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.

Gigabit Ethernet connectivity with the AM389x processors

What is the best way to interface the AM389x to another processor via Gigabit Ethernet? The other processor has GMII interface, can the 2 be connected directly via GMII to GMII? Or must there be Phys in between?

Charles

  • Charles,

    The AM389x does not have an integrated Ethernet PHY, so an external PHY is required.  The EVM is a good reference.

    Regards,
    Marc

  • Marc,

    Thanks, I guess the real question is, if we are connecting the Gig E to another processor can you connect the GMII ports of each processor together? or do you need to go from

    GMII to PHY --- then to the PHY to GMII of the next processor? These are both on the same board, so this is not going off board or across cards..Not sure if this is possible.. or would work.

    Charles

  • Charles,

    Thanks for the additional details - I was thinking that the connection would be outside of your system.  In this case GMII can be cross connected (without using PHYs):

    • TXD goes to RXD
    • TXEN goes to RXDV
    • RXER is tied low
    • COLL is tied low
    • CRS is tied low
    • Need to run in gig mode fullduplex
     
    Regards,
    Marc
  • Thanks Marc, This was very helpful

    Charles

  • Marc,

    Can you verify that this is GMII to GMII speeds supported this way(and NOT just MII to MII)? Also can auto-negotiation be done with the connection done this way?

     

    Charles

  • Hi Marc,

    1) What about GMTCLK, RXCLK and TXCLK signals?

    2) Could CRS be shorted to TXEN or it has to be tied LOW? I found TI's datasheet about direct connections in the document "Inter-DM6467 Communications Using Direct GMII Connections".

    2480.Inter-DM6467 Communications Using Direct GMII Connections E2E .pdf

    Regards,

    Adam

  • There are other pins that may need to be addressed:

     

    Here is my guess:

     

    MPU 1

    Direction

    MPU 2

    Notes

    EMAC Module

     

     

     

    EMAC_TXCLK

    I

    2.5/25 MHz

    10/100 Common on both

    EMAC_GMTCLK

    O

    open

    1000

    EMAC_TXD[0-7]

    O

    EMAC_RXD[0-7]

    Crossed data

    EMAC_TXEN

    O

    EMAC_RXDV

     

    EMAC_COL

    I (GND)

    GND

    Not sure of polarity

    EMAC_CRS

    I (GND)

    GND

    Not sure of polarity

    EMAC_RXCLK

    I

    2.5/25/125  MHz

    10/100/1000 Common on both

    EMAC_RXD[0-7]

    I

    EMAC_TXD[0-7]

    Crossed data

    EMAC_RXDV

    I

    EMAC_TXEN

     

    EMAC_RXER

    I (GND)

    GND

    Not sure of polarity

     

     

     

     

    MDIO module

     

     

     

    MDIO_MCLK

    O

    Unused

    Unsure

    MDIO_MDIO

    I/O

    Unused

    Unsure

     

    Ground all _COL, _CRS, RXER signals on both MPUs

  • Can you help resolve these questions:

     

    1)      Is the phy in the new AM389x equivalent to that used in the DM6467?

    2)      What are the details for using the EMACs for MCU to MCU Communication?

    3)      Does EMAC_CRS get tied to GND or does it tie to the same MCUs RXDV?

     

     

    MPU 1

    Direction

    MPU 2

    Notes

    EMAC Module

     

     

     

    EMAC_TXCLK

    I

    Unused on both

    EMAC_GMTCLK

    O

    EMAC_RXCLK

    Cross Clocks

    EMAC_TXD[0-7]

    O

    EMAC_RXD[0-7]

    Crossed data

    EMAC_TXEN

    O

    EMAC_RXDV

    Cross ETn/RVa

    EMAC_COL

    I (GND)

    Ground Both

    EMAC_CRS

    I (???)

    GND or EMAC_RXDV_1

    Please see attached diagram

    EMAC_RXCLK

    I

    EMAC_GMTCLK

    Cross Clocks

    EMAC_RXD[0-7]

    I

    EMAC_TXD[0-7]

    Crossed data

    EMAC_RXDV

    I

    EMAC_TXEN

    Cross ETn/RVa

    EMAC_RXER

    I (GND)

    Ground Both

     

     

     

     

    MDIO module

     

     

     

    MDIO_MCLK

    O

    Unused

    MDIO_MDIO

    I/O

    Unused

     

     

     

     

     

     

     

     

     

     

     

     

     

     Thanks,

     

    Rinzai Bell