GPU/NB Water block (development thread) - Page 5

Blackeagle · 11-10-2003, 08:52 AM

I don't have exact data Tex707. But that 30X stronger wasn't just pulled from the air. At McMaster-Carr's site which sells both that is how they describe Lexan vs acrylic, although they also say that standard Lexan is easier to scratch than acrylic. But there is a sort of higher grade Lexan with a coating called mar-gaurd (MR-10) that makes Lexan 10X more scratch resistent than is the standard Lexan.......for a few $$ more of course.

EDit:

There is also a bullet resistent grade of Lexan made where the Lexan is blended with a couple other materials. While mar-gaurd is a few $$ more that stuff is $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$.

But McMasters sells it as well in a .78 & 1.5" thickness.

www.mcmaster.com

tex707 · 11-10-2003, 10:58 AM

Well, my almost-impossible-to-be machined GPU waterblock is finished. Despite the fact that almost identical results (maybe even better) could be accomplished wit a simple acrylic top and two angled barbs, I just had to share this over-complicated development of Jaydee’s design with you guys since I have spent quite a few hours working on it.

If anyone wants to have some details enlightened, I’ll be happy to explain, with cross-section drafting and all…

jaydee · 11-10-2003, 11:38 AM

That looks good tex!

BE I have no clue about the Radeons. Not sure if all the manufacturers hold tight to ATI's reference either. Might be best to just get the card and measure it up before installing it. Some of those cards had that raised spacer around the GPU. I think they got that fixed though.

My GF4 MMX look pretty weird. it has 4 holes around the GPU and then 2 mounting holes that the current HS is using. tex sent me a IGES file of a GF2 but it doesn't look anything like mine.

I thinkI am going to use the 4 holes around the GPU insead of the standard mounting holes anyway. That should give superior mounting pressure.

Also note that card is now installed in my Antec SX830 with the Asus A7V8X-X. One thing ASUS did nicely on that board was rotate the CPU socket 90 degrees so hose routing will be easier. Going to look pretty good with my KingPin CPU block, King Pin GPU block, and King Pin North Bridge block.

Going to use clear tops on all of them as of now. Might go with Copper top and solder on some 45's for the GPU though. I also have another plan for the GPU's in/out's which I will try and draw up when I get a chance.

bigben2k · 11-10-2003, 12:23 PM

FWIW...

I like that last design too, Tex.

If I can suggest... First, making round pins is unecessarily difficult. Given that, and that the flow is incoming at 45 degrees, you might consider making square pins, that sit at a 45 degree angle from the base, pointing towards the barbs.

The main benefit is that you'll now have a base with twice the surface area. You can play with the angle, and the remaining baseplate thickness, to optimize the whole thing.

jaydee · 11-10-2003, 03:58 PM

Quote:

Originally posted by bigben2k
FWIW...

I like that last design too, Tex.

If I can suggest... First, making round pins is unecessarily difficult. Given that, and that the flow is incoming at 45 degrees, you might consider making square pins, that sit at a 45 degree angle from the base, pointing towards the barbs.

The main benefit is that you'll now have a base with twice the surface area. You can play with the angle, and the remaining baseplate thickness, to optimize the whole thing.

Not sure if you caught it but his pins are not round. They have mutiple sides which adds alot of extra surface area. I think his design would be MUCH better for one of the newer Radeons or GF FX's as they are in the 50+ watt range now. I think this is sever overkill for a north bridge cooler. But who complains of overkill around here.

Anyway here is were I stand. I got the ASUS A7V8X-X with the GF 440 MX from leadtek. I need a CPU cooler (done), GPU cooler, and north bridge cooler.

The design for the north bridge is done, just needs milling. The GPU however is going to need a whole new design around the 4 holes on the card. When I get home I will measure it all up and start on it. it will also need some kind of angled barbs being it WILL be installed in my Antec SX830 case. I have an idea for this already, but need the card in hand to varify it will work. Will try and work on that tonight some.

I was going to do some milling this weekend but have to change plans. Now I am going to gear up for ThanksGiving as I should have 3 days to mill stuff. This is good though because I have a LOT of tool path and G-Code editing to do and doubt I could get it done by this weekend anyway.

These blocks are going to be installed inline. Pump>Rad>CPU>NB>GPU>Res/airtrap.

Had the daughter this weekend and spent most of my time with her so I didn't get much done.

tex707 · 11-10-2003, 05:58 PM

Quote:

Originally posted by jaydee116

<snip>
My GF4 MMX look pretty weird. it has 4 holes around the GPU and then 2 mounting holes that the current HS is using. tex sent me a IGES file of a GF2 but it doesn't look anything like mine. I thinkI am going to use the 4 holes around the GPU insead of the standard mounting holes anyway. That should give superior mounting pressure.
<snip>

Do you intend to use some kind of backplate to take the load off the video card PCB?

tex707 · 11-10-2003, 06:04 PM

Quote:

Originally posted by bigben2k
FWIW...

I like that last design too, Tex.

If I can suggest... First, making round pins is unecessarily difficult. Given that, and that the flow is incoming at 45 degrees, you might consider making square pins, that sit at a 45 degree angle from the base, pointing towards the barbs.

The main benefit is that you'll now have a base with twice the surface area. You can play with the angle, and the remaining baseplate thickness, to optimize the whole thing.

Glad you like the design...

The pins are, actually, hexagonal, not round, but it is not so visible from those screenshots. The flow is not incoming at 45 degrees, but closer to 90...please see the cross-section...

I have attached a model of 45-angled pins...is this close to your suggestion?

tex707 · 11-10-2003, 06:07 PM

I've forgotten to attach the screenshot...

jaydee · 11-10-2003, 06:07 PM

Quote:

Originally posted by tex707
Do you intend to use some kind of backplate to take the load off the video card PCB?

If two holes spread much father apart don't need it then I don't think 4 holes close together will need it. But, it does give me an idea. The backside of the cards get's damned hot under the GPU. I could make a backing plate that doubles as a 60mm fan mount.

jaydee · 11-10-2003, 06:14 PM

Quote:

Originally posted by tex707
I've forgotten to attach the screenshot...

Pretty sure he ment like the pic attached by taking the square pin and making it a diamond. This would not be required with your current design though.

tex707 · 11-10-2003, 06:16 PM

Quote:

Originally posted by jaydee116
If two holes spread much father apart don't need it then I don't think 4 holes close together will need it. But, it does give me an idea. The backside of the cards get's damned hot under the GPU. I could make a backing plate that doubles as a 60mm fan mount.

Actually, if the GPU W/B is properly attached the backside under the GPU is going to be cold...

I use the backside clamp made of brass...

Blackeagle · 11-10-2003, 07:18 PM

Quote:

Originally posted by jaydee116
If two holes spread much father apart don't need it then I don't think 4 holes close together will need it. But, it does give me an idea. The backside of the cards get's damned hot under the GPU. I could make a backing plate that doubles as a 60mm fan mount.

A very good idea Jaydee!

The plate can be open in the center and hold the fan just a bit off the card, say 3/8" or so. A low noise fan used here would be enough to move the needed air and you'd not hear it.

jaydee · 11-10-2003, 07:28 PM

Quote:

Originally posted by Blackeagle
A very good idea Jaydee!

The plate can be open in the center and hold the fan just a bit off the card, say 3/8" or so. A low noise fan used here would be enough to move the needed air and you'd not hear it.

As tex mentioned it will not matter much on my card, but I think on newer cards it will. The newer Radeons and FX's are suposed to put out 50-75watts. No regular GPU block is going to keep the backside cool. Not even the best CPU blocks can keep the backside cool on CPU. I might forget the idea this round, but when I get my Radeon 9600 or better that will change. Not planing on a vid card upgrade anytime soon though. This little GF440 MX handles all my games peachy and I havn't even attempted overclocking it yet. Probably the best $40 I ever spent on a vid card. Once the 9600 drops under a $100 I will be eye balling it.

Blackeagle · 11-10-2003, 07:40 PM

JD,

Yeah, quite right about the currant card you have. But the instant I saw the idea I was thinking in terms of a 9800 Pro most of all if O/Ced.

