Beta testing of v2.3

[gskjold]

Attached is a new version with changes outlined below. If anyone could test this on their setup, it would be highly appreciated.

!Warning!
This is a firmware which is still under development. Testing this firmware is on your own risk. Please only test if you have possibility to flash via TTL or USB if anything goes wrong.

!Please note!
If you upgrade to this version and then downgrade, you will loose all your data.

Changes:

• Custom price configuration, option to add tariff, VAT etc
• Realtime energy use plot
• Option to disable LED (Pow series hardware depends on version)
• Multiple languages
• Dark mode
• Support for ethernet boards
• Graph per phase where available (wattage and amperage)
• Support for meters with pulse output
• MQTT LWT
• IPv6
• Option to use as Access Point in normal operation
• Option to set context path of webserver (When used behind proxy)
• Changed color for export graph
• Reconnect to correct host/ip if hostname/ip changes on config page or via config file

Please let me know when you have tested this new firmware and include the following information:
Country:
Meter:
Hardware:

23cbcf9

esp8266.zip
esp32.zip
esp32solo.zip
esp32s2.zip
esp32c3.zip

f4de3e6

• Fixed language switching
• Fixed ESP32 UART2
• Added support for HDLC data segmentation
• Added automatic unit scaling for various counters (wh/kwh/mwh)

esp32.zip
esp32c3.zip
esp32s2.zip
esp32solo.zip
esp8266.zip

7c4c096

• Fixed support for HDLC data segmentation
• Added support for more OBIS codes from some Swiss meters
• Changed unit scaling limits
• Added timeout to MQTT to avoid unresponsive GUI on connection error

esp32.zip
esp32c3.zip
esp32s2.zip
esp32solo.zip
esp8266.zip

766849d

• Fixed overflow in data storage for day and month graph
• Changed data storage strategy when list 3 is lost just after new hour
• Added different JSON formats (NOTE: data.id renamed to data.meterId)
• Adjusted colors on graphs
• Forcing GPIO for known board profiles on boot

esp32solo.zip
esp32s2.zip
esp32.zip
esp8266.zip
esp32c3.zip

db90dbf

• Fixed stacking labels on phase graphs
• Fixed incorrect loading of day and month data
• Fixed slow loading of realtime plot
• Fixed real time update of realtime plot
• Fixed incorrect zero values in data for certain meters
• Introduced sequential loading of data from device
• Restored old icons for header shortcuts

esp32s2.zip
esp32solo.zip
esp32.zip
esp8266.zip
esp32c3.zip

[ArnieO]

Country: Norway
Meter: Kamstrup Omnipower
Hardware: Pow-K v1.5.1

Have not seen any functional issues.

Minor issues:

• Language switching not working (it did work on previous test version). When I selected an other language it jumped back to English with then "English" appearing two times in the dropdown menu:
  

• Icons:
  

  • The (i) icon seems too small compared to the others
  • The cogwheel on previous test version is replaced by a sun?

[ElMyggo]

Screenshot from Samsung Phone, android.

[ElMyggo]

This is from my raspberry with browser chromium

[ArnieO]

This is from my raspberry with browser chromium i think

The link does not work.

[ArnieO]

Screenshot from Samsung Phone, android.

I confirm; I see the same (Samsung phone, stock browser, Android)

[ElMyggo]

Now fixed

[ElMyggo]

Installed on Mini D1 ESP32 and same problem with language but just do a refresh in browser (F5) and its working

But no contact with meter (Kamstrup DK8660 OMNIPOWER)

Earlier i use a old beta 3d1d99b with UART2 (16) (generic ESP32) but same settings in this beta and no contact

Settings

this is my old working setup:

[ElMyggo]

Yes! now UART2 working!
But Power per Phase seem not showing any data
Is it possible not to scale the meter counter?

[gskjold]

Are you sure your payload contains this data?

I will adjust the scaling

[ElMyggo]

Aha i think it was just a calculation, and no i dont think my meter have this information. Kamstrup DK8660 Omnipower
Thanks for the scaling change

[ArnieO]

Are you sure your payload contains this data?

