Jump to content
Corsair Community

H80i Cooler Issue


Some Guy

Recommended Posts

Ok, so I'm building a system for a friend with the following..

 

i7 4790k

Asus Maximus VII Gene

H80i cooler

Silverstone TJ08B-E

(other stuff which really doesn't matter)

 

..and like a lot of other people, I'm having a lot of problems with the H80i cooler. It installs exactly like it ought to, following directions for the screw orientation on the backplate, and the standard X pattern of tightening the bolts, but the thing gets wicked hot way too fast under load. At stock speeds it starts off around where you'd expect (mid 30s), and under a full load it will immediately shoot up to the high 80s. It will work it's way up to around 90, but the strange part is that the cooler itself is only reporting a water temperature of around 25-35 (beginning to running for ~60min), and the radiator itself doesn't feel very warm at all. It's got the fans in a push/pull, as exhaust out of the case (yes I know it's not optimal, but we're talking about 3 degrees difference not the 30 I'm seeing), with the block's text oriented right-side-up along with the rest of the text on the board. It's been on it's side of the most part, just because I haven't bothered standing things up without it being 100%.

 

Naturally, I dug around the internets a bit** and found a couple fixes, namely re-seating the block and tossing a few nylon washers behind the backplate. I re-seated the block a couple times, and the paste spread has looked good (using AS5), though the block does feel like it could be a smidge tighter, and it's tightened down as much as it can be. However, I'm a little hesitant to start tossing washers behind the backplate, since I know what kind of pressure it SHOULD be putting on the socket, let alone what it would with extra pressure the washers would add.

 

So my question at this point, is there anything I can do to make sure the pump is ACTUALLY working, before buying some washers? Even with a silent house I'm not hearing much out of the thing (though admittedly years of overly loud concerts have taken their toll), despite the software reading ~2200rpm. Also, is there a way to ensure that I've actually got fluid in the loop? I could swing by my local Microcenter and grab another unit to test, but I really don't want to waste their time if it's a technical issue that could be easily fixed, especially since the one I have would need to go back to Tiger.

 

Thanks in advance for any ideas you guys have!

 

**Typically when I have issues, googling around I stumble onto the Corsair forums with a fix, though this is the first time I've actually registered. A big thanks to RAM Guy and Yellowbeard for all of their efforts over the years; whenever I see the avatars, either on or off these forums, I know it's something I should probably read and commit to memory.

Link to comment
Share on other sites

Is the Corsair logo "lighting" up when it's powered on?

 

Putting the H80i to MAXIMUM speeds for both fans and pump SHOULD sound like a jet engine to you, despite too many Ozzy concerts..

 

Ensure you have fans in FULL speed mode in BIOS. Then set Corsair software to your liking. My 4790k runs a bit warm, but nowhere near where you're at. Are you overclocked? are you running more CPU voltage or cache voltage?

 

Your BLOCK should stay cool, mine jumps up to the low 30's when my CPU is stressing to the 50's in AIDA... Usually in idle mode, it's within a degree of the CPU.

Orientate the BLOCK per instructions... not sure about why you'd install it sideways...??

Link to comment
Share on other sites

Yep, the light on the block is lighting up.

 

Everything on the 4790k is stock (until I get this whole thing sorted out), but I've got the fans going straight into the motherboard fan headers to use the AI Suite fan control, and they go full tilt as soon as the load hits the CPU.

 

Ok, I guess it's just really weird to me that the block would stay so cool, but the CPU itself would go ape so fast. It sounds like I'm not making enough contact, but the spread on the paste is good, and I don't have anything blocking the block that I can see. It's not like a lot of other photos I've seen where someone only gets paste on half of the lid, it's exactly what you would expect good contact to be.

 

Thanks for the reply!

Link to comment
Share on other sites

Sounds good then..

ENSURE you've updated to the latest BIOS, as earlier bios have been known to OVERVOLT our 4790ks... ALSO, check to see that your CORE SYNC is in AUTO. NOT sync all cores, because syncing all cores in essence IS overclocking. The 4790k default is 44,44,43,42. If you're syncing all cores to 44, (or 4.4Gigs) AUTO or adaptive volting WILL jack your voltage upwards to 2.5-2.71. You're temps will look like you're describing now. At AUTO cores or default, you'll peak around 1.156v to 1.171. These voltages will lower your thermal beast quite considerably...

