Project Update (Pics) 56K unfriendly
 

Hi guys,

I thought I would share with you where I am with my project.

I wanted to focus on the water cooling circuit and how I would measure and monitor temperatures as well as how I tested for circuit integrity.

The critical water temperatures we would like to capture are:
1. The temperature of the water at the pump outlet which equals the temperature of the water entering CPU1.
2. The temperature of the water exiting CPU1 and entering CPU2.
3. The temperature of the water exiting CPU2 and entering the chipset block.
4. The temperature of the water exiting the chip set block and entering the radiator.
5. The temperature of the water exiting the radiator and entering the air trap.

In addition we would like to moitor temperatures at the bottom of the case, the top of the case, as well as the temperature between the radiator and it's shroud.

We also need to capture and monitor temperatures aof each CPU.

A compunurse with 2 LED displays will monitor the temperatures of each CPU respectively.

What we want to focus on today is the use of a Digi-Doc to montitor the other temperatures.

Due to the size of the case the Digi-Doc's leads had to be lengthened considerably
http://www.extrememhz.com/images/Com...idoc/ddoc5.jpg

In order to read the temperature of the water the Digi-Doc's thermisters were replaced with a smaller glass bulb encapsulated ones. The new thermisters were then inserted into holes that were drilled into the fittings that were going to be used. The holes were then sealed with multiple layers of marine goop to ensure a good seal.

Here is an example of a thermister inserted into an elbow fitting
http://www.extrememhz.com/images/DCPower/tempssr5.jpg

In these pictures you can see a thermister inserted into the pump outlet fitting
http://www.extrememhz.com/images/DCPower/tempssr9.jpg
http://www.extrememhz.com/images/Digidoc/tempssr11.jpg

In this pictures you can see a thermister inserted in an end cap fitting http://www.extrememhz.com/images/Digidoc/tempssr4.jpg

In this pisture we show the test setup. In this phase we are looking for leaks. The circuit was run successfully for a continuous 48 hour period.
http://www.extrememhz.com/images/08-...TED/pjset1.jpg

The coolant used during this phase consisted of distilled water, Silkolene ProCCA, radiator leak detector dye, and methylene blue.
http://www.extrememhz.com/images/08-...ED/pjset12.jpg

Here is a closeup of the water blocks showing the elbows that are fitted with the thermisters that will be used for temperature monitoring.
http://www.extrememhz.com/images/08-...TED/pjset2.jpg

In order to ensure that the electronics are protected in the event of a leak ocurring in the future, all PC boards will be protected by applying sveral layers of conformal coating.
http://www.extrememhz.com/images/08-...ED/pjset13.jpg

That's it for now.

Cheers!

What is Silkolene? As far as my searches go, it looks like either an antifreeze for racing, motorcycles or snowmobiles, but is a british product?

AH! I got it! (and in the USA too!)

Neat, but is it a corrosion inhibitor? (for those of us that mix Al and Cu)

According to Fuchs (a very well respected German Company), ProCCA is a corrosion inhibitor.

This product is a surfactant just like "Water Wetter" or "Purple Ice". I find it to be preferable to the aforementioned products in that a lower concentration can be used e.g. a 1% solution of ProCCa is recommended Vs a 3.125% solution of Water Wetter.

For best cooling results (in an environment above the freezing point) it is always preferable to maintain the highest ratio of water to additives possible. Additionally, the product has a relatively pleasant smell and for the aesthetically minded, it is blue.

The down side is that it is relatively expensive, but I only paid $12.00 for 8 ounces of fluorescent dye Vs the +/- $15.00 that is commonly charged for 1 ounce by on line shops...so I do not feel too bad about the overall cost of the coolant mixture.

Cheers!

Tip
 

BTW guys,

If you want to effectively color your water blue. Stay away from products like food coloring. Use 1 drop (and I mean only 1 drop) of methylene blue. This liquid is commonly used as an antiseptic and can be found in any drug store where the alcohol and iodine products are kept for less than $1.00.

It is unabrasive and non-corrosive (will not damage your pump etc.). It will remain in a liquid form and not separate into particulates either.

Don't thank me...just throw money! LOL

Cheers!

Nice... very clever.

I'll have to score some Silkolene ProCCA for my windshield wiper fluid. :D (don't do this at home, boys and girls!)

What's the purpose of the Methyl Blue?

