iCue fan presets question

[RequiemNoctem]

I'm running a H150I Elite + 3 x 140mm case fans.
Using the stock controller in the 7000d airflow case.

Questions #1:
All 6 fans are set to "balanced" but somehow fan 2,3,4 scale up harder when the CPU heats up, which is nice because they are the cooler fans.  The other case fans scale up slower. This is all very nice, but what black magic is this?  How do the CPU fans know to scale up faster? As they are all on the same preset, all 6 fans should use the same sensor and scale uniformly?

Question #2:
My first fan is the rear exhaust fan and is the only fan pushing air OUT of the system, (all other 5 fans pull fresh air IN). So I would like to throttle it a bit higher than the others (at least faster than the 2 front fans).  What's the right sensor to base my curve on? I'd like something that scales with the general internal case temperature and not necessarily with the CPU. I have access to 3 GPU sensors, and 7 MSI motherboard sensors (just labeled #1 through #7). Mobo is a MSI PRO Z690-A DDR4.

Cheers.

[c-attack]

17 hours ago, RequiemNoctem said:

This is all very nice, but what black magic is this?

That's about right.  The invisible presets tend to be confusing and none of your fans should be reacting to CPU temp at all.  My standard advice for everyone is to get off the presets and make your own curve.  You can replicate the AIO presets for water cooling directly from the shape tools in the custom curve area.

 

1) In the Cooling tab for the Commander Core, click + for a new fan profile.  A graph should appear below.

2) Go to the lower right corner and click on the third shape tool.  A small "balanced" label should appear to the left of that.

3) Change Sensor to "H150i Elite Temp" (coolant temperature).  This is critical and the graph is scaled for that value.  

4) Change all fans above to the new "custom 1" cooling profile (can be renamed).  

 

Now you can see and edit the control points and add some meaning to the behavior.  Radiator fans work to reduce the temperature of the coolant and that is their function, but case fans can use this too.  Case fans are regulating your internal case ambient temperature and the coolant temp is always tied to this as well.  Case temp goes up +5C?  So does coolant temp regardless of load.  This is a place to start and now that you can see and move data points, you can make adjustments as you go.  The baseline curve holds flat to around 31C, so that looks to be about right for you based on the screen shot above.  Ideally you figure out about where you top out during your normal use (most people about +10C) and then set an acceptable moderate speed for 40C.  Save the 2000 rpm blast for up near 50C and this will act as an alarm system if there is a problem.

 

As for the rear fan.... three options.  

1) Create a second curve and set the sensor to your GPU #1 temp.  Change the scale and speeds to match your normal GPU temp range.  However, be aware you will still get the dynamic change in fan speeds as the GPU loads/unloads, like on map or loading screens.  There is no need for that and it just draws your attention to it.  That rear fan is removing GPU waste heat in the air.  It's not really doing anything about cooling it off.  The GPU's own fans need to do that.  This also requires the CUE software to be active in order to fetch the data from the GPU.

2) Create a copy (or new) custom curve like above.  Same parameters using H150i Temp, but adjust the curve to make it more aggressive so you get increased fans speed in the 30-40C range.  Set to fan #1 only.  Works at all times, software active or not.  

3) Same as number 2, but get a 10K thermistor temp probe that connects to the Commander Core.  It's a long wire with a flat sensor end.  Route it out the back of the case and tape it so the exhaust air blows across it.  This is effectively exhaust air temp and the ideal control variable.  Change control variable to Commander Core Temp #1 (the temp probe).  It will work at all times, software running or not, just like when you use H150i Temp.  The sensor can be inside the case or outside, it makes no difference.  This is really the ideal setup, but the Com Core doesn't come with one so you can't just go set it up.  These are cheap (under $10), but maybe not the first thing you try.  #2 can be done immediately.  Also the effectiveness of #2 may depend on the radiator location.  Up top you will have a strong connection between GPU activity and coolant temp.  With the radiator as front intake you may not and then #3 becomes the better option.  

Edited February 16 by c-attack

[RequiemNoctem]

Yep, I already knew how to make curves but thank you for the exhaustive and informative answer.

I like the thermistor temp probe idea as it will really measure what I care about for that fan.  I'll give that a go.  Thanks much!

[c0rrupt]

I think you would have better temps if you had more then 1 fan running exhaust. You can slide the temp sensor inside your gpu (depending on card). You could always look into undervolting your card.I never can understand why people setup fan curves (not talking about gpu) and not just run them at 100%. You spend xxx to get the best fan possible why limit it:? Just my opinion 😛

[RequiemNoctem]

7 hours ago, c0rrupt said:

I never can understand why people setup fan curves (not talking about gpu) and not just run them at 100%. You spend xxx to get the best fan possible why limit it:? Just my opinion 😛

Oh, that one is easy, because those fans (SP140), running a 100% make a level of noise that make me want to throw my case out a window =D  Plus, running them at 100% is absolutely overkill.

To be clear, I don't have any issues with temperature right now.  Everything is good.  CPU is one of the coolest setup I've had in years, and GPU is well within safe parameters.  I'm just taking an interest in the exhaust fan a bit preemptively.  I'm going to stick 3 more NVMe in there, so general case temp is a thing I want to keep an eye on.

[LeDoyen]

Corsair fans are not known for their silent operation 😛

Running GPU fans at full blast all the time sure makes it cool, but it kills the fans bearings in no time. and they are not as readily available as spares as case fans are.

 

[c0rrupt]

How often/common do bearings fail in fans? Maybe i've just glaced over it whenever someone talks about it on any forum I on. I've had 3 Scythe Ultra Kaze 120mm Case Fan - 38mm Deep running 3000 rpm for 6 years with no issue. Fan curves are great for GPUs,along with undervolting (Red Devil RX480 user)

[LeDoyen]

i've had failed fans on my GTX270, GTX970, GTX 1080, and various case fans from Corsair, EK and Arctic. It's not as uncommon as you may think.

All those have failed before i used watercooling (except the EK fans) because they needed to run faster when aircooling.

For the GPUs the only solution was to get a complete cooler from a dead card on ebay for the first one, or to order from aliexpress because nothing was available in europe. The colleague i sold my 1080 to had to get one there too when one of the fans started rattling, and you better be sure you got the right fan 🙂

I suppose they all feature zero RPM modes now for a reason.

So yea, everytime you can have a fan slow down, i'd advise slowing it down.