Solved PSA to anyone using an AIO CPU cooler

spvtnik1

Community Contributor
Jan 13, 2020
186
415
1,970
Visit site
OK so I was having a little issue here, and I think I resolved it. I'm surprised I can't recall any of the big heads (Jayztwocents, Gamers Nexus, et al) mentioning this, and it seems really important. So I'll review here.

Symptom: Corsair H100i LEDs go to "hardware" mode, not communicating with iCUE. USB disconnect chime sounds on PC.

What's happening?
AIO coolers are designed to run at a constant 12 volts from their USB header. At some point, I went in to the BIOS and adjusted PWM (fan curve) settings for the pump. This was causing the pump to receive less than the required 12 volts, resulting in the connectivity issues. Apparently, running at fluctuating voltages can seriously damage the pump "anything less will degrade their performance overtime" as quoted from Corsair's sticky on the issue.

The solution:
I read several things indicating that I needed to disable the fan curve and set full voltage for the pump. In my particular BIOS, this wasn't an option. I did, however, have an option to set all fans to full speed. I did this, then set the fan curves on the regular case fans.

Please note that these curves are also completely separate from the profiles for the pump and radiator fans in iCUE. I believe the intended setup is to set the pump to full voltage, and then allow iCUE to adjust the pump speed based on monitoring variables.
 
Last edited:
  • Like
Reactions: ElangRizki and Faiq
So solution is don't adjust pump settings for the AIO in bios if you also use Icue.

I believe the intended setup is to set the pump to full voltage, and then allow iCUE to adjust the pump speed based on monitoring variables.
That makes sense, as otherwise you get conflicts between them, Icue expecting 1 thing and getting another. The pump speed is based on what performance profile you run it at - those being Quiet, Balanced or Extreme. I normally have my H100i on Balanced for pump, I think that was what it was set on when I got it.

Never had problems with rgb on fans doing that, well, maybe a few times but a restart normally fixes them. My ram would go into hardware mode occasionally but I have set the same colors for both logged in and on the hardware now so I may not notice.
 
  • Like
Reactions: Faiq and spvtnik1

spvtnik1

Community Contributor
Jan 13, 2020
186
415
1,970
Visit site
So solution is don't adjust pump settings for the AIO in bios if you also use Icue.

Yes, I would say the solution is to not mess with the fan curve settings for your pump in your BIOS, if you purchased a pre-built. On the other hand, it might be worth checking to see that was configured properly by the integrator.

Those building/installing their own AIOs are going to have to set it full speed at some point, unless it happens to default to that.

My RAM had gone in to hardware mode too, previously. But it was only one time I can remember instead of frequently like the pump was. So I think that had more to do with iCue, since restarting the application resolved that, iirc.
 
  • Like
Reactions: Faiq
FYI, if you dug through Tom'sHardware you'll find that the voltage going below or above 12v is detrimental to the pump. You've mentioned what happens when the voltage is low, when it's high, the pump starts to give off a buzzing sound. The solution to that was to introduce a resistor in order to maintain 12v.

Secondary FYI, by high, even a .1 or .2 volts is considered high for the pump. If you're yet within the warranty period, I'd contact Corsair and RMA it.
 
  • Like
Reactions: spvtnik1 and Faiq

spvtnik1

Community Contributor
Jan 13, 2020
186
415
1,970
Visit site
I appreciate that bit of info Lutfij. I can hear the pump when it's... well, pumping. But I wouldn't call it a buzzing sound. And, despite the facts, I don't think I have much of a case for RMA. The pump is functioning as intended.

If the lifespan on these things is only 5-6 years (because of permeation), then my plan is to probably replace it with a heavy duty heatsink and fan later. I don't know, would the .1 or .2 volts even really do anything within that time span? I feel like the worst that would happen there is the pump motor eventually burns out. It will, anyway, but I'm guessing permeation would happen before that.

And besides, it's called the 12+ rail, right? So I'd feel like the engineers are accounting for at least a little bit of extra voltage. I would even go so far as to say, a well engineered pump wouldn't be as sensitive to over-voltage as a poorly engineered one. If it is constantly pulling an extra .2 on the 12+ rail, I'd actually be more worried about the PSU and my mobo than the pump. ¯\_(ツ)_/¯
 
I still have an H50 in working order on a build that still has the i5-750 on an Asus Maximus III Formula, so that's well beyond 5 years. Knowing Corsair and Razer gear, they work well and when they fail they fail spectacularly, so I always take their things with a tablespoon of salt.

If the pump fails, how does the fluid circulate inside the loop(to take the heat from heat source to the radiator)?

Corners need to be cut to make profit, then again, when you factor in that they account for units failing and keep an overhead of units you're advised to contact Corsair for an RMA, provided you're within the warranty period. That faulty unit will also be entered into their statistics which in turn would help the community with a better product. It's been seen in the past that a unit that was faulty beyond a threshold was recalled or a V2 issued as a placeholder.
 
  • Like
Reactions: spvtnik1

spvtnik1

Community Contributor
Jan 13, 2020
186
415
1,970
Visit site
I take all computer equipment with a grain of salt. It's amounted to cups' worth. I hear what you're saying, but there isn't any more indication that the unit is faulty, at least as far as I can tell now. I just installed hwinfo64 and checked my +12V reading on my motherboard. Averaging about 12.14 with a .02 range in flux. I'm not really getting much fluctuation in voltage at idle. Things seem pretty dandy. Obviously a test under load is needed. But I haven't had the pump LEDS disconnect from iCUE since I adjusted the pump PWM settings.

I can also see the pump is operating at a solid 2,350 RPM here on the dekstop while I write this.

I found this on the EVGA website:

The 12v rail should read between 11.8v and 12.4v . It should be stable, with little to no fluctuations. A slight fluctuation of up to .05v is generally fine as long as it happens no more than once every 10-15 seconds. A larger fluctuation, up to .1v can be safe as long as it happens no more than once every 1-2 minutes. A fluctuation higher than .1v, or more often than every 10 seconds could be an indication of an issue with the 12v rail, and you should consider having your power supply tested or replaced.

These readings are taken at an idle state, and may not indicate a more severe issue that occurs when the system is under load. Testing under load with a multimeter is the most accurate way to test a power supply for issues.

I think if I had ignored the issue and let it go for a couple more weeks (or maybe even days), then yeah, I woulda had a problem.

If the pump fails, it fails. If it fails within warranty, I send the PC in. If it fails out of warranty, I'm hopeful thermal protection routine does it's job and I replace the pump.

I don't even actually know if the pump is getting a "high" amount of voltage.
 

TRENDING THREADS