Antec Kuhler H2O 920 with i7-3770K - high temps

METROID4

Distinguished
Aug 17, 2012
19
0
18,520
Hello guys!

After searching on Google for [strike]hours[/strike] days, I've still not found anything quite like this problem as I have.
My problem actually now is finding out what's causing me high CPU temperatures.

So my setup:

First of all, ambient temperatures are around 19-23 degrees.

Corsair Carbide 500R with this setup:
- 2x (stock) 120mm front fans blowing in
- 1x (stock) 200mm side fan blowing in
- 1x 140mm fan on top (on the more to the front slot, not the back one) blowing out
- Water cooler radiator push-pull setup in back (both 120mm fans blowing out of the case)

i7-3770K:
OC settings:
Had an OC for a year (4,2 GHz with 100 base + 42 multiplier, fixed/manual Vcore voltage of 1,090 V as auto gave somewhat higher, I fine tuned to this and it ran 24 hours of Prime95 and already a year without crashing so I quess it's stable).
Idle temps are 40 degrees for all cores.

Antec Kühler H2O 920 as CPU cooler, with Arctic Silver 5 (applied with instructions of the Arctic site).

Other parts include:
ASRock Z77 Extreme4 motherboard
ZOTAC GTX 670 AMP! Edition sitting on lower PCIe slot, no temperature problems (70-75 degrees at full load, 1170 MHz core, refitted screws with custom rings in between and reapplied TIM which lowered full load from 87 degrees to 74)
Corsair TX750 PSU
4x4 GB Corsair Vengeance RAM @ 1600 MHz
Corsair ForceGS 240 GB SSD
2x WD Blue 1 TB HDD

The whole (custom built) computer is around 1,5 years old now.
I did all the CPU OC around a year ago, made sure everything worked (after having it for half a year), then left alone everything.
Story of how I got to notice this:
Around a week ago in Natural Selection 2, I noticed the GPU fans blowing really loud (NS2 [@max settings] uses more resources than BF4 [@ultra]). However, that was quite normal for me, as a year ago while in Furmark (400x300, 8x MSAA) testing it reached the same 85-87 degrees with loud fans, and I thought it was normal for such high OC.
However, I looked into my GPU thing, and I noticed when I pushed the cooling harder onto the chip itself, the temps went down 10 degrees instantly, so hence I refitted the screws with several of those small rings in between to tighten the screws, after this the GPU runs much cooler (even in Furmark it's around 75 degrees, 80 degrees after hours of testing).

So that's when I thought I'd have a look at everything, how everything runs, what temps I get and so on.
I ran HWmonitor (x64) on my second screen (1x 1080p and 1x 1024x768 "support screen" setup) while playing Natural Selection 2, and I noticed that my CPU cores were reaching (although after an hour) 90 degrees and after several hours, all the cores were sitting at 95 degrees.

To get to the point, after lot of tests, I concluded the following things:
CPU runs way too hot, and more important, I removed all OC settings (even reset BIOS settings to make sure, Vcore auto, Turbo on), running Prime95 still resulted in 80 to (after 20 minutes) up to 90 degrees in all cores.
This is of course not normal, as even the stock cooler could cool it down better with stock GHz.
I reapplied the part of the cooler on the CPU, fastened everything, reapplied Thermal Paste (AS5 again) (first cleaned with alcohol and one of those very fine cloth), nothing changed, still too hot.

Then I ran the ChillControl software, the utility and driver of the H2O 920 cooler.
Pump speed, since the first time I booted up 1,5 years ago, is still 3100 RPM (over the max), and sits there +-100. Fan speed behaves normally (600-700 RMP while idle, ramps up when liquid gets hotter). Made sure the fans were working, I could feel the blow coming out.
When I turn fan settings on "Extreme" (100% both fans on) while being extremely loud, cools the liquid down slightly (from 52 degrees back to 45 of the liquid temp) and then stops it from increasing for a while, then temps sit around 47 degrees (liquid temp). Core temps are not (or barely, unseen by the nature of CPU cores going +- up to 10 degrees in a second, up and down) affected.

