MultiPlus-II 48/3000 ESS grid-parallel: discharges 30s, relay clicks

MultiPlus-II 48/3000 ESS grid-parallel: discharges 30s, relay clicks, then charge-only — broke after Venus OS update

Hardware

  • MultiPlus-II 48/3000/35-32 (PMP482306000), integrated GX, single phase in a 3-phase German house
  • Venus OS v3.73 Large, VE.Bus firmware v560
  • 16S LiFePO4, self-built, no CAN BMS — battery monitor: “Multi on VE.Bus”
  • Kostal Plenticore on AC-IN (not behind Multi)
  • Carlo Gavazzi VM-3P75CT 3-phase CT on AC-IN

Symptom
Island mode: discharge works fine, indefinitely.
Grid-parallel ESS: discharges ~30s, audible relay click inside Multi, then charge-only until next restart. “Battery low” alarm fires despite 53V at DC terminals (confirmed with multimeter). Charging works fine throughout.

History
Worked fine June 2024 – October 3, 2025. Broke around October 15, 2025 — coincides with Venus OS updating from ~v3.40 to ≥v3.60. Tried multiple VE.Bus firmware versions, no change. Problem reproduces when writing ESS AC setpoint directly from Remote Console — HA is not involved.

Already ruled out
LOM (A/B/Disabled), grid code (None/VDE-AR-N 4105), ESS mode (Optimized/External control), Min SoC, DVCC, “Configured for VE.Bus BMS” unchecked, ESS battery type set to “Other Lithium / no smart BMS”, DC voltage OK (53V, no sag), all dbus charge/discharge flags = 1.

Questions

  1. Single-phase Multi, 3-phase house, 3-phase Carlo Gavazzi — correct Multiphase regulation setting?
  2. Did Venus OS ≥3.60 introduce a stricter “no battery service” check that kills grid-parallel discharge after a timeout?
  3. Any known ESS Assistant battery wizard value (sustain/cut-off voltage) that triggers this exact behavior?
  4. Where to find the relevant logs — /data/log/CGwacs/ or /data/log/dbus-systemcalc/?

Can provide: VEConfigure .rvsc, Remote Console screenshots, HA DB timeline.