where is a linear operator, is known, and is to be determined. Let be expanded in a series of known functions , , ,...in the domain of , as where the are unknown constants to be determined. We shall call the expansion functions or basis functions. For exact solutions, the summation is infinite and the form a complete set of basis functions. For approximate solutions, finite summation is employed. Since is linear, we have . It is assume that a suitable inner product has been defined. Now definte a set of weighting functions, or testing functions, , , ,... in the range of , and take the inner product of the above equation with each . The result is , The set of equations can be written in matrix form as where , , If the matrix is nonsingular, its inverse exists. The are 1 then given by . The solution may be exact or approximate, depending upon the choice of and . The particular choice is known as Galerkins method. Example 1. Exact solution : Let Choose and . Then, We have . Then, . 2 For , . For , . For , 3 , which is the exact solution. 4 Example 2: Pulse function: Triangle function: Let , , then 5 6 HW#3: follow example 1, solve by letting Plot the result with the exact solution . 7 Antenna Problems In free space, Therefore, there exist a dyadic satisfies is called Greens function. In general, in any structure, Greens functions exist for computing the electric fields generated by a given source. Procedures: 1. Specify excitation . 2. Identify PEC. 3. Identify the induced surface current on the PEC due to and let 4. Satisfy B. C. On the PEC. 5. 8 where For , we have Similarly for slots on PEC, complementary formulation for magnetic currents exist. Let . If two regions are separated by PEC with slots on it, then the slots can be closed by putting equivalent magnetic currents on them. Let the regions be named and . Let be the normal vector pointing from to . Let the electric field on the slots be . Let the magnetic current in region and be and respectively. Then, At the slots, the magnetic field must be the same calculating from region and . We have Let 9 . Then, where For , we have Example: Slot-coupled microstripline-fed patch antennas 10 Pocklingtons Integral Equation for a dipole Let the length and the radius of the dipole be and , where such that we can ignore the circumferential current, therefore and . Assume the conductivity of the dipole is vary large that is a surface current exist on the surface of the dipole. Since , we can also assume doesnt vary circumferentially. That is is only a function of . Then where The approximation in the above equation is under the assumption that the surface current can be approximated by a line source at the center since . If only consider the field at surface of the dipole, . If an incident wave , the field will induce a current distribution on the dipole surface. The induce current will produce scattered field . The boundary condition requires that on the surface of the dipole 11 Integral Equations The equation derived in previous section is called Integral Equation. The right hand side is the known incident field , while the left hand side is the unknown current distribution to be solved. Approximate by a series of known expansion functions such that where are coefficients of the expansion functions and are unknown. For instance, choose impulse functions as defined below 12 Point Matching For conveniences, rewrite the integral equation as
Let be the mid-point of each segment, then Rewriting in matrix form, we have 13 where For , we have Source Modeling 1. Delta gap: or for . 2. Frill Generator: where 3. Plane wave: 14 15 The input impedance becomes if the applied voltage is located at -th section. 16 17 18 Weighted Residuals and the Method of Moments Let weighting functions be a set of known functions. Instead of point-matching, the integration equation is integrated with the weighting function as follow: The matrix formulation will be the same except . Obviously, if , the result is the same as point-matching. This general formulation is called the Method of Moments. If the weighting functions are chosen to be the same as the expansion function, i.e., , its called Galerkins method. In most cases, Galerkins method gives better result than point-matching. 19 Other Expansion and Weighting Functions 1. Triangle functions (piecewise linear) 2. Piecewise sinusoidal: 3. Sinusoidal interpolation: 20 Adding Lumped Loads If an ideal lumped load is inserted into a wire antenna at coordinate having current , the electric field across can be considered as a delta function having amplitude . The original integral equation becomes Therefore, the new impedance matrix is related to the unloaded one by 21 22 Method of Moment in Electrostatic Applications Example: Stripline Capacitance For a line charge , the voltage generated by this charge is The potential produced by a distribution of charge is If the distribution of the charge is discretized to sections as in the figure, let the charge density be constant in a cell and be denoted as . Let and be the coordinate and potential at the center of section . Let be the length of section . We have where 23 As shown in the figure, let the plates be divided to 36 sections with the same section length . Eight sections are on the central strip and 28 sections are on the top and bottom strips. Also let and . Therefore, By solving the above equation, the capacitance can be found by dividing the total charge with voltage . HW# 4 Let , find the capacitance. Compared with the formula in Pozars Microwave Engineering, pp 154157. 24