Thickness of material between water & die & your temps

D

Dumass_Freakboy

Gawd
Joined
Dec 20, 2000
Messages
590
I have custom made my own waterblock. Right now if i'm not mistaken the actual aluminum is 0.76mm or thereabouts thick between the core & the fluid.

This question is for everybody.

What is the thickness of the material (copper & alum) between YOUR cpu & the cooling liquid? (And what material?)

I have applied some thermal paste (perhaps too much) and sanded the block down /w 2000 grit water sandpaper. My current idle temps are about 37 degrees C.

What are the temps you guys are seeing /w copper? /w aluminum?

I've measured the water @ 25.

Do you think if i milled the block down a bit more to 0.5mm it would bend when I applied the pressure on the waterblock?

Anybodys thoughts or comments are welcomed.
 
BUMP

Somebody has got to have something to say!!

Your temperatures on your watercooled rig?? The type of metal your block uses?? I'm trying to figure out why there is a 13 degree diff between water & the cpu temp...
 
What kind of internals does your block have? 0.76mm is way too thin unless you have some serious jet impingement going on.
 
Not sure what exactly you want, here is the swiftech copper block dimensions, thinner will yield better temps, but of course you want the thing ridgid.
 
I have made several homemade blocks and I do believe that it is the thickness of the material that is in direct contact with the cpu that has the MOST effect on the temp. I had a simple aluminum block that I drilled out. It measured about 2"x3" solid aluminum and it worked GREAT, but looked like trash... Now I have a beehive made from a 1 .25" copper end cap and several smaller ones inside but the plate is only about .25 " thick... And I definitely don't get the performace that I did with the aluminum. My guess is the heat capacity of a solid block of metal is so much higher than a plate, that is aborbed more heat and thus gave it off to the water there by cooling the cpu much better. of Course finding a block of metal like that is slightly difficult as I just happend to find it going through a metal recylcling plant near my place. And finding a solid block of copper would be too heavy for the MB...
 
I bought a big hunk of aluminum from a metal supermarket... about 2' long... 2" wide by 1" deep... cost me $30... a similar piece of copper.... 16" would have cost $120 :eek:
So I could have gotten some thinner stuff... like 1/2" or 5/8" which would do and probably cost less... the block I is 1" high... so it's pretty big.

I machined it down... so basically it's like a giant hole in the middle... with 0.76mm of aluminum at the bottom between the cpu core and the water (and the compound), a rubber seal and a plexiglass top.

[edit] oh, and those dimensions for the swifty (was it?) dont provide the thickness of the material between the cpu core and water. :( [/edit]
 
Machined it down? You have access/skills for miling? Cool.

For a relatively featureless Al block like you made, you'd want a baseplate thickness of 8 to 10 mm. The idea being that there needs to be enough material between the core and the water to distribute the heat over the widest possible water-contact area. You also want the restrict the depth of your water chamber to a millimeter or two (no point to water that flows through your block without getting near the metal).

With a mill, try a finned block. Say, 3 or 4mm baseplate, 2mm thick 8mm high fins 1 or 2mm apart. Don't have an open area above the fin, force the water to flow through them. Have your finned area extend beyond the core by about the same thickness as the baseplate.

Another good one, do-able with just a drill press, is a pin block. Drill a grid of 3 or 4 mm holes, spaced so the holes barely overlap, again leaving 3 or 4 mm for the baseplate. With a dremel or an endmill, open water passages among the pins. Make the pin height about 3 time the drill diameter.
 
vbbartlett said:
I have made several homemade blocks and I do believe that it is the thickness of the material that is in direct contact with the cpu that has the MOST effect on the temp. I had a simple aluminum block that I drilled out. It measured about 2"x3" solid aluminum and it worked GREAT, but looked like trash... Now I have a beehive made from a 1 .25" copper end cap and several smaller ones inside but the plate is only about .25 " thick... And I definitely don't get the performace that I did with the aluminum. My guess is the heat capacity of a solid block of metal is so much higher than a plate, that is aborbed more heat and thus gave it off to the water there by cooling the cpu much better. of Course finding a block of metal like that is slightly difficult as I just happend to find it going through a metal recylcling plant near my place. And finding a solid block of copper would be too heavy for the MB...
Click to expand...


First, heat capacity of the metal is not an issue, it only would be under non-steady state heat conduction. When considering equilibrium conditions, heat capacity does not affect transfer.

About base thickness... you need to find a compromise between low thermal resistance and ability to spread heat laterally away from the core. It really depends on many things such as materials properties, block internal design, and flowrate. For example, if you had the most efficient design for transferring heat from the base to the water, you would want as thin a baseplate as possible without compromising structural soundness, which minimizes the thermal resistance of the base. With less efficient (more realistic) heat transfer, a somewhat thicker baseplate may be beneficial in order to "spread the heat out" over a larger area of the base.

t's a tricky problem, and really nobody is going to be able to tell you exactly what to do. You could play around and do some experimenting to see what works best with your design and the rest of your system.
 
Yeah... thats what I'm currently doing. My dad bought a milling machine about half a year ago, and when I started my summer, I started working on a baseplate, and collecting all the materials I needed for a wc setup.

I drew up a diagram for a new baseplate... just wish I had a scanner or camera... doh.. gotta borrow one from a friend @ work :) So you guys can see some pics of what I got.

