Two-dimensional Networks using C

To test larger systems, a C program was written. The extra flexibility of C was needed in order to handle the much larger arrays and the more complex interconnections. The source code for this program is on the floppy disk included with this thesis. After SPICE had performed its simulation, there was an output file with lists of node voltages and currents. ICAP is designed to display and analyse transient voltages at individual nodes, but for viewing simple operating points over a grid of nodes, a separate program was written to interpret the output files. This program reads in the voltage and current tables and outputs several postscript files with the input image, the output image, and an output image with dotted lines added representing those resistive fuses that have tripped. The source code for this program is also on the floppy disk attached to this thesis.

One of the first simulations carried out was on a two by three pixel array as shown in Figure 5.4. The first image shows the input data, with high voltages shown as light shades and low voltages shown in dark shades. The input data was designed to show the flaws inherent with the simple resistive fuse grid. The most striking feature of the input image is that it consists of a light half and a dark half. However, the fact that the image is more grey at the centre is likely to cause the resistive fuse network to miss this fact.

Figure 5.4: Simulation Results of Six Pixel Array.
\begin{figure}\par
\begin{center}
\begin{tabular}{ccc}
\psfig{figure=spresults/s...
...al & Segmentation\\
\vspace{0.5cm} & \\
\end{tabular}\end{center}
\end{figure}

The second image shows the output of the resistive network. Because segmentation of the image is incomplete, the whole image has been smoothed. Thus, there is less contrast between the left side and the right side. The final image shows the segmentation as dotted lines. The central segment of the line running down the middle has not tripped, because the voltage difference is not large enough. A human would say that either the image is of a dark half and a light half, or that the image is of a grey line running horizontally. The network has not produced either result, it being half one and half the other.

Matthew Exon 2004-05-23