I started reversing the way IKEA implements the SEN54 sensor in VINDSTYRKA, as they seem to be doing some funky processing on raw and undocumented values…
The way ESPHome implements this sensor yields different (inaccurate!) results for temperature/humidity when compared to the values reported by VINDSTYRKA (as reported here: IKEA Vindstyrka & ESPHome Integration - #4 by Laess3r).
I hope to come up with some sort of compensation value/slope that can be used to correct the sensor output.
Docs:
I am a bit confused. Where have i to put this code?
I am a bloody newby, so perhaps i ask something ridiculous
Greatings
Christoph
It’s an automation.
if you have zha toolkit installed GitHub - mdeweerd/zha-toolkit: 🧰 Zigbee Home Assistant Toolkit - service for "rare" Zigbee operations using ZHA on Home Assistant then you can go to automations and create it.
Or you can add it to the appropriate yaml file.
Has anyone had any luck contacting IKEA about technical specs of this device beyond those in the ‘assembly’ document? Specifically, I am trying to figure out what type of PM2.5 sensor is being used–I am most familiar with Plantower laser sensors but this one appears to be different.
As mentioned before, IKEA VINDSTYRKA contains a Sensirion SEN54 sensor module. There is plenty of documentation available on the website of the manufacturer: Smart sensor solutions.
Great post, thank you.
I’m new to IKEA smart home products, and wondering if I can add a Vindstyrka to my HA ecosystem by itself, or must I also add their Dirigera hub?
Thanks again.
You can add it as a Zigbee device either with ZHA or Zigbee2MQTT. You will obtain Temp, Humidity and PM2.5 data from the unit into HA. However, as this thread defines, there is an issue with obtaining the VOC index in ZHA which is an issue that I have. Apparently it works with Z2MQTT but I have not gone there. I am a ZHA user.
I am hoping for a resolution in the VOC index as my intention is to use my IKEA unit in a 3D Print Enclosure to control carbon filtration blowers installed in the enclosure.
This is my automation to poll for updates every 15 seconds from multiple sensors, and does not require modifying quirks or patching HASS. Hope this is useful!
Edited to add allow_create: true and removed duplicate attr_read.
Edit: Turns out manf is critical when quirks aren’t set.
alias: "Climate: Fetch AQI data"
mode: single
description: poll for AQI data
trigger:
- platform: time_pattern
seconds: "/15"
condition: []
action:
- parallel:
- repeat:
for_each: # list your sensors here
- lounge_aqi
- office_aqi
- balcony_aqi
- bathroom_aqi
sequence:
- service: zha_toolkit.attr_read
data:
ieee: sensor.{{ repeat.item }}_particulate_matter
endpoint: 1
manf: 4476
cluster: 1066
attribute: 0
state_id: sensor.{{ repeat.item }}_particulate_matter
allow_create: true
- service: zha_toolkit.attr_read
data:
ieee: sensor.{{ repeat.item }}_particulate_matter # Intentional, because the VOC index parameter may not exist
endpoint: 1
manf: 4476
cluster: 64638
attribute: 0
state_id: sensor.{{ repeat.item }}_voc_index
allow_create: true
I hooked this up through Deconz on Home assistant, and I’m definitely confused about what item is what on the output:
The PPB only showed up this morning 
The ppb is definitely interesting. I don’t get a ppb numer via Zigbee2mqtt. The VOC Index is on a different scale, unfortunately not an absolute number in ppb so I wonder what that is all about…
Where is that button? On the device’s diagnostic panel in ZHA I only have an Identify button.
after updating to hassio 2023.10, the tvoc sensor created in this way:
sensors:
keittio_ilmanlaatu_tvoc:
friendly_name: “Keittiö ilmanlaatu TVOC”
icon_template: mdi:weather-dust
value_template: “”
and the automation suggested by @chendo give me back an error message
Any idea?
I am using ZHA with conbee 2
If you want (long term) statistics for the sensor, you have to set the device class. It will also show the history not as state changes but as a graph. Also, set unit to empty and the initial value to None so that it doesn’t try to interpret the empty string as a number on start and throw a warning:
vindstyrka_tvoc_index:
unique_id: vindstyrka_tvoc_index
friendly_name: Vindstyrka TVOC Index
device_class: aqi
unit_of_measurement: ''
icon_template: mdi:weather-dust
value_template: '{{ None }}'
Thanks for this.
This solves my ‘entities not available’ when doing a quick reload.
Can I suggest that @jonixx updates their post with these changes to the template sensor?
I have a more general question. How to interpret these values: In ther morning they are normal because the window was open during the night. Then the values increase sharply until I open the window. What can cause these high values?
what you see is proprietary Sensirion’s VOC Index, basically it say’s when it’s rising that your VOC load increases while it decreases with an open window. these are no absolute values, but relative values. a value of 100 indicates that the VOC load (or „smell“) of the room is fine. values below indicate fresh air, open window (decreasing „smell“). the values are relative on a flexible scale and are supposed to mimic how our nose works/smells.
basically sensirion recommends to clean the air with a air cleaner/purifier with minimum fan power when the index is between 150 and 250, with medium fan power between 250 and 400 and maximum power above 400. or in other words: open the window when the index is above 400. values up to 300 like your graph shows should be absolutly fine (and only indicate that the room is slightly more „smelly“ than fresh air).
this is all explained in more detail in https://sensirion.com/media/documents/02232963/6294E043/Info_Note_VOC_Index.pdf
OK. Thank you for your quick reply. I understand the values as such but I do not understand the source of the VOC load. Does it only come from my breathing? There are no other sources in the room but myself.
it’s a VOC sensor 
sources are anthropomorphic (sweating, breathing, farting, burping) but also evaporate from food or cooking, from solvents in cosmetics or disolved paint, after-shave and so on. more or less things that smell.
For those who don’t speak or read German, this is the English equivalent: