CADForums.net

mauricio · Guest

Is there any way to get the resistance of a selected polygon or does
anybody has a rutine for that?
Thanks

Tim · Guest

The problem I see with a skill routine is that the function won't know
which edges of the shape are the 2 ends of the resistor unless if you
do a partial select of the 2 edges that define the ends. Then any shape
can be broken up into rectangles and/or use the fomula for
non-rectangular shapes for non 0/90 shapes.

Tim

mauricio · Guest

If it is a rectangle, then it's as simple as you said, but if I have a
polygon then I think the simples way to do it is to reduce it to
rectangles and that is where I'm stuck. I found some papers talking about
it, but I hope someone has a rutine already written or an example of how
to do it.
Anyway, thanks Partha, I'm going to check in the ROD doc to see if I can
find something.

tattvamasi@gmail.com · Guest

It is pretty simple to write your own, you can get the length and the
width of the selected rectangle, and grab the resistivity of the layer
from the technology files, to do the necessary calculation to get the
resistance - rho*l/a

I beleieve there is an example of this in the ROD documentation as part
of the installs
Partha

Gerry Vandevalk · Guest

Of course not.

Resistance ( a 3d phenomena abstracted into a 2d problem. )

Your assumption that a "polygon" has resistance makes the assumption that
you identify the 2 ( hopefully not more ;^)
end points.

What you probably want is some sort of idealized routine to calculate an
approximate DC resistance of a rectilinear shape when
you assume that the current flow is between your two "ideal ends"

Even this is a tricky problem.

I have created path based resistor structures that were mostly rectilinear
and that computed the "number of squares" of each segment and
added (a user over-ride-able ) corner factor of ( 0.56 or 0.566 squares ...
I think this was a fab choice !? )
and then multiplied the total # of squares with the rho of the material to
get you a first (or second)
order resistance ( when added to a Fab supplied end correction factor that
was proportional to the width ... )

something like R = RHo * ( #squares ) + endfactor / WidthOfEnd1 + endfactor
/ widthofEnd2

assuming that the resistor can be broken up into rectangular sections and
corners ...
and each segment contributes Lseg/Wseg to #squares and each corner adds a
cornerfactor of 0.56square.

This was good enough in most cases ... ( unless you really needed a more
exact 3d sim run ... )

But if you require better accuracy ( or more complex structures ) either you
understand this problem much better or
you will not get the accuracy you are expecting ...

( Rant ...
Try calculating the real value of a corner in an ideal case ( say a 90
degree bent resistor with each leg being 10 times longer
than wide and then calculate the ideal corner factor. ( assume everything is
perfect ... )

Then Try rounding the corners by 1/10 of the grid and do an exact mesh
extraction that models this corner.

The accuracy of this corner factor is very suspect !!!

But I have spent many hours at different times being asked to include a very
precise cornerfactor.

Note that even though the new buzzwords at DAC of DFM and DFY are implicitly
only hooked into DSM issues,
I still faced these problems in 5u technologies .. and they had a hard time
telling me the corner rounding variations then!

.... Rant off
)

If you want accurate resistors (or matched ones ) making them rectangular
and larger than minimum helps/

-- Gerry

"mauricio" <mcontaldo@gmail.com> wrote in message
news:eb9b4633a06b41995255963e1d9bc05a@localhost.talkaboutcad.com...

	CADForums.net Forum Index -> Cadence	All times are GMT
Page 1 of 1