It does not wait for the smoke to rise. It goes and fetches it.

An ordinary ceiling detector waits. It sits still, and it lights up when the smoke reaches it by whatever route the smoke chooses to rise. An aspirating system does the opposite: a small fan continuously draws air through a network of pipes with holes spread around the enclosure, and takes that air to a chamber where a beam of light measures how many particles it is carrying.

The practical difference is in the sensitivity and in the time. That detector can measure smoke concentrations far below what sets off a point detector, and so it can raise the alert when there is no fire yet: when a cable is heating up, when a power supply starts to smell, when something is slowly toasting. That alert arrives at a moment when you can still go, look and unplug.

Technical rooms and data centres. There are two reasons and both are good. The first is that what burns there starts as a cable heating up over hours, and raising the alert at that stage means not having to put anything out. The second is the air: in a room with the air conditioning at full tilt the smoke does not rise to the ceiling, the airflow carries it away. A ceiling detector can take a very long time; a pipe sampling in the air return finds out straight away, because it samples exactly where everything passes.

Tall industrial units and large volumes. At twelve or fifteen metres, the smoke dilutes and stratifies before it gets to the top, and going up to maintain a detector at that height also needs a platform and a shutdown. With aspiration, the detector goes at a reachable height and only the pipework goes up. In museums and archives the same reasoning holds for another reason: what you see is a small hole instead of a detector hanging from a vault.

The unit is the aspirator, a filter, the detection chamber and the electronics that decide. The network comes off that: plastic pipe across the ceiling or wherever it is needed, with holes drilled in calculated places. And the word calculated is literal: the length of each branch, the diameter and the size of each hole are sized with the manufacturer's software, because whether all the holes draw a similar amount of air depends on that.

There is a second number that rules: the transport time, which is how long the air takes from the furthest hole to the detector. It has a permissible maximum, and it is checked at commissioning by releasing smoke at the worst point and timing it. It is the test that tells an aspirating installation that has been commissioned from an aspirating installation that has simply been switched on.

First the building work. Pipework has to be run across the ceiling, under the raised floor, inside cabinets or wherever the risk is, fixed, taken between compartments with its seals, and done in a place that is often in use and cannot be stopped. That is real installation work, not hanging a detector, and it is the line most underestimated in quotations.

Second the sensitivity tuning, which is a process and not a parameter. A unit like this also measures the normal dust of the enclosure, the fumes from a diesel engine coming in through the door, the steam from a clean. For the thresholds to make sense you have to leave it measuring for a while with the real activity of the place —including the worst hour of the worst day— and set the levels above that floor. A system configured with factory values in a store ends up disconnected in three weeks.

Third the maintenance, which is not that of a detector. There are filters that get changed, an aspirator that is a mechanical part with a limited life, and one check that few people do and that is the one that matters: releasing smoke at the furthest point again and confirming that the transport time is still what it was. A network that has been in a ceiling for five years has changed, because somebody has done building work and touched something.

It is not a more expensive ceiling detector for an office, and it is no use if the air in the enclosure does not pass where you are going to sample. And there is one case where we say no outright: when the client wants the aspiration but does not want the maintenance that goes with it. It is a unit that gives an extremely valuable alert very early, and that alert is only worth something if there is somebody who receives it, understands it and goes to look. Without that, the first thing that happens is that the early alerts get ignored, and the second is that somebody raises the thresholds «so it stops being a nuisance». From then on you have paid for a very early detection system in order to have an expensive ceiling detector.

The whole point of aspiration is to raise the alert so early that nothing needs putting out yet. In a technical room that is the entire objective: that somebody goes in, looks, finds the power supply that smells and unplugs it. If it has got to the stage of extinguishing, the system raised the alert late or nobody did anything with the alert.

What we do not do, here as everywhere: sprinklers, dry risers, hose reels, pressure sets, tanks and their pipework. It is another trade with another licence. And if the room is going to have gas suppression, the detection and the release are ours and the bank of cylinders with its pipework is not. We say it in the first meeting, because in a technical room those two quotations almost always go together and it is worth knowing which is which.

This is operated differently from a panel, and it is worth saying plainly: what the client has to learn is not how to work a unit, it is how to respond to an alert. When the first level comes in, somebody has to go to that room, look, smell and search. Not silence it. That procedure is written down, stuck on the wall and rehearsed once, because the first time it sounds it is going to be a Tuesday at seven in the morning.

What nobody inside can do is touch the thresholds. It is temptation number one and it is what disables these systems: nudging the sensitivity up a little every time it is a nuisance, until it no longer detects anything early. If an alert is a nuisance and is not an outbreak, the problem is in the tuning or in the activity of the enclosure, and that gets analysed, not raised.