CU-waterblock in CNC

Hello to you XS :) Long time reader, new poster.

Me and a friend plan on making our own low-budget(very low!) watercooling, getting parts in the cheapest way possible, while still having some performance, and very low noise.

While this is the first time eigher of us try watercooling, i have okay basic-knowledge of it. My friend has access to an CNC machine, so it is pretty obvious where we get our waterblock :) We only plan on cooling the CPU for starters.

Anyhow, here is the idea I had.
http://img382.imageshack.us/img382/5...erblock5et.jpg

Blue is water-in, and red is water-out. Dont mind the weird pins, just me being a bit too lazy there :).

I hope that it is not too restrictive. Maybe we should make the block smaller in heigh, thus increasing the flow a bit(right?)

Since this is a low-budget DIY watercooling, the pump we will use might be a cheap Eheim-1000 moddet to be inline. But i just dont know if the waterblock will require a stronger pump.

We do not need extreme low temperatures, just something in the leauge of high-end air or something, but the noise must be kept at a minimum.

I hope i am not asking very stupid questions or something.. But what do you think of the block-design?

The design looks pretty good. but what is the red dot for?

Quote:

---

Originally Posted by Bail_w

The design looks pretty good. but what is the red dot for?

---

Way to go on actually reading his post.:rolleyes:

With water cooling the heat does not get very far from the core of the CPU. There is no need to double back on yourself in the design, not to mention you might end up creating clearance problems with the tubes.

Quote:

---

Originally Posted by Bail_w

The design looks pretty good. but what is the red dot for?

---

Water-out, did you read the first sentence beneath the picture?

The best desing is... coolwater on the core, and quick exit out of the block

the rad is the exit of the water

Well, second design.. Dont mind the sides, i am just a sketchup-fool.. :) The sides, which are closed without water now is supposted to be solid.
Might start over on the design, but how is it now?. Now the water does not come direct on core, but a little below, but passes over core.

Dont mind the weird angles and all the stuff that looks very.. wrong - i rushed it a bit :) Especially the left side is weird.

http://img371.imageshack.us/img371/1...rblock24xc.jpg

tbh, that design doesnt look that good to me. keep playing with ideas though, they will always get better :)

design* on my post... the thing is... in both youre designs... the water enters the block, then passes over the core, then exits the block..

in the highest performing blocks... water enters AT the core, and exits somewhere immediately. So the water isnt there long enought to get very warm.

Quote:

---

Originally Posted by ReD.SkY

design* on my post... the thing is... in both youre designs... the water enters the block, then passes over the core, then exits the block..

in the highest performing blocks... water enters AT the core, and exits somewhere immediately. So the water isnt there long enought to get very warm.

---

His design is fine for the performance he is trying to reach. Only more recent waterblocks use the "on core" inlet, everything before that (like the Maze 4) use the idea he has drawn up.
With that said, I would remove the inlet columns and extend the rows just a little. The columns won't do anything for where you have them.

looks a bit restrictive for a lower head pump, but that all depends on how many pins are really indeed used....

If anything, increasing the number of fins will increase the surface area and therefore performance. I know there is a limit depending on the CNC machine and the heads you use, but if you can squeeze in more, that will help.

Edit: Which app are you using do do the sketches?

I use Google Sketchup, it's easy and free - even though i still suck at it :)
http://sketchup.google.com/

What is most restrictive - fins or pins?
Right now they are combined, but would i benefit from cutting out one of them, and making it all fins or pins?

About the CNC machine. I think i remember my friend talking about 3mm at the very least, preferable 4-5.

I'll try to make another one tomorrow, one that enters in the centre, and going two ways and meeting on the side. I just dont know if i should focus on pins or fins.

The model i make is in no way final-like, and i have no idea about how many pins we will be able to squeeze in - i'm just drawing the basic block-idea.

btw. thanks for all the help so far. Great to get so many good answers in so little time.

round pins will be the least restrictive...

Quote:

---

Originally Posted by ReD.SkY

round pins will be the least restrictive...

---

But pins are more restrictive than fins, due to turbulance.

At 3mm, if you want the least restriction, don't bother with the pins. You should however, bring the fins to a point, so the water stream splits, instead of hitting the top face.

Edit: What CPU are you cooling?

Quote:

---

Originally Posted by Fuzzy_3D

But pins are more restrictive than fins, due to turbulance.

