Jump to content
Corsair Community

Corsair H115i making my CPU overheating?


SgtDeathAdder
 Share

Recommended Posts

Yesterday I switched from i7 2600 non-k to i7 3770k (@ stock 3.5ghz)

 

When I had the 2600 non-k I noticed temps were normal but the air blown out from the radiator was hot, with the 3770k the air is stone cold and one tube is hot.

 

Long story short: installed the 3770k applied some artic silver thermal paste (not too much and not too little)

 

Turned on the PC first thing I did was running Aida64 stress test after 2 minutes of running it crashed giving me blue screen of death

 

it said this: sbRSbLy.png

 

After the crash I tried to play a match of PUBG, crashed again after 1 minute in-game

 

after that crash I ran aida64 again it ran for 44 minutes without crashing (I stopped it manually) with this temps

 

RZPPNXU.png

 

after that I was able to play 3 matches of PUBG without any problems so I went to sleep

 

This morning I turned the PC on first thing I did was run Aida64 again, temps are totally different now 84º after only 1 minute running

 

Again the air coming out from the radiator (exhaust) is very cold! should be hot right?

 

My idle temps right now are: http://prntscr.com/isi2vu

pump is set to performance and fans go crazy loud when cpu is under load, I have the latest Corsair Link version and fan profile is set to default

 

I already checked the pump and mounted again temps still high, already repplied new thermal paste and still the same, I don't know what to do, I was not getting this problem with the 2600 non-k but seems like a cooling issue

 

EDIT: I have the latest Bios version https://prnt.sc/isi6uc

Edited by SgtDeathAdder
Link to comment
Share on other sites

AIDA will stop the test at a certain temperature threshold unless you alter it. In this case over 80C.

 

The physical symptoms you are describing sound like some type of flow failure, either pump or an actual restriction. However, the screen shot of Link only shows 37C H115i Temp (coolant temperature). That appears to match your other internal and if you have a cooler specific issue it is normally very high, above 50C.

 

You need to be able separate a cooler issue from all the other possibles. If there is some kind of flow issue, the H115i Temp will start climbing the moment you boot/wake and continue to slowly creep up, even when at the desktop. It is normal to idle about 4-7C above your room temperature. If you keep going past that and higher, that suggests an issue.

 

Your second AIDA test and Link graph look normal, but obviously there was some kind of issue. Check your H115i SATA connection to make sure it secure. If this is the issue, the entire assembly will lose power (fans, lights, etc.). That should be noticeable. However, if there is an internal power issue, the pump may be cutting in and out. Enable the Link graphing feature and take note of the H115i Pump speed. It should stay steady at ~3000 for performance or ~2000 for Balanced. If you see it drop to 0, obviously there is an issue.

Link to comment
Share on other sites

AIDA will stop the test at a certain temperature threshold unless you alter it. In this case over 80C.

 

The physical symptoms you are describing sound like some type of flow failure, either pump or an actual restriction. However, the screen shot of Link only shows 37C H115i Temp (coolant temperature). That appears to match your other internal and if you have a cooler specific issue it is normally very high, above 50C.

 

You need to be able separate a cooler issue from all the other possibles. If there is some kind of flow issue, the H115i Temp will start climbing the moment you boot/wake and continue to slowly creep up, even when at the desktop. It is normal to idle about 4-7C above your room temperature. If you keep going past that and higher, that suggests an issue.

 

Your second AIDA test and Link graph look normal, but obviously there was some kind of issue. Check your H115i SATA connection to make sure it secure. If this is the issue, the entire assembly will lose power (fans, lights, etc.). That should be noticeable. However, if there is an internal power issue, the pump may be cutting in and out. Enable the Link graphing feature and take note of the H115i Pump speed. It should stay steady at ~3000 for performance or ~2000 for Balanced. If you see it drop to 0, obviously there is an issue.

