• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 33 subscribers
  • Views 1918 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.

DP83848J connected to KSZ8895 Switch by capacitive coupling

Yves Willener
Prodigy 100 points
Other Parts Discussed in Thread: DP83848J, DP83848K

Hello

 

My goal is to connect an Ethernet PHY (DP83848J) to one of the ports of a KSZ8895 switch. Both components are on the same PCB. Due to this fact I assume I do not need the two transformers (e.g. H1102NL) and I can connect them directly via capacitive coupling. This approach is described in the AN-1519 application note from TI. Have you got any experience with this solution? I am wondering if that is a well-known way of connecting two Ethernet devices and whether there are some limitations which I have to follow?

 

For example:

What is the maximum trace length of the Ethernet signals when they are coupled by capacities?

 

Are there some best practice guidelines for the layout design especially about the signal crossover? As far as I know crossing of the signal traces should be avoided. But in case of the capacitive coupling it is necessary because TD+ and RD+ is connected together.

 

Would it reduce the risk if I would use the transformers anyway? (PHY - Transformator - Crossover - Transformator - Switch)

Or would that make it even worse because transformators decrease the signal quality?

 

Thank you for your answers.

 

Yves Willener

  • 0
    TI__Mastermind 31600 points
    Yves,

    Capacitive coupling should be OK from the DP83848 perspective. There is not a strict maximum trace length. Given that both devices will be on the same PCB, I would not expect an issue. What distance do you expect between the two devices?

    Have you compared the datasheets for the DP83848K and the KSZ8895 to confirm that there is a crossover? It may be that the transmit and receive pins line up.

    Patrick
  • 0 in reply to Patrick OFarrell
    Prodigy 100 points

    Thank you Patrick for your answer.

    At the moment it looks like the traces are somewhere between 10cm and 20cm. I think that is a bit too long and I will rearrange a few components so that I get to a trance length around 5-10cm.

    As far as a I can see both devices support auto-MIDX or MID-MIDX. In theory, a crossover is not needed. Do you agree with that?

    What do you think about the Ethernet transformer? I am not an expert but I can see three advantages of such a transformer:

    1. Impedance matching
    2. Noise filtering: Common mode rejection
    3. DC isolation: no ground offset issues

    If the two devices are on the same board none of them is relevant and the transformer can be omitted, right?

    Yves

  • 0 in reply to Yves Willener
    TI__Mastermind 31600 points
    Yves,

    Yes, I agree. I do not believe a cross over is needed. You should be able to connect the transmit pins of the PHY to the receive pins of the switch and vice versa.

    Note that the transformerless application note recommends operating in 100M full-duplex. The best case scenario would be to configure both the PHY and the switch in this configuration. Is this possible in your application?

    I agree with your assessment of the transformer. It should not be necessary given that both devices are on the same board.

    Patrick
  • 0 in reply to Patrick OFarrell
    Prodigy 100 points
    Yes, both, the SWITCH and the PHY are set to auto-negotiation and they advertise in a way that they use 100M and full-Duplex as the highest priority.

    Thanks for your help.

    Yves