Pro/Forums

Pro/Forums (http://forums.procooling.com/vbb/index.php)
-   Water Block Design / Construction (http://forums.procooling.com/vbb/forumdisplay.php?f=37)
-   -   Made my own waterblock! (http://forums.procooling.com/vbb/showthread.php?t=6142)

Graystar 03-23-2003 10:34 PM

Made my own waterblock!
 
Introducing the "WhatBlock?" micro waterblock

http://home.nyc.rr.com/graystar/wate...waterblock.jpg
http://home.nyc.rr.com/graystar/WB3.jpg
http://home.nyc.rr.com/graystar/WB4.jpg

This block was made from sheet stock and copper tubing from OnlineMetals. Almost every cut was made with a Skil 4580 jigsaw and metal cutting blade. The holes were made with regular twist drill bits. A Dremel cutoff wheel was used to make the internal cuts on the middle piece.

I made this block to test a couple concepts. The first was to see how well a thin base can perform. The second was to see if a narrow channel would help heat up more of the water.

I can say that this block is performing slightly better than the Ahanix Iceberg1 waterblock. Even with a cheap pump, the flow through the block is substantial, about the same as the Ahanix block. I believe the limiting factor in my setup is the radiator, so I think I'll have even better results when I get a good rad.

Right now the CPU with the micro block is at 43C while the other aircooled processor is at 52C (I got a really hot computer.)

There were concerns about the base bowing. Let me tell ya...this thing is rock solid. I can't bow the base even if I grab the tubes and squeeze really hard. After having it on for a day the base didn't even have a mark from the die. So it's more than strong enough.

#Rotor 03-23-2003 10:54 PM

HHAAAAAAA-SHOOOOOOO !!!!!! :drool:


oooops sorry :D

bigben2k 03-24-2003 10:24 AM

Welcome to ProCooling!

I believe that your block is suffering from a thin baseplate. At 1.1 mm thick, I assume that those lines are some kind of grooves that you traced/milled/scored, making the bp even thinner, at that area.

Your baseplate is too thin, if it's not going to have any fins.

I don't know what your pump and cooling solution is, but I bet you could improve performance significantly by going to a bp thickness of 3 to 5 mm, IMO.

Graystar 03-24-2003 10:44 AM

Why would making the base thicker improve performance? The idea was to have as thin a base as possible. That would reduce the temperature gradient within the copper base, and get higher temperatures to the water. As we know, heat transfer is better with a greater temperature differential.

I have the Ahanix Iceberg1 kit. So the pump is underpowered (just 3ft. of head) and the radiator is very small. I just put a second fan on the radiator and now my temp is 43C. BTW, this machine folds, so both processors are at 100%.

I think that with a Eheim 1048 and a decent rad I'll easily be in the 30s.

In this version of the block I did *not* make the cross scoring as indicated in the drawing. The base is smooth on the inside. I wanted to see what kind of flow I got with it smooth. Now, I can make another with the cross scoring and see what effect it has on the flow rate. The cross marks are just very lightly scored into the base plate. No real ridges are being made.

bigben2k 03-24-2003 11:10 AM

Yes, a higher temp will yield a better delta T, but with that logic, you have to look at the other side: if it's too hot, then the water can't pick up the heat efficiently, and you end up with higher temps.

This is where it becomes counter intuitive, because you would normally figure that by slowing down the flow, the water would be able to absorb the heat, when the reality is that it will actually make things worse, because your coolant will heat up, and your baseplate is still very hot. (or something like that...)

The solution is to decrease the resistance of the heat transfer, to the water. You can do that in a number of ways, classified into two categories:
1- make the flow turbulent
and/or
2- Add fins/pins

#1 is extremely hard to achieve, because the heat transfer is exponentially proportional to the flow, in an unfavorable way: i.e. double the flow rate, will lower the resistance by 1/4 (a very rough approximation, meant as an example).

Optionally, one can use jets, or small orifices to project the water at a higher speed, against the baseplate, which creates a turbulent area, right where you need it.

#2: pins and/or fins have the effect of spreading the heat upwards, keeping the baseplate temp a smidge lower. Don't dismiss the scores/grooves, because that's actually a fin, in reverse.

There are a number of references available on all this.

Graystar 03-24-2003 12:25 PM

It would be absolutely impossible to improve the performance of the block by increasing the base thickness alone. Getting more heat to the water is always better. The hotter the base, the better the transfer. Just gotta make sure I can get rid of the heat.

