L
Admin
[Technical article / FAQ, rfforum Editorial Team]
A via fence drawn as a continuous wall on the layout can still leave a cavity that rings. Pitch, depth, return path breaks, and openings for connectors invent modes that show up as narrow spurs or gain peaking long after the schematic looked clean.
1. What the fence actually does
- Via pitch must be small relative to the highest frequency of interest in the dielectric. A fence that is fine at 6 GHz can be a picket fence at 40 GHz.
- A fence that stops at a connector cutout or a bias opening is an aperture. The cavity does not care that the rest of the wall looks dense.
- Return current that detours around a slot redefines the cavity size. A "stitched" keep-out with a broken reference is two cavities sharing a slot antenna.
2. Freeze list before you add lossy material everywhere
- Via pitch, stack depth, and whether the fence lands on a continuous reference on both sides of the cavity
- Near-field or cavity probe: does the spur peak move when you touch a known opening?
- A temporary copper tape or gasket over the largest aperture: if the spur dies, stop the schematic ECO
- Compare spur frequency to a rough cavity estimate (length/width/height) before blaming an IC
3. Habit
Never close a "mystery spur over ground pour" without pitch, opening map, and an aperture A/B. Archive fence pitch and the opening that killed the mode next to every cavity claim. If tape over a connector cutout kills it, the IC pin was never the patient.
4. Practical five-minute gate
- Measure largest opening and via pitch vs wavelength in dielectric
- Tape the biggest aperture; re-measure the spur
- Only then add absorbers or schematic changes
Discussion: on your last cavity spur, did the fix land in via pitch, an opening, or a return-path break — and do you require an aperture map before approving a fence-only layout?
— rfforum 编辑部 / Editorial Team
A via fence drawn as a continuous wall on the layout can still leave a cavity that rings. Pitch, depth, return path breaks, and openings for connectors invent modes that show up as narrow spurs or gain peaking long after the schematic looked clean.
1. What the fence actually does
- Via pitch must be small relative to the highest frequency of interest in the dielectric. A fence that is fine at 6 GHz can be a picket fence at 40 GHz.
- A fence that stops at a connector cutout or a bias opening is an aperture. The cavity does not care that the rest of the wall looks dense.
- Return current that detours around a slot redefines the cavity size. A "stitched" keep-out with a broken reference is two cavities sharing a slot antenna.
2. Freeze list before you add lossy material everywhere
- Via pitch, stack depth, and whether the fence lands on a continuous reference on both sides of the cavity
- Near-field or cavity probe: does the spur peak move when you touch a known opening?
- A temporary copper tape or gasket over the largest aperture: if the spur dies, stop the schematic ECO
- Compare spur frequency to a rough cavity estimate (length/width/height) before blaming an IC
3. Habit
Never close a "mystery spur over ground pour" without pitch, opening map, and an aperture A/B. Archive fence pitch and the opening that killed the mode next to every cavity claim. If tape over a connector cutout kills it, the IC pin was never the patient.
4. Practical five-minute gate
- Measure largest opening and via pitch vs wavelength in dielectric
- Tape the biggest aperture; re-measure the spur
- Only then add absorbers or schematic changes
Discussion: on your last cavity spur, did the fix land in via pitch, an opening, or a return-path break — and do you require an aperture map before approving a fence-only layout?
— rfforum 编辑部 / Editorial Team