If the meter does not report phase power, the graph should not be shown?

[ElMyggo]

Thats correct, i have force the setting to activate, when in auto it doesnt show

[Cartoon-Dude]

error: HAN: Unknown data received, check meter config
Country: Switzerland
Platform: esp8266
Meter: Iskrameco AM550

[ArnieO]

OK, so it is missing in the payload - and there is of course nothing we can do about that.

[Cartoon-Dude]

I can use the tension from the inverter and combine the data together, it is all right. At least the most important is there, and can finally know the self usage amount of the produced energy like this.

I just need to add the price info after the official release.

[ArnieO]

I just need to add the price info after the official release.

Price should be available also in the test version.

[Cartoon-Dude]

I think i did something wrong because it crashed and not stared any more, needed to reflash.
I would wait for the faq update with the release and create a bug if there's still a problem.

[gskjold]

Added current sensors in the latest build

[Buruglen]

Country: Norway
Meter: Aidon 6484SE
Hardware: Pow-U+

The (simulated) meter reading has up to now been in kWh. With 2.3 it is scaled and rounded and for me it is shown in MWh. I would like it to show kWh. Same goes for consumption.

[Buruglen]

Firmware f4de3e6.

In list of languages I think German should be Deutsch, not Deutsche. It is Die deutsche Sprache, but in a list of languages it should be Deutsch.

[gskjold]

Fixed

[Buruglen]

To be picky, Norsk should be Norsk bokmål, and Norsk (Nynorsk) should be Norsk nynorsk. Both bokmål and nynorsk are Norwegian (Norsk). Norwegian is a macrolanguage by ISO 639, with two written forms.

[ArnieO]

Picky is good; so thank you!
I would personally prefer both with brackets, like this:

• Norsk (bokmål)
• Norsk (nynorsk)

[Buruglen]

Yes, that is fine!

[gskjold]

Updated language files

[Buruglen]

Is there a place to understand how to use the new feature Custom price configuration?

I can see some discussions here: https://github.com/UtilitechAS/amsreader-firmware/issues/252.

But a short wiki would be of help (at least for me).

[ArnieO]

Will be done when 2.3 is released.

[Buruglen]

I have Pow-U+. I cannot see an option to turn off the LEDs. Is this feature not possible on this hardware?

[ArnieO]

Hardware change was needed to implement this for the blue part of the LED.
This is a new feature from Pow-U layout v1.8.1 / Pow-K layout v1.6.1 (in sale from next month) / Pow-P1 layout v1.2.1.
I believe your card is older.

[Buruglen]

My Pow-U is layout 1.8.1. Where in the settings should I see this feature?

[gskjold]

I think this is a problem when upgrading from 2.2 to 2.3. If you go to the status page and click on "Pow-U+", you will go to the vendor menu where you choose hardware. Just save this once and then go the the config page. You should now have a "LED behaviour" selection under "Hardware"

[Buruglen]

Thanks. That worked! I see that the green USB-port LED is still on. During the beta process I keep the unit powered via USB.

[ArnieO]

The green USB-port LED is there to indicate that the USB port is powered. You turn it off by removing the USB power.

[Buruglen]

Just out of curiosity: What information will be sent from amsleser to MQTT when it receives an LWT (Last Will and Testament)?

It must be something it has "kept for later", but should send ASAP before MQTT dies.

[gskjold]

The only thing we are listening on is the "online" event from HA, then it will re-publish sensor configuration. There are no actions taken if LWT is received.

Firmware will send LWT ("offline") upon disconnect.

[Buruglen]

OK.

[Buruglen]

A small typo from 2.2.28 that I suspect is still in there (hard to reproduce).

Message from the Price API PriceAPI: Server says its too early. Should be PriceAPI: Server says it's too early or Price API server says it is too early.

[ArnieO]

Thank you for noticing!

[gskjold]

Updated language file

[Buruglen]

Probably a feature request, unless it is part of the Price configuration-feature: Add the monthly prices for each level in Tariff thresholds for Norway and Sweden.

