L
Admin
[Technical article, rfforum Editorial Team]
A coax probe through a thick substrate adds series inductance that the simple cavity model never shows. You can land the feed at the textbook resistance point and still see an inductive loop on the Smith chart.
1. Where the inductance comes from
The probe is a vertical current through a dielectric slab. Its inductance scales roughly with substrate height and with ln(pitch/probe diameter). On thin FR-4 or Rogers boards (h << λ) it is a small series jX; on thick foam or stacked patches it can dominate the match.
2. What it does to S11
Series L rotates and stretches the resonance loop. The frequency of best |S11| shifts, and the real-part-only feed formula under-predicts how far from the edge you must go. People then chase patch length while the real issue is probe reactance.
3. Fixes that usually work
- Shorten the probe (thinner stack, or a recessed connector).
- Add a series capacitive compensation (gap, series C, or a capacitive hat on the probe).
- Switch to an inset or aperture feed when h is large and bandwidth already forces a thick stack.
- Simulate the probe as a real coax, not a lumped port on the patch surface.
4. Bench habit
After etch, if the resonance is right but the loop sits above the real axis, do not only slide the feed hole — measure or EM-estimate probe L first. Two boards with different connector pin lengths often explain "the same Gerber, different match".
Discussion: on your last thick-patch build, how much series L did you attribute to the probe, and did you compensate in EM or with a discrete C?
— rfforum 编辑部 / Editorial Team
A coax probe through a thick substrate adds series inductance that the simple cavity model never shows. You can land the feed at the textbook resistance point and still see an inductive loop on the Smith chart.
1. Where the inductance comes from
The probe is a vertical current through a dielectric slab. Its inductance scales roughly with substrate height and with ln(pitch/probe diameter). On thin FR-4 or Rogers boards (h << λ) it is a small series jX; on thick foam or stacked patches it can dominate the match.
2. What it does to S11
Series L rotates and stretches the resonance loop. The frequency of best |S11| shifts, and the real-part-only feed formula under-predicts how far from the edge you must go. People then chase patch length while the real issue is probe reactance.
3. Fixes that usually work
- Shorten the probe (thinner stack, or a recessed connector).
- Add a series capacitive compensation (gap, series C, or a capacitive hat on the probe).
- Switch to an inset or aperture feed when h is large and bandwidth already forces a thick stack.
- Simulate the probe as a real coax, not a lumped port on the patch surface.
4. Bench habit
After etch, if the resonance is right but the loop sits above the real axis, do not only slide the feed hole — measure or EM-estimate probe L first. Two boards with different connector pin lengths often explain "the same Gerber, different match".
Discussion: on your last thick-patch build, how much series L did you attribute to the probe, and did you compensate in EM or with a discrete C?
— rfforum 编辑部 / Editorial Team
