• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 33 subscribers
  • Views 963 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.

DP83848C and Conducted Emissions Test

Allison McKenzie
Prodigy 90 points
Other Parts Discussed in Thread: DP83848C

We are using a DP83848C as a 10/100Mb ethernet Phy with an STM32 as the MAC and wurth 7499011121 magnetics.  The circuit is laid out the same as the App Note, and the component values have been checked. 

The unit is failing Conducted Emissions testing in 100Mb/s mode.  There is a constant load of chatter on the Tx and Rx lines (that looks like a valid 100Mb/s signal), even when we are not transmitting anything.

Is there a way to turn this off, or slow it down?

Thanks

  • 0
    TI__Mastermind 31600 points

    Allison,

    We have some application notes and reference designs that might be useful.  As a start, I thought I would bring these to your attention:

    To answer your main question, there are ways to affect the signaling in 100Mb/s mode, but before I make any recommendations, I would like to understand your implementation a bit better.  To that end, I have some questions:

    1. Which conducted emissions specification are you attempting to meet?  What failing point(s) are you seeing (frequencies and magnitude of failure)? 
    2. What is the MII functional mode, MII or RMII?  What is the source of the X1 reference clock?
    3. How is the RJ-45 connector connected to the 3.3V supply and to ground?  Is there a chassis ground separate from the main board ground?
    4. What is the length and type of cable connected to the RJ-45 during testing? 

    The Wurth 7499011121 magnetics looks like a good choice.  I am not very familiar with this particular component, but the specifications are certainly compatible with the DP83848. 

    Patrick

  • 0 in reply to Patrick OFarrell
    Prodigy 90 points

    Hi Patrick,

    Thanks for your reply and I am sorry my original description was a bit sketchy.

    I and one of my colleagues have been through the application notes and tried everything we can think of with no luck.

    The test is Conducted Emissions to CISPR22, I have copied the results in below.  You can see that we are only a couple of dB over in the 10 to 30MHz region.  We are using about 1m of standard CAT5.   

     

    The system operates in RMII mode with a FXO-HC736R-50 crystal.   The circuit is shown below.

    Any help you can offer will be greatly appreciated.  One thing we thought it might be is the lack of connection between the secondary of the transformer and ground.  The board is placed to provide no connection at all.

  • 0 in reply to Allison McKenzie
    TI__Mastermind 31600 points

    Allison,

    In our testing, the spikes in the 0-30MHz frequency range are typically well below the spec.  We do see distinct peaks and valleys for the energy in that range though.  By comparison, what you are seeing looks more like broadband noise. 

    I suspect that the problem may be caused by the lack of connections for the transformer secondary and the connector shield.  Is there any way you could wire in a connection from those connector pins to ground?  Obviously, it would not be an ideal connection, but it might show enough improvement to validate my suspicion.  Given that you are only missing the spec by a couple dB, it might even be enough to clean up those spikes that are out of spec. 

    On a related note, from the information that you sent, it appears that you have voided the power and ground planes beneath the magnetics (consistent with our recommendations).  Could you please confirm?

    Patrick

  • 0 in reply to Patrick OFarrell
    Prodigy 90 points

    Jo Patrick,

    We have vopided the planes under the magnetics.  I appreciate your help.  We will try the grounding idea.

    Thanks!

    Allison