Connecting a thermistor to very long cables

Question

I'm working on a design for a multi-sensor temperature alarm for a greenhouse.

I plan on using PTC thermistors as the temperature sensors as I have a few 1k PTCs lying around.

Can I connect the PTC to a very long (about 20-30m) cable and then use a capacitor in parallel on my PCB as a low pass filter set to a couple of Hz (i.e. 10Hz or lower) and then feed the signal into a window comparator? Will this work or is the long cable and noise a problem? For the cable I will try to find something like shielded twisted pair, if that's available and cheap.

I have looked around on the internet but have not found any info for more than like 5-10m of cable length, for which the info suggests twisted or shielded cable and a capacitor to form a low pass filter.

Edit: I have not yet created a full schematic or even planned every part of the device. Anyways:

As requested I quickly drew a schematic. This is not in any way final, but it is roughly what I was thinking about. The output would drive some other circuits, but that is not relevant here. Also the parts are not neccesarily going to stay the same, as I am just at the very beggining of this project. The PTC is a KTY 81-110 for which you can easily find a datasheet online and that is the only part that I would like to keep as it is, because I have some of them lying around.

As I said, basically everything here is subject to change. My question was simply meant to find out if this is possible without any other electronics in the sensor.

Answer 1

You long cable is effectively going to change the thermistor part of your circuit to this:

^{simulate this circuit – Schematic created using CircuitLab}

As long as R1 and R2, the series resistance of the cable wires are low with respect to the 500 Ohm - ~2K Ohm resistance of your sensor you should be ok.

The value of C1 will affect how quickly you can read a change in temperature. Again this value is related to your cable. As long as you can live with this added delay you will be ok.

The biggest concern in a situation like this is noise which will affect your readings. You have a capacitor in your circuit which will help but using a twisted pair cable will also help.

Answer 2

Better to add some series resistance on the signal line. Something like 1K and 100uF. Note that R2 is on the other side of the added resistor from the chip inputs and the capacitor.

Add a GDT to earth at the inputs to help protect from induced currents from nearby lightning strikes. A brief transient of a few hundred V is acceptable if you use reasonably large resistors, and the discharge tube can deal with anything above that.

Answer 3

Measuring a variable resistance over a long cable is possible. One suggestion is to feed the cable with a constant current source, and measure the voltage across it. Depending upon the resistance being measured, and the resistance of the wires, and the accuracy you need, you may need to compensate for the voltage drop along the length of the wires.

If greater accuracy is needed, you may apply a 4-wire, or "Kelvin" method. Apply a constant current source to one pair of wires. These connect to your variable resistance. A second pair also connects to the variable resistance, and you measure the voltage across the pair that is NOT fed with a current source. Because little current is flowing through the second pair, the resistance of the wires will have less impact on the voltage at your measuring end.

Noise, may be a problem, so it is best to use twisted pair. It is also best if you have a low resistance common mode inductive choke on both ends of the pair with the return voltage (if using the 4-wire method) or the single pair if there is only one.

Depending upon your knowledge of shielding and grounding, shielded twisted pair may be better than unshielded. However, improper use of shielded cable may actually increase noise. If you use shielding, it must only be grounded on one end, and the ground needs to ultimately connect to a real earth ground. You must avoid creating ground loops.

Addendum: Since your sensor is measuring temperature, I have two further suggestions.

One, the current supplied to the thermistor should be small. Although you can compensate for it, whatever current flows through the thermistor will heat it above the surrounding temperature. The datasheets of most temperature measuring devices will specify the size of this effect. I would suggest keeping your driving current to 1mA or less, but check the datasheet for your thermistor. (Or maybe a few degrees off doesn't matter to your application).

Two, since it is unlikely that you will need high speed in your application, use a low pass filter before whatever ADC you are using, to reduce noise.