All I really need is to increase my flow and get a better rad. There doesn't seem anything counter-intuitive about the flow. Dousing something in water cools it off. The more water you throw on it, the cooler it gets. A waterblock is simply dousing from the inside. I don't know who out there would "normally" think otherwise. *That* seem counter-intuitive.

Like I said, the block is actually working well, considering the rest of the equipment This is the lowest temps my processors has been in a long time.

bigben2k 03-24-2003 12:36 PM

Best of luck to you then!:(

winewood 03-24-2003 12:46 PM

There are certain efficiencies to thermodynamics that Bigben has a grasp on. One thing about watercooling, the most obvious answer at times isn't. I had some doubts about this myself, but after hearing the reasoning out from Bigben, cathar, unregistered, and countless others, I found that their offhand remarks cary more educated direction than my best logic.
If I could make a suggestion.. don't discount Bigben so quickly, but research or question the logic behind it. He is really good about explaining, and will put your cooling in a much better position.
I don't doubt your grasp on cooling at all, as your block is quite nice and more advanced than anything I could possibly tool out myself. Good job, this is one of the most unique designs I have seen. Can you take a picture of your mounting and post temps, and your reference equipment?

Graystar 03-24-2003 02:31 PM

I apologize to bigben if my post had a "discounting" tone about it. I certainly don't mean to discount anyone's ideas, as I certainly am no expert on this subject. I appreciate all comments and actually do give lots of thought to suggestions.

However, when someone makes such a direct design comment ("Your baseplate is too thin") then I would expect a more scientific explanation as to why, all other things being equal, a thicker base would have a positive benefit. The explanation given (" if it's too hot, then the water can't pick up the heat efficiently") doesn't make sense, as a greater delta T always leads to more efficient heat transfer. That's basic heat transfer. If there's some other factor involved, I didn't get it, so, I don't understand the reasoning. I can see how the other suggestions, relating to turbulence and fins and such, would improve the block. But those ideas would work regardless of the base thickness.

Bigben, if you can explain it better I'd love to hear it cause, in the end, it's the lowest temp that I'm after. :)

There should a picture of the mounted block in the first post of the thread. I used the clip from an old SK6 heatsink. The equipment I'm using is not very good. I am using the Ahanix Iceberg1 system and I simply replaced the kit block with my own. The pump is rated at 150 GPH with a head of 3 ft. Not very powerful at all. The radiator is as wide as an 80mm fan and about 1 1/2 times as long. It's a very small radiator that has terrible airflow through it.

I have a dual processor board that runs very hot because of the model 6.8.0 2200+ processors. These run about 5 watts hotter than the 6.8.1 2200+ processors. Since I don't have any form of "usual" pump or rad, I gauge performance by comparing the water cooled processor to the air cooled processor. With the old block the water cooler was only 1 degree cooler than the air cooler. This was expected, as the Iceberg1 is a poor performer. With the new block the temperature dropped by 3C to 4C. I then decided to add a second fan to the radiator. That dropped the temp another 3C. So right now, the new waterblock is sitting at 8C to 9C below the air-cooled processor. My current temperature is 43C. I hope that when I get a real pump and radiator (very soon) I'll be in the 30s. Oh, and my machine folds, so these temps are at 100% processor.

bigben2k 03-24-2003 02:59 PM

No apologies needed, it's your design!

If you search through these forums, you'll find that we tackled the baseplate question, to some extent.

The general conclusion, was that for a flat plate type of block, such as a Swiftech, where the water hits the plate, then exits, 5mm will give you the best results. There is actually a performance decrease, if the baseplate is thinner.

So we all went over different types of designs, and found that Cathar's block, which is a combination of multiple jets and densely finned area, allowed him to lower the thickness under 1.0 mm. Now that's a tribute to Cathar's design, because I think few of us thought it was possible to get to a baseplate of 1 mm.

In your design, there is no jet, there are essentially no fins, but you do benefit from a fair pump, so I estimate that you would fare better with a bp of 3 to 5mm.

As for the science of it, I can't throw numbers or equations (few can, ask Les), but I can attempt to explain the general idea.

It might be simpler to look at the system from the point of view as a series of thermal resistances, through the TIM, through the baseplate, and finally into the water. By decreasing the baseplate thickness down so far, you actually increase the resistance of the transfer of the heat, to the water.

