Pro/Forums

Pro/Forums (http://forums.procooling.com/vbb/index.php)
-   Water Block Design / Construction (http://forums.procooling.com/vbb/forumdisplay.php?f=37)
-   -   New design idea. (http://forums.procooling.com/vbb/showthread.php?t=6101)

bigben2k 03-19-2003 05:54 PM

New design idea.
 
1 Attachment(s)
Going over the designs presented here (and elsewhere), notably Nicozeg, and a few #rotor style ones, I came up with a theory/idea for another block (be nice if I finished Radius first!):

In short, the idea is to combine jet inpingement, with a drilled block.

bigben2k 03-19-2003 05:55 PM

1 Attachment(s)
Here's the "theory in action":

bigben2k 03-19-2003 05:56 PM

1 Attachment(s)
... and the pattern I've selected, with an AMD core outline.

Don't ask me for dimensions. I'd guess at a 1 mm hole in the top, for a 3mm hole in the base.

Comments?

Cathar 03-19-2003 06:10 PM

Looks good to me. Could be done with 3 plates.

Bottom plate with the drilled dimples.

Middle plate with the jet array and exit arrangement, and an O-ring groove to seal to the bottom plate.

Top plate that's hollowed out with an inner and outer O-ring. Inner O-ring seals the inlet which leads to the jet area. Outer O-ring seals the block and the outlet area which could be a single outlet barb.

Sounds to me like an excellent idea. Often been pondering how to get an effective arrangement of jets over the core area. You could even have little cut-out inserts which fit into the upper jet arrangement cavity below the inlet to alter the jet array to suit particular die sizes, or have the insert have the jet holes drilled into it itself of varying sizes focussing flow at various sections of the block as required (and determined via a thermal spread analysis of the block for a particular CPU die).

"Dimples" could be as little as a 2mm deep indent of 2mm in diameter. Probably better to arrange them in a honeycomb fashion rather than a grid.

The whole basic block could be completely customisable.

I like it.

SysCrusher 03-19-2003 06:30 PM

That's basically what I did Ben but with out the dimples and a slight convex instead. It works great.

I like the idea myself. What Cathar said with the middle plate with inter-changable jet arrays is a great idea too. Something I didn't think about. Intel has some white papers about using that similiar idea as well. Just have to dig through their site to find them.

bigben2k 03-19-2003 08:26 PM

1 Attachment(s)
Thanks all!

I remembered that I have been suggesting this same honeycomb pattern for some time: 'bout time I do it too!

bigben2k 03-19-2003 08:31 PM

1 Attachment(s)
However... I certainly didn't intend to suggest a middle plate. Instead, I thought I'd use some wise words from a recent thread about plastic tops, and a resource from a shelved idea using copper microtubes.

So here's an overall view, with a theaded PVC fitting, glued to the top, for a polypropylene (or nylon) barb.

Why plastic? because it doesn't corrode.

the only issue I have with it so far is the milling of the area outside of the dimples: maybe I'll just leave it solid, that way I can build the whole thing with a drill and a hacksaw.

bigben2k 03-19-2003 08:36 PM

Hum... those microtubes would allow the water to clear out nicely, but might create a huge pressure drop.

As for the outlet, it'd be "Radius" style, but much smaller:D

edit: the only reason I'm sharing this idea, is because I recently noticed that out of all the blocks that BillA tested here, Owen's block was a high performer (details here).

It amazes me to see this type of block perform so well, and with such a steady performance curve. It's also the easiest/cheapest to build, and just about everyone can do it.

So I'll try to keep this concept under the same spirit;)

Fixittt 03-20-2003 03:10 PM

WOW im totally blown away. Bill, that was a GREAT little write up. Short sweet to the point, and it did not posess all the negative comments that I was used to. Althought they presented them selves. Bill touched base on the problems he encountered with the block, but in a tactfull manner that doesnt leave you say "Gee this guy is an as........" well you get the idea.

not to mention the graphs were typical Bill style. Easy to read and not full of cluter.


CUDOS Bill...... Would like to see more work like this. And I bet you enjoyed writing it too...........


