Thanks for fixing whatever was causing issues. I can finally use my two purifiers again on HA. Please maintain for us 
Iâve attempted a rewrite of the existing integration. It doesnât really solve the connectivity problem, but it brings all sorts of other new features to the surface, like:
- configuration via UI (only)
- new fan model with speed settings
- attributes as sensors
- diagnostic sensors for filters and water level
- control switches and selects for child lock, display backlight, light brightness, humidity target and humidifier function
- dynamic icons to show fan state, humidification function or warn for upcoming filter change or low water level
- original Philips icon set is included as well
- new models AC2936, AC3039, AC5659
9 Likes
Thank you so much for the work you invested into this integration!
Thank you for this, I have updated to your version.
Adding my purifier via the UI worked perfectly!
Strange. Here my AC3829/10 used to work but since 3 days I cannot get it connected to HA anymore⌠also tried the integration from @kongo09 but it stops with an âunknown errorâ.:
Logger: coap
Source: /usr/local/lib/python3.9/site-packages/aiocoap/transports/udp6.py:525
First occurred: 07:22:37 (2 occurrences)
Last logged: 07:42:42
Connection loss was not expected.
Please provide more logs as described on the repository page.
There is a new beta: v0.9.3-beta.1 which brings an attempt at reconnections (thank you @mhetzi ) Please try if that helps you and report back.
This error is not a fatal one, it appeared with me after Philips changed the FW. With the updated aioairctrl it should work. Sometimes a power cycle may be required (as @kongo09 also describes on GitHub)
i have hundreds of these errors per day by the way. Best to add an ignore in your logger config.
Thank you 
,
I just upgrade and it worked at first try (I am without hacs - just place the git content in custom_contents folder).
Both my purifier (AC3033 and AC2889) are working without issues.
Ps. could you confirm that these values are hours?
Yes, these are hours counting down.
By the way: you seem to miss the Philips icons. Did you refresh your browser? And I notice something else from your screen shot: the sensors donât have units. It should look like this:
Still no success. Now the attempt to connect lasts longer, but still with the same result (unknown error). Mobile app (Clean Home+) is working fine.
My purifier is on a different vlan than my home assistant. Wasnât a problem beforeâŚ
You are probably victim of the connection problem of Philips. Not much anyone here can do.
The Clean Home+ app is not using the local CoAP connection but the cloud. Unfortunately, the API is not public. Philips has invested a bit into this and now Alexa and Google Home are integrated. IFTTT uses it as well. Iâve written to Philips and asked for access but they havenât even replied.
Mine are also on a different vlan, that should work. To avoid further firmware upgrades following the last f*-up by Philips i blocked all internet communication.
Can you try to power cycle ha? It might be dependencies such as aioairctrl have not been pulled. Also try to turn on the purifier whilst doing this. I had some odd behavior in the past where a reboot of ha would render the purifier unavailable unless it was on during reboot.
I should have done that long time ago⌠itâs really too bad. But that also means, the Clean Home+ app is not working anymore for you, right? It doesnât have a local mode?
Sorry for the late reply, yes I refresh the browser and clear cache but the âumâ are still missingâŚ
Could this issue depends by the fact that I am not using hacs?
As said I copied the philips_airpurifier_coap folder on your git into the custom_components and I got 1 device and 9 entities for the 2889 and 10 for the 3033.
I tried with safari and firefox and both shown the same data
This is what I got on the 3033, the same is on the 2889.
EDIT: I removed both the purifier and add them again but the integration shown the same data (no without the um) 
I thought it was something related to css but in the html there is no trace of the unit of measurement
I notice that sometimes after restart HA or reboot the pi, the integration takes long time to start
This is the log from HA and from the container
I am getting this error but I am not sure that is from the purifier integration
2022-04-10 11:55:33 ERROR (MainThread) [pyatv.protocols.companion.connection] failed to handle frame
Traceback (most recent call last):
File "/usr/local/lib/python3.9/site-packages/pyatv/protocols/companion/connection.py", line 142, in data_received
payload = self._chacha.decrypt(payload, aad=header)
File "/usr/local/lib/python3.9/site-packages/pyatv/support/chacha20.py", line 42, in decrypt
decrypted = self._enc_in.decrypt(nounce, data, aad)
File "/usr/local/lib/python3.9/site-packages/cryptography/hazmat/primitives/ciphers/aead.py", line 62, in decrypt
return aead._decrypt(backend, self, nonce, data, associated_data, 16)
File "/usr/local/lib/python3.9/site-packages/cryptography/hazmat/backends/openssl/aead.py", line 133, in _decrypt
raise InvalidTag
cryptography.exceptions.InvalidTag
Well, first of all, you now see the custom Philips icons. Thatâs great.
I donât see any relationship at all between using HACS or not and the display of the units. HACS is just a sophisticated downloader. In the code, the units are part of the attributes. It looks something like this:
self._attr_unit_of_measurement = self._description.get(ATTR_UNIT)
where self is the respective sensor and ATTR_UNIT points to the actual unit. The only custom units that the integration provides, are:
LEVEL = "Level"
INDOOR_ALLERGEN_INDEX = "IAI"
AIR_QUALITY_INDEX = "AQI"
All other units are built-in, as documented here:
So I would suspect that this is a HA issue. Maybe a language setting of the frontend? I donât know. To figure out, if this is just a frontend error, please go to the Developer Tools and lookup one of your filters. The attributes should show the units as unit_of_measurement attribute:
If it shows there, but not in your frontend, then you should file this as a HA bug.
1 Like
I solved thank you!
It was an issue on the HA, I moved from the latest version to the stable and everything is ok now
this is the yaml in my composer:
homeassistant:
container_name: home-assistant
image: homeassistant/raspberrypi4-64-homeassistant:stable # before it was :latest
restart: unless-stopped
environment:
- TZ=Europe/Rome
network_mode: host
volumes:
- ./home-assistant:/config
- /etc/localtime:/etc/localtime:ro
- /etc/letsencrypt:/etc/letsencrypt:ro
- /var/run/docker.sock:/var/run/docker.sock
healthcheck:
test: ["CMD", "curl", "-f", "http://127.0.0.1:8123"]
interval: 30s
timeout: 10s
retries: 6
the AC2889 now 
1 Like
Maybe I am blind or something, but I honestly cannot find the child lock switch anywhere. I have an AC3829.
Some icons are missing as well, what seems to be odd too.
Otherwise it looks shiny and pretty. Amazing job!
Thank you!
Thatâs interesting. According to the Philips manual, you should indeed have that switch.
The way the integration is deciding on whether a switch should be offered is by looking at the data received from the device. There is field called cl (for child lock) and if that is present, the switch is setup.
So can you enable debugging and post a log of the device response?
1 Like