[ArnieO]

We'll have to think about that, as it is virtually impossible to make this estimate reflect your bill 100%.
In Norway, the power threshold "penalty" incurred in one calendar month is added to the grid fee bill for the next month.

[ElMyggo]

When i flashed latest beta build (7c4c096) i cant get the card to start in APmode (no wifi showing)
If i flash (f4de3e6) it works.
Im using "flash_download_tool_3.9.5" for windows and i erase chip before flashing
My card is "AZ-Delivery D1 mini 32" with ESP32-WROOM
Is there any trick i can test?

[gskjold]

No,but the size probably increased. I know at some point (99% flash use or so) it does not boot anymore

[ArnieO]

OK, that could explain it.

[ElMyggo]

Find a ESP32-S2-SOLO module and have to build a test jigg....
Flashed with ESP32-S2 files and up and running on latest beta now,
only trouble i have was i have to change from UART0 to GPIO44 in settings to get HAN data
Maybe different config in S2 and S2-SOLO
(I should test ESP32-WROOM board when i get the new board i order)

[ArnieO]

Good! 👍

Maybe different config in S2 and S2-SOLO

"-SOLO" is only a form factor difference, the GPIOs and config are the same. We use S2-MINI in our products due to smaller physical footprint.

[ElMyggo]

Beta #766849d up and running on "AZ-Delivery D1 mini 32" with ESP32-WROOM so there is something with the previous version

[avandorp]

Has anyone managed to get the beta firmware running on an esp32c3? OTA update from 2.2.28 to any of the beta versions stays at 2.2.28 and a full flash via USB gives an unresponsive unit which has to be reflashed with 2.2.28 to work again.

[gskjold]

That is interesting, I will need to look into that

[ElMyggo]

I tested beta 766849d on C3 and now works
but not 7c4c096 (same issue as avandorp)

[avandorp]

OTA update for new firmware with meter power didn't work for my C3, OTA with USB power didn't work either, flashing directly via USB finally worked. Yay!
Voltage, amperage, power factor, phase power for a Regionalwerke provisioned E450 still don't show anything. Total in/Total out reactive power strangely enough switch between 0 and some kVArh (interestingly in is reported in VArh and out in kVArh, even though in is 42 kVArh and out 822 kVArh)


EDIT: Ooh, while the graphs don't show any phase related data, I get amperage data per phase via MQTT, yay. Accumulated reactive import and export switch between 0 and some possibly correct values not only in the web interface but also via MQTT.
EDIT 2: Amperages drop to 0 sometimes too, always simultaneously with the reactive values.
EDIT 3: The amperage and reactive in/out totals dropping to 0 happen, when the meter sends the additional tariff 1 and 2 values every 60s:
0-9:25.9.0; 2 Object list push settings end user information 2
0-9:25.9.0; 1 OBIS Code Push Settings end user information 2
1-1:1.8.1; 2 Energy import +A (QI + QIV) Tariff 1
1-1:1.8.2; 2 Energy import +A (QI + QIV) Tariff 2
1-1:2.8.1; 2 Energy export -A (QII + QIII) Tariff 1
1-1:2.8.2; 2 Energy export -A (QII + QIII) Tariff 2
I think they should be ignored.

[gskjold]

I have added another update today

[avandorp]

Mine has "recovered" but with half the month strange data:

[Buruglen]

Country: Sweden
Meter: Aidon 6442S
Hardware: Pow-P1

I updated from 2.2.8 to 766849d yesterday.

The Energy use last 29 days (kWh) immeditately after installation showed strange values. All numbers were the same and totally wrong. See next image.

Then today only yesterday's numbers show up even if the Pow-P1 had been up more than 30 days. See below.

Luckily I have my consuption number also on Home Assistant.

[ArnieO]

@Buruglen This is exactly what happened to me too - and it never recovered the historic values.

[Buruglen]

I looked at the Energy use last 29 days (kWh) both using Firefox and Chrome right after the beta firmware installation, same strange values. Then after a while, could have passed top of the hour, on iPad and iPhone, it showed no values at all in that menu. Now this menu is updated normally, but all values prior to firmware update (for this menu) is gone. The other numbers seem corect.

