5 Replies Latest reply on Feb 21, 2018 10:52 AM by weston_beal

    Digital Oscilloscope with IBIS-AMI

    pjs

      Am I right in thinking the Digital Oscilloscope does not use the full algorithmic part of the IBIS-AMI model and it just uses the I-V analogue model in the IBIS file?

       

      If so, I'm not sure I understand how representative waveform captures are from the Digital Oscilloscope when using AMI models? Can it be relied on for any kind of output when using AMI models or should you always use the AMI channel analyser for this?

       

      Does that mean it's not possible to use the automated sweep features for parameters such as transmission line length, impedance, passive values etc? Is there a way to do this with the channel analyser?

       

      It would be good to understand what exactly the Digital Oscilloscope is doing to understand if the results are useful in this scenario.

        • 1. Re: Digital Oscilloscope with IBIS-AMI
          weston_beal

          In short, you need to use the IBIS-AMI Wizard with IBIS-AMI models. I have seen IBIS-AMI models in which the analog portion (standard IBIS table-based model) is a simple TDR and has nothing to do with the real behavior of the buffer. The actual buffer behavior was included in the algorithmic compiled portion of the model. The basic simulation in HyperLynx SI (or any circuit simulator) uses only the analog portion of the model. The algorithmic portion requires post-processing in the IBIS-AMI Wizard. This is the way IBIS-AMI models work in all tools that support them.

           

          Regards,

          Weston

          • 2. Re: Digital Oscilloscope with IBIS-AMI
            pjs

            Thanks Weston - that's very helpful. Do you know if there are any plans to add transmission line sweep functionality to the AMI wizard in the future?

            • 3. Re: Digital Oscilloscope with IBIS-AMI
              weston_beal

              Can you elaborate on your idea? I think these are 2 separate tasks, sweeping net topology, and sweeping model settings, but there might be a case for doing it all together. Generally, we have a designed or proposed channel layout and we want to sweep buffer settings to see if we can make the buffers work on that channel. Or, on the other side, we have buffers defined and want to see how far we can alter a net topology and still work with the buffers. If you have a proposal for something different, you should submit a new idea on this site. Just like you create a new discussion, you can create a new idea for marketing to think about.

               

              Regards,

              Weston

              • 4. Re: Digital Oscilloscope with IBIS-AMI
                pjs

                Thinking from a purely board design perspective, I think the two go hand in hand. Take for example if you have two interconnected ICs and you need to determine the maximum viable distance apart these components can be placed, you'd ideally sweep the transmission line length parameter to determine the maximum trace length your buffer can support. However, depending on the quality of the IBIS-AMI model, the digital oscilloscope + sweep tools only support the IBIS-IV analogue model which as discussed above may not be representative. Even if you were able to sweep the transmission line length parameter in a representative manner, then the maximum distance will probably heavily depend on pre-emphasis and equalisation parameters and so ideally you'd want to be sweeping those at the same time to get your best chance at getting the longest distance.

                • 5. Re: Digital Oscilloscope with IBIS-AMI
                  weston_beal

                  Yes, I think you should create a new idea with this description. If enough people think that it is a good idea then it will get considered by marketing.