High temps with H115i Elite Capellix on a I9-9900k

[xLogick]

First my specs:
CPU: i9-9900K
MOBO: Aorus Z390 Master
RAM: 32gb (4x8) Corsair Vengeance LPX
GPU: MSI Gaming X trio RTX 3090
SSD: Samsung 970 Evo Pro
Case: Corsair Carbide Air 540 + DEMCiflex Dust Filters (For the record the filters are placed at the top, bottom, front(along with the included filter), side, and back as pictured in the included link)        

Now the cooling:
Front Fans(intake): 3x 120mm Noctua NF-S12B redux-1200 PWM
Back Fan(exhaust): Noctua 140mm NF-A14 PWM
CPU Cooler: Corsair H115i Elite Cappelix + 2x Noctua 140mm NF-A14 PWM
Paste: Thermal Grizzly Kryonaut
SSD Heat Sink: Sabrent M.2 2280 SSD Rocket Heatsink

Ambient room temps are about 18c-20c. (according to my phone)

I used to hit 30-32 c cooling when I had an OG H115i, but now I'm hitting around 35 c when idle and I have to put all my fans to the max. To make things worst I left my PC on for a few hours and somehow Disco Elysium was downloaded and running on the title screen (I'm still not sure how that happened, I finished and uninstalled months ago). My coolant temps were at 50 c!!?!?! I'm wondering what could be the cause? I've reapplied thermal paste twice. Could this be a bad cooler(happened to me with the OG H115i, I RMA'ed it)? Or could it be the dust filters? Or event bad airflow? Below is the inside of my PC any advice would be appreciated!

 

[c-attack]

39 minutes ago, xLogick said:

My coolant temps were at 50 c!!?!?!

Not TIM or contact issues.  That would prevent efficient thermal transfer.  CPU temp would be high and coolant would be low.  

 

+15C is a pretty large jump and beyond what you can generate from the CPU alone without shutting off the fans or pinning them at the minimum.  However, I would not make any decisions based on a strange incident like this.  Try a moderate CPU stress test like the one in the CPU-Z (bench tab).  It's linear so it is very easy to watch the +1C coolant to +1C CPU climb after the 100% load start.  I would expect +8C in about 10 min and it levels off after that.  You only need to run it for 5-10 min to establish functionality.  A cooler with a partial blockage or other flow issue with start skipping up the temp scale right from the start.  If you do go +15C in 5 min, something is not quite right.  

 

Keep in mind a +3C coolant temp variance is pretty easy to generate, especially if you have uneven room temperature zones, like in Summer.  I don't know if the filter are new, but you can lift off the top one above the radiator during use to claw back another 1-2C if required.  

Edited August 4, 2021 by c-attack

[xLogick]

Ok I removed the top filter and ran two stress tests, one set to balanced, one set to quiet.  Both where at 6 minutes with an ambient of 25c.

balanced
cpu: 45 c
coolant: 37.5c  
 

balanced
cpu: 46 c
coolant: 38.1c  
 

It seemed the cpu shot up but the coolant barely did.  Is this normal?

 

[xLogick]

Note: the above the CPU power was at 80% and I am unable to edit the post for some reason.

Ok I removed the top filter and ran two stress tests, one set to balanced, one set to quiet.  Both where at 6 minutes with an ambient of 25c.

balanced
cpu: 72c
coolant: 40.2c  
 

balanced
cpu: 74 c
coolant: 41.8c  
 

It seemed the cpu shot up but the coolant barely did.  Is this normal?

 

Edited August 4, 2021 by xLogick
note to explain why the double post

[c-attack]

Yes, it's normal to see a coolant to CPU differential in that range at full voltage.  The two have an additive relationship and coolant temperature represents the minimum possible CPU at 0v.  However, your "CPU cooling" is mostly conductive.  You apply electricity to one side of the CPU.  It heats up.  You clamp a metal plate on the other side to try and conduct it through and out to somewhere else.  All CPU cooling works this way and ultimately you are limited by the materials and their physical properties (things you have no control over) and then the one you do -- voltage.  Pretty much any cooler is going to have a similar CPU temp when you put sample A in the socket and apply 1.xx volts -- at least for about 1 second.  It's what happens after that differentiates the various cooler types and sizes.

 

The coolant, radiator, and fans are really waste heat disposal.  It's picking up the heat at the CPU block and dumping it somewhere else.  It takes a specific amount of heat (watts) to raise the coolant temperature +1C, less the amount of heat you are able to dissipate through the fans.  Since the coolant temp is the minimum CPU temp, you are altering end CPU temp by lowering/raising the baseline. You can take your estimated coolant to CPU differential from the CPU test to make a prediction for possible max CPU temps and give your self a maximum allowable coolant temp.  Your differential from the 100% load was approximately +32C.  That is good number and suggests you are not heavily overclocking.  Most people pushing their CPUs toward the limit will see a differential of +50C or so.  If your coolant had been 30C, your CPU-Z test would have been 62C.  If your coolant temp had been 60C, you would have been at 92C.  So in this manner, if your personal limit is no CPU temps over 80C, you know 48C coolant temp is the maximum you can allow.  That should not be an issue.

 

Back to the present issue, I don't see anything unexpected with the cooler.  I am going to assume the starting coolant temp was around 35C, like in the earlier screenshot.  That puts the coolant rise right at expected levels.  Your two tests were consistent for differential (expected), the coolant temp difference is small and if I had told you to standardize settings (fixed fan speed, pump, etc), they would still turn out the same.  Pump speed should not make too much difference on the Elite unit, maybe 1C between Quiet and Extreme.  You can set that for noise.  Fan speed you should also set on noise.  You can penalize yourself by keeping them off or pinned at the minimum, but the difference between 1000 rpm and 1300 rpm might be 2-3C at 300W.  You are going to be running half that at most for normal use, so you can run relaxed on the fans until you feel the CPU temps are not acceptable.  You have plenty of safe headroom above.

 

The hard part is over-analyzing coolant temp numbers.  Coolant temp is affected by the room temperature, the local case temperature, other sources of heat (GPU), the CPU heat added directly, less the heat expelled through the radiator.  It is extremely difficult variable to make bold assessments on, except when it is drastically out of range.  I do over analyze this stuff all the time and even with thermal cameras it can be difficult to make sense out of the numbers.  One of the things probably in the back of your mind is why is the coolant idling at 35C if the room is 25C.  It will go to 25C if you power off the PC for an hour or two.  That is the minimum.  Most people will see an idle coolant temperature of around +4-7C above their room temp, but again it is influenced by a lot of variables, like radiator location, size, CPU idle power, Windows power plan or activity level, fan speed, pump speed, case design, and more often location in the room.  We generalize that our room temp is "21C" (or whatever), but the room has differences between floor and ceiling and other parts of the room.  This is especially true in Summer where you are blowing cold streams of air from the AC vents to try and lower the temperature of a room that is much warmer.  You also will heat up the area around the case, especially after longer load sessions.  The PC case is a hot rock.  All of that is a mess of possibilities, but the bottom line is if your internal case temperature is 31C, then a 35C idle coolant value is right on the expected mark.  Take a peek at your MB temp sensor at idle.  When the GPU is not blasting against it, this is a decent indicator of general case temp as its usually in the middle of the motherboard.  

Edited August 4, 2021 by c-attack

[xLogick]

Thank you for the info.  I'm less worried about my PC temps now!