Jump to content
Corsair Community

H100i Platinum RGB fan curve not 1:1?


Recommended Posts

A fun little observation here.


I noticed that if I set a fan speed value of ~10% on a curve, the fan speeds on my H100i Platinum RGB drop to zero (and actually do stop). Only until the temperature creeps up to around halfway between the current and next control point do the fans begin to spool up.




Once the fans do start up, their indicated RPMs are lower than the unit's theoretical lowest possible RPM (according to the official spec: 400-2400rpm). Their speed also doesn't correlate to what it should be as a percentage of their maximum, ~2400 RPM (in reality, just over 2500).




Setting fixed 100% RPM in iCue makes both fans run steadily at 2550 RPM ±10 RPM.


Does iCue internally reinterpret custom fan RPM % curves or apply a nonlinear multiplier?

Or are the ML120 fans not accurate below 200 RPM?

Or is it the cooler which isn't capable of accurately controlling the fans by smaller increments below a couple hundred RPM?

Or does iCue have a baked-in curve which overrides very low percentage values for design reasons?


(...hello c-attack!)

Edited by Synergist
Link to comment
Share on other sites

We've been talking about this in a different context over the last several days, but short answer is PWM % x Max RPM of a fan does not always give you the actual resulting RPM. PWM % results often are skewed at the extreme ends of a curve. In my experience the low end can be very flat (your fans shouldn't run at 200 rpm) and the top end can be very steep (jumpy changes in the 90+% range). Most fans have a minimum PWM % in the 20-25% range. I think the ML are more linear in behavior than most fans, but below the 20% line is going to be erratic for most fans.


The normal +- for a fan RPM is 10%. At the 2400 rpm limit, that's a hefty 240 rpm and the end result will likely be both a function of fan motor characteristics and actual fan resistance.


Not really sure how the Platinum specifically manages the very low loads. On most controllers, you get a burst when starting from a stop state. This is one of the reasons I recommend using very low fan speeds vs zero rpm states. At 200-300 rpm, you are not moving more than a whisper of air through the radiator, but at least it won't pulse on when it needs to go up. My best guess is you are dropping below the start point and the controller needs some minimal PWM % to get the fan moving.


Not sure if this is just curiosity or if there is a goal. Try bumping the low end to flat 15% and see if that is more stable. If not, try 20%. That should work.

Link to comment
Share on other sites

It was curiosity :D


I also prefer lower states. Following my earlier thread (in which you were most informative) I've been experimenting with fan speed & noise versus cooling performance, seeing how gentle I can make the fan curve without getting the coolant too toasty. Now I'm sat at this PC for 10+ hours each day, having good gaming temps versus fan noise is making me more pragmatic ;-).


Mostly working out optimum speed-temp curves for work and play.


I did wonder what margins there were in the actual achievable RPM amounts. It's just something I never thought about really investigating until now, plus I shamefully didn't realise until recently that this great community exists where I can ask the questions and get decent answers!


These ML fans are also new to me, so interesting to learn more about their behaviour. I also noticed their burst idle before settling down.


Too much time to kill earlier made me wander into thoughts about current demands for multiple fans coming off one 12V or 5V, how many I could in theory run in a start-stop profile without causing rail sag, and how much current they might briefly draw when they spin up.


I also wondered if driving the fans high, 90% or above for a prolonged period, is causing any lifespan issues, or whether ML bearing life is basically unaffected (I'd hope so...)

Link to comment
Share on other sites

Fan lifespan - Generally manufacturers spec this stuff out in terms of hours of use vs a specific number of revolutions. I suspect they have that data too and it's hard to make an argument that if Fan A does 10 times the number of revolutions as Fan B, it doesn't have a greater change of premature bearing failure. However, in the time since the ML launched, I can only remember one single instance of a bearing failure. People aren't wearing them out. Too many other factors are more likely to end the life of the fan first, whether its RGB or user preference for something else. So to that end, you run them how you want to run them. What if it does wear out at 4.5 years vs 5? Would the six months longer lifespan be worth 5 years of micro-managing it to squeeze of 6 more months? Not for me, but that is an individual decision.


It would be hard to drain the 12v line on most PSUs for fan motors alone. A big, hulking 38mm thick industrial fan is still only 6-10W per fan. All our ML and QL and XL and whatever are usually around 4W at 100% fan speed. It just doesn't rate when compared to GPU, CPU loads. However, the 5v line is something quite different. Most PSUs will only give you 20-25A to work with. A white LL or QL is near 0.6A. So lets take a Lian Li O11 XL with 12 QL fans and you are up to 7A+ before accounting for anything else. I know some of the users with really high fan counts in massive 1000D cases do need to monitor this. Even for normal users, my 5v rail sat at 5.00 for years. I had no idea what it was for.:) Now if I go instant white globally I will most certainly see a drop. Or at least I did. I split some of my hubs onto different mainline PSU rails and put my AX1200i into multi-rail mode this week. Now my 5v rail is really steady. I just repeated the max whiteout and it stayed at 5.00v(!). I am actually really excited. It previously would drop to 4.7xv.

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


  • Create New...