Jump to content
Corsair Community

H80i Cooling issues


bhorv67

Recommended Posts

Under load, my i7 4790k is averaging temps in the mid 90s. Corsair link says the H80i temp is around 37. Can't figure why the cooler isn't working better. I'm fairly certain the pump is seated properly and there's good contact. I've tried removing the pump, cleaning the thermal paste on both the cpu and the pump, reapplying a minimal amount, but still can't seem to get the cpu to reasonable temps.

The cooler hoses are warm to the touch so it appears the pump is working.

 

Any suggestions?

Link to comment
Share on other sites

We probably need some more information to interpret what you've stated.

 

1) What load and what voltage? There is a big difference between Prime95 and other stress programs, and further still when discussing games or other normal programs.

 

2) Usually, we can gain some insight by comparing the room temperature to the H80i Temp (water temperature) both at load and idle. 37C looks like the load temp. What about the other two? Be aware is may take the water temperature several minutes to cool down after a stress test. If possible, attach a screen shot of the home page in Link. The other readings can help determine your approximate case temperature, which is usually different than the room temp. It also shows pump speeds and other helpful things.

Link to comment
Share on other sites

We probably need some more information to interpret what you've stated.

 

1) What load and what voltage? There is a big difference between Prime95 and other stress programs, and further still when discussing games or other normal programs.

 

2) Usually, we can gain some insight by comparing the room temperature to the H80i Temp (water temperature) both at load and idle. 37C looks like the load temp. What about the other two? Be aware is may take the water temperature several minutes to cool down after a stress test. If possible, attach a screen shot of the home page in Link. The other readings can help determine your approximate case temperature, which is usually different than the room temp. It also shows pump speeds and other helpful things.

 

Under load of 100% and voltage set @ 1.32 & OC to 4.6, I'm getting temps at 100c. I am noticing at that temp I'm only getting 4100mhz, most likely because the cores are too hot.

I don't use Prime95 or stress tests in general. These programs tend to unrealistically stress the system. Since my work deals a lot with rendering 3d models using C4D, that's when I typically get temps that high.

for gaming, I mostly use P3D (flight sim) which is quite cpu intensive.

 

Room temp is around 74 with the air on in the summer. Winter is a few degrees cooler, but still getting these high temps on the cpu no matter what the room temp seems to be.

Today I did a deep clean of my system - took all the fans out, vacuumed everything. Checked the h80i pump is seated tight on the heat spreader, made sure the back plate is not loose.

Temps are still the same. A few questions I have:

 

Should the exhaust air from the H80i be hot? - normally this is slightly warm which seems odd for how hot the cpu is getting.

Is it possible I may have a defective cooler? I've had it for quite a while and just figured these temps are normal.

The hoses from the pump are warm, not hot. According to the temps, the water should be at boiling.

Something for sure is not right here.

 

Thanks for any help you can offer.

Link to comment
Share on other sites

First glance does suggest you have a contact problem between the cold plate and CPU. Your water temperature (H80i Temp) is within the expected range and your difference between case and water temperature is good for a rear mount single radiator cooler. When there is a problem with the pump or the water flow, the H80i temperature will get very warm and stay warm even at idle. You are well below that level. When there is a contact problem, your CPU idle temps are considerably higher than the idle water temperature, often very reactive to any kind of load, or even erratic or "jumpy" in their behavior. Full 100% loads often bring near shutdown temperatures instantly. When there is a problem with the cooler, those CPU temps start of at moderate levels but them climb slowly to the upper reaches and don't cool down.

 

Unfortunately, I am not familiar with the Cinema 4D program and can't interpret whether the results are high or low. Flight Simulator will be a good test for the cooler, but right now I want to eliminate GPU loads that also have an effect on the water temperature.

 

Can you run a mild stress test like AIDA64 or Intel XTU? Both of these are cupcakes compared to Prime95 and I would expect loads in the upper 50's for XTU. Run the stress test for 5 minutes (not the benchmark). It should jump right into the upper 50's and then follow a sine wave loading pattern in that range. If the CPU core temperatures immediately jump to 80C+, stop and report back. If OK, let in run and keep the Link window open during the run. The H80i water temp should start at your case temperature (~31C) and will likely rise 6-8C during the test. After you stop the test, watch to see how long it takes the water temperature to return to the starting value. In a 5 min test, there is no heat soak and the water temp should come back to normal in just a few minutes. If it stays at a high level, let us know.

Link to comment
Share on other sites

First glance does suggest you have a contact problem between the cold plate and CPU. Your water temperature (H80i Temp) is within the expected range and your difference between case and water temperature is good for a rear mount single radiator cooler. When there is a problem with the pump or the water flow, the H80i temperature will get very warm and stay warm even at idle. You are well below that level. When there is a contact problem, your CPU idle temps are considerably higher than the idle water temperature, often very reactive to any kind of load, or even erratic or "jumpy" in their behavior. Full 100% loads often bring near shutdown temperatures instantly. When there is a problem with the cooler, those CPU temps start of at moderate levels but them climb slowly to the upper reaches and don't cool down.

