R
Member
A λ/4 whip with a 50 Ω feedpoint "should" be trivial. On a real board the input impedance lands anywhere from 12 to 90 Ω depending on decisions nobody wrote down. This is the checklist I wish someone had handed me in year two.
The five things that move impedance
1. Radial count/length. An ideal ground plane is infinite; a solid pour works until a resonant cavity (edge slot) couples. For a printed monopole: the *shape of the void under/near it* matters more than the plane area. A quarter-wave monopole assumes image plane ≈ mirror; if your plane edge is less than ~0.1 λ away on one side, radiation tilts and impedance shifts — both fine if the simulation knows it. Vias only work as an RF ground if: ≤ λ/10 apart at the highest band, ≥ 2–3 rows, short stubs terminated on the opposing layer (via stubs at GHz resonate if left blind).
2. Feed inductance. Whip-to-pad + ground lug: every 2 nH of ground-L moves a 2.4 GHz λ/4 a noticeable fraction. Keep the ground connection a via right at the sleeve, not "somewhere along the connector".
3. Matching network parasitics. That π-pad's shunt C has a via stub; at higher bands series-C becomes series-L and the "match" is gone.
4. Housing/reflector. Plastic boxes: nothing. Metal brackets near-field: 5–8 dB notches and polarization skew — measure inside final enclosure, always.
5. Cable common-mode on the feed. Coax fed monopole without chokes radiates with the cable braid as its "second half", so the antenna you tune is actually the cable you route. A 1–2 turn ferrite sleeve fixes more boards than any matching retune.
Quick empirical workflow
- Simulate the board stackup (even coarse 3D) with real connector model;
- Tune with the final casing/cables, and log gain patterns with the EUT at realistic orientations (not only boresight);
- Verify bandwidth shift vs hand effect: SAR-ish coupling test;
- Record: S11 sweep, radiation efficiency at center, realized gain vs angle (the number that decides link budget).
One trap specific to sub-GHz ISM bands
Designers tune a λ/4 on 868 MHz to S11 −15 dB with a 0-ohm jumper + tiny L... and ship boards whose "antenna" is half the PCB trace — because at λ/4 ≈ 86 mm nothing fits, so the trace becomes part of the monopole. Define the antenna boundary in the schematic with a keep-out and test the worst-case solder-jumper variant.
If you have a measured story of "impedance walked 3:1 from lot to lot" — the culprit is almost always the pad-mask/plane-etch tolerance near the feed — I'd love forum data.
The five things that move impedance
1. Radial count/length. An ideal ground plane is infinite; a solid pour works until a resonant cavity (edge slot) couples. For a printed monopole: the *shape of the void under/near it* matters more than the plane area. A quarter-wave monopole assumes image plane ≈ mirror; if your plane edge is less than ~0.1 λ away on one side, radiation tilts and impedance shifts — both fine if the simulation knows it. Vias only work as an RF ground if: ≤ λ/10 apart at the highest band, ≥ 2–3 rows, short stubs terminated on the opposing layer (via stubs at GHz resonate if left blind).
2. Feed inductance. Whip-to-pad + ground lug: every 2 nH of ground-L moves a 2.4 GHz λ/4 a noticeable fraction. Keep the ground connection a via right at the sleeve, not "somewhere along the connector".
3. Matching network parasitics. That π-pad's shunt C has a via stub; at higher bands series-C becomes series-L and the "match" is gone.
4. Housing/reflector. Plastic boxes: nothing. Metal brackets near-field: 5–8 dB notches and polarization skew — measure inside final enclosure, always.
5. Cable common-mode on the feed. Coax fed monopole without chokes radiates with the cable braid as its "second half", so the antenna you tune is actually the cable you route. A 1–2 turn ferrite sleeve fixes more boards than any matching retune.
Quick empirical workflow
- Simulate the board stackup (even coarse 3D) with real connector model;
- Tune with the final casing/cables, and log gain patterns with the EUT at realistic orientations (not only boresight);
- Verify bandwidth shift vs hand effect: SAR-ish coupling test;
- Record: S11 sweep, radiation efficiency at center, realized gain vs angle (the number that decides link budget).
One trap specific to sub-GHz ISM bands
Designers tune a λ/4 on 868 MHz to S11 −15 dB with a 0-ohm jumper + tiny L... and ship boards whose "antenna" is half the PCB trace — because at λ/4 ≈ 86 mm nothing fits, so the trace becomes part of the monopole. Define the antenna boundary in the schematic with a keep-out and test the worst-case solder-jumper variant.
If you have a measured story of "impedance walked 3:1 from lot to lot" — the culprit is almost always the pad-mask/plane-etch tolerance near the feed — I'd love forum data.