What I noticed is that I think the warmth does not stay in the CPU, as when running Prime95 (all testing done in Stock GHz clock), I can feel the thing on the CPU getting warmer.
The liquid temperature, however, this is one of the things I could find nothing about, increases at a great speed. A minute after running P95, liquid temps rise 0,2 degrees every second, in 3 minutes the liquid temp goes up from 29 degrees to 52 degrees and keeps increasing. One of the tubes (I presume it's the one leading liquid from the CPU to the radiator) gets really hot, other tube feels cold. I can feel quite some warmth coming out of the backside over the radiator as well, so I have somewhat the feeling the radiator cools properly.
Now I don't know if something like this could happen if the pump stopped working.
I don't see any leakages.

Also, don't rule out any of the options what I have mentioned previously (like it still could be that the radiator is not cooling properly for example), as I have not done an extremely in depth research, and I don't know for each part if it's normal (like the liquid getting hot).

Please let me know what tests to run, what things to try if you guys need more info.
I hope all my spoilers kept my post organized.
And, also, I should write a book.

Thanks in advance!

Regards,
M4

Edit 1:
I still have lots of warranty for quite everything in my computer, so replacing is a viable option if necessary.
 
Solution
Having 2 cards in SLI will increase the heat in the case, but your temps are still high enough that something isnt working correctly.

Yes of course you can try the intel cooler, although they arent great - I would go back to stock clocks and maybe compare temps between the two that way.

My suspicion is that either the pump could be going faulty in your H2O Cooler as you have already tried re mounting it, or you have some sort of air lock (kind of like you get in a cars cooling system sometimes).

METROID4

Distinguished
Aug 17, 2012
19
0
18,520


That is actually a pretty good thing to try.
However, to see if the Antec cooler the problem, I could put back the stock cooler (I still have it) as well, the stock cooler also has some effectiveness, if it's then under 70 degrees in load (I don't remember temps of the stock, but in the time I had it [water cooler came 3 weeks later], temps were not above 80 degrees in full load with P95), then at least I know the problem is with the watercooler.

I also don't know what else could cause so high temps except the CPU cooler. Even removing all case fans, putting 2 GPU's in SLI (more heat) etc. would not increase the CPU temp to 90-95 degrees under load in just 5 minutes, would they?

So actually, I'd be glad to hear (as I could not find anyone on Google mentioning this) if some one has a liquid temperature rising so rapidly to 50 degrees, how hot the tubes are (like only one hot, or both tubes to and from the radiator are hot), whether the pump actually shows different speeds than 3050 RPM and so on.
Or, someone else with knowledge of watercoolers, knowing what to diagnose to effectively rule out options and get to the core of the problem

Thanks a lot for you input, RobCrezz! I greatly appreciate your help.

By the way, I'm not in a hurry as of the system is functioning fine, real world applications and even games such as BF4 don't push the temps so high as a P95 session. Now with the CPU forced on 3,5 GHz (I turned turbo off), I still have 60+ FPS and the CPU is "only" around 80-85 degrees after long hours of gaming, so as long as this won't have to go on for years there isn't any damage.
 

RobCrezz

Expert
Ambassador
Having 2 cards in SLI will increase the heat in the case, but your temps are still high enough that something isnt working correctly.

Yes of course you can try the intel cooler, although they arent great - I would go back to stock clocks and maybe compare temps between the two that way.

My suspicion is that either the pump could be going faulty in your H2O Cooler as you have already tried re mounting it, or you have some sort of air lock (kind of like you get in a cars cooling system sometimes).
 
Solution

METROID4

Distinguished
Aug 17, 2012
19
0
18,520


I've just remounted it completely, reapplied thermal paste, to no avail. Temps are the same hot, running stock and getting 85 degrees in 3 minutes in Prime95.
I'll search more, and will let you know if I've found anything.

