Hallo George,

for me it's a bit difficult to follow the discussion. You are posting so many questions and you are mixing everything together... Grounding must be discussed in so many different aspects, that some simple answer would never be appropriate.


Can the common ground between each RS-485 node be the same conductor as the shield of the twisted pair? I would expect that this may not work if each node was also individully grounded to earth as ground loop currents may result.

Also, if the transmitter is galvanically and optically isolated, is a common ground for communication no longer required? If not, and if the device is battery powered and has no external metal parts, can individual nodes totally float without any grounding?

Shield of cable has the same function like metallic enclosure or housing for a electronic circuit: It must completely surround electronics inside, AND no signal current must flow through it!! So, cable shield must not be used for connecting two separately located grounds (0V of circuits). No signal current is allowed to flow through cable shield.
Cable shield is part of Faraday cage, which idealy completely surrrounds whole application: Metallic enclosure, where electronic circuit sits inside and cable shield, where wires are running inside. (Unfortunately this concept can also lead to ground loops, which can sometimes only be overcome by making a cut in the shield somewhere.)

As consequence, a cable shield will most often be connected to metallic enclosure, where cable penetrates enclosure. Cable shield has nothing to do with signal ground, it's part of Faraday cage.

If you want to connect grounds (0V of circuit, means signal ground) of two RS-485 nodes you must use an additional wire, which directly connect those two separately located grounds. This wire should also sit inside this cable shield.

Why is it necessary to connect the two grounds (0V of circuits)?
One might think, that due to nature of symmetrical signal, there shouldn't be any need to transmit signal ground, because it can be recovered by the reciever. But keep in mind, that common mode input impedance of a RS-485 transciever is very high, some dozens of kOhm, in order to be able to connect many of them to the same net, means to keep the common mode loading minimal. But in combination with common mode noise this would be deadly. Assume you want to connect some nodes, where one is mains powered by the help of safety class II transformer. If one other node sees a connection to PE (safety ground, earth, yellow green earth cable), mains voltage is injected into RS-485 lines via interwinding capacitance of mains transformer and common mode input voltage range of trancievers would be overloaded.

So, if voltage supply of two different and separated nodes is not identical, you must connect individual 0V of circuits by a wire, which should run in the same shielded cable as RS-485 lines.
If one node does not have it's own supply voltage, but must be powered by some other node, it's wise to connect these both nodes by a shielded cable, containing two twisted pairs: One pair for the two RS-485 lines, the other for Vcc/0V.

I dont understand where you tie your shield I have some confusion amoung US prevalent terms. Ground you said GND is this related to PE , when you say your cable runs from GND to GND 'O' My. Do you mean you connect your shield to circuit common at one place ,or PE at one place.

First, you must understand, that cable shield has nothing to do with signal ground, it's part of Faraday cage. But where to connect this Faraday cage, and where to connect signal grounds?

When working only with battery powered systems or safety class II devices, theoretically, Faraday cage could float, but it's wise to have it connected to earth potential somewhere.
Working with safety class I devices, which must see a direct connection between metallic enclosure to PE, cable shield connected to enclosure is automatically connected to PE, too.
If at more than one node Faraday cage must be connected to PE because of fullfilling safety class I requirements, cable connecting these two nodes must have a cut in the shield. Otherwise ground loop will result. In this case cable shield is connected to metallic enclosure at only one side.

Signal ground must see a connection to Faraday cage at each node (modul), at least for high frequencies. Otherwise, Faraday cage acts no longer as shield but as an antenna injecting noise into circuit. Only if Faraday cage is forced to have same potential as 0V of circuit at each modul, no interference can couple into circuit of each modul via stray capacitance.
Working with safety class I devices containing mains transformer complying with safety class I standard, not only metallic enclosure must be directly connected to PE, but also signal ground (0V of circuit)!
When working only with battery powered systems, safety class II devices or safety class I devices containing mains transformer complying with safety class II standard (designed properly!), 0V of circuit need not to be connected directly to metallic enclosure. Connection can also be made by the help of 10...100nF capacitor paralleled by some resistance of about 1Mohm.

PE= Physical Earth => Earth Ground = place where you dig the Earth and place some stuff there and it gives you so much trouble.

Not at all! All sources of common mode noise and interference will be completely shunted to earth, away from RS-485 line.

Circuit common = 0v => Electrical ground which is internal to the circuit and always and always seperated from PE.

That's not valid either. Using safety class I mains transformer forces you to directly connect signal ground (0V of circuit) to PE!!

If your node is battery powered and enclosure is made of plastics so that no Faraday cage does exist, cable shield can directly be connected to 0V of circuit.

Tell me clearly does shield connects to masters electrical GND ,but that would contaminate the GND ,RF ,ESD , Lightning all flowing into the poor little GND raising its potential.

Omitting shield would result in some even worse, because then, interference would flow directly THROUGH transciever to 'poor little GND'!!

You can also left cable shield unconnected. But then, shielding efficiency against high frequency interference and ESD events is highly worsened. Or you can connect cable shield to 0V of circuit via 10...100nF capacitor. Then, shielding efficiency is increased again.

LTC datasheet 1535fs.pdf page 13 schematic connects all the Floating commons to the F#$@ng SHIELD and some filter there down to PE Right . Whats that B1 (CN2R) I have seen only 100E between 0v and PE at many places . But why the HELL PE in all the cases?

You have put your eyes into datasheet of a RS-485 transciever, which provides an isolation. And because isolation is necessary, whenever two nodes containing saftey class I mains transformer have to be connected, it's natural that PE is involved. Don't forget that both signal grounds must be connected to PE, then!

As I stated above, shield should be directly connected to 0V of circuit, when there's no metallic enclosure. This is done in datasheet. Nothing else!
Datasheet also takes care for a certain condition: What happens, if cable is disconnected from other node? Then, cable acts like an antenna injecting noise, HF interference and especially ESD (!!) directly into inputs of RS-485transciever and further into isolation barrier! Then, using a cable with shield, which is connected to 0V of circuit (isolated RS-485 common) and using a 10nF capacitor connected from RS-485 common to PE will shunt all this noise and ESD directly to earth potential. So, damage of transciever and isolation barrier is prevented!!

Do you apply same termination resistances to a 4000Feet cable (Vcc----730E----60E-----730E----0v) and other end 62E , and to a 100Feet cable?

Why not Vcc--750E--130E--750E--0V and other end 120E?
For 4000feet cable, 100feet cable and any other length you will always need the same termination.
Why?
Because characteristic impedance of cable is independent of length! If L is inductivity per length and C is capacitiy per length of cable, then characteristic impedance of cable is about SQRT(L/C), which is independent of length of cable!

Any Leftover twisted pair in the cable what to do with em ,I hope they will resonate to some frequency ,and if grounded at one end would still resonate ,what to do .

Your RS-485 line twisted pair will also undergoe these resonances!! One way to suppress it, is to connect Faraday cage at several points to earth (PE): If factory is equipped with a proper earth routing, inductivity of earth routing can be much smaller than inductivity of cable and resonances will be short circuited to earth potential. Cable looses ideal antenna characteristics and possibility of resonances is highly reduced. So, you see, your application can highly profit of these earth connections, especially in terms of suppression of high frequency interference.

Kai