"The ideal water block"

B

Bruno Facca

New Member
#1
I'm designing a new water block (for commercializing), my goal is to maximize heat transfer inducing as much turbulent water flow as possible without creating too much flow resistance, I would like to know what are your thoughts on a couple of issues, please help me to give you what you want in a water block:

1 - How would you like a block that is one piece only (avoiding any leaks)? Do you feel the need to open your block for cleaning (or whatever) from time to time?

2 - What is your preferred retention system and why? I have some new designs here but there are no pictures available yet. The link below has images of some well known water blocks from various companies, they're all nice blocks, each one has it's own retention system. Please take a look and tell me wich one you like most and why, also any new ideas are welcome. I'm not going to copy anyone else's retention system, this post is just some research about the what the public want in a water block.
http://paginas.terra.com.br/lazer/bjfacca/waterblocks.htm

3 - Would you like the hose fittings to be 3/8 or 1/2?

My water blocks for cooling CPU, chipset and GPU will be available next week, I will post pictures when they're ready. I will also be making water cooled PSUs.

Thank you
Bruno Facca
 
Last edited:
MuFu

MuFu

Moderator
#2
Watercooled PSUs are a great idea (as we've seen from the likes of Bladerunner on this forum). I'm sure you'll find a decent market for them.

1. I don't think it is an issue whether the block is one piece or not as long as it doesn't leak (not what you were asking, I know, hehe). Being able to open it up and clean it would be a bonus, yes, although due to the anti-corrosive/anti-fungal coolant mixture most people run it won't be absolutely necessary. I wouldn't sacrifice performance of the block to make it possible.

2. By far the best two waterblock retension mechanisms I have seen are those made by Koolance and Innovatek. They are also the most complicated, however (especially the Koolance one). I can't really comment on those in the pics as I have never used them, sorry.

Hope this helps! :D

MuFu.
 
B

Bruno Facca

New Member
#3
@Mutha Funker: Bladerunner's watercooled PSU is a lot like the idea I have except for block size, location and mounting.

I wouldn't sacrifice any perfomance either :) I will make it openable.

I looked into the retention mechanisms you mentioned, they are complicated indeed but I'm giving them some studying.

Thank you
Bruno Facca
 
Todd a

Todd a

New Member
#4
I personnally like the inside design of the Danger Den Maze blocks. They lots of nice wide channels for heat transfer. The other thing I would do it use a plexiglass top on it. That just looks cool. You can hang it over the edge and mount the block through the holes in most Athlon motherboards (of coarse this is not a "requirement" by AMD anymore and they might disapear completely over time).

This is what most people would likely want, but I don't really like doing this. I prefer the tabs on the ZIF socket instead. A 6 cleat hook would be fine, like most of the better heatsinks have switched to. This has the disadvantage that the waterblock is limited to about 60x70mm. This might not be big enough to cool some CPU (especially when using peltiers). If you make the block taller with very deep, but slightly narrower channels (1/4"x1"), you should be able to get plenty of serface area. The waterblock would be about 60x70mm and about 40mm tall. This is actually still smaller than most heatsinks still.

This is the Maze3:



This is the Maze2:



This is the Maze1:



This is basically like what I would want but twice as thick with more, narrower and deeper channels.

A clip like this would likely do just fine:

 
B

Bruno Facca

New Member
#6
@Todd a: These are nice blocks, the finnish on maze 1 and 2 doesn't look very good to me.. Thanks for taking your time to draw your design idea, I'm not able to make this in time but I may build one of these for testing later, you can e-mail me at [email protected]

Thank you!
Bruno Facca
 
B

Bruno Facca

New Member
#7
I made sketches of the block and of 2 retention mechanism ideas I had today, some detail is lacking in the block because I had no time for doing a better model in 3dsmax. Please give me some feedback.



If the image doesn't load go here http://planeta.terra.com.br/lazer/bjfacca/

Thanks
Bruno Facca
 
ThreeOnTheTree

ThreeOnTheTree

New Member
#8
I have never used a water cooling system before, but i was curious about the results of a block that had a top that looked like a traditional heatsink (with fins). This would aid a little bit in cooling at the source. Since you are working on a new concept, maybe you could try making a cap of this design and see what kind of results you get? Anyone think there would be a benefit?
 
D

dfgsd

New Member
#9
seems like a decent idea to me. it would work a bit as a heat sink as well..i would think
 
B

Bruno Facca

New Member
#10
Maybe it's something to try but in my opinion it doesn't worth the hassle, it would probably take a fan to get a significant temp drop and the retention mechanism would get even more complicated, sorry...

Bruno Facca
 
Todd a

Todd a

New Member
#11
The springs on the retension clip is a bad idea. You are not pushing dirrectly down on the CPU. Just have the thumbscrew put pressure dirrectly over the core. Getting the heatsink or waterblock 100% flat is VERY important. That is one reason I do not care for using the holes in the motherboard. They tend to use screws and such and you need to adjust it several times to get it perfect. It is almost impossible if the motherboard is already installed in the case. With a single contact point pushing dirrectly over the core, you cant go wrong as long as it is easy to connect the heatsink and apply pressure. Inovatek has a pretty snazzy one, but it might be harder to make.

