Incorrect inverter information being displayed in HA w/ 2.1.1

[turtle2472]

Description

After updating the Span Panel firmware to spanos3/r202633/04 (~24 hours ago), the integration now only surfaces one solar inverter instead of two, and the reported nameplate capacity is incorrect.

Environment

• Span integration version: 2.1.1
• span-panel-api: 3.4.1
• Home Assistant Core: 2026.9.1
• Panel firmware: spanos3/r202633/04
• Two physical inverters connected: SolarEdge SE7600H-US, confirmed via a separate SolarEdge Modbus Multi integration talking to the same inverters directly. Serials/model strings:
  • SE7600H-US000XXXX (original unit)
  • USE7600H-USMNXXXX (newer unit — replaced under warranty after a hardware failure on the original)

Possible root cause worth checking: the two inverters' model-string identifiers have different prefixes (SE7600H vs USE7600H) because the second unit is a newer replacement, not the originally-installed model batch. If the integration's inverter discovery/parsing logic expects a single fixed prefix format, this mismatch could explain why only one inverter is recognized and why the aggregated nameplate value comes out wrong.

Expected behavior

With two SE7600H-US inverters connected, the integration should either:

• Expose per-inverter vendor/product/nameplate data for both units, or
• If intentionally aggregated into a single set of sensors, report a combined nameplate capacity of 15200 W (2 × 7600 W).

Actual behavior

Only a single inverter's data is surfaced:

Entity | State -- | -- sensor.span_panel_pv_vendor | SolarEdge sensor.span_panel_pv_product | SE7600H-US sensor.span_panel_pv_nameplate_capacity | 11680.0 W

11680 W doesn't match either a single unit's real nameplate rating (7600 W per SolarEdge's spec for this model) or the correct combined total for two units (15200 W) — it doesn't appear to correspond to any real calculation from the actual hardware.

Logs

No errors or warnings related to this appear in the Home Assistant log (checked a 48-hour window covering the firmware update, filtering for span_panel/span-panel-api plus inverter/pv/nameplate/solaredge/discover). This looks like a silent data-mapping issue rather than a connectivity failure or exception — the integration successfully retrieves data from the panel, but appears to incorrectly collapse or miscalculate the two-inverter response into a single incorrect value.

Additional note (separate from the above, optional)

Unrelated deprecation warning also seen in the logs, possibly worth a quick fix in the same codebase:

WARNING (MainThread) [homeassistant.helpers.frame] Detected that custom integration 'span_panel' calls `device_registry.async_get_device`, which is deprecated because device identifiers and connections are no longer unique across config entries; use `async_get_device_by_identifier`, `async_get_device_by_connection` or `async_get_devices` instead at custom_components/span_panel/extension.py, line 676. This will stop working in Home Assistant 2027.8.0.

[turtle2472]

I just looked into the angle of the newer inverter vs the older in the SolarEdge Modbus integration. The two inverters also differ structurally, not just by model-string prefix — only the newer USE7600H-USMNXXXX unit has an attached revenue-grade meter (SE-RGMTR-1D-240C-C) reporting alongside it; the original SE7600H-US000XXXX unit has no meter attached. If Span's discovery logic branches on the presence/shape of an attached meter, that could be an additional or alternate factor in why only one inverter gets picked up correctly.

[cayossarian]

The best way to determine the root cause is to look at the data coming from the panel directly with a tool like MQTT explorer (MAC) or MQTT Studio (Windows) or similar. You can contact me for additional assistance if needed and we can connect live. I'm in CA/USA. Using these tools you can see exactly what the panel is publishing.

When connecting turn off 'Validate Certificate" but use TLS Encryption. Point to the panel's IP and port 8883, the user name is the panel serial number and the password is your pass phrase from the mobile app (same used to connect the integration). You subscribe to the topic ebus/# which in MQTT explorer is in the advanced tab and set before you connect.

Once connected your devices show up in the root list of the topics. If you're comfortable posting that here please do so with the inverters expanded so we can see what their children are. It could be something as simple as the panel was not properly configured to see the inverter when they swapped it or there could be something more serious in how the data is published or yes the integration is not interpreting that data properly.

[turtle2472]

Connected directly to the panel's eBus MQTT broker per your instructions (mosquitto_sub, TLS with cert validation off, port 8883, subscribed to ebus/#). Here's what I found:

Only one PV device exists on the panel's own eBus tree — confirmed by counting every device type published:

Device type | Count -- | -- energy.ebus.device.bess | 1 energy.ebus.device.circuit | 25 energy.ebus.device.distribution-enclosure | 1 energy.ebus.device.lugs | 2 energy.ebus.device.mid | 1 energy.ebus.device.pv | 1

The single PV device's full topic tree:

ebus/5/<panel-serial>-se7600h-us/$state ready
ebus/5/<panel-serial>-se7600h-us/$description {"homie": "5.0", "version": 1789063251170, "type": "energy.ebus.device.pv", "name": "<panel-serial>-se7600h-us", "nodes": {"info": {"name": "info", "type": "energy.ebus.capability.info", "properties": {"vendor-name": {...}, "model": {...}, "serial-number": {...}, "firmware-version": {...}, "nominal-power": {"name": "Nominal power", "datatype": "float", "unit": "W"}}}}, "children": [], "root": "<panel-serial>", "parent": "<panel-serial>"}
ebus/5/<panel-serial>-se7600h-us/info/vendor-name SolarEdge
ebus/5/<panel-serial>-se7600h-us/info/model SE7600H-US
ebus/5/<panel-serial>-se7600h-us/info/nominal-power 11680

Note: no serial-number value was published for this device during my capture window, even though the schema declares it as a property — only vendor-name, model, and nominal-power came through.

This looks like it confirms your first theory — the panel itself only has one registered PV device, and the nominal-power: 11680 is coming straight from the panel's own eBus data, not from anything the integration is calculating or misparsing. The integration appears to be faithfully reporting exactly what the panel provides. This points at the panel's inverter-discovery/registration process not picking up the second (replacement) inverter as its own distinct eBus device when it was swapped in, rather than anything on the HA integration side.

[cayossarian]

I have a query out to the API dev whether the panel can recognize/publish more than one PV.

Since you swapped the original hardware the issue is a bit muddled but it could be a flaw in the firmware API or a bad registration at SPAN so let's see what comes back from my query.

[turtle2472]

If it would help your query with the API people, I can give you SN. Just let me know and I’ll PM it to you.

[cayossarian]

Yes please DM me the serial number.

[turtle2472]

I'm genuinely curious what the response comes back on this one.

[dcj]

@turtle2472
I have confirmed that as-of-now SPAN API doesn't represent more than one commissioned PV systems accurately
I will investigate & explore how to add this feature, on the stated "as time and resources permit" basis.
Thank you for reporting this, I appreciate it.
Please contact me directly to discuss/explore some options that might enable me to develop & test a solution more-easily/faster.

[turtle2472]

@dcj
You can email me at my username here at gmail. I’ll happily help as I can.