I’m interested in adding liquid cooling to my computer(after installing a professional package in my brothers 3 years ago).
This time, however, I want to do away with all the fans, and that means a large radiator system. I have several ideas including a spill tower constructed from some 4 inch ABS pipe with 1/2 inch copper pipe sections penetrating through the ABS to serve as heat sinks and baffles to divide up the spill.
Something like that if that makes sense, but taller. More pipes. Tighter grouping. 16 inches tall at least.
In a pinch I will just purchase a automotive heater core, but then I will need to add a fan. I dont want to do this. Nor do I really want to use a professional product.
I am making use of a thermaltake pump for the project, which is a professional pump designed for the application. I am thinking about using distilled water rather than real radiator fluid. I will probably have around 1 litre of coolant.
So comes the question: Can I possibly jam enough surface area in that? How can I calculate? Would 1/4 inch copper line be better?
I dont need an answer as to whether I am crazy or not. That calculation is easy.
A bean machine? url was redirected to that, fwiw. It might be going right over my head, though…
I, too, am having to look at non-OEM cooling with my desktop. I like what you are doing, but have little to offer on this experience-wise. Reducing fan noise (and the cleaning of 'em!) is a strong selling point for me, and this sure is heading in that direction. I’d love to see some pics if/when it all comes together
If I understand your drawing, you are making a complicated geometry for the water, but externally to the device it’s just a cylindrical surface. THAT’s the surface you should care about, the one that air and radiation will have to carry away your heat from.
You can figure that a surface in air without any fan will convectively dissipate about 5 to 10 watts per square meter for every Celsius degree the surface temperature is elevated above the ambient air temperature. If you make a chimney for it, to use its hot air to draw more cool air, this will increase. You can look up the Stefan-Boltzmann law to find out how much radiative dissipation you will have.
Ionizer: exactly! I hate dealing with noise and dirty fans. Of course I will share pictures.
wbeaty: Thanks for that, I didnt realize that about copper. You are correct about chimney effect.
Napier: Yes, a complicated geometry. What you say about the external area is important(I suppose i could use some metal heating duct instead). The plastic wont shed heat well I think. The copper is to go through the walls, but as wbeaty said, there is no chimney effect.
The numbers are exactly what i wanted. I can simply compare that to the wattage rating for actual CPU radiators.
So I have to perhaps change the orientation of my device so that the through holes are oriented vertically, Then the accumulated heat will flow better.
I went and bought myself an automotive heater core. Its the device that transfers the heat of the engine into the passenger compartment. It cost me 45 dollars. Basically it works the same as a room radiator.
It measures about 8 inches tall, 6 wide and 2 deep. In centimeters that is 20x15x5. I havent figured out what the surface area is in regards to shedding heat, but its easily 100% bigger than the commercial unit I installed in my brothers computer.
My pump is orders of magnitude less powerful than an automotive one, so I am a little concerned about having good flow through this thing. The commercial computer radiators are more like tubes running through heat sinks. This thing is a plenum style device.
On the other hand, that means huge surface area in comparison to the commercial ones, and a larger liquid dwell time inside the radiator. It also means I probably dont need a reservoir at all. I am hopeful that I wont need a fan. If I do, it can be large and slow moving.
Next I have to acquire a water block to attach to my CPU. NewEgg and tigerdirect seem perpetually out of stock for what I need, so I might try for a local solution. I know a machinist that might be agreeable to milling one from a copper block.
That is sorta what I was thinking of myself when you started this thread. It would seem, though, that there will be a need to have air moving through that ‘core’ just a tiny bit or the heat might not move away well enough (?).
Curious ? -> If you gently blow against one side of that core, is airflow detectable on other side? Or is there a somewhat restricted airflow that will likely not allow much warm air to ‘escape’ on its own (following me here??). When those cores were designed/produced, it was known that there would be a decent-cfm fan to force air to move through it rather than designing a very airflow-efficient design to the core itself. Easier/cheaper to make and produce a simpler core and use a bigger fan to make it work acceptably, so to speak.
