Commit Graph
2 Commits
Author SHA1 Message Date
Jean-Pierre MartineauandClaude Opus 5.5 dee4674942 The SPI radio listens for the preamble the mesh actually sends
MeshCore transmits a 32-symbol preamble up to SF8 (16 above), but the
node and the bridge configured the SX1262 for the library's fixed 12.
The chip waits for the sync word only about as long as the preamble it
expects, so it locked onto each packet, gave up, and locked on again
further along the same preamble - decoding a packet only when it
happened to catch the tail. On the uConsole that meant 0 of 5 detected
packets decoded after a send; with 32 symbols, 4 of 4.

The node now derives the preamble from the spreading factor as MeshCore
does, and re-sends it to the listening chip when Device config changes
the SF. preamble_length leaves the SPI wiring table (it was protocol,
never wiring) and is retired; the bridge defaults to the same rule.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-24 03:12:52 -04:00
Jean-Pierre MartineauandClaude Opus 5.5 9538bfedce MeshTerm drives an SPI radio itself, without the bridge service
A LoRa chip on the host's own SPI bus (the uConsole AIO) is now a
transport like serial, BLE, and TCP: --spi, an SPI profile, or the
picker's row for /dev/spidev1.0. MeshTerm starts a software node as a
child process under whichever Python has openhop-core, talks to it over
a loopback port, and ends it on disconnect; the node also exits with its
parent, so the pins are free whenever MeshTerm is not running.

The node keeps identity, prefs, channels, and contacts under
~/.meshterm/radio/<spidev>/, and on first use copies the bridge's
identity, contacts, and radio settings so it stays the same node.
Board wiring lives in a profile's spi table; radio settings are the
node's own, edited on Device config. Closes #21.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-24 01:48:32 -04:00