Jump to content
Corsair Community

Setting up a proper silent fan curve without compromising cooling.


Recommended Posts

Hello everybody 😀,

first of all: this is my first post here and I hope I'm doing nothing wrong and selected the correct Sub to post my problem. If I'm doing something wrong please don't hesitate telling me 😁

I researched a lot now also through this forum and tried to set up a custom fan curve and I need to say: I read so much that I'm currently more confused than anything else. I need to admit, that I'm not a pro when it comes to this, so I would really appreciate to get some help here, as I'm currently really desperate as I don't know what to do.
 

MY SETUP:

In my profile I already filled out my hardware components under "Gaming PC". There you can see my hardware: 
My Fan Setup: 
In total, i have:

  • 6 Case Fans (Corsair QL) 
  • 3 AIO FANs (CORSAIR ML which came with my Corsair ICUE H150i ELITE CAPELLIX)
  • 1x Commander Pro (6 QL Fans attached to it)
  • 1x Commander Core (3 ML Fans attached to it)

NOTE: I'm using a Push and Pull Setup, this means: I have 3 Fans at the front of my Case (QL Fans) than the radiator behind and after the radiator 3 ML Fans (so 6 Fans attached to the radiator).

I use my pc in the living room plugged in into my LG 4k TV and I'm sitting on the couch using a Headset playing demanding games like RDR2, Control..., but also streaming my console gaming using the capture card where i simply stream via Discord and use OBS.

ATTACHMENT:

I attached 4 Files to this post:

  • Icue Commander Curve: shows my custom created fan curve using the Cooleant temperature as base (Controlling all 6 QL Fans)
  • Icue Elite Capellix: shows my custom created fan fan curve using the Cooleant temperature as base (Controlling all 3 ML Fans)
  • ICue Dashboard: Showing a benchmark running RDR2 in the backround using my custom fan curves
  • ICue Dashboard Intensive: Showing a benchmark running RDR2 in the backround using the Intensive Preset.


What is the problem: In ICue there are three presets: 

  • Quiet
  • Balanced
  • Intensive

All of them make a very loud noise during Gaming. If you are Gaming and you can hear the fans through the  Headset and the gaming audio, then this is really annoying. 
The Quiet preset is okayish from noise point of view, the only issue is that on normal tasks and idle the fans keep ramping up a bit and slowing down the whole time, so you can always here this annoying sound this makes when accelerating the fans and slowing them down again. (And as far as I also read in the forum in other threads here I learned this is the worst case scenario). 
Balanced and Intensive are making always a lot of noise -> I decided to make an own curve, BUT: I performed a gaming benchmark using Red Dead Redemption running in the Background cranked it up to max Settings and what is worrying me are the temp differences: 

On my custom curve I can see a temperature difference (comparing to the intense preset) of 6 Degrees (maybe more if would have run an hour (check the two Dashboard pictures)).

What is worrying me are the temps in general and also of the DRAM. Browsing through the forum I read that people have something aroung 50-60. 

Which temperatures are okay? What should I do because of the noise which is bugging me? Which temperatures are unhealthy for my system? 
Could anybody maybe help me create a fun curve where I simply finally can relax and enjoy gaming because I can trust that this is a good curve? As already said: Im not an expert and far away from that and while playing im constantly checking temps and to be honest: This takes the fun out of it. I'm worrying that I'm damaging my system but on the other hand using the intense preset is not acceptable as it is as loud as on an airport. I simply would like to get assistance in creating the perfect fan curve for my specific setup and situation. This would be really kind 🙂

Thank you in advance for your time and help! 

Icue Commander Curve.png

ICue Dashboard Intense.png

ICue Dashboard.png

Icue Elite Capellix.png

Edited by HardCodedCoder
Link to comment
Share on other sites

the GPU is toasty and heating up the whole case....

It seems despite the fan speed the heat has a hard time coming out. the 570X being all glass.. it can be tough to manage.

You can check you removed the dust filters on the fans exhausting, as they serve no purpose other than impede airflow.

Maybe you can also make a more agressive fan curve on the commander pro. On one side of the AIO you have ML pushing at 1700 rpm, and on the other side, the QL, with lower static pressure, are spinning at just over 1000 rpm, doing very little.