Oh, and the main problem I think is the heat transfer from the aluminum to the water. Because i've taken the temperature of the aluminum /w a laser thermometer and the temp of the water just above it... and the temperature diff is like 20 degrees C.

The cpu is 46 degrees oc'd under load from 2500@3200. The side of the block at this time is 46.5 degrees... temp of the water entering the block and leaving the block is almost the same temperature, about 0.5 deg higher leaving.

The water temp entering the tank is 25.8 and leaving it is 23.8 (temp of tubing).
(I have the pump submersed in a home built tank about 6" by 8" by 7")

Plus my dad has access to a CNC milling machine :O Which If i can figure out how to get an autocad drawing up.. i may try and make a better block which is circular 'n stuff. :)

[edit] Went out and bought a digital camera today.... Olympus D-580.. nice camera... anyhow, heres one picture of the waterblock... and coming soon, pics of the block itself open... and other stuffs
P1010009edit.jpg
[/edit]
 
Took some more pictures, of the block i'm working on, and the milling machine. Thought HeThatKnows would like to see. Also, anybody have thoughts on this picture I drew up of a block?

P1010009.JPG

P1010008.JPG

P1010007.JPG

I was really thinking about the design directly over the cpu core, because that is where I want the thinnest aluminum. But the fins in the cascade got me thinking whether that could help me.
ddTDXag4.jpg

Plus I was thinking about copying the jet idea /w myne because I'm also using 1/2" plexi, it wouldn't be out of the question.
 
is that last picture a picture of the block after you cut it? if so, could you get one with less glare so we can see what it really looks like? thx.

my block is cut from a .25" thick piece of copper (1 on top of another, soldered together) i've got channels cut in the base that go about 1/2 way through the block. it has an angled piece of 5/8"OD pipe that is crimped at the end to a pretty thin opening (no exact measurements) that angles to the center of the block, then a 5/8" OD barb outlet. don't know temps yet on my current chip, as i've been running it with a 5 gallon bucket reservoir and no radiator, but i'm planning to get a heatercore today or tomorrow. then i just need to get some 120mm fans, make a shroud, and i'll be set. whee :rolleyes:

i'll post temps/outside pics of the block (can't reopen it... it's soldered together. i hope no solder leaked and blocked off the waterflow... that would really suck) as soon as i get it running.
 
That's some nice milling there. I see you've cut the same sort of wavy fins the TDX and RBX use.

How thick are your fins? With aluminum's thermal conductivity, it generally does no good to have the fins more than twice as high as they are tall; all the heat is passed to the water before it travels that high up the fins. With copper (higher thermal conductivity), you could go as high as 3:1 or 4:1 for the fin's height:width ratio.

For the base thickness, how about an experiment? Start with a thick base, see what temperatures/overclocks you get, then mill it a bit thinner, repeat. What you'll find is that a really thin baseplate limits your overclocks, because you'll have hot-spots directly below the channels

How do you keep the laundry from being covered with metal chips? :p
 
nonononono
that is the cascade... the one made out of silver oooOOOO
I want to do something like that, but was wondering if it was the best thing to do. I was going to first try some different blocks and see what kinda results i can get... then get some copper later and make it out of copper once i find the most effecient design.

That's not my cascade either, it's just somebody elses pic which I stole. :eek:

The laundry is far enuff away that the chips don't fly that far! :) And if any do get on it... I just move it away and flap it out in the air.

Gotta keep the place clean... note the vacuum cleaner :p
 
Hehe, okay. Guess that explains why it looked so much like the TDX -- it is a TDX. :D

Prototyping in aluminum won't help a great deal with making copper blocks. The difference in thermal conductivity is enough that a good geometry for Al will be mediocre in Cu (and a good design for Cu will suck if made from Al).

By the way, the base of the real cascade has an array of cylindrical cups (no fins), while the middle plate has a matching array of jet tubes that stick down into the cups. "Submerged Jet Impingement."
 
I know, i have those pictures too...
I don't know whether I could get a good jet impengement going on. I'm mainly going for the best temps I can get. I don't care how I get it, but I want to make my own block. I just fuxored up one block, milled out the bottom when I was going around an edge... doh... you know that little spot on the socket for Socket A where it rises. I milled that out on the bottom, and I was milling out the top, and on one little corner I milled a little hole :eek:

Anyhow, this what you talkin about??
c22.jpg

c23.jpg

xxx-1.jpg


And some pics of my blocks...
P1010001.JPG

P1010002.JPG

P1010005.JPG
 
Yep, that Cathar does some kickass work. Wish I had one (or the skills to make one).

I've never bothered with the the cam-box, seemed like a pain in the ass to make that notch. Where I always screw up is tapping the holes to bolt down the top. Nothing uglier than having to make a new hole a little ways away.

So whats that on the bottom of the bottom picture? I can't quite make it out with the high contrast.
 
The second to bottom is the screwed up block... and the bottom is the block which I cross milled to get a flat top to it. I have plans to work on it today a bit before i go to work, and hopefully get it done tomorrow morning.

My dad did the tapping for me. He said i'd probably break the tap, knowing his experience :eek: he's probably right. I'm a pretty big guy and have less patience than him.

Just as well, less work for me to do :p
 
You must log in or register to reply here.
Back
Top