As for your waterblock it is a pretty basic design. The channels are nice and wide for good flow, but the amount of surface area is pretty limited. That is why I suggested a taller block with much more channels. Think about it, if you make the channels 1/2"x1/2" and (by my guess) about 6" long, even counting top, bottom, and sides, you are only looking at 12 square inches of surface area. With my block you are looking at 1"x1/4" and about 12" of channels. You now get 30 square inches of surface area. The wavy edges will increase surface area, but increase turbulance quite a bit. Most of the water will get trapped in those waves and ridges and most of the water flow will only be in the middle. If you could make the ribs run with the channels, it would work better, but be harder to make.

Also the lack of sharp corners will increase flow and the large chambers at the inlet and outlet will help cut down turbulance as the water changes to the tall and narrow channels. You might even taper the input and output holes at the bottom to curve into the channels at the bottom. With your block the sharp corners will decrease flow in the corners and increase turbulance. The more turbulance you add, the harder your waterpump needs to work. If you know much about water pumps, they generate very little pressure and most can not even pump water above 6 feet.

I've been looking into water cooling pretty heavy for the last two years. If you like my ideas and want to use then, all I ask is for some free stuff. :) I do not have much money to spend on my water cooling though. I did build a water cooling system out of scraps and copper tubing. Total I spent about $60 on it and the MAG 250gph pump cost $35 of that. I just upgraded it a bit with some elbo peices and a smaller radiator so I could fit it INSIDE my case. Here are some pictures of the new system:
 

Attachments

Todd a

Todd a

New Member
#12
The picture is kind of small (took it with my Sony HandyCam TRV820), but this picture shows the radiator a bit better. I used a box of 3/8" copper tubing (20 feet for about $12) and I coiled it into a spiral about 5" across and 3" high. I then coiled it over itself to the same height. Then I loosened up the coils so there was a small gap between the coils. That is about 25 square inches of surfas area. My old one was about 15' with fins cut out of the copper tubing soddered into it. It was very bulky and about 5x8x4" and used a 120mm fan. I slapped it onto my K6-2 450 running at 2.6v and 504Mhz. At idle it sat at 75'F (room temp) when run for 24 hours. I then put it under load for 24 hours and the temp rose to about 85'. This was with the fan turned off.
 

Attachments

Todd a

Todd a

New Member
#14
Heatpipes have some very nice features but they are limited on how much heat they can transfere and how far. Without a pump to increase the flow rate the heat transfer is limited. Heat pipes will never be able to compete with good water cooling systems. Water cooling can be increased by increasing water flow or surface areas of the waterblock or radiator. The big problem is they are pretty big and require at least a little mantenance.
 
B

Bruno Facca

New Member
#15
About the springs, a hard mounting is what I was aiming at in the first place, some people said springs would help to make even pressure on the block and prevent you from crushing the CPU core. I was thinking: a couple of heatsinks I used for cooling my CPU before going water cooled were so hard to fit I had to push down the metal clip very strongly for it to get to the socket lugs, that was an enormous pressure on the core, I think if that didn't break my core a hard mounted water block wouldn't either, unless you screw it really hard using pliers or something. I will make a single contact point over the core, a big contact point and maybe with a thin sheet of hard rubber so it will hold the block in place better without decreasing the pressure.
Also a guy at procooling's forum gave me this simple yet effective idea: you can also post into bios and use the temp monitor to adjust the pressure on each corner of the block untill you get the lowest temp.

@Todd a: About your block, here are my toughts: The big surface area on the inside sounds good, the thing is, causing turbulence is one of my main objectives, it makes a little harder for the water to pass (that's why the channels are wide) but it can even double heat transfer, making the ribs run with the channel would give me the same surface area but it will cause the flow to be more laminar, making it easier for the water to pass but also decreasing heat transfer between the copper walls and the water. The closer you get to the copper wall the slower will be the water flow, in fact the water molecules that are touching the copper are not moving at all, between this still molecules and the "free stream area" that would be the middle there is something called boundary layer, this is the part of the flow that is moving but it's slow because it suffers from the "friction" caused by the contact with the water block (and the still water molecules), this is bad because the boundary layer kind of insulates the hotter water that is closer to the copper from the colder water that is running at a higher speed at the center of the block decreasing heat transfer a lot. When you have small, smooth channels the boundary layer might even be the size of the whole channel. Turbulent flow decreases very much the size of boundary layer making the heat echange much better even if the flow rate is suffering a bit from it.
Nice rig you got in there, especially that radiator, looks small and effective, I want to try one of these soon, have you ever tought about building one with 2 or 3 coils in parallel? That would probably increase flow rate and decrease pump stress. too bad copper tubing is kind of expensive here.
Have you ever calculated how much watts of heat your processor dissipates? I guess mine would be about 70W. I would like to know that because your temperatures are quite impressive.
 
