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.

TPS23756: TPS23756 Emissions

Part Number: TPS23756

We opted to use the TI TPS23756 to be compliant with the standard 802.3at PoE+ (30W).

We used as reference the design of the evaluation board TPS23756EVM (http://www.ti.com/lit/ug/slvu329a/slvu329a.pdf)

Paragraph 6.2 page 9 of this user guide are reported the emissions measurements of the evaluation board. 

 

We measured the emissions on the board developed by us obtaining very different results then the expected ones, so we brought to the test laboratory the development board to better understand the origin of the emissions and we found that the evaluation board has very similar issues to the ones encountered on our own developed board, such results differ considerably from the ones outlined in the user guide.

 

We tried to limit the polarising voltage Vc of the gates at 12V. We still have to understand the reason why, in our design, Vc is normally around 15-17V against 11-12V measured on the EVM. 

As expected, after having measured EMC on the EVM, decreasing the gate voltage helps in reducing the emissions but it does not solve the problem.

 

The conclusion is that the TPS23756EVM has some failures at circuit level which causes the issues we’ve seen on the emissions.

We ask your support to verify the status of this board and give us the instructions needed on the fix to be implemented. 

  • Hello Stefano,

    We will get an answer to you, shortly.

    Thanks,
    Thomas Amlee
  • Hi Stefano,

    Could you send me your schematic & layout? I will take a look & let you know.

    Thanks,
    Thomas Amlee

  • Stefano,

    Additionally, be sure that your board & test environment follows the guidelines described in this app note:
    www.ti.com/.../slua469.pdf

    - Thomas A.
  • The schematic is exactly equal to the EMV.
    About the layout we already verified the need to change it changing the opto-isolator position to prevent feedback signal goes across a critical area.
    Nonetheless, the fact that the evaluation board has similar conducted emission issues  leads us to conclude that even revising the layout would not solve the problem and we need to go through the analysis pf the suggested circuit.

  • Hi Stefano,

    I suspect the testing setup was different in the UG. Do you have CE data from your testing? Are you using a resistive load? Is the end equipment completely floating with only the PoE connection (or are there other connections like earth ground connection somewhere on the board?
  • Hi Darwin,

    the measures on TPS23576EVM were done with resistive loads and the equipment completely floating. We have already sent the measures to Thomas Amlee. Let us know how to share with you these measures.
  • Hello Stefano,

    It appears I have not received the your measurements. I will direct message you, so you can share your email with us.

    Thanks,
    Thomas A.
  • Hello Darwin,

    effectively in our M27 board we have a ferrite between GND and the shield of the ethernet connectors. Also the entire system metallic enclosure is connected to GND.
    Is this a problem?
    Could we have an added noise current loop using an ethernet cable Cat 6 S/FTP?
    Can be useful to insert a CM Choke on Vin ?

    But we believe this still doesn't justify the conducted emission on the EVM where there is not GND to Earth connection.
    We are however verifying to have no undesired connections to earth also in this environment.

    Do you have some other suggestions to improve the conducted emission related to the suggested schematic?

    For example, do you think useful to increase the values of R2, C2 and R10 on input MOSFET of EVM board to reduce dv/dt?
    Could you suggest some reasonable values?
    Take in account that in our environment we use less than 20W, so we have no problems in term of maximum power and efficiency,

    Thank you and best regards,

    Stefano

  • Hello Stefano,

    By connecting –Vout  (GND on M27 board) to the system enclosure, you are connecting the negative output to the test system’s ground. Have you tried testing with a floating load? If you are intending to design for an earth grounded load, you need to use a common mode choke instead of ferrite beads on the input (L30 & L35 on M27). 

    The emission graph Davide sent looks similar to figure 1-9 of app note "slua469" where ferrite beads are used with an earthed output. If this is occurring on both your board & on our EVM, then that means you are testing with a load grounded to the test system’s ground.

    By Ethernet connector shield, are you referring to the Bob smith ground plane the Ethernet transformers are terminated to?

    As for schematic recommendations, you can increase those gate resistors & capacitors, but that would only affect high frequency emissions. To address the bigger issue of low frequency emissions, you should use the common mode choke. Also, you can increase isolation capacitance (C33 on M27 board) or the inductance on the low pass filter (L28 on M27) to improve noise performance, slightly.

    Thanks,

    Thomas Amlee

  • Hello Thomas,

    it seems reasonable also if the system enclosure is not connected to earth and the measures on our board were done on both Ethernet and AUX power input without any earth grounded load.

    We have done also a measure on  TPS23576EVM using AUX power input and a isolated resistive load. Also in this case we found the same problems on the conducted emissions. In this case we haven't any earth grounded load.

    Since we need to certify also this configuration we'd like to have a suggestion to fix it.
    Do you think necessary to use a common mode choke also in this case?
    Do you have done the conducted emission measures on AUX Power input?

    About your suggestion to increase the L28 inductance and the insulation capacitance C33, could you suggest us some reasonable values and the maximum acceptable?

    Thanks,

    Stefano

  • Hello Stefano,

    Yes, it would be necessary to add a common mode choke.
    Conducted emission measurements are performed through the PoE port, however the same ferrite beads and low pass filters are present on the AUX power input. A common mode choke can be added there as well.

    As for the inductance & capacitance values, I would first focus on the common mode choke. You can experiment with a higher inductance values (2x for example) to move the cutoff frequency closer to common mode, however it will not make a significant difference compared to the common mode choke.

    Thanks,
    Thomas Amlee
  • Hello Thomas,
    I have got a feedback from our EMC experts who have done the measures and they confirm that on TPS23576EVM all the measures were done with isolated loads.
    Also on this board the conducted emissions are strong and over the limit with a completely isolated load.

    We need to consider that this strong interference on the board will not be sufficiently reduced by the common mode choke. It can even increase the problem with radiated emission. We need to dig further into this detail.

    Before to proceed with our redesign we need to be sure to have no problem with the reference schematic we took as reference.
    So, could you confirm to have no problem on the conducted emission on your evaluation board? Could you share with us your test environment and the report of these measures to submit to our experts?

    Thank you and best regards,

    Stefano

  • Hello Stefano,

    Unfortunately, I was not involved with the testing of the TPS23756 in the EVM user's guide. However, those emissions tests performed were in accordance with the testing guidelines and environment recommended in SLUA469. So please review this app note & let me know if your setup is similar to the setup in the app note.

    Thanks,

    Thomas A.

  • Hello Thomas,

    TPS23576 evaluation board fails the conducted emission test with both the PoE and AUX power inputs.

    Our test setup was constructed per EN55022 and the tests were performed in two different labs.

    For that reason we are asking you to find the test reports of the TPS23576EVM.

    Thanks,

    Stefano

  • Hello Stefano,

    Unfortunately, we are unable to find the test reports.

    Update: customer issue was addressed offline.

    Thanks,
    Thomas A.