WARNING: Delidded CPU + Monoblock = poor mating!

VanGoghComplex

VanGoghComplex

2[H]4U
Joined
Apr 5, 2016
Messages
2,286
I've moved this post from the Watercooling sub for visibility's sake.

Guys, I've got a big problem. My 7700K is relidded, having used superglue on the IHS edges as suggested by the maker of Rockit88's delid tool. The result is a lower Z-height on the IHS, and as a result, smaller gap between die and IHS internally. Win win, right?

Except, it seems that EK's monoblock, which is NOT height adjustable, relies on the surface of the CPU IHS having at least a minimum Z-height. I've got it all replumbed, but looking between the card and block, I can SEE a gap. It's minor, but it's there. My temperatures are unacceptable.

The solution seems fairly obvious to me, though I hate it; delid and relid the CPU again, using RTV as typically suggested, and don't squeeze much on the relid bolt. I'll be sacrificing a larger gap between IHS and die to get rid of a gap between the EK monoblock coldplate and the IHS surface.

Is anyone else in this boat? How did you guys deal with this, if at all? I'm so bummed I can't check out the performance on my shiny new monoblock. ;_;
 
Okay guys, I'm posting here with an update because it seems to be a real thing. I've found a few Reddit posts asking about this, and guys saying that it's an issue, but those are usually drowned out by people saying "no way man that's bull EK is perfect" or something along those lines.

And to be fair, this is NOT EK's fault. It's just a part of their design; they rely on Intel's dimensional tolerances on finished CPUs so that their springless, zero-adjustment monoblock will meet the processor's IHS when torqued down.

THIS IS THE IMPORTANT PART, if you are considering a monoblock: If you delid your processor, and either 1.) float the IHS without reattaching it to the PCB or 2.) use an adhesive like Super Glue that allows the IHS to sit FLAT on the PCB, you may wind up with THIS:

113618_InkedIMG_20180329_201845046_LI.jpg


I'm sorry for the fuzzy photo. This is about as good as I could get it to focus on the relevant point, but I think it conveys the needed information.

Okay, so before anyone calls user error on this, I've had this corroborated by James with RockitCool. I reached out to him asking for advice (since he was so helpful with my delidding adventure to begin with.) He said the exact same thing happened to him when he tested his EK monoblock on a delidded processor, and he's making a custom solid copper IHS for me with a little extra thickness to it. I'll install it, glued (not RTV'd) down just like my stock IHS is, but it'll be thick enough to bring the overall cpu thickness back within Intel's design spec.

Cheers everyone!
 
If they sell these lids I might get one.
I dont have an EK block but my lid is not flat where it meets the die.
This caused 10C difference in temps before I delidded and after the delid causes rapid movement of the paste from between the lid and die and a jump of 10 to 14C on some cores.
I'm holding out for the new delidded block, but no predicted time to release.
https://hardforum.com/threads/ncore-v1-no-frame-water-block-prototype-sneak-peek.1955801/
 
Nenu said:
If they sell these lids I might get one.
I dont have an EK block but my lid is not flat where it meets the die.
This caused 10C difference in temps before I delidded and after the delid causes rapid movement of the paste from between the lid and die and a jump of 10 to 14C on some cores.
I'm holding out for the new delidded block, but no predicted time to release.
https://hardforum.com/threads/ncore-v1-no-frame-water-block-prototype-sneak-peek.1955801/
Click to expand...
They do sell them. It's been reviewed by a number of sites and the general consensus is that it's of high quality and does help temps.

I don't know if the typical IHS made by RockitCool has the extra height needed to offset the dimensional change on a floating IHS mount, since that's really only a problem if you have a monoblock. I had email correspondence with James at RockitCool about my problem, and the one he's making for me is custom in that regard.
 
VanGoghComplex said:
They do sell them. It's been reviewed by a number of sites and the general consensus is that it's of high quality and does help temps.

I don't know if the typical IHS made by RockitCool has the extra height needed to offset the dimensional change on a floating IHS mount, since that's really only a problem if you have a monoblock. I had email correspondence with James at RockitCool about my problem, and the one he's making for me is custom in that regard.
Click to expand...

