Disconnect every few days and then "No ECU found at this IP-address or ECU is recovering from restart"

[kjroberts03]

Requirements for submitting a bug

Help us by sharing us the relevant home-assistant.log part.

`Traceback (most recent call last):
  File "/usr/src/homeassistant/homeassistant/helpers/update_coordinator.py", line 380, in _async_refresh
    self.data = await self._async_update_data()
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/src/homeassistant/homeassistant/helpers/update_coordinator.py", line 281, in _async_update_data
    return await self.update_method()
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/config/custom_components/apsystems_ecu_reader/__init__.py", line 151, in do_ecu_update
    return await ecu.update(
           ^^^^^^^^^^^^^^^^^
    ...<3 lines>...
    )
    ^
  File "/config/custom_components/apsystems_ecu_reader/__init__.py", line 112, in update
    data = await self.ecu.query_ecu(port_retries, show_graphs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/config/custom_components/apsystems_ecu_reader/ecu_api.py", line 145, in query_ecu
    _LOGGER.debug("Inverter raw data: %s", self.inverter_raw_data.hex())
                                           ^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'hex'`

Expected Behavior

System should be running

Current Behavior

Every few days it disconnects like this, I have to log into manager app and connect to the device then it seems to work. The ECU is never not able to be accessed through the EMA App.

PV-hardware (Environment)

System is in basement (one story house) but never loses connection via app.

• ECU-B

• [X ] ECU-R (2160 series)

• ECU-R-Pro (2162 series)

• ECU-C

• ECU-3

• DS3 series

• QS series

• QT series

• YC series

2.3.0 installed

[HAEdwin]

The error tells me that the ECU is receiving empty datastring from the inverter(s). What is the signal strenght on the inverter(s)? Try to bring the ECU closer to the inverters for a while and see if that helps. And I'd recommend installing 2.5.1

[kjroberts03]

It just cut out again, so before I reset it the inverters were showing anywhere from 12-21 dBm, with two of them showing -27 dBm at the time of cut off.

Let me see what I can do with the inverter and moving it closer. Will update.

[HAEdwin]

I have converted this thread to the discussions section because it is not a bug, the error message was solved in a later version.

[kjroberts03]

Hi, seems like the fixes you've suggested have worked. Thank you!

The last few days have shown a different total in kW produced between the HA integration and the EMA app. Any idea why?

[HAEdwin]

Yes, unfortunately the values that are pushed to EMA are different over time than the data from the pull this integration does. Think of it as sampling the windspeed at two different moments, they might not give the same value. APsystems says: "The ECU-C is not a continuous energy meter like the one from the electricity company".

Even then with a five minute interval it should already be a sort of average of what happened in those five minutes (but I'm not sure if that is the case).

So overall it's more like an indication, the most accurate measuring point would indeed be the energy meter from the electricity company. Also take into account that the maintenance the ECU does at night sometimes causes different day values ​​than initially seen. So also check the values ​​of the previous day they might match better.

Great that moving the ECU helped. Over the years I now know the ECU inside and out :)

[HAEdwin]

@kjroberts03 For the sake of clarity I will close this discussion. Good that it helped you to improve the connection between the inverters and the ECU. Of course the loss of connection will only affect the monitoring but not the actual continuity in feeding back to the grid. Feel free to start a new discussion if necessary.

[kjroberts03]

@HAEdwin Sorry to bug you again. After reconnecting the system this morning (I was away on vacation when it disconnected a few days ago), it has now just disconnected again. Most inverters were reading -8 to -16 before the disconnect, and another two read -20 and -24. The EMA app didn't even register a blip the data I can see on it. Is there a way I can fix this so I'm not having to reconnect to the ECU every day or several days? I could try to move the ECU to the attic (currently on the first floor of a single level home), but that would be a pain to do. Thanks in advance!

[HAEdwin]

No problem, what exactly do you mean by reconnecting the ECU - what do you do then? The integration should be able to retrieve data again when it is available again. Do you have an automation that reboots the ECU with a smartplug? It is best to have both the Zigbee and the WiFi connection optimal, that causes the least disruption but does not rule out that you will have to restart the ECU-R 2160 series periodically if it gets stuck. What does the time graph look like when using the cache?

