6

Is using twisted pair for the supply in a long cable a good idea? The cable contains two twisted pairs; one twisted pair for CAN Bus, and the other twisted pair would be used for GND / Supply.

What would be the effects of using such a cable? Good / Bad?

JYelton
  • 32,302
  • 33
  • 134
  • 249
SensorMan
  • 211
  • 3
  • 5
  • See: Power over Ethernet. – Passerby Dec 02 '13 at 08:29
  • What supply: voltage, current, DC or AC, frequency, ripple? – Wouter van Ooijen Dec 02 '13 at 08:36
  • 12 to 28 volt supply from truck batteries on industrial machines. It's a bus system, so sensors will be added in parallel. The sensors don't use much current, 30mA to 90mA each and not more than 10 will be connected. My previous experience tells me that there is lots of electrical noise coming from bad generators. – SensorMan Dec 02 '13 at 08:56

3 Answers3

5

A possible negative effect could be the resistance of the wire. That's not caused by the twisted pair, but rather the wire diameter. If they're really signal wires the diameter will be rather small, and have a non-negligible resistance. The resistance will cause a voltage drop, which may become noticeable at longer distances.

If the wire is thick enough you won't have any negative effects, but no positive effects either. The power supply has a very low impedance, so it won't pick up from the signal, and at DC it won't inject any disturbances into the signal wires either.

radagast
  • 1,794
  • 3
  • 17
  • 28
2

Power over Ethernet (PoE) can provide up to 15, 25 or 51 W over twisted pair data cables by passing power over unused pairs or by applying a common mode voltage over the signalling pairs. Using more pairs obviously helps allow greater current. PoE tends to use higher voltages (up to 57 V) which may also help overcome resistive losses.

So, supplying power over twisted pair signal cables is well-established.

RedGrittyBrick
  • 14,602
  • 5
  • 37
  • 77
  • 1
    As far as I know they use one pair for ground and one pair for positive (4 wires totally), so this is not the same as I want to do. – SensorMan Dec 02 '13 at 09:54
2

Theoretically a cable having a twisted pair for data and two other conductors for power and return is OK - induced currents from the power conductors will tend to equalize on both data wires (because they are twisted) and only produce a small common-mode corrupting signal that can be dealt with by using appropriate components on the CAN bus receivers.

However, if a twisted pair is also used for power, performance can get worse if the twists on the power and twists on the data line up - now you will get a potentially substantial transverse interfering signal on your data.

If the twists are at different pitches then there will be much less of a problem but it all depends on cable length, number of twists and the bandwidth of the current flowing down the power conductors. This is alleviated by having good decoupling caps at both ends of the power cable to try and ensure current down these wires are only slow moving relative to the data.

It's the same problem with phantom powering a CAN bus (or any low power serial) module down the data wires - the power has to not be jumping around too much compared to the magnitude of the data.

Andy aka
  • 434,556
  • 28
  • 351
  • 777