In brief
A smart home in a residence starts with the cables in the walls, and those go in once. Before anybody shows you an app, you have to settle what the house is supposed to do on its own and write it down before the electrical designer places the first back box on the drawings. Some decisions are made together with the electrical design, and after the plastering every one of them that is missing means chasing walls: how the cables are run, the space for the distribution boards and the technical cabinet, power and control for the shading, routes outdoors; the rest can wait or should stay separate, the alarm first of all. I write about what I require from the system and from the integrator, what such a system depends on, and whose project it is once the house is standing. I do not program these systems and I do not choose them for you.
What the house should do by itself
The electrical installation is my favourite stage of a build and, at the same time, the hardest one for clients. In an ordinary house the difficulty is this: to place the sockets and the switches you already have to know where the beds will stand, what the kitchen will look like and which lights should come on from one button – with a year still to go before the finishing. In a residence the same difficulty grows, because the house is supposed to do more things on its own, and every one of them is a cable somebody has to lay before the plaster.
So I start the conversation about automation with a sheet of paper. What should happen by itself, and why? What annoys you today in the house you live in? Which things do you do every single day? The answers are usually simpler than the brochures suggest: all the lights go off when you leave; at night the corridor comes on just enough to get to the bathroom without waking up properly; the shading comes down by itself when the sun reaches the terrace; the heating knows the house is empty. Everything else follows from that list – how many cables and what kind, how big the boards are, and even whether you need a system at all or a few modules will do.
At home, when my wife and I wrote such a list, it ended at the window shading. We needed nothing more, and that is how it stayed. A short list is fine. Because the other ending is this: a guest who came for the weekend stands in the living room and does not know how to turn on the light. So much for the intelligence.
The second thing I put before the electrics is the interior design. Without it, planning automation is guesswork: nobody knows where the table will stand, which side of the bed the button goes on, or whether the wall light above it will exist at all. On a residence the interior design is part of the documentation from the start – I wrote about it in the text on why a residence takes two years. On the builds I run, the list of functions for every room exists before the electrical designer places the first back box on the drawings.
What is decided together with the electrics
An ordinary installation is simple: a cable runs from the board to the switch, and from the switch to the lamp. The switch breaks the circuit and that is all. In an installation built for automation, the controlled points – lighting circuits, shading motors, selected sockets – usually each get their own cable run straight back to the board, or to a sub-board on that floor. The switch on the wall no longer breaks anything itself; it sends a signal to a module or onto the bus, and what happens next is written in the program. It sounds like a detail. From that detail it follows that there are several times more cables, the board is several times bigger, and an electrician who does not know the house is getting automation will build an installation that cannot be reprogrammed afterwards, because physically there is nothing to reprogram. So the automation design should be made together with the electrical design. If it turns up after the plastering, the scope shrinks and the cost grows: chasing, new plaster, painting, and a good part of it cannot be done at all.
A few things on that list are decided earlier than people assume.
The first is space. A residence has several distribution boards – per floor, for the garage, for the pool plant, for the garden – and every one of them needs more room than it needs today, because automation modules take up far more rail than the protective devices alone, and something gets added during the build. I have not yet had a build where nothing was added. On top of that comes the cabinet for the electronics – network, the automation server, battery back-up – which needs a dry, cool room with access for servicing, preferably not next to a bedroom, because fans hum. That room has to be drawn in the design. After the fact, it ends up in the garage, next to the mower.
The second is the routes outdoors. The gate, the intercom, garden lighting, irrigation, cameras, the technical outbuilding, the charger in the garage – all of it needs a cable that passes through the foundation and the ground before the facade and the paving exist. On every such route I lay one empty duct more than the design calls for – separately for power and for signal cables – with a draw wire inside. An empty duct is a trifle in the bill of quantities. Digging through a finished terrace costs you the terrace.
The third is the shading. With large glazing, external venetian blinds and roller shutters are an installation in their own right: every opening needs power and control, above every opening there has to be room for the housing, and somewhere high up a wind sensor has to stand, one that raises the blinds before the wind breaks them, set to the thresholds the manufacturer specifies – a lowered blind works like a sail, and in a windy spot I treat that sensor as compulsory, which is usually the blind manufacturer’s condition too. With this many windows, crank-operated shading stops making sense; I have not planned it for years. The housing, the sleeve and the power supply to every opening have to be on the facade and electrical drawings; the shading supplier comes later and finds what is there.
The fourth is the small stuff people forget most often and remember while looking at fresh plaster: ceiling tracks for motorised curtains, speaker cables for the cinema and the audio zones, a wired network wherever a television or a desk will stand (a cable does not depend on signal range or on how many devices are hanging on the Wi-Fi), deep back boxes throughout the house, because a shallow one has no room for a module, connectors and a spare length of cable, and the price difference is nil. And power: if a charger is one day going to stand in the garage and a heat pump by the pool, the question of the supply capacity comes up already with the plot for the residence.
