Sofar 3000TL-G3 inverter shows spikes

[sarmbruster]

I do have a Sofar 3000TL-G3 inverter with LSW3 logger using sofar_lsw3.yaml config file. Randomly ~ 4 times per hours I do see spikes in sensor.sofar_garage_pv1_power sensor:

The value reported when spiking is 220660 W (220kW) - which is a ridicilously high number. The spikes last for ~ 5 secs.
When using the older solarman plugin (https://github.com/StephanJoubert/home_assistant_solarman) there were also spikes, but they show up as 0W instead of 220kW.

The 3000TL-G3 does have a stable network connection via a WIFI repeater. Ping timings to the repeater are in the double digit ms. I'm confident that networking issues are not the cause for the spikes.

On a separate notice: I do have a Sofar HYD4000-EP as well, with doesn't show any spikes at all.

Any idea why those spikes are happening and why the are now going 220kW instead of 0W with the older solarman integration?

[davidrapan]

Hi @sarmbruster, I want you to record this happening w/ debug logging enabled, thanks!

[sarmbruster]

Will record tomorrow when inverter switches on again and post logs here. Thanks.

[sarmbruster]

Here's the debug log as requested.
For reference the config/custom_components/solarman/manifest.json contains "version": "25.02.10_04".

home-assistant_solarman_2025-02-13T09-03-53.498Z.log

[davidrapan]

I think I have found the issue:

2025-02-13 09:28:31.599 DEBUG (MainThread) [custom_components.solarman.device] [serial] Querying 03 ❘ 0x03 ~ 0000 - 0039 ❘ 0x0000 - 0x0027 #040 failed, attempts left: 4 [AttributeError: 'NoneType' object has no attribute 'registers']
2025-02-13 09:28:31.599 DEBUG (MainThread) [custom_components.solarman.pysolarman.pysolarman] [serial] SENT: a5 17 00 10 45 0d 00 9f 27 71 8d 02 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01 03 00 00 00 28 45 d4 84 15
2025-02-13 09:28:32.638 DEBUG (MainThread) [custom_components.solarman.pysolarman.pysolarman] [serial] RECD: a5 33 00 10 15 0d 1a 9f 27 71 8d 02 01 45 d5 2e 00 33 1c 00 00 ec d7 7e 67 01 04 20 00 0e 53 41 33 45 53 32 33 33 51 33 32 35 33 30 47 34 33 30 56 32 30 30 56 34 33 30 56 34 33 30 02 b3 52 15
2025-02-13 09:28:32.638 DEBUG (MainThread) [pymodbus.logging] Processing: 0x1 0x4 0x20 0x0 0xe 0x53 0x41 0x33 0x45 0x53 0x32 0x33 0x33 0x51 0x33 0x32 0x35 0x33 0x30 0x47 0x34 0x33 0x30 0x56 0x32 0x30 0x30 0x56 0x34 0x33 0x30 0x56 0x34 0x33 0x30 0x2 0xb3
2025-02-13 09:28:32.638 DEBUG (MainThread) [pymodbus.logging] decoded PDU function_code(4 sub -1) -> ReadInputRegistersResponse(dev_id=0, transaction_id=0, address=0, count=0, bits=[], registers=[14, 21313, 13125, 21298, 13107, 20787, 12853, 13104, 18228, 13104, 22066, 12336, 22068, 13104, 22068, 13104], status=1) 
2025-02-13 09:28:32.639 DEBUG (MainThread) [pymodbus.logging] Frame advanced, resetting header!!
2025-02-13 09:28:32.639 DEBUG (MainThread) [custom_components.solarman.device] [serial] Querying 03 ❘ 0x03 ~ 0000 - 0039 ❘ 0x0000 - 0x0027 #040 succeeded

Inverter sometimes replies for some reason w/ data from different memory block. We asked for holding memory block data and it replied w/ input memory block data...

Very strange... But I think I can easily fix it by simply checking received data and repeat request if the data don't match the request.

Hi there @githubDante, maybe you would like to take a look too?

[davidrapan]

feat: Add modbus response pdu validation

[sarmbruster]

Wow, that was a fast one 🎉

How can I testdrive this? Is a "Download again" in HACS good enough to fetch your commit?
(sorry I'm not deep in HACS at all)

[davidrapan]

Unfortunately not, but I'm in the middle of releasing a new version right now. 😉

[sarmbruster]

Updated to v25.02.14 and there are no spikes any more - running now for 90min on the new version. Thanks for fixing this so quickly.

[davidrapan]

Awesome!

Can you record one more debug log so I can confirm that it works exactly as intended?