If you can match the speed somewhat, that could increase the flow of fresh air to the GPU, and ultimately slow down all the speeds as the case cools down.

The exhaust fans will also need to spin faster to get rid of waste heat. 

Where your case is situated may also play a lot in how cooling performs. If it's in some enclosed space, tucked between the TV shelf and the wall or something, it will recirculate ambient heat and your fans will constantly accelerate with time.

Link to comment
Share on other sites

Posted (edited)

Hey, thank you for your answer 🙂

I wanted to say, that I made a mistake in the original posts, i have 6 LL fans and not QL Fans. 

I can not really take out the dust filter as I have cats and there are a lot of hairs the dust filter keeps aways from the system. They are making a really good job 🙂

I attach a picture of the location of my pc 😄 I think the location should be ok, right?
My Question is: Which RPM should I set? I checked it out now, and tbh 1000 rpm are very silent 1300 already hearable and I would say 1500 maximum I would tolerate. But it doesn't matter anyway because the GPU is spinning with 2600 rpm.
Should I attach maybe all the Fans of the AIO 3x ML and 3x LL to the commander pro and the other 3 LL to the Commander Core?

WhatsApp Image 2022-06-10 at 14.32.36.jpeg

Edited by HardCodedCoder
Link to comment
Share on other sites

I Also would like to know: I looked up the specs of the fans:

LL: 600 - 1500 rpm
ML: 400 - 1600 rpm

but why is setting icue at 100 % for example the speed to 2300 etc? The technical sheets tells me 100 % should be 1500 / 1600 rpm?!

Link to comment
Share on other sites

for the dust filter i meant removing the top filter 🙂 you should definitely keep the front one!

the top fans are exhaust so cat hair or dust won't come in from there. And it will increase airflow by removing it.

Regarding the ML fans, there are several models. with two different speed ranges. the AIo ones are indeed 2300 rpm max.

For the speed you have to find a good balance. the more you reduce speed in the fan curve to keep silent, the hotter the GPU and case will be and the harder the GPU fans will work.

If you increase the fan curve speeds, you'll reduce case temperature and their speed may stabilize at a more tolerable level with the GPU fans running a bit slower.

If i were you i'd increase the fan speed quite a bit to see what temperatures the case can achieve. Removing the top exhaust filter should help a lot, since LL fans don't really like obstructions.

 

And damn cat tail is blocking the air gap between glass panels.. he knows how to get his tummy warm lol

  • Like 1
Link to comment
Share on other sites

For my Case Fans I created now a curve like this: 

image.thumb.png.6639e08fbae198f787fa80bdf9a56a1b.png

This means: (Temp based on Cooleant) 

  • Lowest Value: 900 at 36 C°
  • Middle Value: 1300 at 42 C°
  • Highest Value: 1500 at 45 C°

I know that the LL fans can spin higher (some sources say up to 2000), should i add another point at higher cooleant temps, or is this good like this? Dont forget, using a curve like this will make my DRAM go to round about 75 C°. Still fine?

For the ML Fans i have now the exact same curve only calculated with a % and would look like this: (good like this?)

image.thumb.png.232744caedd1370f7332d5eb46fd3c14.png

Link to comment
Share on other sites

Noise is subjective, but the cooling is not.  If you bump the fan speed to 1500 rpm and it sounds loud, then it had better knock a decent chunk of temperature off a key component.  Otherwise it is likely a bad trade.  Most users are better off tuning for noise tolerance, unless there is a piece of hardware reaching critical or performance degrading temperatures.  I don't see anything quite there.  For most people 1300 rpm on 120mm fans is around the tipping point between tolerable and mildly annoying.  Go with what sounds right.  +-100 rpm has little effect on cooling.  It requires large jumps in speed to make meaningful changes, except at the very bottom of the speed scale.  I think your curves are fine, neither too low or too aggressive.  

 

I think LeDoyen covered most of the critical points.  #1 - get off the presets.  That's done.  Everyone has a unique environment mixed with varied hardware and case designs.  There is no universal preset that can cover all those possibilities, including cats. 