 

the last test I made it reached 100ºc I had to take the picture with my phone since I knew it was about to crash but for some reason, it didn't

 

evmzI7M.png

 

so I removed the pump, cleaned and re-applied the thermal paste from a different tube of artic silver this time, a little less paste let it sit for about 2 hours and turned on the PC

 

I connected the Sata plug in a different one this time

 

Temps on Aida64 are 70º to 75º

 

and idle: 45º - 50º

 

still no hot hair from the radiator only cold

 

What can I do? I have the stock cooler that came with the cpu but I don't think temps will be lower with it at 100%

 

this is how the pump graph looks like: http://prntscr.com/islx54

Edited by SgtDeathAdder
Link to comment
Share on other sites

Unfortunately, this does not clear things up. I can see the end CPU temps are out of line, but we are still looking for the source.

 

There does not appear to be any odd drops in the pump speed and I think you would notice a loss of system power where the fans and lights stop.

 

In the new screen grab, your H115i Temp is 40C. That is warm, but not where I would expect it to be with a pump or flow issues. If your coolant temp is is 40C, then the exhaust air coming out of the radiator will also be 40C. Most people would not describe 40C air being cool, so we need to focus on this. 1) What is happening to the H115i temp when you sit at the desktop? Does it go up all on its own? 2) After the test is over, does the H115i Temp cool down relatively quickly? (it should drop 5C is 5 min).

3) What is the approximate room temperature? How does this compared to the idle non-loaded H115i temp?

 

Without the coolant temp information, your problem reads like a contact issue. It is a very annoying thing to have to constantly set, clean, and re-paste hoping for better results. If we can resolve the question of whether the cooler is working, then the list of suspects can be narrowed down. You don't have to screen grab stuff if its difficult. Watch the H115i temp and see what it does.

Link to comment
Share on other sites

Unfortunately, this does not clear things up. I can see the end CPU temps are out of line, but we are still looking for the source.

 

There does not appear to be any odd drops in the pump speed and I think you would notice a loss of system power where the fans and lights stop.

 

In the new screen grab, your H115i Temp is 40C. That is warm, but not where I would expect it to be with a pump or flow issues. If your coolant temp is is 40C, then the exhaust air coming out of the radiator will also be 40C. Most people would not describe 40C air being cool, so we need to focus on this. 1) What is happening to the H115i temp when you sit at the desktop? Does it go up all on its own? 2) After the test is over, does the H115i Temp cool down relatively quickly? (it should drop 5C is 5 min).

3) What is the approximate room temperature? How does this compared to the idle non-loaded H115i temp?

 

Without the coolant temp information, your problem reads like a contact issue. It is a very annoying thing to have to constantly set, clean, and re-paste hoping for better results. If we can resolve the question of whether the cooler is working, then the list of suspects can be narrowed down. You don't have to screen grab stuff if its difficult. Watch the H115i temp and see what it does.

 

Hi there man, thank you for your reply I feel with the water on my neck since no one else is replying

 

I also had a H100i GTX in other PC, so I dismounted and installed that one in the 3770k re-aplied the thermal paste and turned on the PC, temps were 95ºc as soon it displayed desktop I was alarmed, then the pump made a noise and I noticed the fluid started running, then temps went down very quickly to 35's 40 idle

 

Ran Aida64 again and here's the result:

o8HYSnV.png

 

Fans go mad loud I checked the profiles and it was in quiet, so I changed it to default but still Loud as hell

 

Max temp I saw while running the test was 85º I think that's pretty hot since I'm not even overclocking which was the reason I purchased this CPU but that doesn't give me any room for it

 

this is a different AIO, you think the CPU can be defective?

 

the Air I felt wasn't warm in this one too, just cold is in Push config

Link to comment
Share on other sites

Defective CPU seems like long odds. So does having two coolers drop dead at the same time. I notice the H100i GTX is also sitting at ~40C coolant temp. Drive temps seem to be near 30C. Based on your time stamp, it looks like you are East or Central time US, so it can't be 30C in your room right now, although now i see the 33C motherboard temp, so the case is a bit warm. Where it the radiator positioned? (top, front?) Is it intake or exhaust.

