9800 pro full water-cooling
 

Not mine, got beaten to it this time around, :) nice though and a one slot solution. I question whether the VPU & Ram on the frontside are both making good contact, as solving this satisfactorily is why I still prefer separate ram & VPU blocks

http://www.zfz.com/pictures/projects/r9800prowc.jpg

It looks really really good. You should still make one, but you can make it better :P

Nice finish, though.
Is it alu or nickel-plated? ;)
Is there a difference between height of GPU and height of RAM above the card?
LAN-ing too much, drinking too much, sorry for spelling errors, if any, can´t tell myself.
regards
Mikael S.

The forum BladeRunner points to only has the pictures, and absolutly no info about the blocks. I read the thread to be sure (it's in braziliian btw).

http://akiba.ascii24.com/akiba/news/...44096-000.html
Original link.

Translated : http://babelfish.altavista.com/babel....html&lp=ja_en

Final product in stainless steel ...?

This is just educated guessing on my part, but I think it's silver because he has coated it all in solder then polished up the solder, (study the other images in the picture linked topic).

The effect is nice but let down a little by areas of the base plate being left with copper showing through.

On all the 3d cards I've owned there is a difference in height between the Ram & VPU/GPU. It differs from card to card depending on Ram type, VPU/GPU type, and card design, on the 9800 pro it's around about 1mm

I'm making a cooler atm ;) had a 9800 pro for a while just been side tracked by other things:-

here's where I'm at.

http://www.zfz.com/projects/r350/r350wc016.jpg

http://www.zfz.com/projects/r350/r350frb003.jpg

Copied from the link (thanx TerraMex):

"With the mechanism that, as for the water cooling head the base copper, the shell has become the stainless steel make the VGA tip/chip and on-board memory from both sides put with the water cooling head. Originally, with the product which the user of the individual made with hobby, first as for the water cooling head part all it was copper, but when it commercializes, jointly using the stainless steel, it lightened, the ことだ. "

Babelfish seems to work better in The Hitchhiker´s Guide to the Galaxy. ;)
Any known galvanic problems between copper and stainelss steel?
regards
Mikael S.

I realy doubt stainless steel is what is 'meant', especialy WRT "jointly using the stainless steel, it lightened, the ことだ" as it's as heavy as two fu##ing elephants! :eek: ...

SS is almost impossible to machine as well, it's very hard!. It'll blunt a drill bit first time (and you'll not finish your hole :D )...

hey bladerunner what happened to the FXCool the 5800 cooling for gainward.

oh and just to make you work faster. You might want to "add some more logs to the steam train, because others are on the way as well" this about the 5900 line :)

this time around i didnt have to measure a thing. got cad's from the devil himself. (or hir right hand man)

I've had a 5900 ultra for a while, (all be it a non working example). Before that however I had to make my own NV35 from images, and tech specs because I needed one to work with, but no real card couldn't be sourced at that time.

Here's the one I made earlier, wish it wasn't just a dummy.............

http://www.zfz.com/projects/fx/toyah...y_card005s.jpg

I could tell you more, but then I'd have to kill you :p

Radeon 9800 now running water-cooled. It was the most complex cooler I've made to date, but has worked out quite well, considering the 9800 pro's layout is annoyingly even worse than 9700 Pro for making a ram block solution.

Will have a write up about it soon hopefully, but for now a few pics. design & shape dictated mostly by where it will end up eventually, along with the desire for one 8mm festo input & output for both frontside and backside ram blocks.

http://www.zfz.com/projects/r350/r350wc300m.jpg http://www.zfz.com/projects/r350/r350wc301m.jpg
http://www.zfz.com/projects/r350/r350wc304m.jpg http://www.zfz.com/projects/r350/r350wc305m.jpg
http://www.zfz.com/projects/r350/r350wc404m.jpg http://www.zfz.com/projects/r350/r350wc508m.jpg
http://www.zfz.com/projects/r350/r350wc500m.jpg http://www.zfz.com/projects/r350/r350wc502m.jpg

There is just nothing else too say then!

NICE!!! :drool: :D

Your work as allways is, top of the line!

-P-

‚± ‚Æ ‚¾ = kotoda?

Nani?

Alchemy

Can you give my an idea of the waterpath in the ram blocks?

Very nice stuff.

Do the chips near the mounting bracket (voltage regulators?) get as hot as they do on my 9700?. I have a heatspreader plate on mine that I plan on putting a passive HS onto cause it gets damn hot!...

Any plans for doing this or something simular (H2o?) on your 9800?...

Nice work as always. I'd like to be sure of the water path as well. I take it it 'wraps' around the card otherwise that join would be pointless. How well does it contact the chips?, is it easy to fit/slip on?...

The ram block is two parts that bolt together sandwiching the ram chips between them. the coolant enters via one of the 8mm festo fittings that is fitted in the front ram block, and is then split to a loop channel in both front & rear ram blocks to exit agin by the other fetso fitting in the front ram block. The connection between both blocks is done with rubber O rings that are compressed to seal when it's bolted together. There was also a fair bit of base milling in key areas to avoid the surface mounts near the ram chips, and the lower "finger" that joins both blocks together has a relief milled into it so to avoid the coil on the card in this area. Took a fair bit of thought to solve this one the way I wanted, and as said it's layout shape has other constraints to its design that wont be so obvious atm. It allows a very thin, (4.85mm thick), backside ram waterblock.

