L
Admin
[Technical article / FAQ, rfforum Editorial Team]
Antenna pattern and gain plots from EM tools often look smooth long before the domain is large enough. An absorbing boundary or radiation box that sits too close to the radiator reflects energy back into the near field, and the "far-field" number becomes a function of box size rather than of the antenna.
1. What domain size actually buys you
- Too-small radiation boxes couple boundary reflections into gain and null depth; refining mesh inside a bad box mostly burns CPU
- Frequency matters: electrically large antennas and arrays need more free space than a single patch at the same mesh density
- Ground planes and cables that leave the box create artificial truncation currents that show up as pattern ripples
2. Freeze list for a pattern / gain claim from EM
- Box / PML / absorbing distance in wavelengths at the lowest frequency you publish, on the same sheet as gain
- Whether you swept box size and saw gain / null depth converge — one box size is not a convergence study
- Feed and cable model: a truncated coax or lumped port that does not match the measured launch will not converge by enlarging the box alone
- Mesh refinement focused on the radiator and launch, not uniform fine mesh everywhere
3. Habit
Treat box-size sweeps like mesh convergence: if gain moves more than your reporting precision when you enlarge the domain, you do not yet have a pattern number. Archive the final box size next to every published dBi claim. The same discipline applies to pattern null depth — a null that fills in when the box grows was never a real null.
Discussion: on your last sim–chamber disagreement, did the fix land in radiation-box size, feed/cable model, or material / ground detail — and do you require a box-size sweep before freezing gain?
— rfforum 编辑部 / Editorial Team
Antenna pattern and gain plots from EM tools often look smooth long before the domain is large enough. An absorbing boundary or radiation box that sits too close to the radiator reflects energy back into the near field, and the "far-field" number becomes a function of box size rather than of the antenna.
1. What domain size actually buys you
- Too-small radiation boxes couple boundary reflections into gain and null depth; refining mesh inside a bad box mostly burns CPU
- Frequency matters: electrically large antennas and arrays need more free space than a single patch at the same mesh density
- Ground planes and cables that leave the box create artificial truncation currents that show up as pattern ripples
2. Freeze list for a pattern / gain claim from EM
- Box / PML / absorbing distance in wavelengths at the lowest frequency you publish, on the same sheet as gain
- Whether you swept box size and saw gain / null depth converge — one box size is not a convergence study
- Feed and cable model: a truncated coax or lumped port that does not match the measured launch will not converge by enlarging the box alone
- Mesh refinement focused on the radiator and launch, not uniform fine mesh everywhere
3. Habit
Treat box-size sweeps like mesh convergence: if gain moves more than your reporting precision when you enlarge the domain, you do not yet have a pattern number. Archive the final box size next to every published dBi claim. The same discipline applies to pattern null depth — a null that fills in when the box grows was never a real null.
Discussion: on your last sim–chamber disagreement, did the fix land in radiation-box size, feed/cable model, or material / ground detail — and do you require a box-size sweep before freezing gain?
— rfforum 编辑部 / Editorial Team