 

Your radiator fans from the AIO kit are "ML-Elite OEM".  Different fan model from the prior ML-PRO RGB that was meant for case fan duty.  Also matters if you hand tune your RGB settings.  The ML-Elite OEM is an "8 LED fan" whereas the ML-Pro RGB is 4 center LEDs.  I think you have the curves sorted now.  Keep in mind +1C to coolant temp (H150i Temp) is +1C to CPU temp and the same for cooling at -1C to -1C.  That means you don't need to fight for 2C of coolant temperature.  It's only 2C at the CPU.  

 

The only temperature that is higher than average is the RAM, but I am not sure there is a lot you can do about it.  First, the Vengeance run a bit warmer to begin with.  You can observe this the next time you cold boot and do desktop work for a while or even simply leave the PC powered on.  In a 23C room, the RAM is likely to start in the low 30s and warm up to just shy of 50C with no real heavy activity.  This is your standard differential to case ambient temp and it will hold during other activities.  Let's say it's +25C over case air temp.  So when you fire up a game with a combination of heavy GPU and moderate CPU load, the waste heat from the GPU and the front radiator are being directed at the RAM.  So now it's 40-45C in the case and the RAM is closing in on 70C.  There is no way to provide it with cooler air, aside from strapping a fan to the top of it and blowing.  That would defeat the purpose of having RGB RAM and may not help a lot anyway.  The additional heating of the RAM is a common consequence of running radiators as intake and/or standard GPU air cooling.  

 

I won't get into the cat wrangling, but one possible source of improvement is to add a house floor fan somewhere off the front or lateral side area.  Between the cat, wall, and cabinet, you are going to get some heat trapped in the area behind.  You probably can feel it when you get close.  A small house fan (when seasonally appropriate) aimed at the front of the case to displace air between case and cabinet or off to the right to blow the air out from behind may help with overall ambient temperature and that is a boon to all hardware in the box.  It does not need to be blasting.  The lowest speed is usually enough to keep the heat from building up around the case, which then become intake air and heats up the box even faster.  

Edited by c-attack
  • Thanks 1
Link to comment
Share on other sites

I also forgot : for the RAM, white lighting will heat up the sticks substantially. if you use that color, reduce the brightness a fair bit, and see how much it improves.

  • Thanks 1
Link to comment
Share on other sites

10 hours ago, c-attack said:

Noise is subjective, but the cooling is not.  If you bump the fan speed to 1500 rpm and it sounds loud, then it had better knock a decent chunk of temperature off a key component.  Otherwise it is likely a bad trade.  Most users are better off tuning for noise tolerance, unless there is a piece of hardware reaching critical or performance degrading temperatures.  I don't see anything quite there.  For most people 1300 rpm on 120mm fans is around the tipping point between tolerable and mildly annoying.  Go with what sounds right.  +-100 rpm has little effect on cooling.  It requires large jumps in speed to make meaningful changes, except at the very bottom of the speed scale.  I think your curves are fine, neither too low or too aggressive.  

 

I think LeDoyen covered most of the critical points.  #1 - get off the presets.  That's done.  Everyone has a unique environment mixed with varied hardware and case designs.  There is no universal preset that can cover all those possibilities, including cats. 

 

Your radiator fans from the AIO kit are "ML-Elite OEM".  Different fan model from the prior ML-PRO RGB that was meant for case fan duty.  Also matters if you hand tune your RGB settings.  The ML-Elite OEM is an "8 LED fan" whereas the ML-Pro RGB is 4 center LEDs.  I think you have the curves sorted now.  Keep in mind +1C to coolant temp (H150i Temp) is +1C to CPU temp and the same for cooling at -1C to -1C.  That means you don't need to fight for 2C of coolant temperature.  It's only 2C at the CPU.  

