AWR eBooks

Best Practices for Efficient and Effective Planar EM Simulation

Issue link: https://resources.system-analysis.cadence.com/i/1326178

Contents of this Issue

Navigation

Page 11 of 21

Best Practices for Efficient and Effective Planar EM Simulation 12 www.cadence.com/go/awr Chapter 3: Meshing Best Practice Tip #5: Understand how the mesh is set. The minimum facet size and mesh density can be controlled in a variety of ways. The AWR AXIEM simulator, or any other EM simulator, is only as good as its mesh. There are several ways to obtain a reasonable mesh, by which is meant one of reasonable size with well-formed triangles and rectangles. The best-behaved triangle is an equilateral one and the best rectangle is a square. Long, skinny rectangles and triangles can lead to badly condi- tioned matrices, which are difficult to solve. The use of geometry simplification on the layout before it goes to the mesher has already been discussed. The mesh itself can be controlled in several ways: the designer needs to understand how the mesh is created and how to set the size of the mesh. How is the mesh in AWR AXIEM software determined? The simulator uses heuristical algorithms based on the circuit environment, including frequency, substrate thickness, and dielectric constant. The word "heuristics" really means a good guess, but AWR software developers are relying on years of lessons learned in the development of this type of planar solver. For example, meshing a microstrip bend or via has been studied for over 30 years in the community. The user does want to help the AWR AXIEM simulator with the size of the mesh by specifying the minimum facet sizes needed. (Facet is the term used for one shape or cell of the mesh.) The software will then try not to go below this minimum facet size as it determines the mesh. The designer is faced with a tradeoff of performance versus accuracy in selecting the minimum facet size: the larger the facets are, the fewer are needed, and the simulation will run faster, but the results may not be sufficiently accurate. A good rule of thumb is to choose a minimum facet length of one-tenth of a typical line width in the circuit. The user shouldn't be too worried if different lines have different widths. For a distributed filter, choose one of the narrower sections. If the filter sections are tightly coupled with gaps narrower than the line widths, use the gap width instead. Typically, reasonable results require three to five facets across as one facet is usually not adequate to accurately calculate phase changes in the circuit. Remember, each facet is a linear approximation of the current, so enough of them must be used to capture the changing nature of the current across the line. In this spirit, one of the standard tricks is to use an edge mesh on lines. An example of an edge mesh for a microstrip line is shown in Figure 13. Figure 13: 2D view (left) and 3D view (right) of an edge mesh on a line and via The edge mesh is the narrow, rectangular mesh at the sides of the line. The current density on a microstrip is larger at the edges of the line, therefore a narrow mesh can more accurately model the current for the same number of facets. This is a good example of meshing being based on experience and this edge-mesh trick has been used for decades in planar simulators. To run on the edge mesh setting, go to the EM project options in the browser, and select the mesh tab where all the mesh control settings reside. Selecting the field for Edge Mesh Thin Metal will give an edge mesh and the user determines the width of the edge mesh by setting the minimum facet size. There is also a secondary setting in the meshing menu, No Edge Mesh for Thick Metal, for edge mesh for lines with non-zero thickness. Normally it is checked, and no edge mesh is used. It is not clear why an edge mesh should not be used for thick metal. Often, the default is overridden, and the edge mesh used. The argument can be made that even if the edge mesh isn't helping as much, it isn't hurting either. The designer might try running a simple problem with and without an edge mesh to see if it matters. The edge mesh can be set globally up in the Options menu under EM and every new EM structure in the AWR Microwave Office project will have the new edge meshing default. PDKs usually have their own settings as defaults. The AWR AXIEM simulator provides two different methods for setting the minimum facet size. The first method relates the minimum facet size to the grid of the layout and the second one sets the minimum size by assigning absolute dimensions in the mesh options. The grid of an EM layout is the dots on the screen and is used to help draw polygons. Lines naturally snap to the vertices when the grid is turned on to aid the designer in drawing shapes, resizing them, and moving them. A grid is used in almost all drawing software as an aid in drawing.

Articles in this issue

Links on this page

view archives of AWR eBooks - Best Practices for Efficient and Effective Planar EM Simulation