Cooling Temp Sensor on CPU Fan Header?

Hi,

I'm going liquid cooling very soon, my NF7-S' temp sensors have proven to be very unreliable, apparantly i'm at 70°C atm, it'd be unusable at the temp in my experience. Are there any sensors avaliable that could hook up to the yellow wire on the CPU Fan header that i could then configure in MBM5 into a correct temperature?

 

You're better off with a standalone.

Try an AeroGate or Cooldrive w/probes. Much more accurate.

 

I could but i'm on a bit of a budget, any ideas?

 

I'm not sure you can use a fan header for temperature readings (in fact, I'm almost certain you can't). Fan headers read pulses, so unless you have a way of converting temperature to pulse rate, and then calibrate the thing, you're out of luck on that one.

 

Ive never seen anything that would do this 'off the shelf' I suppose in theory you could use a temperature sensor (thermistor perhaps?) and convert the analogue signal from that into a frequency you could drive the fan header with. My first thought was a PIC, but look on Bit-tech for the article on using a 555 to generate a 'false' fan signal, when you're watercooling and you have a board that needs a fan connected to boot. Replace one of the resistors there with a thermistor, thatll vary the 555 output frequency depending on temperature Only thing would be calibrating it properly.

I have a standalone LCD temperature sensor module I use on mine, £5 from ebay and its bl00..

 

I did some research and I stand corrected. Go :here: for the SMT 160-30 temp sensor which comes in a basic transistor package. The sensors are powered with +5VDC. The output signal is a square wave from 1-4 kHz of which the duty cycle is a measure for the temperature of the sensor. A fan header can interpret that.

 

Wow great job Nexxo Will this be safe, no danger of killing the board? If so looks like i'll go for one of those jobbies. It's all powerable from the fan header right? There won't be any conflict between it and the CPU?

 

Basically you connect the transistor + and - to the fan header + and - respectively, and the signal out to the fan header RPM detection line (yellow line). You'll have to put a resistor between the fan header + and the SMT 160-30, as it takes 5V and the fan header, of course, spits out 12V. The current consumption is 200µA (i.e. 0.2 mA) and fan headers should be able to handle up to 0.5 A max. so I can't see any problem there. The resistor value you're looking for to take voltage down to 5V in this case is (R=Vdifference/I = 7/0.0002= 35000) 35K Ohm.

The only question is whether the output signal is going to be strong enough to be detected by the fan header... (else there may have to be a resistor bridge between fan header + and signal line, but I'm not sure about the value that one needs to have --I'll have to do some research on that). The above setup won't do any damage for sure, as long as you use the correct resistor for the transistor (else you blow the transistor, not the board). Perhaps we should take this to the Electronics forum and see what they say!

One more problem though is how you convert the pulse rate to temperature values. MBM won't let you do that, but Samurize is more flexible and allows you to write scripts (I think) that display values which are a formula of measured values. In my flow sensor's case for instance, I have to divide the RPM signal put out to the fan header by 6000 to get the exact l/min. flow. value.

 

Good idea mate Thanks for all your input, looks like i've got a new feature for my project

 

No prob. Thanks for thinking out of the box. I never considered the possibility that there might be a way (or a part) that would allow you to sense temperature straight through the fan header! I think we've got a scoop here!