Sudden high idle temps on 9900k with Corsair Elite Capellix H150i

[Jooren]

Hello,,

 

I'm new on this forum.

I've had my Corsair Elite Capellix H150i for about 1 year and 4 months now.

Since today, I noticed my CPU was going up to 75° Celsius in idle (right after startup) and 78-80°C while just browsing the web. Usually I get about 30-40°C in idle.
Pump is in quiet mode (always is), fans (Corsair QL120) are in balanced mode (always are).
I then saw that there were firmware updates for the Capellix and for the LCD screen, as well as an iCue update. I thought these updates would fix the issues. They did not. Restarted computer, hight temps remain. Checked temps with CoreTemp, temperature readings were the same.

I was thinking that maybe my pump failed, but iCue says my liquid temperature is around 23,40°C, and my pump is running at around 2265 RPM.
I have changed litterly nothing about my setup in the past days, have not moved my computer, and the ambient temperature in my room has not gone up.

No idea what is causing the problem, but I am now 'afraid' of starting up a video game or letting my computer do any heavy tasks.

Any help would be appreciated.

[Jooren]

1 minute ago, Jooren said:

Hello,,

 

I'm new on this forum.

I've had my Corsair Elite Capellix H150i for about 1 year and 4 months now.

Since today, I noticed my CPU was going up to 75° Celsius in idle (right after startup) and 78-80°C while just browsing the web. Usually I get about 30-40°C in idle.
Pump is in quiet mode (always is), fans (Corsair QL120) are in balanced mode (always are).
I then saw that there were firmware updates for the Capellix and for the LCD screen, as well as an iCue update. I thought these updates would fix the issues. They did not. Restarted computer, hight temps remain. Checked temps with CoreTemp, temperature readings were the same.

I was thinking that maybe my pump failed, but iCue says my liquid temperature is around 23,40°C, and my pump is running at around 2265 RPM.
I have changed litterly nothing about my setup in the past days, have not moved my computer, and the ambient temperature in my room has not gone up.

No idea what is causing the problem, but I am now 'afraid' of starting up a video game or letting my computer do any heavy tasks.

Any help would be appreciated.

EDIT: changing pump speed and fan speeds to 'extreme' does not lower the temperature on my CPU (9900k).
Could it be a problem with contact between the cooling plate and the CPU, or cooling paste?
If so, how did this problem suddenly arise?

[c-attack]

When you first cold boot the PC after a decent period of shutdown, does the CPU temp start close to normal and then steadily build over several minutes to that 70-80C baseline?  That is the sign of a blocked flow in the coolant system and once this has happened and the liquid in the cpu block heats up, the cpu temp will stay elevated until the heat can naturally dissipate. That could take several hours. 
 

Another think to check is your Vcore. If it’s pegged at ~1.40v or more, you are in a constant state of load with a rather large amount of voltage. This has happened to some users in the past, but is not cooler related. Usually it comes from auto overclocking or a bios mishap. 
 

On the other hand, if your Vcore is sitting at 0.65v and you’re at 80C, something is wrong. Bad contact or TIM issues are normally very erratic. So it might drop to 45-50C when the voltage drops off, but then hits 95C when the voltage is full. The cpu temp jumps up and down continuously. Steady and level at 80C seems like a cooler problem, despite what CUE lists as the coolant temp. 

[Jooren]

12 minutes ago, c-attack said:

When you first cold boot the PC after a decent period of shutdown, does the CPU temp start close to normal and then steadily build over several minutes to that 70-80C baseline?  That is the sign of a blocked flow in the coolant system and once this has happened and the liquid in the cpu block heats up, the cpu temp will stay elevated until the heat can naturally dissipate. That could take several hours. 
 

Another think to check is your Vcore. If it’s pegged at ~1.40v or more, you are in a constant state of load with a rather large amount of voltage. This has happened to some users in the past, but is not cooler related. Usually it comes from auto overclocking or a bios mishap. 
 

On the other hand, if your Vcore is sitting at 0.65v and you’re at 80C, something is wrong. Bad contact or TIM issues are normally very erratic. So it might drop to 45-50C when the voltage drops off, but then hits 95C when the voltage is full. The cpu temp jumps up and down continuously. Steady and level at 80C seems like a cooler problem, despite what CUE lists as the coolant temp. 

I'll check again tomorrow after cold boot, if it slowly rises and stays, or if it hits 75-80° C almost instantly.

I'll look into the voltages as well.

If these temps/voltages are as you say, would you recommend RMA'ing the thing, or do I try reapplying paste and reseating it first?

If RMA, how long does that usually take?

 

Thanks for the input.

[c-attack]

If it’s erratic and jumpy, then re-paste and the-mount. Continuously building temperature that then stays elevated and flat suggests cooler problem (RMA). 