It sucks that blocks are not there to be ordered. I would be leery of a shop making something unless they knew exactly what they were doing and realized the purpose for which they would be constructing. I know little about the hands-on with making such things, but it really seems that a cooling block is out of the experience of the average machine shop (ime). Hopefully, your ‘friend’ could work the requirements into a block properly without the need to test for efficiency/performance upon your computer, etc…
The auto mini-radiator is a good idea. If you want, make it the floor of a tall cardboard box without a bottom or top, so it heats the air entering the bottom of the box, which rises through the box. This will make a chimney. The beauty of chimneys is that all their height contributes to the driving force for air to rise through them.
A small pump will probably move enough water through this core so that the water everywhere in your loop is within a small range of temperature, and most of the temperature difference throughout the system is between your radiator and the ambient air. When this is the case, further increasing liquid flow will contribute little additional performance - that is, your pump is about as good as you can use. I think the most expensive element of performance is getting the core to dissipate heat to the air effectively, so if any other link in your chain is the weak one, you’re allocating your investment poorly.
I think a machine shop could make your block well enough. Use a copper alloy like 101 or 102; these common alloys are actually quite pure copper and have the maximum thermal conductivity (as well as electrical conductivity, for which they are also often chosen). If noplace else is more convenient, you can get it from Mcmaster-Carr Supply C0. (mcmaster.com). It is a little sticky to drill, unfortunately.
A shop making you a copper block will probably turn out pretty well. You want the surface that mates with whatever it is cooling to be nice and flat and smooth. Buy a “thermal compound” (thermally conductive paste) to mate it to your chip, and spring load it somehow. I have made “cooling plate” blocks by drilling two parallel large holes the hard way through the block (which has something like 4:1 aspect ratio) so that they are just inside two opposite edges. Pipe tap these holes at one end and make them blind so they nearly penetrate. These will let you put hose barbs in to plumb them. Drill the holes from the same face if you want your tubes to come out next to each other, or on opposite faces if you want your tubes to exit in opposite directions. Then, drill a bunch of smaller mutually parallel holes in the other hard way, at right angles to the large holes, so they intersect. These too should be blind, and at the start of each one, you need to tap and plug the hole (probably with a little cap screw because pipe threads are way too big). You must size all your holes so that the total cross sectional area of all the smaller holes is somewhat less than the cross sectional area of one of the larger holes, like maybe half or less. This will cause the velocity through the small holes to be nearly the same everywhere. The larger the number of the small holes, the better your block will work, though your machinist will point out that drilling holes in copper with a large length-to-diameter ratio is difficult. Without more calculation than I feel like doing, I can’t tell you if substituting aluminum or brass will hurt performance much (they’re 3x lesws conductive). When you are done making this block, the passageways inside will look like the ones in a (large) car radiator.
I just tried. Yup. Its very diffuse though. But in any case, I can see light through the rad(the gaps are about 0.125 inches/3mm).
Right. That was a prime concern of mine. I will be using a fan matched to the side. On my brothers radiator they matched a standard 4 inch case fan to it.
It sucks and its fun all at once. I like a journey much more than a destination.
The machinist surfaces motor components, and we had him craft an item to mate to my SLR camera lens, so hes definitely up to par. His shop was spotless but he apologised for the mess. If he’ll take the job, he’ll do a great job.
Right. And I actually feel pretty safe about home brewing it because the delta of temperature change is going to be low enough that I can kill the power in a panic situation.
One of the interesting problems will be figuring out where the inefficient lies. I am pretty sure I want to use regular distilled water, and unlike most liquid cooled computers, I will have the water shielded from any light, so less grows in it.
Thanks for this information. I knew that copper is tricky to machine, but I wasnt sure what alloy would be best.
Wow, lots of information there. I’ll have to draw it out to get an idea of what you mean. Thanks.
You might try printing this whole thread out to show a machinist. Also, machinists often do odd jobs out of their garages or basements, or on the side. Informal side-job machining of this sort is often less expensive, though you’d have to be more flexible about the timing. The quality, though, is likely perfectly good. In fact, you’re likely to get a better-than-average machinist this way. They are the ones who like doing it so much that they find ways of doing more of it.
Why? Since you will have metal components, you will have a corrosion issue. Although pure distilled water has a neutral pH to begin with (7.0), it will quickly become acidic. Automotive antifreeze compositions are specifically designed to prevent corrosion in, among other things, the heater cores of automobiles, so someone else has even done all the work of figuring out exactly what buffers and inhibitors to use, just mix it per the directions and go. Be sure to use a long life antifreeze and use distilled water to dilute it.