It's a color wheel!!!

Scarlet-Tech
It's a color wheel!!!

Nateman quits, going to teach kindergartners now!

EVGATech_DaveB
I'm not sure how, but this MUST be something for making air-cooled waterblocks, has to be.

You really want that block huh?

nateman_doo
yea, right?  He just won't:

Can't blame the guy though. The idea of the air cooled water block is exciting and promising. I still love the idea.

i have not given up on the idea, trust me.  Just too busy making projects to do R&D stuff.  I am building a second monster machine that I can devote to R&D projects, or work on different batches at the same time, so I do plan on working on this more. 

Looks familiar.  The 3 rings are chambers where you can put parts in, set a weight on top of them, and then the base table rotates or the top also rotates and it polishes / laps.
 

AnonymousGuy

Scarlet-Tech
So, myself and a few other users currently have the EK Supremacy EVO CPU block and we are getting subpar performance.  I do love EKWB and the awesome things they provide when it comes to water blocks, but this block is not performing as we had hoped.  This is a call to Nateman_doo, to see if we can rally enough people to fund a solid CPU block for the x99 platform that will be able to perform top notch on these rigs.  
 
I know Nateman is ALWAYS under high demand, so if you are interested, be patient and lets rally together to help him out so that we can get some fantastic blocks on these fantastic systems.
 
Currently, I have tried 2 separate Supremacy EVO blocks.  The first couldn't keep the temps of the 5960x below 82c, no matter how I tried to mount it.  The second, I was able to get it down to 73c, but both blocks have flex to the bottom of them, and I am thinking that this is causing contact issues.  I would like to see what Nateman  an provide, as I want my system to run a good consistent temperature at all times. 
 
Who is with me on this?
 
If I have to purchase a cheap Motherboard, and another CPU, I will do what I have to... I will be looking for a dead board and cpu if that is necessary. 
 

On this subject I don't think the waterblock is really the problem.  If you compare any of the blocks from the Supremacy to the 380i to the Raystorm, the temperature difference isn't that significant (few degrees C).  So privately (well not so much now) I've asked nateman if he wants to go all YOLO with me and use his CNC to shave off at least 1.9mm from the IHS on a second 5960X I have sitting here.  FirstCut is making a socket fixture that I threw together in 10 minutes of Solidworks and I'll have that by the end of the week and from there it's up to nateman if he wants to give it a go or otherwise I have to find another machine shop locally that can do it.
 
I also got the new nickel Heatkiller IV Pro waterblock from Aquatuning so want to see how that does compared to my Supremacy non-EVO gold.

I like your style. This will be very interesting to see the reaults of.

seta8967
 
I believe the block is the issue, I got Natemans Z97 block, and the Evo. Scarlet was having issues with his block and I decided to test out a new EVO that I bought on a 4790k. Got the block, and noticed it had the same issue as Scarlet and had this off shaped design, got a new one and it was flatter. The temps were roughly 5C difference. It's not a lot, but I put on a Supremacy (non EVO) and it runs a little worst than the EVO, finally did the Classified Supreme, and it of course ran the worst.  Overall a 5C improvement on top of the EVO, is a significant change for me.

On the subject of a better waterblock (because I do think that the EK base plate can be made better through tighter mfg. tolerances), I emailed EK and asked for the Solidworks part file from them - unsurprisingly "no". 
 
Anyways, what I'm getting at is that I don't think if Nateman were to make a waterblock he would really need to/be able to come up with a better design than what's already out there.  I would simply take the design of an existing block and manufacture it to exacting tolerances - to the point where you wouldn't even need a o-ring gasket (which deforms the baseplate) due to the surfaces being so perfectly polished they seal themselves.
 
That was plan A - making a Supremacy base plate out of .999 silver, but it's just not very practical for the cost.  Plan B which is the current plan was to knock down the IHS, which should bring down overall temperatures by quite a bit more than a better waterblock would.

