Advanced mesh and boundary conditions
Advanced mode opens up every parameter Auto mode picks for you. This page walks through the panels in the order they appear on the Simulate tab: mesh and cell sizes, PML, air margins, energy ringdown, NF2FF, and field export. The same parameters back the Auto run. Advanced mode lets you read and edit each one directly.
How the frequency band sets the floor
Before tweaking anything, it helps to know which numbers the band ends up driving:
- Fmax drives cell count. The mesh has to resolve the shortest wavelength with at least 15 cells per lambda in the densest dielectric. Doubling Fmax roughly halves the cell size on every axis.
- Fmin drives runtime. The Gaussian pulse has to ring out at the slowest cycle before the simulation can stop, so a lower Fmin means more time steps.
- Fmin also drives air margins. Low frequencies need more space between the geometry and the PML to avoid near-field clipping at the boundary.
Mesh and Cell Size panel
The Mesh and Cell Size panel shows the actual mesh RayRF will use. The inner grid is sized to the dielectric. The outer cells expand toward the PML at the configured grading ratio.
- dx / dy / dz. Explicit Yee cell size on each axis in mm. Smaller cells mean more accuracy and more VRAM.
- Refine margin. Extra cells held around fine features so the resolution does not snap back to the base grading immediately at the edge.
- Per-axis multipliers. Scale dx, dy, or dz independently. Useful when one axis (usually Z) needs a finer cell for a thin dielectric.
- Mesh grading. When on, the outer cells grow by the grading ratio as they march toward the PML, which keeps cell count down without changing the inner resolution.
Boundary Conditions panel
The only boundary condition RayRF currently runs is Open Air (PML): a perfectly matched layer that absorbs outgoing waves so they do not reflect back into the simulation. The Boundary Conditions panel exposes:
- PML thickness in cells. Default 8 to 12. Bump to 16 to 20 if you see PML reflection signatures.
- Auto PML. Lets RayRF pick the thickness from Fmin and the air margin. Turn off if you want to hand-set the value.
- Cells per mode. Controls how aggressively the PML grades into the absorbing region. Higher is more absorbent at the cost of cells.
Air Margins panel
Air margins set how much empty space sits between the geometry and the PML on each face. Each face has its own value in mm. Warning triangles appear in the panel when a margin is too narrow for the lowest frequency in the band.
Energy Ringdown panel
The energy ringdown criterion decides when the FDTD run can stop. The simulation tracks total energy in the volume and stops once it drops below the configured ratio of the peak.
- Target energy ratio. Default 1e-4 of peak. A lower target waits for cleaner decay at the cost of runtime.
- Max steps. Hard upper bound on the time step count regardless of ringdown. Acts as a safety stop.
- Early-stop fraction. Optional. Stops a run sooner once the energy curve is plainly past its tail.
NF2FF panel
The near-field-to-far-field transform is what populates the Radiation Pattern tab.
- Compute Radiation Pattern. Master switch. Off means no radiation data.
- Frequency points. How many frequencies the transform is run at across the band. Patterns are stored per point and the Radiation tab interpolates between them.
- Angular resolution. Theta and phi step size on the far-field sphere. Finer steps cost a bit of CPU and disk.
Field Export panel
- Save Surface Currents. Stores per-frequency surface-current snapshots for the Field Viewer.
- Frequency-domain points. How many discrete frequencies the field viewer can scrub between.
- Capture every N steps. Sub-samples the time-domain field for the Field Viewer's Time mode. Smaller N gives smoother playback at the cost of disk.
- Export full E/H fields (expensive). Writes the dense volumetric field. Off by default. Output can run from hundreds of megabytes into several gigabytes per project.
Mesh preview
Before committing to a long run, click Preview Mesh on the Simulate tab.
RayRF builds the first mesh pass and renders it in 3D so you can confirm the cell density around the features that matter. Zoom in to see the per-cell grid and catch under-resolved features, such as a thin slot that ends up with only one cell across it, before the convergence study points them out later. The final mesh is refined further during the actual run.
Reading the convergence study
With Perform Convergence Study on, the right-bottom panel reports the delta between the first and second mesh passes at a key resonance frequency. A small delta means the answer is mesh-stable. A large one means you need more cells.