I have a chisel tapping machine as well. BIG HAMMER!

bigben2k 03-20-2003 03:20 PM

Quote:

Originally posted by Fixittt
WOW im totally blown away. Bill, that was a GREAT little write up. Short sweet to the point, and it did not posess all the negative comments that I was used to. Althought they presented them selves. Bill touched base on the problems he encountered with the block, but in a tactfull manner that doesnt leave you say "Gee this guy is an as........" well you get the idea.

not to mention the graphs were typical Bill style. Easy to read and not full of cluter.


CUDOS Bill...... Would like to see more work like this. And I bet you enjoyed writing it too...........


I have a chisel tapping machine as well. BIG HAMMER!

(In reply to this thread, maybe?)

I don't suppose you have any comments on my idea, while you're here?:D

leejsmith 03-20-2003 04:28 PM

ben is this what you have in mind ?

(not to scale)

http://homepage.ntlworld.com/lee_smith/bb1.jpg
http://homepage.ntlworld.com/lee_smith/bb2.jpg

here with a multiple jet in the inlet tube

http://homepage.ntlworld.com/lee_smith/bb3.jpg
http://homepage.ntlworld.com/lee_smith/bb4.jpg

i have been playing with 3d studio max and the trial version of real flow on some block designs.


lee

jaydee 03-20-2003 05:57 PM

I have been pondering such a block since my REV 2.0.

http://www.customcooledpc.com/REV3/base.jpg

I just havn't had time to do it. Every weekend for the last several months has been occupied. Same with this weekend and probably the next. :(

bigben2k 03-20-2003 07:03 PM

Pretty darn close lee! but I'm proposing to direct individual jets to each dimple, to:
a-allow to have fins over the core
and
b-allow the maximum usage of a jet's trajectory

What baseplate thickness did you have in mind?

BTW, Cathar recently revealed (on OC?) that WW's bp thickness is less than 1 mm (!).

LiquidRulez 03-20-2003 11:39 PM

[quote]Originally posted by bigben2k


BTW, Cathar recently revealed (on OC?) that WW's bp thickness is less than 1 mm (!).
[/QUOTE ]

Less than 1mm??? Damn thats even thinner than I thought it was.
Gonna be even harder to beat that one than I thought!

krazy 03-20-2003 11:45 PM

the general design idea is an excellent one. jets create turbulent flow that rapidly exposes more surface area of water to the hot surfaces wishing to disipate heat. Jets have been used before on a small scale, but I havent seen anything like this before. The cooler I was brainstorming earlier today would have a open area over the core with a smaller diameter tube blasting coolant onto the area above the core from a right angle. The clearance between the nozzle's tip and the plate below it would be something like 1mm to force the coolant to spray turbulantly outwards and possibly crash over a few bumps before returning out around the outside to the outlet pipe.

I'm not sure how this could be incorperated on a large scale, but it would definitely be a very effective block.

The splitting down of the input flow to many small jets woud be difficult. What about if the block had three layers. The input would come into the side of the outermost layer and fill a largeish resiviour area. The bottom of the resiviour would be made frfom the top of the next layer. This would have many small (1/8" maybe) holes going through the ~.5" thick layer. The bottom layer would be the copper base block and have the wells machined into it. return would be around the outside edge or something.

any thoughts? am I bad at this, or might this design work?

davidjr 03-21-2003 01:14 AM

Hi,

this is my first message at procooling. Sorry for my poor english.

I'm interested in this waterblock, that bigben2k proposed, and draw a first prototype (w/ errors). The pictures can be found at:

http://www.be31020022s.hpg.ig.com.br/wb/index.html

-it was made using three plates;
-the holes in the bottom are 2 mm deep and have 2 mm diameter;
-rectangular space in bottom and midplate for 1mm o'ring.

Volenti 03-21-2003 01:45 AM

I've had thoughts about doing something like this too(re-my experience with jets in direct die cooling) This could be done in 2 layers, with a thick perspex upper layer, I'll knock up a prototype over the weekend if I have time.

TerraMex 03-21-2003 02:20 AM

davidjr :

I think that's it, but i also think that a reduction in the diameter of the "nozzles" would be better. Having a nozzle the same size as the "pimple" may not be as efficient.

