Open question #6 in docs/architecture.md §11. Requires the full cable drum (same hardware as #3).
Problem. Modbus RTU at ≤ 19.2 kbps over ~450 m is far inside the RS-485 envelope on paper, but the actual error rate of the real cable + transceivers + connectors has not been measured.
Test plan
- Full-length trunk, master + at least two nodes, continuous polling at the production cadence.
- 24 h soak; count CRC errors and timeouts on both ends.
Pass criteria. An error rate low enough that the retry policy makes losses invisible at poll cadence — document the actual numbers.
Mitigation if it fails: drop the baud rate — D05 reserves 10× headroom.
Open question #6 in
docs/architecture.md§11. Requires the full cable drum (same hardware as #3).Problem. Modbus RTU at ≤ 19.2 kbps over ~450 m is far inside the RS-485 envelope on paper, but the actual error rate of the real cable + transceivers + connectors has not been measured.
Test plan
Pass criteria. An error rate low enough that the retry policy makes losses invisible at poll cadence — document the actual numbers.
Mitigation if it fails: drop the baud rate — D05 reserves 10× headroom.