 

I still need to know what is happening with the coolant temperature (H115i Temp or H100i GTX Temp) at idle. When not running any kind of test, does it creep up? How fast does the coolant temp drop when you stop the AIDA test?

 

You have a +55-60C core/package temp delta over coolant temp. That is high and this reads like a contact issue. Given it is happening on two coolers, first thing to check is the back plate. Is it turned the right way? Something keeping it from laying flat?

 

However, the coolant temp still seems a bit warm and when you have a contact issue you don't get heat into the coolant. It sits at 35C, you turn on the stress test, it still sits at 35C because the heat does not move efficiently through the cold plate. This is why I need to know what's going on with the coolant temperature. Your AIDA graph shows a clear creep up as the test progresses. This suggests the coolant temp is climbing during the test (normal).

Edited by c-attack
Link to comment
Share on other sites

Defective CPU seems like long odds. So does having two coolers drop dead at the same time. I notice the H100i GTX is also sitting at ~40C coolant temp. Drive temps seem to be near 30C. Based on your time stamp, it looks like you are East or Central time US, so it can't be 30C in your room right now, although now i see the 33C motherboard temp, so the case is a bit warm. Where it the radiator positioned? (top, front?) Is it intake or exhaust.

 

I still need to know what is happening with the coolant temperature (H115i Temp or H100i GTX Temp) at idle. When not running any kind of test, does it creep up? How fast does the coolant temp drop when you stop the AIDA test?

 

You have a +55-60C core/package temp delta over coolant temp. That is high and this reads like a contact issue. Given it is happening on two coolers, first thing to check is the back plate. Is it turned the right way? Something keeping it from laying flat?

 

However, the coolant temp still seems a bit warm and when you have a contact issue you don't get heat into the coolant. It sits at 35C, you turn on the stress test, it still sits at 35C because the heat does not move efficiently through the cold plate. This is why I need to know what's going on with the coolant temperature. Your AIDA graph shows a clear creep up as the test progresses. This suggests the coolant temp is climbing during the test (normal).

 

Right now the AIO temp shows 33.8º in Corsair Link at idle

 

I'll guess my room temp is 28 maybe, doesn't feel hot or cold, I live in South America, Venezuela

 

Ran the test again this time with Pump in performance (first test with this AIO was in quiet sorry)

 

Here's the screenshot: http://prntscr.com/isorkn

 

H100iGTX temp went to 36ºc and fans are not that loud now acceptable with headphones, H100iGTX temp is decreasing slowly back where it was before the test

Link to comment
Share on other sites

Whoops, should have figured that out from the motherboard temps and followed the meridian south. OK, well then some good news.

 

A normal coolant idle temp is 4-7C over your room temp. You are within that range. A normal coolant delta when running AIDA64 is +6-8C, depending on how long you run. It looks like that is also the case with a baseline of 33-34 and and a max of 40-41C. This means the cooler appears to be working and we can move on to other stuff.

 

Quiet mode (2000 rpm) vs performance (3000 rpm) should not matter for this test. If it gives clear differences in coolant temp for this short less than 10min AIDA runs, that is something entirely different and needs to be addressed.

 

Your temps in the most recent test above look quite good (in the 60-70C range). So we have this dramatic shift of almost 20C between some runs or gaming time. If the pump is not shutting itself off, then it stands to reason the pump's cold plate is somehow shifting or losing contact with the CPU lid. Just by a tiny amount. If it were anything readily visible you would overheat on boot, so this really is a tiny amount if true. The next time the CPU is kicking out 80C like temps, press down on the pump block with two fingers and see if the temps drop. That would be a clear sign things are not set.

