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.

AM335x ICE V2 schematic

Other Parts Discussed in Thread: TLK110, CDCE913

Hi,

Our custom AM355x board is close to the AM335x ICE V2 board. I'm now implementing Ethernet on our board. By chance I have seen this E2E thread: http://e2e.ti.com/support/interface/industrial_interface/f/142/t/206970.aspx

And also found the following threads concerning TLK110/RMII Clock connection:
http://e2e.ti.com/support/arm/sitara_arm/f/791/t/214462.aspx?pi301021=1
http://e2e.ti.com/support/arm/sitara_arm/f/791/t/243640.aspx

As we also use the TLK110 I'm asking the question why is TI using the TLK110 CLKOUT for RMIIx_REFCLK on AM335x on the AM335x ICE V2 board although TI knows that this is a bad thing to do?

It would be very helpful if EVM designs are bullet proof as they are often used as standard.

Did TI do any Ethernet tests on the AM335x ICE V2 board with RMII interface?

I finally managed to implement Ethernet communication on our board and I'm able to ping @100Mbit Full Duplex but I'm feeling pretty insecure having also used TLK110 CLKOUT as AM335x RMII Ref Clock.

Thanks and best regards,
Patrick

  • Hi Patrick,

    Pruf said:
    As we also use the TLK110 I'm asking the question why is TI using the TLK110 CLKOUT for RMIIx_REFCLK on AM335x on the AM335x ICE V2 board although TI knows that this is a bad thing to do?

    Where exactly did you see this on the ICE v2 schematic? Could you post the net name?

  • Hi Biser,

    First picture is from sheet 4 of schematic (dated Friday, January 31, 2014) showing AM335x side:

    Second picture is from sheet 11 showing the one source of the clock (CLKOUT of TLK110):
    (The other one is the same on sheet 12)



    Used schematic:

    Hope this helps.

    Regards,
    Patrick

  • Yes, I saw them now. The TLK110 datasheet is vague on the CLKOUT pin, saying that it's timings are dependent upon the XI input characteristics. I think that this works on the ICE because a synthesizer chip is used for clocking the PHYs.

  • We use the same clock synthesizer (CDCE913) but the mentioned E2E threads make me feel very insecure about the stability of this circuit. Can you please further investigate this issue. Thanks.

  • Yes, I have sent a request to the factory team to explain.

  • Biser, thanks and best regards.

    Patrick

  • The issue with the TLK110 not working in clock out mode with the AM335x is not related to AM335x or the PCB design. CLKOUT on the TLK110 is simply a 50 MHz reference clock that doesn't have any timing relationship to the RMII signals. Therefore, it can not be used as the 50MHz RMII reference clock.

    The ICE board has the TLK110 devices connected to the PRU via MII becasue PRU only supports MII.

    CPSW in AM335x supports RMII, but with the restrictions discussed in the E2E forum threads you have referenced.

    Are you trying to connect the TLK110 to the PRU or CPSW ports in AM335x?

    Regards,
    Paul

  • Hi Paul,

    As I know from TI, AM335x ICE V2 board is intended to use the Ethernet ports both for Ethernet TCP/IP and EtherCAT although not at the same time. That's why the TLK110 are connected to the PRU MII interface as well as the RMII interface of the CPSW.
    Our board should be able to do EtherCAT on two Ethernet ports. One of this ports can be switched to normal Ethernet if the device is the Ethernet Master. Then we have a third Ethernet port doing normal Ethernet. All three ports using a TLK110. We have been in contact with TI while designing our board. That was the at the time TI was designing the AM335x ICE V2 board.

    Why shoul TI connect the 50Mhz Clock to the AM335x RMII Ref Clock when they do not use RMII?

    In the datasheet of the TLK110 under the description of "CLKOUT" it is written: "In RMII mode, this pin provides a 50MHz clock output. This feature allows other devices to use the reference clock from TLK110 without requiring additional clock sources."
    Maybe this sentence has to be changed or extended with a notice.

    But most important I'm looking forward to answers from the questions of my first post.

    Best regards,
    Patrick

  • Hi Biser,

    Any news on this topic?

    Regards,
    Patrick

  • No feedback so far.

  • Sorry for the delay, I am escalating this...

  • Pruf said:

    Hi,

    Our custom AM355x board is close to the AM335x ICE V2 board. I'm now implementing Ethernet on our board. By chance I have seen this E2E thread: http://e2e.ti.com/support/interface/industrial_interface/f/142/t/206970.aspx

    And also found the following threads concerning TLK110/RMII Clock connection:
    http://e2e.ti.com/support/arm/sitara_arm/f/791/t/214462.aspx?pi301021=1
    http://e2e.ti.com/support/arm/sitara_arm/f/791/t/243640.aspx

    As we also use the TLK110 I'm asking the question why is TI using the TLK110 CLKOUT for RMIIx_REFCLK on AM335x on the AM335x ICE V2 board although TI knows that this is a bad thing to do?

    It would be very helpful if EVM designs are bullet proof as they are often used as standard.

    Did TI do any Ethernet tests on the AM335x ICE V2 board with RMII interface?

    I finally managed to implement Ethernet communication on our board and I'm able to ping @100Mbit Full Duplex but I'm feeling pretty insecure having also used TLK110 CLKOUT as AM335x RMII Ref Clock.

    Thanks and best regards,
    Patrick

    This issue has been highlighted with the ICE team and the board errata will be updated with this item to ensure that customers understand that they should not use TLK110 50MHz output as RMII_REFCLK.