Smoke and CO alarms: get the basics right, then make them smart
Interlinking matters more than anything you can buy. Here is correct placement, the optical-versus-ionisation question, why CO needs its own device, and what smart genuinely adds.
Interlinking matters more than anything you can buy. Here is correct placement, the optical-versus-ionisation question, why CO needs its own device, and what smart genuinely adds.
Smoke alarms are the single most cost-effective safety device in a home, and most homes have them installed in a way that measurably reduces how well they work. Making them smart is worthwhile, but it is the second improvement. The first is more basic and matters more.
An interlinked system means that when one alarm detects smoke, every alarm in the house sounds. Standalone alarms only sound where the smoke is.
This is not a subtle difference. Most fatal house fires happen at night, and the failure mode is a fire starting downstairs while people sleep upstairs behind closed doors. A single alarm sounding in the kitchen may not wake anyone. Every alarm in the house sounding will.
Research consistently shows that time-to-waking is the variable that determines outcomes, and interlinking is the cheapest way to reduce it. If you do one thing after reading this, make it that.
Interlinking is available in three forms:
Wired. Mains-powered alarms connected by an interlink cable. The traditional approach, required by regulation in new build in many countries, and the most reliable. Retrofitting the cable is disruptive.
Radio-interlinked. Alarms that talk to each other over a dedicated radio link, independent of your Wi-Fi or hub. Retrofit-friendly and just as effective for the core function.
Hub-interlinked. Smart alarms that interlink via your home automation system. Convenient, and it introduces a dependency worth thinking about — see below.
Placement errors are common and they cost you detection time.
For kitchens, use a heat alarm instead. It responds to temperature rise rather than particles, so toast does not set it off, and it belongs on the ceiling of the kitchen with a smoke alarm in the adjacent hallway.
Two sensing technologies exist and the distinction is practical rather than academic.
Optical alarms detect light scattered by smoke particles. They respond faster to smouldering fires — soft furnishings, overheating cable, a cigarette in a sofa — which are the fires that produce the most smoke and the most fatalities.
Ionisation alarms detect the disruption of a small ionised current. They respond faster to fast-flaming fires with small particles, and they are considerably more prone to nuisance alarms from cooking.
The consensus has moved decisively towards optical for domestic use, and many markets have effectively phased ionisation out of retail. Buy optical, or a combined multi-criteria alarm if you want both.
Then check for a sealed ten-year battery, which removes the two commonest failure modes at once: a flat battery and a battery deliberately removed after a nuisance alarm.
Smoke alarms do not detect carbon monoxide, and CO alarms do not detect smoke. Combined units exist and are fine, but do not assume one covers the other.
Carbon monoxide is produced by any incomplete combustion — gas boilers, gas hobs, wood burners, open fires, paraffin heaters, and a car running in an attached garage. It is colourless, odourless, and its early symptoms resemble flu, which is precisely why it kills people who had time to get out.
Place a CO alarm in every room with a fuel-burning appliance, and in rooms where people sleep if there is any combustion appliance in the property. Unlike smoke, CO mixes fairly evenly with air, so mounting height is less critical — head height on a wall or a shelf is fine, roughly one to three metres from the appliance rather than directly above it.
CO alarms have a finite sensor life, typically seven to ten years, printed on the unit. They expire whether or not they have ever alarmed. Diarise the replacement date.
Now the genuinely useful part, which is not the alarm itself but everything around it.
Notification when you are out. The most valuable addition by a distance. A fire in an empty house, or one with only pets in it, otherwise goes undetected until a neighbour notices. Minutes matter enormously.
Automatic responses. On a confirmed alarm, a hub can turn every light in the house on at full brightness, unlock a smart lock on the escape route, stop a ventilation system from circulating smoke, and shut off a gas valve if you have one fitted. None of these are gimmicks; lighting an escape route in a smoke-filled house at 3am is meaningfully useful.
Self-testing and battery reporting. Alarms that tell you they are healthy remove the monthly test that almost nobody performs.
Distinguishing which alarm triggered. Knowing the alarm was the utility room rather than the landing is useful information when you are deciding whether to investigate or evacuate.
A smart alarm adds capability. It must not subtract reliability.
The test is simple: does the alarm still sound, locally and loudly, with no internet, no hub and no cloud service? Any alarm worth buying does, because the alarm function is self-contained and certified, and the smart features are an addition on top. Confirm this before buying rather than assuming.
Similarly, if you rely on hub interlinking rather than radio or wired interlinking, then your hub going offline reduces your protection to standalone alarms. That is an argument for choosing alarms that interlink among themselves independently, and letting the hub handle notification and automation rather than the life-safety function.
Steps one to five cost very little and matter more than anything smart you could buy. Step six is where the technology earns its place.