Building a new water block

[Bender]

Here is my Water Orb cooler I have been making today. I still have to do some finishing work but it’s mostly done. The components consist of a 1.75" copper cap, 2 pieces of 3/8" copper tubing and an orb cooler I had sitting around. The first thing I did was to drill tons of shallow holes in the top of the orb were the fan would normally sit. This makes a kind of heat sink affect for the water going over the dimples. Next I drilled 2 3/8" holes in my copper end cap and soldered the pieces into place with a torch. My next step which I haven’t done yet will be to make a cut with a hole saw so the copper cap is counter sunk into the base of the orb. In the picture the cap is just sitting on the orb but once it’s done it will be flush with the top of the orb fins. To secure the copper cap I will use some type of sealant or epoxy. So far it seems to be going very easy so I should have it done by tomorrow. The total cost so far is $1.66; the price of the copper cap at Home Depot. The orb and the copper tubing I found lying around my basement. A copper orb would probably make for a better connection between the water and the processor but for $1.66 I guess I can’t complain.
http://daocplace.com/~bender/watertop.jpg

http://daocplace.com/~bender/waterside.jpg

[cjlax5]

looks nice

[UserName]

A small suggestion.

Cut the cap down so you have less water volume in the cap.
This will cause a higher velocity in the water over the heat sink. Velocity is gooooood, just like in air cooling. It will aslo increse pressure marginally which is also good but very marginally.

You might want to clamp it together to see if it cuts flow down too much as most of our pumps cant handle much back pressure.

[Bender]

I will think about cutting it down but I'm not shure thats the best thing to do. The cap will be counter sunk into the orb but the dimples from the drilling will stick up into the chamber about 3/8 to 1/2".

[UserName]

Then i would cut it down to a little more than the biggest dimple.

Think of it this way. Imagine a piece of ICE 3/8" in diameter and the same length as the distance between the centers of the 3/8" holes, say 1 1/4". (You might want to measure this). Got that in our heads? Now lay the ICE in the heat sink and let it melt. How much would it fill the heat sink? We are imagining this so it is constrained by the diameter of the cap. I think you can picture that it wont go more than 1/4 or even 1/8th of an inch up the sides. In fact you could do the math an tell us how far up the sides it would go.

If you make a space under the cap larger than the volume of the pipe the water must slow down. If you make it smaller it will speed up.

Again here fast is good.

[Bender]

I think I'll try that but first I'll have to get my hole saw. By the way does solder stick well to aluminum? I may just solder the cap to the orb instead of using epoxy.

[UserName]

I don't understand about the hole saw?

I was thinking cutting down the cap with a hack saw or a band saw.

I would put a little RTV on the edges of the cap and clamp it in place with one of the strip and screw setups you see. That way you could take it apart a bunch of times to experiment with your design. Then when you get a c/w of .0001 you can sell the idea. :burn:

[Bender]

The hole saw is a 1.75" wide drill with no center section. I will be using it to cut a hole into the top part of the orb. Once I have made the cut the endcap will be able to fit prfectly inside the hole with the dimples sticking about half an inch inside the cap. I hope this is clear now.

[UserName]

Great Great. Yes of course.
Sorry about that.
Why are you hole sawing the orb insted of cutting down the cap?
Just curious.

[Aesik]

=edited for me having a brain fart while be doped up on drugs after having a root canal=

[Bender]

I still might chop down the cap but it all depends on how far the cap is countersunk. Here is a very crude picture of the thing.

[UserName]

That's a great idea. Should make it easy to seal. Just dont go to far of course. :)

Aesik;

You have the right idea. Now carry it through. If the cap was real tall the speed over the core would be slow. If the cap was just like benders it would be fast. What if it was real narrow? Would it be faster?

You are right that the flow is determined by the tubing. But that is only one of the factors. How much pressure is in the tubes also inportant. And any restrictions also reduce flow.

Have you herd of the Venturi tube or maybe Bernoulli

If you restrict a flow of a fluid, the fluid in the restriction accelerates. The downside is the total flow is reduced. No free lunch. We use this to speed the water over the die. The faster it goes, the cooler the Core.

So you see Flow volume and Flow speed are not purely functions of the tubing diameter.
The upper limit of Flow Volume is determined by the smallest tubing diameter. As flow speed is increased Flow volume reduces, and vice versa, all other things being equal of course.

[Bender]

Well I don't think I'll have a problem with going to far down. The base of the orb is a good 3/4" of solid aluminum. I just hope the hole saw idea works out well. If it does this will be a very tough block. I don't like the idea of my water block falling apart while in use.

[Aesik]

=edited for me having a brain fart while be doped up on drugs after having a root canal=

[UserName]

Yes. So we have some questions about water cooling.
Is velocity important?
Is wetted area important?
Which is more important?
If one is more important than the other should we maximize one and minimize the other?

How should we design water blocks insides?

Tell me what you think and why you think it and we can compare notes.

Or if you think i am being an ass just ask me what i think.

[Aesik]

.

[Bender]

I still havn't formed any solid conclusions about what would be best. If I didn't cut the cap down I would probably get more tubulance in the water. If I can chop it down like you said I may get more efficient water coverage. If there is less water in there there is less warm water sitting around. Any ideas.

[UserName]

Yes bender i have some ideas, but that don't make them good.

I think velocity is king. Nothing else is as good. But the pumps we use dont have any pressure, so we do the best we can.

As i said i would cut it down slowly and see for yourself.

You could even write it up and send it into overclockers. Might even get paid. Might learn something too.

If you cut it too short and you have to build another one $1.66 wont kill you.

Most might think $1.66 and a little time to have some fun and learn some physics a great deal.

[Aesik]

.

[Bender]

The block is finished now so I just have to give it a try. I may put it in my athlon system just to see how it compairs to my Maze2. I decided to not cut the cap down any for my first test. With my hole saw I drilled about 3/8" groove for the cap to sit in. Instead of epoxy I soldered the cap to the base of the orb with my torch. This made for an indestructable joint. Since the bottom of the cap is now 3/8" lower than the top of the divits there is about 5/8" space between the top of the cap and the divits drilled in the aluminum base. The top of the cap is now flush with the top of the orb fins. I'll get a pic of it soon so you can see what I'm talking about.

[Aesik]

Are you going to mount it on the processor using the stock orb clip? Or have you come up with something spiffier (and safer!)?

[Bender]

So far I think I am going to use the stock orb clip. If I wanto to use my peltier I am going to have to think up a new design for the coldplate and mounting.

[xxgg]

I don't know how you have designed the inside of that block but...

How about having a flat piece of copper or whatever and attack that to the cap that you made and have the water go in to #1 and let the water spin to its path (this type of water movement good? no?) and to #2

I just want to give my idea.

[Bender]

Welcome to overclockers.com XXGG.

If I understand you right it sounds like a very good Idea. I may be building another water orb for a friend so mayby I'll try it.

[UserName]

Aesik

You have told me what you think and have stated your theory. I would like to question you on it

You say
“The total amount of heat transfer that occurs in a water block is determined by the temperature differential between the block and the fluid, the total surface area, and the mass flow rate of the fluid (putting material concerns aside).”

So I thought up a little thought experiment.
Imagine a flat plate, one inch square, which we will cover with a plastic water block top. We pump water in one side of the plastic block, over the flat plate, and out the other. For the rest of this conversation lets assume the “temperature differential between the block and the fluid, the total surface area” remain the same. The only thing we are going to change is the third element, the mass flow rate.

Scenario One
Assume we have an inlet and an outlet of 1 inch by 1 foot.
Assume we replace the whole 12 cubic inch volume of the block every hour. That’s not so good is it? That water is moving too slow and getting too hot by the core.

Scenario Two
What if we quarter the volume by changing our inlets and outlets to 1 by 3 inches, but we changed it 3 times an hour? That’s 3/4 the mass flow. How would the cooling compare?

Scenario Three
What if we went even smaller to 1 by 1 but we changed it 6 times an hour? That’s half the mass flow.

Scenario Four
What if we went crazy and made an opening 1-inch by ¼ and changed it 48 times an hour. That’s nearly once a second. That is also half the mass flow.

Which scenario would cool the best?

Your theory, as I understand it, would say Scenario One, Correct?
Do the rest of you believe that Scenario One is the best?

I asked some questions before but you ignored them. Now you tell us you have written an article explaining how every one is mistaken and you have it right. I am very eager to read it. This kind of revolution comes along very rarely.

[Aesik]

.

[UserName]

I find it so funny you do all the work and then you get to the line “Let's continue your purest scenario until we are down to where you have barely a molecules width of a channel but it's flowing and an obnoxiously huge velocity, would that be better in your mind?” and then dismiss it with a “ I don't think so =P”

Why don’t you think so?

[Aesik]

.

[UserName]

I am sorry. To me it was the meat of the post. The rest of the post was based on the assumption that the fullest extention of the idea was ridiculous, with out any reason given. Not one. Not a hint.

In your answer you chose the third one cause the last one would lead you down the path to the thing you reject in the begining. Again no reason.

Now we both agree one molecule would be impractial to do. But the power of the thought experiment is to understand the principals. The principal is a water molocule bumps the copper gets some heat and leaves. The faster that happens the cooler it gets. Only the ones on the bottom touch the copper. Thery are the only ones necessary.

Untill this is understood the foolishness about trubulent water flow won't be straighted out. Yess, that was baiting you.

Cooling copper has to do with how fast cold water molecules hit copper. When the flow is slow trubulant flow means more water hits copper. As the velocity increases turbulent flow is no longer possible as the energy to increase velocity grows exponentialy. In order to increase velocity ans increase cooling laminar flow is the only option and will begin to cool more than turbulent ever could.

Trubulent flow works better because the flow is slow. The real potential for water cooling is in fast smooth flow of small volumes of water.

[ButcherUK]

Um, turbulence is caused by fast flowing water, if you flow slow you won't get any turbulence because your water flow will be too slow and will just slowly ooze over the block in a laminar fashion. I don't see how fast laminar flow is better than fast (or even slightly slower) turbulent flow either - laminar flow means a very slow flow by the copper which inherently means less molecules hitting the copper and hence less heat, a bad thing.

[Aesik]

.

[Aesik]

BTW Bender, sorry about hijacking your thread. If this continues on, perhaps a moderator could move part of it to a new thread.

[UserName]

ButcherUK
read the words around exponentialy

Aesik
I am sure i can teach you nothing on this subject. And three times now you have avoided explaining what is wrong with the natural extention of the thought experiment i have described, so i guess it's not worthy of reply, or you can't do so.

I must run along now and buy some books. Should i start with the ones on how to avoid things i don't want to talk about?

[Aesik]

.

[Aesik]

.