when I read that I had visions of a 9700 AIW with lots of holes in it :p

Only thing I've noticed over a long period is the cards pcb will warp a small amount. extra supports would be advisable if you were to move the pc around a lot

G33k,

Yeah, I have thought about it... not a light solution... but as light as I can make it.

BladeRunner,

Whoops... meant "I plan on lightning up the blocks too... " :eek: :D

hmm... didn't think about the board warping... I may shave more off the blocks than just drilling holes then. I'll have to do some sketches to see what looks good and is still VERY light.

I guess my main concern with lightening up the blocks too much is the copper might bend when clamped to the board... although that might be a good thing (solid contact with ram and GPU). :shrug:

I think you'd have to make it pretty damn thin for it to bend on mounting it, and there shouldn't need to be big clamping loads anyway. Bending during construction / assembly is more of a possibility with making it thin. The board warpage isn't an issue. I'm not sure of your cards final weight when finished but my 9800 pro cooler is far from light, and it's been in my tower for the best part of a year unsupported.

On removing it recently the card is a little warped now (I've fitted the cooler to the Asus 9800xt), but to be honest so what... it still works fine :) put it this way unless you were told it was warped I doubt you'd notice it. The main problem with card weight is moving the case. If you move or bang the case about you could put significantly more loads on the card and agp slot, unless you have additional support.

ummm yea i have issues with my Ti4600 bending slightly even though its is supported ALSO in the corner (the one you would find the power connector on the 9800 for example)

i think for the next solution i am going to make an aluminum braket to support the card.

BladeRunner,

Thanks for the info... I may see just how lightweight I can make this then. :D

I will see how far it bends with the plates clamped on the card (before I solder the cover plates on) and see if I need to lighten it up, but I am kind of leaning toward lightening the blocks up now that my mind has gotten up and running on the idea. :dome:

BladeRunner,

How about trimming this...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0174.jpg

to this...

http://pages.sbcglobal.net/mmz_tl_01...ionTrimmed.jpg

Whadda yah think? :shrug: :confused: :cool: :p :dome:

It will still cover the RAM and GPU (including shim). The back side would be similar except for the square area around the GPU.

NOTE: originally the design was roughly like this...

http://forums.procooling.com/vbb/att...achmentid=2764

Looks good to me :) I was about to suggest that you round off the bottom left corner... until I remembered what shape a GPU is! *slaps forehead*

Well... I took BladeRunner's advice and devised the lighter weight design above... here you see the template glued to the back plate and the milling of the surface to RAM height has begun. I had already drilled and milled the recesses for the various SMCs that stick up.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0194.jpg

Here I've milled the surface to 0.058" depth for the RAM and left a small square for the back of the GPU area.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0195.jpg

Here I've begun milling out the rough shape of the block... did you bring your shovel? :p

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0196.jpg

Two corners milled away and one just short so I could still use it for an anchor...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0197.jpg

Here's the roughed shape next to the back of the card... the side of the block you are seeing will fit on the side of the card you are seeing. Again, note the cooling square for the back of the GPU... :D

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0199.jpg

That's it for today... I will get the final shape milled tomorrow and begin roughing in the front block. Cheers! :cool:

looking like something good now ;) , its great when you start to see the design concept form into a real item.

The terms "light" & "copper" used together is kind of an oxymoron, but I always try to now approach my builds from the make it as light as I possibly can direction. It isn't easy unless you are willing to use Perspex / acrylic etc.

Only other thing I can see, (and don't jump on me if you have it hand), is you'd ideally want something on the square section that will touch the pcb opposite the core, mica shim possibly

Looking good MMZ... Can't wait to see the finished piece! I had a load of suggestions for you, but Bladerunner got there first (again).

@Bladerunner : Perhaps "Lighter" / "Heavier" would be more appropriate? :) I was at a mate's house the other day and saw one of the first WB's I made, and my GOD it's heavy! 's quite amazing how far we've come in just a few years... *start reminiscing (sp?)*

BladeRunner,

Yeah, "Copper" and "Lighter" don't necessarily go together... as you say... "as light as possible" is my goal. :dome:

As for the pad on the back side of the GPU area... I was planning a layer of Arctic Silver II (I don't have any Arctic Silver III). My understanding is that ASII and ASIII are not conductive. Am I mistaken? :shrug: :confused:

G33k,

I too pulled one of my first CPU blocks (simple drilled passage block) and it's defintely heavier than my current impengement CPU block. :eek: :D

When u referd to the back, is that the back of that part of the block, or the back block of the CPU? Because if it is the back part of the block i can see a potential major design floor. The big gap over the #rotor style cooling for the GPU will give a different path for the water to travel. This will reduce the cooling effect, as not much coolant will be travelling close to the core.

