Should long wires be kept separated from each other, to reduce interference?

Question

Let's say someone has four CAT5 cables that are running along the ceiling for the length of a room (30 feet), of which most of the twisted pairs are connected to Arduino input/output pins on one end. Would it be okay to keep these wires bundled together when running along the ceiling? Or is it safer to keep them separated by a few inches? Is interference usually a problem only in very specific situations, or should it be considered for all long wire runs?

How about if there is a telephone cable (hooked up to a landline telephone) running along the same length? Is there any chance of interference if this is too close to the Arduino cables?

How about if there are 120v mains lines running the same length? Should these be kept at a distance from the Arduino cables?

On the other end of the cables that are hooked up to the Arduino, some of the following are connected: temperature sensors, light sensors, servo motors, LCDs, relays, and push-buttons.

Answer 1

A benefit of using CAT-5 is that the pairs are twisted. You would want to look at how you are using the pairs so that you're not introducing cross talk inadvertently, but the twisted pairs tend to resist RFI which is why its used. Telephone wire, however, can be a completely different story. You may find some that is twisted pair, but most if it is simply flat cable. Worse, much of it is very cheap flat cable with what amounts to a piece of thread with a very thin foil wrapped around it. This is very high loss and very vulnerable to RFI.

Answer 2

It depends on the frequency/frequencies you care about transmitting on your cables. One simple approach is to put on both ends a filter that lets pass only those frequencies you care about. You can also consider using some error detection/correction encoding or check if there is some existing library that suits your needs. This last approach will work if you do not get too many interferences and your application can tolerate that data is re-transmitted every now and then, making the communication timing not 100% deterministic.

Answer 3

CAT5 should work well with what you are trying to do, however stay away from power lines unless they are in a grounded conduit. The phone lines would most likely generate interference when the phone is ringing as there is a lot of voltage transients on the line at that time.

To get the full benefit of the twisted pairs send and receive your signals with differential transceivers, one transceiver per pair.

Personally I will not connect a microprocessor pin to the outside world without some type of buffering isolation etc. Those pins were not designed to drive long lines, although it may work eventually it will fail, that may be momentary to many years in the making but it will fail.

Lines connected directly to the microprocessor pins will all act as bidirectional antennas. They are not only are sustainable to interference they also generate it. Remember the wires are small inductors which can generate some huge voltages when the transition time is short.

I typically use Schmidt devices such as a 74C14, 74HC etc. as my input receivers for switches etc. I normally use an active low input signal, this help to keep noise out of the power supply. The input to the gate is pulled up to the 12VDC with a 10K resistor, this keeps a minimum of about 1 mA through the contact. The other end of the resistor is connected to a 100K resistor that in turn connects to the microprocessor input pin. You can add a cap to de-bounce at the gate if you want. The software is easier as it is looking for an active high which is a logic "1". This circuit has been replicated hundreds of thousands of times and is very stable and reliable.

What you are doing will work for dorm room, testing etc...