We have two nRF54L15 tags acting as BLE centrals and CS reflectors. Each tag connects to three anchors acting as BLE peripherals and CS initiators.
3.4.2: 3 nrf54L15 as anchors and 2 mobile tags.
The six links use 100 ms ACL > intervals, persistent non-IPT CS, RAS, one antenna path (A1_B1), and one 11 ms subevent. V sounds every 200 ms per tag. A1 and A2 sound every 800 ms per tag.
The same tag, E3:DD, intermittently logs SoftDevice Controller ASSERT: 131, 1138 followed by HARD FAULT. That tag then drops from all three anchors with BLE reason 0x08 (supervision > timeout). In one timestamped run, the assertion reached the collector 2.16–2.43 seconds before the three disconnect reports.
We captured the assertion with both 5 ms and 7.5 ms tag ACL event budgets and with different Ethernet host arrangements. Captured failures occurred after about 2:26, 4:31, and 17:30 of concurrent ranging. A separate run with all six streams completed 30:51, so the time to failure varies.
With both A2 ACLs kept connected and A2 sounding disabled, an eight-minute run completed without an assertion. A two-hour repeat is underway. We also built an all-six-stream variant > that disables only the failing tag’s diagnostic controller anchor reports. That variant awaits hardware testing.
Can you map file ID 131, line 1138 for nrfxlib revision 43e5c2a78592, identify the failed invariant, and advise whether there is a controller fix or a scheduling workaround that preserves all three active sounding links per tag? What additional trace would help isolate it?