• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 4 answers
  • Subscribers 63 subscribers
  • Views 631 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TPS23861: Isolation between VDD and VPWR / DGND and AGND needed?

Marcel Jahnke
Prodigy 50 points
Part Number: TPS23861

Hello experts,

i have developed a POE+ switch with the TPS23861. All functions are ok, but we have made a 2250VDC test (shortened DATA Lines which include the POE PWRD voltage) measured against GND, which doesn't passed successfully.

So i asked me, if the VDD and VPWR (and DGND and AGND) of the TPS23861 are isolated? We use an external AC/DC power supply, which have 48 VDC output. After it, the 3,3VDC (VDD) is generated.
AGND and DGND are connected together.

I am not sure how the 2250VDC test is done correctly (relating on IEEE 802.3). Is the test measured from the Data lines to GND or to Earth? Perhaps you can help me.

I think about an isolation between the 48VDC and the 3,3VDC if it needed for the IEEE802.3 high voltage test..  For this the VPWR and the KSen pins belong to AGND and VDD to the digital DGND.

Thank you for your help.

Marcel

  • 0
    TI__Mastermind 41780 points

    Hi Marcel,

    For the PSE, the isolation is on the 48V rail (in this case the adapter). Is the adapter isolated and rated for the expected high pot testing you're doing?

    The 3.3V does not need to be isolated and is always non-isolated.

  • 0 in reply to Darwin Fernandez
    Prodigy 50 points

    Hi Darwin,

    thank you for your answer. Yes the Ethernet adapter is rated up to 2250VDC, but the 48VDC pin of the adapter goes directly on the board. There is no galvanic isolation of the 48VDC and i don't find an Ethernet Poe adapter, that has there an isolation.
    So the Hi Pot test goes directly (48 VDC Pin and its GND) on the board and to the TPS23861 (VPWR pin and Drain pin).


    It could be that the power regulator on my board (48VDC -> 5 VDC -> 3VDC) or the pcb is not enough isolated.
    So a solution could be to use an isolated DC/DC converter (48VDC -> 5VDC), but anyway then the TPS23861 has at the Drain pin up to 2250 VDC hi pot test voltage.
    Does these TPS23861 pins tolerate this hi pot voltage?

    Marcel

  • 0 in reply to Marcel Jahnke
    TI__Mastermind 41780 points

    Hi Marcel,

    I don't believe the hi pot on the TPS23861 is referenced to any of the ground pins of the IC so I don't expect the IC to have issues. I can see that there would be issues if the isolation components like the data transformer cannot withstand this voltage and so there is some current flows through the isolation barrier.

    In short, the ground pins of the TPS23861 are not isolated from each other and I have yet to see an isolated converter used for the 3.3V rail from the 48V input.

    Can you send a quick diagram on how you're measuring the hi pot test and the setup? Thanks!

  • 0 in reply to Darwin Fernandez
    Prodigy 50 points

    Hi Darwin,

    i have attached for you a short diagram, which shows the setup.
    Do you know, if it is right to measure against GND or should it be Chassis Shield?

    I think the problem could be the non isolated 48VDC. All power regulators or ics on the board which have the 48VDC connected, have to be proofed at 2250VDC.

    The POE+ power (48VDC) rail has no transformer in the RJ45 Ethernet jacket. It also could be that my pcb has an isolation problem, i will examine it. Perhaps i can see an electrical spark.

    If i use an isolated 3,3V rail from the 48VDC input, the 48 VDC (VPWR) is still not isolated.

    Thank you.

    Marcel

  • 0 in reply to Marcel Jahnke
    TI__Mastermind 41780 points

    Thanks for clarifying this block diagram helps. This should NOT be tested in reference to GND of the PSE. What needs to be tested is the isolation of the 48V rail. So the return of the generator needs to be connected to the bob smith termination of your PSE. As a reference, it is TP11 of the TPS23861EVM.

    Thanks!

  • 0 in reply to Darwin Fernandez
    Prodigy 50 points

    Hi Darwin,

    thank you for your help. In the reference (TPS23861EVM), TP11 is the Shield GND. So for my circuit this is a good news.


    But what i don't understand is, why the POE Ethernet jackets have transformers, if the hi pot measurement is against Shield GND (Chassis).
    I thought we have to protect the system circuit and this has the GND as signal.


    So for my understanding, if i use POE+ then i have not the 2250VDC protection for my system circuit, because of the 48VDC, which bypass the transformers.
    If i use only a standard RJ45 jacket (without POE) then i have the 2250VDC protection against GND also.

    Do you know, that this is usually in the industry?

    Thanks.

    Marcel

  • 0 in reply to Marcel Jahnke
    TI__Mastermind 41780 points

    Hi Marcel,

    The data transformers allow the differential ethernet data to pass through while keeping the common mode (DC 48V voltage) away from the PHY.The data transformers also isolates the high voltage front end to the lower voltage secondary side where there is usually physical access to the end user (like a USB connector).

    All PoE designs in the industry will have the data transformer in the design.