Maybe stupid question… how do I send a DM here…?
Just click on my name and “message” 
My bad. Changed it.
It worked
There’s this automatic measures the forum takes for spamming. You needed level 1 to be able to PM people or so.
Changed it to 0 
1 Like
This matter is all in all a little bit tricky. Had the all off experience myself with the very same switch.
Back in the days, I’ve thrown this PR into the field, but there’s apparently other side effects which make an amendment not as easy as thought.
Regardless, there’s a workaround to cease the negative side effects (it doesn’t cater for the root cause). You can set the group being responsible for switching everything off to deleted inside deconz’ DB. For that, find the DB named zll.db on your file system. Make sure deconz is NOT running and then open the DB on the command line with sqlite3 (assuming here you go with Linux). Once connected to the DB, issue the following command:
update groups set state="deleted" where gid="0xFFF0";
and then close the DB again with .exit. Afterwards, you can start deconz again and there shouldn’t be any further trouble.
Hi guys, any further news here?
Any update on this issue? I’ve been experiencing the same issues and I can’t get the workaround mentioned above by @Swoop to work.
We are investigating
Hi guys, any real update here…?
This thread is open since july now. Is there any plan to find a solution for these 4 channel switches or should i replace them with 4 button ones? (if not thats also a fine answer by the way ;-))
Still more than happy to send you a device if it helps fixing this. Thanks!
Hey,
not realy so far as we aren´t had time to test anything in the last weeks. Our main-programmer will take a look at it on the weekend. If there isn´t anything new about it I will come back at your offer.
seems this 4 channel switch is jinxed and wont make it as supported… 
Any recommendation of one which actually works? I could really benefit from one in my setup, so if buying new hardware is a faster solution i’m all open for it.
Sorry @tompijls , we had so much devices to integrate at least that our integration programmers have barely kept up. That’s why some of the equipment has been left behind.
I asked again, sry for that delay.
Hi everybody, sorry it has been a while. After many tests and discussions later the problem is a bit unique and turned out a rabbit hole, this switch is a Sunricher also rebranded by SLC, Namron and ROBB.
Details
With latest deCONZ version the switch and button events are working correctly in the REST-API, it also uses proper unique groups instead broadcasting to all lights which was a problem in the initial post.
The Phoscon App switch editor shows all buttons for configuration generically based on the data in button_maps.json, which is good so far…
However the generated rules for the rows 2-4 point to the wrong sensor. The switch could have been implemented as one sensor covering all buttons, but has instead four sensors one for each row, and here the generic approach breaks.
Although it would be simple to combine all sensors into one with a small code change, this would break existing setups which only use the REST-API but not Phoscon App…
Fix
I have extended the button maps handling to provide a bit of extra information in PR Button maps support specify endpoint for buttons by manup · Pull Request #5901 · dresden-elektronik/deconz-rest-plugin · GitHub and adapted the Phoscon App.
With the next version the button assignments for rows 2-4 need to be re-created in the Phoscon App Switch Editor but should work afterwards.
@manup Replied on the respective DDF PR as well. You’re currently making wrong assumptions on the model Identifier to expose the button row correctly in Phoscon. Also, our beloved binding to group 0xFFF0 plays a role here.
Like said, I got the device and it works perfectly (if you know how to dance around the annoyances of binding tp the ALL group) 