MAGIC Magic Mailing List
 
 

From: R. Timothy Edwards (tim AT stravinsky DOT jhuapl.edu)
Date: Thu Oct 03 2002 - 14:20:04 EDT

  • Next message: Jeff Sondeen: "Re: problems with connectivity in magic"

    Dear Mika,
    
       The source of the problem seems to be this:
    
    << m2contact >>
    rect 168 396 172 400
       .
       .
       .
    << polycontact >>
    rect 168 396 172 400
    
    The "SCN6M_SUBM.10.tech27" does not support (I don't know if this is correct
    or not according to the process design rules) via overlapping poly contact.
    The intended via (drawn first) at this location is being blown away by the
    drawing rules.  Why magic retains the original connectivity information is
    another question, probably worth looking into, but not your root problem.
    
    Your layout is also using layer "pm12contact" in several spots.  The root
    problem may have something to do with this, as this stacked via is showing
    up in places where no poly exists, whereas the poly contact + via
    (separately drawn) is showing up in places where "pm12contact" was probably
    (?) intended.
    
    The quick solution to the problem you saw is to add, in the "compose"
    section of the tech file:
    
      paint pc m2c pm12c metal1
      paint m2c pc pm12c metal1
    
    which will keep the via from being erased when a poly contact is painted
    on top of it.  Unfortunately, the syntax "compose pm12c pc m2c" is
    illegal when both types are contacts (maybe this should be fixed---magic
    should be able to handle cases like this).  And the "paint" syntax above
    is not quite correct for m2c painted over pc (the poly contact continues
    to exist).   But at least it should help you see what your automatic
    layout generator is generating.
    
    						Regards,
    						Tim
    


  •  
     
    Questions? Contact Rajit Manohar
    cornell logo