 

The only temperature that is higher than average is the RAM, but I am not sure there is a lot you can do about it.  First, the Vengeance run a bit warmer to begin with.  You can observe this the next time you cold boot and do desktop work for a while or even simply leave the PC powered on.  In a 23C room, the RAM is likely to start in the low 30s and warm up to just shy of 50C with no real heavy activity.  This is your standard differential to case ambient temp and it will hold during other activities.  Let's say it's +25C over case air temp.  So when you fire up a game with a combination of heavy GPU and moderate CPU load, the waste heat from the GPU and the front radiator are being directed at the RAM.  So now it's 40-45C in the case and the RAM is closing in on 70C.  There is not way to provide it with cooler air, aside from strapping a fan to the top of it and blowing.  That would defeat the purpose of having RGB RAM and may not help a lot anyway.  The additional heating of the RAM is a common consequence of running radiators as intake and/or standard GPU air cooling.  

 

I won't get into the cat wrangling, but one possible source of improvement is to add a house floor fan somewhere off the front or lateral side area.  Between the cat, wall, and cabinet, you are going to get some heat trapped in the area behind.  You probably can feel it when you get close.  A small house fan (when seasonally appropriate) aimed at the front of the case to displace air between case and cabinet or off to the right to blow the air out from behind may help with overall ambient temperature and that is a boon to all hardware in the box.  It does not need to be blasting.  The lowest speed is usually enough to keep the heat from building up around the case, which then become intake air and heats up the box even faster.  

Thank you for your detailed answer! This helped a lot!

one question concerning the RAM: now all four RAM Slots started glowing permanently red after experimenting a bit around with the colour and their temperatures. The hardware diagnosis tool of windows 11 tells me, that there is a hardware issue and I should contact the manufacturer. Did I damage now the ram with the heat? Could this be? Are 75 degrees to much for the DDR 4 Ram?

what should I do now? 
 

Link to comment
Share on other sites

75C is getting pretty close to the point where it is going to cause problems, but it's not like melting ice cream where the event creates an irreversible change.  Higher heat on DDR4 can cause memory instability, which would lead to crashes or freezes.  If that's not happening, then I would not be too concerned but I also would not move the voltage past the likely 1.35v you are running now.  You can't go for the 4400 MHz heavy overclock with a hot case.  

 

I think LeDoyen was suggesting you turn down the intensity on the RAM LEDs as they can contribute to the heat on some models.  First step would be to turn them off (static black 0/0/0 or remove the effect).  See if it makes a difference.  If it does, then you try your normal effect and either turn down the color intensity by lowering RGB values or another method is to put a shade on top.  CUE lighting has a top down hierarchy.  If you put a "custom static" black 0/0/0 layer above any other lighting effect, then reduce the opacity of that black static shade, it will let increasing amounts of the underlying effect(s) through.  This trick can be used on Corsair LED device to precisely adjust brightness.  The RAM will also have a 3 position brightness slider in its settings, but I find the "shade" to be a more useful tool, easily created or deleted, and also profile specific.  The brightness slider will apply universally across all profiles and states, so that can be a quick one touch option if the 2/3 or 1/3 setting are pleasing.  

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

Posted (edited)
7 minutes ago, c-attack said:

75C is getting pretty close to the point where it is going to cause problems, but it's not like melting ice cream where the event causes damage.  However, higher heat on DDR4 can cause memory instability which would lead to crashes or freezes.  If that's not happening, then I would not be too concerned but I also would not move the voltage past the likely 1.35v you are running now.  You can't go for the 4400 MHz heavy overclock with a hot case.  

 

I think LeDoyen was suggesting you turn down the intensity on the RAM LEDs as they can contribute to the heat on some models.  First step would be to turn them off (static black 0/0/0 or remove the effect).  See if it makes a difference.  If it does, then you try your normal effect and either turn down the color intensity by lowering RGB values or another method is to put a shade on top.  CUE lighting has a top down hierarchy.  If you put a "custom static" black 0/0/0 layer above any other lighting effect, then reduce the opacity of that black static shade, it will let increasing amounts of the underlying effect(s) through.  This trick can be used on Corsair LED device to precisely adjust brightness.  The RAM will also have a 3 position brightness slider in its settings, but I find the "shade" to be a more useful tool, easily created or deleted, and also profile specific.  The brightness slider will apply universally across all profiles and states, so that can be a quick one touch option if the 2/3 or 1/3 setting are pleasing.  

