• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 33 subscribers
  • Views 640 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.

How much repeater equalization is needed to recover the TX De-emphasis completely, over a lossy channel?

Mahesh KC03
Intellectual 1360 points
Other Parts Discussed in Thread: DS125BR820, DS80PCI800

To rephrase the question, what is the maximum pre-channel loss for a repeater so that it recovers the De-Emphasis (DE) from the TX completely?

For example, assume that the TX ASIC provides a data with 5 dB of DE and the channel loss from TX to repeater is 15 dB. With maximum 10 dB EQ of the repeater, will it be able to reproduce the same 5 dB DE at its output? The repeater DE is set to 0 and VOD/VID = 1. This ability to recover TX DE is essential in the link training in PCI, SAS protocols.

 

So, I just wanted to know how to allocate the loss budget.

 

Thanks in advance.

  • 0
    TI__Mastermind 29655 points

    Hi Mahesh,

    I am not sure exactly how much pre-channel loss there can be before the DE is no longer recovered. However, I do know that it is more important for the Rx to see that the waveform from the Tx ASIC changes when the Rx requests for different precursor and postcursor information. The DS125BR820 is able to retain the precursor and postcursor information from the Tx source much better than its prior generation, the DS80PCI800. With previous generation repeaters, the Tx and Rx had a hard time optimizing reliably because the repeater would compress the source signal such that the repeater outputs did not look much different across a variety of Tx source precursors and postcursors. In this case, it was hard for the Tx and Rx to train, because the Rx sees approximately the same signal for each setting. Therefore, I do not think it is as important that de-emphasis from the Tx source be perfectly reproduced at the repeater output. Rather, it is more important that the repeater does not disturb the ability of the Rx and Tx to determine the optimal signal between them. As long as the Rx can see that the Tx responds to its requests for different precursor and postcursor values, the pair can theoretically find a mutual signal conditioning setting. Thus, it is possible for our part to work even with 20-25 dB of channel loss before the part, despite not perfectly recovering the de-emphasis in the waveform the repeater receives.

    To summarize, the DS125BR820 will lower the overall attenuation or reduce the effective trace distance.  See the waveform groups in the attached graphic, the colors align by trace length.  Notice how the redriver waveforms “look” like the non-redriver waveforms from shorter PCB traces.

     

    Light Blue - @ Tx

    White – 4 inches

    Yellow – 8 inches

    Light Green – 12 inches

    Red – 14 inches

    Pink – 18 inches

    Blue – 20 inches

    Green – 23 inches

    Putting all this into an eye diagram – in this case redriver compensates for approximately 9 inches of test trace length.

    Thanks,

    Michael