Sungrow SH10RT-20: add feed-in curtailment support

[FH1957]

Is your feature request related to a problem? Please describe.
The sungrow-hybrid template currently does not provide curtailment support for the SH10RT-20 V11. This prevents using the inverter for §9 EEG / LPP curtailment in evcc.

Describe the solution you'd like
Add curtailable support to the sungrow-hybrid template.
According to Sungrow Communication Protocol of Residential Hybrid Inverter V1.1.9 (2025-06-20), feed-in limitation can be controlled via:

Register 13087 – Feed-in Limitation
Register 13088 – Feed-in Limitation Ratio (0–100.0 %)

With a smart meter connected, the control point is the grid connection point. In a master/slave system, the parameter only needs to be sent to the master inverter.

Describe alternatives you've considered
Hardware curtailment via the inverter's ripple-control inputs.

Additional context
Device: Sungrow SH10RT-20 V11

The relevant Sungrow protocol documentation is attached.

TI_20250620_Communication Protocol of Residential Hybrid Inverter_V1.1.9_EN.pdf

[github-actions]

Thanks for the detailed report and the protocol reference.

Current state: templates/definition/meter/sungrow-hybrid.yaml (SH series) declares only capabilities: ["battery-control"] — no curtailment support today.

Existing precedent: The other two Sungrow meter templates already implement curtail/curtailed using the same "enable/disable switch + ratio register" pattern you describe:

• templates/definition/meter/sungrow-inverter.yaml (SG string inverters) — added recently in Sungrow string inverters: add EEG §9 curtailment #32976, using registers 5006/5007 (0-based, i.e. one less than the doc numbering).
• templates/definition/meter/sungrow-ihm.yaml (iHomeManager) — using registers 8027/8030.

That #32976 PR explicitly notes the SH hybrid register map differs (e.g. its register 5008 is used for internal temperature on the SH series), which matches your report that the SH series needs its own registers — 13087/13088 per the documentation, i.e. 0-based Modbus addresses 13086/13087 following the same convention.

This is a reasonable, well-precedented feature addition, but adding it correctly means deciding whether it belongs to the grid or pv usage block in the template (the existing precedents attach curtail to pv, while feed-in limitation logically applies at the grid connection point), handling the master/slave behavior you mention, and validating the actual register writes against real SH10RT-20 hardware — as was done with a live device for the SG-series PR. That validation isn't something I can do without access to the inverter, so I'd suggest this be picked up by someone with an SH10RT-20 who can test the writes directly (similar to how #32976 was verified).

🤖 Generated with Claude Code

[FH1957]

I have an SH10RT-20 V11 with a CHINT smart meter and a master/slave setup available for testing. I can test and validate the implementation on the actual hardware if someone provides the required test instructions or a test build/PR.

[harmendp]

Could you test #33210 ?
It looks OK on my SH4.0RS.

[FH1957]

I can test this on a master/slave setup with 2 × SH10RT-20 V11, using the master inverter's direct LAN interface (Modbus TCP) and a CHINT smart meter. Let me know when the PR is ready for hardware testing and how you would like me to test it.

[FH1957]

Could you trigger /build docker for #33210? I can then test the PR image on my 2 × SH10RT-20 V11 master/slave setup.

[harmendp]

Could you trigger /build docker for #33210? I can then test the PR image on my 2 × SH10RT-20 V11 master/slave setup.

I have the docker image available, just don't really know how to get it to you.

[FH1957]

Thanks. The easiest way would probably be to trigger /build for PR #33210, so I can pull the official PR image directly.
I can then test it on my 2 × SH10RT-20 V11 master/slave setup and report back.