| |
|
|
 |
Simulation methods and techniques ( 40 ~ 60 minutes, 169 pages ) |
| |
|
Table of contents
Chapter 1 Conformal FDTD Method
Chapter 2 Boundary Conditions
Chapter 3 Adaptive Mesh
Chapter 4 3-D Modeling
Chapter 5 Parallel Processing
Chapter 6 Web Based Interface (WBI)
Chapter 7 Data-post Processing
Chapter 8 Parameter Optimization
Chapter 9 GEMS Box System
Chapter 10 Hardware Acceleration Techniques
Chapter 11 Practical Examples
|
| |
|
|
| |
|
|
|
Basic steps in the GEMS simulation |
| |
1. |
Select the project unit (default unit is millimeter). Although you can use any unites to input a parameter, the project unite cannot be changed any more once you start the project design. Use the project unit as close to the minimum dimension in the domain as possible to ensure a correct model. You should avoid using the cell size like 0.000001 since the GEMS tolrence is 1.0e-5. |
| |
2. |
Select the type of excitation pulse, the modulated Gaussian is a good choice for the most applications. The width of pulse spectrum should cover the frequency band of interest. Since the actual sampling time window should be larger than the summation of the pulse width and system response time, we usually select a narrow pulse so that the simulation time window becomes narrower to reduce the simulation time. |
| |
3. |
input objects, excitation sources and output options, design the mesh distribution, and specify the boundary condition. The far field and plane wave excitation settings require the mesh information, therefore, you need to design the mesh before you specify the plane wave excitaiton and Huygens' box. |
| |
4. |
Design the processor distribution in the parallel processing for the GEMS cluster version. |
| |
5. |
Save the GEMS project |
| |
6. |
Specify the number of time steps and convergent condition in the PCF/GPV file generation window.
(a) For GEMS PC version, click on the Start button to generate the PCF file;
(b) For GEMS cluster version, check the Validation Only box in the Pre-Calculate File box, and then click on the Start button to generate the GPV file (PCF file will be gnerated in the cluster using domain decomposition technique. |
| |
7. |
Click on the Ok button to close the PCF file generation window. |
| |
8. |
For the GEMS PC version, launch the GEMS Solver module, and select the PCF file in the GEMS solver window. Click on the Start button to start the GEMS simulation. GEMS solver will use the all the cores inside this computer to simulate this problem. You will have chances to extend the simulation before the simulation completed. |
| |
9. |
During the simulation, or after the simulation is completed, you can launch the GEMS display module to view and process the GMES results. GEMS display module allows you to open more than one GEMS project so that you can compare the results generated from different projects. |
| |
10. |
For the GEMS cluster version, launch FireFox or EI web browser, and tyoe the cluster address in the adress box, and then login the cluster using your userID and password. Zip the GPV file in the project folder in the local PC, create a new folder in the cluster and upload the zipped GPV file to the cluster. |
| |
11. |
Simulate the project using GEMS solver in the cluster. Zip the output foder and GDX file, and then Download them to your local PC. Launch the GEMS display module to view and process the GEMS results. |
|
Distance selection between excitation and output position |
| |
A uniform microstripline is excited by using a voltage between the microstripline and PEC ground plane, and the voltage and current outputs are measured at the same position along the propagation direction. The computational domain is truncated by using a 6-layer CPML in the five directions except the PEC ground wall. The reflection cofficient calculated from the voltage and current outputs varies with the position of the outputs, and is plotted in the figure below.
*we did not optimize the cell size, domain size, and excitation location for the low reflection coefficient.
|
| |
|
|
|
Reflection from lumped element |
| |
A uniform microstripline is terminated by using a 50 Ohm lumped element at one end, and the reflection from the mismatched lumped element is plotted in the figure below (assuming the no reflection from the CPML truncation). A lumped element can be used as an inner resistance of a voltage source or terminator of a port.
|
| |
|
|
|
Terminator (matched load) |
| |
A coax is terminated by using a terminator at one end, and the reflection from the mismatched terminator is plotted in the figure below (assuming the no reflection from the CPML truncation). If you need to calculate a 3-D far field pattern, you cannot let the Huygens' box to truncate the coax or waveguide feed, the terminator (matched load) allows to truncate a coax or waaveguide insode the computational domain.
|
| |
|
|
|
Object order |
| |
The latter input object will replace the former obect in the overlapped region, therefore, the object order is very important. When any infinitely thin objects are involved in the domain, the object is critical when the infinitely thin object tuches other objects. The wrong order may remove the infinitely thin object completely. For example, the correct order should be the substrate of a patch antenna first and then the patch and feed line. Change this order, you will not see the infinitely thin patch or feed line.
The excitation, output and lumped element can neither be overlapped by other objects or options, nor overlap other options or objects, namely, their property is not affected by their order.
|
| |
|
|
|
3-D animation |
| |
Specify a field/current output plane in the time domain in the GEMS designer, you can view the variation of field distribution with the time after the simulation is completed or during the simulation. In the GEMS display module, open the GEMS project file *.GPV and GDX, open the project configuration (set transparent), and, select the feld distribution option in the result tree and then click on the "Add current window" in the toolbar. Click on the Play button to view the variation of field distribution on the selected plane. You can export any distribution to a data file and then plot or process it in MatLab software (surf or Mesh command).
Follow the similar procedure to view the field distribution in the frequency domain. Even it is one shot, you still need to click on the Play bitton to view the field distribution at the pre-specified frequency on the selected palne.
|
| |
|
|
Understand the field distribution along a line |
| |
Field distribution in the time domain along a line indicates the field variation with both time and posion, which means that the x- and y-axes indicate the time and position, respectively. Using the Pickup Curve option in the Result menu in the Toolbar in the display module, you can view the field distribution varies with time or position.
Field distribution in the frequency domain along a line indicates the field variation with both frequency and posion, which means that the x- and y-axes indicate the frequency and position, respectively. Using the Pickup Curve option in the Result menu in the Toolbar in the display module, you can view the field distribution varies with frequency or position.
 |
| |
|
|
|
Correctly use the power calculation in GEMS software |
| |
Power conservation can be expressed as:
where:
For a lossless system, the real part of power supplied by excitation source equals to radiation power or the power going through a closed surface that encloses the system. The power conservation can be validated for a lossy system.
There are two important things in the GEMS simulation:
(1) If a feed system such as coax, waveguide or microstripline,is open to PML or matched load (lumped element or terminator), only half of the incident power is supplied to the system, therefore, the efficiency should be doubled and the gain should be added 3 dB in the data processing.
(2) You should avoid placing an excitation source inside PEC object, otherwise, the incident power may be not accurate.
|
| |
|
|
