THVD8000: Ground difference waveform, are these waveforms expected or do we require a 3rd wire between the Transceivers Ground to make the system more stable

Part Number: THVD8000

System Architecture & Testing setup
We have attached below the image of the detailed architecture of our system, we have kept the oscilloscope GND common with the master THVD8000 GND ( which comes directly from the external power supply GND) and probed the GND of the slave node, the 0.5V - 0.7V GND difference is due to the bridge rectifier ( which ensures the slave device gets powered when connected in reverse polarity ). 
The architecture which we have used is very similar to the architecture proposed in the Application section of the Datasheet "THVD8000 RS-485 Transceiver with OOK Modulation for Power Line Communication" - Section 9.2, Figure 9.1 .
 
Waveforms Captured :
When no communication is taking place the ground difference comes out to be 0.7V ( due to diode rectifier at slave ) & when communication is taking place we observe a peak to peak of 7V difference between the ground of Master THVD8000 vs Slave THVD8000. I have attached the waveforms for your reference.
Fig1: When no communication is taking place
 
Fig2: When communication is taking place 
 
Fig3: When communication is taking place
 
We have used isolated probe as well for taking measurement, but results are same as seen in the waveforms. 
 
Question: 
1) Can the ground difference as seen in the waveform cause any communication failure ?
2) Should we use an additional wire to connect the transcievers ground together & will that make the system more stable in terms of communication reliabilty ( additional wire causes these ground differences to disappear )
 
  •  Question 3:

    In this architecture, during communication it seems to disturb the GND wire by modulating differential signal, can the same GND wire which carries data be used to filter any high or low frequency noise present in the slave node

  • Hi Ashutosh,

    Today is a holiday. Please help give until next week for some feedback, thanks. 

    Best Regards,

    Michael. 

  • 1) Can the ground difference as seen in the waveform cause any communication failure ?

    It shouldn't since the communication is differential. The receiver (slave) will simply look at the A voltage and B voltage and do a subtraction (A-B) to calculate the actual state. If there is a GND shift of 7V on A then there will also be a 7V shift on B. 7V-7V=0V so the device wouldn't even see the GND shift from that perspective. This assumes that the cable length of A is equal to B and the signal arrives at A and B at the same time. If A is longer than B then this math function may not work due to time of flight mismatches.

    2) Should we use an additional wire to connect the transcievers ground together & will that make the system more stable in terms of communication reliabilty ( additional wire causes these ground differences to disappear )

    Are you actually seeing communication issues? You might need to probe the A/B pins on both sides (master node and slave node) and verify if the communication is what you would expect it to be. 

    From your set up, I don't immediately expect this to be some kind of GND shift problem since the communication is differential.

    In this architecture, during communication it seems to disturb the GND wire by modulating differential signal, can the same GND wire which carries data be used to filter any high or low frequency noise present in the slave node

    You can try to filter the signal but you need to think in terms of differentials. If you place a cap on the B node, you need to place the same cap on A node to keep both nodes balanced. 

    -Bobby