• State
  • Locked Locked
  • Replies 10 replies
  • Subscribers 33 subscribers
  • Views 2883 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

TLK110: TLK110 stops working with high-traffic

Kummi
Mastermind 6220 points
Part Number: TLK110

Hi,

We have issues with TLK110 PHY(RMII mode) with high-traffic multicast broadcast packets.
The PHY communication stops working(hangs) whenever there is a high-traffic
or when used for long duration(few days). It works normal when there is a low traffic.

The communication was stopping with a traffic more than 4000pps and
after adding PHY software reset command we are able to get the speed upto 20000pps,
but still the communication hangs in between.

Do we have any such related issues with TLK110, which stops working during high-traffic or long run?

Best Regards
Kummi

  • 0
    TI__Mastermind 18425 points
    Hello Kummi,

    Can you provide more details about the test?
    - Packet Size
    - Interpacket Gap (IPG) and Channel Utilization
    - Operating speed (10M or 100M)

    What the problem symptoms when the PHY hangs?
    - Does the PHY lose link with link partner?
    - Does the PHY lose communication with the MAC?
    - Does the PHY become unresponsive to MDIO-MDC communication?

    -Regards,
    Aniruddha
  • 0 in reply to Aniruddha Khadye
    Mastermind 6220 points

    Hi Aniruddha,

    Thank you very much for the reply.
    Below is the detail.

    - Packet Size
      Average 350 Bytes,Multicast packets are used for the test.

    - Interpacket Gap (IPG) and Channel Utilization
      Each pps(packets/s) of 5,000pps, 10,000pps, 20,000pps, 30,000pps
      are tested, issue occurs if the pps increases and also with long run.

    - Operating speed (10M or 100M)
      100M

    - Does the PHY lose link with link partner?
      Yes, packet communication is Lost with no Trasnmit/Receive packets.  

    - Does the PHY lose communication with the MAC?
      Yes, no more communication with PHY, cannot receive Trasnmit/Receive interrupts.

    - Does the PHY become unresponsive to MDIO-MDC communication?
      We can read the PHY registers through MDIO-MDC communication.

    Just in case I am attaching the PHY registers dump before the issue
    and after the issue.　PHY Status.pdf

    Recently we also have cases where we cannot access the PHY registers after this issue.
    In that case MDIOUSERACCESS0 register's bit 31 was set to "1" which didn't change to "0".

    Best Regards
    Kummi

  • 0 in reply to Kummi
    Mastermind 6220 points
    Please let me know if there is any comments on the above answers.
  • 0 in reply to Kummi
    Mastermind 6220 points
    any updates on this?
  • 0 in reply to Kummi
    Mastermind 6220 points
    Hi Aniruddha,

    One thing I forgot to mention.
    Actually the CLK_OUT from TLK110 is connected to the RMII_REFCLK of MAC.
    Will this could cause problems during High-Traffic communication?

    Best Regards
  • 0 in reply to Kummi
    TI__Mastermind 23675 points
    Hi Kummi,

    Can you send me a schematic?
    From the register dump it looks like the link is stable but you are receiving RX_ERs.

    Thank you,
    Ross
  • 0 in reply to Ross Pimentel
    Mastermind 6220 points
    Hi Ross,

    Thank you for the reply.

    I am sorry that it is difficult to send you customer's Schematics.
    But, this schematics is exactly same as the AM335x ICE EVM.
    www.ti.com/.../tmdsice3359

    After some testing we found that, after connecting the RMII_REFCLK of MAC
    directly to the External 50MHz Clock(instead of CLK_OUT) , we were able to reduce
    the Hang up issue to some extent.

    Could you please let us know if there is any information that
    TLK110's CLK_OUT clock could cause problems during High-Traffic communication?

    Best Regards
    Kummi
  • 0 in reply to Kummi
    TI__Mastermind 23675 points
    Hi Kummi,

    is the RMII_REFCLK (in the MAC) an input or output?
    I am not aware of any issues with our CLKOUT pin.
    Have you confirmed the ppm of the CLKOUT? Any jitter measurements?
    Also, can you provide the load capacitance you are using and the crystal?
    A problem I often see is that the load capacitance is off slightly and it results in too much ppm offset.
    When sending higher length frames and/or higher utilization the issue becomes apparent.

    Thank you,
    Ross
  • 0 in reply to Ross Pimentel
    Mastermind 6220 points

    Hi Ross,

    Thank you.

    RMII_REFCLK (in the MAC) is the input,
    the CLKOUT is connected to the RMII_REFCLK of the MAC(AM335x).

    With regards to the CLKOUT issue, I was talking about the below E2E.
    e2e.ti.com/.../243640
    It mentions that "The output clock (CLKOUT) of the TLK110 device cannot be used
    to source the RMII clock to AM335x as doing so would violate the interface timing"

    The customer is using external oscillator as shown below.
    With Capacitance 0.01u.



    Best Regards
    Kummi

  • 0 in reply to Kummi
    TI__Mastermind 23675 points
    Hi Kummi,

    It might be an issue with the AM335x because I do not see why the CLKOUT is not ok to use here.
    The AM335x might have tighter timing requirement.

    Kind regards,
    Ross