Build Advice I've had a Corsair 150i AIO in my i9-10850k build for about a year. It ALWAYS ran hot. I just fixed it 2 days ago and wanted to share how...

[JCole$$$Tech]

Hi All,
Long time lurker, 1st time Poster, so here it goes.....
The Thread title explains the problem pretty well, but to elaborate: My system has a Z490Msi Mobo and I have the Corsair AIO H150i RGB PRO XT pump securely fastened to the i9-10850k CPU and the 3 Corsair fans that came with it plugged into the three ports on the cable coming from the pump and the pump plugged in to the CPU_FAN header. All the fans worked, they were 4 pin fans into 4 pin connectors off the pump, but seemed to just run at a constant speed.
After powering the PC on within minutes the temperature went to 80+ degrees, and it watching full screen 4k video, not even gaming the temperature would push 100 and HWInfo64 reported some throttling due to the heat. iCue Software, MSI Center and BIOS settings didn't have any settings that provided an adequate solution. I had one chassis fan secured to the inside roof of the case that started making noise on start up, so it was time to replace it. I knew the 10850 had a reputation for running hot, but not like this. As I was doing so and straightening up the cable management, I had the realization that one thing I hadn't tried yet was plugging the Corsair AIO fans straight into the motherboard instead of the cable off of the pump and attached them to headers SYS_FAN1, SYS_FAN2, and SYS_FAN3.
There was a night and day difference as soon as I powered it up and went into the bios. Not only did they spin faster and louder, but they also responded to the BIOS Smart Fan Control settings when before they seemed stuck in DC mode instead of PWM. Now the happy ever after part: Now, the fans stay low, around 800 rpms until the temperature rises enough to call for higher speeds and the CPU fan reading,(for the pump) isn't pinned at 100% now (again it seemed like it was stuck in DC mode for some reason, and the CPU Core temp stays around 42-45 degrees under idle or ight work conditions and doesn't even get to mid-70's under a decent load. It's crazy how much more responsive it is now and the CPU_Fan reading hovers at around 33-35% instead of being stuck at 100%, which I imagine couldn't have been good for the lifespan of the pump. So now I just have my fingers crossed that the throttling mechanism of the CPU protected me from myself and the inadequate cooling it was receiving for little over a year and that the pump will still have a decent lifespan after being run at 100% for that time.
The problem had bugged me for that whole time as I knew the CPU shouldn't run that hot but just couldn't figure out why. I saw a post on here describing the same situation, same brand and realized this may be a thing. Then when I saw how many Corsair AIO overheating threads there are I knew I should detail my experience of it here in hopes that it'll save some others the time, mental energy and possibly their hardware that running a high end pc with inadequate cooling can bring.
I'm theorizing that it's an issue with the amps put out by the pump as that would determine fan speed and obviously the fans weren't getting enough connected to the pump, but do when connected to the mobo. I don't know if it's an equipment flaw, or inadequate instructions but it definitely runs A LOT better connected the way it is now over the way described in the manual.

Hope this helps someone that has been stumped by their overheating and that it helps you to utilize your build's true potential.

Live long and build well.

 

[Theresa N]

Actually, I believe it's commonly held that the pump should run at 100%.

 

[Paperdoc]

Many AIO systems are designed to have the pump run full speed all the time. Corsair, however, added another option using iCUE. In that tool you CAN set the pump speed to less than full but fixed. The way to decide how to set this is to check how your rad fans are running to keep the CPU temperature in control. Ideally those fans should almost never run full speed (so that even at max workload and heat the CPU CAN be cooled properly), but you do not need them near full speed all the time. OP does not seem to have noted this. I do not know how iCUE is set up by default, but it MIGHT have been set to less than full speed.

Since OP refers to trying to use iCUE I must assume he has installed the required cable from the pump to a mobo USB2 header - otherwise iCUE would show errors of no system to deal with. So, assuming that, iCUE SHOULD be able to manage the three rad fans' speeds according to CPU temperature, and use the FULL range of fan speed for that

I am puzzled why OP reports that those fans connected to SYS_FAN headers are now running much faster. Of course, the SYS_FAN headers are using a DIFFERENT temperature sensor on the mobo, NOT the one inside the CPU chip. If the case temperatures are high requiring fast case ventilation fan speeds, that might explain why this speed change. And it raises the question of whether or not the actual CPU temperatures were high enough to require faster rad fan speeds.

