Making PVC Block

[iggybaseball]

I was thinking about making a PVC waterblock. One kind of glue should i use to inert the nozzles and i also need to cement it to the coldplate. thanks.

[Tiger]

Without trying to discourage you in anyway, I tried this using acrylic sheet around a socket 7 heatsink. I used a twin pack epoxy glue rated -40C - 180C. It was really great didn't leak and all the connectors were really strong. My final test was to simulate a pump failure and I left it on a hot surface for a while. This it failed miserably. All the joints with the metal sheared. Looking at the results my own opinion is that the rate of metal expansion exceeded the plastic expansion causing the joint to shear. If you feel confident about the residual water keeping it sufficiently cool to avoid this then go for it. It would probably work if an auto shutdown was incorporated so that the proc never gets to this kind of temp. I just figured it was cheaper to get a commercial block than risk 4 times that value in hardware.

[Diggrr]

Hey Ig,

I've been using the pvc endcap waterblock for a year and a half now, love it. My pump did fail (actually a terminal had a loose screw) and I booted into windows, logged onto the net and was checking e-mail before I noticed. The block held fine.
There's a couple of reasons for this. There's only one glue joint versus one made out of acrylic, the built in hold down keeps it held securely against the coldplate, pvc has a much higher temperature rating than most plastics, and it is slightly softer allowing more flex (and machinability). It's also maleable to some degree.

Now to answer specifically...Use Aroldite or JBweld. Rough up the plastic with 100 grit sandpaper first, and I like to clean the top of the copper plate with a sanding too, same grit.(alcohol too)
I just tap holes for the fittings, but Aroldite and JBweld would work fine too...Leave it for 24 hours, don't push your luck. You want it to cure well before adding water/heat.

*edit* bacause of the flexible nature of the plastic, you drill a hole slighty smaller, and just force it to screw in. Take it back out, apply the adhesive, and screw it back in.

[The Overclocker]

i have a waterclock held together with aroldite (long setting stuff is best) and i once booted up without the pump on, i got into quake without noticing, it still works

[Silver]

I also tried a cheap aluminum heatsink surrounded with plexi and had no problems with it however I would expect failure in the event of the pump not working correctly. In fact it ran 1c cooler than my dd 2/1 block. I must also state that it had 3 inlets and 3 outlets. Flow was fantastic and allowed me to open up and additional 50 holes on the evap head. Used a 500 gph danner pump.

[iggybaseball]

thanx a lot.

[Silver]

I think that a copper or possibly a copper/aluminum pin block with better fin/pin design than the one I used would perform ever better. I have considered (and may do ) a glaciator with soldiered plates. This of course would give much increased cooling surface and negate the fear of a seam (glue) giving out. FYI I ran the voltage on this chip up to 2.35 just to see if the heat would cause problems and it did not.

[Stephen Castles]

heres what i made :

[Stephen Castles]

:

[Silver]

Yes, I have seen this and think it rather ingenios. I look at a pvc cap and see a pvc cap, but I am learning to look at things differently. I saw one article wherein they did the same thing with a heat sink attached to the plate. Other than that it was identical. The article failed to state test results though. Have you read the article posted on the front of overclockers today? He used pipe to surround a heat sink. Now this opens a whole new area to those of us without the machinery requied to groove copper blocks. Cut the pipe down and soldier it to a plate in the design of your choice.

[Stephen Castles]

yeah, i saw that SK6 water block, pretty neat, probly has a TON of surface area. i made the block that i showed above, and i am using it at the moment. MBM5 says 33c, however, it's a mobo in socket temp probe so it reads a bit high, the block is temporary, but it performs well enough. this block cost me about $5 because i had the copper already. I am working on a new design, as soon as i get some more copper bar stock it will be done.

[Silver]

Just a note to those that might read this. I got my new shuttle board and modified the socket on it yesterday. I was looking at putting the plexied heatsink on it yesterday when I noticed what appeared to be a little crack on the edge attached to the heatsink. As I looked closer I found about 8 different stress cracks located all around the edges attached to the heatsink. Needless to say this watersink will no longer be used. I have some copper on the way though.