[ArnieO]

Exactly the same I saw. After passing of whole hour, the graph points were reset to zero, but all historic values were gone.

[Cartoon-Dude]

Firmware: 766849d
The text of the scale is overlapping.

Just vor curiosity, the reactive power is a data sent by the meter or calculated?

[gskjold]

Reactive comes from the meter

Will have a look at the overlapping text

[ElMyggo]

OK, now beta 2.3 #766849d up and running on 3 board with different hardware.
It also working on my ESP32 wroom module but not on previous version (7c4c096)
I also can confirm that previous beta doesnt work on my C3 but this version works

[ArnieO]

previous beta doesnt work on my C3 but this version works

Good to hear!

[ElMyggo]

Time scale on Real time plot is to fast when i have browser stay open...
Looks like the plot updates every 10 sec (same as meter update) so the scale of minutes goes wrong
If i do a refresh (F5) its correct but continue to update to fast if browser stays open

screenshot from new installation whitout refresh

After a refresh (F5)

[ArnieO]

Thank you for that observation, it seems like the horizontal scaling of the real time plot is initially incorrect, until a first refresh (F5).

[ElMyggo]

I have planned to use ESP32-C3-DevKitM-1 in my next build and they have a Addressable RGB LED (WS2812), driven by GPIO8
Can i configure this led or need to rebuild the code
And found this setting with no instruction...

[ArnieO]

Can i configure this led or need to rebuild the code
And found this setting with no instruction...

This is not yet documented, as it is new in v2.3:
The newest versions Pow-K, Pow-U and Pow-P1 have a facility to deactivate the onboard RGB LED. We did that in response to customer requests, for situations where the power meter is visible and the LED is disturbing at nighttime.
If Pow-K, Pow-U or Pow-P1 is selected in the initial configuration, the user gets this config tile where the RGB LED can be deactivated:

• The red part of the LED is used for error signalling, and will not be deactivated.
• The green part of the LED is controlled by the microcontroller
• The blue part of the LED is controlled directly by the incoming data signal, and can on these new boards be shut down via a MOSFET transistor. That transistor is controlled by a GPIO from the microcontroller.

When you configure the board using the "Generic ESP32xx" setting, you get some more detailed choices on the config page. One of those id the ability to select the GPIO that on the Pow-x boards can be used for controlling the LED as described above. This is the "LED dis. GPIO".

However, the "Hardware" tile (where the "LED dis. GPIO" can be set) is not shown when you use the "Generic ESP32xx" setting. Anyhow, all you could use it for (unless you modify the code) is to set that GPIO either high or low, depending on the "LED behaviour" setting.

PS:
Unless you use an ESP32 module with more than 4 MB flash, I do not think you will be able to fit the library to run a WS2812 LED.

[ElMyggo]

OK, thanks for clearing this out, i skip the WS2812 led function for now ;-)

[ArnieO]

I love those addressable LEDs, and have used variants of it in several projects.
But on this hardware we have specifically chosen to use an ordinary RGB LED, so that one colour (blue) is controlled directly by the incoming data signal, not by the micro controller. We than have a sure way to know if the blue LED flashes, the power meter delivers data - disregarding configuration issues or whatever in the micro controller.
You could consider using the same setup on your own hardware, where the R and G part of the LED are controlled by the micro controller.

[ArnieO]

Can i configure this led or need to rebuild the code
And found this setting with no instruction...

This is not yet documented, as it is new in v2.3:
The newest versions Pow-K, Pow-U and Pow-P1 have a facility to deactivate the onboard RGB LED. We did that in response to customer requests, for situations where the power meter is visible and the LED is disturbing at nighttime.
If Pow-K, Pow-U or Pow-P1 is selected in the initial configuration, the user gets this config tile where the RGB LED can be deactivated:

• The red part of the LED is used for error signalling, and will not be deactivated.
• The green part of the LED is controlled by the microcontroller
• The blue part of the LED is controlled directly by the incoming data signal, and can on these new boards be shut down via a MOSFET transistor. That transistor is controlled by a GPIO from the microcontroller.

