I agree BTW with the other commenters. The more robust solution is to use a separate device to do the control (determine valve settings based on measured and desired temperature) and use HA to do the automation (determine and communicate which temperature is desired when)
2 Likes
Versatile Thermostat in HACS works well with my underfloor heating system. I use it to command a set of switches which operate the zone valves and relay box to start the pump.
I added zigbee temp sensors in each space and sonoff 4chan switches to the existing system.
The boiler knows its temperature at all times. It knows this because it knows the temperature it isn’t. By subtracting the current temperature from the desired temperature, or vice versa—whichever is greater—it obtains a difference or deviation. The control system uses this deviation to generate corrective commands to adjust the boiler’s fuel and air inputs, driving the temperature from where it is to where it isn’t. Arriving at a temperature that it wasn’t, it now is. Consequently, the temperature where it is now is the temperature that it wasn’t, and it follows that the temperature that it was is now the temperature that it isn’t.
In the event that the temperature where it is doesn’t match the temperature that it wasn’t, the system has acquired a variation. The variation is the difference between where the boiler temperature is and where it wasn’t. If this variation is considered significant, it too may be corrected by the PID control system. However, the boiler must also know where its temperature was.
The boiler control computer scenario works as follows: Because a variation has modified some of the information the boiler has obtained, it may not be entirely sure what its temperature is. However, it is sure what its temperature isn’t, within reason, and it knows where its temperature was. It now subtracts where its temperature should be from where it wasn’t, or vice versa. By differentiating this from the algebraic sum of where the temperature shouldn’t be and where it was, it is able to obtain the deviation and its variation, which is called error.
I know a thing or two about control systems. But you completely lost me with this description 
. Anyhow, I wasn’t the one asking the question, I was just trying to suggest to the OP how I thought their question would be more accurately phrased.
Lol this is “The Missile Knows Where It Is…” meme just with boilers, I like using the meme to describe pids because its overall an advanced pid system.
Its probably only partially true when it comes to boilers.
But anything that has to do with motion over time usually has some pid thing like drones need pids just for the motors. (heat is motion btw).
Yes you do. HA can talk to Loxone.
1 Like
Hah. Now I get it. I just didn’t know this meme
I have an electrical floor heating system that is now controlled by NS Panel flashed with ESP Home. Very efficient
1 Like
Thank you Kolia,
I will try to remember that you’ve had great success with NS Panel, in floor electrical, and HA.
Your input is valued!
thanks for the info.
they are ~1m above ground - as with all typical underfloor heating setups (at least in europe)
maybe that really makes the most sense.
would be good if there is an “official” recommendation
At least my setup doesn’t have a boiler.
It’s using a heat exchanger - so no need to 
If you would have read all the posts, you would have noticed that I also mentioned that it of course also depends on the outside temperature.
You can still “learn” the heating curve when accounting for the outside temperature.
Loxone can do it - but unfortunately it’s a proprietary closed-source solution.
Would be nice to have an open source solution available.
the “I” in my example was relating to the “new user”.
I myself can of course also continue to use the Loxone solution.
But I would like to switch to an open, non-propriatary solution instead.
Seems like that’s not existing.
So that’s what this WTH is about: no simple solution to get existing underfloor heatings into Home Assistant (at least not without buying additional/dedicated hardware)
What I have discovered in decades of searching is “if it was easy, somebody would have done it already, and written about it”. I first learned about control systems many decades ago. There are simple ones like “bang-bang” that work well enough for many things, but in-floor heating is not one of those situations.
This is a good article that goes into some of the issues: Optimal PI control parameters for accurate underfloor heating temperature control
This is a reasonably easy to understand description of the type of controller needed: PID Control made easy | Eurotherm Limited
It doesn’t go into the details, which get really complicated really quickly.
This is a repo with some information on one person’s attempt to use AI/ML to help solve the problem: tilkb/thermoAI: Heating system control with Reinforcement Learning
I use HA with the in-floor heater (mine is electric) in my bathroom. I have a Zigbee floor thermostat that controls the heating element. HA adjusts the set point based on automations that attempt to determine if the space is likely to be occupied. Since the response time of the heat is on the order of an hour or more (to raise the temp several degrees) this is not an easy task.
Think about what you are really trying to achieve and why. WHY do you want to switch from Loxone? WHAT do you hope to gain from doing that?
My goal was to reduce the energy usage of heating the floor. Since the heat went into the house, I was not too worried about it in the winter, but in summer time, there was likely some wasted energy. In the year I have had the new system in place, I have reduced my energy consumption for floor heating and only moderately decreased my enjoyment of a warm floor (definitely some room for improvement, but physics will limit the results until I get future presence sensors working which current understanding of physics also limits)
Why? Simple: money 
I’ve got a quite old Loxone Miniserver. If it’s faulty, the only option would be to replace it by “something”.
A new Loxone Miniserver costs ~780€. In my case, I would also need the KNX extension which is not integrated in the newer version anymore: another ~680€.
So ~1460€ in total.
Using Home Assistant would be much cheaper: 100€ for Home Assistant Green + ~160€ for a KNX to Ethernet adapter. (+ maybe buying some dedicated hardware for the underfloor heating)
If this truely is your WTH, that is out of scope for the WTH. WTH purposely excludes new integrations, new DIY devices, or devices outside HA. I recommend re-reading the WTH guidelines again.
what do you mean by new devices?
the proposed solution would only require code in Home Assistant
And that would be a new integration. Please make a Feature Request, link below:
"I really want this new integration or add-on to be implemented, so I can use my devices. Is this the right place for it?”
No, this event is not for requesting new add-ons or device/services integrations. Please use the “Feature Requests ” forum category instead.
Oh yes that was so obvious /s