Add H3 register values

[TonyM1958]

Plenty of reference data around for H2 holding registers. Should modbusH3 file be added?

[TonyM1958]

Added modbusH3.yaml and spread sheet tab - needs to be tested by someone with H3 inverter. Modified generic template.yaml so it works with both single phase and three phase inverters.

[cosmico89]

HI! I own a H3.10kw, if you need to test something I can try! 😄

[TonyM1958]

Thanks @cosmico89. What's your current setup in terms of home assistant and monitoring? Don't want to disrupt stuff too much, so any context will help. There are definitely some registers and behaviours I'm not certain about.

[TonyM1958]

Bit more info... I've basically taken the known register set for H3 and added in new registers we found on H1 and KH but these have not been tested. So one thing would be to load up the sensors and check for illegal address errors in the HA log where modbus cannot read what we have. As part of this, review all the sensor values and confirm they read OK, including decodes and model id for inverter, bms and batteries.

Another area is we have 8 x fault registers on H1 but only 3 on H1. With no fault, they read 0 so are hard to detect unless the addresses flag as illegal. So, add some more and see if they read OK.

Final area, after whatever gets flagged above is how Load Power is calculated. This needs to be validated against Fox cloud data.

Ideally, we would also check load power results with a second solar inverter connected via meter 2 but that's not easy or common so I'm not expecting to be able to do that.

[TonyM1958]

A scan of H1 inverter gives the following active address blocks:

Holding Registers

• 30000
• 31000
• 32000
• 41000
• 44000

Input Registers

• 10000
• 11000
• 40000
• 41000
• 42000
• 43000
• 44000

The same scans on H3 give the same range of Holding Registers. There are NO active Input Registers.

Rescanning H1 shows the holding registers at 41000 mirror the input registers at 41000 BUT you cannot read the Time Period registers (41001 TO 41006) individually. Also, there is no holding register 41016 (Ground Enable).

[TonyM1958]

Scanning 31000 range, registers are as expected up to 31040. Then:

• 31041 Inverter State Code
• 31042 unknown = 1
• 31043 unknown = 1
• 31044 Fault 1 code
• 31045 Fault 2 code
• 31046 Fault 3 code
• 31047 Fault 4 code
• 31048 Fault 5 code
• 31049 Fault 6 code
• 31050 Fault 7 code
• 31051 Fault 8 code

That's the end of that block of registers

[TonyM1958]

Scanning 41000 range, the holding registers are:

• 41000 Inverter Work Mode
• 41001 - 40116 not available, Charge Time Periods (read as a block?)
• 41007 Max Charge Current
• 41008 Max Discharge Current
• 41009 Min SoC
• 41010 Max SoC
• 41011 Min SoC On Grid
• 41012 not available
• 41013 not available
• 41014 unknown 1
• 41015 not availabe
• 41016 unknown 4621
• 41017 unknown 0
• 41018 unknown 0
• 41019 unknown 4621
• 41020 unknown 0
• 41021 unknown 0
• 41022 unknown 3171
• 41023 unknown 1847
• 41024 unknown 16

[TonyM1958]

This is what we see scanning holding registers from 44000. Looks like:

• 44007 Power Out Limit
• 44008 Power In Limit

Others TBC - this block may be related to input registers 11026 - 11033 on H1?

[TonyM1958]

Expected Inverter Model, and Firmware versions at holding register block 30000. Model number has leading and trailing spaces.

[TonyM1958]

Power calculations: meters are combined from raw data and not split by phase, Template calculations updated to reflect this and now appear to agree with Fox Cloud data.