feature: Growatt WIT support via solax_modbus — likely already works in VPP mode, needs validation + wiring

[johanzander]

Overview

Support the Growatt WIT hybrid series (WIT 4-15kW / 8000-HU / 12000-HU / 15K-HU / 100000-HU) through the wills106/homeassistant-solax-modbus integration, reusing the existing solax_modbus_growatt_min (GEN4) platform.

User-visible outcome: a user with a Growatt WIT installs solax_modbus, runs the BESS setup wizard, sees "Growatt MIN/GEN4" light up green with sensors auto-mapped, and gets a working optimized schedule — no manual sensor mapping, no cloud integration.

Reported by @tissson in #627, where a WIT was misdetected as growatt_server_sph via the cloud integration.

Background — what #627 showed

The reporter's WIT has no working control path today:

• The official growatt_server cloud entry exposes only a plant-level device ("Malax Total"), so BESS's device resolution picks that, and every growatt_server.read_ac_charge_times call returns 500.
• Detection is elimination-based (ha_api_controller.py:3118-3131): no -tlx_ marker → assume SPH. WIT never enters the logic.
• His actual WIT entities come from a third-party HACS integration, 0xAHA/Growatt_ModbusTCP (domain growatt_modbus), which BESS knows nothing about. He mapped all sensors by hand.

What is already true today (verified, not assumed)

solax_modbus classifies WIT as GEN4 — plugin_growatt.py:9703-9707:

    # WIT
    "0PE": HYBRID | GEN4 | X3,  # WIT 8000-HU, 2 MPPT
    "0PC": HYBRID | GEN4 | X3,  # WIT 12000-HU, 2 MPPT
    "0PH": HYBRID | GEN4 | X3 | MPPT10,  # WIT 100000-HU, 10 MPPT
    "0HU": HYBRID | GEN4 | X3,  # WIT 15K-HU, 2 MPPT

Every entity in SOLAX_GROWATT_MIN_SUFFIX_MAP is reachable by a WIT. Each of the 31 mapped keys was resolved to its allowedtypes in plugin_growatt.py. All are gated on GEN4/HYBRID with no X1/X3 constraint, and the only serial blacklist in the plugin is SPF_SINGLE_MPPT_SERIAL_PREFIXES (SPF only). A HYBRID | GEN4 | X3 WIT satisfies all of them:

BESS map group	allowedtypes in plugin
battery_soc, battery_charge_power, battery_discharge_power	GEN4 | HYBRID
total_forward_power, total_reverse_power, total_load_power, total_yield	GEN4
lifetime energy counters	GEN4 / GEN4 | HYBRID
ems_charging_rate, ems_discharging_rate, ems_*_stop_soc	HYBRID | GEN4
charger_switch	HYBRID | AC | GEN4
vpp_status, vpp_remote_control, vpp_power, vpp_time, vpp_allow_ac_charging	GEN3 | GEN4
limit_grid_export, grid_export_limit	GEN2 | GEN3 | GEN4
time_1..9_{enabled,begin,end,mode,update}	HYBRID | GEN4

So detection already fires. _GROWATT_TOU_MARKER_SUFFIX = "time_1_enabled" (ha_api_controller.py:3801) is created for any HYBRID | GEN4 inverter, so a WIT on solax_modbus is detected as solax_modbus_growatt_min with no code change — sensors mapped, wizard pill green.

And BESS already has the control mode WIT most likely needs. SolaxModbusGrowattController supports both tou and vpp, and the Growatt VPP path (registers 30100 / 30407-30410) landed for #118. GEN4 currently defaults to tou (battery_system_manager.py:_resolve_control_mode).

The open question this issue exists to answer

Does WIT honour the GEN4 numbered-TOU registers, or is VPP the only working write path?

