L
Admin
[Technical article / FAQ, rfforum Editorial Team]
A coax-to-microstrip or SMP-to-board launch is where many S-parameter arguments start. A lumped port at the pad is fast and often wrong above a few GHz; a wave port on a correctly sized coax cross-section with a clear reference plane is slower and usually the number you can trust against a measured thru.
1. When each port earns its keep
- Lumped port: early matching / DC-block exploration at low GHz when the launch is electrically short and you only need relative trends
- Wave port: connector launches, vertical transitions, and any claim that will be compared to a calibrated VNA plane — mode purity and reference-plane position own the phase
- Mixed mistake: wave-port size that admits higher-order modes, or a lumped port placed after a via stub that the model never named
2. Freeze list for a launch S-parameter claim
- Port type (wave vs lumped), reference-plane note, and whether a de-embed / thru cal matched the hardware plane
- Mesh convergence on the launch (not only on the DUT body)
- Connector PN / footprint and whether the model included the real dielectric and outer geometry
- Overlay: simulated launch vs measured thru or known-good launch coupon
3. Habit
Never freeze a filter or antenna "EM pass" whose launch was a lumped guess at a frequency where the connector is a significant fraction of a wavelength. Name the port class on the traveler. A 0.1 mm reference-plane error at 20 GHz is already a non-trivial phase argument in a filter dispute.
4. Practical gate
- Below your house frequency (often ~3–6 GHz depending on launch size), lumped exploration is fine for relative ranking
- Above that, require wave-port + measured thru overlay before accepting absolute S11/S21 for release
- If EM and VNA disagree only in phase, check reference plane first; if magnitude disagrees, check footprint dielectric and outer geometry before mesh density theatre
Discussion: on your last launch mismatch between EM and VNA, did the fix land in wave-port sizing, reference-plane shift, or a footprint detail — and do you forbid freezing launch S-params from a lumped port above your house frequency gate?
— rfforum 编辑部 / Editorial Team
A coax-to-microstrip or SMP-to-board launch is where many S-parameter arguments start. A lumped port at the pad is fast and often wrong above a few GHz; a wave port on a correctly sized coax cross-section with a clear reference plane is slower and usually the number you can trust against a measured thru.
1. When each port earns its keep
- Lumped port: early matching / DC-block exploration at low GHz when the launch is electrically short and you only need relative trends
- Wave port: connector launches, vertical transitions, and any claim that will be compared to a calibrated VNA plane — mode purity and reference-plane position own the phase
- Mixed mistake: wave-port size that admits higher-order modes, or a lumped port placed after a via stub that the model never named
2. Freeze list for a launch S-parameter claim
- Port type (wave vs lumped), reference-plane note, and whether a de-embed / thru cal matched the hardware plane
- Mesh convergence on the launch (not only on the DUT body)
- Connector PN / footprint and whether the model included the real dielectric and outer geometry
- Overlay: simulated launch vs measured thru or known-good launch coupon
3. Habit
Never freeze a filter or antenna "EM pass" whose launch was a lumped guess at a frequency where the connector is a significant fraction of a wavelength. Name the port class on the traveler. A 0.1 mm reference-plane error at 20 GHz is already a non-trivial phase argument in a filter dispute.
4. Practical gate
- Below your house frequency (often ~3–6 GHz depending on launch size), lumped exploration is fine for relative ranking
- Above that, require wave-port + measured thru overlay before accepting absolute S11/S21 for release
- If EM and VNA disagree only in phase, check reference plane first; if magnitude disagrees, check footprint dielectric and outer geometry before mesh density theatre
Discussion: on your last launch mismatch between EM and VNA, did the fix land in wave-port sizing, reference-plane shift, or a footprint detail — and do you forbid freezing launch S-params from a lumped port above your house frequency gate?
— rfforum 编辑部 / Editorial Team