Link to comment
Share on other sites

Generally anytime the temps spike up that fast it's going to be a contact problem. Be careful not to over-tighten the mounting posts. Hand tightening should be enough for anyone with strong hands or simply turn until it stops and give 1/4 turn more with a screwdriver. It's possible it's a bubble, but if your confident in the paste job, then focus on the way the block is sitting. If the pump or fluid level was a problem, it would take longer for the temperatures to build up to that level.

 

Go to the bottom of this thread for the picture. Make sure the backplate orientation is correct.

 

http://forum.corsair.com/v3/showthread.php?t=135990

Link to comment
Share on other sites

Sounds good then..

ENSURE you've updated to the latest BIOS, as earlier bios have been known to OVERVOLT our 4790ks... ALSO, check to see that your CORE SYNC is in AUTO. NOT sync all cores, because syncing all cores in essence IS overclocking. The 4790k default is 44,44,43,42. If you're syncing all cores to 44, (or 4.4Gigs) AUTO or adaptive volting WILL jack your voltage upwards to 2.5-2.71. You're temps will look like you're describing now. At AUTO cores or default, you'll peak around 1.156v to 1.171. These voltages will lower your thermal beast quite considerably...

 

Yep, latest BIOS on the board, and firmware on the cooler too. Good call on the core sync; I dialed it back to AUTO, but the problem persists.. spikes to 85-90ish, then when I cut the test it drops back down to 25-30ish.

Link to comment
Share on other sites

Generally anytime the temps spike up that fast it's going to be a contact problem. Be careful not to over-tighten the mounting posts. Hand tightening should be enough for anyone with strong hands or simply turn until it stops and give 1/4 turn more with a screwdriver. It's possible it's a bubble, but if your confident in the paste job, then focus on the way the block is sitting. If the pump or fluid level was a problem, it would take longer for the temperatures to build up to that level.

 

Go to the bottom of this thread for the picture. Make sure the backplate orientation is correct.

 

http://forum.corsair.com/v3/showthread.php?t=135990

 

Yeah, the backplate is correct (two notches on the backplate going around the screws).. although admittedly it was dumb luck that they ended up that way in the first place.

 

Yeah, now that you mention it the pump being busted or fluid being low would still raise the temperature, but it would do it over time.. it does sound like a contact issue, but I really don't know what I'm doing wrong here. I've been installing high-end cooling for a decade and this is the first cooler that I've had any issues with.

Link to comment
Share on other sites

Is it possible you have a board where a capacitor is interfering with the block which will not allow a proper sealing...

 

I thought the same thing, but there's nothing on the board that would interfere:

 

http://rog.asus.com/wp-content/uploads/2014/04/M7G_2D-with-SupremeFX-Impact-II-copy.jpg

 

The closest stuff around the board that might are the caps on the right of the socket, which are directly underneath the tubing, but there's still quite a bit of clearance for them.

Link to comment
Share on other sites

Ok, so I got bored and tried the last two options I had... with no good results.

 

I tossed a nylon washer behind the mount, and while the block now certainly seems much more firm once everything is tightened, the impact was minimal: 1-2c at best (though I didn't leave it on long, since it was still wicked hot). Then I tried a second cooler, and the results were similar. For kicks, I tried to rotate the block 90 degrees, and it's still more of the same. All of the above were also using NH-T1 instead of the AS5, though that shouldn't make a huge difference here.

 

I also confirmed I'm at stock speeds, thanks to Zorlac's suggestion, but even after that and everything else I'm still raising to well over 80c almost instantaneously. The backplate is correct, and I'm positive I'm installing the block properly, but I have no idea WTF I'm doing wrong here. I find it really hard to believe that I've got two bad H80i's, though I guess it's not impossible, but other than that I'm out of ideas.

Link to comment
Share on other sites

Please report a screenie of CPUZ running intel extreme tuning utility. I need to see your voltage and CPU speed. ENSURE YOU'RE not USING xmp profile. Run your ram at default 1333 speed so we can properly address this temp issue. XMP tends to overvolt the cache as well...

