Status field always shows as Unknown in Home Assistant, but values are published on lxp/.../inputs/all (EG4 6000XP)

[presto8]

I have an EG4 6000XP and lxp-bridge listening to its traffic on port 8000 and then publishing the events via MQTT. Most of the values show up properly in Home Assistant, but in particular "status" and "fault_code" always show as "Unknown" in home assistant.

The lxp-bridge logs show a value for "status" and "fault_code" every minute or so when the datalogger broadcasts it status:

[2024-03-08T14:51:24.247 INFO lxp_bridge::mqtt] publishing: lxp/BJ0000000/inputs/all = {"status":16,"v_pv_1":0.0,"v_pv_2":0.0,"v_pv_3":414.1,"v_bat":53.2,"soc":84,"soh":0,"internal_fault":0,"p_pv":0,"p_pv_1":0,"p_pv_2":0,"p_pv_3":0,"p_battery":-10,"p_charge":0,"p_discharge":10,"v_ac_r":243.8,"v_ac_s":0.0,"v_ac_t":614.5,"f_ac":59.98,"p_inv":0,"p_rec":49,"pf":0.0,"v_eps_r":243.4,"v_eps_s":513.6,"v_eps_t":416.2,"f_eps":59.99,"p_eps":0,"s_eps":0,"p_grid":416,"p_to_grid":0,"p_to_user":416,"e_pv_day":0.0,"e_pv_day_1":0.0,"e_pv_day_2":0.0,"e_pv_day_3":0.0,"e_inv_day":0.0,"e_rec_day":0.9,"e_chg_day":0.3,"e_dischg_day":2.5,"e_eps_day":2.2,"e_to_grid_day":0.0,"e_to_user_day":10.9,"v_bus_1":415.7,"v_bus_2":374.5,"e_pv_all":0.0,"e_pv_all_1":0.0,"e_pv_all_2":0.0,"e_pv_all_3":0.0,"e_inv_all":0.0,"e_rec_all":22.9,"e_chg_all":19.9,"e_dischg_all":10.6,"e_eps_all":9.3,"e_to_grid_all":0.0,"e_to_user_all":33.9,"fault_code":0,"warning_code":0,"t_inner":12,"t_rad_1":26,"t_rad_2":32,"t_bat":12,"runtime":367123,"max_chg_curr":150.0,"max_dischg_curr":300.0,"charge_volt_ref":57.4,"dischg_cut_volt":0.0,"bat_status_0":0,"bat_status_1":0,"bat_status_2":768,"bat_status_3":0,"bat_status_4":0,"bat_status_5":12912,"bat_status_6":43967,"bat_status_7":782,"bat_status_8":24800,"bat_status_9":149,"bat_status_inv":3,"bat_count":3,"bat_capacity":300,"bat_current":0.0,"bms_event_1":0,"bms_event_2":0,"max_cell_voltage":3.332,"min_cell_voltage":3.328,"max_cell_temp":10.0,"min_cell_temp":9.0,"bms_fw_update_state":531,"cycle_count":5,"vbat_inv":53.1,"v_gen":1310.7,"f_gen":133.67,"p_gen":12336,"e_gen_day":1361.9,"e_gen_all":212.9,"time":1709938284,"datalog":"BJ000000"}

I see in the source code that the status codes are mapped to human-friendly strings, but I don't see anything like that published to Home Assistant.

Any tips on what I should do to get the status working? Thanks!

[celsworth]

I think this is probably a configuration option that needs turning on - see https://github.com/celsworth/lxp-bridge/wiki/MQTT-Publishes#lxpdataloginputxparsed-payload--string-added-in-v012

Try setting mqtt.publish_individual_input true - this should probably become the new default to be honest.

[presto8]

Setting publish_individual_input: true fixed it immediately, thanks!

[presto8]

Moving on to the next field I'm interested in :-), ac_first/1 (lxp_bjxxxxxx_ac_first_timeslot_1 in Home Assistant) is also showing as Unknown. Setting the publish_holdings_on_connect: true didn't seem to help with that variable.

Because the 6000XP does not have a concept of day of the week, I would like to set up an automation to make AC first all day on weekends (when my utility charges cheap rates). Then on weekdays, there would be a schedule for AC first versus battery first.

[celsworth]

Hm that one is a bit trickier, that should already work by default.

For debugging you can check the following. There should be a retained topic on your MQTT broker for homeassistant/text/lxp_3333333333/ac_first_1/config (replacing the serial). This is the one that tells HA that there is something of interest and where to find it. In that, you can see which topic we tell HA to watch for for updates, for example:

"state_topic":"lxp/3333333333/ac_first/1"

So you could try watching for that topic being broadcast. I'm just trying to track down where it actually gets sent, wondering if this got missed out. If you specifically try to read that register, does that send it (and then populate HA?)

See https://github.com/celsworth/lxp-bridge/wiki/MQTT#lxpcmddatalogreadac_chargex-payload--empty-added-in-v010 for an example of how to do that (publish to lxp/cmd/$datalog/read/ac_first/1)

[presto8]

Thanks for the tips! I will explore around. This is my first time really using MQTT, so I am having to learn some fundamentals along the way. One thing, when I use MQTT explorer to subscribe to Mosquitto, I see the topic lxp/BJ3333333/hold/1 through 79 only. According to the documentation, they should go a lot higher? I have AC charge start 1 set to 10 pm and it shows up as hold/68 = 22, so it seems to be working for the hold registers that do show up. It looks like AC First is 152-157, which I don't see under hold at all.

[presto8]

With the excellent instructions you gave I was able to make a lot of progress!

I wrote a Python script that subscribes to the appropriate topics. When I issue an explicity read using lxp/cmd/.../read, it works perfectly (and the value shows up in Home Assistant). However, on initial startup, and also if I change the value on the web dashboard, the value does not get re-published automatically. The next time I do a manual read, the value gets updated. I ran the experiment below to confirm.

Is there possibly an issue with publishing the values when they are not explicitly read using lxp/cmd/.../read? Everything else seems to be working perfectly, reading, setting, home assistant, etc.

(Lines starting with NOTE: were added by me after the experiment)
Connected to hedghog.local:1883 with result code 0
sub lxp/cmd/#
sub lxp/result/#
sub lxp/BJ33333333/ac_first/3/#
rcv lxp/BJ33333333/ac_first/3 payload=b'{"start":"04:00","end":"05:00"}'
  NOTE: Stale value, web dashboard shows 1:00 and 2:00
slept 1 second
pub lxp/cmd/BJ33333333/read/ac_first/3
rcv lxp/cmd/BJ33333333/read/ac_first/3 payload=b''
rcv lxp/BJ33333333/ac_first/3 payload=b'{"start":"01:00","end":"02:00"}'
rcv lxp/result/BJ33333333/read/ac_first/3 payload=b'OK'
  NOTE: This is the (correct) value shown on the web dashboard
slept 1 second
pub lxp/cmd/BJ33333333/set/ac_first/3 payload={"start":"20:00", "end":"21:00"}
rcv lxp/cmd/BJ33333333/set/ac_first/3 payload=b'{"start":"20:00", "end":"21:00"}'
rcv lxp/BJ33333333/ac_first/3 payload=b'{"start":"20:00","end":"21:00"}'
rcv lxp/result/BJ33333333/set/ac_first/3 payload=b'OK'
  NOTE: Confirmed it was changed on the web dashboard
sleeping 15 seconds, change ac_first/3 on web dashboard...
  NOTE: After about 5 seconds, I manually set start=6 and stop=7 on web dashboard
  NOTE: Waited about 10 more seconds, no further messages received
slept 15 seconds
pub lxp/cmd/BJ33333333/read/ac_first/3
rcv lxp/cmd/BJ33333333/read/ac_first/3 payload=b''
rcv lxp/BJ33333333/ac_first/3 payload=b'{"start":"06:00","end":"07:00"}'
rcv lxp/result/BJ33333333/read/ac_first/3 payload=b'OK'
  NOTE: This is the correct value, but needed a manual read to retrieve

[celsworth]

If you restart lxp-bridge, does HA get the current values then? It should broadcast all the holding registers on restart (if publish_holdings_on_connect is set).

Actually I think I remember this has been covered before that lxp-bridge doesn't automatically re-broadcast the new values if they change externally - unfortunately it's a bit of a tricky fix and was deemed a bit of an edge case so not really high priority. See here (this is about force discharge but the principles are all the same: https://github.com/celsworth/lxp-bridge/issues/207)

[presto8]

Thanks, Chris. You are correct! If I set publish_holdings_on_connect: true, then it does indeed publish everything on startup and updates the values to what is shown on the web dashboard.

So if I set both publish_individual_input: true and publish_holdings_on_connect: true, then it makes sense. On initial startup, it will publish all of the current data. But it may not rebroadcast new values without an explicit read. And as per linked issue, it is tricky to fix. That's easy to work around however, now that I know that it is working as intended :)

Thanks for helping out with this issue. I learned a lot in the process!

[presto8]

Did some more digging: ac_first/3 start time and stop time are hold/156 and hold/157 respectively. However, it seems there is an error when processing the status announcements on or after hold/79 is processed. hold/79 appears in the MQTT subscription, but nothing above that, unless an explicit read is made.

I noticed this in the lxp-bridge debug logs - could it be related?

[2024-03-11T13:28:22.624 DEBUG rumqttc::state] Publish. Topic = lxp/BJ333333/inputs/all, Pkid = 100, Payload Size = 1563

[2024-03-11T13:28:22.783 DEBUG lxp_bridge::lxp::packet_decoder] 291 bytes in: [161, 26, 5, 0, 29, 1, 1, 194, 66, 74, 51, 53, 48, 55, 48, 52, 56, 56, 15, 1, 1, 4, 52, 48, 51, 51, 55, 52, 48, 48, 51, 53, 127, 0, 254, 184, 4, 184, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 9, 0, 0, 0, 133, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 186, 4, 188, 4, 0, 0, 0, 0, 0, 0, 0, 0, 20, 0, 47, 0, 0, 0, 0, 0, 24, 0, 128, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 147, 35]
[2024-03-11T13:28:22.783 DEBUG lxp_bridge::coordinator] RX: TranslatedData(TranslatedData { datalog: BJ3333333, device_function: ReadInput, inverter: 4033740035, register: 127, values: [184, 4, 184, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 9, 0, 0, 0, 133, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 186, 4, 188, 4, 0, 0, 0, 0, 0, 0, 0, 0, 20, 0, 47, 0, 0, 0, 0, 0, 24, 0, 128, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0] })
[2024-03-11T13:28:22.783 WARN lxp_bridge::coordinator] ignoring unhandled ReadInput register=127 len=254
[2024-03-11T13:28:22.783 WARN lxp_bridge::mqtt] ignoring unhandled ReadInput register=127 len=254

[celsworth]

This isn't related; this message related to input registers (read-only, transient data, like power data/temperatures/etc), and we're concerned with hold registers (read/write settings).

This message does mean that we're missing some data from the inverter in those higher registers, but it's not related to the current issue.

[presto8]

And just to document in case anybody else should run into this issue, changing the value from Home Assistant side works as expected. Home Assistant updates the value, it is sent to lxp-bridge, and then the value is rebroadcast as changed. Since I plan on using the EG4 web interface as little as possible and do everything via Home Assistant automations and UI, it will be seamless without any further changes needed.