iCUE temperatre sensor issues?

[Inoeie]

So I purchased an XSPC 20k temperature sensor for a watercooled PC. It is on the output of the reservoir so I would imagine that it should be accurate enough for the temp of the water it is in contact with.

 

The issue now is that the GPU temp reading is a few degrees higher than the water which is impossible... So, is there any way to calibrate or fix the sensor or is it just wrong/broken or is it the commander?

 

https://gyazo.com/304e46c51da30f5b1df8c854b8978b1f

[BlaiseP]

I use a Phobya inline sensor in my loop that works as expected BUT it is 10kΩ , not 20kΩ . The Commander Pro uses 10kΩ sensors NOT 20kΩ.

 

BTW. Once the loop temperature normalises, it doesn't really matter where you place the sensor within the loop so you can place it at the most convenient location. Mine is after the CPU just before the bottom 240mm rad.

[Inoeie]

I use a Phobya inline sensor in my loop that works as expected BUT it is 10kΩ , not 20kΩ . The Commander Pro uses 10kΩ sensors NOT 20kΩ.

 

BTW. Once the loop temperature normalises, it doesn't really matter where you place the sensor within the loop so you can place it at the most convenient location. Mine is after the CPU just before the bottom 240mm rad.

 

Ok so I miss remembered, it was indeed a 10K sensor. Do you know if yours is similar to mine then? Or is this just abnormal?

[DevBiker]

It could very easily be a calibration issue between sensors.

[BlaiseP]

My package (at idle) is currently 27°C and the water is 22°C.

[c-attack]

My package (at idle) is currently 27°C and the water is 22°C.

 

Those will always be different unless you manage to sit on your desktop at 0.00 Vcore.

 

As for the difference between in-line water sensors and the GPU, remember the GPU value is typically the diode, not the coolant itself. While it is simple to explain to others everything in the loop is the same temperature and the coolant is minimum possible temperature, the reality is there still is some thermal transference between things and that is not a 100% perfect energy transfer. At a desktop idle state, it is every day normal for my GPU to be about 2C below both of my 10K water sensors, neither of which is connected to my C-Pro. The most likely reason for the observed behavior is the exact location of the GPU diode sensor and with the GPU in an idle state and 0.0v, heat is not being fully conducted into the sensor. After all, the sensor's job is to measure GPU component temperature and not coolant temperature. Typically at load this evens out, although I would expect to see +-1C differences at times. That may be within the sensor calibration limits.

[BlaiseP]

Quoting an earlier post of mine...

I use a Commander Pro to control my radiator fans (mix of ML-120 & ML-140) on an custom EKWB loop (CPU & 2 x 1080 TI's, 420 & 240 Rads). Here's the settings I use on the for the 420mm radiator (using ML-140 fans) gratuitously stolen from my old H115i fan curves. I have included the tests I performed and as you can see, the differences between my "Balanced" and "Extreme" curves are identical within a margin of error. As a result, I didn't bother with the combined test (CPU/GPU's) for the "EKWB Extreme" fan curve.

 

Notes:

1. The custom fan curve points used by ICUE are expressed in RPM, not PWM duty cycle so I plugged the info into Excel and use a formula to convert the PWM percentage into RPM.
   
2. Water temperature was measured using an inline 10kΩ sensor in the loop.
   
3. Tests performed were Intel Extreme Tuning "Stress Test" set for 10 minutes for the CPU and Heaven Benchmark looping for 10 minutes for the GPU's to normalise temperatures.
   
4. The "combined test" is Intel Extreme Tuning and Heaven Benchmark being run simultaneously for 10 minutes.
   
5. This loop also runs a 240mm radiator which has 2 x ML-120 fans running the same duty cycle as the ML-140 fans.
   

 

No code has to be inserted here.

 

Hope this helps.