4 Replies Latest reply on Oct 6, 2010 1:34 AM by jerome.thomas

    PEX R+C+CC extraction mode without Grounded Capacitances ....


      Hi ,


      When i launch PEX in R+C+CC mode , grounded capacitances don't appear ,there are just R an CC columns

      In R+C mode , PEX extract R and C , but C value is excatly the same than CC value extracted in previous mode R+C+CC .

      Has someone ever encounter this mistake?

      Thanks in advance


      calibre version 2010.3_19.12

        • 1. Re: PEX R+C+CC extraction mode without Grounded Capacitances ....

          Hi Jerome-


          In RC mode, PEX is reporting "lumped capacitance".  The coupled capacitance is "lumped in" with the grounded capacitance in the report. (This dates back to when the capacitance to substrate was usually much greater than the coupled capacitance.)


          In RCC mode, the capacitance to substrate/grounded capacitance and coupled capacitance are separate. In the older versions with which I am most familiar, "grounded capacitance" includes coupled capacitance to nets that have been excluded, but I am unsure what 2010.3 does.


          For a visual of where the parasitics go, take a look at the "Producing Parasitic Models" chapter of the Calibre xRC User's Guide.


          That lumped capacitance equals coupled capacitance makes it sound like you may have the substrate defined as a ground net. (In the very old versions, before the multiple grounds command was added, that was the method for handling different domains.  I've also seen some setups where people, trying too hard to be precise, explicity define ground resulting in it being seen as a net rather than the substrate.)


          There was a big retitling of PEX SVRF commands in 2009, so I am unsure what commands to tell you to check in the __rules__ file. You could contact customer support for help with this.


          Another possibility, though you would probably expect this outcome if that is what the layout was, is that you are extracting parasitics in layers M7 and higher only. In most modern processes, starting around M4 or M5 the coupled capacitance dominates the capacitance to substrate; by M7, the capacitance to ground may very well be too small to show up. I have heard of this happening in conjunction with the top level routing in large, fast chips when worried about inductance. But like I said -- you would probably expect these results in that case.





          • 2. Re: PEX R+C+CC extraction mode without Grounded Capacitances ....

            Hi Samantha ,

            Thanks for your answer .


            I would like correct what I said previously because with a bigger cell lumped Capacitances are detected in R+C+CC  mode .

            But i don't understand something else :

            I work with  SOI technology , how Calibre can extract lumped Capacitance in this techno with the following value where any bulk appear ?


            CAPACITANCE ORDER  ndiff pdiff p1trm shield1 met1sl shield2 met2sl shield3 met3sl

            shield1 = ((p1trm AND met1sl) AND (met2sl AND met3sl)) AND capdef
            shield2 = copy shield1
            shield3 = copy shield2
            CONNECT shield1
            CONNECT shield2
            CONNECT shield3

            RESISTANCE SHEET p1trm   [45  100]

            RESISTANCE SHEET met1sl  [0.150  100]
            RESISTANCE SHEET met2sl  [0.150  100]
            RESISTANCE SHEET met3sl  [0.070  100]

            RESISTANCE CONNECTION met1sl ndiff   [100    0]
            RESISTANCE CONNECTION met1sl pdiff   [150    0]
            RESISTANCE CONNECTION met1sl p1trm   [100    0]
            RESISTANCE CONNECTION met2sl met1sl  [1.44   0]
            RESISTANCE CONNECTION met3sl met2sl  [1.44   0]


            CAPACITANCE CROSSOVER PLATE  met1sl ndiff [0.037]
            CAPACITANCE CROSSOVER FRINGE met1sl ndiff [0.049]
            CAPACITANCE CROSSOVER PLATE  met1sl pdiff [0.037]
            CAPACITANCE CROSSOVER FRINGE met1sl pdiff [0.049]

            CAPACITANCE CROSSOVER PLATE  met1sl p1trm [0.039]
            CAPACITANCE CROSSOVER FRINGE met1sl p1trm [0.050]

            CAPACITANCE CROSSOVER PLATE  met2sl ndiff [0.024]
            CAPACITANCE CROSSOVER FRINGE met2sl ndiff [0.043]
            CAPACITANCE CROSSOVER PLATE  met2sl pdiff [0.024]
            CAPACITANCE CROSSOVER FRINGE met2sl pdiff [0.043]

            CAPACITANCE CROSSOVER PLATE  met2sl p1trm [0.025]
            CAPACITANCE CROSSOVER FRINGE met2sl p1trm [0.044]

            CAPACITANCE CROSSOVER PLATE  met2sl met1sl [0.070]
            CAPACITANCE CROSSOVER FRINGE met2sl met1sl [0.055]

            CAPACITANCE CROSSOVER PLATE  met3sl ndiff [0.017]
            CAPACITANCE CROSSOVER FRINGE met3sl ndiff [0.044]
            CAPACITANCE CROSSOVER PLATE  met3sl pdiff [0.017]
            CAPACITANCE CROSSOVER FRINGE met3sl pdiff [0.044]

            CAPACITANCE CROSSOVER PLATE  met3sl p1trm [0.018]
            CAPACITANCE CROSSOVER FRINGE met3sl p1trm [0.044]

            CAPACITANCE CROSSOVER PLATE  met3sl met1sl [0.032]
            CAPACITANCE CROSSOVER FRINGE met3sl met1sl [0.051]

            CAPACITANCE CROSSOVER PLATE  met3sl met2sl [0.046]
            CAPACITANCE CROSSOVER FRINGE met3sl met2sl [0.050]

            // ................ Coupling Capacitances ..................

            CAPACITANCE NEARBODY ndiff WITH SHIELD ndiff [0.022 5]
            CAPACITANCE NEARBODY pdiff WITH SHIELD pdiff [0.022 5]
            CAPACITANCE NEARBODY ndiff pdiff             [0.022 5]
            CAPACITANCE NEARBODY p1trm WITH SHIELD p1trm [0.022 5]
            CAPACITANCE NEARBODY met1sl WITH SHIELD met1sl [0.039 5]
            CAPACITANCE NEARBODY met2sl WITH SHIELD met2sl [0.039 5]
            CAPACITANCE NEARBODY met3sl WITH SHIELD met3sl [0.023 5]

            • 3. Re: PEX R+C+CC extraction mode without Grounded Capacitances ....

              Hi Jerome-


              I think I've found the root cause of the problem: you say you are using Calibre 2010.3, but the rules you quote are a very old format -- the Capacitance rules were made obsolete back when I was still working with xRC, about 2006. Resistance Sheet went out with 2007. It could well be that even the error message about incompatible versions has expired. The reason they went obsolete? They weren't complex enough to handle new features like SOI that occur below 90 nm.


              I believe you can get up to date rules from the foundry that will be manufacturing your design. The current style (2008.1 and newer) of xCalibrate-generated rules are encrypted, but they should have a plain-text header that will say what version they are. (The default names are rules.C and rules.R for capacitance and resistance respectively.)


              If you are still seeing these odd results when using up-to-date rules, I highly recommend filing a service request. That team's information will be much more current than mine, and they may have run into similar situations with the layouts for other customers.


              Best of luck-







              • 4. Re: PEX R+C+CC extraction mode without Grounded Capacitances ....

                Hi Samantha ,


                I have just open a SR about this and ask Fab support recommended Calibre version to use .

                Thanks again .