New installation - pv_power value unknown

[WyndStryke]

Hi,

I had a KT7 + EC4300-H4 installed yesterday, with master v1.34, manager v1.28, slave v1.01, running the forked version of this integration. There are two arrays, one connected to MPPT2, and the other connected to MPPT3. I connected it up via a waveshare RS485 to POE Eth(B) adaptor.

Interestingly the connector was a 16 pin thing like the H1, pinouts appeared to be the same, and it appears to be working (feedback on the wiring instructions might be to unhook the CT clamp from the line at the start, taking a photo first to ensure that it is reattached in the correct orientation and to the correct tail - reason being that it started buzzing as soon as the connector was detached).

So now I'm trying to get the software side of things running.

I'm very much a newbie so I'm sure I've made plenty of dumb mistakes.

Currently predbat is unhappy with the sensor.pv_power value, which is unknown. However, the pv2_power, pv3_power, and also sensor.solar_energy_total, do all have data ... ? (presumably pv_power is just the sum of pv1_power, pv2_power, pv3_power and pv4_power)?

16791 | 2025-01-09 09:00:03.457288: Warn: record_status Warn: Return bad float value unknown from pv_power 16790 | 2025-01-09 09:00:03.421212: Warn: Return bad float value unknown from pv_power using default 0.0

sensor.connection_status = connected
sensor.inverter_state = on grid
sensor.pv2_voltage 224.9V
sensor.pv2_current = 0.7A
sensor.pv2_power = 0.164kW
sensor.pv2_power_total = unknown
sensor.pv3_voltage = 250.0V
sensor.pv3_current = 0.3A
sensor.pv3_power = 0.096kW
sensor.pv3_power_total = unknown

pv4 all zero
pv1 has some slight voltage for some reason even though nothing is connected (1.5v)

Many of the sensors are picking up data as expected.

sensor.pv_power = unknown as already mentioned
sensor.pv_load_power = unknown
sensor.grid_ct = 0 (the grid data shows up in the Fox app)
sensor.grid_consumption = 0
sensor.solar_energy_total = 0.5kWh
sensor.solar_energy_today = 0.2kWh
sensor.total_yield_total = 6.2kWh
sensor.grid_consumption_total = 16.7kWh
sensor.input_energy_total = 14.2kWh
sensor.load_power_total = 8.6kWh

`Device info

KH7 - AUX
by FoxESS

Automations

No automations have been added using this device yet. You can add one by clicking the + button above.

Scenes
No scenes have been added using this device yet. You can add one by clicking the + button above.

Scripts
No scripts have been added using this device yet. You can add one by clicking the + button above.

Controls
Force Charge Power 7.0 kW
Force Discharge Power 7.0 kW
Max Charge Current 50.0 A
Max Discharge Current 50.0 A
Max SoC 100 %
Min SoC 10 %
Min SoC (On Grid) 10 %
Work Mode Self Use

Sensors
Ambient Temp 24.5 °C
Battery Charge 0 kW
Battery Charge Today 12.5 kWh
Battery Charge Total 14.2 kWh
Battery Discharge 0.228 kW
Battery Discharge Today 1.2 kWh
Battery Discharge Total 6.3 kWh
Battery SoC 88%
Battery Temp 18.9 °C
Connection Status Connected
Feed-in 0 kW
Feed-in Today 0.0 kWh
Feed-in Total 0.2 kWh
Grid Consumption 0 kW
Grid Consumption Today 14.3 kWh
Grid Consumption Total 16.7 kWh
Grid CT -0.0 kW
Input Energy Today 12.8 kWh
Input Energy Total 14.2 kWh
Inverter Battery Current 0.9 A
Inverter Battery Power 0.228 kW
Inverter Battery Voltage 238.6 V
Inverter Current 2.3 A
Inverter Power 0.0 kW
Inverter State On Grid
Inverter Temp 17.6 °C
Load Energy Today 2.9 kWh
Load Energy Total 8.7 kWh
Load Power 0.0 kW
Max Charge Current 50.0 A
Max Discharge Current 50.0 A
Max SoC 100%
Min SoC 10%
Min SoC (On Grid) 10%
PV Power Unknown
PV1 Current 0.1 A
PV1 Power 0.0 kW
PV1 Power Total Unknown
PV1 Voltage 0.4 V
PV2 Current 0.7 A
PV2 Power 0.156 kW
PV2 Power Total Unknown
PV2 Voltage 222.1 V
PV3 Current 0.5 A
PV3 Power 0.133 kW
PV3 Power Total Unknown
PV3 Voltage 249.7 V
PV4 Current 0.0 A
PV4 Power 0.0 kW
PV4 Power Total Unknown
PV4 Voltage 0.0 V
Solar Generation Today 0.3 kWh
Solar Generation Total 0.6 kWh
Version: Manager 1.28
Version: Master 1.34
Version: Slave 1.01
Yield Today 1.3 kWh
Yield Total 6.3 kWh
+8 entities not shown

Logbook
8 January 2025
Work Mode changed to Self Use 23:45:39 - 10 hours ago
Version: Master changed to 1.34 23:45:39 - 10 hours ago
Inverter State changed to On Grid 23:45:39 - 10 hours ago
Version: Slave changed to 1.01 23:45:39 - 10 hours ago
Version: Manager changed to 1.28 23:45:39 - 10 hours ago`

So I'm not really sure how to proceed. To start with, was ocl80's forked v1.13.0-beta (which has the register changes and the HA 2025.1 fixes) the correct one to install for a KH on 1.34? Reference is #678 (comment)

-Cheers

--- Edit:

Just tried ocl80's v1.13.0-beta2 which was published last night, but that one doesn't seem to read the KH registers. Reverted to v1.13.0-beta and it can read the registers fine with that version.

Connection Status changed to Error reading registers. Type: RegisterType.HOLDING; start: 36001; count: 3; slave: 247 from tcp://192.168.68.200:502: Exception Response(131, 3, IllegalAddress)
11:49:52 - 5 minutes ago

--- Edit 2:

Swapping to the new beta2, and then back to the original beta, seems to have got pv_power etc working. Perhaps initialisation or something was the issue?

[FozzieUK]

yes as you say pv-power is an internally maintained sums - They do require some time to collect but i’ve seen them hang a few times on recent installs and I suspect you may have ‘fixed’ this with the several restart of HA as it would have forced state changes through them.

I’m not sure on the latest beta version status, the original fork was done as the KH register mapping changed and it needed someone with a KH system to modify and test - i’ll leave @ocl80 to comment on that.

[ocl80]

I had understood that the key firmware element regarding the address locations is the "Manager" firmware, and with your Manager firmware at 1.28 @WyndStryke you might be able to run the "normal" version of the integration here (i.e. not the one I forked).

I think that the reason that you found it worked with my v1.13.0-beta, but not v1.13.0-beta2 is that in v1.13.0-beta (same as in my fork of v1.12.2) I hadn't changed any of the PV register addresses. For my system it worked OK on Manager 1.35 without changing these because I have only 2 strings, and these seem to be "backwards compatible": queries to both the "old" and "new" address structures work for PV1 and PV2 (but I did have to change various other register addresses, notably grid_ct, rpower and loadpower to make it work with my KH). But this seems not to be the case with PV3 and PV4, so in v1.13.0-beta2, I moved all PV1-PV4 measures to use the "new" addresses.

If (my guess) your KH with Manager 1.28 is using the "old" ones, that's why v1.13.0-beta2 didn't work for you... have you tried the regular v1.13.0?

My forked v1.13.0-beta2 seems to be working OK on my KH8 with 2 strings on Manager 1.35, Master 1.39, Slave 1.01.
If its confirmed as working OK for more strings, I'll tidy up the code changes so they're easier to read and remove the "pre-release" flag.

