was thinking about something like that. But maybe there is some device that can do the same as the arduino with relay that you can mount on standard DIN mount that is certified because I need to recertifie the electricity in the house and I don’t think they are going to like arduino’s (living in Belgium so would be the same option for me 230V relays)
but then you don’t have a solution for the normall in wall switches.
I wonder if there’s a complete solution where you can switch from in wall switch(wired) and from mqtt or something else conform EU regulations
Fellow belgian then 
So far I haven’t found a complete solution for that, at least not where the wall switches have an actual wired connection to control the lights… I should have a meeting with our electrician soon where I can ask whether Arduino’s are certified or not, but as far as I understand the regulations, they check if the system functions as designed, and whether any 230V lines are directly exposed. There are also DIN mounts for Arduino’s, and the relays should be doable by using standard low-power relays (like you would have with a teleruptor installation), not sure yet if they actually work on 5V…
What you could do (which I might do as well) is first wire the normal light switches to the lights directly (after coming to the central location), and later on swap the system to use the Arduino’s, then you don’t need to recertifie I think.
Take a look at this guys work, I like the solution he has, central relays with simple switches. He sells some of the components needed too. No idea if its EU compliant. The ability to dynamically assign buttons to different tasks is pretty awesome.
My only concern with this solution is if the system goes down you lose local control (zwave, wifi, etc. light switches should still work locally when the main hub is down)
you can look at UniPi an EU based company
https://www.unipi.technology/
They have recently opened up their API, that might probably work with HA
https://www.unipi.technology/news/introducing-evok-2-0-105
They have support for many opensource projects
Personally, I would run CAT 5E cable and not CAT6 cable. The CAT6 cable is much more bulky and a bit more difficult to deal with. You can run gigabit ethernet drops over CAT5 cable, while CAT6 cable can support 10G Ethernet, which is likely overkill for most wallplate drops vs. interconnects to servers or network device trunking.
I’ve done both, and most recently figured I’d “upgrade” to CAT6 cable for some new cable runs, and I now regret that decision (and buying a whole box of cable… and new/different connectors.)
I recommend that you make a thoughtful, deliberate choice here, and not just choose the “latest and greatest” without really thinking of the implications. For example, the reduction in crosstalk in CAT6 cable comes from structures in the cable that manage the physical separation of the 4 pairs of wire from each other. If you manage to crimp the cable while pulling it through holes drilled in wall studs, you might render that mechanism moot if you “kink” the cable too badly or in too many places. Something to think about.
You need to be careful when running any low voltage wire. Sure, cat6 is a little more delicate than cat5e, but that doesn’t mean you can rip cat5e through kinks and around bends. Cat6 is marginally more expensive than cat5e. Limiting yourself to gigabit when that might be all you need now when you could easily upgrade to 10G to marginally future-proof something that isn’t easily replaced is silly.
Having pulled a few thousand feet of wire while wiring a couple of the houses I’ve lived in over time, the mechanical practicalities really do matter. The size of the CAT6 cable is noticeably larger than CAT5e and this means more or larger holes in wall studs. It also poses a challenge in cable management and bulk termination in a wire closet or whatever location(s) you pull cable back to. If you really do end up commited to CAT6 cable, then you should seriously consider more than one distribution point to offset the additional labor and termination complexity.
In my builds, I pull all the CAT5 runs to 110 punch blocks for higher density termination. I can successfully cross-connect gigabit ethernet on the 110 blocks; maybe not actually “legal” but it works.
I would also consider how many devices need in excess of a gigabit/sec of bandwidth? Sure, you might have some aggregation points where more is needed. But that’s probably media servers and similar, and you’d need to consider what residential media server can sustain more than 120 Mbytes/second? And if that really is a bottleneck, you could do channel bonding to get get 1,2,3, or 4 Gbit/sec to a high bandwidth traffic source/sink.
Moving to 10Gb/s ethernet is still beyond commodity pricing these days, much less 40G (and now, 25G which is scaled-up 10G) Ethernet MACs. I use this stuff all the time at $WORK in our data center builds where we uplink a whole rack of servers with 4x40G, moving to 2x100G in some cases. So yeah, this stuff is out there, but you need some moderately expensive hardware on the other end of the wire to source/sink these more intense bandwidth.
Go find some CAT5e cable and CAT6 cable and imagine what bundles of each feels like before you make your decision.
What’s very common in Europe is to use knx.
Switch are not physically connected to bulb or device, but are all connected to a central control that then switch the light or the associated device.
There is also some component to link knx to hass.
I’m in the same boat – planning and spec’ing things for a new house. I’m having a hard time deciding what switches to use. I certainly wouldn’t want to replace 30+ switches a few years from now.
The building would like to install Leviton devices, which seem to be available in Z-Wave, WiFi, and possibly ZigBee though they call it Lumina RF and it’s not clear that it is completely compatible.
I would really stay away from wireless for a permanent installation, some wire going to a central location let you be much more flexible. You can use any standard switch, so when you want to change in 10 years, you wont have to find some old Z-Wave stock ^^
I’ve been thinking of a relay board + Pi for mine, though I haven’t heard anyone use anything similar in Hass - but they’re affordable & scalable:
unipi is more suited, has pi & relays/digital input all included in DIN format with opensource software
Unipi looks really interesting going to examine this. It has a pi built-in so you can program what you want
I think that’s good solution for development or small project, to control a whole house, I would look for something more reliable than a raspberry. It’s unlikely it will run for 10/20 years without problem/sd card corruption.
And one bug per year (with a few hours downtime) is more than enough to make the rest of your house hate your installation ^^
Why dont go for Xiaomi Switches? Then you can controll them via HA and via the physical button.
And some WiFi smart plug for window lamps, that you can controll via HA and the button on the smart plug.
Motion sensors/ door senors you can use Xiaomi, sure you need to change battery sometimes but thats not a super big problem.
Easy and not to expensive soulotion.
their new product line does not include raspi but another processing power basically you could use nodered & hass control everything with wired connections.
You still need a server where hass needs to run, so if your server fails…
https://www.unipi.technology/hardware_documentation/technical-intro-to-neuron-products-41
That’s exactly the point, you need something that run without hass too. Like Knx where you simply define “this switch -> this relay”.
So if hass fail, you can still use the switch to control your house.
And the rest of the time (hass not failing) you can control your relay/pool the switch and relay status, through hass.
Like @Daniel_Gronlund says, the XiaoMi switch are well conceive for that purpose, because they let you turn the light on/off even if the server/connection is lost.