 

http://www.intel.com/content/www/us/en/motherboards/desktop-motherboards/desktop-boards-software-extreme-tuning-utility.html

Link to comment
Share on other sites

Attached is a photo of CPUz while running Prime95 small FFTS. Here's what I'm seeing..

 

Intel XTU stress test: ~60-65c

Intel CPU diagnostics: ~40-45c

Prime95 large FFTS: ~60-65c

Prime95 small FFTS: ~88-93c

 

Obviously the last is the most troubling. This is with an XPS profile that's modifying the RAM to 1600 and 1.35v, but I'll dial that back and re-run everything. Obviously this is all at stock, but I had hoped to get a little more out of this thing before I hand it over (after all, why else pay for a K!), though I'm not going to take it very far.

Prime.png.6f977f930700f69e65c46c71d17a9ae6.png

Link to comment
Share on other sites

OK.... here's the deal: you need to download and run P95 version 26.6 ONLY. This version doesn't run insane AVX instruction. You're temps look good man, only 1 thing puzzles me... WHY does your CPUZ show 4.4G. If you're default, it should only go up to 4.2G which is max on ALL 4 cores. 1.35 volts is INSANELY too much voltage for stock speeds.

 

60-60C ETU looks great! Thanks for the info SG...

Link to comment
Share on other sites

No XMP profile enabled, RAM at 1333...

 

Intel XTU stress test: ~50-55c

Intel CPU diagnostics: ~38-43c

Prime95 large FFTS: ~48-53c

Prime95 small FFTS: ~63-68c

 

Obviously a HUGE improvement! Attached are the CPUz shot along with XTU.

 

So what exactly is going on here where it's getting that much more hot (25c in prime95) where the only difference that I can see is that it's running an XMP profile that modifys memory speed and voltage? Especially since the voltage is up .15v?

Intel1333.png.fc6a928c5f33d18aa779e7b92b0c6f62.png

XTU1333.thumb.png.0ca0ef4174e45738c6837b3e66ce6fc5.png

Link to comment
Share on other sites

CACHE voltage is modified to run an OVERCLOCKED ram speed. Default ram bus is 1066/1333. If you're MOBO sees an XMP profile that is ABOVE the aforementioned speeds, it AUTOMATICALLY increases the cache voltage in order to keep up with a faster ram speed. Increasing ram speed ALSO increases cache bus resulting in an "overclock." Which.....in turn, results in higher voltage, which...in turn, results in thermal increase.
Link to comment
Share on other sites

Your temps look really good SG... thanks for hanging in there..

Are you running P95 26.6v?

 

*just noticed, you're cpu voltage is set to adaptive, try AUTO. Adaptive WILL increase turbo voltage a bit more than AUTO will. (looking at ETU settings right column)

Link to comment
Share on other sites

Here's a few more things you can do to help "tame" the 4790k's voltage/heat issues.

1. set load line calibration to level 1. (no need for ANYTHING higher unless you OC.)

2. set LLC to DISABLED...

3. Voltages set to AUTO NOT adaptive. BOTH cache and cpu.

4. Run the H80i pump at FULL speed at all times. Make profiles for fan speeds, but set pump to MAX.

5. Try push pull method from OUTSIDE to INSIDE of case for H80i.

6 Prior to block installation "STAIN" the block with thermal paste..

Link to comment
Share on other sites

Well now, guess who's feeling like an idiot? You guessed it, it's THIS GUY!

 

I tweaked all of the aforementioned, got all my fan settings dialed in, have the multiplier for 1/2/3/4 cranked up to 48/46/45/44, and with a full load on the XTU stress test the CPU sits between 60-65c. Other tests with a lower core count are all sitting below 60c, which I'm pretty happy with too, especially since the guy it's for isn't going to be doing a whole lot that will be long-term CPU intensive.

 

All that's really left to do on it is to try and tidy up the cabling (though I doubt it will get very good) and clean up the drives a little bit. I'm going to leave the radiator as an exhaust though; when I was testing before it was actually out of the case entirely with the fans attached, and temps aren't any different now. I don't really think it's worth it in this scenario, since there's a decent filter on the front, and I don't feel too comfortable telling the guy to pop the side and dust it out every couple of months.

 

Thanks to everyone for the help, it's greatly appreciated!

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...