4 Replies Latest reply on Apr 29, 2014 8:42 AM by Ed Bartlett

    Several questions about connectors and simulation in Hyperlynx

    cchen120

      Hi Steve,

       

      Thank you so much for your answer of the previous question about batch simulation. I have three more questions.

       

      1. Batch simulation do not support SPICE model, on the other hand, we can’t assign IBIS model to connectors in Hyperlynx. So how can we use batch simulation to simulate multiboard through connector?
      2. How to set up simulation time in the interactive simulation, especially when I am simulating the eye diagram?
      3. I assigned the pins in the model of the connector to the actual pins on the board. When I select one net, some other nets that are not connected are selected also. What happened? I thought some pairs of the pins in the model are shorted. But when I simulate them in Linesim, I found that all pairs are independent. The following image shows the problem. Only one pair of nets should be selected but there are 5 pairs are selected.

       

      Thank you for your help!

       

      question about the selecting nets.JPG

        • 1. Re: Several questions about connectors and simulation in Hyperlynx
          Steve_McKinney

          Hi cchen,

           

          Sorry I didn't get a response to you sooner on this - I've been traveling.  I think you should open a support ticket for these issues.  The customer support team should be able to get you up and running with the items you list below. 

           

          Here's a quick response to your questions though:

           

          1.  This is a current limitation of the tool that we're developing a solution for.  In the mean time, I would suggest characterizing the behavior of the connector from your SPICE simulations and then approximating it in the simple model with delay, impedance, capacitance, inductance.  That will get you a good first order approximation and then the nets that might be failing or close to failing, you can go back and interactively simulate them with the SPICE model for the connector for higher accuracy.

           

          2. Simulation time (I'm assuming you mean the transient time) is controlled through the Oscilloscope.  In standard mode, you set the ns/div scale and whatever time is displayed in the Oscilloscope, that will be how long your simulation is for (say 10ns or 20ns).  For eye diagram mode, the transient time is a combination of your data rate and the number of bits in your stimulus.  So if you have a data rate of 3.125 Gbps you have a UI of 320ps and if you simulated with a PRBS of 2^7 (127 bits), your simulation time would be 40.64ns.

           

          3.  This is one customer support will need to look into with you.  Since it's specific to your design setup, I would only be guessing which isn't that helpful.

           

          -Steve

          • 2. Re: Several questions about connectors and simulation in Hyperlynx
            cchen120

            Thank you so much for your answer.

             

            For the 1st problem, I think will use the interactive simulation with crosstalk enabled instead since the number of nets with large crosstalk are not too many.

             

            For the 3rd problem, I found AppNote 5612 which talks about how to use connector in Multiboard and solved that problem. It told me to assign the model in LineSim and export as a .hyp file to multiboard. The the nets are independent now.

            • 3. Re: Several questions about connectors and simulation in Hyperlynx
              sandeep.kumar

              App note 5612 is not there on supportnet,

              if anyone os having kindly attach .

              • 4. Re: Several questions about connectors and simulation in Hyperlynx
                Ed Bartlett

                This app note is made obsolete with HyperLynx version 8.0. If you upgrade to a newer version of HyperLynx you will get many new features and bug fixes, plus you will not need this app note procedure.

                Best regards,

                Ed.