Simple GPU block - improvements?

|kbn| · 03-13-2005, 05:27 PM

As I now have a x850XTPE

I thought Id watercool my gpu again

Ive done two simple GPU block before. The first was very simple, cut up T reducer pipe soldered to a base - resonable performance for the price/effort it took.

The next was just modified.. and it killed all flow in my system.. probably one of the mot restrictive blocks ever made (less than 10mm2 cross sectional area).

I will be using the same top as before, as shown in the first attached pic.

I have decided to use a middle plate, essentially just a copper frame... the middle is cut out (20*40mm). It is 4mm thick. This gives me 20*4 = 80mm2 cross section, so there hould be little restriction (the barbs are 11mm ID which is a bigger restriction). The middle plate was not cut from stock, but instead sintered.. because im lazy

The base, so far, is flat - well no .. its not...it needs a lot of lapping.... - but its size is:
60*38*4.8mm

I am thinking of making quite shallow (maybe 2 or 3 mm deep) holes in the wet side, but I wont want to waste effort if you all think it wont help atall. So thats the point in this thread.

Alternativly I have the option of using the base in the second pic.
It is made of silver, at the thinnest point in the channels about 2.5mm thick. The problem with this, is I dont think much moving water will enter the very small channels becuase they are restrictive (<10mm2 C.S.A) and the conduction of copper is better than water..
If I use this, then the only differnce from my second block attempt will be the 4mm midleplate to make it less restrictive.

Block will all be soldered together hopefully tomorrow.. and I will do some flow rate testing with it (not pro/c standards... but then I dont have £80 for a proper flowmeter.. suggestions welcomed though

)

Moto7451 · 03-19-2005, 02:59 PM

The Silver base will probably be an improvement. If you can, try getting the base to be a little bit thinner by making the channels deeper. That will help flow & a thinner base generally means better performance.

FL3JM · 03-19-2005, 04:08 PM

I do not agree that you should make your bp thinner, silver has superior thermal conductivity and thus you can have a thicker baseplate than lets say a copper one. A thicker baseplate means that the heat will be distributed over a larger surface area, though at some point it will begin to insulate the heat rather than cool it. The optimal baseplate thickness depends on the design of the block ofcourse and the material used, since your design does not rely on some advanced jet impingement technique and it hasn't got a central placed inlet, the heat needs to be more evenly distributed across the baseplate, thus a thicker baseplate is prefered.

Though i must agree that the bp you are currently using is kinda thin overall, the shallow channels you have will make the block quite restrictive, though if you have a crap flow in your system allready you won't notice any differance.

Please correct me if am totally wrong you thermal gurus.

Maybe silver = even more thinner bp? O_o?

|kbn| · 03-22-2005, 01:05 PM

Decided to ue the silver base, 4mm spacer middle, and modified the top.

The silver base is currently the same as in the pic. I think I would be sacrificing mechanical strength if I made it much thinner - silver is very soft... Also I think BP thickness depends a huuge amount on the chip and block design, and also I think it would take many trials to get it just right..
I will however cut between the remaining channels as water will not move in them if there is a 4mm gap above them in whihc water can move freely...

The thermal conductivity of silver over copper is not a noticable quantity, I think the bp is about right.

The block will not be restricitive becuase as I said there will be a 4mm middle spacer, which is 40*20 ID.

For the top, I took out the exitisting pipe/barbs as they are a pita the way I originally done them (using copper wire to make the difference between 12mm and 15mm pipe.
I was going to do them the way jaydee116 does his, with end-soldered joints, but as Im uing elbows, I doubted the strength.. also I think he used braize and not soft solder.

So instead, I used a hand drill, and a 16mm FLAT wood bit (this type - which gave be REALLY sore hands with several blisters..!) to drill two 16mm holes.
Then I braized the elbow in and it never coming out, and its never going to leak.

My flowrate is very high imo (see link above), havent tested this yet but I expect the full system rate to not be much lower than the test with the rads. I mnight replace the rads later on if I think it will improve flowrates.. I wantto put them in parrellel anyway..