The reduction of the nozzles would , like bb2k said, implicate a large headloss due to the restrictions. However, having a considerable number of those just might compensate.

pHaTtYaSs2x4 03-21-2003 03:35 AM

This is my first post. I like what i've seen around these forums.
Leufken Tech. had a "direct water injection" piece, dont know how it performed but also had a peltier option as well but thought i'd let ya know.(his black heater core rad is the shiz)

leejsmith 03-21-2003 03:39 AM

Quote:

Originally posted by bigben2k
Pretty darn close lee! but I'm proposing to direct individual jets to each dimple, to:
a-allow to have fins over the core
and
b-allow the maximum usage of a jet's trajectory

What baseplate thickness did you have in mind?

BTW, Cathar recently revealed (on OC?) that WW's bp thickness is less than 1 mm (!).

how are you going to hold down the the top ? if your base plate is too thin and depending on the o-ring and screw placement you could bend the base plate. (i did !). I made the mistake of not cutting the o-ring channle deep enough and maing the hold down screws to close to the edge. when i tighten the screws the edged of the base plate bent.

I am working on a jet for my own wb to spray over the hole cpu area if possible well let you know how progress.

From my 3 stage design i did before i can cut that down to 2 layers buy making the top higher than the fins then making a changable jet plate then will sit directly on top of the fins and also inside the inlet hole of the top. if that make sense.

Volenti 03-21-2003 03:47 AM

Quote:

Originally posted by pHaTtYaSs2x4
This is my first post. I like what i've seen around these forums.
Leufken Tech. had a "direct water injection" piece, dont know how it performed but also had a peltier option as well but thought i'd let ya know.(his black heater core rad is the shiz)

I don't think it was ever tested on a "live" cpu, judging by the experience I've had with direct die jet impingement, it would have performed like crap, IMHO.

Gilhooley 03-21-2003 07:47 AM

A simple variant
 
1 Attachment(s)
A simple variant that you could make with a drill press.

Drill the middle hole in block 4, then clamp 2+3+4 together. Drill the small holes in lock 3. Use block 4 as a template and drill through 2+3 .Also drill the other big hole through 2+3+4. Unclamp and reattach block 1+2+3, then drill the holes in the copper block using 2+3 to lineup the holes. Unclamp and cut a larger hole in block 2 connecting both holes. Screw / Glue all blocks together.

Gilhooley 03-21-2003 07:48 AM

Addenum
 
1 Attachment(s)
Thickness idée:

bigben2k 03-21-2003 09:29 AM

Wow! Lots of replies, and a few new faces too! Welcome to ProCooling!

I'm not going to get into the direct die approach, I think we've beaten that horse to death, and it's clear that it's not a viable option. Certainly the work that has been done may be used as reference here though.

I hear you Lee, on the top support. I could extend a bunch of copper rods, from the fins, up to the top.

The baseplate thickness is what's holding me back right now: if it performs very well, then I want a thin one, but if I were to use a standard or low pressure pump, then I would want/need more thickness.

David jr: thanks for the excellent render! One note, I was planning on using a square ended endmill, in order to avoid the conical shape that a standard drill bit leaves.

Gilhooley: I like your idea, and I don't have a problem with using an adhesive for the plastic-to-plastic connections, but I'm not sure what kind of adhesive I'd use for the copper to plastic region. Also, threading a top has been shown/discussed to be a no-no, so I'd opt for gluing a threaded adapter to layer 3, and drop layer 4 alltogether. Alternatively, I could glue a nylon barb, and just forget the thread alltogether.

Gilhooley 03-21-2003 09:51 AM

Use copper =)
 
Quote:

Bigben2k: I'm not sure what kind of adhesive I'd use for the copper to plastic region.
Well, you can just use four blocks of copper instead. The top block is mostly for deeper threading. Without it, the third block is harder to make. There is also plenty of space for drilling screw holes - so you don't have to glue or solder. (With a Dremel you could make small groves for o-rings.)


All times are GMT -5. The time now is 09:25 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...