IKEA FLOALT panel not reacting any more (but still "seen" by HA)

Usually, my IKEA FLOALT panel is working as expectet. But every few days or weeks, it does not react any more and I have no idea why (no HA update between working and not working any more).
On the ZHA integrations page, the “last seen” updates every few minutes
image
and the visualization map looks OK (as far as I understand it):

On the HA log page, I don’t see an error. But in the log file I see entries like:

2021-04-12 07:57:33 DEBUG (MainThread) [homeassistant.core] Bus:Handling <Event call_service[L]: domain=light, service=turn_on, service_data=entity_id=light.ikea_panel_arbeitszimmer_level_light_color_on_off>
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.zigbee.application] Sending Zigbee request with tsn 240 under 241 request id, data: b'01f001'
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.api] Command Command.aps_data_request (18, 241, 0, <DeconzAddressEndpoint address_mode=2 address=0x5EB7 endpoint=1>, 260, 6, 1, b'\x01\xf0\x01', 2, 0)
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.api] APS data request response: [2, <DeviceState.APSDE_DATA_REQUEST_SLOTS_AVAILABLE|2: 34>, 241]
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.api] Device state changed response: [<DeviceState.128|APSDE_DATA_REQUEST_SLOTS_AVAILABLE|APSDE_DATA_CONFIRM|2: 166>, 0]
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.api] Command Command.aps_data_confirm (0,)
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.api] APS data confirm response for request with id 241: 00
2021-04-12 07:57:33 DEBUG (MainThread) [zigpy_deconz.api] Request id: 0xf1 'aps_data_confirm' for <DeconzAddressEndpoint address_mode=ADDRESS_MODE.NWK address=0x5eb7 endpoint=1>, status: 0x00
2021-04-12 07:57:38 DEBUG (MainThread) [homeassistant.components.zha.core.channels.base] [0x5EB7:1:0x0006]: command failed: 'on' args: '()' kwargs '{}' exception: ''
2021-04-12 07:57:38 DEBUG (MainThread) [homeassistant.components.zha.entity] light.ikea_panel_arbeitszimmer_level_light_color_on_off: turned on: {'on_off': TimeoutError()}

I use Conbee Stick II on a Intel NUC and Home Assistant 2021.3.2 (HA OS). All other ZHA devices (32 total) work without that issue, some of them are further away from the Conbee Stick.

Here is the full HA log:
https://raw.githubusercontent.com/der-optimist/test/master/log_zha_ikea_01.txt

Did you ever find out how to fix it?

I have a IKEA Floalt panel within a “kitchen” group in ZHA and every few days the other 3 lights in the group will work fine (off/on/dim) but the Floalt panel will not react to commands. It usually works again after a few tries but still it’s not a nice experience…

No, I did not solve the “Floalt - ZHA” problem, at least not the “debug way” :slight_smile:
I disassembled the LED driver, removed the ZigBee Board and replaced it by an ESP32 with esphome.

I checked the behavior of the IKEA Board with a logic analyzer. Then I removed it, used the 3.3V power supply for the ESP32 and used the cwww of esphome to drive the FLOALT. It works very well, and I can dim the panel to a very low value now.

1 Like

Hi,
same issue here, sometimes Floalt panel not react any more.
All the other Zigbee stuff nearby work fine.
How did you connect ESP32 to Floalt?
Thanks,
-f

I soldered the Output of the ESP to the pads wehere the IKEA controller was before.

After a while, I had trouble with the 3.3V supply of the IKEA board. So I used the higher DC level of the board (iirc 30V) and connected it to a step-down converter that deliveres the 3.3V for the ESP now:

I think the interesting part is inside the LED Driver :innocent:
Is the replaced Zigbee module soldered on the board:

or is it a standalone section:

in your Floalt version?
I ask it because I’m not a soldering master…
-f

I’m not a soldering master either :grinning: This one is the version I have and I marked the module I removed:


The pinout is marked on the board. As I destroyed some soldering pads when removing the module, I searched “the other end” of them and soldered my wire there.

Cool, thanks.
Could you please share ESP32 config as well?

Sure:

esphome:
  name: esp-esp32-nodemcu-1
  platform: ESP32
  board: nodemcu-32s
  comment: Ikea Panel

wifi:
  ssid: "abc"
  password: "xyz"
  use_address: a.b.c

  # Enable fallback hotspot (captive portal) in case wifi connection fails
  ap:
    ssid: "ESP32-NodeMCU-1"
    password: "xyz"

captive_portal:

# Enable logging
logger:

# Enable Home Assistant API
api:

ota:

web_server:
  port: 80

output:
  - platform: ledc
    pin: GPIO18
    id: output_component_cw
    frequency: 200Hz
  - platform: ledc
    pin: GPIO19
    id: output_component_ww
    frequency: 200Hz

light:
  - platform: cwww
    name: "Ikea Panel esphome"
    cold_white: output_component_cw
    warm_white: output_component_ww
    cold_white_color_temperature: 4000 K
    warm_white_color_temperature: 2200 K
    constant_brightness: true
    default_transition_length: 1s

sensor:
  - platform: wifi_signal
    name: "WLAN Signal ESP ESP32 NodeMCU 1"
    update_interval: 600s

Many thanks.
I can see only 4 cables in/out to LED Driver so VCC, GND, PWM1 and PWM2, I suppose. Is it correct?

Yes. The red one is on the 30V level which is reduced to 3.3 by the external step-down-converter

Same mistake, I destroyed some soldering paths too (PWM2 for example) :frowning:
Fortunately, I only use warm light so PWM1 is enough.
ESP32 is powered with an external tiny USB charger (following your tip to avoid powering it from Ikea board).
With ESP32 is a completely different story so, thanks again for the hint!


Each PWM pad is directly connected to pin of a MOSFET, iirc.


(you can check with a continuity tester)
I think I used that place for soldering my PWM wires instead of the pads.

I’m not so sure about your power supply - How do you assure you have the same ground potential?
If ground potential is not the same, this may lead to malfunction or destroyed parts.

The USB charger is connected to the same Floalt hub hidden in the light panel - this one:

Is this an unsafe scenario in your opinion?

This is the AC connection box, right?
Sounds unsafe for me, yes. The USB charger creates its ground potential “somehow” and the FLOALT power supply creates it “somehow”. As your setup is working, it looks like they are doing it pretty much the same. But there may be little too high or low levels on the PWM pins now - which may cause problems in the future.
I suggest using ground of the FLOALT + 30V DC from the FLOAT and use a DC-DC converter

Chad move right there