 

Something else I noticed is your motherboard temp reads 40C in AIDA during the last test. If the whole case is that warm, then the coolant will be as well. The coolant temp is the minimum possible CPU temp, and thus when the coolant is 40C, your idle core temps will 40ish as well, even with power saving features like c-states and adaptive voltage. Is there any reason for the Sabertooth to read this warm? It is just a single sensor, so it can be heavily influenced by a GPU load in close proximity or something else venting heat.

 

The only other possible thing that might fit this situation is an air bubble or some other restriction is blocking the coolant from entering the pump head. This would reduce flow and should create higher coolant temperatures and might certainly be erratic in frequency. However, it usually also comes with bags of irritating noise and you might notice. Nevertheless, you can try tilting the case back and forth a little (while the pump is running) if the temps spike again. The problem with this is I can't see it happening to two coolers simultaneously.

Link to comment
Share on other sites

Whoops, should have figured that out from the motherboard temps and followed the meridian south. OK, well then some good news.

 

A normal coolant idle temp is 4-7C over your room temp. You are within that range. A normal coolant delta when running AIDA64 is +6-8C, depending on how long you run. It looks like that is also the case with a baseline of 33-34 and and a max of 40-41C. This means the cooler appears to be working and we can move on to other stuff.

 

Quiet mode (2000 rpm) vs performance (3000 rpm) should not matter for this test. If it gives clear differences in coolant temp for this short less than 10min AIDA runs, that is something entirely different and needs to be addressed.

 

Your temps in the most recent test above look quite good (in the 60-70C range). So we have this dramatic shift of almost 20C between some runs or gaming time. If the pump is not shutting itself off, then it stands to reason the pump's cold plate is somehow shifting or losing contact with the CPU lid. Just by a tiny amount. If it were anything readily visible you would overheat on boot, so this really is a tiny amount if true. The next time the CPU is kicking out 80C like temps, press down on the pump block with two fingers and see if the temps drop. That would be a clear sign things are not set.

 

Something else I noticed is your motherboard temp reads 40C in AIDA during the last test. If the whole case is that warm, then the coolant will be as well. The coolant temp is the minimum possible CPU temp, and thus when the coolant is 40C, your idle core temps will 40ish as well, even with power saving features like c-states and adaptive voltage. Is there any reason for the Sabertooth to read this warm? It is just a single sensor, so it can be heavily influenced by a GPU load in close proximity or something else venting heat.

 

The only other possible thing that might fit this situation is an air bubble or some other restriction is blocking the coolant from entering the pump head. This would reduce flow and should create higher coolant temperatures and might certainly be erratic in frequency. However, it usually also comes with bags of irritating noise and you might notice. Nevertheless, you can try tilting the case back and forth a little (while the pump is running) if the temps spike again. The problem with this is I can't see it happening to two coolers simultaneously.

 

Yes this temps looking more forgiving now: http://prntscr.com/isp3kx in this screenshot I have the air conditioner of my room on temp 18ºc, I notice better temps with the H100i but I can't keep that one, so I have to go back with the H115i which last time marked 100ºc I also have an old cooler master V8 its gonna be a pain to install but what you think should I try it?

 

Just to clarify: did you said its possible for the CPU to be defective and not the coolers? I have that doubt about the CPU since its a second-hand CPU and the person might have sold it if it had any kind of problems

Link to comment
Share on other sites

The defective cpu theory seems the least likely. If damaged (in whatever way) the results would be permanent and not erratic. It would be hot all the time and no cooler would make any difference.

 

The air tower is often a real pain to install and there is no clear evidence the H115i is not working. I think it would be better to continue testing with the H115i and try to solve this.

Link to comment
Share on other sites

The defective cpu theory seems the least likely. If damaged (in whatever way) the results would be permanent and not erratic. It would be hot all the time and no cooler would make any difference.

 

The air tower is often a real pain to install and there is no clear evidence the H115i is not working. I think it would be better to continue testing with the H115i and try to solve this.

 

got it, will keep trying with the H115i and let you know, Thank you very much man

Link to comment
Share on other sites

The defective cpu theory seems the least likely. If damaged (in whatever way) the results would be permanent and not erratic. It would be hot all the time and no cooler would make any difference.