---

This was tested, and found to be not quite true. I don't have a link to this article, but it was tested by an expert.

At the TS, look for reviews of waterblocks and try to make a block that is close to the best performing blocks on the market today. Look for small fins and pins and most surface area right above the core. Depending on the design of your waterblock 2-5mm base tickness is best.

For your machine, and lack of skill of forming blocks... do not try pins.

Use fins, just take some pointers from the Maze 4, its still an amazing block and easy to duplicate.
Even swiftech can't machine the pin design correctly without breaking some and leaving crap in the block.

Quote:

---

Originally Posted by alexio

This was tested, and found to be not quite true. I don't have a link to this article, but it was tested by an expert.

At the TS, look for reviews of waterblocks and try to make a block that is close to the best performing blocks on the market today. Look for small fins and pins and most surface area right above the core. Depending on the design of your waterblock 2-5mm base tickness is best.

---

I just read an article that pitted 3 GPU blocks againist each other, the maze 4 lost by 2C temp wise, but had the least restriction, the other two blocks were pin designs.

lol... why dont you just copy a MP-05 or something similar?

hehe, I see that I made a minor discussion about pins vs fins :)

I made a new design, in this one it enters right on core, exits at both sides, and then team up again on the other side of the block. I have realised that the CNC machine we will use might not be the best for small pins :)
http://img133.imageshack.us/img133/4...rblock37mi.jpg
So.. How is this? Should i just forget the fins in the end, near the exit? Do they make the block unnessacary restrictive without giving performance?

And again.. Thanks a lot guys - you are really helping here :)

EDIT: We plan on using it on a conroe when it is out. What sizes should the block approximately be?

This thread is right on the money as I'm contemplating some designs in my head right now. Thanks for the application info ;)

Do you have any 1inlet/2outlet block design ideas?

Also, guys, how important is the depth of the block or height of pins/fins if you will?
Which is better lower or higher?
________
INDIAN COOKING

Answers to all these questions are readily available at Pro-Forums.

i think it would be sweet to use that same design but with that extra part on both sides and have a dual output block

Quote:

---

Originally Posted by MarioMaster

i think it would be sweet to use that same design but with that extra part on both sides and have a dual output block

---

I think so too...

Clokker thx for the link:)
________
Montana marijuana dispensary

Basically mirror the outlet to the other side and would probably work pretty well. I suggest taking a look at how cathar designed the white water and using some pointers there. Might want to make the top of those fins lower, by the way.

Here's what I came up with.

http://members.shaw.ca/nick_anti/Block3s.jpghttp://members.shaw.ca/nick_anti/Block1s.jpghttp://members.shaw.ca/nick_anti/Block2s.jpg

Click to enlarge.

Would something like this be allright? About the rounded sides, that is only because i just added something to my previous model :)
http://img468.imageshack.us/img468/9...rblock45tr.jpg

I'm still thinking it might be a little too restrictive, but am I way off on that one?

Fuzzy_3D - Maybe it is just me, but i think one end of that design will have still-water(Pardon my english). I mean, wont the water stand still, and just flow directly from intake to outtake?

About the pump, we are currently thinking about Hydor L20/L30. Since this is a small loop, consisting only of CPU, will they be enough?

As long as the inlet is above the core, I dont see a problem. Though you are right, the inlet should be down about 5mm more.

As for your design, that would be extremely restrictive. I would suggest a layout that does not require the water to change direction within the block.
If you have allot of time on your hands, it would be interesting to try out different designs.

Edit: My version is just a ripoff of the AquaXtreme IceFloe

So basicly, I should go for a White Water replica with two outlets? No funny 90-degree angles, and a lot of fins.

Check this model then :)

http://img432.imageshack.us/img432/7...rblock51op.jpg
http://img222.imageshack.us/img222/8...block5b6ov.jpg

Basicly a rip-off, but with holes so the water gets in the side, too.

I'd say thats your best design yet, but I would have an odd number of fins, leave and un-cut fin in the middle, and bevel the top of all the fins, so you don't have dead faces.

Also, are you taking into account that you are using 3mm heads?

What do you mean when you say bevel all fins?

Deadfaces is a big no-no or what?

Also, i am a little unsure what you mean about the odd number of pins, and uncut fin in the middle..

And no, i havent thought about exact measures yet. I will contact my friend soon to find out what head-sizes we have avalible.

