Interesting Failure Mode for WB Top

[zackbass]

Normally I wouldn't make a big deal about a WC system failure but the cause of this one made me say "What were they thinking!" I don't know whether this is a chronic problem or not but it seems like pretty stupid design anyway.

In moving my system home from college for the summer a kink worked itself into the line for my Northbridge block that effectively cut off flow to the block completely. I didn't notice it, by some miracle the system ran fine for several days, but it finally BSOD'd and I opened the thing up and found the kink. The block (DangerDen A8N block) looked like it got a little toasty but was otherwise fine so I went ahead and restarted the system.

Again it worked fine for a while until it BSOD'd (the famous "not less than or equal" error any experimenter gets familiar with) again. I just went over to try a restart this time but it was a no-go. Nothing, no power on at all.

I opened it up and found the mobo quite wet around the NB:


The waterblock looked to be in pretty bad shape so I pulled it apart:



Notice the nice ring formed by the block's O-ring. Next up is the warp on the plate:



And more detail on the O-ring indentation, measuring from the edge to the center (total depth) of the deformation:



My theory on the failure which I'm quite sure is correct is that without water the block heated up to the point that the block's top plastically deformed to the O-ring. Since the ring wasn't being squished anymore its ability to seal was badly damaged and with flow restored began to leak.

I was wondering, has this happened to anyone else? It just struck me as a pretty bad material choice for the application.

[jman1310]

oh yes
there is/was a thread on this server that polled uesr's w/c failures
acylic melting is one i've seen before

did you have a splitter in your system or did the whole system stagnant and overheat?
or graphicly...
pump>cpu>chipset>gpu>pump
or
pump>cpu>splitter>chipset>splitter>pump
xxxxxxxxxx>splitter>gpu>splitter>xxxxxxxx

what other blocks do you have in your system?pump? what coolant were you using? any luck running it now?

[jaydee]

I have always tried to make my screw hole pattern as close to each other as possible to keep the top flat on the block to prevent warping. Looks to me like those holes are spaced way to far apart for a Lucit top. Polycarb or delrin would have been less likely to warp so bad.

In the end though 100% metal blocks with proper screw spacing or soldered together like the MCW6000 series is the way to go. I think that block has an optional brass top now worth looking into.

[zackbass]

I've got everything back up and running now, there were several close calls along the way but it ended well.

I actually run a MCW6000 series block on the CPU because I don't like the Lucite tops but there wasn't much choice on the chipset block. I really like the construction of the MCW6000 except for the short single-beaded barbs.

The system runs in parallel through a distribution manifold so none of the other parts of the system got starved. The pump is an Iwaki MD20 and the other blocks in the system are a Swiftech MCW6000 and one of the huge DangerDen Geforce 6800 blocks.

I was able to rehabilitate the block with some careful sanding but decided against using it again. The ORB thread seals were pretty screwed up too and I didn't want to take the risk. I'm going to go with the original chipset fan for the time being.

[phpCypher]

you are going to see warping/seperation like that anytime you have a junction of 2 dissimilar materials in an extreme temperature condition.

Sucks

Aluminum heads pull away the same way when they are connected to an iron block in automotive applications.

Quick question, which mobo is that ? Those 8 sata connectors look like the business !!

Good luck, and I hope you get your next system up soon

[Talcite]

Looks like the Asus A8N SLI premium to me...