Control is decided before the fitting, not after

Control and technology

Choosing the luminaires and then asking how they dim is the wrong order: the protocol constrains the fitting, not the other way round.

The wrong order, and why it is so common

On a contract project the question "how does it switch on" almost always arrives after the fittings have been chosen. It is understandable: fittings can be seen, control cannot, and meetings happen about the things that can be seen. But by then the control protocol becomes a constraint to endure.

There are two consequences, and they arrive together: some fittings do not support it at all and have to be replaced; others support it with an accessory that doubles the cost of the light point. Neither shows at selection stage, because at selection stage control was not yet a question.

The opposite route: the scene first, the fitting after

Designing the scene before the supply runs the other way: you decide how many scenes are needed, who calls them and from where, and from there you come down to the fittings that can do it. It is less natural, because it forces a conversation about how a room will be used before the room exists.

In exchange it is the only order in which control costs what it is worth. A fitting chosen knowing what it will have to do is bought once; the same fitting chosen first and adapted afterwards is paid for twice, and the second time the site is paid for too.

How many scenes, really

The question "how many scenes?" almost always gets an inflated answer, because at design stage every scene looks free. In practice three scenes that get used are worth more than twelve programmed and never called, and the twelve cost in programming, in staff training and in confusion at the panel.

The number is decided with whoever will run the room, not in the technical office. The people who open in the morning and close at night know which moments of the day are real; the technical office knows which are possible. They are two different lists, and the one you need is the first.

Who controls it, and from where

A wall panel, a tablet and integration with the property system are three different installations, not three ways of doing the same thing. The cabling changes, who is allowed to touch it changes, and what happens when the person who knew how to use it moves on changes too.

Choosing after the supply means running cables where none were planned, and in a finished building every new cable is a chase, a making good and a repaint. It is the line that blows control budgets more than any software licence.

Compatibility is tested, not read

Two products declaring the same protocol do not always talk to each other. Control families leave room for interpretation, and every manufacturer fills it their own way: the declaration is true, and the pair still does not work.

The moment to find out is the sample, not the commissioning. A test with a real fitting, the real driver and the real control costs a day; the same discovery at commissioning costs the delay of an opening, and arrives when there are no alternatives left.

The cabling is the part that does not get redone

The cost of changing protocol after the supply is not the cost of the drivers: it is the cost of redoing the cabling in a finished building. Drivers are swapped in an afternoon, cables are not.

A practical rule follows: control is chosen while the ceiling is still open, that is, far earlier than it feels necessary. If the decision is not taken by then, you at least lay the cable that keeps most routes open, because one extra cable costs a fraction of one chase.

Deep dimming is a promise to be verified

"Dimmable" on a data sheet does not say how far, nor how. In hospitality the part of the range that matters is the bottom: from ten per cent down, where a room turns into an evening room. Many fitting-driver-control combinations behave well at the top and badly at the bottom — steps, flicker, sudden cut-outs.

This too is tested on the sample, and it is worth testing at the final colour temperature and with the source that will actually be installed. It is the test that separates an installation people use from one that merely switches on.

Who will maintain it, and with what documentation

A control system is the one part of an installation that has to be understood by somebody, years later. If the programming lives in the head of whoever did it, the first change starts from scratch, and the cost is that of a new commissioning.

Three objects have to be asked for with the supply: the schedule of what controls what, the programming files delivered in a format that opens, and the name of whoever answers for it. Without the first two, the third becomes compulsory forever.

Control is not for switching off: it is for changing the room

There is a misunderstanding behind many unused installations: control gets thought of as a way to save, that is, to turn down. In hospitality its job is another and more important one: the same room has to be three different rooms at three moments of the day, and none of the three is "the same one, darker".

A breakfast room at seven in the morning needs level and evenness; at eight in the evening, with the same tables, it needs contrast and few lit zones. That is not a dimming setting, it is a different scene: which fittings are on changes, not only how much.

That is why an installation designed to dim everything together is abandoned after three months: it does one thing, and that thing is of use to nobody. The operator switches half the circuits off by hand and the scene stays a design idea.

The range that matters is the bottom one

"Dimmable" on a data sheet says neither how far nor how, and the part of the range that decides whether a room works in the evening is the bottom: from ten per cent down. That is where many fitting-driver-control combinations behave badly — visible steps, flicker, sudden cut-outs.

The useful test is trivial and almost nobody makes it: take the real piece, with the real driver and the real control, and go slowly down to off while watching a wall. It takes five minutes and separates an installation people use from one that merely switches on.

Whoever will use it is not whoever buys it

Control is the one part of an installation that has to be used by somebody every day, and that somebody almost never attends the meetings where it is chosen. The commonest defect follows: an interface designed for the person who designed it.

The check worth making before handover is simple: ask a member of staff to call up the scenes, without explanation, and watch. If it takes them more than ten seconds to find the right one, they will not use it — and at that point it is not a training problem, it is a problem of how many scenes there are.

When automation helps, and when it irritates

A presence sensor in a service corridor is almost always right; the same sensor in a room where somebody sits reading is a defect, because it switches off while they are still there. The difference is not the product: it is how much people move in that room.

The practical rule that follows: automation goes where behaviour is predictable and passage is brief. Where people stop, a person is in charge. And in any case a manual override is needed, because the first thing staff do with automation that gets it wrong is disable it forever.

Documentation is the only thing that stays

A control system is the one part of an installation that has to be understood by somebody years later. If the programming lives in the head of whoever did it, the first change starts from scratch, and the cost is that of a new commissioning.

Three objects have to be asked for with the supply: the schedule of what controls what, the programming files delivered in a format that opens, and the name of whoever answers for it. Without the first two, the third becomes compulsory forever — and that is exactly the dependency nobody wants to discover they have.

The cost that does not show: people

Into the cost of a control system go the drivers, the controls, the cables and the programming. What almost never goes in is the most recurring item: staff time, both to learn and to use. Twelve callable scenes mean twelve things to explain to every new person.

That is why three used scenes are worth more than twelve programmed, and it is not an aesthetic preference: it is arithmetic. Three scenes are explained in two minutes on the first shift; twelve are explained badly, remembered worse, and in the end two get used.

What to test before signing

The tests that separate an installation that works from one that seems to work are five, and they are all made on a sample:

• The bottom of the dimming range, taken slowly to off while watching a wall
• the dialogue between fitting, driver and control, with the real parts
• a scene called up by a member of staff, without explanation
• the behaviour after a power cut: what comes on when it returns
• how long a scene takes to start — under a second nobody notices, over two everybody does

What to put in the specification, before choosing the fittings

Five lines, and they belong in writing while the fittings are still a hypothesis:

• how many scenes, decided with the operator, and what each one does
• who calls them and from where: panel, tablet or integration
• The dimming range required, with attention to the bottom end
• that compatibility is demonstrated on a sample, not declared
• which documents and programming files are handed over, and to whom

• A bespoke piece
• The properties we have lit
• Free lighting calculation
• Lighting knowledge