When you configure the board using the "Generic ESP32xx" setting, you get some more detailed choices on the config page. One of those id the ability to select the GPIO that on the Pow-x boards can be used for controlling the LED as described above. This is the "LED dis. GPIO".

However, the "Hardware" tile (where the "LED dis. GPIO" can be set) is not shown when you use the "Generic ESP32xx" setting. Anyhow, all you could use it for (unless you modify the code) is to set that GPIO either high or low, depending on the "LED behaviour" setting.

[gskjold]

Added another update to the first post. Will probably be the last update before release

[gskjold]

I am curious to why OTA does not work in some situations, I have yet to reproduce this issue and if anyone else have this issue, please run a serial debug.

[ArnieO]

I am curious to why OTA does not work in some situations, I have yet to reproduce this issue and if anyone else have this issue, please run a serial debug.

The symptom I saw was that it got stuck on one of the "install" pages - and the device was unresponsive. No life on the LEDs if I pressed Reset or held AP button. Same each time I tried to roll back to v2.2.28 from any of the beta versions. So I assume you'll be able to reproduce.

[avandorp]

The zeroing of amperage values every 60 seconds is fixed (as seen via MQTT).

Possibly due to the fix for this issue I see a new issue (with an E450): Once active export has a value above 0 it stays there until another value above 0 is encountered:

[gartenbauer]

Hi @avandorp, which meter configuration settings did you use to connect to the E450 provisioned by Regionalwerke Baden?

[avandorp]

Hi @gartenbauer. I use the following settings:

[avandorp]

I am curious to why OTA does not work in some situations, I have yet to reproduce this issue and if anyone else have this issue, please run a serial debug.

I've tried to update to 2.3.1 OTA and had a telnet session running:

(D) Received valid DLMS at 27
(D) handleFileUpload file: /firmware.bin
(D) handleFileUpload Open file and write: 0
(D) handleFileUpload Written: 0
(E) Serial buffer overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(D) handleFileUpload Ended
(W) Used 22941ms to handle web
(W) loop() used 22977ms
(I) Successfully connected to MQTT
(I)   Subscribing to [homeassistant/status]
(I) Received online status from HA, resetting sensor status
(D) Sending translation file for language: en
(D) Language file /translations-en.json
(I) Rebooting

The overflows are certainly suspicious but also reconnecting to MQTT before rebooting seems strange. Anything else I should try?
EDIT: A verbose session isn't much more insightful.

