0

I have element 'A', the receiver, connected to earth, which powers modules 'B' and 'C', the signal sources, unconnected to earth. 'B' sends a differential line signal, and 'C' sends an analog signal. My cables from 'A' to 'B' and from 'A' to 'C' have ground wires.

I read we usually connect the cable's shield to the ground on the signal source's side. In my case, that would be to elements 'B' and 'C'. However, elements 'B' and 'C' are powered by element 'A', the signal receiver. Therefore, should I connect my shields to the ground on the side of element 'A', the signal receiver, instead?

Edit:

  • Differential line of 'B' is RS-422 with twisted pairs and a maximum frequency of 10 MHz
  • Analog signals of 'C' are from temperature, humidity and pressure sensors.
  • Power and ground are twisted pairs in both cables
  • Analog and ground are twisted pairs in cable 'C' to 'A'
  • 'A' has its ground plane connected to earth via 4 screws.
  • There is no galvanic isolation.
  • Both cable lengths are a maximum of 3 meters.
Tommy95
  • 87
  • 6
  • With your description of the system and not knowing anything about the signal or the cable, there is no one thing that is better. In some cases it might be better to connect shield to ground on both ends, or use AC coupling on one end and DC coupling on the other end. For example, USB2 has power to one way and data to both ways, and diffrent devices may connect the shield differently. – Justme Dec 02 '23 at 08:26
  • Thanks for your reply! Sorry for the limited information, I'm a physicist so my intuition in electronics has its limits. Could you please tell me which information I can add regarding the signal and the cable? – Tommy95 Dec 02 '23 at 11:46
  • So you should think about how to explain even to another physicist how to replicate your experiment. What devices they are, what signals go between them, how to connect them together. If these exact devices are not available, could e.g. another generic oscilloscope be used in place of a certain generic oscilloscope. It makes a huge difderence if these signals are e.g. very weak and low voltage signals from brain electrodes, or kilowatts of RF power at 10GHz for an antenna. – Justme Dec 02 '23 at 12:42
  • 1
    It is almost always better to ground the shield on both ends. Generic advice to the contrary can be disregarded with wanton abandon. Specific advice to the contrary, is possible, i.e. there are certain situations where it is acceptable, or preferable even, but this happens far less often. – Tim Williams Dec 02 '23 at 14:18
  • Thanks @Justme. My differential lines going from 'B' to 'A' are RS-422 signals and the analog signals going from 'C' to 'A' are from temperature, humidity and pressure sensors. The RS422 signals are at maximum 10MHz. – Tommy95 Dec 02 '23 at 17:02
  • Thanks Tim Williams. In my case, I am using RS-422 differential signals at a maximum of 10 MHz in one of the cables (signals going from 'B' to 'A', with ground wire) and so I believe this at least is sensitive to grounding issues, such as ground loops. There is no galvanic isolation, and the ground plane of 'A' is connected to Earth via 4 screws. Reminder: module 'A' powers both 'B' and 'C' via these cables for which I am trying to figure out the shielding. – Tommy95 Dec 02 '23 at 17:09
  • @TimWilliams Where is it said it is almost always better to connect shield to ground on both ends? Especially in this case where there is a ground wire inside the shield, and the last thing you want is to pass current in the shield, or even share current with both shield and the ground wire. Now, of course what to and in which case depends on if the 10 MHz RS-422 wiring is twisted pair or random wiring so how it will couple to shield and radiate as antenna, or if it even matters for the 0.01 Hz temperature etc signals. – Justme Dec 02 '23 at 17:21
  • @Justme thanks for your input. The differential lines of RS422 are indeed twisted together, and power and ground are another twisted pair in this same cable. – Tommy95 Dec 02 '23 at 17:29
  • 1
    @Justme Yes, exactly, the shield is there to carry common mode currents; which it can't do if it's open one end, obviously. Tommy: RS-422 has excellent noise immunity (common mode input range), and doesn't usually require shielded cabling for commercial test levels. Since your B and C modules are apparently single-cable units, does this answer your question? https://electronics.stackexchange.com/a/689138/311631 – Tim Williams Dec 02 '23 at 18:54
  • @Tommy95 can you clarify more about the analog signals involved? Does each have its own wire, or pair of wires, or? Are these simple DC-level voltages, or differential, or are they frequency-encoded in some way, or...? Transmitting DC analog signals over any substantial distance is probably not a good idea, as it can be affected by resistance/inductance of the wire and other factors (and is very prone to interference, crosstalk, etc). If you're using some other method, then it's important to know what kind of analog signal we're talking about (frequency ranges, balanced/unbalanced, etc). – Foogod Dec 02 '23 at 19:13
  • Dear Foogod, I was currently thinking of twisting each analog line with a ground wire. I wouldn't mind making my analog signals differential but I don't know how to do that. I'm open to good ideas. I cannot make these analog signals digital as my microcontroller is very busy and cannot afford to waste time reading SPI. It must also be able to work years without buggs and from my experience, I2C and SPI end up bugging after days, weeks or months and require a system reset. Cable lengths are a maximum of 3 meters for both 'B' and 'C'. – Tommy95 Dec 02 '23 at 19:43
  • Thanks Tim Williams. I read your post and I feel a bit overwhelmed. I am responsible for making one of our instruments pass CE marking at work and I am a bit scared of what is to come. I'm just a physicist, but I'm learning a lot on these issues. If I understand correctly, in the case I described here, if my cables have shielding, you would connect it to ground on both sides of the cables for both the analog and RS422 modules? – Tommy95 Dec 02 '23 at 20:09
  • As for an answer -- likely one will be more forthcoming with a diagram or schematic provided (preferably both module and system level). A description of signals, levels and bandwidth in the cables would also be helpful (power lines at what V, I; signals at what level, accuracy, rate of change or bandwidth; and, digital being RS-422 up to 10Mbps, is sufficiently descriptive for that case). – Tim Williams Dec 03 '23 at 03:09

0 Answers0