Tex707,

I don't see why that would be impossible. But what if the hole being drilled from the water chamber were drilled on a drill press with a tiltable table. Or could be done by clamping the Lexan in a drill vise. Then you could drill the hole on a slant and avoid the 90 degree corner. Instead it would be say 65-70 degrees, saving a bit of flow for you while still allowing you to have the side mounted barbs you are making use of. Anything to avoid 90 degree bends!

jaydee · 11-10-2003, 07:47 PM

Quote:

Originally posted by Blackeagle

I don't see why that would be impossible. But what if the hole being drilled from the water chamber were drilled on a drill press with a tiltable table. Or could be done by clamping the Lexan in a drill vise. Then you could drill the hole on a slant and avoid the 90 degree corner. Instead it would be say 65-70 degrees, saving a bit of flow for you while still allowing you to have the side mounted barbs you are making use of. Anything to avoid 90 degree bends!

Use a drill press with a tilting vice and a ball nosed endmill and you golden.

Blackeagle · 11-10-2003, 09:29 PM

Quote:

Originally posted by jaydee116
Use a drill press with a tilting vice and a ball nosed endmill and you golden.

Ah, I didn't think about use of a endmill in a drill press. A mill has to make things go easier, and better.

I just was doing some work on a Lexan top, I'm sorry to say it didn't survive. Fence sliped, and the router took a bite where it wasn't desirable to have it.

while I'm going to wait to see just what the size is on a 9800, I've done a small bit of work today. Took off part of the excess pins and filed them down smooth. Decided to make a practice top to learn with.

Did get the feel of routing the Lexan, and also learned to set the fence tighter.:shrug:

jaydee · 11-10-2003, 09:33 PM

Quote:

Originally posted by Blackeagle
Ah, I didn't think about use of a endmill in a drill press. A mill has to make things go easier, and better.

A ball nosed endmill will make life a lot easier tryingto drill at an angle into anything. I drill bill will not like trying to go into material at an angle as it isn't designed to.

jaydee · 11-11-2003, 12:42 AM

Here is the layout of the 4 holes around the GPU on my Leadtek GF4 440MX and the overall demensions I wantto use. Going to be a tight fit....

jaydee · 11-12-2003, 12:42 AM

Here are some pics of my vid card and the mounting situation.

The mounting holes.

Full view of card.

Stock heat sink. The little circled milled can only be for a thermal probe. it fits just slightly over the GPU. Can think of no other reason it is there.

The case this will all fit into when done.

Notice in the above pics the Aluminum HS on the north bridge. That came off my dead Asus A7V333. Now one would think Asus would at least keep the same cooling for a NEWER board eh? Well this is what the stock HS was on the A7V8X-X. And no, there was NO fan on it.

jaydee · 11-13-2003, 01:59 PM

This is what I have so far on the GPU bock. It is a very tight fit if I use those 4 holes. Even with the smaller area and less pins there is still plenty of surface area as the GPU is smaller than the NB. Only problem I see is the inlet and outlet. I plan on using a Copper top and soldering pipe that bends around the card and up in the direction of the NB block so I think I will be fine there.

The CPU and NB blocks will have clear tops. The CPU block is already done as per the other thread.

So what I have left to do is confirm the mounting holes for both the GPU and Asus NB block. I should be able to do that tomorrow on the laser. Just the same as I did with the standard NB block.

I setup the test bench last night but it failed the leak test.

I need a couple more hose clams in one spot. that 3/4" tubing just dopesn't such up as much as the smaller sizes. Especially with a 500GPH pump. Once that it sealed up I will start testing the King Pin against the Maze 4 and the Lemon Block Cu. Should be doing that all next week.

EDIT: For got the pic.

jaydee · 11-13-2003, 04:30 PM

I ordered the copper and the endmills today. The copper should be here in a few days but not sure about the endmills. Never ordered anything from this company but they had 3 flute 1/16" endmills! I even got a 3 flute 1/16" Ball nosed endmill. Also picked up 2 3-flute 1/8" endmills for the actual milling of these blocks. They were under $5ea. to boot!