(V) Using application data:
(V) 02 06 01 06  02 04 12 00  28 09 06 00  09 19 09 00 
(V) FF 0F 02 12  00 00 02 04  12 00 28 09  06 00 09 19 
(V) 09 00 FF 0F  01 12 00 00  02 04 12 00  03 09 06 01 
(V) 01 01 08 01  FF 0F 02 12  00 00 02 04  12 00 03 09 
(V) 06 01 01 01  08 02 FF 0F  02 12 00 00  02 04 12 00 
(V) 03 09 06 01  01 02 08 01  FF 0F 02 12  00 00 02 04 
(V) 12 00 03 09  06 01 01 02  08 02 FF 0F  02 12 00 00 
(V) 09 06 00 09  19 09 00 FF  06 00 13 BF  9B 06 00 17 
(V) 03 8D 06 00  00 05 0F 06  00 00 02 0F  
(V) LNG
(D) handleFileUpload file: /firmware.bin
(D) handleFileUpload Open file and write: 0
(D) handleFileUpload Written: 0
(E) Serial buffer overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(E) Serial FIFO overflow
(D) handleFileUpload Ended
(W) Used 23326ms to handle web
(V) HDLC frame:
(V) 7E A0 84 CE  FF 03 13 12  8B E6 E7 00  E0 40 00 01 
(V) 00 00 70 0F  00 2B 50 0F  0C 07 E8 04  06 06 10 07 
(V) 05 FF 80 00  00 02 0E 01  0E 02 04 12  00 28 09 06 
(V) 00 08 19 09  00 FF 0F 02  12 00 00 02  04 12 00 28 
(V) 09 06 00 08  19 09 00 FF  0F 01 12 00  00 02 04 12 
(V) 00 01 09 06  00 00 60 01  00 FF 0F 02  12 00 00 02 
(V) 04 12 00 03  09 06 01 00  01 07 00 FF  0F 02 12 00 
(V) 00 02 04 12  00 03 09 06  01 00 02 07  00 FF 0F 02 
(V) 12 00 00 16  C5 7E 
(V) LLC frame:
(V) E6 E7 00 E0  40 00 01 00  00 70 0F 00  2B 50 0F 0C 
(V) 07 E8 04 06  06 10 07 05  FF 80 00 00  02 0E 01 0E 
(V) 02 04 12 00  28 09 06 00  08 19 09 00  FF 0F 02 12 
(V) 00 00 02 04  12 00 28 09  06 00 08 19  09 00 FF 0F 
(V) 01 12 00 00  02 04 12 00  01 09 06 00  00 60 01 00 
(V) FF 0F 02 12  00 00 02 04  12 00 03 09  06 01 00 01 
(V) 07 00 FF 0F  02 12 00 00  02 04 12 00  03 09 06 01 
(V) 00 02 07 00  FF 0F 02 12  00 00 
(V) GBT frame:
(V) E0 40 00 01  00 00 70 0F  00 2B 50 0F  0C 07 E8 04 
(V) 06 06 10 07  05 FF 80 00  00 02 0E 01  0E 02 04 12 
(V) 00 28 09 06  00 08 19 09  00 FF 0F 02  12 00 00 02 
(V) 04 12 00 28  09 06 00 08  19 09 00 FF  0F 01 12 00 
(V) 00 02 04 12  00 01 09 06  00 00 60 01  00 FF 0F 02 
(V) 12 00 00 02  04 12 00 03  09 06 01 00  01 07 00 FF 
(V) 0F 02 12 00  00 02 04 12  00 03 09 06  01 00 02 07 
(V) 00 FF 0F 02  12 00 00 
(W) loop() used 23486ms
(I) Successfully connected to MQTT
(I)   Subscribing to [homeassistant/status]
(I) Received online status from HA, resetting sensor status
(V) HDLC frame:
(V) 7E A0 7D CE  FF 03 13 D0  45 E0 40 00  02 00 00 6C 
(V) 02 04 12 00  03 09 06 01  01 01 08 00  FF 0F 02 12 
(V) 00 00 02 04  12 00 03 09  06 01 01 02  08 00 FF 0F 
(V) 02 12 00 00  02 04 12 00  03 09 06 01  01 05 08 00 
(V) FF 0F 02 12  00 00 02 04  12 00 03 09  06 01 01 06 
(V) 08 00 FF 0F  02 12 00 00  02 04 12 00  03 09 06 01 
(V) 01 07 08 00  FF 0F 02 12  00 00 02 04  12 00 03 09 
(V) 06 01 01 08  08 00 FF 0F  02 12 00 00  B7 43 06 
(W) Boundry flag missing
(W) Used 1977ms to read HAN port (false)
(W) loop() used 3134ms
(W) No HAN data received last 30s, single blink
(E) Ended up in default case while unwrapping...(tag 06)
(W) Unknown data received
(V)   payload:
(V) 06 00 2A C3  78 06 00 00  07 1E 06 00  00 B7 43 06 
(V) 00 00 00 09  06 00 00 0F  05 06 00 0D  28 28 12 00 
(V) 0F 12 00 5B  12 00 3C 1D  10 7E 
(D) Sending translation file for language: en
(D) Language file /translations-en.json
(I) Rebooting

EDIT: And that's with external USB power and not connected to the meter:

(W) No HAN data received last 30s, single blink
(D) handleFileUpload file: /firmware.bin
(D) handleFileUpload Open file and write: 0
(D) handleFileUpload Written: 0
(D) handleFileUpload Ended
(W) Used 22494ms to handle web
(W) loop() used 22680ms
(I) Successfully connected to MQTT
(I)   Subscribing to [homeassistant/status]
(I) Received online status from HA, resetting sensor status
(D) Sending translation file for language: en
(D) Language file /translations-en.json
(I) Rebooting

[ArnieO]

Potentially very useful - thank you !!

