• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 33 subscribers
  • Views 341 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.

DS125DF111: 10G SFF-8341 host TX eye_mask fail issue.

Jimmy Lai
Expert 6845 points
Part Number: DS125DF111
Other Parts Discussed in Thread: DS125BR401A

Dear Team,

  Attached for DS125DF111 transmitter eye_diagrame and Tjitter measurement using the best of de-emphasis 0x14 (-6dB) VOD setting 0x85 . I first adjusted the VOD setting within spec and then adjusted the de-emphasis. Do you have any comments for de-emphasis adjustment and do we need to add ac coupling caps on SFP+ side?

Board to board connecter s-parameter report refer to the 

PCB material - FR4 

REV05 SFP test report.pdfSI-BTB-51154_51159-Ac_20200716(PCIe gen 3).pdf

Many Thanks,

Jimmy

  • 0
    TI__Mastermind 36123 points

    Hi Jimmy,

    Could you please clarify the value of register 0x15 since this has the de-emphasis settings?  May I also know the output driver VOD that you are using?

    Also, do you have an estimated insertion loss on the TX side of the device.  I see the report for the board-to-board connector, do you have any information regarding the ~2.4 inch trace length after that?

    AC coupling caps are required on the high speed inputs and outputs of the DS125DF111.

    Thanks,

    Drew

  • 0 in reply to Drew Miller1
    TI__Expert 6845 points

    Hi Drew,

       The best of de-emphasis is 0x15 = 0x14 (-6dB)   VOD setting( 0x2d )= 0x85

    Many Thanks,

    Jimmy

  • 0 in reply to Jimmy Lai
    TI__Mastermind 36123 points

    Hi Jimmy,

    Thanks for the clarification.

    In general, it seems like you don't have much insertion loss between your retimer and SFP connector.  I did notice that the board to board connector insertion loss has "notches" around 5-6 GHz, which might be adding an additional challenge to meeting eye mask since the nyquist frequency of 10.3125 Gbps is 5.15625 GHz.

    I am wondering if it is possible that -6dB is too much de-emphasis.  Have you tried using less de-emphasis?  If so, how does this impact your eye?

    The table below is from another product (DS125BR401A), but it shows some suggested de-emphasis levels for certain use cases.  Based on these suggested values, it seems as though -6dB is too much de-emphasis.

    Thanks,

    Drew

  • 0 in reply to Drew Miller1
    TI__Expert 6845 points

    Hi Drew,

    We appreciated your great support always!

    According to the above table, set FR4<5 inch 0db,  VOD = 1V / 1.3V doesn't seem to be the best value and the eye height is very small. Are these two retimers the same in the tx equalization compensation?

    The following are our comparison measurement data,  please help to check, thanks!

    DS125111 retimer tuning result.docx

    Many Thanks,

    Jimmy

       

  • 0 in reply to Jimmy Lai
    TI__Mastermind 36123 points

    Hi Jimmy,

    Thanks for the detailed report showing different tx settings.  I am looking into this and will get back to you.

    Thanks,
    Drew

  • 0 in reply to Drew Miller1
    TI__Mastermind 36123 points

    Hi Jimmy,

    I suspect that since the de-emphasis table above was taken from a redriver, values were selected with the assumption that there would be a receiver at the end of the link that could also offer some equalization.  This likely contributes to the discrepancy we are seeing between recommended values compared to what is working well in your system.

    Looking over your eye mask data, it appears that you get best results with VOD=1.3V, so let's stick with that.

    It also appears that more de-emphasis also helped with eye margin.  Does this trend continue for de-ephasis < -6dB, or is -6dB the best?

    Thanks,

    Drew