Nexa Wall Switch

[Mirar]

I have lots of things going on on my home automation. But I don't have user input - at least not without using the web interface or waving to motion sensors.

So I wanted to see if the Nexa wall switches could work with Telldus.

It _almost_ works.

It's so close it's really annoying.

A few packets work, sometimes. They come out fine, as (raw packets):

Code: Select all

class:command;protocol:arctech;model:selflearning;house:417958;unit:1;group:0;method:turnon;
class:command;protocol:arctech;model:selflearning;house:417958;unit:1;group:0;method:turnoff;

Isn't that lovely? The problem is, they don't always show up like this. They also show up as (on):

Code: Select all

      1 class:command;protocol:arctech;model:selflearning;house:14775410;unit:10;group:0;method:turnon;
      1 class:command;protocol:arctech;model:selflearning;house:50436137;unit:5;group:1;method:turnoff;
      1 class:command;protocol:arctech;model:selflearning;house:8414730;unit:10;group:0;method:turnon;
      1 class:command;protocol:everflourish;model:selflearning;house:3274;unit:2;method:turnoff;
      1 class:command;protocol:sartano;model:codeswitch;code:1101100110;method:turnon;
      6 class:command;protocol:arctech;model:selflearning;house:208979;unit:9;group:0;method:turnoff;
      6 class:command;protocol:sartano;model:codeswitch;code:1110110011;method:turnoff;
     12 class:command;protocol:arctech;model:selflearning;house:417958;unit:1;group:0;method:turnon;
     12 class:command;protocol:sartano;model:codeswitch;code:1111011001;method:turnon;
     13 class:command;protocol:everflourish;model:selflearning;house:9738;unit:2;method:turnoff;

and (off) (notice the overlap between on and off...)

Code: Select all

      1 class:command;protocol:arctech;model:selflearning;house:104489;unit:1;group:1;method:turnoff;
      1 class:command;protocol:arctech;model:selflearning;house:25335818;unit:9;group:0;method:turnon;
      1 class:command;protocol:arctech;model:selflearning;house:26744842;unit:9;group:0;method:turnon;
      1 class:command;protocol:arctech;model:selflearning;house:33658921;unit:1;group:1;method:turnoff;
      2 class:command;protocol:arctech;model:selflearning;house:208979;unit:1;group:0;method:turnoff;
      2 class:command;protocol:everflourish;model:selflearning;house:6154;unit:2;method:turnoff;
      2 class:command;protocol:sartano;model:codeswitch;code:1111100110;method:turnon;
      2 class:command;protocol:sartano;model:codeswitch;code:1111110011;method:turnoff;
      8 class:command;protocol:arctech;model:selflearning;house:417958;unit:1;group:0;method:turnoff;
      8 class:command;protocol:everflourish;model:selflearning;house:9738;unit:2;method:turnoff;
      8 class:command;protocol:sartano;model:codeswitch;code:1111111001;method:turnon;

The number is the number of packets received that looks like this.

It seems like something needs to be better at guessing where the packet starts. 104489, 208979 is shifted 2 and 1 steps from the correct house number. Everflourish and sartano shows up if it guessed on the wrong protocol (everflourish is independent on the button "off" and sartano is "on").

Has anyone else looked into this? Did the Telldus people look into this?
I'm cloning the repository now to see if I can prod around somewhere to help the situation...

[Mirar]

If anyone have a tip of a button that _does_ work, I'd like to know.

I ordered some Flic, but that'll be a completely different adventure...

[Mirar]

Ok, a little more investigation in what the telldus-duo firmware emits as raw data:

Code: Select all

protocol:arctech;model:selflearning;data:0x1982990;
protocol:everflourish;data:0x982990;

on the "on" event. "off" gives

Code: Select all

protocol:arctech;model:selflearning;data:0x1982980;
protocol:everflourish;data:0x982980;

These are different packets - sometimes (usually?) the first packet is broken.

A guess is that this isn't "everflourish" data, but the transmitted repeat packet comes so close that the first bit(s) are dropped and the first one is quite often broken... so telldus receives the first packet but misses it, and the following repeats are broken due to ... timing issues?

[Mirar]

Ok, running all "everflourish" as nexa "selflearning" but adding 0x1000000 makes the switch show up >99% or the buttonpushes.

I'm going to settle for that. If anyone wants this _exceptionally_ kludgy patch, tell me.

Correct solution would be to fix the firmware so it doesn't miss the beginning of the repeat of the packet.