--- /dev/null
+/*
+The code below for calculating the voltage at a point i,j, given the
+voltages and permittivities at the four adjacent points (i,j+1),
+(i,j-1), (i-1,j), and (i+1,j), although I am not convinced it is
+100% correct in the case of mulitple dielectrics, the errors are
+still very small.
+*/
+
+
+#include "config.h"
+
+#ifdef HAVE_STDLIB_H
+#include <stdlib.h>
+#endif
+
+#ifdef HAVE_STDIO_H
+#include <stdio.h>
+#endif
+
+#include "definitions.h"
+
+
+extern int width, height;
+extern double **Er;
+extern unsigned char **oddity;
+extern int dielectrics_to_consider_just_now;
+extern double r;
+extern int coupler;
+
+#include "exit_codes.h"
+
+/* The following function updates the voltage on the matrix V_to given data about the
+oddity of the location i,j and the voltages in the matrix V_from. It does this for n interations
+between rows jmin and jmax inclusive and between columns imain and imax inclusive */
+
+void update_voltage_array(int nmax, int imin, int imax, int jmin, int jmax, double **V_from, double **V_to)
+{
+ int k, i, j, n;
+ unsigned char oddity_value;
+ double Va, Vb, Vl, Vr, ERa, ERb, ERl, ERr;
+ double Vnew, g;
+
+ if (dielectrics_to_consider_just_now==1)
+ g=r;
+ else
+ g=1;
+ for(n=0; n < nmax; ++n)
+ for(k=0; k < 4; ++k)
+ for (i = k&1 ? imax : imin; k&1 ? i >=imin : i <= imax ; k&1 ? i-- : i++)
+ for (j = (k==0 || k ==3) ? jmin : jmax; (k ==0 || k == 3) ? j <= jmax : j >= jmin ; (k == 0 || k ==3) ? j++ : j--){
+ oddity_value=oddity[i][j];
+
+ if( oddity_value == CONDUCTOR_ZERO_V ){
+ V_to[i][j]=0.0;
+ }
+
+ else if( oddity_value == CONDUCTOR_PLUS_ONE_V ){
+ V_to[i][j]=1.0;
+ }
+
+ else if( oddity_value == CONDUCTOR_MINUS_ONE_V ){
+ V_to[i][j]=-1.0;
+ }
+
+ else if( oddity_value == TOP_LEFT_CORNER ) { /* top left */
+ Vnew=0.5*(V_from[1][0]+V_from[0][1]);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+ else if( oddity_value == TOP_RIGHT_CORNER ) {
+ Vnew=0.5*(V_from[width-2][0]+V_from[width-1][1]); /* top right */
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if(oddity_value == BOTTOM_LEFT_CORNER) {
+ Vnew=0.5*(V_from[0][height-2]+V_from[1][height-1]); /* bottom left */
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == BOTTOM_RIGHT_CORNER) {
+ Vnew=0.5*(V_from[width-2][height-1]+V_from[width-1][height-2]); /* bottom right */
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ /* Now the sides */
+
+ else if( oddity_value == ORDINARY_POINT_LEFT_EDGE ){ /* left hand side */
+ Vnew=0.25*(V_from[0][j-1]+V_from[0][j+1] + 2*V_from[1][j]);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == ORDINARY_POINT_RIGHT_EDGE){ /* right hand side */
+ Vnew=0.25*(V_from[width-1][j+1]+V_from[width-1][j-1]+2*V_from[width-2][j]);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == ORDINARY_POINT_TOP_EDGE ){ /* top row */
+ Vnew=0.25*(V_from[i-1][0]+V_from[i+1][0]+2*V_from[i][1]);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == ORDINARY_POINT_BOTTOM_EDGE ){ /* bottom row */
+ Vnew=0.25*(V_from[i-1][height-1]+V_from[i+1][height-1]+2*V_from[i][height-2]);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if ( oddity_value == ORDINARY_INTERIOR_POINT || (oddity_value>=DIFFERENT_DIELECTRIC_ABOVE_AND_RIGHT && oddity_value < UNDEFINED_ODDITY && dielectrics_to_consider_just_now == 1) ) {
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=(Va+Vb+Vl+Vr)/4.0;
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ /* I'm not sure the following equations, which compute the voltage
+ where there is a metal around are okay. One line of thought would
+ say that the same equations as normal i.e.
+ v_new=(v(i+1,j_+v(i-1,j)+v(i,j-1)+v(i,j+1))/4 should be used
+ but then since the electric field across the metal surface is zero,
+ the equation that was used to derrive that equation is not valid.
+
+ Another thought of mine is that voltage near a metal will be more affected
+ by the metal than the dielectric, since the nearest part of the metal is at
+ at the same voltage as the node, whereas for a dielectric is less so. Hence
+ the following seems a sensible solution. Since the metal will have twice
+ the effect of a dielectric, the voltage at i,j should be weighted such
+ that its effect is more strongly affected by the metal. This seems to
+ produce reasonably accurate results, but whether this is chance or not
+ I don't know. */
+
+ else if( oddity_value == METAL_ABOVE ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Va/3+2*Vb/3+Vl+Vr);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_BELOW ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vb/3+2*Va/3+Vl+Vr);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_LEFT ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vl/3+2*Vr/3+Va+Vb);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_RIGHT ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vr/3+2*Vl/3+Va+Vb);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_ABOVE_AND_RIGHT ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vr/3+4*Va/3 +2*Vl/3 + 2*Vb/3);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_ABOVE_AND_LEFT ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vl/3+4*Va/3+2*Vr/3+2*Vb/3);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_BELOW_AND_LEFT ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vl/3+4*Vb/3+2*Vr/3+2*Va/3);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ else if( oddity_value == METAL_BELOW_AND_RIGHT ){
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ Vnew=0.25*(4*Vb/3+4*Vr/3+2*Va/3+2*Vl/3);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+ }
+
+ /* Again, when there is a change of permittivity, my equations may
+ (probably are wrong). My logic is that if there's and RF field,
+ the impedance is inversly proportional to Er. So if the material
+ above a node is of a higher permittivity, then the
+ voltage will be closer to that of the node above, becuase of this.
+ The same applies for other directions of change in Er. */
+
+
+ else if(dielectrics_to_consider_just_now > 1){
+
+ Va=V_from[i][j-1];
+ Vb=V_from[i][j+1];
+ Vl=V_from[i-1][j];
+ Vr=V_from[i+1][j];
+
+ ERa=Er[i][j-1];
+ ERb=Er[i][j+1];
+ ERl=Er[i-1][j];
+ ERr=Er[i+1][j];
+
+ Vnew=(Va * ERa + Vb * ERb + Vl * ERl + Vr * ERr)/(ERa + ERb + ERl + ERr);
+ V_to[i][j]=g*Vnew+(1-g)*V_from[i][j];
+
+ }
+ else if ( (dielectrics_to_consider_just_now == 1 && oddity_value == UNDEFINED_ODDITY) || (dielectrics_to_consider_just_now > 1)) {
+ fprintf(stderr,"Internal error in update_voltage_array.c\n");
+ fprintf(stderr,"i=%d j=%d oddity[%d][%d]=%d\n",i,j,i,j,oddity[i][j]);
+ exit(INTERNAL_ERROR);
+ } /* end if if an internal error */
+
+ } /* end of j loop */
+}
projects / debian / atlc / blobdiff