waterblock idea

csimon

I've been mulling over this waterblock for a long time now.
The basic design is pretty self explanatory from the renderings. I want to try and use hexagonal pyramids in a fin type array for the base and inverted hexagonal pyramids cut to 1/3 height for the jets leaving a 2/3's spray distance.
Keep in mind that this is just the jet and fin area ...this needs to be hashed out before moving on.

I've been in touch with a few guys over at procooling the past few days and I'll probaby start to get feedback soon.
Check it out and let me know what you think.

profdlp

I'm intrigued. It certainly looks like you'd have plenty of surface area for good heat transfer.

csimon

Thanks prof that's one of my objectives ...and the pyramids can be higher pitched for even more surface area. Other objectives are adequate turbulance and flow.
If I can get the jets to spray into the pyramid array using the pitches of the jets that would be another acomplishment perhaps.
If you notice the wedges between jets should reduce flow resistance and concentrate it through the jet itself.
Each jet is actually set directly over each pyramid. There is a 2/3's gap from het to pyramid and the jet itself is a pyramid with a 2/3's fulcrum.
It probably still needs a bit of refinement but I wanted to publish the concept here before I went anywhere else with it.
Actually fins and jets and even pyramids are nothing new to waterblock design ...I've just never seen them done with a hexagonal.

ryko

not sure if i am completely understanding your concept, but from my perspective the only thing i would be concerned about is the flow of water. with that pattern it doesn't seem like all of the used (warm) water would have a direct path back to the loop. some of it might get caught inside the edges of the block and not make its way back to the radiator right away which would make for an in-efficient cooler.

you might need more flat space in-between the hexagonal pyramids to allow the water to flow more evenly. try to eliminate anyplace where water could get stuck. the edges seem problematic to me.

other than that, i find it to be an interesting design. the hexagonal pyramids have more surface area to cool off the water so that's good. is the top (blue) part the same as the bottom? as in identical? maybe just a flat smooth top with the pyramids more spaced out on the bottom.

good luck!

lemonlime

Looks like a really interesting design csimon..

Are you thinking about getting this machined?

scott

lemonlime wrote:

Are you thinking about getting this machined?

It does look pretty cool but, as a machinist I can tell you it won't be easy to make ( i.e. Expensive )

I have been playing with making waterblocks for about a year now and I am really suprised at how little difference the blocks really make. It seems that getting the heat off the block is much easier than actually transfering the heat to the block. I have had my biggest gains in better radiator efficiency and optimizing flow for the system. I will attach some pics of a design similar to yours ( Except the top plate ) I tried this design with center inlet and dual exhaust. But the ( Shown ) cross flow worked better on my setup.

Food for thought


Scott

csimon

crossflow works better?
Did you jet the water over the fins in the center configuration or just flow it through? I'm sure that makes a difference.
At any rate ...I would prefer to use a single inlet and single outlet.

csimon

lemonlime wrote:

Looks like a really interesting design csimon..

Are you thinking about getting this machined?

Yeah once I get it all together I plan on machining it ...well the base anyway. I will make the jet top myself maybe.

csimon

scott wrote:

It does look pretty cool but, as a machinist I can tell you it won't be easy to make ( i.e. Expensive ) Scott

It would be a breeze if I could cast it.

scott

csimon wrote:

crossflow works better?
Did you jet the water over the fins in the center configuration or just flow it through? I'm sure that makes a difference.
At any rate ...I would prefer to use a single inlet and single outlet.

With the center inlet design I tried both. Wide open flow and restricted ( Jets ) They worked out about the same. I had somewhat reduced flow with my jets. If I had the same flow with the jets as I had with the wide open configuration It may have been better. But neither of those worked as good as the crossflow.

Most of all keep in mind that with all the designs I have tried there was only a couple of degrees difference. In my present setup my worst block runs at 45C and my best block runs at 40C ( Full load XP-M 2500+ @2505 )

Scott

csimon

40c full load is incredible really. What is your vcore setting for that?
I have run my xp3000/400 @ 2400 1.825v for over a year now. In the beginning when

I was on air I ran about 43-45c and I was able to clock to 2.5g with no issues and with 2-2-2-11 timings. Since I switched to my ww I haven't seen those temps or clocks again. I've tried everything with this equipment that I can think of and I'm even in a new house with way better ambients and the best I see is 48-52c @2400@1.825. The biggest difference is that now I don't have my vga silencer which helped with sys temps and in turn helped cpu temps somewhat. I've always leaned towards blaming the radiator ...it's a pro-120 core.

So...I have to say that I really like the crossflow best because of it's symetrical design. It's a well balance block. I couldn't figure out how I was going to have a center inlet while producing a single outlet on one side ...that would through off the symetry.

As I mentioned earlier I still have plenty of refinement to do on the design.
I will change over to crossflow then ...are the jets still necessary or should I reconfigure the fins? I can align the fins or change the pitch of the jets to get way better flow.

One of my biggest questions is ...what size should I make the fin area and what size should the in/out be. I have an idea that the fin area needs to cover the die for sure.
My target cpu is 90nm 939 ...preferably x2 if that makes a difference.

Thanks for advice scott and everyone else!
csimon

Nightwolf

AWSOME!!!!

csimon

Here is a simplified version just to show the geometry. I modified it a little ...especially the base ...just to show how flexible it is.
Now with the tetra series on base I have much better flow as suggested earlier. Still in keeping with the hexagonals. At the first angle you can see the V path. This would also be easier to machine than the hexs.
There are many variations to go with but I just wanted to get my concept across.
They render way faster too this way!