OK thanks for that.
If I get no response from Nudecnc about the delidded blocks timeframe I'll get this lid.
 
VanGoghComplex said:
They do sell them. It's been reviewed by a number of sites and the general consensus is that it's of high quality and does help temps.
Click to expand...
I went through every page of their blocks and parts but cant find it.
fyi
 
You can always try using a thinner thermal pad for the VRMs. That is if your monoblock came with an 1mm pad so that you can change it with a 0.5mm thermal pad.

I have gotten the EK Monoblock for the Asus Rampage VI Extreme but for now I do not intend to delid my CPU. I will though down the line and if this turns out to be an issue I will try a thinner pad or go for a standalone CPU block + VRM Block. I guess though that the thinner pad + maybe a tiny thin layer of silicone could fix it save for the trouble of trial & error.
 
JargonGR said:
You can always try using a thinner thermal pad for the VRMs. That is if your monoblock came with an 1mm pad so that you can change it with a 0.5mm thermal pad.

I have gotten the EK Monoblock for the Asus Rampage VI Extreme but for now I do not intend to delid my CPU. I will though down the line and if this turns out to be an issue I will try a thinner pad or go for a standalone CPU block + VRM Block. I guess though that the thinner pad + maybe a tiny thin layer of silicone could fix it save for the trouble of trial & error.
Click to expand...
I'm considering this just because the 1mm pads were super squished, but I don't think that the pads were holding my monoblock off the CPU lid. All the mounting screws were tightened to a "solid stop", the posts were sucked all the way down onto the mobo.

Of curiosity, assuming the gap between the coldplate and VRM is less than 0.5mm to begin with, is there any reason to use 1mm vs 0.5mm? Does a thermal pad gain effectiveness when compressed more?
 
VanGoghComplex said:
I'm considering this just because the 1mm pads were super squished, but I don't think that the pads were holding my monoblock off the CPU lid. All the mounting screws were tightened to a "solid stop", the posts were sucked all the way down onto the mobo.

Of curiosity, assuming the gap between the coldplate and VRM is less than 0.5mm to begin with, is there any reason to use 1mm vs 0.5mm? Does a thermal pad gain effectiveness when compressed more?
Click to expand...

Hmm if your screws are fully tightened then its the lack of silicon that creates the gap. But then this suggests that the naked die is "squeezed" since the IHS should make contact with it in an effort to "reach" the plateau...around the cpu. I can understand too much silicon raising the height of the IHS but how comes that no silicon creates a gap when the cpu die should stop the ihs from getting lower..

The only possible explanation is that Intel's poop is thicker than Liquid metal applied to the CPU die and in conjunction with the removal of the silicon the IHS is indeed being lowered by a small distance. Then the "thicker" (default 1mm) VRM thermal pad creates an angle since it is now higher.


About the VRM thickness vs effectiveness - I honestly have no idea although it sounds not so possible to me. Test it and report back with your findings.
 
JargonGR said:
Hmm if your screws are fully tightened then its the lack of silicon that creates the gap. But then this suggests that the naked die is "squeezed" since the IHS should make contact with it in an effort to "reach" the plateau...around the cpu. I can understand too much silicon raising the height of the IHS but how comes that no silicon creates a gap when the cpu die should stop the ihs from getting lower..

The only possible explanation is that Intel's poop is thicker than Liquid metal applied to the CPU die and in conjunction with the removal of the silicon the IHS is indeed being lowered by a small distance. Then the "thicker" (default 1mm) VRM thermal pad creates an angle since it is now higher.


About the VRM thickness vs effectiveness - I honestly have no idea although it sounds not so possible to me. Test it and report back with your findings.
Click to expand...
You're correct in your second paragraph. The stock IHS is actually elevated off the surface of the PCB by a small amount by the silicone. In my install, my IHS is positioned lower because I'm using super glue, which thins out a great deal more than Intel's silicone.

The CPU die doesn't "elevate" the IHS. The pocket in the IHS is slightly deeper than the CPU die is tall. That's one of the places where you get better performance from a delidded processor: Intel's layer of goop occupies a thicker gap between IHS and die than you wind up with after relidding.
 