 

[Phaaze88]

Many boards have 1Amp CPU_FAN and SYS_FAN headers. Fewer boards have 2A versions of those.

One of the H150's included ML120s is about 0.30A max. Three = 0.90A.
The pump is around what, 0.50A..? Either way, the CPU_FAN header on the unspecified Msi board was likely 1A, so it didn't have enough room for 3 fans and a pump.

 

[Paperdoc]

The way that Corair AIO system is SUPPOSED to be installed (I don't thnk it will work otherwise) is it MUST have its wide connector on a cable end plugged into a SATA power output from the PSU. ALL power for that system is drawn from the PSU directly, and the ONLY function of the cable plugged into the CPU_FAN header is to report the PUMP speed of that header. Effectively there is NO power limit to the pump and rad fans.

 

[Phaaze88]

The way that Corair AIO system is SUPPOSED to be installed (I don't thnk it will work otherwise) is it MUST have its wide connector on a cable end plugged into a SATA power output from the PSU. ALL power for that system is drawn from the PSU directly, and the ONLY function of the cable plugged into the CPU_FAN header is to report the PUMP speed of that header. Effectively there is NO power limit to the pump and rad fans.

So THAT's the 'tradeoff' of the mandatory iCUE software? Other AIOs can be powered by either SATA or mobo headers... no, NZXT's Kraken would be just like Corsair.

 

[JCole$$$Tech]

So THAT's the 'tradeoff' of the mandatory iCUE software? Other AIOs can be powered by either SATA or mobo headers... no, NZXT's Kraken would be just like Corsair.

Actually, I believe it's commonly held that the pump should run at 100%.

Yeah, I've read that as well. Thank you for that. The good thing is that seems to be reacting to the BIOS setting for it now, and switching it to DC has the CPU FAN measurement at 100%. Probably going to set it back, since the sound is negligible over any fans but I haven't seen any heat issues so far at 5 days in, PC turned off at night but running all day.

 

[JCole$$$Tech]

Many AIO systems are designed to have the pump run full speed all the time. Corsair, however, added another option using iCUE. In that tool you CAN set the pump speed to less than full but fixed. The way to decide how to set this is to check how your rad fans are running to keep the CPU temperature in control. Ideally those fans should almost never run full speed (so that even at max workload and heat the CPU CAN be cooled properly), but you do not need them near full speed all the time. OP does not seem to have noted this. I do not know how iCUE is set up by default, but it MIGHT have been set to less than full speed.

Since OP refers to trying to use iCUE I must assume he has installed the required cable from the pump to a mobo USB2 header - otherwise iCUE would show errors of no system to deal with. So, assuming that, iCUE SHOULD be able to manage the three rad fans' speeds according to CPU temperature, and use the FULL range of fan speed for that

I am puzzled why OP reports that those fans connected to SYS_FAN headers are now running much faster. Of course, the SYS_FAN headers are using a DIFFERENT temperature sensor on the mobo, NOT the one inside the CPU chip. If the case temperatures are high requiring fast case ventilation fan speeds, that might explain why this speed change. And it raises the question of whether or not the actual CPU temperatures were high enough to require faster rad fan speeds.

Right. Sorry about the missing info, my bad. First time posting here.

The board is the MSI MPG Z490 Gaming Carbon Wifi. The SATA power cable AND USB2 connector are plugged in, I can control the RGB features through iCue just fine. iCue is updated, same as all MSI firmware. I'm not sure why the fans run faster now either. I don't think it's responding as it was designed or I got something wrong somewhere BUT the main point of focus,for me anyway, is that for the first time in this build's run (as I wanted to just do it by the book, or manual's instructions), the temps don't get near hot enough to throttle (as reported in HWinfo64). Before with the rad fans plugged into the pump, the pump was steady at 100% and the fans steady about 1200 rpm and did not increase or decrease in speed, BUT the temps were not healthy. I'm pretty sure the problems were with the CPU TEMPS as that's where the highest temp reading was coming from and it was always significantly higher than the SYSTEM temps like it was a hot spot. HWinfo64 would report that the Temps hit 100 and that throttling was engaged to compensate for the higher temperature.

