• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Subscribers 32 subscribers
  • Views 2081 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

SND65HVD78 ground loop problem

mahmut turk
Intellectual 300 points
Other Parts Discussed in Thread: SN65HVD78, ISO3088

Hello,

In our design we used sn65hvd78 ic for one way data transmission between subsystems. A subsystem pcb has one data receive part and one data transmit part on it. Those parts have sn65hvd78 ics that are set to receive and transmit modes.  Subsytems are connected with cat5 cable with each other. By that way we planned to send data over subsystems. The cat5 cable only have data+ and data- differential pairs, no ground is going over cat5. Common ground is only at power supply parts of the subsystems. 

We have many subsystems that are mounted to a large area. One subsystem's average current consumption is about 3A. More than 10 subsystems are powered with a uniqe power supply.

Our problem is that sometimes the data communication fails. When we investigate the problem, we saw that the sn65hvd68 at transmit part of subsytem pcbs is damaged and not working. So data communication fails. We think that the reason of failure is ground loop since there is no isolation in send or receive parts of pcbs.

We have two options. One option is adding some componenets to pcb so that the isolation will be assured. The other option is to replace sn65hvd78 with an isolated differential line driver.  

Could you provide a solution to us? 

Could you also suggest an isolated differential driver for 10mbit data rate?

  • 0
    TI__Mastermind 28265 points

    Hello Mahturk,

    the good thing is that you do not have a ground loop. However the ground potential difference between the transmitting node and the receiving node might be larger than the +/- 7V specified in the EIA/TIA-485 standard. While the receiver section in a transceiver has high-impedance input resistors that prevent damage to the receiver in the presence of high common-mode voltages, the driver section is low-impedance when transmitting data. High-common mode voltages can therefore force the driver into current limit which can lead to data errors and even device damage.

    As you correctly stated, eliminating the high common-mode voltage requires signal and supply isolation.The following schematics provide you with two solutions:

    1) using the SN65HVD78 with a separate digital isolator, and

    2) using an RS-485 transceiver with integrated isolation.

    I think these are the solutions you are looking for.

    Best regards, Thomas

     

     

  • 0 in reply to Thomas Kugelstadt
    Intellectual 300 points

    Thank you very much for your really fast reply Thomas.

    ISO3088 will be the best choice as you recomend.

    As you mentioned above, the damage is at the transmit node. If i changed the transmit node of my subsystem pcb (so taht it will have isolated differential driver), will it be enough for solving the problem? Or should i also change the receive part of my subsystem pcb (isolating the differential driver part also at receiver node)?

    My second question is that if sn65hvd78 and ISO3088 can work together? I mean can ISO3088 decode the differential signal that is the output of sn65hvd78?

     

  • 0 in reply to mahmut turk
    TI__Mastermind 28265 points

    Hello Mathurk,

    to your first question:

    - if you have a point-to-point connection, only one isolator is required because this isolation already eliminates the common-mode potential between the two nodes.

    - if you have a bus with multiple nodes, every node should be isolated to ensure there is no common-mode potential between the other nodes. However, sometimes busses with multiple nodes can have the master node non-isolated. This is known as single ground referenced bus design.

    To your second question:

    The beauty of differential signaling is that all transceivers only care about the magnitude of the differential signal. The RS-485 standard specifies that a standard compliant transceiver, independent of its supply voltage (5V or 3.3V) must be able to drive a differential load of 54 ohms with a differential voltage of +/- 1.5V minimum, while its receiver must have a maximum input sensitivity of +/-0.2V.

    This assures that your 3.3V SN65HVD78 can reliably communicate with your 5V ISO3088 and vice versa.

    Best regards, Thomas

     

     

  • 0 in reply to Thomas Kugelstadt
    Intellectual 300 points

    Hello Thomas,

    We worked on the design taking your advises into consideration.

    I want to emphesize some points about the design that i have not mentioned yet:

    In our design the nodes are too close to each other (about 30-40cm). We used differential drivers since the output of cmos level drivers's output voltages(3.3v) in submodules are  disappearing since ground levels start to differ when many subsystems are connected back to back. Our power supplies for subsystems are far away from subsytems (about 20m). Many subsystems are fed from one power supply. We also carry Vcc and Gnd between subsystems. But when the length is at limit because of power carrying capacity of subsytems, we carry another power cable(vcc,gnd) from power supply to other subsystems that are connected to each other.  However the differential data path goes from one subsytem to another without taking care of power supply paths. By that way, we tried to seperate data and power path so that data will not be effected from ground potentials differents. But then we observed the problems(damage at differential driver part of pcb, only the sender part) that i mentioned above.

    I am confused about that; can there be such a potential differents between our nodes (30cm seperation) that will damage our sender part(differential driver that is set to drive mode) of the pcb? May the problem caused by other part of the design?

    Thank you,

     

  • 0 in reply to mahmut turk
    TI__Mastermind 28265 points

    Hello Mahturk,

    first I'd like to say that I am not entire certain I fully understand your exact system setup, although I do have some idea. If you could send me an overview schematic, it would significantly enhance my understanding.

    Also if you could measure the ground-potential-differences (GPD) between an active driver and some close as well as remote receiving nodes, it gives you a clearer picture about what voltage levels we are talking about. Typically the GPD between node grounds consists of some 80% of the 3rd harmonic of the mains frequency (50 or 60 Hz), and some minor percentage of the 5th and higher harmonics. years ago I wrote a technical article about this issue with the title: "When good grounds turn bad". I will search for it and send it to you.

    In my years with RS-485 I have experienced various power supply concepts to power multi-node busses. If you have multiple power supplies supporting multiple systems, the magnitude of GPD depends largely from which location within the electrical installation your supply receives its AC input. Secondly, the GPD also depends on during what noise environment the power supply cable are routed.

    One example: I had an applications lab whose supply was coming straight from a distribution panel. Parrel to the lab was an office area whose printer was powered from a different installation that also provided power to hundreds of other office equipment. So in the first case the lab received the cleanest power you can think off. The outside office area however was supplied with a severely noise AC. When connecting a computer in the lab to the outside printer in the office area using RS-485, we measured a ground potential difference of 35V. Note that the cable length was 2m (6ft) only).

    Even if you have clean supplies for different groups of subsystem, if one subsystem's supply cable run through noise environment, it might pick up noise transients from all kings of motors and generators (i.e. Fan, lighting tubes, solder stations, etc), whose peak magnitude can easily exceed the common-mode range of an RS-485 driver.

    I hope this information sheds a little light into the various possibilities of the generation of large common-mode voltages.

    Best Regards, Thomas

  • 0 in reply to Thomas Kugelstadt
    Intellectual 300 points

    Hello Thomas,

    Thanks for sharing your valuable experience with us. 

    I will send the details of our design to you via email.

    As i understand from your interpretation, we have a grounding problem. Since we have to repair/enhance the system, we have to take action immediately .  

    Now, we start to change our subsytem pcb's sender part with an isolated differential driver instead sn65hvd78. We will not change the receiver part.

    I hope this will make the system works without problem.

    I also want to know if we use fault protected differential line drivers instead of isolated drivers.

    Thanks, 

  • 0 in reply to mahmut turk
    TI__Mastermind 28265 points

    Hello Mahturk,

    as I suggested before, please try measuring the voltage difference between an active driver and a receiving node. These measurements can show that the voltage difference between grounds is too high, exceeding the common-mode range. They also allow to determine the magnitude of your differential bus signal.

    regards, Thomas