diff --git a/docs/FINDINGS.md b/docs/FINDINGS.md index aa4a049..28ccb9a 100644 --- a/docs/FINDINGS.md +++ b/docs/FINDINGS.md @@ -146,3 +146,20 @@ - Narrower RX bandwidth (156 kHz) vs current 234.3 kHz on weak meters. - Long-run stability at >=100 and >=500 verified repeat events. - Early-stop RX (poll GetRxBufferStatus) to cut the ~20 ms deaf tail. + +## CONFIRMED (power supply, field debugging 2026-09-01) + +- A weak USB supply silently kills radiated TX while everything else looks + healthy: SX1262 reaches TxDone, firmware counts success, LED double-blinks + -- but the SDR 50 cm away sees nothing. A/B proof: laptop USB 7/7 TX + events RF-aligned (median +37 dB over floor); wall adapter minutes later, + same position: 0 of ~expected 5-7 across 80 s (and 0 across four earlier + captures). Suspected mechanism: rail sag during the KCT8103L PA burst. + "TX done" is chip-level packet completion, not radiated power. +- Practical rule: power the repeater from a supply that holds 5 V through + ~400 mA bursts (laptop port, 2 A charger, power-bank USB-A, or LiPo). + After any supply change, verify with one 60 s SDR capture -- the LED + cannot distinguish radiating from non-radiating TX. +- The recurring 10-24 ms bursts at +30..+43 dB ("monster singletons") are a + neighbouring meter railing the dongle, NOT the repeater: their rate is + unchanged across our firmware versions, reboots, and power sources.