DT Direct 2011 waterblock

[DetroitAC]

My next new waterblock, this one mounts directly to the IHS, sealed with an Oring. This is an LGA2011 specific block, will not work with any other sockets. I have not tested this 2011 version, I did all the testing and development work on my 1366 test rig. What I learned on 1366 is applied to this 2011 block, hopefully others can test this and report it's performance. This version was "finished" first to try to get it included in Stren's roundup of current blocks, 1366 version will be finished soon. My testing on the 1366 version showed better performance than either Apogee HD or Sniper with PK-1 TIM.

Restriction is less than Apogee HD, more than Raystorm, no where near as low as Sniper. I tried a lot of things to get the restriction down but performance suffered. Ultimately I chose a production geometry that maximizes thermal performance, and the restriction is what it is.

I tested for three months (1366 version), and had bad leaks between the block and the IHS when I would forget to use an O-ring. No catastrophies, no ruined hardware just some excitement. I used the old socket 775 prototype of this concept for two or three years with no issues. The processor IHS on that 6600 was lapped to the copper, so I have no long term results with water against Intel's nickel plated IHS. I have a feeling Intel knows how to keep a part clean before it gets plated.

I should mention, some of the credit for this block goes to XS member Wez. We were discussing new concepts in a thread, and I mentioned that I've done this for socket 775, performance was a little worse than EK Supreme, and used it successfully for a few years with no problems. I did not have thermal performance as good as this current concept, and I speculated about machining fins directly into the IHS. I thought it was a bit too extreme to do something that risky for watercooling. Wez offered to loan a Bulldozer to me to try it out, never took him up on it, but it reversed my thinking about the idea. I was in the process of screwing up my courage to try it, and was thinking about the problem when this current idea occurred to me (while playing with a garden hose on concrete). I gave it a shot, got decent results and decided to do a full development effort on the concept. Thanks for the brain kick Wez...
Attachment 129715
IMG_1944crop.jpgIMG_1943crop.jpg

[stren]

Excited to try this one

[SugarJ]

Wow. I'm sure that works great by eliminating the block between the water and the IHS, but I foresee a few customers crying because they didn't mount it right and flooded their systems.

[Manicdan]

if we need guides on how to instal tim, then its clear this product is not ready for the masses.
block should be shipped with boxes, bags, and RMA forms for various hardware

love the idea, scared of the results.

[El Mano]

This may take WC to a new level, I long to see results, the idea looks briiliant to me!

[DetroitAC]

Thanks Alex, cheers

Yeah guys, this concept isn't for everyone. Even if it does work, many will not be willing to try this. That's understandable, hardware is expensive. This concept uses an O-ring that is a face seal from the block to the copper base, the block and base are clamped together with 4 screws. This is exactly how I would describe the DT Sniper, the Koolance CPU-370, the EK Supreme, the Raystorm, etc. In this case the "base" stays with the mobo when you take your block apart to clean it.

The block is poke-yoke to install in the correct way or leave a large gap between the processor and the block. It has some little features that center the block and O-ring on the IHS, and do not allow it to go down if you install it 90deg incorrect. When you install it correctly, the features engage the IHS edges, it feels solid and purposeful. When you install it wrong there is a huge gap and the block is free to spin and float around on the processor. We'll see what reviewers think of it.

[bds71]

interesting idea to say the least. i know with the current mount there is no way to do this, but (if you could) how well would this work with direct die? (ie: removing the IHS) i've heard of folks using liquid metal under the IHS and their temps dropped 20C on the cores. since water is no more/less conductive than liquid metal...well, you see where i'm going i hope. but again: very interesting idea - and cutting fins into the IHS to improve performance even further....nice....<drool>

[Vetalar]

i've seen some direct cooling solutions from different enthusiasts but they perform worse than top-end WB on the market. or even worse than homemade piles of copper and plexi which we call waterblocks. now i see brilliantly made WB and very exited about results!

[Manicdan]

i've seen some direct cooling solutions from different enthusiasts but they perform worse than top-end WB on the market. or even worse than homemade piles of copper and plexi which we call waterblocks. now i see brilliantly made WB and very exited about results!

the surface area is quite low, so flow speed will really matter. however flow speed on smaller blocks is already quite high where thers more surface area. so ignoring the surface area point, the only other difference might be how much material there is between water and cpu. those low surface area blocks with high flow are still only a few mm thick. i think what we will see is a constant increase is cooling quality as flow rates go up, where as with normal blocks its a minor change after a certain point. it might be a real winner if its being fed some serious gpm.

i think what we will see is that for a low wattage build, this could do great, and for idle temps it could be right at ambient. however needing 50w of pumping power to get there sorta would defeat the point of a low wattage overclock

[st0ned]

Any version for LGA1155 in the near future ?

[El Mano]

I like the "do it at your own risk" aspect of the block and the oportunity it gives to users to go beyond just placing the block on the mobo.
I mean, with other blocks all you have to do is lock them in position and that's all about it. However, this block will probably require extra work from the user, which is great IMHO. Like, not just lapping the IHS but perhaps trying to engrave micro channels with clockmaker tools or - who knows.
I can see people creating their own IHS out of copper sheets, with fins and channels and what-not.
Great idea Erik!

If you ever produce a block like this compatible with older sockets I'll buy one!

By the way, is there any dielectric liquid, heavier than water? Sinking the mobo in such liquid would minimize damage if there's a leak as water would float over the isolating liquid, I guess. Am I correct?

[DetroitAC]

Leaks are not much of an issue, I think it's hard to convey this, and I can understand the skepticism. I think I might make a video about mounting the block, I'll be as sloppy and fast as possible about mounting the block.

I'm currently working on Sniper for AM3, but Direct-1366 will probably be finished next, and then Direct-115x. These will be more work and more complicated than 2011, the 2011 socket with built-in threaded bosses and mobo stiffener made this one easy.

[zalbard]

Interesting idea!
Have you measured the performance compared to the other blocks you made?

[tool_462]

Interesting idea!
Have you measured the performance compared to the other blocks you made?

From OP:

My testing on the 1366 version showed better performance than either Apogee HD or Sniper with PK-1 TIM.

[Vetalar]

i think laser cutter with apropriate settings can easily make engravings on IHS - it will increase surface, create unique CPU design, made this CPU incompatible with any other cooling solution and transofrm boring RMA process to real adventure

[paulbagz]

i think laser cutter with apropriate settings can easily make engravings on IHS - it will increase surface, create unique CPU design, made this CPU incompatible with any other cooling solution and transofrm boring RMA process to real adventure

!!!!!!!!!



Lost my haha.

-PB

[DetroitAC]

I made videos describing some block features, and showing mounting and draining.

[meanmoe]

This is really awesome. I think I'm going to have to have one.

[El Mano]

If you ever make a 775 block like this I'll be the first in line

[Vetalar]

after this videos i have only one suggestion: thumbscrews for block mounting.

[kolombo]

Let see some results .This idea is too old and abandoned.

[Nathanial]

Very interesting, I'd be delighted to own one of these for my 1366.