[Bram-Verhagen]

To control (dis)charge the batteries with the Deye inverters they expose register 1100 till 1121.
I have added these registers to the deye_p3 definition to expose them in HA.

Attached the explanation and some examples how to use these register to automate (dis)charge:
wiki_remote_control_deye_p3.md

[trilobyte303]

It would be very helpful if that could be incorporated!

[therealprof]

That would be super useful to have.

[tRoOlos]

Hi @Bram-Verhagen ,

I would be happy to help review and test this on my Deye SUN-15K-SG05LP3-EU-SM2, with three Seplos battery packs and Home Assistant/EMHASS.

Firmware:

• HMI: 1001-C05F
• MAIN: 2107-1184-1809
• ARC: D20A

Deye supplied me with three attached documents and answered several implementation questions directly. Comparing their documentation with this PR, I noticed these differences. All addresses below are decimal:

Register	Deye documentation	PR #978
1106	Combined/per-phase AC control selection	Constant-voltage setpoint
1107	Battery voltage setpoint (0.1 V)	Battery current setpoint (0.1 A)
1108	Battery current setpoint (0.1 A)	SOC setpoint (1%)
1110	Battery target SOC (0.1%)	Reactive power control mode
1111	Total AC power setpoint (0.1%)	Power factor setpoint
1112–1114	AC power setpoints for phases A/B/C	Reactive power and Volt-VAR curve points
1115	Maximum solar power (0.1%)	Volt-VAR curve point 2 voltage

Register 1109 does agree as the battery power setpoint, scaled in 0.1% of rated power.

There is also a contradiction within Deye’s documents concerning 1100: the general protocol describes 0 as disabled, whereas their EMS document describes 0 as enter and 1 as exit. Their email response pointed to the latter interpretation for my firmware.

Could you clarify which protocol revision, inverter model and firmware this mapping is based on? These may be variant-specific differences rather than errors.

Regarding transport, Deye explicitly replied “Currently only supported by RS485 adapter” when I asked about local LSW-5/Solarman access. Has this PR been tested through a logger or a direct adapter?

The three documents supplied by Deye are attached to this comment. I can also share the relevant email excerpts and help with supervised testing: read-only access first, followed by zero/low-power tests, register readback, verification of the actual battery response, and watchdog fallback to TOU. Testing would use one command owner, with safeguards against unintended grid charging or battery export.

RS-485-Remote mode(1).docx

MODBUS RTU三相储能通信规约V105.4-20240814(4)(1)_1.doc
ems control English version.docx