RS422/485 and ground wire.

[HellasTechn]

Greetings dear friends.

Today I was reading about RS422/485 protocols. I do know that these are differential line but that raise a question. Do they really need an extra signal ground wire or not ?
Ive seen it many times used without a ground wire, just the data + and data- lines.
How does the receiver get the information? Where does current flow to if there is no ground wire ?
I guess the line that is at 0V potential probably sinks current back also practically I have never seen rs422 systems use common ground. Is it because they share common ground through their power supplies? If that is the case then is it the way it is meant to be?
Not to mention that I have seen rs232 devices reading rs422 directly by interpeting 0V as -12 and +5V as +12V which i would NEVER do because this way you put the other data line directly to ground.

 

[Minder]

The date is sent between two conductors, that is all you need, in many cases the shield is connected to earth GND and/or to the common of the transmitting device,
They may, or may not use the same supply, generally each has its own.
M.

 

[Harald Kapp]

Do they really need an extra signal ground wire or not ?

Differential doesn't mean that no ground is required. At least in most cases.

In an ideal world the currents in both lines of a ideal differential signal would be identical in magnitude but opposite in sign.
In the real world you never have currents completely identical in magnitude. The difference in current has to be balanced. With a dedicated ground wire this is the shortest path for the difference current to take and distortions and susceptiubility to noise are minimized.
Often differential system will work seemingly without a ground wire, But that doesn't mean there is none. There is no dedicated return wire in these systems, but the differential current "finds" a way along other connections, e.g. the earth connections of the sending and the receiving system. This opens a current loop with a large enclosed area and makes the system susceptible to electromagnetic disturbances and also creates an antenna which radiates electromagnetic radiation.

I always recommend to have a return (ground) wire with differential signalling. Using the shield, as @Minder mentioned, is one possible method, but imho only second best. I suggest you use a return wire within the cable, connected on both sides to ground of transmitter and receiver respectively. Connect shield to earth to protect the signal wires from external electromagnetic disturbances.
Note: gnd is not necessarily equal to earth, but it can be, depending on the ground concept of the system. This is often confused or at least not handled in a defined way. Not a simple topic to be explained in a few words. If you're interested, read up more e.g. here.

 

[Harald Kapp]

Not to mention that I have seen rs232 devices reading rs422 directly by interpeting 0V as -12 and +5V as +12V which i would NEVER do because this way you put the other data line directly to ground.

1. You can do that if the two systems have separate power with grounds not being connected.
2. You can do that by using only one of the differentiual data lines plus a common ground (the second data line is then not connected to gnd but left floating).

In both cases features of the receiver are used (i.e. the sensitivity to input voltages in the range 0 V ... 3 V) that are not conform with the RS-232 standard. SO this may or may not work, depending on the parameters of the circuit componnets used.
It will not work withthout additional circuitry for bi-directional RS-485 or RS-422 circuits.

 

[HellasTechn]

Interesting what Wikipedia writes about this.
https://en.wikipedia.org/wiki/RS-422

RS-422 specifies differential signaling, with every data line paired with a dedicated return line. It is the voltage difference between these two lines that define the mark and space, rather than, as in RS-232, the difference in voltage between a data line and a local ground.

 

[Harald Kapp]

paired with a dedicated return line.

... a ground plane in the case of routing on a PCB being the best option. This minimizes the area enclosed by the signal lines and the return path thus minimizing radiation and susceptibility to such.

 

[HellasTechn]

Yes that’s exactly what ground plane does but in this case i believe the text is referring to the A and B lines. In other words that at a given time one acts as tx and the other as return.

 

[Harald Kapp]

Sorry, I disagree. When you read on, you'll find this statement:

Moving to dedicated return lines and always defining ground in reference to the sender allowed RS-422 to use 0.4 V, allowing it to run at much higher speeds. RS-423 differed primarily in that it had a single return pin instead of one for each data pin.

which tells us that these dedicated return lines are really separate return lines, one for each data line A and B.

 

[HellasTechn]

Say i have a battery powered rs422 devise and want to talk to an other battery powered rs422. No ground between them. This will never happen if i do not use the A and B lines plus a 3rd line as common ground. Yes

 

[Minder]

Not for me it hasn't, just two conductors A&B.
If you happen to pick up the $1 USB to RS485 convertors on ebay they just have the two terminals only.
If shielded cable, I just earth GND the shield only.
M.