 

Unfortunately, I am not familiar with the Cinema 4D program and can't interpret whether the results are high or low. Flight Simulator will be a good test for the cooler, but right now I want to eliminate GPU loads that also have an effect on the water temperature.

 

Can you run a mild stress test like AIDA64 or Intel XTU? Both of these are cupcakes compared to Prime95 and I would expect loads in the upper 50's for XTU. Run the stress test for 5 minutes (not the benchmark). It should jump right into the upper 50's and then follow a sine wave loading pattern in that range. If the CPU core temperatures immediately jump to 80C+, stop and report back. If OK, let in run and keep the Link window open during the run. The H80i water temp should start at your case temperature (~31C) and will likely rise 6-8C during the test. After you stop the test, watch to see how long it takes the water temperature to return to the starting value. In a 5 min test, there is no heat soak and the water temp should come back to normal in just a few minutes. If it stays at a high level, let us know.

 

Thanks for the help - Attached is a screen shot while Intel XTU was running. As you can see, as soon as I hit start, the cpu shot right up to 100c. The H80i's temps didn't change. I stopped the test after a minute as it was reporting throttling and was dangerously hot.

I'm completely stumped. I opened the case and checked the hoses, they were room temperature. The fan speed didn't even come up from idle.

1142190530_IntelTestTemps.thumb.jpg.d2b3b8ed1031adda86907fb35952d853.jpg

Link to comment
Share on other sites

That looks like a contact problem. Your package temperature was already quite high even before the load was initiated. Two basic types. The Thermal Interface Material is a mess and heat is not transferred well. Usually very jumpy, even random looking temperatures at both idle and load. I don't think this is your issue.

 

The other and most common is there is a physical gap between the cold plate and CPU. If you take the block off, you may see this as unutilized or an uneven TIM pattern. Even the smallest of gaps beyond what you can physically see can cause this issue. The most common culprit is the backplate. It should be loose while installing, but then tighten as you clamp down. This is not about clamping force. Do not over-tighten the screws. Unfortunately, this can be a very tedious problem. Look at the installation guides at the top of the cooling forum and see if anything stands out.

Link to comment
Share on other sites

That looks like a contact problem. Your package temperature was already quite high even before the load was initiated. Two basic types. The Thermal Interface Material is a mess and heat is not transferred well. Usually very jumpy, even random looking temperatures at both idle and load. I don't think this is your issue.

 

The other and most common is there is a physical gap between the cold plate and CPU. If you take the block off, you may see this as unutilized or an uneven TIM pattern. Even the smallest of gaps beyond what you can physically see can cause this issue. The most common culprit is the backplate. It should be loose while installing, but then tighten as you clamp down. This is not about clamping force. Do not over-tighten the screws. Unfortunately, this can be a very tedious problem. Look at the installation guides at the top of the cooling forum and see if anything stands out.

 

Thanks - All makes sense. I agree, sounds like a contact issue.

I'll have a look. The thing is, the backplate has no way to tighten or loosen from the back. the stand-offs are screwed in under the pump block. The only way to loosen the back plate would be to loosen the stand-offs. Usually I have the pump block screws finger tight, with maybe a touch more from a screwdriver.

Link to comment
Share on other sites

That's perfect. No need for more.

 

ok, fixed the backplate. Idle temps much better, however stress test temps are still too hot.

Doesn't seem the H80i is absorbing the heat as I thought it would.

I followed the install guide on the forum site and made sure the block is centered and there doesn't appear to be any contact issue.

 

Attached is the latest test, post correcting the backplate.

1068377160_IntelTestTemps2.thumb.jpg.c1351fc45d71ce523cfcc7fb5ea1e528.jpg

Link to comment
Share on other sites

If you can't see anything else you can do, it may be time to try another set of washers to make the backplate more snug. Make sure they are non-conductive. You are not supposed to need them and this is the last resort, but some people have only found success this way and a lot them have Asus boards from Z87-97 series.
Link to comment
Share on other sites

If you can't see anything else you can do, it may be time to try another set of washers to make the backplate more snug. Make sure they are non-conductive. You are not supposed to need them and this is the last resort, but some people have only found success this way and a lot them have Asus boards from Z87-97 series.

 

Thank you for all the guidance. I've finally had success in getting the temps in a more reasonable range. The fix is as you said - need backplate washers. This should be something Corsair includes in the kit. Even though there was contact to the cpu, it wasn't enough. The backplate was still a bit loose. Once I put a few plastic washers in between the backplate and the back of the mobo, the backplate is now snug against the mobo. There's now full contact to the thermal heat spreader and the pump block.

Load temps are within the ~65-70 deg range, as well as the H80i temp is now reporting 35.2. Much better!

Link to comment
Share on other sites

There are a couple of buried threads with some discussion whether the backplate is defective or poorly designed versus the Asus boards being too thin. I don't know the answer. Most boards don't need them. A lot of the ones that do are Asus. Beyond that, I am not sure there has ever been a sorting of specific models with this trouble. At the time, Z87/97 were the most popular by far and the most of the people having issues had one in those families. It is times like this I am glad I overpaid for my 2011-3 socket.

 

I'm glad it's sorted. Sorry it took so long to come to a conclusion.

Link to comment
Share on other sites

Archived

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

×
×
  • Create New...