I often find it useful to look at extremes, in a multiple of five or ten, to try to explain something.

Since you're already running it, you can do a simple test: compare your water temp to your CPU temp. If the difference is more than 10 degrees C, your design stands to be improved.



Not to slam your design, but if I were you, I would be concerned about one of the hoses getting tugged, and crushing a corner of the core. The Athlon pads are great for preventing this, to some extent. I'm sorry, I just had to bring that up. I'll be using a fairly large res (a 6" tall prototype), right on top of my block, and to prevent any issues, I'll not only use the pads, I also picked up a shim. Accidents happen so quickly!

Balinju 03-24-2003 03:06 PM

i don't know if i am seeing correctly, but do you trust your pipes in the barbs without any clips or ties?? or i am not seeing them :confused:

Balinju 03-24-2003 03:13 PM

Quote:

Originally posted by bigben2k

Since you're already running it, you can do a simple test: compare your water temp to your CPU temp. If the difference is more than 10 degrees C, your design stands to be improved.

so now i have a question, the difference in 10 degrees C that you are talking about is the difference when the cpu is under load or when it is at idle??

bigben2k 03-24-2003 03:28 PM

A fair question, Balinju!

I'm going to refer you to these results, for my reply.

If you go down to Bill's test results, you'll find (as of today), 9 blocks, rated with a c/w between 0.19 and 0.26. That's degrees Celsius per Watt.

So for Cathar's block (0.19), assuming a 70 Watt load, the temp difference is 13.3 degrees C. This is, in the case of this block, a temp difference between CPU and water, at a flow rate of 1 gpm. (at 2.5 gpm, you can actually get close to 0.17 C/W). This is the best block known to the entire watercooling community, and I mean worldwide!

To answer your question, I was referring to an idle temp difference.

Graystar 03-24-2003 03:32 PM

Quote:

Originally posted by bigben2k
Since you're already running it, you can do a simple test: compare your water temp to your CPU temp. If the difference is more than 10 degrees C, your design stands to be improved.
Hey, since I got a res, and the water goes from the block to the res, I can measure that. I'll see if I can pick up a thermometer today.


Quote:

Originally posted by bigben2k
Not to slam your design, but if I were you, I would be concerned about one of the hoses getting tugged, and crushing a corner of the core. The Athlon pads are great for preventing this, to some extent.
It's amazing how much tension those clips have. I actually had my rad hanging off the block and the block didn't move. So I'm not too worried about, but I do think of it.

Quote:

Originally posted by Balinju
i don't know if i am seeing correctly, but do you trust your pipes in the barbs without any clips or ties?? or i am not seeing them :confused:
You are seeing correctly. The amount of pressure produced by typical watercooling pumps is but a drop in the bucket compared to the amount of pressure required to get that tubing on and off. Like I said earlier, I had my rad hanging from that tubing. It ain't going anywhere!

Balinju 03-24-2003 03:34 PM

thankyou bigben. i will find this very helpful

Fixittt 03-24-2003 04:49 PM

Grey,

Let me start by saying good job, I like it.......

Now let me tell you what "I" would do a bit different and see if you can agree or disagree.

What I would do, is make the base thicker, for the simple fact, that I would try and get as much surface area as I could. now with a thicker base, I would put the micro channels as you show. But I would take them deeper. I would also sand blast it, as the rough surface will yeild a bit more surface area. I to would also score (Micro channel the top plate and blast it as well, because a small block like this will probably indeed be transfering the heat to both plates.)

I think that the smooth surfaces of the block are what may be holding back better temps. And the other components that you have. But with a dual system those are not all that bad of temps.

Good job.

Fix

Graystar 03-24-2003 05:32 PM

Quote:

Originally posted by Fixittt
Let me start by saying good job, I like it.......
Thanks! It was a challenge and I had fun making it.


Quote:

Now let me tell you what "I" would do a bit different and see if you can agree or disagree.

What I would do, is make the base thicker, for the simple fact, that I would try and get as much surface area as I could. now with a thicker base, I would put the micro channels as you show. But I would take them deeper. I would also sand blast it, as the rough surface will yeild a bit more surface area. I to would also score (Micro channel the top plate and blast it as well, because a small block like this will probably indeed be transfering the heat to both plates.)

