• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 63 subscribers
  • Views 367 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: Quesition about the PSE(TPS23816) protection mechanism

Hsinwen Wang
Prodigy 10 points
Part Number: TPS23861

Hi Sir,

Good day, this is Hsinwen who is a hardware engineer.

i have a quesition about the PSE protection mechanism.

In my test condition, there're two PoE PSE devices, both using the TPS23861 solution.

May i know if i connect these two devices directly through ethernet cable, is there any protection mechanism for the TPS23861?

Thank you.

  • +1
    TI__Genius 15275 points

    Hi Hsinwen,

    There are a lot of factors at play here but in general PoE goes through a handshake process which is IEEE802.3 defined. This standard defines the handshake between the PSE and PD which includes detection, classification, and inrush. The PSE starts this procedure by sending a detection current signal to calculate the detect resistance based on measured voltage. This detection must fall in the range of 19kohm-26kohm with 25kohm being the typical value. If the PSE does not have a valid detection it will continue to repeat this process until a valid resistance is found. 

    Therefore connecting PSE to PSE there will be no valid detection and the recieving PSE will have open FET protecting the device. However, if you operate the PSE in manual mode and disable detection and classification and manually turn on power, the PSE will send power and you would have to disable short circuit faults. The ptoection on the reciving PSE will once again be the open FET on the drain. 

    Very Respectfully,

    Brandon Beader

  • 0 in reply to Brandon Beader
    Prodigy 10 points

    Hi Brandon,

    Tkanks for your reply, and sorry that i also have two more questions to ask here. List my questions below.

    1. After i rised the question, i did this experiment and also measured the waveform. May i know if the measurement method is correct? And what's the different between these 3 waveforms? (Is the PSE trying to send a different sense current signal?)

    The picture below is the measurement point.

    The pictures below are the measurement results.And the signal is continuously change.(1->2->3->1 repeat).

    2. Have you heard that TPS23861 PSE device burning the non-PoE devices? And waht's the root cause?

    Thank you.

    Hisn Wen

  • 0 in reply to Hsinwen Wang
    TI__Genius 15275 points

    Hi Hsinwen,

    Please see my comments and follow up questions below.

    1. What is your configuration for connecting the two PSE ports together? Are they from the same system therefore have the same Vpwr or different system? What mode is each PSE operating in, such as manal, auto, or semi-auto?

    I am assuming from your waveforms that you are operating in auto mode for both channels and the PSE's are attempting to pulse their detection current signals and having some form of interference with each other. Sinc neither PSE will return a valid detection, neither PSE will turn on and will continue to cycle through detection.

    2. We have heard of PSE's burning non-PoE devices and will really depend on the architecure of the non-PoE system you are connectiing to. If you are connecting directly to Ethernet PHY which does not support PoE then there will be no way to filter out PoE DC voltages and will therefore burn up the end equipment. All of our EVM's and reference designs you will see data transformers to protect ethernet PHY. 

    Very Respectfully,

    Brandon Beader 

  • +1 in reply to Brandon Beader
    Prodigy 10 points

    Hi Brandon,

    I appreciate your reply. And sorry that i did'n discribe it clearly. Answers to my follow-up questions are below.

    1. The experiment I ran were using the same PSE device with the same VPWR and both of them were operated in auto mode.

    2. "All of our EVM's and reference designs you will see data transformers to protect ethernet PHY."
    Are the transformers you're talking about placed on TPS23861'EVM?

    I'd liek to add that our product is a conference soundbar which built in ethernet port with PSE capability. So If we follow TI's reference design and EVK to place the transformer, could it avoide burning the non-Poe devices?
    (Actually, our design have followed TI's EVM and reference design for transformer placement.)

    Thank you.

    Hsinwen

  • +1 in reply to Hsinwen Wang
    TI__Genius 15275 points

    Hi Hsinwen,

    1. Okay I understand the setup now, since both these devices are operating in auto mode, they are both pulsing detection signals and not seeing valid detection. Detection will continue to cycle until valid detection occurs in auto mode. This why these waveforms you have sent seem different and and are cycling between different waves. This is due to the detection cycles infereing with each other.

    2. Yes TPS23861EVM-612 has these data transformers that are built into the RJ45 port (J7-9 and J19-21). Our EVM's offer a data only port for customers to polug their data into EVM to ensure system works.

    Is your goal to send power out of the conference soundcar? The data trasnformer would be on the reciving side to protect ethernet phy. It will also depend on what operating mode you are in. If you use TPS23861 in auto mode, it must go through detection and classification before power is applied to the port, therefore when you connect a non-PoE device it will not pass detection and classification. What we have seen is non-PoE devices burning up due to detection signals from PSE, only resolution here would be add data transformer on the recieving side which I am assuming you have no control over. 

    Very Respectfully,

    Brandon Beader

  • 0 in reply to Brandon Beader
    Prodigy 10 points

    Hi Brandon,

    Totally understood, thank you.

    Hsinwen