Heater ineffective fires on a working heater during every normal warm-up: the brain alerts when the tent stays below target − 0.3 °C for 10 min while heating, b

Found 2026-09-13 by the dev-bed autonomous LLM test loop (recurring ‘unexplained’ heater_ineffective on near-healthy runs) and a healthy-baseline probe (heater_ineffective ON at 10 virtual min with tent 24.45 °C, target 25 °C, heater running normally). Confirmed from firmware: the hub turns the heater on after the arbitrated tent is below target − 2.0 °C for heat_wait and off at target − 0.5 °C, so while heater demand is on the tent is below target − 0.5 by construction. The brain’s ineffective test (below target − 0.3 for 10 min, with ≥ 8 min runtime) is therefore satisfied by any warm-up that takes longer than 10 minutes — a false ‘Heater ineffective’ on a working heater, which PR #224 introduced when it corrected the old inverted (overshoot) condition. Secondary: the hub heats toward the ARBITRATED tent’s target (4×8 or 2×4 by priority) but the brain always compares the 4×8 temp_c to number.dsc_hub_target_temp. The simulator’s ladder thresholds (on < target − 1.5, off > target − 0.3) also differ from firmware, so the dev bed did not exercise the real band. Expected: judge ineffectiveness by lack of progress (tent not warming across the hold while the heater runs), or by the tent still being at/below the firmware ON threshold (target − 2.0) after the hold; read the arbitrated tent; align the simulator ladder with firmware.