Pro/Forums

Pro/Forums (http://forums.procooling.com/vbb/index.php)
-   Water Block Design / Construction (http://forums.procooling.com/vbb/forumdisplay.php?f=37)
-   -   Need some socket A design assistance... (http://forums.procooling.com/vbb/showthread.php?t=10090)

pilsn3r 07-23-2004 09:26 AM

Need some socket A design assistance...
 
1 Attachment(s)
Planning to wc my computer (barton 2500+, nforce2, radeon9700).
The goal is mostly silence but also some OC...
Only have access to drillpress with X-Y table (Also a lathe and other common machinery. No mill.)

Planned loop:
..................../ gpu-rad \
-pump-cpu <.................> small tank/airtrap-
....................\ nb- rad /

Nb and gpu blocks are to be identical inside.

Cpublock as follows?

http://forums.procooling.com/vbb/att...tid=3472&stc=1

Hole diam. 3mm.
Channels: 1mm. (Dremmel cutting wheels).
Barbs: 10mm.
Thickness: 10mm. (Got this already, wont change)

Pump: most likely eheim 1048 (big/expensive/noisy pumps are not an option...)
Rad: Cut up condenser from industrial AC made from copper pipes and Al fins spaced 1mm apart. (Probably wont change since i can make this exactly the size i want to fit in my PC-60. Will end up about 250*3-400mm)

Some questions:
Drill depth: big question... 7mm?
Channel depth: 6mm (leaves small cup that causes turbulence at bottom?)

Should i make the holes bigger? 4mm?
Or aim for a simpler/less restriktive design?
Thicker base in over cpu? Or cut down these pins?

Thanks to: #rotor , Jaydee and everyone else here who have given me ideas and inspiration.

Etacovda 07-24-2004 05:54 AM

Looks extremely restrictive, and also very time intensive. With a 3 mm base, you likely wont need any extra cooling in an area greater than a 30mm square around the core area, possibly even smaller... Just look at moving to 10mm diameter bits greater than that, and design an easy path for the water to move out of thats not restrictive. I recommend a row of 3 10mm drill holes, connecting finally to a 12.5 or 15, if you have the bits; depending on how you are planning to barb or pipe the outlets.

Go about 4mm between the drill holes; it appears you have them at 3mm? and centre punch each one; believe me, when you're drilling that many holes, you dont want them to be that close to each other; you'll probably have the bits slipping into the adjacent holes and its not very fixable if youve punched the holes already.

For depth, use feeler guages lined up to the base of the x-y table; makes it easy as hell. I use a car scissor jack (dont laugh, it works well!) underneath the drillpress table and leave the table unclamped, raise it higher than you need then use the scissor jack to drop hieght; really nice way of having extremely small increments for depth checking; you can be accurate to less than 0.1mm for the depth, which is really handy; clamp up the drillpress table once you've got the depth you want.

Any other q's, just ask and ill help if i can

davidzo 07-24-2004 09:04 AM

nice looking block, but it looks like very much affort for only ok performance.
the cooling area is to big.
The effective surface where you should cool the block ist about 14*14mm in the middle of it, depending on which processor core is under it.
The rest of the block has nothing to do with its cooling capability, so you should keep the restrictivity very low.
The water should flow trough the channels very fast, the faster it hits the copper walls of the channel, the better it cools. But trough which channels? Not trough the channels which don't really cool effectively because theuy are a many milimeters away from the die. If the most water flows trough channels which are not really used to cool the cpu, the flow in the important channels in the middle is less high as it could be. So a more restrictive block, but effective block can be better than a less restrictive where the water is flowing anywhere but not where its hot.
Most commercial Blocks out there have the same problem, their cooling surface is not optimized for the real diesize and they are simply too restrictive. I bet a swiftech mcw6000 would be a few k better if it would be half as big, has only pins on a surface of 14*14mm and would have half the height to increase friction on the baseplate.

Try some round shaped jets, that could give you some nice performanceincrease if you place them right.
Try to design the block for your barton, an jetpattern with the size of 15*10mm would be a fine choice then. The cooling structure on the base would be something like 17*12mm.
The pattern looks ok, but i would try some smaller holes, less thick material and less thick effectivebaseplate.
The key to good performance is a very small thermal resistor of the block.
The idea behind this is a direct die cooler. The advatage of a good direct die cooler is, that it can save about 5k (depending on cpu) only because of the missing thermal gerase and missing 1mmthick copperplate which is isolating the hot cpu.
The advantage of a CPUblock is that it can increase the surface of the cpu.
So to combe this two important facts, we must try to find a pattern which maximises the surface and has a very small thermal resistance (e.g. small baseplate thickness).
Many cooling designs only maximise the surface without thinking on the resistance of a thick copper plate. Thats the point where the cascade SS gets its power from. Inside the cups, the silver baseplate is so thin, that the thermal resistance of the heatsink is almost 0.
The baseplate must be at least 2mm thick to make it stable. But you can use a few tricks to get it thinner where it should be thin. One of this tricks is to cut wholes into the bp like cathar does it whitch his cascade.
Another is it, to create a micro pinpattern, where the pins support the thin baseplate that it cant bend under the pressure of the mounting.
Another point of baseplatedesign is the flow. Flow simply means to get as much water as its possible there, where it is needed (directly on the small hotspot) an to get it away from there without much restriction.
Many designs have the problem, that the flow is not enough devided and there where it is needed.
A Jet pattern whith many small jets helps to direct the water directly an exactly to where it is needed.