[avandorp]

OTA from 2.3.1 to 2.3.3 worked.
Issue with active export keeping its last non-zero value is fixed with 2.3.3.
Thanks!

[avandorp]

While OTA from 2.3.1 to 2.3.3 worked I can't get it to work from 2.3.3 to 2.3.4.
EDIT: And suddenly it did work. A timing issue?

[ArnieO]

We think there could be a timeout issue deep down in the code somewhere, as this is something some users experience from time to time. Potentially not at a level where we are able to do something in our code. Workaround is to download file and upload locally.

[Buruglen]

Country: Sweden
Meter: Aidon 6442S
Hardware: Pow-P1

Updated from 766849d to db90dbf.

Went smoothly.

Energy use last 31 days (kWh) suddenly show export value for 28.03. See below. I do not export, and 766849d shows correct values.

Before upgrade:

After upgrade:

[Buruglen]

Soon after the upgrade to db90dbf I realized that 2.3.2 had been released and decided to do an OTA upgrade.

That upgrade never ended (using Firefox on Windows) and I had to refresh the window. The upgrade had in fact completed successfully.

The values shown above from db90dbf are still here after the 2.3.1 upgrade.

[ArnieO]

2.3.2 had been released

That is a typo, you meant 2.3.1 ;-)

[Buruglen]

Returned to Sweden after a week in Norway. The Pow-P1 still ran 2.3.1.
The graphs under Energy use last 31 days (kWh) showed all 0, except for the last two days.
Updated OTA to 2.3.4. Same issue.
Reporting to Home Assistant (HA) via MQTT seems to have worked all the time.
Very strange. There has not been anyone in the house and power has been stable all the time.
Any good explanations?
Snapshot below is from today (two days later).

[Buruglen]

Screen dump from HA for 2024-04-12 (the last day with 0).

[ArnieO]

No - sorry. Can't see any obvious explanation for that. It seems that old data for some reason got deleted on the device early on the 13th.

[Buruglen]

Yes, it is really very strange. The Tariff peaks are not confused by this. They show hours/days in the "zero days"-range. It seems that this information is kept separately, probably only updated when the just ended hour mandates a change.

[ArnieO]

Yes, I believe the Tariff peaks are stored separately.

[yoabau]

Device information
Chip: esp32s2 (160MHz)
Device: [Pow-K+]
Meter
Manufacturer: Kamstrup
Model: 6841121BN243101040

Just update firmware from v2.2.28 to v.2.3.1 on my POW-K+. It was setup with static IP address, but turned back to DHCP after the firmware successfully updated. I was able to access the web server by forcing POW-K+ to AP mode. To my surprise, the network settings was unchanged (static), but I managed to get it back by first turning on DHCP and then back to static.
However, POW-K+ was only stable during 9 hours (from 7AM to 4PM) as it published payloads to MQTT broker during this period. When I was back home I saw red led blinking but I managed to get it back alive by un/plugging it.

[gskjold]

I probably need an account with ubidots to test this... Will put it on my list

[ArnieO]

If I remember correctly: I think @yoabau wrote in an email that we could "borrow" his account for testing.

[yoabau]

Sure I can send you PM with credentials, but you can also sign up for a 30-days trial.

[yoabau]

Hi there.

I was just wondering if @gskjold made a test with POW-K+ publishing data to Ubidots without crashing? or should I create a new post in "issues" to get updates on this problem?

[ArnieO]

Hi there.

I was just wondering if @gskjold made a test with POW-K+ publishing data to Ubidots without crashing? or should I create a new post in "issues" to get updates on this problem?

Salut! Yes, probably a good idea to lift this by creating a new "issue".

[gskjold]

We are not beta testing anymore. Any problems should be posted in issues.