Differential Signaling and Ground Referencing

[rudyb]

I have been asking this question but have not been able to find a reasonable answer for it.

When we have differntial signaling, we don't need a ground reference anymore, correct?
And that is because one of the pairs is acting as the ground.
But then I noticed that in different application when we are sending differntial signals from one unit to another, we somehow care about grounding, and I am not sure why?

For example, in RS-485 transactions, we normally connect a third wire as a ground reference.
Why do we need to send a third wire (ground) in RS-485 commmunications?


And in other forms (for example ethernet), we use isolation transformers for galvanic isolation.
It is not that the isolation transforms are only used for short circuit protection. I have read plenty of articles talking about that isolation transforms are also used to remove ground reference dependecny as well !!

Coming out of ethernet PHY, the signals is already differential pair (and to me that is well isolated, since there is no need for ground referecing). Then why do we even care about isolating it even more by the transformers?

I though that you could care less about ground reference differences when you are dealing with differential signals !!!

From as far as I can see, the receving end can have as much as ground difference as it wishes, and tha shouldn't matter for differntial signaling.

 

[Analog Ground]

The common ground connection helps to prevent any common mode voltage which is too large for the differential receiver by connecting the driver and receiver "grounds" together with a low resistance conductor. This is necessary since the RS-485 specification calls for direct connections of the signals between a driver and receiver.

 

[rudyb]

I am not an expert, please bare with me. How would common ground help to prevent any common mode voltage?

 

[andre_luis]

For example, in RS-485 transactions, we normally connect a third wire as a ground reference.
Why do we need to send a third wire (ground) in RS-485 commmunications?

As far I know, RS-485 electrical standards could not be classified as a differential signaling, due both TX and RX are referenced by ground reference.

Coming out of ethernet PHY, the signals is already differential pair (and to me that is well isolated, since there is no need for ground referecing). Then why do we even care about isolating it even more by the transformers?

Differential pairs signals are not necessarily isolated signals. It is achieved through galvanic isolation of transformer.
Although a self referenced pair can perform proper signaling between each board, additional ground reference serves for equalize potential difference which could damage circuitry.


+++

 

[rudyb]

Maybe I am not sure what is considered an isolated signal ! What is an isoalted signal. Why we cannot say all differential pairs are isolated signals?
And, I thought isoaltion transforms are to remove the ground reference. Maybe I am still not sure why it should provide additional ground reference ?

 

[andre_luis]

...Maybe I am still not sure why it should provide additional ground reference ?

Ground reference is not an additional reference, but the only one.

This is required just for protective purpose, because the GND signal at one board have not necessarily the same potential on another board. In other words, If you could hypothetically measure the absolute potential at the GND1 at board #1 and the absolute potential on GND2 at board #2, on real world you could expect difference of up to several volts .


+++

 

[Analog Ground]

I am not an expert, please bare with me. How would common ground help to prevent any common mode voltage?

Let's say we have two boxes A and B. Each box has a local circuit ground we can call Ground A and Ground B. For a number of reasons, there may be a potential difference between the two grounds. Now, a differential signal is sent out of Box A and the two signals produced as referenced to Ground A. When received at Box B, the potential difference between the two grounds appears as a common mode signal on the two received signals. However, the voltage between the two signals is the same at Box A and Box B. Within limits, the common mode voltage is rejected and only the differential voltage is detected. Now, back to the wire. If a heavy copper bar is connected between Ground A and Ground B, the bar short circuits whatever source is causing the potential difference between the two grounds. The ground wire going along with the differential signals acts like the copper bar but smaller. If the potential difference between Ground A and Ground B is within the common mode range of the receiver, then no ground wire is needed. If the wire is not enough to limit the common mode voltage, then the receiver can have a problem.

Is RS-485 true differential signaling? I think so. For signaling purposes the wire is not needed. So, the signal is truly differential.