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.

DP83848I issue with Gigabit Switch

Other Parts Discussed in Thread: DP83848I

I have an issue that hopefully someone can help me out with. I have a DP83848I chip connected to an ARM STM32F407. Everything works fine until I turn off my circuit with the ethernet cable still plugged in. After about 30 seconds the link light on the switches shows a link and the port starts getting hammered with sent bytes.

I have tried 2 different gigabit switches Asus and Netgear and both do the same thing. The netgear is a prosafe + so I was able to connect and see that the port gets flooded with sent bytes. Also what happens is the other ports get taken down because the switch gets overloaded.

I then tried 2 other switches 10/100 and everything works normal. No link light when my board is turned off and no flooding of the port.

Any idea what might be happening or what I should be looking for.

Thanks

  • Rob,

    Could you provide the specific part numbers for the Asus and Netgear switches that you are using? 

    Also, could you provide the schematics for your board?  If you are concerned about posting them to the forum, let me know and we'll work out an alternative.

    Patrick

  • Thanks for the quick reply.

    The Netgear switch is a GS105E and the Asus switch is GX01081. 

    One other thing I noticed, I found the design and layout tech note and it recommended voiding the planes under the connector which I have not done yet.

    Below is the Ethernet schematic.

  • Turns out one of the switches that was working normal is a gigabit switch as well. Its a netgear GS116

    The other switch that was working fine is a Linksys EZXS55W.

  • And a correction, the Asus is GXD1081

  • Rob,

    We do not have any of the exact Gigabit Ethernet switches that you specified in our lab, but we have tested using other switches and have not been able to re-create your results using our boards.  Do you have any other discrete boards, e.g. an EVM or another of your designs, that you could test with those switches as well?

    For the most part, your schematic looks good.  I do not see any major issues.  Relative to the noise, I have a few notes and questions:

    1. As you noted, we do recommend voiding the planes beneath the magnetics.
    2. We also recommend using a Bob Smith termination on the cable side of the magnetics.  The chosen connector with integrated magnetics does not appear to include a Bob Smith termination.  Could you please confirm?
    3. We often include capacitive connections between the shield or earth ground and the board ground in our designs?  I do not see any such connections in your design.  Could you confirm if these connections exist elsewhere in your design (but are not shown in the portion of the schematic you shared)?

    Patrick

  • I actually built another prototype yesterday to see if it was a bad board or parts. I even populated it with a jack with the bob-smith termination (Same series jack but MB not MA in the part number) but still had the same issues. My earth ground is separate from the board ground. I have it terminated to building ground. I tried removing this as well as tieing it to board ground with all the same results. 

    I will try the capacitive connection between the two grounds. Also i will try a few other types of switches to see how common this issue is. 

    Ill let you know how i make out. 

  • Just to give you an update. I ran new boards and removed the Ground and power planes underneath the RJ45 connector. That seemed to fix the issue. It's Interesting how sensitive that is.

  • Rob,

    Thank you for feeding this back.  I have not seen this functionality manifest itself in that way before.  Nor is the problem behavior immediately intuitive to me based on the symptoms you initially described.  I will be on the lookout for this in the future. 

    Thanks again for bringing this to light and closing the loop on what you discovered.  I appreciate your diligence.

    Patrick