0
\$\begingroup\$

Is it necessary to connect isolated grounds for 2 different isolated RS485 nodes? For instance the following isolated RS485 breakout board doesn't have any GND connection in the board.

https://www.mikroe.com/rs485-isolator-click

If I connect them together, will it improve the performance in a noisy environment?

According to my understanding signal ground is non isolated part of the supply as shown in figure. Is there anything wrong if I connect signal GND to chassis GND and connecting all isolated GNDs together(Without connecting to signal GND)?

Please correct me if I am wrong.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Welcome to EE.SE. \$\endgroup\$ – Sparky256 Nov 14 '18 at 4:35
  • \$\begingroup\$ For a proper design you always need a signal ground. There are plenty of the hobbyist level transceiver boards on the market, but they can be used maximum for a point-to-point test to see if the communication works at all \$\endgroup\$ – GAttuso Nov 14 '18 at 7:34
4
\$\begingroup\$

RS-485 standards mandate that a separate signal ground (Not always the same as Earth ground) wire be used to connect signal grounds at remote nodes so that the common mode of the signal wires stays close to the center voltage of 2.5 volts, or 1/2 of the 5 volts that RS-485 IC's normally use.

Without a common signal ground wire hopping from node to node then nodes can drift away from the common center point and data corruption will occur. It will not hurt the RS-485 IC's as they are designed to be electrically 'tough'. The signal ground is defined as the RS-485 IC's own ground connection on the board it is soldered to.

A better design would have offered you a 5 pin connector or a separate ground screw. If just 2 nodes there should not be much drift, but 8 nodes spread out over a hundred feet would need a signal ground wire (18 gauge is good enough) from node to node.

As you can see from this drawing there can be only 1 Earth ground if it is used. This is copied from page 6 of:

Tyco Electronics RS485 and Modbus Protocol Guide

Link to Tyco Electronics

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Does it apply for isolated RS485(Both power and signals)? If yes, then which ground needs to be connected together? Non isolated GND or isolated GND? \$\endgroup\$ – Supreeth Nov 14 '18 at 5:10
  • 1
    \$\begingroup\$ Isolated ground is normally your signal ground. This is what connects the signal grounds of the other nodes together. Do not earth ground the signal ground. \$\endgroup\$ – Sparky256 Nov 14 '18 at 6:26
  • \$\begingroup\$ I am a little bit confused about the term "signal GND". Is this same as GND in non isolated part of the supply as shown in figure? I have edited my question with the figure. \$\endgroup\$ – Supreeth Nov 14 '18 at 8:25
  • \$\begingroup\$ I think you should clarify where the signal ground should be connected to the supply ground. Because I read this answer as "connect the signal ground of each node to supply ground then daisy chain them", which is obviously incorrect, as doing so will let loose all manner of ground noise. Also, star connection is ideal for ground but not ideal for the RS-485 signals! Don't mix these two up, the question is about grounds, not how to design a RS-485 network without stubs. (Practically, you might not be able to mix the two topologies.) \$\endgroup\$ – Lundin Nov 14 '18 at 10:35
  • 1
    \$\begingroup\$ And why would AWG18 matter? If you have higher currents going through the signal ground, you are doing it wrong. The only current going there should be what drives the galvanic isolation + termination. \$\endgroup\$ – Lundin Nov 14 '18 at 10:40
1
\$\begingroup\$

I wouldn't be concerned too much. In an industrial RS485, usually a shelded cable with two wires (A and B) is used. The shield only sheilds EMI noise and it is connected to the ground. For fairly long distances it is not recommended to connect two grounds together due to compensating currents that will flow trough the shield.

Since the two transceivers are floating, they will reach the same potential when connected together. Usually there are also high value resistors connected to signal wire and to ISO_GND and to ISO_Vcc, these are called bleeding resistors and they leak some current to equalize the floating potential.

What I really miss are some TVS diodes to protect large potential difference before two devices get connected, but I guess the ADM2682E has inbuilt protection also, so don't worry.

\$\endgroup\$
0
\$\begingroup\$

In principle, being a differential standard, as long as the common mode potential of the interface IC is not exceeded, it should make no difference. These ICs tend to be able to tolerate common mode differences of a few tens of volts.

In practice the different ground potentials are commonly brought in close proximity by some form of common-mode termination of the differential lines or just leakage from the ICs. Keep in mind that the output driver specifies its voltages to its own ground, a split termination is all that’s needed to reference the grounds to each other. That saves you the need to add a third wire to the interface.

If you expect large AC common-mode disturbances, the communication modules are sufficiently isolated from the equipment, and common-mode termination is not practical or adequate, then yes. Tying the isolated grounds together via a third “field ground” wire would reduce the chance of garbled data and damaged ICs.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.