one of the extracted nelist for spectre has the following pattern :

inline subckt pmos (s g d b sub)

MP (2 3 3 0) pmos l=4u w=4u $bulk=bulk

inline subcircuit has 5 nodes and device instance has 4 node and 5th node is at the end as variable.
Spectre simulation gives warning saying it expects 5 nodes and there are 4 nodes.
How to attach the variable to the 5th node in inline subckt ?

Andrew -
I'm not entirely sure about the question, but it looks like krishnap has a problem that the netlist has

MP1 (d g s b sub) pmos l=1u w=4u

and the library has

.model pmos bsimsoi type=p ...

where (I'm guessing) the device "bsimsoi" has four fixed terminals and then some optional terminals that are passed to the model evaluation code as parameters.  (As I recall, in the spice3 code, there are up to 3 optional terminals, and there's some funny business in recognizing the tokens as terminals.)  I don't know what model krishnap really has, nor if Spectre has any built-ins that have terminals passed as parameters (it appears that Spectre's bsimsoi has optional terminals but they're inside the parentheses, not passed as parameters).

Or maybe the library has a .subckt definition of pmos.  Do subckts support optional terminals?

-Geoffrey

John O Donovan wrote on Aug 28^th, 2007, 4:46am:


What was the syntax for that MODEL statement?  Does it use a $ as the original poster did?

It sounds like the extracted netlist now has an extra terminal (presumably, the schematic netlister put out only 3 terminals), and perhaps the poster was trying to use the existing library files with the extracted netlist, perhaps overriding the bulk node in the MODEL statement with the layout node.

But, I'm just guessing, so maybe we'd best wait for the original poster to clarify his question before we try to solve a different problem.