I kinda wished the 5960x used thermal instead of solder, so I could delid it..

nateman_doo
Man, as if i didn't have enough stuff to do :P 

   Ill put in some volunteer time for ya Nate. For the EVGA community, that's not a problem!!! :)

Scarlet-Tech
Over on OCN, there is a lot of people curious why I lapped the block instead of the CPU, saying I ruined the block.  The block performs exactly the same as it did before lapping, and the TIM Spreads MUCH better than it did before.  Anyone that says lapping a Haswell-E is a good idea probably isn't considering the warranty.  $45 to replace the copper base of a waterblock vs replacing the whole CPU, I would rather take my chances with a little bit of copper.
 
I am kind of surprised that it isn't performing worse, like they were saying over there on OCN.  It is my first time lapping anything, and I actually enjoyed it, and may do this to a golden CPU if I can ever get one.

 
    I actually got a good chuckle outta this post, thanx scarlet. Please tell us why they think it will ruin the block and perform worse. I really cant wait to hear it. The whole point of lapping is to smooth out the imperfections so you get a closer metal to metal contact. You end up using less TIM too. Its not like your cutting the block down, your taking off a micro layer. I do it with my heatpipe direct contact cpu coolers like EVGA's SC cooler, CM Hyper 212 EVO, Zalman 12x and my Xigmatek Balder. Although, I will say the finish on the EVGA and Hyper were very bad and needed it to be done before you could even use them, imo. The Balder had probably the best finish ive ever seen on a cpu HS/F unit and the Zalman was pretty close to that too, so those two only got nano compound/polish just to clean them up. As far as lapping the cpu IHS, I don't do it and def do not advocate it. Intel are tough with their warranty terms and stick to the rules absolutely. This is especially important to remember now that they are integrating so many things on the cpu itself. Of course, if you just have money to burn then go enjoy yourself, but most of us don't.

I don't mind milling the CPU down.  Just don't want to be responsible if it doesn't perform.  I have no idea how much induced magnetism could happen with a carbide end mill whirring 4 thousand times per second over a super sensitive electronic part.
 
I will be getting to these soon.  Currently finishing up some ASX-11 blocks, cutting some 680's blocks and then on to the stingers/these cpu blocks.  I will make a few to test and see if people get better results then the evo as one stated before which started this whole thread. 

well the thing with grinding is many super fine particles that come off from the grinding process.  Also grinding is generally used for harder metals.  Not that you cant grind softer metals, but I would use the same bit for a single pass as shown here:
 

 
 Then I would lap it on my AA granite surface plate. 

smart.

zophar

AnonymousGuy
Have a sacrificial 5820K on the way.  Think it's smarter to start with that first instead of going straight for the $1000 part :)

Are you testing the CPU before sending it to nateman so you have base line temperatures to go off of?

Yeah I pretty much have to.   
 
I also need to see if the Supremacy LGA2011 studs support bottoming out to the socket or not since the ILM (socket retention) thing has to go away with a naked die.  The Heatkiller IV uses the studs as the thumb screws (instead of EK's stud + thumbscrews) so it should be able to support a naked die physically.
 
Both waterblocks might not provide enough clamping pressure with a naked die.  It's a bit tricky to explain, but both waterblocks use "amount of thread travel" to stop you from over tightening.  In other words, you can only twist the thumb screws so many turns before they bottom out and that's it.  With no heat spreader, that means the springs won't compress as much -> less clamping pressure.  The Koolance 380i (that I used on my naked 3770k) relied on the springs compressing fully to prevent you from overtightening, which meant that it always provided the same clamping pressure regardless of the thickness of the chip.

From drawing to part:
 
 
From here out I'll probably break off into a separate thread dedicated to the purpose.
 




ProtoLabs did an awesome job machining it using nothing but my Solidworks part file.  It's so flat my straight edge can't tell it isn't flat.  It even suction cups itself to the surfaces I put it down on.
 
It's happening!
 
EDIT: And this is the sacrificial 5820K in the fixture.  I need to test it out in my mobo, make sure it can work without the socket retention mechanism, and take measurements and run thermal tests on it before shipping it off to get machined down.  I checked parametric data on it and it should be an average overclocking / average temperature chip.  I expect 4.5 Ghz out of it @ 1.3V.  All this while renovating my watercooling setup to the tune of replacing about half of it with new parts.  Going to be a busy weekend.