VanGoghComplex said:
You're correct in your second paragraph. The stock IHS is actually elevated off the surface of the PCB by a small amount by the silicone. In my install, my IHS is positioned lower because I'm using super glue, which thins out a great deal more than Intel's silicone.

The CPU die doesn't "elevate" the IHS. The pocket in the IHS is slightly deeper than the CPU die is tall. That's one of the places where you get better performance from a delidded processor: Intel's layer of goop occupies a thicker gap between IHS and die than you wind up with after relidding.
Click to expand...

Then since you can't actually increase the liquid metal layer thickness you have to decrease the vrm surface area height (i.e. the thermal pad). If a 0.5mm thermal pad (EK has on I got it) doesn't fit and still creates an angles then you need an even thinner layer of thermal interface material that could be made of thermal paste maybe. I haven't actually tried it but maybe its a solution.

I am now wondering whether I should have gone for a separate CPU / VRM block (both exist for my CPU) in the first place + a Rockit 99 delid tool. In case I come across the same issue on the Rampage VI Extreme with an i9-7940X I will be left with a nice brick of a waterblock that is not working properly anymore. This or no delid. I will have to check with other owners of this monoblock.
 
JargonGR said:
Then since you can't actually increase the liquid metal layer thickness you have to decrease the vrm surface area height (i.e. the thermal pad). If a 0.5mm thermal pad (EK has on I got it) doesn't fit and still creates an angles then you need an even thinner layer of thermal interface material that could be made of thermal paste maybe. I haven't actually tried it but maybe its a solution.

I am now wondering whether I should have gone for a separate CPU / VRM block (both exist for my CPU) in the first place + a Rockit 99 delid tool. In case I come across the same issue on the Rampage VI Extreme with an i9-7940X I will be left with a nice brick of a waterblock that is not working properly anymore. This or no delid. I will have to check with other owners of this monoblock.
Click to expand...
I'm still positive that the thermal pads are not the culprit here. My monoblock isn't being held up off the mobo anywhere; all of the mounting posts are secured tight.

The problem is that my CPU, from the bottom of the PCB to the top of the IHS, is thinner than a normal Intel CPU due to my delid. That monoblock is a zero-adjustment install; it relies on a CPU of a certain overall thickness in order to mate with the IHS.

And therein lies my adjustment point: I'm getting a thicker IHS. The additional thickness will bring the CPU's overall height back within Intel's dimensional specs, meaning the EK block will work with it.
 
VanGoghComplex said:
I'm still positive that the thermal pads are not the culprit here. My monoblock isn't being held up off the mobo anywhere; all of the mounting posts are secured tight.

The problem is that my CPU, from the bottom of the PCB to the top of the IHS, is thinner than a normal Intel CPU due to my delid. That monoblock is a zero-adjustment install; it relies on a CPU of a certain overall thickness in order to mate with the IHS.

And therein lies my adjustment point: I'm getting a thicker IHS. The additional thickness will bring the CPU's overall height back within Intel's dimensional specs, meaning the EK block will work with it.
Click to expand...

There might be differences between yours and the Asus Rampage VI Extreme then. Here is a quote from another user from overclock.net

"I did not re-glue. This led to uneven pressure owing to the slightly shorter package height.

I compensated by using 0.5mm fujipoly pads with some thermal compound on them. Ideally 0.75 would have meant more pressure on the pads, but it worked and produced good thermal results.

18 cores are ~9C apart most of the time from top-to-bottom with even/full/heavy load."


In essence the solution was to reduce the height of the VRM pads but in your case it looks different.
 
Okay everyone, I've received my custom copper IHS from RockitCool today, and installed it into my 7700K, which I then covered with my new EK monoblock. I did a couple test fits, checking the thermal compound spread on the monoblock, and James at RockitCool did it right; there's just enough extra thickness on my CPU package to bring it back within the right dimensions.

My CPU temperatures have dropped around 5c under load, as compared with my old Watercool Heatkiller IV block and stock IHS (was delidded at that point too.) Additionally, my cores have all fallen to within 2c of each other, whereas they used to be 6-7c apart. My VRMs have dropped from about 85c under full heat-soaked load to 45c.

All in all I'd say this was a success, even if it took me a little longer than I'd planned. Cheers!
 
You must log in or register to reply here.
Back
Top