Yeah, because of his comment I tried out the temperatures with and without lighting. And in idle it’s round about 5-6 degrees less. 
when I tried than to start rdr2 I got a blue screen with windows error code: MEMORY_MANAGEMENT (happened twice) and the rgb of the RAM turned static red (all of them). I ran then the memory analysis tool and it says there is a hardware issue and I should contact the manufacturer. 
after the analysis and booting up again the RAM had +60 Degrees. 😮 

i don’t know what to do now tbh. Is the ram defect?

 

For sake of completeness: I’m simply using XMP 2.0 and that’s it. 3200 MHz CL16 without manually doing a thing

Edited by HardCodedCoder
Link to comment
Share on other sites

4 minutes ago, HardCodedCoder said:

I got a blue screen with windows error code: MEMORY_MANAGEMENT

Well, there is a crash.  So it happened when starting a new game from a "cool idle" state?  Not after playing something else and a hot case?

 

Can you elaborate on the glowing red modules?  Meaning you've set it to rainbow wave (or whatever) and instead it's glowing red and you did not set Temperature lighting effects?

Link to comment
Share on other sites

13 minutes ago, c-attack said:

Well, there is a crash.  So it happened when starting a new game from a "cool idle" state?  Not after playing something else and a hot case?

 

Can you elaborate on the glowing red modules?  Meaning you've set it to rainbow wave (or whatever) and instead it's glowing red and you did not set Temperature lighting effects?

Exactly from idle I started rdr2 and instantly the crash comes. 
 

when they were glowing red I could only make them in the ram management module of icue during the selection of an effect make them change the colour, but by switching out of that module it immediately went back red. 
 

after the windows ram management analysis they start up now normally glowing white again I needed to leave for dinner with my parents, so I didn’t have the time to start up again rdr2. But what i quickly saw in icue, is that the ram was in idle at temps of 60ish degrees. And I thought: why??? Maybe because of the ram test? 

Link to comment
Share on other sites

11 hours ago, HardCodedCoder said:

after the windows ram management analysis they start up now normally glowing white again

Did you run the Windows Memory Diagnostic?  I suspect when it was running 1) it severs the connection with software as part of the test parameters; and 2) It effectively running a stress test, so I am not surprised the modules were 60C after it was done.  For Corsair gear, red is sort of the default RGB value when things go wrong, hardware level reset, etc.  So I don't think there is anything critically wrong, but obviously any crash is problematic.  In this case is does not seem possible it was temperature triggered or related to anything we talked about above.  I would treat it like a one off event, until it's not.  I have not played RDR2 in a long while and don't know the current status.  Some games are very sensitive to specific GPU or RAM instabilities that other programs run over like a road apple.  

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

On 6/12/2022 at 2:25 AM, c-attack said:

Did you run the Windows Memory Diagnostic?  I suspect when it was running 1) it severs the connection with software as part of the test parameters; and 2) It effectively running a stress test, so I am not surprised the modules were 60C after it was done.  For Corsair gear, red is sort of the default RGB value when things go wrong, hardware level reset, etc.  So I don't think there is anything critically wrong, but obviously any crash is problematic.  In this case is does not seem possible it was temperature triggered or related to anything we talked about above.  I would treat it like a one off event, until it's not.  I have not played RDR2 in a long while and don't know the current status.  Some games are very sensitive to specific GPU or RAM instabilities that other programs run over like a road apple.  

Thank you again for the valuable information you provide 🙂

I didn't have any problem since then, but I also keep looking at the temperatures more often 😉

i found now a setting for the fans on which i'm pretty happy with despite a single issue im currently facing where I would like to get an advice of you guys: 

As Sensor for my fans I took now the GPU Temperature as input, because I had the following issue with the cooleant temp: 

On Several days I had different base temperatures in idle:

Day One:

Idle - 36 °

Gaming - 42 °

So I made A curve with 3 points based on that: 36°, 42°, 47° --> As suggested by C-Attack in other threads i calculated than +5° to get an audible noise with full speed rotation when something goes wrong (here 47°) 

Day Two: 

Idle - 32° 

Gaming - 37 - 39° 

Fullspeed: 43 °

As this seems to change dependant also of the heat in my living room because of the temperature outside I was really desperate because I didn't know what to, I mean: I can not change the curve each day, this is really tideous. 

