• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 33 subscribers
  • Views 409 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.

SN65LVCP114 Port C Loopback

Kevin Du
Prodigy 20 points
Other Parts Discussed in Thread: SN65LVCP114

We are testing SN65LVCP114 in a SFP+ application. 

On the board, the trace length between SFP+ socket to LVCP114 PortC is less than 2 inches. From LVCP114 PortA to 10G PHY(not KR capable) is another 2 inches. The PortB of LVCP114 is connected from IN back to  OUT with about 2 inches trace without cap. LVCP114 is operate at 2.5V and the ripple is <8mv with dedicated regulator.

Our first test case is that we loopback on Port C and test BER. So the 10.3125G XFI goes in CIN and loopback out COUT. The setting is: EQ=1.3dB, Gain=0.5, VOD=0.6. We are observing bit error at the rate of 10E-12 with DAC cable. And if we use SFP+ optics module, the bit error rate is much higher at 10E-10 (barely working). 

Now for the same board, if we bypass the LVCP114 and connect trace from SFP+ socket to PHY directly (we have backup traces, breakers on board to characterize the design), the bit error rate is 0 after 8 hours of running.

The second test is that we enable DIAG=1, and select PortB, since PortB IN and OUT are connected, so the signal is loopback through port B. This time, we have much worse BER. DAC cable is at 10E-11, and SFP+ optical module does not work.

We are wondering if the LVCP114 is designed for such application and appreciate your insight.

  • 0
    TI__Mastermind 20820 points

    Hi Kevin,

    The SN65LVCP114 is a channel equalizer with a mux and was developed for applications like you described. There is the 1.3dB of EQ that cannot be turned off though. So my worry is that your perfect XFI input eye from the optical module is being over equalized because there is no channel to equalize in your link.

    Do you have the capability of measuring the eye diagram on a scope before and after the mux so that we can see the effect that the SN65LVCP114 is having on the signal? This will be the best way to see what the performance of the channel is. Also, are you AC coupled on the inputs of the device? We recommend AC coupling for the best performance when using the SN65LVCP114. The settings that you described for gain, Vod, and Eq sound right for your application but you can try adjusting the output Vod to 1200mV instead of 600mV to see if this provides you more head room on the output driver.

    Regards,

    Mike

  • 0 in reply to Michael Peffers
    Prodigy 20 points

    Hi Mike

    Thanks for getting back to me promptly.

    There are 100nF cap built in SFP+ module, so the signal is already AC coupled into LVCP114's port C. The PortB's IN and OUT are directly connected without cap though.

    Change Vod to 1200mV does not seem to help much. 

    And thanks for pointing out the 1.3dB minimum EQ. I'd like to follow up on this idea to see if you have further pointers for us to act upon. 

    Our scope is not capable of measuring the eye diagram of 10.3125g. Since we are in San Jose. in case you can help us connect to a local rep in the area, I am happy to bring in the board for a spin.

    Thank you

  • 0 in reply to Kevin Du
    TI__Mastermind 20820 points

    Hi Kevin,

    I am going to close this post since our conversation has moved off line.

    Regards,

    Mike