Another silly question for you: is there a particular reason why you have so many elbows?

Also, since you mention running the 48hr test, did you consider running a pressure test instead?

Nice pics hmale, looks like your killer system is shapping up nicely.

Ok, I've got another question...

If the distilled water is clear, the Silkolene ProCCA is blue , Methyl blue is blue , and the UV dye is whatever color it is, how come the coolant is green?

Hi Ben,

The explanation for the use of the methylene blue is right above your last post.

I have learned alot from and have been challenged by your posts and questions Ben (great job!). The question about the elbows is not silly at all and should be explained. As we all know blocks, headers, tees, elbows et al all inhibit flow to varying degrees. In designing this system I knew that I would have to use more fitttings than I would like. What I did was calculate resultant flow taking into consideration the number and nature of the blocks I was going to use, the distance the coolant must travel through the circuit, the resistance to flow relative to distance travelled, number and nature of the fittings, and the impact of the radiator design. Once I calculated the expected back pressure in terms of feet head, I plotted the results against the pump's performance curve realtive to back pressure. My findings were that the resultant flow would be in the 300-325 GPH range...more than sufficient for what I am trying to accomplish.

I did not test circuit pressure. Since the temperatures I was reading were very acceptable I did not feel that pressure was going to be problematic. If the temperatures would have been higher than anticipated I would have tested the circuit's pressure in PSI's.

Cheers!

The UV dye is yellowish..yellow+blue=green.

If UV dye is not used, the water would be blue depending on the concentration of meth blue used. If UV dye is used, the intensity of green is regulated by the meth blue.

Cheers!

and cheers to you too!

Actually, for the pressure, I meant, testing the system for leaks, by pumping some air in it: that way, if it's ever going to leak, you'll know. It seems quicker, and far more reliable (to me) to test it that way, than a 48 hr test run.

Did you consider using blue UV dye? (I know, it's just a matter of personal preference, but blue all-the-way would have been my choice). BTW, what's your source for the dye?

I still maintain that it's a top notch comp, for the money. Very nice job, with cleverly selected components.

Hi Ben,

OK...now i understand your question. You are right, pressurizing the circuit would lead to a more effective test of the circuit's integrity. I did not do that though (where were you when I needed you?) LOL.

Anyway, my preference is green (I also like vanilla ice cream). One of my friends in the engineering community works for an electronics company that manufactures leak detection equipment. I was able to obtain some leak detector fluid (UV dye) at cost from him.

Cheers!

I prefer black cherry:p

How do you apply the conformal coating? What thickness is needed to adequately protect the PCB? Do you think the heat generated by the PCB will affect the coating? Or vice versa, will the insulation by the coating affect the PCB?

Many questions!

First, you have to mask off the CPU socket, and all the other sockets for that matter (ram, AGP slot, PCI slots, HDD, floppy, ports, etc...). You should also mask all the jumper pins, and any other area that you plan to put a heatsink on (northbridge). Then spray like you're happy, on both sides.

A single coat should do, but if you doubt your spraying skills, go for two coats.

No. The heat comes from the chips, but won't affect the coating.

Yes, The coating will affect the heat transmission, so don't go coating everything!

There are many sites that have an article on this. I'll see if I can dig one up for you.

Ben,

You are the man!!

Thanks for the link.

There is some heat given off by the PCB, but especially on the backside of a chip mount. Feel the backside of that GF4, tell me it doesn't sting!

Looks like some good stuff, where did you find the coating here in the US?

Now I know that you didn't read it...

It says: www.graybar.com

There's also another link in the article, but that one, I'm gonna let you look up all on your own...

As for the VC, You really don't need to worry about it, but make sure that you cover the pci connector. I prefer using black electrical tape myself, just because it comes off easy, and doesn't leave any goop behind.

Actually, I did read (most of) it, so that's not fair to say. I was referring to the acrylic spray that hmale is using. And let me quote the article on that one:

The conformal coating product I used can be found at the following site:

http://www.cir.com/mgchemicals/coatings/coatings.htm

If you do a google search for MG Chemicals you can find other sites that sell a wide range of their products.

Cheers!

Thanks! Do you really think it is necessary to use the coating however? Might not the increased PCB heat outweigh the possibility of a leak? Then again, you do have a lot of joints near the sockets. Well, in any case, please tell us how that conformal coating works out... I assume you will be simulating a leak for testing purposes?;)