6 Replies Latest reply on Sep 11, 2017 9:24 AM by weston_beal

    Slew rate measurement for DDR3

    udaya.kumar

      HI Experts,

      Can "DDR2 Slew rate De-rating"  window be used for DDR3 ?  or

      is there any separate app's for DDR3 slew rate calculations?

       

      Thanks in advance for early response.

       

       

      Regards,

      Udaya

        • 1. Re: Slew rate measurement for DDR3
          weston_beal

          Udaya,

           

          You can use the slew rate calculator for DDR3 signals. The thresholds are pre-set for DDR2, but there also the option to specify custom thresholds. The method of measurement on the waveforms is the same. You just choose the custom thresholds setting and input the appropriate voltage thresholds for your signals being measured.

           

          Regards,

          Weston

          • 2. Re: Slew rate measurement for DDR3
            udaya.kumar

            Thanks Weston.

            I did as you suggested.The slew rate of DQ is 4.19 V/ns. But Datasheet only contains de-rating table till 2 V/ns for AC150/DC100( Our DRAM is DDR1333) .

            How do I calculate  de-rated -setup  for 4.19V/ns? or

             

            Please correct me if I  am looking at wrong table?

             

            Regards,

            Udaya

            • 3. Re: Slew rate measurement for DDR3
              weston_beal

              Udaya,

               

              I've heard designers using two different methods. Some say that you should only use derating values that are declared in the table. Others say that it's better to interpolate values through the table, and extrapolate the values beyond the range of the table. Yet another philosophy is that the signal should not be faster than what is specified in the table, so you should add series resistance to slow the edge. The correct method depends on the priorities and product requirements of your particular design.

               

              Weston

              1 of 1 people found this helpful
              • 4. Re: Slew rate measurement for DDR3
                udaya.kumar

                Weston,

                 

                I agree with you. Device behavior should not be faster than what is specified in DS.

                Here my question is ,

                Why DDR2 slew rate APP's is showing more than what is specified in DS?

                 

                The answer would be , IBIS is not captured correctly / not in-line with JEDEC spec.  otherwise

                 

                Some head around required to ensure slew rate app's is doing it's job well !.

                 

                Any thoughts ?

                 

                 

                Regards,

                Udaya

                • 5. Re: Slew rate measurement for DDR3
                  shashank.d@siennaecad.com

                  Hi Weston,

                   

                  "DDR2 Slew rate De-rating" will give 2 values, Min and Max Slew rate. Which one we should consider for Calculating Derating setup and hold time values?

                   

                  Regards,

                  Shashank

                  • 6. Re: Slew rate measurement for DDR3
                    weston_beal

                    I think that the DDR2 specification is not clear on this overall. It assumes a single measurement (single edge). The DDR wizard measures many edges in the bit stream, so provides the min and max numbers. The very best situation would be to measure each signal edge and derate the setup and hold times for each edge accordingly. This is no feasible, so we use an overall calculation.

                     

                    I would calculate the setup and hold margins using both the min and max slew rates, and then keep the worst case that result in the smallest setup and hold margins. If one of those fail the timing, then go back to see if the worst-case assumption applies to the specific edge that caused the timing violation. If not, calculate new setup and hold margins for that edge. iterate until all edges are correctly calculated, or the worst-case measurement is applicable.