Last edited:
Todd a

Todd a

New Member
#16
I still don't trust the increased turbulance. With standard heatsinks this is a bad think. It puts a lot of back pressure and causes the air to get stagnant. I made a heatsink with an old GlobalWin FOP32-1. I added a low RPM 80mm fan on top. To fit the fan I spread the fins out on the heatsink out to 80mm. The larger space between along the top drastically decreased the horrable turbulance at the top and this heatsink works pretty good. Thye fan does generate more CFM airflow then the YS-Tech 60mm fan I used to have (34 instead of 26), but the amount actually passing through the heatsink is about 3 times what it was befor. It cools better than most copper heatsinks with 700RPM fans on top. 42'C full load is a great temp for an all alluminum heatsink with a 2500RPM fan that generates a minumal 30db.

If you look at the Maze3 above, they did add some very shallow channels in the bottom to sort of mix up the water. I'm guessing it would make the water spin slightly in the channel as it spirals around the water block.

As for the pressure on the core, AMD requires something crazy like 20+ lbs of pressure on the core. Have the tip be a half circle blade and add a small notch in the middle so the block will not slide or twist. A large thumb screw with fine threads would be fine.

You do not want to put radiators in parralel. The water will take the path of least resistance and flow more through the radiator with the least restrictions. If you want two or three then run them in line or series. I was thinking of added a third coil on the inside, but I did not think I could get any smaller coils without crshing the tubing.

Chances are both blocks will work just fine for standard water cool. Peltier cooling will start to push things a bit. Make both and compare them. That is about the only way to really check. Also you have to look at the cost to make each. I have no idea if the ribs will increase the cost or is a spirrellnig channel is harder to cut, but if the performance is similar in both, you likely will want to use the one that looks better and cost less as these factors sell blocks better than most other features.

As to watching the BIOS as you tighten down the four bolt, I'd still be afraid of over-tightening one corner and cracking or frying the CPU. I have little money to replace parts I screw up. That is why I tested my cooling on my K6-2 instead of my Duron 900 or AthlonXP 1600+.

By the way here are some links to pictures of my mod'd heatsink and water cooling set-up (work in progress with testing and my old radiator):

http://www.geocities.com/todd_e_abbott/photopagerock.html

http://www.geocities.com/todd_e_abbott/computer1.html

http://www.geocities.com/todd_e_abbott/computermod3.html

http://www.geocities.com/todd_e_abbott/geforce4ti4200mod.html

If you have not figured it out, I'm bored. :) My wife works 2nd and I work 1st and I'm stuck at home watching a 1, 3, and 5 year old. Computers is one of my only hobies left.
 
ThreeOnTheTree

ThreeOnTheTree

New Member
#17
Heh, I did the same thing to a vantec HS. I didn't want the 6800rpm beast on it, so I widened the fins and stuck an 80mm, just as you have. :)
 
Todd a

Todd a

New Member
#18
Yah I had ordered a FOP32-1 for a friend's computer when I upgraded it because I liked my FOP32-1 I ran on my Duron [email protected] Well the company messed up and sent me two (I double checked, but they only charged me for one). Well I mod'd it about 2 years ago, but it set in my parts box for a year befor I got my AthlonXP 1600+. I was paranoid about using it, but when I finally hooked it up, I was plesantly supprised with the performance.

Well I read a few articles and some turbulance is supposed to be good. They said the s channels aded turbulance (and surface space), but they were showing off a Swiftek waterblock which was basically a box with the inlet in the middle and a bunch of depressions covering the bottom of the heatsink (looks like they just used a drill, but only went in a bit with the tip to make little cone shapes). It perfromed pretty good with a peltier and an AthlonXP 1600+ running at 1.83Ghz. They got the CPU down to 9'C.

I also was bored and desided to design another idea. Think of a block of copper about 65x70x70mm and drill 16 holes in a 4x4 grid all the way through the block (3/8" holes will fit with about 1/8 inbetween and an extra 5mm on the bottom). Then cut some small channels in the end to connect the channels together. Add a plate on each end and some water fittings.

I made a guess on surface area on this thing and it falls in around 50 cubit inches. Thats more than most heatsinks. With all the bends it should have plenty of turbulance, but hopefully no dead spots. It is likely a bit overkill, but should be pretty easy to build. You could even make the end plates out of lexan That would look really cool with some dye and in the water (like Dylite) and a black light. The sides of the block would glow nicely. :D
 

Attachments

Todd a

Todd a

New Member
#19
And here is my clip idea. If you have guessed, I have had some mechanical drawing classes way back in high school, but all are just quick free-hand drawings without the aid of a simple ruler or cyrcle templete, but you should be able to understand them just fine.
 

Attachments

Last edited: