Radeon 9600/9800 sandwich waterblock 3D model preview (images)

[Jackal]

RAM is 0.7mm lower than GPU

[satanicoo]

Ah man, you sure? i measured 1.0 mm...

[Jackal]

Well It may be a little different from card to card...
But at the end I can still shave off a little material on the pad for GPU to make a perfect fit.

[8-Ball]

Personally, I would be inclined to underestimate the difference slightly and use some kind of thick tim pad to make it up.

8-ball

[nicozeg]

Originally Posted by Jackal

But I think the days when I was making my waterblocks on a drill press are finally over. I managed to get an agreement to use the mill at my father's workplace. Of course only when it's free. Unfortunately that's mostly over the night. But that isn't a problem for me. And it's a huge mill (Prvomajska) over 4 meters high with 40HP engine and hydraulic XYZ table with manual or automatic feed.

Unfortunately this last weekend the mill was fully ocupied by others so I couldn't get any work on the block done. The other problem is that the block probably won't be finished before than in two to three weeks, because it's the end of semester at our faculty and I have to study hard for some exams and I live approximately 150 km away from home (I study at the faculty of electrical engineering - electronics & microprocessor systems - 4th year finished now). The good thing is that I choose to make a microprocessor controlled CNC controller for one of the subjects to pass And after that there will be no more milling by hand for me.

And its going to be GNU? Please please

[9mmCensor]

Originally Posted by 8-Ball

Personally, I would be inclined to underestimate the difference slightly and use some kind of thick tim pad to make it up.

8-ball

TIM is horrible for heat transfer.

[killernoodle]

Or you can just clamp the thing tight enough to make up for the tiny difference.

[sandman]

Originally Posted by killernoodle

Or you can just clamp the thing tight enough to make up for the tiny difference.

And possibly crack the core/damage card? Not the best idea.

[FizzledFiend]

I am also working voer an Idea like this for my 9600XT the issues I am running into is the clearance for other electrical components. especially the back side. My powercolor has a few rather small but tall black square "things" I was planing on just milling a hole for this "thing to reside up inside. However your Ideas with the flow patterns exceedes my Idea..mnight just rip that idea off of ya..Excellent man..too bad I know squat about solidworks

[8-Ball]

Originally Posted by 9mmCensor

TIM is horrible for heat transfer.

Maybe so, but it is much more important to get a good contact on the gpu than it is on the ram.

If you overestimate the difference, then the block won't make contact with the gpu. Not good.

I agree, getting it perfect would be the best option, but may not be practical.

8-ball

[Peppo]

Cind of beafy...
Do you really have to make it so large ?
Consider using the Bladerunner style

Good luck // Peter
Sweden

[starbuck3733t]

is it done? pics? :-D

[davidzo]

This block won't work.
The plexi/Polycarbonate Top will crush because of the pressure from the small screws. On some edges, there are 3-5cm between the screws, i don't think that you can get it proof there because polycarbonat is more flexible than you would imagine, especially when it is so thin because of the big channels.
The fin area ist much too complicated to produce it at a realistic level of costs. 0,5mm cutting tools are very rare and expensive. the smallest diameter what i would use ist 2mm.
The block will be much heavier as a zalman heatpipe (330g)and thats the maximum what is possible without risk that your card breaks.
The block wont fit on a 9700 series card, it is a little big in the corner.

The cooler all in all looks nice, but i have to tell you these difficulties, because i made these faults before too.
I know what i am talking about, i constructed a similar cooler and i had to fight the same problems. My cooler now has a less large diameter, less flow, less wheigt, less thickness (1slot!) but some more screws.

Here is the link to the original Thread (in German). Scroll down and view the next page, there are only a few pics, because it is still in the state of prototype, but it will be finished in a few days.

Sorry for my bad english, but i hope you'll understand what i meant

another pic:

[Jackal]

Maybe so, but it is much more important to get a good contact on the gpu than it is on the ram.

That is exactly what I was thinking too.

is it done? pics?

Not yet... As I wrote in one of the previous posts, I will continue with the water block construction after I finish some exams at our faculty.

The plexi/Polycarbonate Top will crush because of the pressure from the small screws. On some edges, there are 3-5cm between the screws, i don't think that you can get it proof there because polycarbonat is more flexible than you would imagine, especially when it is so thin because of the big channels.

Trust me on this one... I've done this before

The fin area ist much too complicated to produce it at a realistic level of costs. 0,5mm cutting tools are very rare and expensive. the smallest diameter what i would use ist 2mm.

I've done that before too . Ever heard of steady hand, dremel and silicon carbide metal cutting plates? That is exactly what I'm going to use.

The block will be much heavier as a zalman heatpipe (330g)and thats the maximum what is possible without risk that your card breaks.

That won't be a problem, because I can always make some kind of suppoprt for the card.
Although my heaviest copper heatsink (totaly custom made) for the graphic card was about 600g and the card is doing ok without any support. (Link to the original thread about this heatsink on Slovenian forum Slo-Tech. You probably wont understand much, but you can at least see the pics.)

[davidzo]

i don't really trust you in the one case with the polycarbonate. I prooved it whith my construction and it diddn't work out. I think, that you'll have to make the space around the screws a little bigger, this is too thin that the polycarbonate will get tears (Risse in german) because of the pressure in a little time.

I know dremels, but i didn't expectet that you would dremel it, because i thought that you wantet to produce it in a little serie.
For a really thin channel you can use a diamond metal cutting, which is thinner than a silicon carbide one (0.2 vs 0,6mm) and is much more conditionable for more channels.
I wouldn't make the finblock out of that shape. The flow resistence is smallest on the side of the channels, because the channels are very short there now. but the most heat and the biggest Temperatur difference delta T is in the middle. so i would make the fins/channels all with the same lenght.

Wow this heatsink looks really heavy, and i thought that i had one of the biggest Graphics card heatsink (alpha pep66).

[Jackal]

Finaly after a long time I found some spare time and finished the project.

Here are the final pics.

The block is designed to fit on Radeon 9600 and 9800. Both cards aren't identical, but there is +/- 1mm at the most and all of that was taken into consideration when I was making this block.

[FireCrack]

nice, noty much definition in those shots though.

[jaydee]

Well done. Hope it works as well as it looks.

[Jackal]

It works great.

[tex707]

Congratulations, man...this is really a great work...!!!!

[fruity]

isn't it too heavy for tha agp port?

[Tempus]

works great huh - numbers?

[ntrsfrml]

watch out for falling video cards

[Jackal]

The block is currently set up on my Radeon 9600 (unfortunately I don't have any 9800 around yet).
From 400MHz Core and 300MHz on mem I pushed it to 615MHz core and 390MHz mem. Voltmoded naturaly.

The block weights about 700g. I think there is no need to worry that it will rip out AGP, because the connected hoses carry come of the weight. Besides I can always make some kind of support for the card.

[Butcher]

AGP cards can carry a surprisingly large amount of weight.