Huge disparity w/watercooling

[Manticor]

I'm running a homemade watercooling setup, and I seem to have run into a bit of trouble:

-At idle, my cpu temps are 44/45c (as reported by MBM5) while my water temp is generally around 26/27c (PC has been on for the past 24 hours).


Some quick info about the setup:

-I know for a fact that there's very good flow going through. (Eheim 1048 going full blast)

-The block is lapped with 1500 grit sandpaper and is EXTREMELY flat (I used my fathers lapping machine with a 40 micron silica solution)

-I've got Arctic Silver II applied as a thermal compound (in a thin layer because of the mirror finish)

-The waterblock is seated on reasonably firm via screw/wingnut system...probably over the 15-24 lbs limit it's supposed to be at

-The cpu goes from 21c -> 44c in about 5 minutes after I boot up.

-The waterblock is of the copper cap variety

The only real thought I have in my mind is that somehow my previous hsf, the GlobalWin FOP32-I, which had a very rough base, somehow scratched the die/was misaligned and screwed up the die seating, and now I'm not getting good contact.

Any thoughts? I'm stuck.

Thanks in advance

 

[Manticor]

This is the base of the block, at its relatively mirror finish (no, I didn't use brasso , just the Hoot technique--Thanks Hoot!)

 

[Manticor]

I modified the base a little bit recently to try to get more surface area...didn't change the temps any

 

[Manticor]

And finally, this is how the water flows in/through the block (in case you were unclear on how it hits the base). The red is the CPU--and no, it isn't exactly centered over the input. I think that would defeat the purpose of the the stalks in the center.

 

[fuzzba11]

Sounds like your water is a little warm, I would think. I'm running a bong cooled system, only ~3 feet high, and my water temps range from ~18 at night to at most 24 C when loaded. You might want to consider changing your method to cool the water

 

[BillA]

that wb would make a good sinker for a trot line

for what it is, I'd say the temps are excellent
lots of threads here on wb design considerations, which you have ignored

be cool

 

[nihili]

Here are some thoughts.

First, BillA is correct. There are a lot of good threads on wb design. Looking through them should give you some good pointers.

Second, from your drawing it looks like you have extended both the inlet and the outlet tubes down near the baseplate. Thats a bad move. Water will tend to move directly from the inlet to the outlet and just eddy in the rest of the block. This greatly reduces the cooling potential of the block. The link you gave was correct to put the inlet directly over the die. With this type of block you won't get as good a result from offsetting it. You want the inlet centered and as close to the baseplate as it will go without reducing the flow rate.

Third, from your picture it looks like you have a very thin baseplate. I suspect that you're not getting enough heat spread through the base. Try a thicker plate.

Remember, this will never be a great block. It's designed for ease of construction, not performance.

nihili

 

[Manticor]

Thanks for the replies you guys. I guess it makes sense how the temps rise--that the die is heating up the entire block in those first 5 mins and that the water isn't pulling sufficient heat away. I'm torn between three things now I guess:

1. Scrapping this and trying my hand at a solid block w/tunnels

2. Direct die cooling with a water jacket

3. Adding a peltier and seeing how that works (anyone know a good place to buy from?)

Any suggestions?

Thanks

 

[cjlax5]

i dont know how great of temps you would get with a pelt using that block...it just may not be able to draw away enough heat. i would go for a new wb. if u dont want to make it, buy one, they arent that much.

 

[nihili]

Thanks for the replies you guys. I guess it makes sense how the temps rise--that the die is heating up the entire block in those first 5 mins and that the water isn't pulling sufficient heat away. I'm torn between three things now I guess:

1. Scrapping this and trying my hand at a solid block w/tunnels

2. Direct die cooling with a water jacket

3. Adding a peltier and seeing how that works (anyone know a good place to buy from?)

Any suggestions?

Thanks

I vote against the pelt. Pelts need very good water blocks.