Leonardo

Disclosure - I have no hands-on experience with watercooling. OK, that's out of the way. Looking at your picks, Simon, I would think the pyramid design will dam up the water, preventing it from flowing smoothly into the exit receptacle (whatever the outlet is called). OK, if it's a cross flow configuration, I think there will be damming. If it vents straight off the top, the pyramids might create swirling, which can be good in an intake header for an automobile engine; but with water, I think it might prevent efficient flow away from the water block. Again, this surmising is only from looking at the pictures.

scott

csimon wrote:

40c full load is incredible really. What is your vcore setting for that?

vcore is 1.65v 200x12.5

csimon wrote:

I've always leaned towards blaming the radiator ...it's a pro-120 core.

Where is your rad? in the case or out ? On my setup above I have it in the case and was originaly sucking air through the rad ( Danger Den Black Ice Pro 120mm )from inside the case. I reversed the flow and am now blowing outside air into the case. That dropped my water temp a few degrees. And then I got tired of the "whine" that the MCP650 pump made and decided to run it at 9 volts instead of 12 and was ready to accept the increase in temps but to my suprise the temps dropped. So I guess it is a matter of balanced flow with the system.

csimon wrote:

One of my biggest questions is ...what size should I make the fin area and what size should the in/out be. I have an idea that the fin area needs to cover the die for sure.
My target cpu is 90nm 939 ...preferably x2 if that makes a difference.

I also tried a few variations here. One of my first blocks was a big chunk of copper with a simple path through it . When in use it was more than warm to the touch. It was sinking the heat away from the cpu but I did not have enough water on it to take the heat away. So my next designs were much skimpier on the heat sink and almost totally covered with water. The one above has the body made from plexiglass ( Very poor thermal properties ) And the copper sink is totally flushed with water. I would agree that you want your "Fins" to be at least the size of the die.
Are you running 1/2" tubing? If so I would make your ins and outs 1/2" as well.



Scott

Garg

Both of you guys have some really cool designs for waterblocks! If you ever considered making a small batch of them, I'm sure you could sell them here on SM. If they're about as good as what I can buy elsewhere, I'd rather have one made by a Short-Median

(how cool would it be to have a waterblock made by a SM member running in SM11? )

profdlp

I was fortunate enough to get to see Scott's computer setup and his shop in person. If he ever decides to branch out into the watercooling business he is well-equipped to make a big splash in the field.

/me apologizes for the pun - it just happened. :o

scott

Garg

SM11 looks gooood

Funny though, I was gonna use a Skyy Vodka bottle for my res. ( Funny what alcohol will do to a modders brain )


Thanks for the kind words Prof !

Scott

csimon

Leonardo wrote:

Disclosure - I have no hands-on experience with watercooling. OK, that's out of the way. Looking at your picks, Simon, I would think the pyramid design will dam up the water, preventing it from flowing smoothly into the exit receptacle (whatever the outlet is called). OK, if it's a cross flow configuration, I think there will be damming. If it vents straight off the top, the pyramids might create swirling, which can be good in an intake header for an automobile engine; but with water, I think it might prevent efficient flow away from the water block. Again, this surmising is only from looking at the pictures.

Yeah you may be right ...I'm no expert in fluid dynamics by any means. I was thinking that water sprayed from above the array (well spray or turbulence if you will) would be disbursed along side the tetrahedrons removing the heat as they flow over and around the surfaces. If water is dammed and accumulated it will just sit there and boil which isn't good.
I need to study more and harder but I do know that even flow won't remove the heat like turbulence will. I invisioned the valleys to be enough to allow flow through but that may not be the case or it may not be the most efficient means.
Perhaps I could make paths between the tetrahedrons to make the paths like so - _/\_/\_ ?

Thanks for your input Leon!

csimon

Here is the base I'm having built. I'll be sending in the top as soon as I lay out the toolpaths for it.

Pretty simplified really. I kept it to a 3mm tool min.

Garg

Looks good

csimon

Gargoyle wrote:

Looks good

thanks garb ...It's getting redesigned though ...again.

csimon

I'm trying to get an evaluation sample of this carbon fiber and nanotube technology atm ...

http://thermocomposite.com/TechPresentation_files/frame.htm

Leonardo

Thanks for the link. Very interesting. In a few years we'll look back at our fancy, big metal heatsinks and laugh at how primitive they were.

csimon

Leonardo wrote:

Thanks for the link. Very interesting. In a few years we'll look back at our fancy, big metal heatsinks and laugh at how primitive they were.

Well this is the second response I get ...

"We only provide license the technology to configure the anisotropic fibers
into a transfer agent. You would need to provide your own materials. All
materials I have are for Gov. contracts."

Brad

Where the heck do I get this? All I need is a small bit to test ...sheesh.

Leonardo

Looks like you have some heavy duty researching, to follow with emails to industry marketing types!

csimon

Leonardo wrote:

Looks like you have some heavy duty researching, to follow with emails to industry marketing types!

Well so far I've found out that it cost a mere $200k to produce this stuff.
Heck Tracie would never go for it ...LOL

I mean I can hear myself now. "Baby, darling, butercup ...would you and jr mind moving in with your mother for a few years ...I need to sell our new house to manufacture nanotubes ..." :shakehead

csimon

This was the final base submitted. Far less restrictive and probably won't perform as well as I'd like but hey ...I'm saving my best for a patent maybe!

csimon

Last I heard today he was milling the next block from silver? I've yet to see pics.

Talked to a patent attorney today. Basically he wants $1200 retainer to patent search and advise then a patent fee or $250 per hour if it's worth persuing.

Oh woe is me!

My neighbor is an attorney not specializing in patenting. He told me this weekend that I should at least try a copyright by mailing the design to myself. I think I'll draw the geometry this weekend and then print it out and mail it to myself. I mean even if it's worthless it can't hurt anything ...unless of course it get lost in the mail!