12 Replies Latest reply on Oct 5, 2012 7:51 AM by oscar.martinelli

    How to convert an .s12p from SE to Mixed Mode?

    cristian.filip

      I know that we can convert an .s4p from SE to Mixed Mode using HL Touchstone Viewer (Convert -> Mode), but this menu seem to be grayed out for S-parameter models with more than four ports. Right now we use the .LIN command in HSpice to do this conversion. Is there any way to do that directly in HL?

       

      Thank you,

      Cristian

        • 1. Re: How to convert an .s12p from SE to Mixed Mode?
          weston_beal

          This presents a few questions. What does it mean to have mixed mode S parameters for this many ports? Assuming that you define all the differential ports from the many single-ended ports, what do you do with the parameters in between diff pairs?

          HyperLynx keeps it fairly simple by working with only 4-port files. For your case of 12 ports, you can use the port reduction function to get down to 4 ports, and then convert that to mixed mode.

          • 2. Re: How to convert an .s12p from SE to Mixed Mode?
            cristian.filip

            Hi Weston,

             

            Any crosstalk analysis for high-speed differential signals has to be done using one victim diff pair and at least two aggressor diff pairs one in each side of the victim. Some people use two aggressors on each side for a total of 5 diff pairs (20 SE ports or 10 differential ports). We have adopted Intel’s practices and we use only 12 SE ports. There is a lot of very good documentation explaining why to use 12 ports, but the paper that I like the most is “Data Mining 12-Port S-Parameters” from DesignCon 2008, by Dr. Eric Bogatin and Mike Resso (see the attachment). Note also that most of the test equipment vendors have 12 port instruments available (e.g. SPARQ Signal Integrity Network Analyzers from Lecroy - http://teledynelecroy.com/sparq/?capid=131&mid=1042) while most of the EDA vendors do support SE to Mixed Mode S-Parameter conversion for unlimited number of ports (e.g. HSpice, Agilent ADS, etc.).

             

            How do you suggest us to measure differential parameters as near and far end differential crosstalk (SDD31, SDD41, SDD51, SDD61) without this capability?

             

            Thank you,

             

            Cristian

            • 3. Re: How to convert an .s12p from SE to Mixed Mode?
              weston_beal

              Thanks for the reference to Dr. Bogatin's paper. Here's my summary. If you work for a VNA company, show what you can do with a VNA. If you work for an EDA company, show what you can simulate. In HyperLynx, I suggest that you instantiate the 12-port S-parameter block in LineSim with differential buffers at the appropriate ports. Simulate the differential crosstalk in the time domain.

               

              If you really need to convert a 12-port S-parameter file to mixed mode, you should suggest this enhancement on the Mentor Ideas site. I think there is at least one other customer who would promote this idea.

               

              Regards,

              Weston

              • 4. Re: How to convert an .s12p from SE to Mixed Mode?
                cristian.filip

                Hi Weston,

                 

                My question was about simulating the differential crosstalk in frequency domain rather than time domain. We do have a system specification that asks for at least -100dB of isolation between two adjacent differential pairs and FD simulations make sense for us. It is also common practice when designing backplanes to define channel characteristics in FD, not in TD since the characteristics of the driver/receiver might be unknown. Same principle applies to connectors, where differential crosstalk between adjacent pins needs to be specified in FD (see Figure 3, page 2, in the attached datasheet). Anyway I was expecting from you some technical solutions instead of your comment “I think that there is at least one other customer who would promote this idea” which discourages the customers to suggest any other tool enhancements or even to ask questions on this blog.