 

The air tower is often a real pain to install and there is no clear evidence the H115i is not working. I think it would be better to continue testing with the H115i and try to solve this.

 

hey man, me again

 

mounted the H115i again with fresh thermal paste, this time I didn't attach the radiator to the case to try different positions

 

idle temps were between 45 and 50

 

ran Aida64 in no moment temps went 100º and 105º thermal throttling went 73% in 7 minutes

 

DbpaoW2.png

 

This is what you'll find interesting: H115i temp was 65.6ºC

6e4DpxR.png

 

room temperature is same than yesterday

 

Radiator air was cool, but a part of the radiator where the warm hose its connected was hot

 

I tried moving the radiator in vertical horizontal but temps were the same, while running at those temps I unscrew a little the screws nothing changed so I tighten up again

 

now I'm using the Intel stock cooler, idle temps are 55 to 60º. with Aida64 gets to 95 - 100ºC

Edited by SgtDeathAdder
Link to comment
Share on other sites

The idle temps at 3.7GHz for 3770 are 50-60????? I have not used a stock fan in ages, but that seems really high. I would like someone else to add this.

 

 

As for the H115i, you clearly have some type of blockage. It seems to be mobile at this point and thus the inconsistent results. In your last AIDA run, you can see temps take a huge jump a few minutes into the test. The AIDA blend test you ran is not one of those that has a short warm up and then drops the hammer. It is consistent from the start. That is probably the moment when the obstruction shifted to block the strainer. There is no reason for us to try and pin that part down. All you need to prove malfunction is the Link screen shot above with the crazy high 65C coolant temp and the completely wacky 6000+ rpm pump speed. No more testing with that cooler.

 

Unfortunately, I am not sure where that leaves you. If you have the purchase invoice, you will still be under the 5 year warranty. The speed or nature of that process in your current location may be another matter. You may want a working cooler now and then to save the replacement as a back-up, secondary system tool, or just market value sale. I know you have some reservations about the CPU, so getting another working cooler on there seems like a priority to confirm things are as they should be. The test with the H100i v2 really complicates things.

Link to comment
Share on other sites

The idle temps at 3.7GHz for 3770 are 50-60????? I have not used a stock fan in ages, but that seems really high. I would like someone else to add this.

 

 

As for the H115i, you clearly have some type of blockage. It seems to be mobile at this point and thus the inconsistent results. In your last AIDA run, you can see temps take a huge jump a few minutes into the test. The AIDA blend test you ran is not one of those that has a short warm up and then drops the hammer. It is consistent from the start. That is probably the moment when the obstruction shifted to block the strainer. There is no reason for us to try and pin that part down. All you need to prove malfunction is the Link screen shot above with the crazy high 65C coolant temp and the completely wacky 6000+ rpm pump speed. No more testing with that cooler.

 

Unfortunately, I am not sure where that leaves you. If you have the purchase invoice, you will still be under the 5 year warranty. The speed or nature of that process in your current location may be another matter. You may want a working cooler now and then to save the replacement as a back-up, secondary system tool, or just market value sale. I know you have some reservations about the CPU, so getting another working cooler on there seems like a priority to confirm things are as they should be. The test with the H100i v2 really complicates things.

 

I just submitted a ticket for RMA explaining the case, I went ahead and installed the old V8, idle temps are 40 to 45º Aida64 temps are 65 max

 

U0cdOie.jpg

 

5XSDwhQ.jpg

 

this thing is silent! even when CPU is at 100% I might even try OC what you think? if I do what maximum temps should I accept?

Link to comment
Share on other sites

My magic line has always been around 80C. I skipped over the Ivy Bridge line so I can't offer any specific experiences. If your AIDA peak is 85C, but actual load use is less, then I don't really see an issue.

 

got it, thanks so much for all your help man, you have been very helpfull

Link to comment
Share on other sites

 Share

×
×
  • Create New...