next up previous contents
Next: 9.3.5 Running the Inverse Up: 9.3 Inverse Modeling of Previous: 9.3.3 Deriving a Metric

9.3.4 Handling Multiple Devices and Bulk Biases

Since we have measurements for devices of different gate lengths, we create a network-model as depicted in Figure 9.7 which contains multiple instances of the match evaluation model ("match.mod").
\begin{Figure}
% latex2html id marker 9387\centering
\includegraphics[width=0....
...omprehensive match vector at the output port
named \textit{match}.}
\end{Figure}

\begin{Example}
% latex2html id marker 9395\centering\scriptsize
\begin{minipa...
...he global match
(see also Figure~\ref{fig::invmod-global-match}).}
\end{Example}

Each of these instances has individual settings for the default values at the ports gatelength, vb, idvg-refcurve, and idvd-refcurve. By means of this configuration we are able to include all our measurements into the inverse modeling procedure. Each of these submodels evaluates our doping profile (defined by the settings parameters at the model ports) for different devices with a specific gate length measured at specific values of bulk bias. To derive an overall measure of match we again combine all match metric vectors into a single one by concatenating them.



Rudi Strasser
1999-05-27