The 1/16" is for the O-ring groove on the GPU block and some other projects in the design stage.

Hopefully the endmills will get here before ThanksGiving.

Socko · 11-13-2003, 04:48 PM

Where did you order from?

jaydee · 11-13-2003, 05:39 PM

Quote:

Originally posted by Socko
Where did you order from?

The copper I got from www.onlinemetals.com. they are only 300miles from me and stuff gets here quick.

The endmills I order from http://www.cetdirect.com. The prices suprized me. Usually places like this charge a small fortune for tooling they make themselfs. Hope I get them and hope they are of decent quality.

MadDogMe · 11-17-2003, 10:59 AM

Quote:

Originally posted by jaydee116
heat is spread out over the GPU and NB. If it wasn't then there would be no reason to make the chipset as big as it is. Every section of the chip has parts underneath it and they all make heat.

Might be a bit late to influence design, but: The chip is the size it is to acomadate the pins that fix it to the mobo. Every chipset I've lapped (GPU's & NB's) have been about 10mm x 12mm in the centre of the plastic(usualy, or Cu~GPU) 'chip'. You can see a depression from the actual chip in the plastic(or Cu on GF/Ti's!!) when lapping. That is where the heat is produced. It looks like a CPU package with the chip in the middle and the pins around it, but not under it...

*****************************************

I'd do something like this. The central part would be continous fins(not three rows) and would be 'more suitably proportioned'

...

Code:

________________________
|                      |
|        INLET         |
|                      |
|                      |
||||||||||||||||||||||||
||||||||||||||||||||||||        
||||||||||||||||||||||||
|                      |
|                      |
|       OUTLET         |
|______________________|

Edited for 'CODE' only twelve times! new low record!!...

jaydee · 11-17-2003, 11:51 AM

I still have yet to measure the heat of a NB. All my probes are in use. If there are pins around the entire body then there will be heat anyway. Pins carry heat. I will agree though, most of the heat is in the middle. On the KT133A it has a metal circle in the middle and I assume that is where most the work is done. That circle also appears on the GF4 I have as per the pics above. Interestingly the stock HS was NOT covering even half of the GPU! The thermal compund marks where non existant there. Attribute that to the mounting holes being way to far apart and weak mounting system. Those little plastic spring loaded pins are near useless.

11-10-2003, 08:52 AM	#101
Blackeagle Thermophile Join Date: Oct 2002 Location: U.S.A = Michigan Posts: 1,243	I don't have exact data Tex707. But that 30X stronger wasn't just pulled from the air. At McMaster-Carr's site which sells both that is how they describe Lexan vs acrylic, although they also say that standard Lexan is easier to scratch than acrylic. But there is a sort of higher grade Lexan with a coating called mar-gaurd (MR-10) that makes Lexan 10X more scratch resistent than is the standard Lexan.......for a few $$ more of course. EDit: There is also a bullet resistent grade of Lexan made where the Lexan is blended with a couple other materials. While mar-gaurd is a few $$ more that stuff is $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ $$$$$$. But McMasters sells it as well in a .78 & 1.5" thickness. www.mcmaster.com Last edited by Blackeagle; 11-10-2003 at 09:00 AM.

11-10-2003, 12:23 PM	#104
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	FWIW... I like that last design too, Tex. If I can suggest... First, making round pins is unecessarily difficult. Given that, and that the flow is incoming at 45 degrees, you might consider making square pins, that sit at a 45 degree angle from the base, pointing towards the barbs. The main benefit is that you'll now have a base with twice the surface area. You can play with the angle, and the remaining baseplate thickness, to optimize the whole thing. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

11-13-2003, 01:59 PM	#120
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	This is what I have so far on the GPU bock. It is a very tight fit if I use those 4 holes. Even with the smaller area and less pins there is still plenty of surface area as the GPU is smaller than the NB. Only problem I see is the inlet and outlet. I plan on using a Copper top and soldering pipe that bends around the card and up in the direction of the NB block so I think I will be fine there. The CPU and NB blocks will have clear tops. The CPU block is already done as per the other thread. So what I have left to do is confirm the mounting holes for both the GPU and Asus NB block. I should be able to do that tomorrow on the laser. Just the same as I did with the standard NB block. I setup the test bench last night but it failed the leak test. I need a couple more hose clams in one spot. that 3/4" tubing just dopesn't such up as much as the smaller sizes. Especially with a 500GPH pump. Once that it sealed up I will start testing the King Pin against the Maze 4 and the Lemon Block Cu. Should be doing that all next week. EDIT: For got the pic. Last edited by jaydee116; 11-13-2003 at 02:07 PM.

