L
Admin
[Technical article, rfforum Editorial Team]
Radiation-box size and absorber thickness are two of the easiest ways to get a beautiful but wrong antenna simulation. The solver will still converge; the answer just describes the box, not free space.
1. Distance to the boundary
A common starting rule is at least λ/4 at the lowest frequency from any radiator or significant current to the absorbing face, then check sensitivity. Electrically large arrays, high-gain horns and endfire antennas often need more clearance in the main-beam direction than a short monopole does. If moving the box by λ/8 changes gain by more than your uncertainty budget, the box is still part of the antenna.
2. Thickness and absorber quality
Thin absorbers reflect. At the low end of a broadband sim the electrical thickness shrinks and reflections rise; that shows up as ripple on S11 and as standing-wave artefacts on realised gain. Prefer the vendor's broadband absorber model (or a measured reflectivity table) over a single-frequency "perfect" ABC when you sweep more than an octave.
3. What to watch besides S11
S11 can look calm while gain, efficiency or front-to-back still walk with box size. Converge the metric you will publish: realised gain / total efficiency for antennas, not only match. Plot the same cut after ±λ/8 box moves; if the pattern lobes move, fix the domain before you tune the geometry.
4. Symmetry and ground planes
Infinite ground approximations and finite boards need different boxes. A finite ground with edge diffraction needs lateral clearance too; shrinking only the +Z height is a common false economy.
5. Practical checklist
1. Set box from λ_min, not from "what fitted last time".
2. Sweep box size once on the frozen mesh settings; record Δgain and ΔS11.
3. Only then run adaptive mesh / parametric geometry sweeps.
4. Re-check box when you change frequency band or add a reflector/director.
Discussion: for a sub-6 GHz PCB antenna with a short ground, how large a radiation box do you actually freeze before you trust a ±0.3 dB gain number?
— rfforum 编辑部 / Editorial Team
Radiation-box size and absorber thickness are two of the easiest ways to get a beautiful but wrong antenna simulation. The solver will still converge; the answer just describes the box, not free space.
1. Distance to the boundary
A common starting rule is at least λ/4 at the lowest frequency from any radiator or significant current to the absorbing face, then check sensitivity. Electrically large arrays, high-gain horns and endfire antennas often need more clearance in the main-beam direction than a short monopole does. If moving the box by λ/8 changes gain by more than your uncertainty budget, the box is still part of the antenna.
2. Thickness and absorber quality
Thin absorbers reflect. At the low end of a broadband sim the electrical thickness shrinks and reflections rise; that shows up as ripple on S11 and as standing-wave artefacts on realised gain. Prefer the vendor's broadband absorber model (or a measured reflectivity table) over a single-frequency "perfect" ABC when you sweep more than an octave.
3. What to watch besides S11
S11 can look calm while gain, efficiency or front-to-back still walk with box size. Converge the metric you will publish: realised gain / total efficiency for antennas, not only match. Plot the same cut after ±λ/8 box moves; if the pattern lobes move, fix the domain before you tune the geometry.
4. Symmetry and ground planes
Infinite ground approximations and finite boards need different boxes. A finite ground with edge diffraction needs lateral clearance too; shrinking only the +Z height is a common false economy.
5. Practical checklist
1. Set box from λ_min, not from "what fitted last time".
2. Sweep box size once on the frozen mesh settings; record Δgain and ΔS11.
3. Only then run adaptive mesh / parametric geometry sweeps.
4. Re-check box when you change frequency band or add a reflector/director.
Discussion: for a sub-6 GHz PCB antenna with a short ground, how large a radiation box do you actually freeze before you trust a ±0.3 dB gain number?
— rfforum 编辑部 / Editorial Team
