Radical Waterblock Design

[Mick64]

So if the cpu is gonna be on its own, use a injector style. You can improve the injector and get better gains. And its easier to mod an injector as it is to mod an entire plate.

NaeKuh, can you show me some examples?

[NaeKuh]

NaeKuh, can you show me some examples?

i snaped my EK injectors while building sakura. :[

i made it too thin.

But basically u channel drive it more so you get more water pressure on the cooling spots.

The best thing to do is to try to lite the block up and see how the water would flow.

Faster velocity water implacting the plate = better heat dispersion into water.

Since water can only pick up a set amount of molecules, low restriction blocks mostly crap out @ base material.

Impact/injector blocks can squeeze a bit more heat off that plate where low restriction cant.

So its all about channeling your pressure to areas thats important, and having a good area for you to peal that heat off.

I hope this makes sense, im doing 5 things at once right now. :X

[Mick64]

I hope this makes sense, im doing 5 things at once right now. :X

I think I understand, but some pix or links would help...

[fox3]

no...

Need to wait on my X58 board so i can get the side wings messured and redone.

Is this what you mean?

http://www.aqua-computer-systeme.de/...num=1224179250

[nysulli]

naekuh, I wasn't suggesting that short, far to high restriction for my tastes, as I would rather run multiple blocks in a single loop. I would go for something with a bit more height, and without the injectors. yeah yeah, i know injectors provide some benefits, but I also look at the drawbacks.

now if we could have all the pump in the world, I'd be all over something like that, or even if I felt like spending the extra cash and fitting all the added gear in my case for 2 loops.

side note: actually working on a project at work with a 3rd party, who has designed a high pressure fire fighting system, they use something like 3000 psi jets, so I know how effecient the injector nozzles can be, but then again, they also have a specifically designed pump to handle their needs.

we almost have that available as well, the rd-30, but for its cost, I'll stick with my L35 or a liang anyday

[Kayin]

Guys, I'm real sick tonight, but I'll try and add what I've found.

First off, NaeKuh is right on the thermal grade on the fins. Also, on the injectors.

However, I'm still thinking that a combination of cups and pins is best. I'll be machining (if you can call my hackery thus) a prototype of my design out Tuesday, and I think it'll be a lot easier to deal with. Right now, the pressure drop out of my second block is so bad that I can't even cool my NB properly-however, that's a 65w NB as well. My first was just a modded Maze GPU block, and while its pressure drop is quite low, its cooling isn't nearly what my one test with my second achieved (before it popped a seam.)

I wholeheartedly agree with him on single loop for CPU. My testing has shown some amazing things thus far, and it's only getting stranger as I learn to do this myself. Mostly, what I found was that I could lose my LAN with warm water coming off a hot CPU and NB. New blocks later, I'd rather run a loop that runs NF200-SB-NB and have that barely heated water hit it last than run CPU-NB-SB-NF200 and toast my LAN controller or drop a card.

Naekuh, if you're interested in a block, I might can knock you up a model, and if I can get heat under control I'll render it too... Had to drop coin on six new fans today, so I can get some flow in Mithril. With all you give to the forum, it's the least I can do.

[leuler]

i snaped my EK injectors while building sakura. :[

i made it too thin.

But basically u channel drive it more so you get more water pressure on the cooling spots.

The best thing to do is to try to lite the block up and see how the water would flow.

Faster velocity water implacting the plate = better heat dispersion into water.

Since water can only pick up a set amount of molecules, low restriction blocks mostly crap out @ base material.

Impact/injector blocks can squeeze a bit more heat off that plate where low restriction cant.

So its all about channeling your pressure to areas thats important, and having a good area for you to peal that heat off.

I hope this makes sense, im doing 5 things at once right now. :X

There may be a better way of achieving the desired result, though maybe
not with the current group of blocks.

As you noted earlier, in any water block there is a temperature gradient
in the fins ( or pins ). Highest temps at the base of the fins. Same goes
for the fluid, not taking turbulence into account. So yeah, speeding up the
incoming fluid to force more of it to come in close proximity to the base
will help cause more heat flow into the fluid.