Is it something like this you mean?(very quickly changed - some things look weird).
The fins are lowered, so the water can flow over them, but is that the best way to spread the water between all fins?
http://img68.imageshack.us/img68/457...block5c8ha.jpg

By bevel, Im refering to bringing them to a point at the top, like in my initial design.

Dead faces create turbulance, which is good in high flow loops, but is bad with low flow pumps.

As for the odd number of fins, you have the right idea, but only the middle one should be whole, leave the rest on ether side with a cut to distribute flow to the outer fins.

The head sizes are critical to doing any sort of concept.

I'll try to do another one assuming 3mm heads are used.

I still dont think i understand what you mean about bevel. Do you mean i should raise the fins?
Also, i am still unsure about the uneven pin-thing. Like this new-one?
http://img122.imageshack.us/img122/9...block5d6gh.jpg
(edit: The water intake is ofcourse in the centre, just made a minor mistake in sketchup, and now it is a bit hard to move it without screwing everything up)

I'm sorry i dont quite understand you, but it is a bit hard to understand spoken illustrations when english is not your main language.

I've just spoken to my friend, and he said that 3mm was the least that they would use. So pretty big distance between the fins i suppose :/.

And thanks a lot for the help - It is greatly appriciated.

Here's another rough design.

http://members.shaw.ca/nick_anti/Block4s.jpg
Click to enlarge.

The fins are 2mm wide with 3mm gaps. Note the 2 outer fins are half the height of the inner ones.

you can also see that the fins are pointed instead of flat, I refer to that as a bevel (trimming edges).

Again, that block you see is a complete ripoff.

Ahh, now I understand. The bevels cut away the rough corners, and making a larger surface area.

Who made the original design of that block? I do not think it looks very White-water'ish.

The design you made - isn't it a bit long? It must be ~5-6cm on the short side, and ~8cm on the long side. Is there room for such a large block on the motherboard? How large can the block be in total? I'm thinking ~6cm approximately, but i am unsure. I do not know which motherboard that will be used, but for an allround block?

I think I will play with sketchup a bit, and see what turns up. I just dont have a clue about the dimentions.

Cute.

But did you estimate your costs yet, between the copper and the seal and the mouting hardware? Assuming the machining will be free as well as your time. I can almost gurantee you can buy a block of equal performance to what you will make for less, or one that will perform better for the same amount it will cost you to make yours.

Quote:

---

Originally Posted by epion2985

But did you estimate your costs yet, between the copper and the seal and the mouting hardware? Assuming the machining will be free as well as your time. I can almost gurantee you can buy a block of equal performance to what you will make for less, or one that will perform better for the same amount it will cost you to make yours.

---

I thought you of all people would understand that theres no fun in doing things the easy way. Or the cheapest way for that matter.

Thats why he posted on this particular forum.

Machining is free, and we allready have 10mm plexiglas, so we have to pay for the cobber(not sure if my friend can get this free as well, i think so, but am not sure), some sealing and some mounting screws/bolts. Not so expensive..

Here in denmark, i havent seen s.775 blocks for under ~22£, 42US$. So i am quite sure it will be cheaper to make a few ourselves(we are three that wants blocks..).

Besides - It's a project, and it is fun to play around with, so I really cant see why we shouldn't go for an DIY waterblock.

Alternativly, we could just buy a Thermaltake BigWater :p: :p:

Here's my idea for you, input onto the pins, output next to the pins on any side.
1.5" block with channels @ .03125"

http://img54.imageshack.us/img54/3699/waterblock7ie.jpg

n00b of 1337 - The block would sure be a great performer, but the CNC machinery we have access to is minimum 3mm(0.118 inch). Unfortunately, if we tried to make a pin-block, there would be very few pins :/ So I think we have set our minds on fins.
Anyway, thanks for the suggestion.

I just played around in Sketchup.. Here's the result.
10mm high(buttom not included) 4-5 mm here.
The fins on the side is 3mm, and 2mm bevels(am i using this word in the correct way?)
The next ones are 5mm, also with 2mm bevels, and the last four are 6mm, plus 2mm bevels.
The block measures 60mm*60mm.

http://img61.imageshack.us/img61/854...rblock62vk.jpg
Comments?

Great, but the very center is bare, thats where I would put the odd fin.

Otherwise, you just need to design a cover and mounting bracket.

Water takes the quickest path.
So that wont work.

But also with my design, you want to provide jets, and you dont need alot of pins, the idea is to cause an impingement effect on or around the fins.