[kjroberts03]

Hi, sorry for the delayed responses. To get the ECU back onboard after it has completely lost connection, I have to open the access point wifi on the ECU and connect via the EMA Manager app. Once I connect to it with the Manager app, I restart HA and it always comes back.

Yesterday I saw that the Current Power reading from the panels had been reporting the same value for about three hours. It had been changing to Cache 1, 2, 3, 4 over and over for a few hours. I hit the reboot ECU option (which isn't an option when it fully disconnects), and that fixed it. Maybe I'll have to set up an automation if the Current Power reading stays constant above 0 for more than 15 minutes.

The graphs go completely flat on all readings when it's in this mode.

[jurasjea]

Hi ! what I don't understand is why the zigbee signal between the inverters and the ECU would be high enough for the ECU to send data to APS servers but not enough for the integration ?

[HAEdwin]

The ECU firmware is designed to provide optimal updates to EMA. Despite this, the firmware also has a maintenance cycle at night to check data and correct it if necessary. The integration uses the installation part of the firmware software and in that respect does not differ in operation from the EMAmanager app or the ECU app. What ultimately helps is a stable connection of the ECU on both Zigbee and WiFi. It makes the operation of the EMAmanager app and this integration more stable, but even then an ECU reboot is sometimes necessary. This may be related to the IO that must take place with EMA and now also with the integration. This could cause conflicts. It's more an ECU issue than an integration issue. But in my home it works pretty good and not much reboots are needed.

this is my production graph (production always runs)

this is my development environment (runs at the same time and lot's of experimenting, extra queries etc.)

Antenna's are best place like \ / and pretend you stick half an orange on the antenna, that is the transmit and receive directions to other nodes. Although Zigbee is a mesh protocol, it can be easily disrupted by other (smart) devices and WiFi because it also works on the 2.4Ghz band. A meter cupboard is a bad place for the ECU as well as reinforced concrete walls, large refrigerators - well everything in the line of sight.

[jurasjea]

Thanks for the explanations ! I think I will buy an external antenna to improve the quality of the signal...

[HAEdwin]

@kjroberts03 you wrote: "It had been changing to Cache 1, 2, 3, 4 over and over for a few hours. I hit the reboot ECU option (which isn't an option when it fully disconnects), and that fixed it". I was wondering if you then came across the log line: "Restarting ECU after x failed attempts to fetch data". What is the "Cache count before auto ECU reboot" set to in your case? The same routine is called as when you manually press the button. I think I also make the automatic reboot more explicit in the notification area on the bottom left in the HA UI.

[kjroberts03]

Right now "Retries when ECU fails to respond" is set to 5 and "Cache count before auto ECU reboot" is set to 5. Should I change that second value to 3 or 4?

[HAEdwin]

Well that depends on what is acceptable in your situation.

The way it works is as follows. Every five minutes the integration tries to contact the ECU, if it fails then the integration waits 1 second and tries as many times as you have set (5x). If it is ultimately not successful, that means 1x cache count (up to the number you have set) if it is not successful the 5th time, then the integration reboots the ECU. You should be able to find that in the Home Assistant log.

For example, if the cache count does not exceed 4, then it has apparently succeeded in contacting the ECU again because the cache counter is then reset. If that keeps repeating for a few hours you would better set the cache count to 3 to enable the integration to reboot the ECU in a earlier stage.

I have mine set to 3 retries, if that fails 3 times in a row (15 minutes total) the ECU will reboot. So 3 retries and 3 cache counts.

The new version will explicitly give a notification when the ECU has been rebooted.

[kjroberts03]

Thank you for the explanation! I will try 3 and 3 to see if this helps the situation.

[HAEdwin]

I'll close this discussion. If for any reasons you'd like to reopen it, feel free to ask
Cleaning the discussions🧹

[kjroberts03]

Hello,

I've been having this issue pretty much since the last time I posted about it. Moved it from basement to first floor (one floor home), and still disconnects a few times a week. So I ran cable to the attic and moved the ECU up there to get it as close to the inverters as possible. It still disconnects about 3-4 times a week. One inverter reads lower than the others (the last few hours it's been between -22 and -24). Any advice? To get it to rejoin, I've been having to go up to the attic and join the ECU's wifi, then connect to it via the EMA manager app. That always brings it back online with HA.