What is not appreciated is that the fluid is forced to abruptly change
direction. Work is being done on the fluid. Since it is not an "ideal" fluid
(no resistance to flow), some of the work being done to change the direction
of flow is converted into heat (this is a contributing factor to heat dump).
Furthermore, to take the EK Supreme and the Fusion V2 as examples, part
of the incoming fluid is hitting the tops of the fins/pins, which are flat and
have sharp edges. This creates turbulence where it is not needed and
generates extra heat. I'm not saying that a tremendous amount of heat
is created, but any extraneous heat created in the flow negates the same
amount of heat from moving from the base to the fluid.

It would be better for the tops of fins/pins directly in the path of the incoming fluid
to have rounded edges, or maybe even better, not have any fins/pins between the incoming fluid and the base.
Maybe have the nozzle even closer to the base. While I am at it, I don't
remember how all of the injector plates for the Supreme look like, but
the one pictured in this thread is not optimal. The injector plate should
not create much turbulence. Its purpose is to convert the pressure of
the flow into velocity.

Can waterblocks be improved significantly? Take a look at this article:
http://www.electronics-cooling.com/html/2008_aug_a1.php

[Kayin]

Leuler, I'm working on these things. However, I'm not a machinist by trade, and teaching goes slow.

I already know how to implement what you've stated, but I have to get my halfway crippled hands to make it. In fact, I'm implementing some of them Tuesday.

I just need a CNC sponsorship...

[AndrewZorn]

Drawn In Sketchup then renderderd in

what? Rendered in what?

I wanted to know so bad!

I didn't know SketchUp was compliant and stuff enough to use other rendering programs on it; I guess I never took it very seriously. I should play around with it some next time I feel like going for SolidWorks.

[Mick64]

what? Rendered in what?

I wanted to know so bad!

I didn't know SketchUp was compliant and stuff enough to use other rendering programs on it; I guess I never took it very seriously. I should play around with it some next time I feel like going for SolidWorks.

You will have to purchase SketchUp Pro if you want to export into rendering software...

[Alexontherocks]

Just wanted to add that ironically I had started building something similar a few days ago...here is the link. I think your idea is sound!

http://www.xtremesystems.org/forums/...d.php?t=205072

Work on it because it is not hard and it is good fun!

[bentleya]

what? Rendered in what?

I wanted to know so bad!

I didn't know SketchUp was compliant and stuff enough to use other rendering programs on it; I guess I never took it very seriously. I should play around with it some next time I feel like going for SolidWorks.

oops forgot that bit lol its Kerkythea which is free and can be used with the free sketchup as well

you need to download Kerkythea and the su to Kerkythea plugin from there website

You will have to purchase SketchUp Pro if you want to export into rendering software...

haha no lol, who told you that? Kerkythea is a plugin you download

[Shocker003]

Your block design reminds me of the CM aquagate RC 80 block/pump combo.
I believe with the right pump and rad it might be an effective design.
The inlet was on the side of the plasticbut channeled in such a way that it gets to the fins first and the outlet has it chamber, where the pump was located. If you copy the D-Tek V1 flow outlet design it will be as good as most high performing blocks in the market.

[Mick64]

oops forgot that bit lol its Kerkythea which is free and can be used with the free sketchup as well

you need to download Kerkythea and the su to Kerkythea plugin from there website



haha no lol, who told you that? Kerkythea is a plugin you download

Thats good to know! Thanks for the info, might give it a try on my waterblock design when I get some time...

[AndrewZorn]

You'll also have to remember that water takes the path of least resistance. I know you're trying to aim the water to the edges to get it to flow all the way across, but you aren't going to get much flow toward the base. Which is where the temperature gradient thing comes in...

[NaeKuh]

very good comments leurler except the fluid temp difference will be very small.

We did calculations and tests when i was setting up a TEC chiller, that water needed about 350W of heat for it to go up 1C when moving 1.5gpm. <-- you see why this number is so important now.

