Hi all, I am running a few 10k thermistors around. The longest away will probably be about 10-15 meters, I'm using cat5e so at least they will be twisted.
I'm measuring them with the arduino, and using the onboard 5v output for the divider circuits. I have got plenty of thermistors around and pinpoint accuracy is not really a necessity, so I'm not using digital temperature sensors at this stage. But id still like to keep it as accurate as possible with the components that I'm using
I think I have got two options, not sure which would be the best..
set up voltage divider at the location of the thermistor, 3 wires (two for power and one for variable voltage output) and measure the return voltage
set up voltage divider at the location of the arduino, use just two wires (connected to the thermistor) and measure the resistance
I think there will be little difference between the two wiring methods, both will have the same performance noise wise in my opinion. A bigger factor will be the actual output impedance value as seen by the analog input pins of the voltage divider. 10K ohms or less is the recommended driving impedance of a signal wired to a analog input pin.
100 metres of cat 5 typically has a total loop resistance of less than 20 ohms so in your case you are looking at a loop resistance of around 3 ohms. At 50 degrees C a 10k RTD has a slope of around 150ohms per degree C. Therefore introducing a 3 ohm wire loop resistance will introduce a measured error of around 0.02 degrees centigrade. It is unlikely that your ADC will detect this so, you can ignore the effect of loop resistance error, irrespective of whether you go for 2-wire or 3-wire installation. Given the choice, I'd go for 2-wire since the field end becomes much simpler and the cable is twisted pairs.
Because its a very low bandwidth signal the only problem likely is RFI, so put a 10nF (or so) cap across the Arduino ends of the wire to suppress this perhaps? With very long wires you might need to take the wire's resistance into account.