P.S., u will have to stop with the hugging, my girlfriend will get jelous, and im sure ur wife wouldnt b to happy either. Probly think we have some kind of afair going on, lol.

Pears0,

Back side of the back block... meaning there are two blocks. The front block (for the front of the video card) has the #Rotor area. The back block (for the back of the video card) has a simple clear path with a pad for contact on the BACK SIDE of the GPU.

So the pattern I posted is for the FRONT of the FRONT BLOCK (faces away from the front of the video card)

http://pages.sbcglobal.net/mmz_tl_01...ionTrimmed.jpg

and this pattern is the BACK of the BACK BLOCK (faces the back of the video card).

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0194.jpg

Hope that somewhat clarifies it... :shrug:

To clarify ... and because I haven't posted them yet...

Here is the full design for the front block...

http://pages.sbcglobal.net/mmz_tl_01...ontTrimmed.jpg

Here is the full design for the back block...

http://pages.sbcglobal.net/mmz_tl_01...ackTrimmed.jpg

Hope that helps someone with their design... :dome:

that looks great, i get here every day to see if there are any new pics :)

AS2/3/5 is not conductive but is capacitive. You will get problems if you put it on traces!
Id use a mica shim or similar, infact you oculd just glue a bit of paper on the back much like how you used the templates in making the blocks. You wont get much heat atall from the back so its not worth cooling it - With my 9700np core at 470/1.75v its still very close to ambient (though Ive only measured with my fingers...)

Also has your card got a heatspreader on the back? This will get very hot (though very low amount of heat) so i suggest you cool that too, change it, theres only one chip that gets hot. It doesnt need w/c, but Bladerunner did ono his...

I like the shape of the skimmed down block set, very nice. having a rotary table helps there I'd find that very had to do on a straight x-y

All I can think of on that is it may be a little weak in the areas shown below until the lid is soldered on (I assume it's being soldered on?)

http://www.zfz.com/pictures/temp/9500wc.jpg

Artic silver is not conductive but it is capacitive, (see their site for more details). I'd be a little concerned having the copper touching the pcb directly without some type of isolator / insulator other than paste. Mica shims are perfect for this as they also have reasonable heat transfer properties. Any radio shack should have some in stock you can cut to size. You'd need to allow for its compressed thickness though.

Not a very good picture but this is an extreme close up of the mica shim in place under my gpu block retainer on the Asus 9800xt to give an idea.

http://www.zfz.com/pictures/jpg/vga/...9800wc_005.jpg

I had to modify the retainer because there were extra components in this area on the Asus, compared to the previous 9800 pro card :rolleyes:

got beaten to the reply by |kbn| :) one other thing do any of the tv tuner specifics get hot....... never had a tv card my self.

I put a water block on the plate as it was still getting pretty hot with a zalman nb block thermal pasted to the spreader. It wasn't really required, but i was pushing the voltage and it was in a fanless enviroment. Hard to be 100% sure as i was trying other things at the time, but i don't think watercooling the spreader helped the vmods / o/c at all.

:p

I got about 8mhz from cooling it on mem and about 30 on core. Mine was getting so hot before that the plastic holding it on had melted and it had slightly brown'ed.
Ive just made a heatsink for the chip thats the perfect size 13mm * 8 mm base made from soldered copper sheet. It has 15 copper fins 45mm long which can be bent out to get mroe air to them. Will try to get a pic of it later once Ive cleaned it up.

Both the tv cards ive had get very hot, infact cooling the tuner chip allowed me to almost watch cable channels whih were previosly scrambled (no decoder box)

interesting.. I thought that tv part would get hot.

I wasn't clear before, but I did see results on cooling the spreder plate, just none was obvious over the zalman nb sink to waterblock...but actyally cooling it helped my ram alot over it's stock cooling finger burning temps

more on it in this old topic Here

|kbn|,

Thanks for the tip... but my card is in a tower and so the back sides of these cards DO soak heat over time and I want to keep the GPU area as cool as possible.

