WIT-only Growatt Modbus Fix (VPP / Remote Control)

[linksu79]

Summary: WIT-only Growatt Modbus Fix (VPP / Remote Control)

This package contains a WIT-only implementation for Growatt hybrid inverters (e.g. WIT 6 HU), based on working, field-tested files.

The goal is to make remote/VPP battery control fully functional and stable without affecting other inverter models (SPH, SPF, etc.).

---

Problems in current integration

1. WIT is treated like SPH/SPF (incorrect)

The Growatt WIT series uses a different control logic and register behavior.
Reusing SPH/SPF assumptions causes:

• broken AC charging control

• missing VPP entities

• unreliable charge/discharge behavior

• conflicts with EMS / CEW / external controllers

---

2. Missing VPP / Remote control registers in WIT profile

The following official Growatt VPP registers were missing or unusable for WIT:

Register | Purpose -- | -- 30100 | Control authority (VPP enable) 30407 | Remote power control enable 30409 | Remote power control value (+charge / −discharge) 30410 | AC charging mode (Disabled / PV first / AC first) 30474 | Actual applied control value (readback)

Without these, external control cannot work correctly.

---

3. Signed values not supported (critical)

Register 30409 is signed:

• positive value → charge

• negative value → discharge

The original implementation expected only unsigned values (0–65535), which caused:

H format requires 0 <= number <= 65535

This made discharge control impossible.

---

4. WIT select entities were never created

select.py exited early for WIT devices, so no select entities were created for:

• control authority

• AC charging mode

• remote enable flags

This made correct configuration impossible from Home Assistant UI.

---

What was changed (WIT-only)

✅ profiles/wit.py

• Added full VPP register map for WIT

• Explicitly defines which registers are readable / writable

• No changes to other inverter models

---

✅ const.py

• Added WIT-specific writable definitions:

  • vpp_control_authority (30100)

  • vpp_remote_power_enable (30407)

  • vpp_remote_power_control_value (30409, signed)

  • vpp_ac_charging_mode (30410)

• Clear separation from SPH/SPF logic

---

✅ number.py

• Added support for signed 16-bit registers

• Proper conversion:

  • HA value → unsigned Modbus (two’s complement)

  • Modbus → signed HA value

• Enables stable number.growatt_remote_power_control_value

---

✅ select.py

• Removed WIT early-return

• Allows WIT to generate select entities from writable registers

• Makes AC charging and control authority configurable in UI

---

✅ growatt_modbus.py

• Ensures negative values are written correctly using:

  value & 0xFFFF
  

• Prevents Modbus write failures for discharge control

---

Result

With these changes:

• WIT inverters expose proper VPP / remote control entities

• Charge and discharge work reliably using one number entity

• AC charging mode is selectable and respected

• CEW / EMS / external controllers work without hacks

• No impact on non-WIT models

This implementation has been tested in real operation with:

• Home Assistant

• Growatt WIT 6 HU

• Cheapest Energy Windows (CEW)

• Grid charging + price-based discharge

---

Design choice

This is intentionally provided as a WIT-only solution, to:

• avoid regressions in other models

• make review and upstreaming easier

• reflect actual hardware differences

It can later be merged upstream either:

• as a WIT-specific branch, or

• behind a model capability flag

---

If you want, I can next:

• condense this into a shorter PR description, or

• adapt the wording exactly to 0xaha’s repository style.

growatt_modbus_wit_only_patch.zip

[0xAHA]

This is great. I won’t get time to take a look until the weekend so perhaps submit a PR so it’s easier for me to see the differences.

Thanks for helping out - I’m flying blind unless people ask for specific registers 😁 so hopefully this can complete the WIT functionality completely.

[0xAHA]

Hi @linksu79 - i've grabbed your files and merged - can you please test v0.4.0b1 and see if anything got missed or conflicted with some of my other interim changes?
https://github.com/0xAHA/Growatt_ModbusTCP/releases/tag/v0.4.0b1

[linksu79]

Hi 0xAHA,

Just to give some context and hopefully help you move this forward as far as possible.

This has been a heavy one to debug on our side, and we didn’t manage to get full VPP control working yet on WIT. That’s not a complaint — more a summary of where we got stuck.

Based on the official Growatt VPP Communication Protocol (v2.03), WIT requires an explicit VPP activation chain before any remote charge/discharge works:

1. Control authority must be enabled

   • Holding register 30100 = 1

2. Remote power control must be enabled

   • Holding register 30407 = 1

Only after those are enabled does:

• Remote charge/discharge power (HR 30409, -100…+100)
• and TOU / remote power logic
  actually take effect on WIT.

In v0.4.0b1 we don’t see any switch entities or services that expose 30100 or 30407, so VPP can’t be activated at all from HA yet. As a result, writing to 30409 alone has no effect on WIT.

This is very WIT-specific behaviour and not obvious unless you dig into the VPP protocol PDF, so totally understandable that it’s tricky.

From our testing:

• Hardware, battery, and Modbus comms are fine
• Smart Mode works (even too aggressively)
• The missing piece for normal operation is the VPP enable / arm path

Hopefully this helps you implement it cleanly in the integration.
Thanks a lot for continuing to work on this — it’s appreciated.