[WyndStryke]

@ocl80 Many thanks. Ah I see, I was looking at the 'Master' (version 1.34) rather than the 'Manager' version. I've only had the system for a couple of days now so I hadn't noticed the distinction between the two. That also explains why I wasn't getting a reading for grid_ct/rpower/loadpower.

In that case, I'll try the regular version.

Many thanks for your help :-)

Regarding the issue of detecting the firmware version, has this been figured out yet by anyone? If I'm able to help in any way by reading registers or similar on my version of the firmware (Manager 1.28, Master 1.34, Slave 1.01) ...

[WyndStryke]

Just as an aside for anyone feeding sensor.pv_power into predbat. Predbat uses watts for this, and converts internally since sensor.pv_power is in kW. But it displays as zero in the charts & on the config page. It's just a display issue since it is displaying the pre-conversion value, behind the scenes it is working fine. If you want to see it on the chart, either modify the chart to multiply by 1000 if the sensor units is kW, or use a template sensor to multiply it earlier.

[WyndStryke]

Suddenly everything stopped working, so just in case, I checked the firmware versions, and I now have:

Master : 1.41
Slave : 1.01
Manager : 1.37

I got the error:

Connection Status changed to Error reading registers. Type: RegisterType.HOLDING; start: 31000; count: 55; slave: 247 from tcp://192.168.68.200:502: Modbus Error: [Input/Output] No Response received from the remote slave/Unable to decode response
19:13:14 - 2 hours ago
watchman_last_updated changed to 13 January 2025 at 19:13
19:13:14 - 2 hours ago
Max SoC became unavailable
19:13:14 - 2 hours ago
Version: Master became unavailable
19:13:14 - 2 hours ago
FoxESS - Modbus Connection error: Error reading registers. Type: RegisterType.HOLDING; start: 31000; count: 55; slave: 247 from tcp://192.168.68.200:502: Modbus Error: [Input/Output] No Response received from the remote slave/Unable to decode response

So I'm going to swap to @ocl80's beta2 fork.

Don't know why my firmware changed, guess the update was pushed out by Fox since this is a new installation.

--- Edit:

Yeah seems to be fixed with beta2, had to delete / reinstall the integration, for some reason the inverter rebooted when I tried to reinstall it initially? Perhaps coincidence?

--- Edit2:

Looks like the inverter reboot was as a result of battery firmware being updated, so unrelated to the RS485 adaptor talking to it.

60KM432xxxxxxx | ECS4300 | 1.14
60KS432xxxxxxxx | ECS4300 | 1.14
60KS432xxxxxxxx | ECS4300 | 1.14
60KS432xxxxxxxx | ECS4300 | 1.14

[williamjeccles]

@WyndStryke Fox did a never-before-seen push of firmware updates across the whole fleet of inverters. I might be to blame for this because I reported that I noticed the time was going out of sync and I suspect they're done a force update for everyone to try and resolve this.

[WyndStryke]

This might mean that @ocl80's beta2 fork suddenly becomes critical for everybody (with a KH).

I seem to recall that the blocker was the need to identify the firmware version reliably, is that still an issue?

[ocl80]

Oh right! - for the time being at least my KH8 remains on

Manager: 1.35
Master: 1.39
Slave: 1.01

so perhaps fixing the clock sync issue was targeted only at a previous one, or maybe they just didn't get to my system yet... Incidentally though I did notice the other day that the local time on the inverter had drifted a couple of minutes slow.

Regarding the firmware version, it remains the case that my fork will work with the newer KH versions but not the old. In the latest release I published, I have tidied up the code so that the old address map is still documented and I use a new inverter profile label for the newer ones, but making it work with both would need new logic in the setup flow which is a bit beyond me at the moment.

[WyndStryke]

Oh right! - for the time being at least my KH8 remains on

It might be that the 'time sync' update was <= 1.35. I have a very new installation, and it had an old firmware on it to start with.