BladeRunner,

I think I have some mica shims from other electronics around here somewhere.

As to the weak points... Yes, I will be using the "Oven Technique" for that after everything is fitted. In addition, I actually want some minor flexing as it will help the card conform to variances in the RAM GPU height from what I measured/milled.

I plan on Arctic Silver II for all RAM and GPU contact areas, but now that I know... I will NOT use it on traces.

Thanks again to both of you for that bit of info..

On a side note... I picked up some Solid Carbide endmills today. It will be fun to see the performance difference that JFetting's always talking about. :D

As to the TV Tuner... I won't be utilizing that part.

I doubt it will flex much when finished at safe claming levels, I was talking more about unwanted flex / twist that, milling, clamping, heat, distortion, dropping, etc could fairly easily cause by accident until the lid is soldered on.

I use AS5 on the gpu and regular white thermal past on the ram chips, as I don't want AS getting under those bga chips.

Is the TV part not on all the time the card is powered creating heat then?

Please tell me you are going to put in the extra effort to make this block look like a mirror and lacquer it... be a crime not to :D

*a tip from experience, full card coolers are complex things to get exactly right & about now is the time to step back from it all, recheck your measurements, etc. Take a deep breath and slow it all down a few notches and make sure your chi is aligned :p . About now is when I often screw something basic up, like snap a tap or mill a little to far in one direction because I can see the finish post. This turns the job into an annoying "how can I recover from my cock-up" venture :cry:

BladeRunner,

Actually I have slowed down... I already snapped drills on this project... that's when I decided to slow it down a notch. I am being EXTRA careful about measuring THREE times and cutting once for the RAM and GPU depth cuts as these are the ones that are CRITICAL to this working right.

Not sure if I'll do a lot of polishing, but I probably will as this is by far the nicest/most complicated video card solution I've ever attempted. :eek: :drool: :p

The carbide endmills seem to work a bit better... but I guess when you work on such an abrasive metal (like copper) you just don't get much better without MAJOR RPM on the spindle.

At this point I had glued the last template for this block on and trimmed up some and am just setting the center for the rotary table so I can mill the curve part.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0200.jpg

This is almost finished as far as rough shaping goes...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0201.jpg

Most of the edges are in a close to finished condition. I did skip cutting that little notch on the left near the smaller circle target.

Here the back block has been attached to the unfinished front block to verifiy alignment of cap screw holes as well as water crossover passage.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0202.jpg

The front block has had it's BACK side template glued on and it is secured to the rotary table.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0203.jpg

Last thing for the day was to drill and mill out the keepouts for the SMCs that are at GPU height on the card(depth for milling out of the block).

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0204.jpg

Tomorrow I'm going to mill away the outline of the block shape first. THEN I will mill the depth for RAM and GPU... that way I won't have to try and line up templates front to back.

That's it folks... L8R all... :dome:

You're gonna have to stop with these Juicy updates MMZ - my work is suffering! :) Looks like this is going to be a very very nice looking block - can't wait for tomorrow's update :D

...I just wish I had a mill like yours!! :(

Looks like my entire page six post got eaten by the ProForums' system crash :cry: :mad:

I will attempt to remember what I wrote...

Here I've milled away most of the outside of the pattern... rough, but it will be finished up after the cover plates are soldered on.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0205.jpg

Here I'm preparing to drill the pilot hole for milling out the GPU area.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0206.jpg

GPU area is now milled to depth... only checking the width and length to do.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0207.jpg

Here I've measured the width and length and it was a bit small... so I made a few passes around the perimeter of the GPU area and took a few thousandths off each pass.

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0209.jpg

Once the size was correct on that... I began rough milling to remove the large corner section... I LOVE roughing endmills... :D

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0210.jpg

Here I've roughed out the smaller area to be removed... did I mention I LOVE roughing endmills? :D :p :dome: :drool:

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0211.jpg

Now that the overall shape is complete, I then milled down the RAM areas like so...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0212.jpg

After a few "adjustment" passes on the mill and several test fittings... we have what is beginning to resemble a full water cooling solution... ;)

Front side...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0213.jpg

Back side...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0214.jpg

Here are the blocks alone as a full unit after some edge trimming...

Front side...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0215.jpg

Back side...

http://pages.sbcglobal.net/mmz_tl_01...s/Dscn0216.jpg

That's all I can remember... Hope to get the cover plates soldered on tonight.

Cheers!