 

[Harald Kapp]

2 wires only works in astonishingly many cases. Having a dedicated return is in my experience definitely more robust.

 

[HellasTechn]

about a year ago i was testing an rs422 to uart converter chip powered from my bench psu (not earthed) with an rs422 gps receiver powerd from an smps wall wart (also not earthed). i remember clearly that it would work with either the A and B line connected or with only B line and both grounds tied together.

I am confused...

 

[Minder]

Most of the Technical sheets from the manuf. don't specify any return or common GND conductor.
Texas Inst.
M.

 

[HellasTechn]

Here is the thing, though they do not show any common line, they dont show any VCC either. That doesnt mean that there is not any. Maybe in these diagrams they assume that VCC and GND is common for all.

 

[HellasTechn]

I beleve they take it for granted that there will be common ground. Just like they take it for granted that there is VCC thus they just not show it on their diagrams.

I found this pdf on www.robustdc.com
On page 4 it says :
Data communication systems involve connecting multiple "systems" together and therefore careful thought must be given to grounding. It is a common misconception that RS-485 requires only "two wires". This is never true. RS-485 always requires at least three conductors: 2 signal wires and 1 signal return path. The EIA/RS-485 standard states:
"Proper operation of the generator and receiver circuits requires the presence of a signal return path between the circuit grounds of the equipment at each end of the interconnection. The circuit reference may be established by a third conductor connecting the common leads of devices, or it may be established by connections in each using equipment to an earth reference."

So the way i see it now, there is no doubt that we either need common ground on all devices (maybe through the power socket) or a 3rd common wire.

Also in page 3 it mentions galvanic isolation and if not present then ground wire can not be used. I dont fully understand this. also what is galvanic isolation ?

 

[Minder]

Galvanic isolation is complete isolation between one section of a system and another, for e.g. a transformer with a separate primary and secondary is said to have galvanic isolation.
How would you explain all the RS485 devices out there that only have two terminations for the RS485 connection,? VFD's for one example.
And the many items cheap RS485 interfaces I mentioned earlier, all work.with two connections.
To expand on the galvanic isolation.
An example is when a 480v transformer has a 120v secondary for supplying control circuitry,
If you leave the secondary as is, you have galvanic isolation and no longer have one side as a neutral and connected to ground.
NFPA79 & NEC allow you the option to connect one side of the transformer to earth ground, this conductor then becomes a grounded neutral.
But the control circuit functions will operate OK in both methods.
M.

 

[HellasTechn]

I think I understand but I can not imagine how signal can pass from one point to another without completing circuit some how

 

[Minder]

To complete a circuit you normally require two conductors, these two input to a amplifier similar to any op amp etc where only two conductors are required.
M.

 

[HellasTechn]

Please excuse my confusion. I either have things mixed up in my mind badly or i am missing alot.
I read that in the rs422 standard the 2 conductors are one way only conductors. In my mind i picture it like the diagram i attach below. How can current return to the power source if some short of common ground is not present ? It is not like an ac signal that when shifting phase what is actually going on is that electrons change direction. do i make sense ?
I picture an rs422 bus without some short of common ground like the top circuit.
Is my thinking wrong ?

 

[Harald Kapp]

Your circuit shows a resistor as a differential component between A and B lines. In that case any current flowing into the resistor via the A line comes out via the B line and no ground connection is required as the receiver only evaluates the voltage drop across R. Current returns to gnd of the transmitter via the "batteries" shown. You have a closed circuit and that is fine.


This assumes a very high impedance of the voltage sensing circuit to minimize any current going not through the resistor but into the circuit. In most cases this will be the case and everything will work under normal circumstance without visible issues even without a dedicated return line.

However, if the voltage sensing circuit is not completely symmetrical or stray capacitances (impedances) due to layout asymmetries lead to different currents in the A line and B line (note that the signal has an AC component therefore capacitances are relevant), the signal can be disturbed. The easy way to fix this is to add a dedicated return line so that the residual current (difference between current in A and B) can return via the shortest path back to the transmitter.

You have to take into account that the real world is not as simple as your schematic makes it seem. There are stray impedances everywhere (between wires, from wires to ground), grounds between two systems may be on different potential due to impedances of the ground connections etc.

In many (maybe most) cases you'll get away with two wires and never notice a problem. But adding a third wire for return from the beginning is possibly less costly than to invest hours into finding a problem and fixing it should you run into problems with data integrity.