possibly a complicated waterblock q
 

OK, i'm designing a waterblock to mount in a very non-standard motherboard. every measurement i've taken is off big time compared to every white paper i've looked at. anyway there are cap's on all sides of the zif socket. if i need to make a 2 tier system so it will actually mount and not hit the caps can anyone think of a problem that i might have with a design like this. in case you dont get what the problem is here is a kinda simple drawing of what it might look like from the side.

More info! What kind of CPU? What mobo? Hard to say without seeing what it will be applied on. If it is AMD you do not need to make the block any bigger than the CPU body. No need to go outside of the socket. Make your hold down with the top peice so it goes over the caps. I don't know though without knowing what it will be mounting to.:shrug:

There really isn't a problem with making a waterblock like that, but if your block exceeds the dimensions of the CPU, or , more specifically, if your block exceeds the AMD defined restricted area around the socket, you might re-think your design: you shouldn't need to exceed the CPU, except for mounting.

Now is your mobo so not-standard that it doesn't meet AMD specs?

Are you doing a socket mount, or are you using holes around the socket?

I agree, just use 3" x 2" material. No problems with capacitors and no worry from the hinge area of the socket hitting your baseplate.
Of course you have mounting holes in your system board, correct?

hopefully these will answer your questions. the mounting holes do not meet amd's spec's in fact they are waaaaaay off. there are cap's in between the mounting holes on 3 of the sides that are large enough i'm going to have to make the block in an H shape (top view) when finished. to be honest i was more concerned about the step than anything else but since it's just direct die cooling that hopefully will not be a problem.

Hasn't direct die cooling been proven to be a bad thing? Last I heard it was worst than a regular WB, and could also corrode you CPU's core away.

But if this is a AMD board then it should have the AMD speced holes or NO HSF on the market that uses those holes will fit it. Unless this is a server type board with specialized holes for HS's made spacifically for that board.

If your going to be elusive about the hardware being used then don't bother asking questions we cannot answer. ;) Can't help design a block unless we know what it is mounting to.

Direct die cooling is not a viable solution

There are many problems with it. Understand clearly that it's only meant for testing purposes, for data gathering of heat dissipation properties of a CPU.

If you don't get that, I'll be more than happy to detail it for you.

Why isn't it an option? I know of a few people who had very good results with it. Sure there are problems, but thats the risk taken, but why would it corrode cores?

sorry i said direct die, but was incorrect. yes a regular waterblock will be used. when i said direct die i was meaning an injection directly to the block over the die instead of directly onto the dir or even a waterblock that flushes the coolant over the die. as for the motherboard here is a link to it however this was taken before the final version was released and does not show all the obstructions on the mb itself that you have to deal with. it is a sff mb for the sn41

http://www.shuttle.com/german/prod/mb/600/fn41.jpg

hope this gives a little more info for everyone.

Isn't that that Shuttle mobo they use in those cubes? Wonder why the holes are so far off. Almost looks like Intel hole pattern.

I would just make your block normal like and just make the TOP spread out to the holes. that way you do not have to worry about the parts on it. Or just make the block fit and make a acrylic/lexan top for a hold down.


Also the core will corrode because it is aluminum. And the water flow itself will errode it away if high enough flow is used, which is should be with direct die cooling.

Those holes are out of spec, all right! But those caps are not. Mount your block the other way...


Direct die cooling is not viable because:
a) the core absorbs the coolant and [edit] no no no... get it right Ben! see below![/edit]

b) if your pumps stops for any reason, your CPU will die right away, and I mean instantly!

where's the evidence for this?

and if so,what's stopping the core from absorbing artic silver?

I have my own theory on what may be causing the failures, but I don't think it's a permeable core.

Oops, my mistake!

What holds the die to the substrate is what absorbs the coolant.

yea that's what I suspected too, as well as the "pumping effect"(core physically flexing with changes in heat) that billa metioned in one of his articles sucking in water through failures in that sealant, allowing water to corrode the BGA joint under the die.

Volenti, I know you have experimented with direct-die, so I would greatly value your input on this thread: http://forums.procooling.com/vbb/sho...&threadid=5665