Bullet proof glue for water blocks

[pudgy-duck]

Hoot;

Last Nov I bought an Athlon XP. This meant that I now had a T-bird 1.2 Gig just laying around on my work bench. Between Xmas and New Year's I was bored, so I made a direct contact water block for my T-bird. Hey, It was just a spare. I took a plug for a 4" pvc sewer cleanout, and used the square nut portion for the water jacket. Cheap, readily available, and I could use existing and proven solvent bonding (Oatey purple primer, and Oatey cpvc/pvc cement) for almost all joints. To bond the pvc water block to the cpu itself, I used Epoxy. I also used Epoxy to "pot" the top of the cpu (bridges, surface mount components). It has worked since New Year's, but I am not 100% comfortable with the Epoxy as a bonding agent. Doing research, I came across an adhesive called ITW Plexus, which is a methacrylate adhesive. According to this url http://www.adhesivesmag.com/CDA/ArticleInformation/features/BNP__Features__Item/0,2101,2457,00.html they are using this stuff to glue a car together! I thought this would be the bullet proof bonding agent for the mating of the pvc to the cpu. However, I cannot find a consumer source for this stuff. I thought that ITW PLEXUS MA300 or MA301 would work, but can't find it. What do you know about this stuff? Is this so nasty that it is sold on an industrial basis only? If you know anything about this stuff, do you agree that it would work for this application?

Hoot, or anyone, I would appreciate any info on this stuff. My prototype encapsulated T-bird is working out so well I am considering doing this permanently to my Athlon XP, but I want a bonding agent I feel better about than Epoxy.

Pudgy Duck

[Diggrr]

Plexus is nasty stuff. I've used it at work to bond laser housings together when the supplier sent them without their sonic weld.
It's thick, requiring a pneumatic glue gun. Ours was a two part syringe style tube.
It also required the use of a filtered air exchange cabinet for drying, as the fumes were sickening and burned the eyes. It smelled like superglue and rubber cement mixed...yuck.

Just use marine goop for the joint, it worked so good on mine I had to dremel the plastic block apart to remove it from the cpu.

[pudgy-duck]

diggrr;

Thanks for the info. You confirmed what I suspected about Plexus. I recall reading in another thread about your experience in tearing your (I assume direct water contact) water block apart. Can I ask why you where tearing the water block back apart? Did direct water contact not work out for you? This is my first experience with water cooling. Seeing a T-bird that even with a case temp of 73 F, ran at 107 to 113 F cpu temp, now run at 77 F cpu temp was SOMETHING! Now I am hooked on water cooling. I tried direct water contact because it was inexpensive and sounded like it would yield the best cooling. If you have tried both direct contact and copper water blocks, which do you prefer? What are the pros and cons of both? I am undecided which way to go with my Athlon XP. I hate to spend $40+ for a DD maze2 if direct contact is better, and on the other hand I hate to seal up my XP if a copper block is best. Appreciate any input you may have.

Pudgy Duck

[Diggrr]

Direct die cooling almost cost me my t-bird. I used nail polish to cover all the bits and goodies on the substrate (never cover the core). After a week, I did an emergency shutdown because I could hear a high-pitched sizzling sound in the waterblock. That was high frequency electricity pulsing through the water. (I build tesla-coils, and know what it sounds like)

After tear-down, I found the polish had bubbled around one of the 5 components next to the core, and the bubble tore letting in water. That component turned to charcoal, but didn't die.

I'm back to using a copper waterblock now, as I can't afford to mess up the processor since the plant layoff. Besides, it didn't really drop my temps much anyway. (I think about a 2-3C drop off the load temp vs. copper block)

I don't know if the polish was bad, or the component was bad when I started. Tom Leufkens recommends using nail polish for his direct die blocks though. He also says that you shouldn't use regular epoxy, cause it doesn't handle prolonged water exposure. I would bet a marine epoxy would.

In short, it's being done by a few here, and it's working to lower the temps. I might do it later, after I'm working again and can afford to replace it should it fail.
Here's a pic I saved of the waterblock, note the dremeled cutout to keep off the L1 bridges. The corners are cutout so I didn't have to remove the pads, and could reuse a copper block later.

[pudgy-duck]

diggrr;

Thanks for the reply. Like I said, I am 100% comfortable with the pvc to pvc bonding. But I did use epoxy to cover or "pot" the goodies on top the CPU, and to bond the water block to the cpu.
I am NOT comfortable with that. The main thing I was looking for was the difference between direct and indirect contact water cooling. If 2 or 3 degrees C is the only dif, I think I will see about a indirect cooling copper water block for the XP. If a totally copper water block is properly soldered, it will never leak. Besides, the XP runs cooler than the T-bird to start with, so that will make up the difference between direct and indirect cooling. Better to be safe than sorry, because potentially there is more at risk here than just the cpu. If the cpu fries, it could take out the MoBo. If it leaks, there are some expensive cards sitting right below the water block. I don't regret doing this little project, I learned A LOT! But I think I will go with a commercial water block for the long run, probably a DD Maze2.

Good luck with the job situation. I've been there and done that, and I know it is not a comfortable situation to be in. I was laid off for 24 weeks in '87. It was a tough time both emotionally and financially. Don't let it get you down.

Pudge

[Dissolved]

thats a very nice water block...
man i wish i was still in school to use metal shop cus i wanna make my own block badly ;)

btw how did you seal the water block to the cpu?

[Diggrr]

Dissolved, I used Marine Goop to fasten the block to the cpu. I made a plexiglass hold down that used the 4 bolts, and pressed the block to the cpu. It also took the weight of the block/fittings/tubing off of the socket and cpu pins. Wait 24 hours for curing.

Pudgy-Duck, I'm not trying to discourage direct die cooling...heck it gave good results and only cost $7 to build. Just make sure your comfortable with the sealant you use on the cpu's bits. You could test a sample in water. The processor is a year old (bought it when the 1Gig first came out), and could very well have been a failing component instead of the coating being bad. I was thinking that the Marine Gel Resin for fiberglass would work excellent for this, or a Marine Epoxy paint. (I know these could take the water)

I'll probably try again, the 1G t-bird is down to $93 bucks. Usually I have more guts than brains.:D

[pudgy-duck]

diggrr;
I have been thinking some more about this. You're right, the pros are inexpensive and probably the best heat transfer if you keep the Gallons Per Hour up. So far the only con is the suitablility of the "potting" and cpu to water block bonding agent. I looked at http://www.epoxyproducts.com/marine.html and found two products that look promising. Bio-Seal 192 marine epoxy and Nova Clear. However, they are both $30+ for 40 ounce "Trial" sizes. Kinda cost prohibitive. I have a prototype block right now that is working great. Maybe I am being obsessive about this, but I am not quite sure about exposing epoxy to water on a long term basis (months and possibly years). I am just one step away, so I am going to keep researching this. What I really need is an adhesive/bonding expert.

[UserName]

I have never sealed mine and it works fine. do most people seal the chips? I just use distilled water and AF. Never had a problem.

[pudgy-duck]

UserName, On the T-birds, there are surface mount resistors and I think the larger items are surface mount capacitors on the top surface of the chip, besides the bridges. On the XP they moved the surface mount components to the bottom of the chip, and only the bridges are on top. But either way, the bridges are connected electrically tothe cpu, and on the T-bird, the surface mount components are also connected electrically. So, most people that I know of, and admittedly it's not many, seal those components and bridges from the coolant and only leave the cpu die's top surface area exposed to liquid.

What kind of cpu do you have, and how did you bond the cpu to your water block?

Pudge