[Jooren]

On 5/10/2022 at 11:06 PM, c-attack said:

If it’s erratic and jumpy, then re-paste and the-mount. Continuously building temperature that then stays elevated and flat suggests cooler problem (RMA). 

Hey, sorry it took me a while to reply.

Right now I have a few browser tabs open + iCue + CPU-Z + windows explorer is copying some files.
 

CPU Load: 20%

CPU temp: 75°C
Liquid temp: 22°C
Vcore: 1,25-1,30V

CPU temps and liquid temps do not drop when I increased fan speeds or pump speed.

I've cleaned, repasted and reseated the cooler twice in the past month and a half, doesn't solve it.
 

What concerns me is that most people talk about their Liquid temps (with same AIO) as around 30-40°C.
Mine is 22°C, which is low, but my CPU temps are high..
Seems like a problem with the cooler? Some sort of blockage or malfunctioning pump?

 

Thanks in advance for the help.

[c-attack]

That reads more like a contact/conductivity problem.  Liquid temp idle level varies, but most people will be 4-7C above the room temp when the PC is fully warmed up at idle after 15 minutes or so.  That's mostly related to environmental issues, radiator location, and how much resistance you get from your case panels.  Now if you are going to tell me it's 17C in the room then the 22C seems OK, but very few people can hold 22C as a coolant temp outside the dead of Winter.  I think the value is a bit suspect and I have some doubts about the integrity of the sensor right now.

 

Everything else you've said illustrates a pretty classic picture of internal flow restriction.  CPU starts off at 30-40C but immediately starts climbing to 70-80C.  Your coolant to CPU temp differential is very large at 55C and that is typically what you see at the limit of overclocking.  I don't think you're quite there yet with that voltage.

 

So I think priority one is figuring out whether the liquid temp number is false.  If so, then it needs to be replaced via warranty now anyway.  Liquid temp should be about the same as the "motherboard temp" or the 4-7C above room temp estimate.  

[Jooren]

1 minute ago, c-attack said:

That reads more like a contact/conductivity problem.  Liquid temp idle level varies, but most people will be 4-7C above the room temp when the PC is fully warmed up at idle after 15 minutes or so.  That's mostly related to environmental issues, radiator location, and how much resistance you get from your case panels.  Now if you are going to tell me it's 17C in the room then the 22C seems OK, but very few people can hold 22C as a coolant temp outside the dead of Winter.  I think the value is a bit suspect and I have some doubts about the integrity of the sensor right now.

 

Everything else you've said illustrates a pretty classic picture of internal flow restriction.  CPU starts off at 30-40C but immediately starts climbing to 70-80C.  Your coolant to CPU temp differential is very large at 55C and that is typically what you see at the limit of overclocking.  I don't think you're quite there yet with that voltage.

 

So I think priority one is figuring out whether the liquid temp number is false.  If so, then it needs to be replaced via warranty now anyway.  Liquid temp should be about the same as the "motherboard temp" or the 4-7C above room temp estimate.  

If it's a problem with contact (seating the cooler?) or conductivity (cooling paste?), then it's weird that the problem suddenly started happening, after changing absolutely nothing (never even changed BIOS settings or anything), and that repasting (once with Noctua paste, once with Kryonaut paste, each time thoroughly cleaning off the previous paste with the stuff that's meant for it) and reseating the cooler TWICE did not fix the problem.

How do I find out if my liquid temp is false? Is there anything besides iCue that can show me the correct liquid temp?
And if my liquid temp is false, and if it's higher, I still have to fix the problem which is high CPU temps with low CPU load.

 

 

Thanks in advance.

[c-attack]

1 hour ago, c-attack said:

Liquid temp should be about the same as the "motherboard temp" or the 4-7C above room temp estimate.  

Open up AI Suite if you use it or something like HWMonitor.  You are looking for the "motherboard temp", which is typically in the motherboard section of the monitoring tool.  Or if you have a general idea of your room temp and whether you are sitting in 17C environment.  

[Jooren]

6 minutes ago, c-attack said:

Open up AI Suite if you use it or something like HWMonitor.  You are looking for the "motherboard temp", which is typically in the motherboard section of the monitoring tool.  Or if you have a general idea of your room temp and whether you are sitting in 17C environment.  

Not at my PC at the moment, but it was least 5°C hotter in my room (like 20-25°C at least).

[c-attack]

And that means probably 27-29C in case and it's hard to imagine the coolant can hold 22C in there even with zero heat conducted in.  However, we need to get you a more solid foundation before going to support.  If you go in with 22C coolant in 25C room, they are going to run you through contact questions again.