Here's a few images that may help to explain how it was done, channel in backside ram block simmilar to the front side one shown. Coolant splits in the entrance chamber to both and rejoins on the other side after some passes through both front & backside block paths:

http://www.zfz.com/projects/r350/r350frb000m.jpg http://www.zfz.com/projects/r350/r350frb002m.jpg
http://www.zfz.com/projects/r350/r350frb011m.jpg http://www.zfz.com/projects/r350/r350frb010m.jpg
http://www.zfz.com/projects/r350/r350rrb003.jpg http://www.zfz.com/projects/r350/r350rrb002.jpg

The 9800 pro doesn't seem to suffer the hot vr chip issue that the 9700 pro did, (it may do when I vmod later on though). I used a zalman chipset sink o the vr cooling plate of the 9700 Pro initially, and it helped my ram overclock. I did later make a water/vr block for it. more about that HERE

Wow, I didn't fully appreciate what was going on until I saw those o-ring pics, I though some of the festo's were just hidden from view. WOW! Now that's how a mill helps :p Orings :D

It appears to works fine, (so far), but in using seals that way, they are acting more like a gasket which is not an ideal. However as long as there is plenty of "squash" they will work fine for water cooling pressure levels. I wouldn't want to put 200psi air through that type of sealing and rely on it not leaking. To use an O-ring correctly it would need a very accurate CNC machine to create a circular grove the size of the O-ring with say a 1.5mm ballend mill bit, not something I can do with a hand mill, (an expensive rotary mill table may make it possible, but still far from easy).

This image shows how the entrance chamber connects to the channelling in the front ram block, the two holes shown are drilled down to the channel path and sealed at the top with plugs later on. This is later covered by the name plate shown in previous finished images. I never like to say, it can't be done, but I very much doubt this type of block would be possible without a mill, and I'm constantly learning more and more about how to get the most out of my mill :)

http://www.zfz.com/projects/r350/r350frb016m.jpg

I should also state for the record I do screw up when making stuff like this. along with basic milling errors going too far or the wrong way, (very easy to do), the bolt holes are M2 and that is quite risky drilling & tapping in copper. I snapped a tap in the finger area so that is why there is a bolt hole on the frontside ram block, (the intention was to have two bolts that hold it together just on the backside block along with the 5 in the chamber area). Once a tap has snapped it's near impossible to get the broken bit out. I then went on to drill the additional hole and unintentionally drilled all the way through..,. arghhhh... this was a case of, I should have put it down and started on it another day :rolleyes: My point here is just about every block I make I screw something up but normally with a little redesign I can save the day :D

Hey Blade...

All I gotta say is... Freaking swwwwwwweeeeeet! :D

I thought it was something like that. Very nice. Bit annoying when taking off though, as you can get bits of water everywhere :(

It's not what I thought, I thought the water entered one side went down and connected through the lower 'join' and up the other side to exit. Do you think it gets good flow on both sides?...

My V~Chips get so hot I can't touch the plate!, even putting a fine~finned HS on that is the same size as the plate and 1inch high does'nt get rid of the heat, it just knocks abit of it off. At least I can keep my fingers on it now though :D I though the caps' on the other side were gonna blow!!...

*Edit* I can't keep my pinkies on the strange little graphite(?) things with copper wrapped round them, even though I have the HS on and they are on the other side of the card!. Have you read/heard of anyone removing the heatspreader/plate?. Mines on there quite firmly and I don't relish the thought of ripping the SMT chips off the PCB. I wonder if putting it in the freezer would work like the GPU epoxied on HS used to?...

Forgot to ask!, how much does it weigh in at with the VPU on there as well?. Does it 'stand up' allright? :) ...

Thanks Joe ;)


pippin88

The connections and input/output are in the top chamber area so as long as the card is held upside down when the ram block is unbolted no trapped coolant should fall on the card, in theory.


MadDogMe

To do it as you say was my original intention, but if you look at the R9800 pro layout there is a one of those graphite copper coil things you mention right where the bottom finger part of the cooler wants to be. To avoid this the front ram block has the underside of this finger part milled out, which would have made it near impossible for it to be a channel as well. The ram chips although very hot to touch naked, are only giving off a small amount of heat. A trickle of flow will be enough to cool them as they are in contact with a largish copper block. There is good flow through both however, touching the backside ram block it feels cold, (16C coolant temp in a 25C ambient atm). The cooler is heavy but I haven't made any extra support. It would be a good idea if the PC was moved about though, there is a hole in the card that could be used for a simple brace.

Best to read the entire topic I linked you to previously for more info about the Radeon 9700 pro cooling as I removed the plate in doing the ram v mod.

Gorgeous once again ! :dome:

I got inundated with pop ups at that site, even had the damn 'auto install' Gator one! :eek: ...

*Edit* strange!, got none this time?. Could'nt find any mention on how to remove the plate though :shrug: ...

from the Rage3D topic
posted by BladeRunner