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.

TIDA-01630: Function of Y capacitance_1nF/2KV

Part Number: TIDA-01630
Other Parts Discussed in Thread: TIDA-010026

The shield wire of RS485 communication line connects the metal of D-sub.We find that if we want to pass the EFT test, Y capacitance(1nF/2KV) is very important.During the EFT test, the noise signal is coupled to the shield.Why does this capacitor make communication better? Why don't you worry about the noise from the shield wire going to the inside of the circuit board?

We can prove that this capacitor is the key to communication stability through a lot of experiments on different circuit boards, but we don't know the principle. I hope you can tell us why it is really reusable.

  • We have tested additionally on the RC connection between Earth and GND.

    These results can be seen in the Design Guide of the TIDA-010026 in chapter 3.2.2.4.1.

    We also see improvements using this RC filter, in that design also the choice of RS485 Termination was tested and also showed improvements can be seen with different terminations.

    Another thing to remember this discussion is also very focused on the use of Shielded cables, if unshielded cables are used other considerations also need to be taken into account.

    In general the capacitor is buffering the noise energy or potential difference charge between the Master and the Slave system to ensure that the common mode signals are within the range of the two communicating Transceivers of the Master and Slave even with a long Earth cable connecting the systems which can cause potential differences pending cable.

    If the Earth and GND signals are not referenced against each other the two GND planes of the master and slave can drift apart due to the noise(and energy/charge of the noise) and cause an Earth potential differences which can eventually lead to the communication to fail.

    Now adding them directly can create ground bounce or a high current part inside an IC or on the PCB, which the IC’s and PCB only can support if the design/IC is done to allow it, here considerations for Unshielded cables should to be taken into account.

    If the time constant of the RC filter is too low it can be seen that after a couple of noise pulses the communication fails as the capacitor keeps increasing the potential difference of the master and slave system.

    On top of these discussions there is also the topic of cable and connector quality, which we do not cover or discuss in our designs, this is also why typically a specific cable is recommended from the end equipment manufacture.

    We picked a good performing industrial cable and keep this as reference for all tests.

  • Glad to hear from you!

    Please see if I understand this correctly. The noise signal may lead to the voltage difference of the ground potential of the master-slave communication system, which leads to the communication failure beyond the common mode voltage range of the communication chip.

    1.Why don't we connect GND directly to shielding ground?

    2.Why the RC value of each driver manufacturer is different?The RC value next to Panasonic's encoder is 1MΩ//100nF;Beckhoff PLC at Ethernet port is 1MΩ//10nF;Servotronix encoder is 2MΩ//10nF.

    3.The manufacturer mentioned above will also use RC to connect GND with the CASE GND at other positions of the circuit board. Will this affect the RC connection near the encoder terminals?

  • As mentioned above “Now adding them directly can create ground bounce or a high current part inside an IC or on the PCB, which the IC’s and PCB only can support if the design/IC is done to allow it, here considerations for Unshielded cables should to be taken into account.”

    This typically means additional protection circuits which can handle high current and voltage pulses, protecting the IC's of overvoltage/overcurrent events. Which is why this typically it is wanted to be kept as noise in the EARTH cable/path by adding a higher impedance path into the PCB than the Earth path.

     

    Now Each customer has done extensive testing on their EFT solution and have found for their PCB/system solution a specific decoupling creating the best performance for used IC’s.

     This is also depending used cable, mechanical setup(in particular the EARTH connection scheme), termination choice and PCB layout which have different effects on the EMC performance of the system.

     

    For your point 3 this is correct, and you have to remember each cable going into the system typically has an EARTH connection which is decoupled for best performance for this sub-system while still keeping overall performance of the other sub-systems.

    The customer is testing all connections of the system and need to find a combined solution for all used cables connected into the system which can pass the IEC61000-4 tests per connection as a system and passing the End equipment requirements on the EMC/EMI standard.

  • “If the time constant of the RC filter is too low it can be seen that after a couple of noise pulses the communication fails as the capacitor keeps increasing the potential difference of the master and slave system.”

    What is the change of RC time constant when there is no a resistance or a capacitance between GND and shielding ground?R-->∞ ,C-->0

    Short circuit directly between GND and shield ground.?R-->0 ,C-->∞ ?

    What combination of capacitance and resistance can make the RC time constant the lowest?

    Resistance affects DC common mode voltage,

    Capacitance influences common mode voltage.

    In the EFT test, should we increase the capacitance to make GND and shield ground short circuit on the AC signal, so as to minimize the AC common mode voltage.

  • There is always a parasitic resistance of the system which will over time discharge the capacitor without the resistor the time constant is not under your control and you can not know how long it will take. Potentially here a second noise pulse can happen before the first has been discharged.

    With the short circuit idea, as mentioned before need to ensure your PCB can tolerate the noise with protection circuits or decoupling of the sub-systems, this is your choice as a designer to make how you protect the circuit.

    What time constant is needed depends on system parameters which is why this is best to test the system at different noise levels and see what time constant/ decoupling circuit fits your system. As this also included mechanical setup a pure theoretical approach is difficult as you need to have a model for your full system including parasitic influence of the system. 

    For the actual configuration I would recommend you to do extensive testing of different decoupling options and this can reveal what decoupling fit your system test setup best.