 

The single definitive assessment of any cooling system not performing is when the CPU temp skip counts up 45..50..55..60..65..70.. etc right after you cold boot.  That is the signature of a low flow state.  Once you have done this one time, the hot liquid in the CPU block can't get to the radiator and so it only cools down when the entire PC is shut off for hours and the heat can naturally dissipate in the liquid.  The next time you cold boot in the morning, try to see if you can catch the CPU doing.  You can stop it at the BIOS and watch from there to get to the temp measurement quicker.  Liquid temp and CUE are not relevant for this.  If we can establish this is happening, then the other stuff does not matter.  

 

On the other hand, if you boot up and CPU is instantly at 80C but then keeps dropping to 30C when the voltage falls off (and back and forth), that is something else.  Possible contact problem or possible voltage problem.  

 

 

[Jooren]

8 hours ago, c-attack said:

And that means probably 27-29C in case and it's hard to imagine the coolant can hold 22C in there even with zero heat conducted in.  However, we need to get you a more solid foundation before going to support.  If you go in with 22C coolant in 25C room, they are going to run you through contact questions again.

 

The single definitive assessment of any cooling system not performing is when the CPU temp skip counts up 45..50..55..60..65..70.. etc right after you cold boot.  That is the signature of a low flow state.  Once you have done this one time, the hot liquid in the CPU block can't get to the radiator and so it only cools down when the entire PC is shut off for hours and the heat can naturally dissipate in the liquid.  The next time you cold boot in the morning, try to see if you can catch the CPU doing.  You can stop it at the BIOS and watch from there to get to the temp measurement quicker.  Liquid temp and CUE are not relevant for this.  If we can establish this is happening, then the other stuff does not matter.  

 

On the other hand, if you boot up and CPU is instantly at 80C but then keeps dropping to 30C when the voltage falls off (and back and forth), that is something else.  Possible contact problem or possible voltage problem.  

 

 

This morning, after cold boot, temperature went from 30-40 to 65-70 an stayed there, with liquid temps at 15°C. Motherboard temps at that time were about 30-40°C.

[c-attack]

Ok, that’s what needs to go in the support ticket. That’s a developing flow issue and a inaccurate temp sensor. Make sure when you describe it you say you watched it climb from 30 to 70C and not that you turned it on and it was 70C. That should make the state of things clear. 

[Jooren]

6 minutes ago, c-attack said:

Ok, that’s what needs to go in the support ticket. That’s a developing flow issue and a inaccurate temp sensor. Make sure when you describe it you say you watched it climb from 30 to 70C and not that you turned it on and it was 70C. That should make the state of things clear. 

So I need to RMA my cooler?

Another thing is, when I 'stress test' my CPU with Prime95 or OCCT, the CPU temp goes up to 70-80 °C, but they don't go below 70°C after stopping the test. During the stress tests, liquid temp does not really go above 20°C

Edited June 27, 2022 by Jooren

[c-attack]

Well, that is a peculiar detail.  If you can run Prime95 or OCCT, you don't have a flow issue.  When that happens the coolant in the block stays trapped there and continuously heats up to 50--70C, which in turn also causes the CPU to idle at that same temperature with minimal voltage.  However, if when you turn on a stress test the CPU should go 30-50C on top of that immediately which would then cause the test to instantly fail when it hits 105C or whatever is TJ Max.  It also removes "contact" as a potential problem.  You can't run Prime/OCCT with a contact problem.  

 

So you have two issues.  1) the weird coolant readings that can't be accurate; 2) A seemingly elevated CPU temp at the desktop, which appears to keep the Vcore loaded.  

 

Make sure you are not running HWiNFO, AIDA, or another full spectrum monitoring program that is polling the H150i Elite Capellix.  That can cause erroneous data, lock ups, or missing data from the CUE panel.  However, that has not come up and usually the data is impossible (65,000 rpm fans) and not a slightly offset coolant temperature.  

 

Open the Windows task manager and kill the following processes.  iCUE App, Corsair.Service., Corsair.Service.CPUIDRemote.  That will stop CUE from running and its background monitoring services.  Now open another monitoring program like HWiNFO or HWMonitor to get some CPU temp data from another source.  Note the Vcore voltage and CPU temps.  Your CPU temps should drop down to the 30s with a Vcore somewhere near 0.67-0.75v when the CPU is truly idle.  It would be concerning if you CPU temps stay at 70C with a Vcore of 0.67v.  

 

Finally, run OCCT or Prime again, still with CUE closed.  See what temps you get.  Run it for about 5 minutes.  The CPU temps should slowly go +1C about every 20-30 seconds.  That's the actual liquid temp baseline increasing by +1C and the CPU elevated by the same amount.  

 

I am not quite sure what's going on here.  The data doesn't connect with any normal or malfunctioning condition.  