Edit:
I also tried to remove any airlocks, as described by others having water coolers. Demounted the whole cooler and gently shook it, turning all directions. I don't hear any strange (hard ticking or swooofff) sounds, and I do hear the typical buzz which was demonstrated on YouTube, the sound that the pump is actually doing something.
Now I got a feeling the warm liquid might not make it to the radiator to cool down properly, as one tube (presumably to the radiator, not from) is really hot (after 5 minutes of P95 you can't hold it, too hot) and the other tube is actually so cold you can't hold it as well (the feeling of cold metal).
Also, now tested, increasing fans of the radiator to extreme (From 600 to 2500 RPM) did hardly any to nothing at all to the liquid temperature.

Edit: I don't have a lot of time at the moment for further testing on my own, but I'll do some further extensive diagnosing later on this week (~Friday).
I'll let you guys know if I have an update.
I'll also post my problem in the Official H2O 620/920 owner's club at an other forum.
 

METROID4

Distinguished
Aug 17, 2012
19
0
18,520
Tested something in the meanwhile: shutting down the two fans around the radiator (pulling their cables out so they are not running at all) did not affect liquid or CPU temperature, as if nothing had changed. So, the radiator is not doing it's job (properly) to cool the liquid down. However, it looks like there is at least some pumping and cooling (but at like 20-30% effectiveness), because in cases when the pump completely stopped working (and everything else was working fine, even the radiator cooled liquid) the CPU overheated (above the TJ) in minutes if not in seconds, as the little liquid sitting next to the CPU got hot, but was not replaced.

I've also contacted Antec about the issue, and send it back if necessary. I tried to solve this here without the hassle of removing (especially the plate on the back of the motherboard is very difficult to remove). In the meanwhile, the cooler is still mounted and I can do further testing. I can also use the computer, including gaming, as temps in games don't get much above 85 degrees, temps are sitting around 75-85 and once in several minutes one core for one short moment reaches something like 87 and then drops back.
I also don't want to (and somewhat can't) mount the stock cooler, as it can't be fitted because of the backplate for the watercooler, which puts threaded holes in the place were the clamps of a normal air cooler should be.
Removing the backplate would mean that I can't remount it again (for further tests and a possible solution). Also, the backplate is glued with a large surface area, and after getting it off (I read it can take 30-60 minutes of hair drying) I don't think it'll fit back properly (glue not fresh anymore).
 

METROID4

Distinguished
Aug 17, 2012
19
0
18,520
Update:
It turned out to be a half-working pump.
I sent it back to Antec, every part included (taking the adhesive back plate was quite difficult, I recommend using a hairdryer, but careful not to heat up everything, and also a fine knife to very carefully cut the "threads" of the glue, beginning at the corner is the best to get off the whole thing).
Having inspected it, after few days Antec contacted me that the pump is not completely broken (which would have resulted in even more extreme temperatures and quickly overheating above 100 degrees), that explains the strange temperatures and behavior.
As the 920 model is not available anymore, they are going to send the 950 model.

In the meanwhile, I remounted the stock cooler, which proved that there was nothing wrong with the CPU and cooling in the case was sufficient. I have overclocked though to 4.0 GHz @ manual voltage of 1.090V, which results in max core temp of 82 degrees peak for a 5-6 hour test on Prime95, without crashing.

I hope someone having these symptoms can solve their problem easier now, and that I contributed knowledge in forms of testing and experience.

Regards,
M4

Last update:
I've got the 950 model. Arrived neatly packed, just as the 920 back then.
Installation went very smooth, everything straightforward.
I've replaced the thermal paste as it was quite thick. I've used my Arctic Silver 5 again (little droplet/rice in middle).
I'm really content with it as it has just finished doing 12 hours of full stress test in Prime95, and being overclocked to 4.0 GHz (same setting as previously described), the cores were going around 55 to 60, with average max peak temps being 61 degrees and one core reaching 64 degrees once. (ambient temp measured was 23 degrees)
I guess those are nice temperatures.
I hope I can be helpful with this information for future problem scenarios.
 

FollowTrainTV

Commendable
Mar 4, 2016
10
0
1,510
I have always had issues with my 920, I think I will finally swap it to something else. I have idle temps of 45 degrees with no OC.

Often when I play with it the stupid thing seems to break completely and I hit 97 degrees. More messing about it seems to start working again.

Really not looking forward to removing the back plate. :-(