Moto7451 · 03-22-2005, 08:38 PM

I use a hand drill & a Flex Shaft drill (think of a dentist drill on steroids, big 120v AC .5A motor that spins at 14,000 RPM) for my blocks.

Right now I'm working on a "pitted" block for my GPU which simply has rows & rows of overlapping drill holes to create turbulence. It almost looks like someone has chewed on it with their molars

. I follow the school of thought that a thin base means you can get heat to the water faster & it wont "insulate" the heat so much. The thicker base school of thought is valid too, the designs end up being different often times. I figure with a GPU like my RV350, thinner is a good thing. Perhaps it isn't but I'll find out by building blocks now wont I. Good luck with your design.

03-22-2005, 01:05 PM	#4
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	Decided to ue the silver base, 4mm spacer middle, and modified the top. The silver base is currently the same as in the pic. I think I would be sacrificing mechanical strength if I made it much thinner - silver is very soft... Also I think BP thickness depends a huuge amount on the chip and block design, and also I think it would take many trials to get it just right.. I will however cut between the remaining channels as water will not move in them if there is a 4mm gap above them in whihc water can move freely... The thermal conductivity of silver over copper is not a noticable quantity, I think the bp is about right. The block will not be restricitive becuase as I said there will be a 4mm middle spacer, which is 40*20 ID. For the top, I took out the exitisting pipe/barbs as they are a pita the way I originally done them (using copper wire to make the difference between 12mm and 15mm pipe. I was going to do them the way jaydee116 does his, with end-soldered joints, but as Im uing elbows, I doubted the strength.. also I think he used braize and not soft solder. So instead, I used a hand drill, and a 16mm FLAT wood bit (this type - which gave be REALLY sore hands with several blisters..!) to drill two 16mm holes. Then I braized the elbow in and it never coming out, and its never going to leak. My flowrate is very high imo (see link above), havent tested this yet but I expect the full system rate to not be much lower than the test with the rads. I mnight replace the rads later on if I think it will improve flowrates.. I wantto put them in parrellel anyway.. __________________ Waterblocks

03-19-2005, 02:59 PM	#2
Moto7451 Cooling Neophyte Join Date: Jan 2005 Location: LA, CA Posts: 53	The Silver base will probably be an improvement. If you can, try getting the base to be a little bit thinner by making the channels deeper. That will help flow & a thinner base generally means better performance.

03-19-2005, 04:08 PM	#3
FL3JM Cooling Neophyte Join Date: Jul 2003 Location: Sweden Posts: 66	I do not agree that you should make your bp thinner, silver has superior thermal conductivity and thus you can have a thicker baseplate than lets say a copper one. A thicker baseplate means that the heat will be distributed over a larger surface area, though at some point it will begin to insulate the heat rather than cool it. The optimal baseplate thickness depends on the design of the block ofcourse and the material used, since your design does not rely on some advanced jet impingement technique and it hasn't got a central placed inlet, the heat needs to be more evenly distributed across the baseplate, thus a thicker baseplate is prefered. Though i must agree that the bp you are currently using is kinda thin overall, the shallow channels you have will make the block quite restrictive, though if you have a crap flow in your system allready you won't notice any differance. Please correct me if am totally wrong you thermal gurus. Maybe silver = even more thinner bp? O_o?

03-22-2005, 08:38 PM	#5
Moto7451 Cooling Neophyte Join Date: Jan 2005 Location: LA, CA Posts: 53	I use a hand drill & a Flex Shaft drill (think of a dentist drill on steroids, big 120v AC .5A motor that spins at 14,000 RPM) for my blocks. Right now I'm working on a "pitted" block for my GPU which simply has rows & rows of overlapping drill holes to create turbulence. It almost looks like someone has chewed on it with their molars . I follow the school of thought that a thin base means you can get heat to the water faster & it wont "insulate" the heat so much. The thicker base school of thought is valid too, the designs end up being different often times. I figure with a GPU like my RV350, thinner is a good thing. Perhaps it isn't but I'll find out by building blocks now wont I. Good luck with your design.

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