Try to think on this ideas. You could save time, money and work an get a better cooling result with obeying this advices.

pilsn3r 07-24-2004 09:41 AM

Its 4mm between the holes that are in line. Between the other holes it [3,1]mm.
Thats very close but it works with sharp drills and plenty of fluid..
Just did a test on a piece of scrap. Didnt take so long to make two half rows and cut the walls. Maybe 20min? (About 2h to make a complete block. Not a problem unless the results will be useless...)
Trying to get a camera so I can take a picture...

The Machine is able to set depth good enough so no need for the jack. Nice tecnique though, have to remember that...

?:
Will making the holes bigger decrease the resistance alot? (about 4mm (c/c 5mm, [4,2]mm) and cuts 1mm). Will draw this up to get a "feel" of it...
Or is it good enough to put the outlets closer together? The outlets will be rather easy flowing since i will cut into the top made out of 20mm acrylic there.
Or just make a side-to-side block and then a splitter to the other blocks? No impingement anyway due to the 3-4mm base.

pilsn3r 07-24-2004 09:54 AM

davidzo:
Using drills smaller than 3mm wont work for me. Neither is changing the material. 10mm copper is all I got.
You have made me lean more towarda a side-to-side block with finns over the core though.
What if i made just 5 or 6 rows 30mm long over the core?. (Base will still be 3-4mm. Less than that and I have to get thinner raw material)

pilsn3r 07-24-2004 10:06 AM

1 Attachment(s)
Made a quick drawing. Outlets and such not done at all.

edit: holes still 3mm and cuts 1mm.
Depth 6-7mm?
Will i benefit from a thinner base as im planning to use the rather weak eheim 1048?

edit2: It looks alot like jaydees lumpy block... But i guess thats a good thing. :D

Kobuchi 07-25-2004 04:36 AM

This is exciting. You haven't got your head crammed with convention yet, pilsn3r, so right now you might invent anything.

D0GMatiC 07-26-2004 12:57 AM

Quote:

Originally Posted by pilsn3r
Depth 6-7mm?
Will i benefit from a thinner base as im planning to use the rather weak eheim 1048?

Why don't you try 5mm depth? so the base will be 5mm thick, it should dissipate heat better than 3mm base. And your 2nd variant should perform better with weak waterpump, I like it :dome:

davidzo 07-26-2004 05:15 AM

the second design looks better to me :)
But why did you cut the 2nd outlet away?

I would make a 3bardesign with 1/4" barbs (more is not necessary for a 1048), with a little smaller channel area than yours.
The Top can be plexi or copper, in which you can mill some basins where the outlets are, to make the block less restrictive.
Then i would use only 2mm Holes for 1mm channels, to guarantee that there is no calm water in the holes (look at lumpy channel block again ;) ). Calm water has a very bad heattransfer, much worse than copper. calm water can isolate the effectively flowing water from the hot copper. This effect is the well known boundary layer effect which has to decrease as much as possible in a good waterblock.

With 10mm Material, you will not really benefit from a thinner baseplate i think.
Thats because the water goes always the shortest way, which is from inlet to outlet on the top of the block and not on the baseplate. you must find some tricks to let the flow go over directly over the bp and not further above in the channels, jets would probably help.

I would also mill a 3mm basin on the middle of the block itself, so you can decrease the channel height to about 5mm (->2mm bp) and insert a polycarbonate plate over the inlet in the middle of the block (little bit like DD RBX) whith some nozzles. I would direct the nozzles into the holes you drilled. Whith this technique you will have a very good flow distribution all over the channels and inside the drilled holes - less calm water, less boundary layer = less thermal resistor = more performance.

pilsn3r 07-26-2004 05:50 AM

I removed the center inlet to be sure that the water got to the bottom of the channels. Due to restriktion the fluid speed will now be uniform over the entire depth over the core. But this is of the cost of impingement.
I doubt that I will be able to construkt at nozzle accurately enough since the opening has to be smaller than the channels. A nozzle also kills flow i guess. Thats why I changed from center inlet with an easy way down the channels to the side-to-side setup.
There will be cutouts or basins in the top. Thats the big area around the in/outlets

How will 2 mm holes prevent stagnant water? My theory with this pattern is that the thin channels will dreate a jet (sideways) into the excentric holes. The jet will then:
1. Bend away from the next channel opening and therefore is destroyed and has to recreate at every hole. In other words, it cant hit the channel on the other side of the hole and just rush past.
2. Create turbulence because it cant expand evenly around its main direction. (This is one of the things that creates the jet to bend.)
Not that laminar flow will occur at any resonable flow anyway...

Then again, maybe this theory is all wrong...

Etacovda 07-26-2004 06:03 AM

Sounds pretty good from here...

I have made a block that looks pretty similar to that one; but with a top jet and one outlet barb. You should have good performance; im happy with mine :)

pilsn3r 07-26-2004 10:03 AM

1 Attachment(s)
One bad hole on the 2:nd line... Ah well.
Now I just have to make the top and lap the bottom.
Oh, and buy a pump and some tubing :D

Info:
Material: 10mm copper.
Hole diam: 3mm.
Depth: 7mm.
Channels: 1mm.

davidzo 07-27-2004 06:10 PM

looks good

a little bit like the block i milled before a few weeks, i also drilled 3mm holes in the walls of the 2mm channels in RBXstyle, but mine is smaller and has two cutouts at the inlet and outlet ...

jaydee 07-27-2004 06:16 PM

PROGRESS! Excellent. Good work so far.

pilsn3r 07-28-2004 04:16 AM

RBX? Hmmm, looks like I reinvented the wheel.
Maybe I should be a complete copycat and make a center inlet like davidzo keep telling me...
Still dont like the restriktion from a nozzle though... And with a 3mm base...
Guess I have to think a little more before doing the top.

"Note to self: Do a little more research on existing products. Will save time." :D


All times are GMT -5. The time now is 03:45 AM.

Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...