1 Reply Latest reply on Feb 27, 2012 8:01 AM by samantha_lizak

    PEX ALIAS makes different result between other layers

    km37.kim

      hi, I have a question about 'pex alias' rule.

      i'm using this rule like this

       

      PEX ALIAS M0 M0a M0b M0c

      CAPACITANCE ORDER M0 M1 M2

       

      and i found that if PEX ALIAS rules is changed like below, the capacitance between M1 and M2 is affected and changed.

       

      PEX ALIAS M0 M0a M0b

      CAPACITANCE ORDER M0 M1 M2

      (M0c is removed)

       

      M1 and M2 is above M0 physically, so i think aliasing of M0 can't affect any capacitances between M1 layer and M2 layer..

      anyone can answer what causes this problem? or does it make sense?

        • 1. Re: PEX ALIAS makes different result between other layers
          samantha_lizak

          Hi KM-

           

          First, which version are you using?  PEX Alias has undergone a lot of enhancements over the years; this could make a big difference.

           

          Second, would you please provide more details on extraction?  As you mention that capacitance between M1 and M2 is affected, I assume you are doing -cc as opposed to -c, but are there any reduction statements?

           

          Last, why did you want to remove it? What is the goal for the change?

           

          Now, the likely explanation for what you are seeing:

          The big picture is that M0c is not being treated as part of M0, capacitively. As you are doing coupled capacitance, that means that some capacitance between M0c and M1 and M0c and M2 is going to be handled differently than before.  If M0c is not in the Capacitance Order statement, it is considered to not interact with the other layers.

           

          Now you may need to either take my word or pull out a physics text book here.  Say you have a slice through the layout where all 3 of M0, M1, and M2 occur.  M1 interacts with M0 and M2. (M0 and M2 also have a bit of interaction, but it is going to be dwarfed by M1.) The coupled capacitance between M0 and M1 is going to weaken the capacitance between M1 and M2. (This is most noticeably in the fringe component; that is, capacitance from the wire's sidewall.)  Total capacitance seen by M1 is greater when it has an upper and a lower neighbor, but the capacitance between M1 and upper is less in the presence of a lower neighbor than it would be if M1 and the upper did not have the third party.

           

          So, by removing M0c from the PEX Alias, you have told xRC M0c is something other than M0.  Unless you have it in the Capacitance Order statement, xRC doesn't see it as interacting with M1 and M2 (or M0 itself, for that matter). So areas where the 3-body interaction includes M0c are going to be treated as a two-body problem.  I predict the coupled capacitance between M1 and M2 in those areas is higher with M0c removed.

           

           

          Incidentally, I see that PEX Alias was deprecated in 2010.4, and if possible Mentor Graphics would like you to move to PEX Map. (We understand that many times the engineer is not allowed to change the rule files without rigorous testing.)  You might want to try that to see if it behaves more as you would expect.  I'm not a PEX specialist, so I am unsure what the differences are.

           

           

          Hope this was helpful-

           

          Sam.