Kinda Newbie

[ibmkg]

I recently made a water block. I used a cross drill method. Block has following specs:

Thickness:10mm
Length x breadth: 76 x 76 mm

Now I have drilled two holes of 60mm on one side and three holes on other. I have then used one for water in and other one for water-out. Rest has been blocked using small screws. Hole size was 5mm.

I am using cross drill because it is the easiest method.

For testing, I took a blowtorch and heated the block until it was hot (the point at which I could not touch it). Then I powered the pump and within 3-5 secs the block returned to its normal temp.

For pump, I used a car's fuel pump. I really want to show you guys what I am up to but now I only have access to my Cell Phone VGA camera.

My questions:

a) Is Cross Dill efficient enough to hold down the temperature of a LGA775 p4 HT 3.2GHz?

b) How can I test the efficiency of my block? One way I could think of is attaching the block to a car's engine heater (found in diesel car engine and works on 12V). First, I would measure the heater temp then attach it to block and measure the drop in heater’s temp.

c) Say if I bolt in this block, is there any chance that I might damage my CPU? I mean like, would it be overheated or something?

d) Which thermal paste should I use?

e)If I increase the thickness of copper from 10mm say to 13mm and increase hole size from 5 to 8mm, would that be a good idea? Also, increase the copper size to say 85mm length and 51mm width (decreased, as this is the only size I can get my hands on) and increasing the number of holes?

For mounting issue, I was thinking of gluing a Plexi plastic to block and drilling holes in it.

Excuse my English but I am in a hurry.

Cheers! DAN

[Brians256]

a) Too many factors to tell. Just test it.
b) Put it onto a CPU and compare it to another known (tested by the big boys) block. It'll give you a rough idea.
c) Yes. I suggest not using a hammer, though. Seriously, the modern CPUs havea thermal cutout. You're more likely to damage things through mechanical action than thermal.
d) Any one of them but toothpaste. Seriously, they all perform fairly similarly.
e) I'm not a block designer.

Your english is fine. Good luck!

[ibmkg]

Thanks for the reply.

I am about to take a leap after my rad and reservoir is made. I was instructed in #Windows95 (Dalnet IRC, Yeah I am still a lamer) that I should use "Arctic Silver Ceramique", I am about to order that. Other than that, I can get 'Transistor Paste' from a local shop.

I am now concerned about mounting as I do not have a slightest Idea what to use (bolts springs?). I would be needing help in this. I would also need socket blue prints for LGA775, can any one help?

I do not want to be involved in math regarding mounting issue. It is extremely boring.


Just for reference, I am using LGA775 Intel 3.2Ghz HT on a ASUS P5WD2 Premium Mobo.

[Brians256]

Arctic Silver is good stuff, but transistor paste probably won't be 1C hotter.

Do a search on the forums for mounting holes on 775. There should be some info for you.

[ibmkg]

One other thing, I am going to use Car's Coolant for WB colant, is that ok?

And if possible, make my search easy by refering LGA775 dim to me. Thanks.

[ibmkg]

Ok I got answer to my question. LGA775 dimensions for any board is 72mm to 72mm(hole to hole), which makes it a perfect square.

Again I have set of questions:

a)Use of engine rad coolant, would that work?
b)Is there any chance of condensation? If yes, will it hurt and how to avoid it?
c)Any help regarding springs and bolt / washers setup would be appriciated.

[Brians256]

a) yes
b) not unless you chill your coolant or CPU to below ambient temperature by peltier, HVAC, or other method
c) Search the forums again. Look for stuff on "which springs" and you should find something.

[ibmkg]

Again, thanks for reply. And once again I have a question:

a) My WB does not has a 100% flat base (ofcourse). What I did is, took a square glass, placed a sanding paper and rubbed the block on it. I did it until every part of the block was polished. Is that good enough?


My WB setup is almost complete. I am going to post pics soon.

[Brians256]

It's probably good enough. Real flatness is difficult to measure and produce. You can look at the reams of threads on lapping to see the evolving approach to the problem. BillA had one of the better solutions, and that involved diffraction.

[ibmkg]

Recently some one told me that since I do not have the 'tech. know how' and that I have not done any math regarding heat dissipation and stuff, I am going to kill my CPU and Mobo. He kept on arguing and he is an Electronics Engineer. It really lowered my morale. Is he right?

His point that struck me was "The heat dissipation from CPU to WB not being enough would yield dire results; you need to do some paper work".

What math should I know?

[bigben2k]

I wouldn't worry about that; it's really hard to screw up making a water block. However...

I'd say "look at other designs", but the cross drilled one is pretty rare, because it take a long time to make. We haven't seen your design yet; I'm assuming that you're cross drilling a whole bunch of small holes (run a search here for a cross drilled block, someone made one)?

Otherwise, be aware of the thickness of your baseplate, where the "baseplate" is the thickness of the copper, before it touches the water. It shouldn't be thicker than 1/4 inch. (btw, the math part is where you calculate the ideal thickness).

If you're not sure about your design, you can create a heatload tester (power resistor) and take some temp measurements. Alternatively, someone here might volunteer to test it, if you ask nicely. Otherwise, I'd say try it on, but keep the computer idling (no CPU load test) until you have a good idea of wether your block is "in the ballpark" or "out to left field".

Your results are also going to depend on the coolant flow rate (determined by the pump) and your cooling solution (radiator-fan combinaion).

Let us know if you have any more questions, and if you have a chance, please post a picture or diagram.

[Brians256]

Ben, he said he's only got three through holes from one side to the other. This is not a xjinn style block.

This block probably won't do well, ibmkg. Your surface area isn't enough to transfer heat quickly. However, it may be "good enough".

The modern CPUs have throttling so that they slow down if marginally cooled and shut off completely if they get really hot. Don't worry. Unless you heat them up quickly, you should get a quick ballpark estimate of how good it is.

Test the temperature with the BIOS setup screen while the CPU is really idle. If it isn't a decent temperature, you'll know that you need to redesign. Very simple. If it is a good temperature, start your OS and run a CPU load to see the high temperature.

Short term tests show how good the block is. Long-term tests show how good the radiator is.

[bigben2k]

Ah I missed that...

Yeah, it'll fall in the same class as the early DangerDen blocks, i.e. Maze 2. Not bad, but no longer great.

Still, there's something to say for a DIY solution.

[ibmkg]

BigBen,

The base plate thickness is about 2-3mm max. I can increase the number of holes if that would help, in my next block. How can I make a heat load tester? Moreover, what would be the math to calculate the ideal thickness?

For rad, I am using a Heater Rad, which comes in small three cylinder hatchbacks (like Suzuki Alto). It is made of Cu. The coolant would be normal car's rad coolant. In addition, for the pump, I have not calculated its flow rate as I am making a reservoir for it from brass.

I am going to post pics as soon as it is in running condition.


Brain,

By increasing surface area, you mean, to increase water flow channels i.e. more holes?


Thanks again, all.

[ibmkg]

One more thing, the fuel pump only generates 1.5 Liters per 26secs. Or roughly 46GPH. Have I failed? Please reply asap too all my questions so as I can either drop everything or proceed with something.

[ibmkg]

Pics of rad & pump+reservior.46GPH just my bad luck again