Evidence that TOU may be a dead end on WIT — from the other integration's WIT profile (0xAHA/Growatt_ModbusTCP, profiles/wit.py):

- Register 30476 (priority_mode) is READ-ONLY - shows TOU default, cannot be changed via Modbus

That integration drives WIT entirely through work_mode (register 202: 0 Standby / 1 Charge / 2 Discharge) plus a remote charge/discharge power rate, gated by control_authority (register 30100) — a continuous VPP-style command model, and it re-asserts work_mode before every power-rate write.

Register 30100 is the same vpp_status register BESS already writes in VPP mode. So the likely answer is: WIT works today on solax_modbus_growatt_min with control_mode: vpp, and TOU mode silently no-ops.

This cannot be settled from source. It needs one user with a WIT and solax_modbus installed.

Proposed work

Phase 1 — validate (blocking, needs hardware). Ask a WIT owner to install solax_modbus alongside their existing setup and report:

1. Does the wizard show "Growatt MIN/GEN4" detected, with sensors auto-populated?
2. In control_mode: tou — does a TOU slot write actually change inverter behaviour, or is it accepted and ignored?
3. In control_mode: vpp — does a charge and a discharge command take effect?

Phase 2 — wire the outcome. Depending on Phase 1:

• VPP works, TOU doesn't (expected): make WIT resolve to vpp rather than the GEN4 tou default. WIT is not distinguishable from a MIN by entity shape alone — both are HYBRID | GEN4 — so this needs a positive signal (serial prefix 0PE/0PC/0PH/0HU via the device/serial, or an explicit user choice). Options to weigh: a WIT-specific platform id, or keeping solax_modbus_growatt_min and steering only the control-mode default.
• Both work: documentation only.
• Neither works: close this and reopen on the growatt_modbus path (below).

Phase 3 — document. WIT in the supported-hardware table at the correct maturity level (experimental until a real install is confirmed), plus a wizard note that WIT needs solax_modbus rather than the cloud integration.

Explicitly out of scope

Adding a growatt_modbus platform for 0xAHA/Growatt_ModbusTCP. It's a real option — it has a dedicated wit.py profile, per-model register maps, and a VPP control surface close to what BESS already drives (number.growatt_battery_remote_charge_and_discharge_power, select.growatt_battery_mode_vpp, select.growatt_grid_control_authority, number.growatt_tou_active_periods, select.growatt_work_mode) — and #627's bundle already gives the full entity vocabulary. But it is a third integration to support, and solax_modbus is the path already built. File separately if Phase 1 rules solax_modbus out.

Also out of scope: fixing the growatt_server elimination-based detection that misfires on WIT (has_tlx → SPH). That's a bug, tracked from #627.

Acceptance criteria

• A WIT on solax_modbus is auto-detected by the wizard with sensors mapped, no manual entry.
• Charge and discharge commands are confirmed to take effect on real WIT hardware (log + inverter state).
• The platform's control mode is resolved correctly for WIT without the user having to know which register family their inverter honours.
• Supported-hardware docs list WIT with an accurate maturity level.

Definition of Ready

Fails criterion 4 (approach agreed) and criterion 2 (reproduction) — Awaiting: user/hardware. Phase 1's answer decides Phase 2's design; the fork is stated above so a later attempt doesn't rediscover it. Nobody on the team has WIT hardware.

Related

• SPH read? (WIT inverter) #627 — WIT misdetected as growatt_server_sph (source of this request)
• Feature: Native SolaX Modbus support for Growatt MIN (TOU control via solax_modbus) #92 — the precedent: native solax_modbus support for Growatt MIN
• Correction to Solax-Growatt integration: Growatt MIN-inverters now have functional VPP registers, and do not need TOU programming #118 — Growatt VPP registers are functional and don't need TOU programming (the mechanism this depends on)
• solax_modbus: https://github.com/wills106/homeassistant-solax-modbus
• Growatt_ModbusTCP (out-of-scope alternative): https://github.com/0xAHA/Growatt_ModbusTCP