As for what to do next, that sort of depends on what your goals are. If you just want to end up with a first rate block, then go buy one. The Gemini seems to be a great block at a very reasonable price. But it seems to me that you're really after a learning and tinkering project and the satisfaction of having rolled your own. If I'm right, then I'd tear that one apart, modify it, record the changes it makes and their effect on the system. Keep doing that till you think you have the best possible copper cap block. Then take what you've learned and apply it to a solid block with tunnels. Along the way, keep your eye on the waterblocks threads.

If I had a shop handy, I'd take the second route.

nihili

 

[Tecumseh]

I say you got pretty lucky with the temps you did get.

Read the thread on numerical WB design.

I couldn't tell, but I think your base was too thin, too.

You need at least a 1/4 inch base and and Easter Basket
full of turbulence.

 

[Manticor]

Ok, I think I'll tinker with this one for a bit then scrap it in favor of a solid copper one. I definitely can't buy it however, since this is a school project I was just thinking with the peltier that it would spread out the heat a little better on my too-thin baseplate (it is 1/8" thick... you're right, very bad for dispersing heat).

Just wondering for the future though, how much is a good 156/172 watt pelt and where's a reputable source I can get one from?

 

[CrystalMethod]

Here are some thoughts.

First, BillA is correct. There are a lot of good threads on wb design. Looking through them should give you some good pointers.

Second, from your drawing it looks like you have extended both the inlet and the outlet tubes down near the baseplate. Thats a bad move. Water will tend to move directly from the inlet to the outlet and just eddy in the rest of the block. This greatly reduces the cooling potential of the block. The link you gave was correct to put the inlet directly over the die. With this type of block you won't get as good a result from offsetting it. You want the inlet centered and as close to the baseplate as it will go without reducing the flow rate.

Third, from your picture it looks like you have a very thin baseplate. I suspect that you're not getting enough heat spread through the base. Try a thicker plate.

Remember, this will never be a great block. It's designed for ease of construction, not performance.

nihili

Another draw back to the inlet and outlet being so low is that it may be hard/impossible to remove all the air completely from the WB. That could be contribute to you having high temps with water cooling.

 

[nihili]

Another draw back to the inlet and outlet being so low is that it may be hard/impossible to remove all the air completely from the WB. That could be contribute to you having high temps with water cooling.

Oooh, very good point. One that had escaped me completely.

nihili

 

[Cougar522]

id go with your first route using what u have learned to create your own wb with tunnels.

 

[Manticor]

I think I am going to do just that. Thanks for the help guys. BTW, anyone know offhand how thick the copper should be left between the die and the tunnel?

 

[shadow]

DISCLAIMER - my opinions.

I don't think you're that far off.

Another draw back to the inlet and outlet being so low is that it may be hard/impossible to remove all the air completely from the WB. That could be contribute to you having high temps with water cooling.

True, he probably does still have air in there, but that's not what is messing him up. The air is not creating a buffer between the metal and the flow of water. The flow of water is from the inlet to the outlet, past two pillars.

Your fatal flaw - water can go right from the inlet to the outlet without hitting 90% of the metal you've got in there. You create nice laminar regions along the base metal piece in your design. That is where all the heat is, the only effective water-metal contact you have are those two pillars in the middle.

There are a few things you can do to the block you have now.

The simplest
A) Drop the inlet and outlet even farther, so that there is only ~1/2" between the tube and the base. This should help elminate the hot spot right above the core. But, as your tube is not on axis with the core, it might not do all that much. How far from the bottom of the inlet to the metal?

The hardest
B1) Make channels, you've shown some soldering skillz, add a V that goes from the bottom of the cap to the top This can be done with very thin copper. Creating a channel. Place the bottom tip of the V somewhere in the middle of the CAP so that the water has to go around the top tips of the V and around the cap before it gets to the outlet. This will increase cooling as you're FORCING the water to contact more metal.

B2) Drop the inlet tube even farther and try to direct it to the middle somehow, to the middle of the V before it makes it's way back. Perhaps place a triangular piece of copper in your block.

*********flame words (if you can't hold back, don't read this**********
This minimum thickness IMO is crap, and I will prove it one day. The numerical analysis will also.
*********flame words**********