Anyhow when you got that much flow even with impact blocks, slight design flaws will be almost non existant.

If you have a good design, draw it up, and let me see it.

[Mick64]

First attempt at rendering with Kerkythea. Needs a lot more work I think...

[bentleya]

what settings are you using?

[Mick64]

what settings are you using?

What settings are you referring to? There are settings for everything!

Just been playing around with it really, there's a lot to get to grips with. Need to have a look at the tutorials and download some more textures, etc...

[migueld]

First attempt at rendering with Kerkythea. Needs a lot more work I think...

Holy, that looks like a spaceship

Hey Mick64 I'd like to see your design tested, It'd be very interesting to see how it performs. Regarding the thermaltake design, I don't think the two can be compared at all. Fins are attached directly to the heatpipes, not the copper base. And the flow path is way different.

The XSPC Edge has a fin design, and it performs better than the fuzion 1:

[link1896]

There may be a better way of achieving the desired result, though maybe
not with the current group of blocks.

As you noted earlier, in any water block there is a temperature gradient
in the fins ( or pins ). Highest temps at the base of the fins. Same goes
for the fluid, not taking turbulence into account. So yeah, speeding up the
incoming fluid to force more of it to come in close proximity to the base
will help cause more heat flow into the fluid.

What is not appreciated is that the fluid is forced to abruptly change
direction. Work is being done on the fluid. Since it is not an "ideal" fluid
(no resistance to flow), some of the work being done to change the direction
of flow is converted into heat (this is a contributing factor to heat dump).
Furthermore, to take the EK Supreme and the Fusion V2 as examples, part
of the incoming fluid is hitting the tops of the fins/pins, which are flat and
have sharp edges. This creates turbulence where it is not needed and
generates extra heat. I'm not saying that a tremendous amount of heat
is created, but any extraneous heat created in the flow negates the same
amount of heat from moving from the base to the fluid.

It would be better for the tops of fins/pins directly in the path of the incoming fluid
to have rounded edges, or maybe even better, not have any fins/pins between the incoming fluid and the base.
Maybe have the nozzle even closer to the base. While I am at it, I don't
remember how all of the injector plates for the Supreme look like, but
the one pictured in this thread is not optimal. The injector plate should
not create much turbulence. Its purpose is to convert the pressure of
the flow into velocity.

Can waterblocks be improved significantly? Take a look at this article:
http://www.electronics-cooling.com/html/2008_aug_a1.php

a design with a fin/pin density of 3 fins/pins per mm wouldn't be impossible with a 0.25mm jewelers separating disks giving you 0.1 mm pins and 0.25 spaces. Very hard, but not impossible.The GTZ has 0.225mm pins.

I like your idea of no coolant flow outside the fin/pin gaps, and the idea of not trying to do a 90 degree turn and accelerate the flow at the same time.

[Mick64]

Mark II - Shorter, thinner fins, but not as short as Supreme or Edge....

What are your thoughts on this one?

[AndrewZorn]

I'm not trying to be mean, or say what you've done is worthless. I also had a "look at the waterblock I designed" thread.

...but the height of the fins has already been optimized. As very, very short. See the EK Supreme, or XSPC Edge (as referenced above). Copper costs are not what is holding back much better temperatures.

It's just that with the constant barrage of high-pressure water, combined with the 'gradient' effect of heat through a soild, that extra height is doing nothing but making the water turbulence less, adding material/machining cost, and having the water run past the coldest parts of the metal.

As Journey has said, "don't stop believing" that you can think up something better, but this idea seems to have been refined to what you see as the top waterblock today.

[Shocker003]

@Mick64
Don´t you think designing your top in such a way that it will have two chambers. In the middle of the chamber will be the divider that will force the water to flow over the core. The inlet chamber will have lower fins so the the water will get to all the fins at the same time and the out let can retain the original height of the fins you started with. Meanwhile where the divider is will have much lower fins. I hope my diagram can throw some light on my idea.

[tgm_of_xxl]

The thinner the base & fins the better the thermal exchange and thermal inertia.