Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations GregLocock on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Will this RS485 network work

Status
Not open for further replies.

ccdubs

Electrical
May 1, 2002
51
Hi All,

I am trying to get my head around RS485, especially earthing. I have a simple question that should help me clarify some misunderstandings.

If I have one PLC powered by a battery and another powered by a seperate battery that are connected by RS485 (2 wire only) will the communications work? Neither battery terminals are earthed and are therefore completely isolated supplies.

What I am getting at is ... Is there a requirement for a return path of any sort for RS485 to work?
 
Replies continue below

Recommended for you

I don't have RS-485 handy, but since RS-422 is similar, there is a requirement that the common-mode voltage does not exceed certain values. Additionally, RS-422 paragraph 7.1 stipulates:

"Proper operation of the interface circuits, whether using balanced, unbalanced, or a combination of both, requires the presence of a signal common path between the circuit commons of the equipment at each end of the interconnection. The signal common interchange lead shall be connected to the circuit common, which shall be connected to the protective ground by any one of the following methods, shown in figure 16, as required by specific application:"



TTFN
 
Here we have 2 points, the operation and noise.
From operation point of view, it will work because RS485 detects the voltage difference between the two lines rather than the absolute value, mathimatically,GND voltage will be cancelled out from the equation, I saw several circuits use this way.
However the input and output circuit can not go to any level in the - what is called - common mode voltage or the voltage that will not be detected as logic change, above several voltaged (7 and up depends on the particular chip used- refer to data sheet) the chip will breakdown and may be permenantly damaged, this common mode voltage can be piced up in long lines from near power lines or static, that is why it is a good practice to Earth a shield or ground it from one end only. Not both ends? - Please note that Ground returns are a major source of noise and Hum pick up.
Since you use battery, Hum problem is a minor issue here.

Maged A. Mohamed
 
Thank you for your comments.

The network described in my question is not the network I am fault finding. The actual network is more complex.

There seems to be some contradiction in the 2 answers provided. The exert from IRStuff says:

"Proper operation of the interface circuits, whether using balanced, unbalanced, or a combination of both, requires the presence of a signal common path between the circuit commons of the equipment at each end of the interconnection."

This is not the case in my example network as there would be no signal common. I guess the definition of "proper operation" is the key. It may mean it will work but not in the way it is supposed to.

So I am still looking for someone to confirm that: "RS485 requires only 2 wires/connections to operate".
 
It means it MIGHT work, and then again, it MIGHT not. If you are willing to have flaky or non-operation, then do whatever you want, which is what you seem to be indicating.

But, you run the risk of frying your differential receivers any time there is some uncontrolled potential difference between the circuits.

I don't understand why you are so reluctant to having a proper common. Bad practice is ALWAYS bad practice. While it MIGHT work THIS TIME, the person after you might do something slightly different and have a complete failure.

TTFN
 
from the article:
Remember when ordering your cable, "two-wire" is really two wires + ground, and "four-wire" is really four wires + ground

TTFN
 
Once again thank you for your help.

I believe I have this sorted now.

IRstuff: just to clarify, the network I am fault finding (which I didn't design :) ) seems to be working on the 2 wire method with no signal ground. This network has blown the com ports on 2 plcs. So now I intend to provide a signal ground plus some isolation and surge protection.
 
ONLY 2? That's pretty good.

My system blew about a dozen diff rcvrs, before I figured out that the two ends where being zapped during power-on, because the power supplies were dis-similar and had a huge potential difference that was being equalized through the emitter-base junctions in the inputs of the receivers. The E-B junctions can only tolerate a maximum of 7V common-mode. Beyond that, they die a horrible death.

BTW, an easy test is to power up the rcvrs and measure their short circuit input current by grounding the inputs through a current meter. A E-B junction short will cause upwards of 100 mA to GND, while a good input should have less than 1 mA. Anything in between is an indication of latent damage that might bite you later.

TTFN
 
Boy that was dumb of me. I had something else on my mind, obviously. IR is right.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor