C
Editorial
Hello from CosmosForum Editors —
Industrial robots, AGVs/AMRs, and machine-mounted telemetry often ship a small patch (or planar array) that looks fine as an isolated antenna: S11 dips at the design frequency on a clean bench fixture. In the metal chassis, next to motors and cable harnesses, or when the beam is steered / the robot attitude changes the scan geometry, the PA or LNA sees a different active impedance — and the pattern you cared about was never the free-space element pattern with neighbours terminated. Radio and science-instrument teams keep rediscovering the same failure mode robot wireless bring-up already knows by another name: “it matched alone” is not a release gate.
We opened a technical companion on CosmosForum with rfforum editorial on exactly this checklist. Cross-link: https://www.cosmosforum.org/t/44
### 1) Isolated S11 ≠ what the radio sees under scan
Quote **active impedance** (or scanned S-parameters) at broadside *and* at the max scan / max attitude your AGV or arm actually uses. A single unmatched “good S11” screenshot from an isolated patch on foam is brochure, not acceptance.
### 2) Embedded element pattern with neighbours terminated
Array-factor painted on a free-space pattern will miss scan loss and nulls. Edge vs centre elements deserve separate honesty; E-plane scan is often the first pain point for patches. Robot parallel: a side-mounted patch next to a steel cover is not the same radiator as the datasheet photo.
### 3) Retune discipline
Freeze an active-impedance check **before** PA matching, then re-check the scan edge after any broadside-only feed tweak. Optimizing one attitude only is how fleets inherit a null at the docking pose.
### 4) What robot / AGV teams can steal as a release gate
- Write **active match at broadside + max scan/attitude** next to the isolated S11 claim.
- Require an EEP-style sanity (or at least a chassis-on vs chassis-off pattern delta) before a patch enters a module ICD or a fleet RF acceptance sheet.
- Version the harness: chassis cutout, cable dress, neighbour modules, and lid — the same habit SI/PI and instrument teams use when a “RF failure” was really a fixture change.
### Acceptance question we are collecting war stories on
On your last small patch / planar antenna for AGV telemetry, Wi-Fi on a robot wrist, or machine-mounted link, which failed first under scan or in chassis — active match, EEP null / scan loss, or feed-network imbalance — and did **lattice spacing / placement** or **matching network** win the fix once you stopped trusting isolated S11?
Companion technical thread on CosmosForum (rfforum editorial note + admin engineering reply):
https://www.cosmosforum.org/t/44
Methods and failure modes only — no product pitches, no sock accounts. If irunion has a production note on robot / AGV antenna acceptance, chassis-on pattern deltas, or active-impedance gates that science / RF teams should steal, a technical reply on either side is welcome.
Clear patterns, honest chassis —
CosmosForum Editors
https://www.cosmosforum.org/
Industrial robots, AGVs/AMRs, and machine-mounted telemetry often ship a small patch (or planar array) that looks fine as an isolated antenna: S11 dips at the design frequency on a clean bench fixture. In the metal chassis, next to motors and cable harnesses, or when the beam is steered / the robot attitude changes the scan geometry, the PA or LNA sees a different active impedance — and the pattern you cared about was never the free-space element pattern with neighbours terminated. Radio and science-instrument teams keep rediscovering the same failure mode robot wireless bring-up already knows by another name: “it matched alone” is not a release gate.
We opened a technical companion on CosmosForum with rfforum editorial on exactly this checklist. Cross-link: https://www.cosmosforum.org/t/44
### 1) Isolated S11 ≠ what the radio sees under scan
Quote **active impedance** (or scanned S-parameters) at broadside *and* at the max scan / max attitude your AGV or arm actually uses. A single unmatched “good S11” screenshot from an isolated patch on foam is brochure, not acceptance.
### 2) Embedded element pattern with neighbours terminated
Array-factor painted on a free-space pattern will miss scan loss and nulls. Edge vs centre elements deserve separate honesty; E-plane scan is often the first pain point for patches. Robot parallel: a side-mounted patch next to a steel cover is not the same radiator as the datasheet photo.
### 3) Retune discipline
Freeze an active-impedance check **before** PA matching, then re-check the scan edge after any broadside-only feed tweak. Optimizing one attitude only is how fleets inherit a null at the docking pose.
### 4) What robot / AGV teams can steal as a release gate
- Write **active match at broadside + max scan/attitude** next to the isolated S11 claim.
- Require an EEP-style sanity (or at least a chassis-on vs chassis-off pattern delta) before a patch enters a module ICD or a fleet RF acceptance sheet.
- Version the harness: chassis cutout, cable dress, neighbour modules, and lid — the same habit SI/PI and instrument teams use when a “RF failure” was really a fixture change.
### Acceptance question we are collecting war stories on
On your last small patch / planar antenna for AGV telemetry, Wi-Fi on a robot wrist, or machine-mounted link, which failed first under scan or in chassis — active match, EEP null / scan loss, or feed-network imbalance — and did **lattice spacing / placement** or **matching network** win the fix once you stopped trusting isolated S11?
Companion technical thread on CosmosForum (rfforum editorial note + admin engineering reply):
https://www.cosmosforum.org/t/44
Methods and failure modes only — no product pitches, no sock accounts. If irunion has a production note on robot / AGV antenna acceptance, chassis-on pattern deltas, or active-impedance gates that science / RF teams should steal, a technical reply on either side is welcome.
Clear patterns, honest chassis —
CosmosForum Editors
https://www.cosmosforum.org/