The fifth is surge protection. A residence with automation is full of electronics, from the modules in the board to the motor in every window, and one surge from the grid can take all of it out at once. Protection is staged – in the board, at the equipment, and in an extensive installation also along the way – and it covers the network lines and the camera cables too, especially the ones that leave the building, which almost nobody does. Without proper earthing no arrester has anywhere to send the energy. Part of this follows from regulations, the rest from common sense. Proper protection needs a design and an electrician who knows the subject. An arrester picked at random, without checking the earthing and without the whole protection concept, mostly gives the feeling that something was done.
What the system depends on
When a system is being chosen, the question that usually comes up is “KNX or a manufacturer’s system”, because most integrators start with the system they sell. That is normal. I start somewhere else: who and what will this system depend on for the next twenty years? If the answer is “one installer, one account and one server”, I slow down.
On the installer. Automation does not end on handover day. The first months after moving in are tuning: a scene that turns on too much in the morning, a blind that comes down too early. Then come children, new furniture, other habits, and the installer comes back many more times. That is good, that is how it should be. It is worse if every such change is possible only through this one man, or through somebody who works on the same manufacturer’s system.
On the manufacturer. You build a house for decades, and the modules, the controllers and the apps will change several times in that period. If the system belongs to one manufacturer – his devices, his software, his licences – then the whole house inherits that company’s fate: takeovers, changes of business model, end of support. An open standard, such as KNX, is built differently: it belongs to no manufacturer (an association maintains it), many manufacturers make the devices and they can be mixed in one installation, and another integrator working in that standard can take over the project, provided you have the project file, the passwords and the keys, which I come to in a moment. I do not write off single-manufacturer systems; some of them are quicker to program and suit a pool or a sauna better. I only want it known, before the contract is signed, who will be able to service this in five years and on what terms.
On the controller. In some systems every device has its own logic: the button talks to the module directly and the light works even if the server, the router and the app are all down. In others everything passes through one central controller, and when it fails, the house fails – no light comes on from any button until somebody arrives. Both arrangements have their place. The requirement I set does not depend on the arrangement: the button on the wall has to switch the light locally – without the central controller, without the internet and without the manufacturer’s server; the module in the board and its power supply are then basic infrastructure and have to be designed as such. The shading has an agreed state in case of failure (closed or open – that is not obvious at all, and it is settled separately for every room). With gates, what happens without power is decided by the drive and its safety devices: stopping, manual release, access for the emergency services; the automation gets a say only after those. I wrote about the manufacturer’s server in the texts on the alarm and on the cameras, so only the short version: a service works for as long as it pays somebody to keep it running.
And finally the power. The cabinet with the electronics gets battery back-up, and after the power returns the system has to come back by itself, with no logging in and no call to the service. At handover, besides the measurements and the test reports, I also run the simplest test there is: I turn off the main switch and watch what comes back on its own.
I do not know which of today’s companies will exist in fifteen years. That is why I require the basic functions of the house to work even when one of them disappears.
Whose project it is
There is one more thing nobody mentions when the contract is signed, and it usually surfaces years later: the whole logic of the house sits in the project file. In it are the functions of every button, the device addresses, the scenes and, in newer installations, the encryption keys as well. Whoever has that file and the password to it can change the house. Whoever does not can, at most, guess.
In many installations that project exists only on the installer’s laptop. No copy with the owner, no device list, no description of what each module does. As long as the installer answers the phone, nobody notices. When he stops – because he changed trades, moved away or the company ceased to exist – a new integrator stands in front of the house like in front of somebody else’s computer without the password. A lost project password usually cannot be recovered (in an open standard there are, by design, no back doors), and what remains is rebuilding the project from scratch. (With ordinary electrics the same applies, only on a smaller scale: that is why the protective devices in the board are to be fitted straight away, and after the installation there is to be documentation with photographs taken before the plastering. If contact with the electrician were lost, it is better to find out at the shell stage than after moving in.)
So in the contract with the integrator I write down what you receive at handover: the project file with its password, the device list with addresses and a description of functions, the network documentation, every administrator password and key, and a guide for the household written in plain language – which button does what, what works after a power cut, whom to call. That is your property, the same as the walls. To this I add two things that have more to do with good order than with technology: a list of the people who have access to the project, and the rule that the service works on its own account, which can be switched off after they leave. This is not a lack of trust in the installer. The point is that access to the house should not depend on whether somebody remembered to log out. It also happens that the settings of the whole house are registered to the installer’s account or the first owner’s, and at the sale it turns out the previous owner still has access. So I ask whose account all of this is registered to.
What stays separate
There are a few things I do not hand over to the automation, even though it could control them.
