• 27.04.2024, 15:20
  • Registrieren
  • Anmelden
  • Sie sind nicht angemeldet.

 

Lieber Besucher, herzlich willkommen bei: Aqua Computer Forum. Falls dies Ihr erster Besuch auf dieser Seite ist, lesen Sie sich bitte die Hilfe durch. Dort wird Ihnen die Bedienung dieser Seite näher erläutert. Darüber hinaus sollten Sie sich registrieren, um alle Funktionen dieser Seite nutzen zu können. Benutzen Sie das Registrierungsformular, um sich zu registrieren oder informieren Sie sich ausführlich über den Registrierungsvorgang. Falls Sie sich bereits zu einem früheren Zeitpunkt registriert haben, können Sie sich hier anmelden.

OCTO : Different results between two temperature sensors

Mittwoch, 25. Oktober 2023, 14:14

Hello everyone
I have been using an OCTO with 2 temperature sensors for 2 years:

Description

Sensor 1 at the outlet of my EK pump
Sensor 2 at the entrance to my radiator
My loop is built like this: EK Pump >> Sensor 1 >> GPU Block >> Radiator 420 >> CPU Block >> Sensor 2 >> Radiator 420

Problem
According to the information provided to me by OCTO, sensor 2 is always one degree cooler than sensor 1 (from -0.8 to -1.2°).
This does not seem correct to me, because in my opinion, sensor 2 receives hot liquid from the Water Block CPU while sensor 1 receives the liquid cooled by the 420mm radiator :!:



My questions are:
What is the margin of error from one sensor to another and who to believe?
How much does a 420 radiator lower the temperature?
How can I check if those sensors are correctly calibrated?

Thanks for your help !

Remayz

Senior Member

Mittwoch, 25. Oktober 2023, 15:00

the short answer is sensors aren't too accurate. I mean they are not calibrated to anything. You'd have to connect them both at the same point in the loop and see how far off they are from each other, then correct on aquasuite with an offset on one of them.

That said it doesn't seem abnormal to me. It's mostly the GPU that heats up the water a lot in normal use. Because you have a rad in between the two blocks, you don't really know what temperature water the CPU block is getting. And your pump speed will make that difference vary a lot.
If your pump runs at full speed a 420mm rad may decrease the water temp by 0.5°C, if you slow it down maybe 3 or 4°C. It's very variable.

Mittwoch, 25. Oktober 2023, 17:30

I never concern myself with absolute accuracy, assuming the error to be fairly constant.
I am more interested in change.
My typical loop compares coolant temperature to ambient air temperature, increasing fan speed as that delta increases.

Donnerstag, 26. Oktober 2023, 01:36

The temp sensors in computer cooling loops are just cheap 10K NTC Thermistors which typically have an accuracy of +/- 5% in the temperature range encountered in a computer. Aquasuite allows you to add an offset to a temp sensor so you can calibrate them by sticking them in a cup of ice water and adjusting the offsets until they both read 0°C. Their accuracy is not linear over the entire range of temperatures they can measure but accuracy is basically the same over the range of temperatures they will measure in a computer. They are generally more accurate at lower temps (like below 50°C) because their resistance changes more per degree C at lower temps. As the temp goes up, the resistance changes less so its harder to measure accurately. Here is a resistance vs temp graph for a typical 10K NTC Thermistor.



I have 3 of the #53206 temp sensors plugged into a Farbwerk360 and coiled up in the back of my case. The temp they report varies a bit, normally a few tenths of a degree C. If you want higher accuracy, Aquacomputer still sells the Calitemp sensors but they report via Aquabus so they can only be used with an Aquaero. They also cost about $20 each. I don't know what type of sensor is in the Calitemp sensor. Thermistors can be Class A or Class B. Class A Thermistor accuracy is +/- 0.15°C @ 0°C, Class B Thermistor accuracy is +/-0.30°C @ 0°C, so the Class A Thermistor is twice as accurate as a Class B Thermistor at 0°C. 10K NTC thermistors are Class B, which are pretty accurate but not as accurate as a Class A Thermistor.