mind the ideas i had before.

The better the contact between the block and the water the better. I might need a heavier stronger pump but it might be worthed. Let me sketch real quick what my idea is, I'll model it in 3d tomorrow. Anyway I was wondering if this would work...

All the thin copper piping/tubing whatever you call it, is soldered together and onto the copper plate on the bottom, so it works like a big heatsink with canals that are all thermally connected and water being forced through. It might kill my pump so I'll need a big one, but do you think this idea might cool 'Well'?


http://forums.procooling.com/vbb/att...achmentid=2533

Instead of drilling all these holes to make a 'cavity' or space in the block, I'll try to solder different copper plates together on the sides, which would be waterproof, and then put 5 to 10 mm acrylic plating on the top, depending what I need and can find.

Solder is a poor thermal conductor, avoid having solder between your water and CPU.

Go with the first design in post #7 - the center inlet, pin-grid spiral.

Silver solder, however, has relatively good thermal conductivity as long as you use hardly any flux.

Links to datasheets?

Exactly, anyway we have plenty of Silver solder left... Anyone know the conductivity of silver 'solder' anyway?

Really quick 3d model, took about 2 min to make, the only thing accurate on this are the mounting holes and the size of the chip 35mm by 35mm (bottom of the block).

I'll design it in greater detail later, also those 2 pipes are the brass fittings that I don't know the sizes from, I should be able to buy some at HomeDepot today and then model them. These DO NOT represent the actual tubing of the system only the fittings.

I'll solder the sides of the water block, 5mm to 7mm thick plates of copper with silver solder then in the middle you see all the little pipes which I'll have to solder together in there too with silver solder. Then on top there is a 5mm thick acrylic plate, which I will close off with an O-ring and some silicone paste. Anyway as you can see this model right now does not have that on.

The other image I included, 'simple.jpg' was the other idea, although this is simpler it has less surface area than my new idea ( 'top.jpg/normal.jpg' ).

There are brazing alloys of 50% silver and 50% copper that would do nicely in terms of heat conductivity. I cannot find any tech sheets on their specific heat conductivity, but alloys combine the features of metals. Both copper and silver are great conductors of heat, so the alloy of the two should perform at least on par with copper.

This is something that might be used:

http://www.lucasmilhaupt.com/assets/...eets/BR559.pdf

I think whatever you try you should go go a centre inlet... into a grid of #roter pins. I have a feeling this woudl be the best performer out of the lot.

Killer noodle: The AgCu alloy mix has been discussed here many times - it is significantly worse performing than Cu

Ewout_vB: IMNSHO the best performing readily buildable block is the Rotor design that so many in this thread have been pointing you at.. Your designi is not going to do well because it has to many thermal interfaces (solder joints) each of which is going to cost you a thermal penalty.

Gooserider

But if i use silver solder and still have all of the copper tubes touching it might decrease a little but it shouldnt be that much would it?

The other way would be to drill smaller than 2mm holes in a copper block, but is that even possible without breaking your drill bit 2 million times?

This is the milling process i used for the Lumpy Channel Block and the Pin Block.

Drilling the holes with a 1/8" drill bit.
http://www.customcomp.us/milling/001.jpg
----------------
Milling the ends for water to merge/spread before entering and exiting through the barbs.
http://www.customcomp.us/milling/002.jpg
http://www.customcomp.us/milling/003.jpg
----------------
milling the O-ring groove.
http://www.customcomp.us/milling/004.jpg

Here is the two blocks that were made from the above process.

Lumpy channel block. (Not Tested Yet)
http://www.customcomp.us/channel/003.jpg
http://www.customcomp.us/channel/004.jpg

And the one shown above. (Tested, works good)
http://www.customcomp.us/pins/001.jpg
http://www.customcomp.us/pins/002.jpg

You can make the holes with a drill press. I used a 1/16" endmill to connect the holes. But a slim dremel blade may work. I also milled out the ends over the inlet and outlet to free up flow a little more. You can just make it pins all the way though. The pin block works better than I expected with only a single inlet and single outlet. I havn't the time to test to other one yet. I might end up sending them to pH in a month or so for real testing.

If you used pure silver to do your soldering, maybe it would work.

But, check your solder. The soft, silver-bearing solders are generally 95% tin/5% silver have a thermal conductivity under 50 W/mK (compared to copper's 390, silver's 420). Even the harder/higher percentage silver solders are poor - I've seen a 95% silver solder at 94 W/mK.

Pure element metals will have a nice regular crystal lattice which conducts heat well. When you introduce another element, even a small percentage, you'll mess up that structure. Exampe: Copper - 390 W/mK, Silicon 150 W/mK, Silicon Bronze (97/3 Cu/Si) 36 W/mK. In general, alloys suck at heat conduction.

If you want the block to look cool, solder those tubes. But, if you want the CPU to be cool....:shrug:

Do you have any designs like this with a blueprint for VGA/chipset ones?

And that is the alternative (just little pilars of vertical copper) if we cannot figure out how to effectively guide the heat to the copper pipes.

For the one you made without the channels, the one i linked earlier, wouldn't a dremel be bad since you want low RPM for cutting through metal and high RPM for soft materials (the dremel is high RPM). Could you do that block with only a drill press?

One thing wouldnt the indentions on the bottom of the pin block work like a golf ball with air and create a cushion/body of water that doesn't get displaced? Or is there enough turbulance? (would you have the blue prints of that design?)

I don't. Maybe #Rotor has some.
#Rotor uses a dremel. I never tried myself. Always had my mill for those little chores. You could do all but the O-ring groove and the ends of the channel with a drill press. You will have to figure a way to connect the holes. You can get a cross sliding vise and an endmill maybe.
It adds water to copper surface area. It is very shollow and with the pressure of the system I think it doesn't hurt anything. Probably will not be measurable. Also if you cut deep enough you can cut through it. My 1/16" endmill would not go any deeper so I couldn't completely mill through the holes. I did however with the Channel block. I am actually expecting the channel block to work "better" than the pin block.

All I have is a ACAD drawing of tool paths. I don't have a detailed drawing as I don't need one to mill it. Just a bunch of 1/8" holes 1/16" apart from each other on my blocks.

When do you think you'll know which is better? I mean the one with channels has more surface area yet at the same time should slow the water down a bit...

If they get sent to pH then I will know. I don't have the equipment to tell within 3C right now.