Edited June 27, 2022 by c-attack

[Jooren]

6 hours ago, c-attack said:

Well, that is a peculiar detail.  If you can run Prime95 or OCCT, you don't have a flow issue.  When that happens the coolant in the block stays trapped there and continuously heats up to 50--70C, which in turn also causes the CPU to idle at that same temperature with minimal voltage.  However, if when you turn on a stress test the CPU should go 30-50C on top of that immediately which would then cause the test to instantly fail when it hits 105C or whatever is TJ Max.  It also removes "contact" as a potential problem.  You can't run Prime/OCCT with a contact problem.  

 

So you have two issues.  1) the weird coolant readings that can't be accurate; 2) A seemingly elevated CPU temp at the desktop, which appears to keep the Vcore loaded.  

 

Make sure you are not running HWiNFO, AIDA, or another full spectrum monitoring program that is polling the H150i Elite Capellix.  That can cause erroneous data, lock ups, or missing data from the CUE panel.  However, that has not come up and usually the data is impossible (65,000 rpm fans) and not a slightly offset coolant temperature.  

 

Open the Windows task manager and kill the following processes.  iCUE App, Corsair.Service., Corsair.Service.CPUIDRemote.  That will stop CUE from running and its background monitoring services.  Now open another monitoring program like HWiNFO or HWMonitor to get some CPU temp data from another source.  Note the Vcore voltage and CPU temps.  Your CPU temps should drop down to the 30s with a Vcore somewhere near 0.67-0.75v when the CPU is truly idle.  It would be concerning if you CPU temps stay at 70C with a Vcore of 0.67v.  

 

Finally, run OCCT or Prime again, still with CUE closed.  See what temps you get.  Run it for about 5 minutes.  The CPU temps should slowly go +1C about every 20-30 seconds.  That's the actual liquid temp baseline increasing by +1C and the CPU elevated by the same amount.  

 

I am not quite sure what's going on here.  The data doesn't connect with any normal or malfunctioning condition.  

Alright.
Today I noticed that 3 of my cores stayed on 100% while 'idle', and temps remained around 70°C.
BUT, when I opened Windows task manager, the cores went down and the temperature dropped to 30-40°C.
Did some searching and found threads on Reddit claiming that it would be some kind of mining virus, that deactivated (and 'hid', so to speak), whenever you opened any kind of task manager.
So, I did some scanning, nothing came up.
Decided to reinstall Windows, then driver, iCue, HWinfo, CPUZ, OCCT.

The 100% problem with high temps, at 'idle', has now been solved (hooray).


So, after that, doing what you suggested.
Switched off everything + iCue related, so CPU was as idle as possible.
Opened up HWinfo. Voltage dropped, CPU temp was about 30-40°C.

After that, I ran 5mins of OCCT, highest CPU temp was 87°C (near the end of the 5mins), but on average my temps were about 70-75°C.

Stopped the test, temps, activity and voltage dropped again.
As I am typing this right now, with 1 chrome tab open + HWinfo open, HWinfo says my CPU Core temperature is 33°C.

EDIT:
Now I just opened iCue again, it reports 35-41°C CPU temp, but still reports just 15°C.
In fact, just ran other 5min of OCCT (this this without HWinfo), but with iCue on...), In 5min48sec, the liquid temperature rose from 15°C, slowly to 20°C.

So, everything is withing acceptable parameters now?

Thanks btw, for your continued help.
 

[c-attack]

OK, the weird temperature stuff appears to have an explanation.  That was really spamming things up as we tried to solve this.  Cross that off.  There is no contact problem.  There is no flow problem.

 

That leaves one thing -- the inaccurate coolant temp reading.  This is not common and there is no way to know if its the hardware (sensor end) or some bizarre software/firmware glitch.  I've got a meter of radiators lined up and I can't keep the coolant below 20C in Winter with load, so there is little question it is off.  The only possible user remedy would be to force update the firmware (check for the update -> None -> Force update).  If you were already stuck, I'd say go ahead and there is nothing to loose.  However, these firmware updates can be a bit odd at times on the Elite series.  You might submit a support ticket to Corsair and wait for them to say the same thing.  Just stick to the coolant stuff.  Turn on PC, coolant reads 15C in a 25C room, run OCCT it goes to 20C.  Then if things do not go well, you have a clear path to replacement.  

 

In the meantime, the normal presets and H150i Temp will be not overly helpful for fan control.  You can either 1) set a comfortable fixed speed prior to gaming/load use; or 2) Make a custom curve for the negative offset coolant.  That might be a bit tricky since the graph minimum is 20C, but that's the part you need.  Put the first point at 20C=750 rpm (or thereabouts) and the slope it up to 1300 around 25C.  That should keep you functional.