Jump to content
Corsair Community

Install Issue - H115i Pro - Not enough clamping force?


Deathman20
 Share

Recommended Posts

So last night got ready to get my fans and new cooler setup in the new case and everything was going smoothly. Then... power up on the new water cooler and issue showed up. High temp on 1 core. After a few attempts of re-seating, adjusting the back plate and failing to bring my i7 6700k temps down I'm figuring there is an issue. Core 0,2,3 all nice and low around 18-20C idle. Core 1 was sitting about 10-15C higher, and then would spike into the 80, 90 and even 100... So I knew there was a contact issue going on.

 

After 4 attempts to take everything apart, reseat the block still getting similar results, actually last attempt Core 3 was sitting about 5-10C higher than it was previously. I see that the back plate is loose on a lot of people's boards, I got that too but the thing that is odd is I can fully tighten the screws for the block all the way down by hand! So I know there is to much slack between the back plate and motherboard to get proper, full contact.

 

I know it works fine with other peoples boards no issue so I'm thinking that my motherboard might actually be thinner than others. That being said curious what people come up with for a solution to help get more clamping pressure? Was thinking of using washers as I've used that with other type of installs before. Either using the ones they got from the kit or purchased some. I got some used plastic ones I've used on my previous heatsink that I might try or run to the store quick and get some plastic ones to use since it looks like I got a few traces around a few of the holes on the backside and probably shouldn't use metal ones.

Edited by Deathman20
Link to comment
Share on other sites

Having one core warmer than the others isn't that unusual in the Skylake+ processors. What is unusual, however, is the 90-100C temperatures that you are seeing.

What is your max vCore during this?

 

(Note: It's not likely that you needed those washers.)

Link to comment
Share on other sites

Having one core warmer than the others isn't that unusual in the Skylake+ processors. What is unusual, however, is the 90-100C temperatures that you are seeing.

What is your max vCore during this?

 

(Note: It's not likely that you needed those washers.)

 

Actually the core temp is unusal. I just uninstalled a heatsink and those temps where a solid 78-80C on it all cores within 1C of each other running Intel Burn Test. Its been that way for roughly 2 years since I got this setup. Actually over the years never had a core with "odd" temps compared to others, maybe I've been lucky but my past ones where all solid within 1-2C of each other at the same load.

 

Voltage is fixed in the bios @ 1.275V for 4.5Ghz.

 

Down clocked it a bit to just to get a base line at least... Room Temp 18C

 

3.2Ghz @ 1.1V

 

Core 0 = 59C

Core 1 = 80C

Core 2 = 43C

Core 3 = 53C

 

Pretty consistent holding temps firmly now for 5min, water temp climbed from 20.1 C to 22.4C on balanced pump so far. So there still has to be an issue with pump contacting the heatsink... hmm

 

Edit: Here is a graph on temps, note CPU1/CPU2 corresponds to Core 0 from what I was mentioning before.

3JR3UvU.png

Edited by Deathman20
Link to comment
Share on other sites

I hate to say this because of the implications, but it is not possible for a single core to be affected in this way by the contact between cold plate and CPU lid. The physical space between the cores is extremely tiny. This also appears to be an interior (inline) core and that really makes things like clamping force or that sort of physical contact more than improbable as the culprit.

 

Your core temp range is really bizarre in the graphs above. I assume each core is the physical core and its corresponding virtual thread. Usually wildly varying core temps trace back to the TIM material -- too much, contaminated, something else.

 

Since your initial attempts, have you taken the pump off and looked and the TIM spread? Is it smooth and even? Or clumps left behind? If it is anything but perfect, the prerequisite move is going to be to clean and apply new TIM, classic line or pea method. Too much is not better.

 

This is not a new processor and you did not have these issue before, correct?

Link to comment
Share on other sites

Correct this is not a new processor it is about 2 years old. Yeah the graph does show the virtual threads, next to the actual core. But yes the graph seems very odd with the temp ranges.