                 

                If you had a positive attitude you could explore the field of FD analysis and understand the beauty of being able to converting from FD to TD and vice versa. You could also discover how powerful tool HyperLynx is in this regards and I am making reference to two new features that were added in v8.2: Time-Domain Responses and TDR Impedance Plot from Touchstone Viewer. I am attaching for you a simple LineSim simulation setup that contains three coupled diff pairs along with SE 12 port parameter extracted from this simulation deck. You will find also a small HSpice netlist that can be used to convert this parameter into a mixed mode parameter. I am providing for your convenience the converted s-parameter and I invite you to open it using a text editor to identify the port mapping. Then you can open it in Touchstone Viewer and start to play with those two features that I mentioned above. You will be able to change on the fly between FD and TD responses and do some sort of “what if” analysis. For example you can see the TD responses of a trapezoidal pulse injected in diff port 1 at any other port: reflected and transmitted pulse, as well as coupled pulse at near and far ends. Ultimately if you use the TDR Impedance Plot option you will find out that you can select Mixed Mode Plot Type, Differential Mode and choose any number of ports from the drop down menu (this feature is not limited to four ports only).

                 

                In conclusion I hope that you will find this interesting and there will be more than only one customer that will benefit from my honest technical opinion (not a sales speech as you suggest).

                 

                Cheers,

                 

                Cristian

                • 5. Re: How to convert an .s12p from SE to Mixed Mode?
                  yu.yanfeng

                  I agree with you and I think Mentor Hyperlynx team may consider to enhance conversion function  for S-parameter mode convertions, function for S-parameter quality check, function for compliance check etc.

                   

                   

                  Yanfeng

                  • 6. Re: How to convert an .s12p from SE to Mixed Mode?
                    weston_beal

                    Sorry if my comment came across negatively. It was not intended that way, but sometimes my terse text does not convey the right feeling. Thanks for providing the differential to single-ended coupler example. I hope that others can benefit from it as well.

                     

                    Regards,

                    Weston

                    • 7. Re: How to convert an .s12p from SE to Mixed Mode?
                      Steve_McKinney

                      Hi Cristian,

                       

                      I passed this thread on to our engineer responsible for s-parameters.  Generally speaking, he agreed with what you had to say on this topic.  He also said this is something that he is currently working on.  I can't say that it will be part of 9.0, but it is something that we feel we must do, so it will be coming as soon as we can in the tool.  One of the key pieces to make this happen is support of the Touchstone 2.0 spec which is what we're working on right now.

                       

                      -Steve

                      • 8. Re: How to convert an .s12p from SE to Mixed Mode?
                        cristian.filip

                        Hi Steve,

                         

                        I am extremely pleased with your answer and I am glad that once again we collaborate for our common benefit.

                         

                        Cheers,

                        Cristian

                        • 9. Re: How to convert an .s12p from SE to Mixed Mode?
                          cristian.filip

                          Hi Steve,

                           

                          It looks like Modua (HyperLynx 3D EM) has this capability (see the attached Mentor presentation – slides 23 to 43). Maybe this helps…

                           

                          Cheers,

                          Cristian

                          • 10. Re: How to convert an .s12p from SE to Mixed Mode?
                            oscar.martinelli

                            Hy Cristian,

                            could you please advise how to visualize with the "TDR Impedence plot" the TDD11 (return loss in time domain)?

                             

                            Regards,

                            OSCAR

                            • 11. Re: How to convert an .s12p from SE to Mixed Mode?
                              cristian.filip

                              Hi Oscar,

                               

                              Using the mixed.s12p parameter that I have uploaded along with Dr. Eric Bogatin's naming conventions and port mapping described inside the 12 port mixed mode S-parameter, you should do the following:

                               

                              1. 1.     Open the mixed.s12p in Touchstone Viewer and check S(1.1). Accordingly to the pin mapping this is SDD(1,1).

                              2. 2.     From  Touchstone Viewer ->View - > TDR Impedance Plot , select Mixed Mode (Plot Type), Ports 1 and 3, Mode Differential, Other Ports Terminated, Stop Time you can leave 10 ns (depends on the system’s delay)

                               

                              I did plot for you a couple of other responses and attached all of them.

                               

                              If Mentor guys disagree or have more to add, please feel free to comment.

                               

                              Cristian

                              • 12. Re: How to convert an .s12p from SE to Mixed Mode?
                                oscar.martinelli

                                Hy Cristian,

                                clear now... thank you for your time!

                                 

                                Regards,

                                OSCAR