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: Getting STRT1 event in SEMI AUTO mode but not in AUTO mode- Why?

Part Number: TPS23861
Other Parts Discussed in Thread: LM5072,

Hi. I am developing POE switch and at the moment I am testing it with different PDs. 

I am implementing power budgeting so I need to use semi auto mode.

Settings of TPS23861:

  • A bit set to 1(by internal pull up). 
  • SEMI AUTO MODE
  • Detection and classification enabled on all ports
  • Two-event classification enabled
  • tSTART=0b10; approximately 120 ms

My problem:

One particular PD which works fine with other PSE equipment do not start with TPS23861 in SEMI AUTO mode.  It power ups fine in AUTO mode.

This is the waveform of detection and classification:

Whats happening in I2C bus:

Port  1 class and detection(0xC) register shows: 0x14, which means Class 1 and Resistance valid.

Then I set POEP1 bit to 0, ICUT Port1 to 1 which gives 110 mA current limit.

Then I enable port with PWON1 bit.

Approximately 400ms after PWON1 bit is set Class/Detection register(0xC) of port 1 changes to 0x3, which means Class Unknown and Resistance too low.

This 0x3 value persists for approximately 1 second, then for about 40ms I read register(0xC) value 0x4, which is Resistance valid and Class unknown.

Then 0xC register shows 0x14 again.

Also when I read Start/ILIM Event Register(0x8) register it shows STRT1 bit active, which indicates that a start fault occurred at port turn on. This may be caused by:

1. Overcurrent (foldback) condition at the end of tSTART.
2. Detect or classification fault following a pushbutton PWON command in Semi-Auto or Manual Mode.
3. Overcurrent or class mismatch on second finger in Semi-Auto or Manual Mode.

From all information above it seems like for unknown reasons Class is changed in the PD which stops port from being Enabled. 

Why this PD is powered in Auto mode?? Any ideas? I would really appreciate help!

Karlis

  • Hi Karlis,

    That's very interesting behavior. Can you capture the port current in semi auto mode and auto mode? I want to verify if inrush current exceeds the inrush limit. In addition, did you see consistent detection and classification result (detection valid and class 1) before issuing power on command through 0x19? Thanks.

    Best regards,

    Penny

  • Thanks for reply Penny.

    I read the current and class/detect registers every 11.5ms and these are the results:

    Current in Auto mode:

    1. 1 measurement of 14.2 mA, <- this might be classification current, but in that case it is not legit because it falls in between Class 1 and 2 currents.
    2. 55 ms of 0 mA, 103ms of 1.46 mA,
    3. fluctuating 40-48 mA of load current when PD is fully turned on

    Class/detect r in Auto mode:

    1. 0x6 -open circuit,
    2. 1 reading of 0x4 - which would mean no classification has still occured but there is a legit detection resistance
    3. consistent readings of 0x14 - class 1 and legit detection resistance - 

    Current in Semi auto mode:

    1. 368ms of 14.7 mA
    2. 862ms of 0 mA
    3. 414ms of 14.16 mA
    4. 1000mS of 0 mA
    5. 414ms of 14.16 mA
    6. 966 ms of 0 mA
    7. 414ms of 12.87 mA
    8. 1000 ms of 0 mA
    9. 437ms of 12.94 mA

    Class/Detect in Semi auto mode:

    •  while PD is not connected it shows 0x6 - Open circuit  - that's ok. Now I connect PD
    • 2 measurements of 0x4 - resistance valid, but no classification has taken place - seems ok.
    • 425 ms of 0x14 - seems ok, class 1 and resistance valid
    • 862 ms of 0x3 - resistance too low.
    • 1 measurement of 0x4.
    • 437 ms of 0x24- class 2 and resistance valid.
    • 138 ms of 0x4
    • 839 ms of 0x3
    • 2 measurements of 0x4
    • 448 ms of 0x14
    • then again 0x4 and 0x3. you get the idea...

    For me it seems like PD has not legit classification current and that's why Class is being changed all the time in TPS register, thus port does not turn on.

    But then why it is powered on in Auto mode? 

  • Hi Kalis,

    In auto mode, the port perform detection and classification once the PD is connected and power on the port right after it detects a valid PoE PD(valid detection and classification). That's why you only saw one cycle of detection and classification in auto mode before the port is turned on. 

    In semi auto mode, the port keeps doing detection and classification after the PD is connected and before the power on command is issued. Once there's a host command to turn on the port through 0x19, the PSE finishes the current cycle of detection and classification and turns on the port. That's why you usually see more than one cycle of detection and classification before the port is turned on in semi auto mode. 

    From your measurement and observations, looks like the PD has issues present consistent detection and classification during continuous detection and classification. In auto mode, the device takes care of turn on automatically as long as it sees a valid detection and classification. While in semi auto mode, it relies on the host. It is possible that while the host sees a valid detection and classification then the host issues power on command, and the PD presents invalid detection and classification in the last cycle and the PSE refuses to turn on and reports start fault. 

    Does that make sense to you? My suggestion is to do a sifos test or get an EA certificate with your TPS23861 board and as long as it passes, your PSE system is compliant and it should work fine with compliant PDs and you don't need to worry about the noncompliant PDs. Thanks.

    Best regards,

    Penny

  • Hi Penny. Thanks for thorough reply.

    Yea, I kinda thought the same way, but it still seems strange that company with 1 billion $ revenue would make non compliant PD.

    I took that PD apart and inside I found LM5072 Integrated 100V Power Over Ethernet PD Interface and PWM Controller with Aux Support.

    That is Texas Instruments part. Any comments?

  • Hi Karlis,

    LM5072 is a TI part as well. The compliance relies on system design not on the IC itself. 

    Can you test the PD with TPS23861EVM-612 and use PI commander GUI to see if you can reproduce the same issue on the EVM? Thanks.

    Best regards,

    Penny 

  • Well ok. Next week I will know if I will be able to get my hands on the EVM.

  • Hi again.

    So we decided to buy the necessary things for legitimate test for this PD.

    Can you please correct me if I am wrong?

    We need to buy:

    1. TPS23861EVM-612 Evaluation Module
    2. USB-TO-GPIO: USB Interface Adapter EVM

    What we need to do:

    1. connect TPS23861EVM-612 to power supply.
    2.  connect TPS2361EVM-612 to PC with USB-TO-GPIO adapter and use PI Commander GUI to set it in Semi-auto mode.

    Is this setup enough for semi-auto mode test or do we need to install CSS and use MSP430G2 LaunchPad Development kit MSP-EXP430G2 ?

    I downloaded the TPS23861EVM-612: Auto-Mode Evaluation Module for TPS23861 but there is nothing much written about 

  • Hi Karlis,

    TPS23861EVM-612 and USB-TO-GPIO are the right tools you need. MSP430 launchpad is another way of running the TPS23861 in semi auto mode besides the PI commander GUI. We provided MSP430 reference code to help customers get started with software. Please let me know how it goes. 

    Thanks,

    Penny