 

TIM spread on the cpu has been good. Just this last round might of used a little too much as it was really on there but didn't affect temps positively or negatively. As well I just tried to rotate the block to see if that made any difference only got 90 deg but nothing different on the quick test.

 

I just removed the whole cooler from the case and re looking at it tonight again with a clean mind about it. Hopefully new results in the next couple hours. First dinner.

Link to comment
Share on other sites

And here is a weird one.... While still not perfect i just drastically dropped the temp! Kicker here is that I didn't use the back plate and standoffs it came with it. Used my old Thermalright Ultra 120 plates. So this was after 2-3 removals/resets on the core (no reapply of paste, running very low till tomorrow),

 

Temps idling right now 19-24C (24C was the really hot core before). Load on balanced performance right now, NO fans on the rad, 2 cores at 53-54C, and 2 cores at 43-45C. Wow major brake on this, sucky that its using my heatsink bracket but least it shows that the mounting plate is bad. Going to try to remount the rad, and get it all re connected back up and clean the block and proc and reapply what little paste I have left.

 

oMCpvqV.png

Link to comment
Share on other sites

And now back up to 4.5Ghz where I was previously. Funny thing is that 2 of the cores are sitting really close to what I was getting with air, and the other 2 are about 10C cooler. Still getting a discrepancy with 2 cores, but they are side by side under load, way better than it was. Now idle, everything is within 2-4C at most.

 

XUc8mIi.jpg

Link to comment
Share on other sites

And final assessment with it. The mounting bracket/system is just not cool.

 

Which I could just upload the images here but links below. Basically guessing what it came down to is I could tell on the pump that the outer edges of the IHS was hitting the plate. I think my CPU IHS is concave so it wasn't making great contact in the center and pushing/tighting the block just pushed it further away. With my old clamp, on the back side of the motherboard (seen in the links below), it actually pushes up against the CPU socket brackets. As well while the stand offs are actually about 1/16" taller on the MB side it sits flush to the motherboard and probably helps make good contact. The screws on the top, now this is interesting. They are bottomed out, tight as can be which really isn't much since its finger tight. After my last remount of the remount I had some washers on the back of the board swapped the 1/16" to 1/32" washer what I had originally on the heatsink I was using. Re-tightend it all down and now temps are solid. Very close to each other (just a few C) and I can go to my previous OC, probably better since its almost 15-20C cooler doing the intel burn test.

 

Post #42 and Post #43

Link to comment
Share on other sites

Interesting. I suppose it is possible that generally poor contact overall was exacerbating a naturally existing temp discrepancy between the cores. Most CPUs have one (or two) hot cores and I had a 5820K where #3 ran +8-10C. That 65-55-45 spread was very strange, but perhaps it was the result of partial contact on the lower half and the expanding temps of the less efficient cores 0/1. You still have an odd spread, but not outside the realm of possibility and hopefully this makes your normal use possible.
Link to comment
Share on other sites

Interesting. I suppose it is possible that generally poor contact overall was exacerbating a naturally existing temp discrepancy between the cores. Most CPUs have one (or two) hot cores and I had a 5820K where #3 ran +8-10C. That 65-55-45 spread was very strange, but perhaps it was the result of partial contact on the lower half and the expanding temps of the less efficient cores 0/1. You still have an odd spread, but not outside the realm of possibility and hopefully this makes your normal use possible.

 

Actually had a new post on the other forums after I remounted the system. Showed some temps and they even dropped more/came inline with each other. So while the stock bracket I was able to start getting it close after so many attempts, now I actually did its spot on. Underload its within just a 2C of each other now. Really think and maybe its related to someone else post I saw. The block pushes down on the CPU socket enough where it bends the board in the area, but by having a plate that is now sitting behind the socket and helping it push against the block as well fixes it. I can see much better contact using something like this, course probably cost an extra $5-10 in the kit because its not some throw away piece of plastic.

 

All in its fixed, sadly not utilizing the bracket that came with the pump but confident it will work out in the end.

Link to comment
Share on other sites

 Share

×
×
  • Create New...