Therefore I thought I should make it maybe simpler and there I thought then: There are two scenarios mainly: Idle with some browsing and youtube watching etc and gaming. Therefore I thought: GPU Temperature would then make sense. In idle state the GPU has 54°. When Gaming in 4K something like RDR2 i have constantly 76°. So I made a curve like:

<= 60° - 1000 rpm

from 60° to 70° - 1000 rpm to 2000 rpm

>=70° - 2000 rpm

But now I found out that this works not for all games I play. Because there are games which use only 40% GPU Load where you also have waiting time like in a lobby where the GPU doesn't do a damn thing. And in this case of course, the fans ramp up and slow down then constantly. This is the case in a game like Dead by Daylight where also in the Lobby and Menu the GPU temperature changes from 60 - 70° the whole time. 

I don't know what to do to be honest. I simply would like to come to a decent solution which does not damage my hardware and where I dont need to swap from game to game the settings.

What would you suggest in my case now? Which sensor should i take in ICue? Which temperature values? 

And another question: Do you think it would be good maybe to exchange the case, to something like the LianLi O11 Dynamic.? Would this increase also airflow?

 

Link to comment
Share on other sites

Yeah, this is a general problem with fluctuating ambient temps and anyone with a heavy shift during the day will have to compromise. My home office is on the sunny side of glass wall and in Summer it will go from 19C in the morning to 28C in the evening with the AC blasting. Room temp is additive to all temps, so that’s a +9 shift across the board. All you can do there is program for the upper part of curve. Noise tolerance is still the main focus, so use that as your guide. If your under temp and the fans are quiet, you won’t be complaining. It’s too loud that’s disturbing. 
 

Another option is to use fixed speed fan curves. This can be done at the fan curve level or as a specific profile you can toggle in/out of when you’re ready to game. You set a quiet 750 rpm or whatever for the desktop (or a relaxed curve). Then create fan profile B as a custom curve and set it to the highest rpm you are willing to tolerate. You switch to it prior to gaming and back when you’re done. This effectively gives you the maximum allowable fan speed during the session and then an immediate reduction at the end. Nothing in the case is particularly reactive to small changes in fan speed. Dropping 100 rpm might cost you 1C at the most, if anything, so still let noise be your guide as much as you can. 

Link to comment
Share on other sites

45 minutes ago, c-attack said:

Yeah, this is a general problem with fluctuating ambient temps and anyone with a heavy shift during the day will have to compromise. My home office is on the sunny side of glass wall and in Summer it will go from 19C in the morning to 28C in the evening with the AC blasting. Room temp is additive to all temps, so that’s a +9 shift across the board. All you can do there is program for the upper part of curve. Noise tolerance is still the main focus, so use that as your guide. If your under temp and the fans are quiet, you won’t be complaining. It’s too loud that’s disturbing. 
 

Another option is to use fixed speed fan curves. This can be done at the fan curve level or as a specific profile you can toggle in/out of when you’re ready to game. You set a quiet 750 rpm or whatever for the desktop (or a relaxed curve). Then create fan profile B as a custom curve and set it to the highest rpm you are willing to tolerate. You switch to it prior to gaming and back when you’re done. This effectively gives you the maximum allowable fan speed during the session and then an immediate reduction at the end. Nothing in the case is particularly reactive to small changes in fan speed. Dropping 100 rpm might cost you 1C at the most, if anything, so still let noise be your guide as much as you can. 

but which sensor should I take here? And which temps are tolerable for the hardware? Where is now my startpoint when it differs from day to day?

Link to comment
Share on other sites

Your CPU will throttle around 100°C. The RTX start reducing speeds above 83°C core temp. GDDR6x on the GPU is 110°C if i remember well. That's like the temps you don't want to linger around, but the hardware will throtte to protect itself.

The farther you are from these figures, the better.

For RAM and M.2 SSD, it's usually good to stay well below 70°C. Those two will often be heated up by the GPU, so if you keep the GPU in check, the whole case will be fine.

Link to comment
Share on other sites

Looks like his exhaust fan is intaking making it hard to cool internal temps. I think the cat lying on the exhausting fans is probably not helping.

Edited by ntzphyr
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...