I think that the smooth surfaces of the block are what may be holding back better temps. And the other components that you have. But with a dual system those are not all that bad of temps.

I agree with you on the base modification, but one of the design goals of this block was to build it with a hacksaw and a drill. Well, I ended up buying a jigsaw for the cuts (which worked out great,) but otherwise there are no special tools needed to make this thing. Microchannels would be wonderful if you can find an easy way to make them without a milling machine.

I agree with the sandblasting in principle, however I had to rely on the flatness of the sheet of copper, as delivered from OnlineMetals, for my flat base. Every cut and hole was made with the sheet sandwiched between plywood, so that it wouldn't deform. Sandblasting or peening would deform the sheet. I don't think I'd be able to make it flat again.

Finally, I agree that the smooth inner channel is holding back performance. But I have enough material to make at least 20 of these blocks. I think this block is a great reference point, in that I can now make single changes, such as base scoring, or a wider channel, and compare results.

I think a slightly wider channel might be the next thing to do. I wanted a wider channel, but was afraid that the base might bow under pressure. HA! What was I thinking? There's no way that base is going to bow! I can't even bow it with my hands squeezing the tubes. It's really solid. Then I'll try the scoring. That should make a real difference (I hope!)

max 03-24-2003 05:42 PM

IMO the reason why a thicker base plate is better is so it can spread the heat out sideways before it gets to the water, hence more surface contact with the water.

Making yours wider IMO will not improve performance because with a BP that thin the only place getting hot is right above the core, maybe it would improve performance by letting more water flow though.

With the blocks I have made, mainly rotor style drilled ones, the differnce between 2mm and 3mm base was about 10C!

BTW i like your design:) all these people making there own is great
:cool:

bigben2k 03-24-2003 05:47 PM

FWIW...

It's possible to mask off a piece, when sandblasting it.

I'm about to place an order myself, with onlinemetals. I don't mind ordering an extra piece, 3mm thick (actually 1/8 inch), and mailing it to you, if you tell me how wide it is. PM me. Is 1 inch long OK?

Fixittt 03-24-2003 05:55 PM

sandblasting should not deform the metal at all.
also ever seen those vibrating engravers? that could be used to make your channels, they are what $12 at a hardware store? I used to use them to engrave on glass, just take the tip and sand it to a finer point. could also be used to give the surface a sandblasted finish.


Hell, rub the metal on some course 60 gritt paper. that will do it too.

FIX

Graystar 03-24-2003 06:01 PM

Quote:

Originally posted by bigben2k
FWIW...

It's possible to mask off a piece, when sandblasting it.

I'm about to place an order myself, with onlinemetals. I don't mind ordering an extra piece, 3mm thick (actually 1/8 inch), and mailing it to you, if you tell me how wide it is. PM me. Is 1 inch long OK?

A piece that you're gonna sandblast for me? Or a piece to test the thicker base idea? Either way, that would be cool, but it has to be 1 1/4 " long. I got 3/8" tubes and I need 1/4" between the tubes for the clip. That's 1" exactly so I need some overlap. The piece is 5/8" wide, btw.

Graystar 03-24-2003 06:05 PM

Quote:

Originally posted by Fixittt
Hell, rub the metal on some course 60 gritt paper. that will do it too.
LOL! That *would* do it...wouldn't it... :)

Unlike you guys (it seems :p) I don't have a sandblaster. So that's out of the question. I was just going to take my dremel cutoff disk and just run it across a bunch of times.

bigben2k 03-24-2003 07:36 PM

Quote:

Originally posted by Graystar
A piece that you're gonna sandblast for me? Or a piece to test the thicker base idea? Either way, that would be cool, but it has to be 1 1/4 " long. I got 3/8" tubes and I need 1/4" between the tubes for the clip. That's 1" exactly so I need some overlap. The piece is 5/8" wide, btw.
1/8 by 5/8 by 1 1/4. OK, no problem. I should be ordering it this week.

No, I don't have a sandblaster either. I found that they're only good for two things: cleaning grafiti, and etching glass details.

Fixittt 03-24-2003 08:52 PM

lol cleaning grafitti ... TURK 182!!!!!!!!!!!!!! (the movie for all you youngins)

bigben2k 03-24-2003 09:17 PM

OT: I remember Turk 182...


All times are GMT -5. The time now is 02:18 AM.

Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...