[linksu79]

Hi,

I’m adding a new register scan for reference and comparison.

This scan was taken with:

Inverter: Growatt WIT 6 HU

Firmware: YE1.0 / YEAA020299

COM: ZDDA-0002

The purpose of this scan is to provide additional register-level data for the integration, not to report a functional bug.

When comparing this scan to earlier ones taken on older firmware, there are some differences mainly in battery / EMS related status registers (312xx range), while the control registers themselves (e.g. 30100 / 30407 / 30409) appear unchanged.

For additional context, this is the guidance I received directly from Growatt regarding VPP / remote control:

When register 30100 is enabled, the BMS commands and the existing EMS configuration should be overridden and will not interfere with charge/discharge control.

Register 30409: a positive value = Charge, a negative value = Discharge.

Based on this, the expectation is that VPP / remote control should remain independent from local EMS logic once enabled, which is why I’m sharing these register scans in case they help identify any firmware-specific behaviour or assumptions at register level.

I’m sharing this purely as reference data, in case it helps with:

firmware or model-specific handling

conditional feature exposure

future debugging or validation

If useful, I can also provide additional scans taken in different EMS modes.

Thanks for your work on the integration.

growatt_register_scan_20260207_230532.csv

[0xAHA]

I've release v0.4.1 - the new sensors had conditions on them (stupid default error i/we keep making).
The sensors should now show.

i'll see if I can do a bit of reading on the rest of the info you provided - but if you come up with a working combination of settings, then that'd be great

[0xAHA]

Ok found a pretty decent oversight in earlier versions. Hopefully rectified in v0.4.3

[0xAHA]

I've looked through the manual from here: https://www.pretapower.com/wp-content/uploads/2025/09/User-Manual-of-WIT-4-15K-HU-Growatt-Hybrid-inverter.pdf

It looks like we are missing a couple of relevant items... as Claude so kindly suggests....

What We Currently Have for WIT Control:

VPP Remote Control (registers 201-203):
∙ work_mode - Standby/Charge/Discharge (VPP remote command only)
∙ active_power_rate - Power percentage
∙ export_limit_w - Export limit in watts
System Mode (register 902):
∙ onoff_change_mode - Hybrid/-H, Economy/-EP, UPS/-UP (hardware mode switch)
Diesel Generator (registers 939-940):
∙ dg_start_soc - SOC to start diesel generator
∙ dg_stop_soc - SOC to stop diesel generator
Demand Management (registers 944-946):
∙ demand_discharge_limit / demand_charge_limit
∙ demand_manage_enable

What the Manual Says Exists (But We’re Missing):

Operating Modes:
∙ Load First / Battery First / Grid First / Solar Only Backup / Idle / PTO / Grid First (ECO)
System Strategies:
∙ TOU (Time of Use) with time period configuration
∙ Self-consumption mode
∙ Grid Peak Shaving thresholds
∙ Micro-grid mode
GEN Port Modes:
∙ Zero Export to Meter
∙ Zero Export to Grid
∙ Smart Load
∙ AC Couple

[linksu79]

Hi,

I’m using your 0xaha integration with a Growatt WIT hybrid inverter, and despite multiple attempts it still does not behave correctly (power oscillation, charge/discharge looping, unstable control).

After carefully reviewing the WIT manual and Growatt support responses, the core issue appears to be a fundamental model mismatch, not a small bug.

Key point

Growatt WIT is not externally mode-controllable via Modbus.

Many of the operating modes and strategies described in the manual (Load First, Battery First, Grid First, TOU, Self-consumption, Peak Shaving, GEN port modes, etc.) are internal EMS states only.
They are configurable only via the inverter UI / installer menus, not via Modbus or VPP.

What is actually exposed for external control

For WIT, external systems are limited to a VPP-style override layer, essentially:

Charge / Discharge / Standby command

Power level (percentage or W)

Export limit

Directional power command (positive = charge, negative = discharge)

These commands:

temporarily override the internal EMS

do NOT change operating modes

do NOT switch strategies

are not persistent system states

Why the integration fails on WIT

If the integration:

assumes Battery First / Grid First / TOU can be set externally

tries to emulate modes by rapidly switching charge/discharge

writes multiple control registers in short intervals

treats WIT like SPF / SPH / MIN / Victron-style EMS

then oscillation and lockups are unavoidable.
This matches exactly what I am observing.

Required changes for WIT compatibility

To be compatible with WIT, the integration must:

Stop treating operating modes as external states

Use only one control authority at a time (single writer)

Limit control to:

direction (charge / discharge)

power limit

Apply slow control cycles (no fast toggling)

Let the internal EMS handle:

zero export

peak shaving

meter logic

GEN port behavior

grid compliance (EN50549, etc.)

Summary

This is not a configuration mistake on my side.
The current integration logic assumes capabilities that Growatt WIT simply does not expose.

I’m happy to help test or clarify behavior, but the control model must be aligned with how WIT actually works.