I run the alarm separately and join it to the automation at one agreed point: the automation receives from the alarm, one way only, as much as it needs – for instance that the house is armed, so that it can turn the lights off – while arming and disarming is done by the alarm alone, on its own installation. Why – I wrote that in the text on the alarm system and I will not repeat it here.
Their own controllers also stay wherever the manufacturer has built safety interlocks into them: the heat pump, the heat recovery unit, the pool plant and the pool hall dehumidification, the sauna, the gate drive, the charger. Each of these devices looks after itself – defrosting, a temperature limit, the photocell in the gate – and that is how it has to stay. The automation gets a reading from them and may ask for a setting within the permitted range; it does not drive the compressor or the heating element directly. One system that rules everything sounds good in a presentation. In practice it means one update can break the heating, the gate and the pool all at once, and the heat pump engineer who arrives and finds no manufacturer’s controller has nothing to work on.
What also stays separate is everything that does not have to be automatic at all. If you prefer to turn on the light with an ordinary switch and have no need for scenes or scenarios, there is nothing wrong with that. Automation is there to solve the problems on the list; if the list is short, so is the system.
When it is not worth it – and what to leave for later
When the list of functions is short. If what remains of the whole catalogue is the shading, the gate and the light on the drive, a full bus system for the whole house is a sledgehammer for a nut – a few simple modules do the same at a fraction of the complexity, with one person fewer on the service list. On one condition: that the installation is laid so it can be extended one day – cables straight to the board, deep back boxes, spare room. That is a cheap reserve for changing your mind.
When the decision comes after the plastering. Then radio is what is left: modules in deep back boxes, battery sensors, control in groups rather than of single points. That works, and in an ordinary house it is often enough. In a residence that door is narrow, because most of the value sits in the things radio cannot do: power to the shading, the routes outdoors, the boards, the cabinet.
When the system is meant to replace the design. No automation will fix a house in which the buttons are in the wrong places because nobody knew where the bed would stand. The function of a button can be reprogrammed; the wall the back box sits in cannot.
What I do not promise: I do not program these systems and I do not choose them for you – that is the integrator’s job, and assessing his design is for somebody who does that for a living; I do not promise that the electronics will survive twenty years without servicing, because they will not; and I do not describe how the automation is arranged in the homes of particular clients – not in texts and not in conversations.
What I ask before I say yes
- What is the house supposed to do by itself – is there a list of functions for every room, written before any conversation about any system?
- Is the automation design being made together with the electrical design, and does the electrician know the cables run straight to the board?
- How much room is left in the boards once everything in today’s design is fitted, and where will the technical cabinet stand?
- Which routes outdoors have a spare empty duct?
- Does the button on the wall turn on the light when the controller is down? And when the internet is down and the manufacturer’s server has been switched off for good?
- What state does the shading take in a failure, and who settled that for every room? And the gates – what does the drive do without power, and how is it released by hand?
- Will the project file, the passwords and the keys be mine at handover, or will they stay on the installer’s laptop – and who else has access to them? Did somebody photograph the cable routes before the plastering?
- Where is the point of contact with the alarm, and can the automation only read the alarm, or control it as well?
- Which devices keep their own controller, and what exactly can the automation tell them to do?
Frequently asked questions
KNX or a manufacturer’s system – which to choose for a residence?
That question is shorthand. I ask what the system will depend on: one installer, the existence of one company, its server, one central controller. An open standard answers most of those questions by design – it belongs to no manufacturer and another integrator can take the project over – but only if you receive the project file, the passwords and the keys. A single-manufacturer system can be a good choice if it answers the same questions satisfactorily and that is written into the contract: local operation without a server, the project and the passwords with you, servicing possible from more than one company. Only after those answers does a conversation about a particular system make sense.
Can a smart home be added to a finished house, after the plastering?
Partly, by radio: modules in deep back boxes, battery sensors, control of shading that already had power. What needs cables laid in advance – separate circuits, motors without power, sensors in the walls, room in the board – cannot be added without chasing walls. In an ordinary house radio is often enough. In a residence it usually is not, because the biggest items are the shading, the routes outdoors and the cabinet, and radio does not do those.
Should home automation control the alarm?
No. The automation may know that the house is armed and, for instance, turn the lights off; arming and disarming belongs to the alarm on its own, separate installation, and the contact between them is one defined point. The reason: an alarm is supposed to be predictable, and automation is by its nature something somebody keeps retuning. Servicing one should not open the other.
How much does automation cost in a residence?
You cannot work it out from the floor area at all. The price follows from the list of functions, from the number of shading units and points, from the number of systems that have to talk to each other, and from the moment of the decision: the same thing planned together with the electrics and the same thing added after the plastering are two different sums. Until the list exists, every figure is plucked from the air.
Answers to common questions about privacy and discretion are collected in the Questions section.
If you are thinking about a house that is supposed to do part of the work by itself, that conversation belongs before the electrical design. Let’s talk about your project.