Pcells/XL Connectivity

[none]

IBM ships pcells with their sige5am design kit that are in skill context
files. Therefore I can not access the source. They have nfet and pfet
pcells that have a button for adding m1 creation over source/ drain
contacts. If I choose m1 creation the source and drains will show up as
incomplete nets. If I do not choose the m1 option the source and drains
of the pcell do not show up in the XL incomplete nets list.

I can not fix the pcells without source code. Do you think the
connectivity could be fixed by writing some skill that is run from the
layout view and not in a pcell. I would like to be able to add
connectivity info to the contact layers that touch drains and sources of
nfet or pfet pcells.

Or maybe it would be easier to write my own pcells that work. Whats the
opinion of comp.cad.cadence users.

Thanks

 

[eda support guy]

none wrote:

IBM ships pcells with their sige5am design kit that are in skill context
files. Therefore I can not access the source. They have nfet and pfet
pcells that have a button for adding m1 creation over source/ drain
contacts. If I choose m1 creation the source and drains will show up as
incomplete nets. If I do not choose the m1 option the source and drains
of the pcell do not show up in the XL incomplete nets list.

I can not fix the pcells without source code. Do you think the
connectivity could be fixed by writing some skill that is run from the
layout view and not in a pcell. I would like to be able to add
connectivity info to the contact layers that touch drains and sources of
nfet or pfet pcells.

Or maybe it would be easier to write my own pcells that work. Whats the
opinion of comp.cad.cadence users.

Thanks

My opinion is that you should not try so hard to get DLE to behave
right. It is but an indication of connectivity, and layouter can go-on
working with a broken DLE support. When DLE is wrong, if the diva check
is OK, then forget about it. When diva is wrong, if dracula|assura is
OK, then forget about it. Quoting Saddam in "South Park" : "Relax, man"
Now, if you tried IBM's library support, they are not responsive, you
don t want to relax, and you really want it bad to fix this... there are
options, but by doing that you'd surely loose IBM's backup. Heck, it
might even be considered reverse engineering and illegal where you work.
We all know how pcell code is a piece of clever engineering and valuable
intellectual property ! You ll find the same situation with CDF
callbacks, and maybe in the future even with specte/verilogA model
libraries. I even saw diva ruledecks that had been skill-encrypted (yuck).