I just wanted to share how this particular configuration affected my build's performance. Any insights you may have to better understand it is appreciated.

Did I miss anything there? Asking genuinely.

 

[Paperdoc]

Just an FYI on pump power and speed with AIO systems. As we said, MOST design to have the pump run full speed all the time, but a few offer other options. (The Corsair system is one of these latter, and it does that by getting all power from the PSU and setting pump speed within iCUE.) There's a "trick" to how this is done, but you CAN cause a small problem if you set things wrong.

The design of older 3-pin and of newer 4-pin fan systems means that connectors and electrical signals allow you to plug either fan type into either header type. They perform as expected if you match fan and header type. If you mis-match by plugging a 4-pin fan into a header using the older Voltage Control Mode (aka DC Mode), the fan's speed WILL be controlled that way, even if not by an ideal signal set. If you plug a 3-pin fan into a header using the new PWM Mode, that fan will always run full speed. So, MANY AIO pumps are wired just like 3-pin fans (the connector will have only 3 holes), and they run full speed all the time if plugged into most mobo 4-pin CPU_FAN headers. BUT that works only if that header really IS using the newer 4-pin fan PWM Mode of signals.

There are two ways users can cause issues. One is to see the three-hole connector and decide that the CPU_FAN header should be manually set to Voltage Control Mode (aka DC Mode) since the connector looks like the one for a 3-pin fan. The other is to leave the CPU_FAN header set to "Automatic" configuration of fan type. In this setting at every start-up the header TESTS the fan by sending out signals to run full speed, then a PWM signal to slow down. If the fan does slow down, then it must be a PWM fan and that is left alone. But if the fan fails to slow down, then it must be an older DC (3-pin) fan, and the header changes to sending out DC Mode signals so it CAN force the "fan" to slow down. BUT for an AIO PUMP, we do NOT want it to slow down. We WANT that header to use only PWM Mode, so we should set it that way and NOT allow the "automatic" detection of "fan" type.

 

[JCole$$$Tech]

Thank you for that. Very informative. I know the fan cables on the Corsair fans are 4-pin, as are the cables coming off the pump, they're 4 pin as well. The SYS_FAN headers are all 4 pin but say (DC Mode) under them in the manual, the CPU_FAN header is 4 pin but the connector from the pump is a 3 pin connector with only a black grounding wire, I assume, so that the mobo has something plugged into it. Also, the CPU_FAN has (Auto) under it in the manual.

Some more information: I have a dual loop cooling system with 2 separate pumps. The AIO on the CPU and an EKG pump on a water cooled GPU, a 1080 Ti Aorus.

The EKG Pump is plugged into the PUMP_FAN1 4 pin header with a four pin cable and the PUMP_FAN1 has (PWM Mode) under it in the manual.

 

[Paperdoc]

Regarding the labels in the manual, those usually tell you the default setting as your mobo arrives from the factory. They do NOT tell you that those headers can do ONLY that Mode. So when you go into BIOS Setup, for each header you usually find that there are two or three choices available to you, and you can set as you wish. You confirm that your CPU_FAN header was set to Automatic (detection of fan type) by default, and I presume you never changed that. However, this should not have mattered in your case because power to the PUMP and its speed is controlled by iCUE. The CPU_FAN header only gets that pump's speed signal (and iCUE should show you the same speed). What MAY be happening for you is that the setting within iCUE for the pump speed is NOT full speed, but some fixed slower speed, so check for that.

MSI often sets their CPU_FAN header to PWM Mode or Auto by default and their SYS_FAN headers to DC Mode, but you can change those in BIOS Setup. On the other hand, most PUMP headers are set ONLY to send full power and max speed signals to their load since they are intended for pumps. SOME such headers are labelled PUMP / FAN because they also have an extra option switch in them you can set to FAN use, and that DOES allow you to have it control the speed of a fan plugged in there. That option usually is set by default to PUMP mode, meaning fixed full speed.