11-10-2003, 11:38 AM	#103
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	That looks good tex! BE I have no clue about the Radeons. Not sure if all the manufacturers hold tight to ATI's reference either. Might be best to just get the card and measure it up before installing it. Some of those cards had that raised spacer around the GPU. I think they got that fixed though. My GF4 MMX look pretty weird. it has 4 holes around the GPU and then 2 mounting holes that the current HS is using. tex sent me a IGES file of a GF2 but it doesn't look anything like mine. I thinkI am going to use the 4 holes around the GPU insead of the standard mounting holes anyway. That should give superior mounting pressure. Also note that card is now installed in my Antec SX830 with the Asus A7V8X-X. One thing ASUS did nicely on that board was rotate the CPU socket 90 degrees so hose routing will be easier. Going to look pretty good with my KingPin CPU block, King Pin GPU block, and King Pin North Bridge block. Going to use clear tops on all of them as of now. Might go with Copper top and solder on some 45's for the GPU though. I also have another plan for the GPU's in/out's which I will try and draw up when I get a chance.

11-10-2003, 07:40 PM	#114
Blackeagle Thermophile Join Date: Oct 2002 Location: U.S.A = Michigan Posts: 1,243	JD, Yeah, quite right about the currant card you have. But the instant I saw the idea I was thinking in terms of a 9800 Pro most of all if O/Ced. Tex707, I don't see why that would be impossible. But what if the hole being drilled from the water chamber were drilled on a drill press with a tiltable table. Or could be done by clamping the Lexan in a drill vise. Then you could drill the hole on a slant and avoid the 90 degree corner. Instead it would be say 65-70 degrees, saving a bit of flow for you while still allowing you to have the side mounted barbs you are making use of. Anything to avoid 90 degree bends!

11-12-2003, 12:42 AM	#119
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	Here are some pics of my vid card and the mounting situation. The mounting holes. Full view of card. Stock heat sink. The little circled milled can only be for a thermal probe. it fits just slightly over the GPU. Can think of no other reason it is there. The case this will all fit into when done. Notice in the above pics the Aluminum HS on the north bridge. That came off my dead Asus A7V333. Now one would think Asus would at least keep the same cooling for a NEWER board eh? Well this is what the stock HS was on the A7V8X-X. And no, there was NO fan on it.

11-13-2003, 04:30 PM	#121
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	I ordered the copper and the endmills today. The copper should be here in a few days but not sure about the endmills. Never ordered anything from this company but they had 3 flute 1/16" endmills! I even got a 3 flute 1/16" Ball nosed endmill. Also picked up 2 3-flute 1/8" endmills for the actual milling of these blocks. They were under $5ea. to boot! The 1/16" is for the O-ring groove on the GPU block and some other projects in the design stage. Hopefully the endmills will get here before ThanksGiving.

11-13-2003, 04:48 PM	#122
Socko Cooling Neophyte Join Date: Nov 2003 Location: Vineland, NJ Posts: 19	Where did you order from?

11-17-2003, 11:51 AM	#125
jaydee Put up or Shut Up Join Date: Dec 2001 Location: Spokane WA Posts: 6,506	I still have yet to measure the heat of a NB. All my probes are in use. If there are pins around the entire body then there will be heat anyway. Pins carry heat. I will agree though, most of the heat is in the middle. On the KT133A it has a metal circle in the middle and I assume that is where most the work is done. That circle also appears on the GF4 I have as per the pics above. Interestingly the stock HS was NOT covering even half of the GPU! The thermal compund marks where non existant there. Attribute that to the mounting holes being way to far apart and weak mounting system. Those little plastic spring loaded pins are near useless.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)