5 Replies Latest reply on Jul 18, 2018 12:17 PM by cristian_filip

    SerDes Wizard (VX.2.3)

    aleksandr.kupchik@oracle.com

      Hello,

       

      I have been SerDes Wizard to test a diff pair.

       

      Bellow is a list of steps:

       

      1. Open a board file

      2. Defined a pair under test as a diff pair using the "differential Pair" tab

      3. Started the wizard, selected the pair, selected the protocol - 10KR

      4. Run the test

       

      Test has a couple failings. One of them - Near End Cross talk Mask.

       

      I believe that according to the 10KR standard, a near end is defined as a crosstalk interference between the  two diff pairs.

      I don't understand why the Near End reported in a case where the only one diff pair is defined.

       

      It is possible I am missing something.

       

       

      Regards,

       

      Alex

        • 1. Re: SerDes Wizard (VX.2.3)
          cristian_filip

          First of all, the IEEE802.3 specification does not define any mask for power sum crosstalk (including near end) for 10GBASE-KR. Please see my blog for more details about that:

          https://blogs.mentor.com/hyperblog/blog/2018/04/01/serdes-design-part-4-frequency-domain-analysis-for-high-data-rate-ser…

          Consequently, in the SerDes wizard there is no mask to check against, as a result we do not report power sum crosstalk failures. If there is no aggressor, then there will be a trace at -200dB in each of the the power sum plots, showing that there is no crosstalk in the system. This value is the lowest noise level in the computations. However, if the "Include crosstalk" checkbox in the Channel Operating Margin page of the wizard is checked, the wizard will account for aggressors of the same type (not alien crosstalk) depending on the coupling settings and coupling thresholds.

           

          Secondly, the power sum crosstalk plots are always included in the HTML reports, regardless if there is crosstalk or not in the system, due to simplifications needed for implementation. Removing those plots dynamically depending on the topology and/or the selected settings would add unnecessary complexity to the implementation. We do consider though that this should not be an inconvenient for the users.

          • 2. Re: SerDes Wizard (VX.2.3)
            aleksandr.kupchik@oracle.com

            Cristian,

             

            Thank you very much. Thank you for the pointers. I have found all information great and very useful.

            You are correct, the Power Sum Near End Crosstalk" is at -200 dB.

             

             

            I  think I failed to clearly state the question. 

            Please see the attached snapshot from a test report. It contains the "Near Return Loss Report", that has failed. There is a "Mask Check" statement right under the graph.  My question was related that that specific graph and statement. Is this a part of the pass/ fail criteria, and how this graph generated? Is this Near End of  the  diff pair itself?

             

            Again my apology if my initial post was not clear.

             

             

            Regards,

            Alex

             

             

            Near_End_Return_Plot.jpg

            • 3. Re: SerDes Wizard (VX.2.3)
              cristian_filip

              Hi Alex,

               

              This is the Near End Return Loss and represents the differential return loss SDD11 as seen from the transmitter side of the link (the diff pair under analysis). The Far End Return Loss is the differential return loss SDD22 and is being measured from the other side (Rx) of the link. Associated to those plots there are two TDR plots, one from the Tx side and the other one from the Rx side that help you to identify the location of the discontinuities on the link. The plots in the HTML report allow you to zoom in/out, place cursors, enable/disable traces, etc. Alternatively you can use the hyperlinks on the last tab of the HTML report and load the s-parameters of the channel in the Touchstone Viewer, where you can do many other things.

               

              Cristian

              • 4. Re: SerDes Wizard (VX.2.3)
                aleksandr.kupchik@oracle.com

                of cause, return loss ... , not a crosstalk.

                 

                My bad.

                 

                I have one more question. I would like to save a report to a different folder from the one where a board file is stored. It does not look like there is an option to that. 

                 

                Thank you,

                 

                Alex

                • 5. Re: SerDes Wizard (VX.2.3)
                  cristian_filip

                  You can just copy the results directory (using Windows Explorer) wherever you want. There is also a pdf version of the report inside the results directory, in case if you want to share it.