Rule 1: the measurement radios transmit no 802.11 frames¶
Cross-capture on 2026-10-13¶
Both A9000s in the Pi (capA stick A, capB stick B), kernel 6.18.50. One stick listens for frames with the other stick's address as transmitter (wlan addr2 <MAC>), 60 s per run. Before each run a 10 s control on the listener: beacons of a BSS of the home SSID on the same channel (wlan addr2 <bssid> and type mgt subtype beacon).
| Run | Radio checked | Its state | Listener | Control | Frames from the checked radio |
|---|---|---|---|---|---|
| 1 | capB |
fixed, 6135/160 | capA on 6135/160 |
about 100 beacons | 0 |
| 2 | capB |
sweep 2437, 5240, 6135 at 20 MHz, 150 ms dwell, 450 hops, failed=0 |
capA on 5240/20 |
97 beacons | 0 |
| 3 | capA |
fixed, 6135/160 | capB on 6135/160 |
96 beacons | 0 |
| 4 | capA |
sweep as in run 2, 450 hops, failed=0 |
capB on 5240/20 |
97 beacons | 0 |
linkside-radio status in the same setup: only-monitor ok=1, nm-unmanaged ok=1 (the monitor interfaces are absent from nmcli device), rfkill ok=1.
The Mac stayed on Wi-Fi (5 GHz, channel 48) during the runs; its frames carry its own address and fall outside the filter.
The address filter does not see frames without a transmitter address (ACK, CTS), frames sent with another address, or the RF burst of the MT792x when it retunes.