Skip to content

v1.12.9 — Accuracy & reliability · 40 models CI-verified

Choose a tag to compare

@elRadix elRadix released this 26 Oct 17:11
· 209 commits to main since this release
a7129bf

⭐ Featured — Daily Summary Tabbed dashboard

Overview Quality Cycles Events

Five interactive tabs — Overview · Quality · Cycles · Duration · Events — with adaptive charts, detail popups on every element, and Dutch/English entity auto-detection.

📥 Download YAML · 📖 Full documentation · 🎨 All cards


🔧 Featured — 40 Altherma models, CI-verified

We now validate 40 Daikin Altherma model definitions against the upstream ESPAltherma project on every pull request. If a model definition ever silently drops a required attribute, our CI fails before it can break your install.

Newly tracked in this release: the HT ERRQ-ERSQ outdoor unit (11-16kW) — a high-temperature outdoor unit that publishes outdoor-side and refrigerant sensors only. Documented as a "partial model" so its limits are visible up front.

Supported model families (40 definitions)

Family Coverage
EBLA / EDLA D Monobloc 4-8kW · 9-16kW
EPRA D / E 14-16kW (incl. ECH2O) · 8-12kW · D7 variants
ERGA D / E 04-08kW (incl. ECH2O) · D/E series
ERLA D 11-16kW · 3kW (ERLA03)
Altherma 4 EPSK / EPSKS / EPBX / EPSX / EPVX / EPVZ 04-14kW
LT Monobloc EBLQ / EDLQ / CA series 5-16kW · Bizone variants
HT series EKHBRD-ERRQ (11-16kW) + ERRQ-ERSQ outdoor unit (NEW)
GEO EGSAH-X / EWSAH-X GEO3 6-10kW · EGSQH-A GEO2 10kW
Hybrid Altherma Hybrid
Multi DHW EKHWET-BAV3 · CKHWS-BJV3 · LT_Multi_DHWHP
Mini chiller EWAA / EWYA D · EWAQ / EWYQ B
Protocol S PROTOCOL_S · PROTOCOL_S_ROTEX

Full matrix: MODEL_COMPATIBILITY.md


v1.12.9 — Accuracy & reliability

Your cycle data is now trustworthy end to end. Three quiet data-quality issues are corrected — both going forward and throughout the full history. If DHW thermal kW ever read 0.0, a single heating run showed up as two cycles, or older cycles had inflated sample counts, this release fixes all three.

🔥 DHW thermal output is real now

DHW cycles used to report 0 kW thermal output when the compressor was heating the tank. Now we measure DHW thermal directly from flow × ΔT, using the physically correct cold-side sensor (R4T — confirmed against live production captures; the previously assumed R5T overestimates by ~45%).

You'll see: realistic DHW kW in your dashboard, accurate SPF, and fewer false energy_daily_anomaly alerts.

🔗 Cleaner cycle boundaries

30–60s transient events (defrost, service routines, sensor hiccups) used to close a heating run early and start a new one. The detector now bridges transients into one continuous cycle via the new _apply_merge() primitive.

You'll see: fewer ghost cycles, duty cycle that matches real compressor behavior, and more reliable pendulum_hourly / high_cycle_rate signals.

🔧 Historical sample counts repaired

Cycles recorded before late October stored cop_sample_count values up to 10× real. A new migration (async_migrate_to_v1_12_5_sample_count_fix) repairs the whole history with a safe relative threshold (>1.5× real).

You'll see: cop_confidence ratings that match actual sample density — no more "high confidence" on 45-sample cycles.

🛡️ Safer reconfigure

_migrate_entity_ids is now idempotent — re-running it preserves unique IDs, translation keys, and history. Dashboards, automations, and history keep working after reconfigure.


🔬 Under the hood

  • New engine/physics.py — standalone Carnot COP primitive (wired in v1.13.0)
  • Model-compat contract: parser + checker + PR-gate + weekly drift workflow
  • Deterministic CI gate: fetch-by-pinned-SHA (upstream branch movement no longer breaks the PR check)
  • 130 tests in test_model_compat.py · 8 tests in test_v1129_132_thermal_mode.py
  • Full suite: 3588 tests · 100.00% coverage · mypy strict 38 files

Full technical details: CHANGELOG.md

📋 Post-release watch

We're tracking the first 48 hours of v1.12.9 via #163. If you see unexpected cycle counts or DHW thermal values, please open an issue with logs — it helps us validate quickly.