NoL, i like yours

n00b 0f l337 - With a minimum of 3mm between pins, that would make a very bad centre.. I think i have to go with fins.

I made a minor change, and i have absolutely no idea if it will have the effect i long for.
http://img135.imageshack.us/img135/3...block6b7ht.jpg

This should make sure that the water is forced to go to the sides, and not just flow directly from inlet to outlets.

You could also take my idea and adjust the cut sizes. Water wont be too forced around by even that design as the water will just flow OVER the fins.

Quote:

---

Originally Posted by n00b 0f l337

But also with my design, you want to provide jets, and you dont need alot of pins, the idea is to cause an impingement effect on or around the fins.

---

Pins and Jets add allot of resctriction to the loop, and with a 3mm head, you're not increasing the surface area either.

As for the last design, you can fix the broblem of water flowing straigt out by adding a fin down the middle, decreasing the height of the fins, and increasing the space between the inlet and the nearest fin (making the block taller).

I do realise the water will flow over the fins, but it will be forced to the sides, and removing quite a lot of heat from around the centre(the way i see it).

About your idea: I know it is great, and all.. But with atleast 3mm space between each pin, you cant have very many pins in the waterblock.

If we assume that each pin is ~1*1mm, and the internal of the block is 4*4cm, we can have 10*10 pins in total. And that is quite small pins, and a very large block-internal, given solely to pins. I'm not doubting your design, we just dont have the tools for it.

Edit: If I have to put a fin down the middle, I'd have to remove the other two middle fins. I just dont know if it's worth is(surfacearea vs middle "spreading-fin").
I do not know what you mean increasing space between inlet and fin. How would that make the block taller?
The way it is now, the fins decend in heigh the more you go to the sides.. Should i lower them even more?

Imagine it like a flashlight. The further from the surface you get, the larger the spot light. Water runs in a similar fashion when entering a lower pressure zone. If you have the nozzle higher above the fins, the stream will spread over more of the fins. Making the fins shorter will allow the stream to spread with less restriction.

Would making the fins twist and turn be a good idea? lol jw. kinda like
this makes me wanna take AutoCad as an elective next year lol
\
/
\
/
\

Anyway for the ends you can't have them touching the side. Though you probably already know that.

Well, from how this is going, it would seem that you've forgotten what the inside of a WW block actually looks like... So here's a reminder:

http://us.st11.yimg.com/us.st.yimg.c..._1900_46710562

The problem with your design, JacobHL, is that, in the most recent revision, very few of the fins will really do much of anything and, in the prior revision, well... really, the same problem applies.

Edit: As a side note, a CSP-MAG (which, iirc, is a little less powerful than the L30) can pump about 1GPM through a WW + BIP + ~5' of 7/16" ID tubing + rotameter (flowmeter) loop...

Quote:

---

Originally Posted by epion2985

Cute.

But did you estimate your costs yet, between the copper and the seal and the mouting hardware? Assuming the machining will be free as well as your time. I can almost gurantee you can buy a block of equal performance to what you will make for less, or one that will perform better for the same amount it will cost you to make yours.

---

when the only expense is copper and sealing hardware, the price rapidly drops towards "less than ten bucks", which counts as signifcantly less than pretty much every other waterblock out tere.
your best bet is a whitewater clone, especially if you borrow the midplate too.

I have thought about the original white water design, but the problem is that the CNC is minimum 3 mm, which would leave an awfull few fins if i were to choose a direct copy og WW.

Perhaps my best bet would be a maze 4 clone? It should be do-able, with relatively few pins.

How is the performancedifference between maze4 and a WW block?

how about you try and copy a WW... but not identical... just make it have as many and as thin as possible fins.

Still, the machinery is not very good at small things, and we only have 3mm heads.
It's just not small enough for WW-clones..

Petra - I'm not really all the way in watercooling jargon yet, is 1 GPM good or bad?

Quote:

---

Originally Posted by JacobHL

Still, the machinery is not very good at small things, and we only have 3mm heads.
It's just not small enough for WW-clones..

---

You should make a WW clone, only with 3mm pins instead of fins. This will make a decent block.

Here is the block in my avatar. Made with 2mm endmill though, but it should give you a rough idea for making your own block.

Water enters the center and leaves through a corner.


http://xs202.xs.to/xs202/06240/Block3dD.jpg

http://xs202.xs.to/xs202/06240/BlockWireD.jpg