openHAB thermostats

Your floor responds much faster than mine did.

Our slab was heated by embedded copper tubing with circulating hot water. Although one could notice a slight increase in temperature after the heater and pump were on about an hour, it took about three hours before significant heating. Similarly, the slab stayed warm several hours after the heat was turned off--it was still very comfortable to the feet at 8am after being shut off at midnight!

The vented, metal thermostat had no anticipator heater, since providing a time constant comparable to the slab would have been impractical.

--
-michael - NadaNet 3.1 and AppleCrate II:  http://michaeljmahon.com

Have a look a PID control methods too.

---druck

That very same problem is already present with simple old-style metal radiators. Turning them on they get hot in minutes, turning them off changes nothing at first. But a small wall radiotor in a large room has a large power to mass/area ratio, so it cools down fairly rapidly. Several tons of pavement material distributed over the whole floor area take a long time to cool down.

So what you have to do is turn off long before the desired temperature is reached, wait an hour or so, and look. That's exactly what the internal feedback by heating resistor does. Simple and easy in analogue electro-mechanical hardware, tricky to emulate digitally. Of course the downside to all this is, that it takes a log time for a room in a colder equlibrium to reach a new higher desired temperature. As a human coming home you turn up the thermostat a few degrees higher until the heating "gets going" appreciably. Having got a running digital emulation that bit should be amenable to be added in, but that's an extra second step for later.

--




/ \  Mail | -- No unannounced, large, binary attachments, please! --

The point of PWM is not for the frequency to be high in any absulute sense (there is none) but too high for the system in question to follow. For floor heating a cycle length of about two hours is PWM and it is what el cheapo wall thermostats with internal feed back enforce.[1] The cycle length with an overshooting on/off thermostat would be more than ten hours. If you are faster than the natural cycle by an order of magnitude, then by definition what you get is PWM.

[1] The valves don't like being switched too often. Ten times higher frequency means ten times lower life expectancy and would become rather a nuisance.

--




/ \  Mail | -- No unannounced, large, binary attachments, please! --

I did, and to elaborate, it can be done using a branch chip, but also if proper care is taken with regards to powering, cabling quality, and similar.

This document proved to be quite enlightening:

--

MN-1334-RIPE           SA0XLR            +46 705 989668 
Should I do my BOBBIE VINTON medley?

Clearly an AI is required to do it effectively. :-)

With on/off control and a situation like that, common practice is to establish a "deadband" around the desired point and turn the system on when below the low value and off when above the high value.

In the case of infloor heating, you need a sensor in the floor as well as in the room and when the floor reaches a certain temperature, the system turns off so the floor doesn't get so hot that the room temp overshoots significantly.

--
Jim H

While this may be desirable you don't actually need it. I've just moved to a flat in a newly built house and had my first experience with floor heating. It took me a bit to understand how the current system works.

I have a bimetal swith with a large hysteresis. It switches on at about the nominal temperature and off about three degrees (six in Fahrenheit) higher. There also is a heating resistor in the case making it switch off in about an hour or so even if the outside temperature does not rise.

According to regulation textbooks this setup switches on an off about ten times as frequently as a highly overshooting setup using a simple switch without hysterisis would and the amplitude of the swings drops by a similar amount. The downsides are:

1) The offset between nominal set point and actual average gets larger. This is easily compensated for by moving the scale for setting the temperature. (As water temperature follows outside temperature, the difference between the regulated inside temperature and that reached with heating full on stays about constant, so the regulator stay inside a limited parameter space.) 2) The system takes longer to follow large changes in the desired set point. As this is rare, you can compeksate manually by setting it higher on coming home and turning down after a while.

It still means that emulating this cheap standard off the shelf unit in software can get quite tricky for someone who has not had any trainig in how to do it.

--




/ \  Mail | -- No unannounced, large, binary attachments, please! --

There is LOTS of software out there that is supposedly written by trained people that is full of security holes and other issues. Often the issues is that the wrong type of people are used to do everything, assumption that doing anything is copy from stackoverflow (etc) because it is just adding a library call toi sort out 'x', or mamagement not wanting to waste time and money on that.

Consider the many publicised problems on data breaches and other issues.

I have reviewed some software systems for customers and in some fields the amount of security and simple errors are outstanding.

See

--
Paul Carpenter          | paul@pcserviceselectronics.co.uk 
    PC Services 
 Logic Gate Education 
     Timing Diagram Font 
    For those web sites you hate

Quite. A lot of it is incompetence, but there's also idiots writing obfuscated, unmaintainable code 'just because they could' and knowing that they would not get the kicking they deserved for doing it.

This has been going on for years: here is a reference to a few choice examples I had the misfortune to encounter in the '70s:

--
martin@   | Martin Gregorie 
gregorie. | Essex, UK 
org       |

As long as I can't get rid of the squatter in my flat (my daughter, back from several years studying abroad) and all my electronics is in (expensive) storage, I can't do much except look round.

I just had a catalogue from the German electronics mail order shop Reichelt in the post with a nice table summarizing the available home automation systems. All can do the silly stuff, five out of ten can control radiator thermostats and only three can do floor heating. All three of those won't work without the Klaut (German for pinch/steal, an approriate name choice imho).

So it is as I and all those of you who chimed in feared: I'll have to do it all by myself from scratch. Ah well, that's how I used to prefer it anyway, but starting something entirely new keeps getting harder with age. And I haven't really done any serious electronics for over twenty years now. What are you saying? "Stop whining and get on with it!" Right you are.

--




/ \  Mail | -- No unannounced, large, binary attachments, please! --

The most advanced, although little known, is the Brit Idratek system.