Aircon surge protection: where it sits decides its reach

A surge is a brief, violent rise in voltage on a circuit that is otherwise working normally. It arrives and leaves faster than anything mechanical can answer. Nothing switches off, no protective device moves, and the supply is still there afterwards. What changed is what the electronics on that circuit were exposed to while it passed.

Lightning reaches equipment by four routes. Only two of them require the building to be struck. IEC 62305-1 is the standard behind Singapore's SS 555 series. It names the four as a flash to the structure, a flash near it, a flash to a line connected to it, and a flash near such a line. The last two matter most to a household. Neither of them needs anything to happen on the roof.

The damage that follows is attributed to the lines as often as to the strike point. Overvoltages induced on connected lines get transmitted into the structure, and once inside they travel the same conductors as everything else. A flash at or beside the building also couples into equipment through the field it throws. Both mechanisms finish in the same place, which is whatever electronics happened to be energised.

Singapore hands those routes an unusual number of chances. The Meteorological Service Singapore puts the average at about 176 lightning days a year, and describes the local strike rate as one of the highest in the world. A parliamentary reply from the Ministry of Sustainability and the Environment states the same figure as lightning on one day in every two. Activity peaks in November, then April and May, and the inter-monsoon months account for more than half the strokes in a year.

Not every surge on a circuit came from the sky. The IEC guidance on protecting electronic systems treats lightning and switching transients together, because a large motor dropping off a shared circuit produces its own. That matters for attribution. A board that failed on a clear evening has not ruled out a supply event.

The board holds the lowest tolerance of anything in an aircon, and that is what puts it first in line. A compressor winding and a fan winding are lengths of insulated wire built for continuous duty. The board runs low-voltage electronics, and what it will survive in the way of a brief overvoltage is a fraction of what the motors will shrug off.

Protection standards treat that tolerance as the number every other choice is set against. IEC 62305-4 requires the let-through voltage of an installed device to be matched to the immunity withstand voltage of the electronic equipment behind it. Read from the owner's side, that says the board's own limit is the target. A device only earns its place if it holds the residue below that limit.

The board is also live for far longer than the machine runs. Anything with a receiver listening for the remote is drawing standby power the whole time the circuit is on. The board therefore sits energised through storms that pass while the unit is switched off at the remote. Exposure follows the circuit, not the running hours.

A damaged board is the most expensive way for an aircon to fail short of the compressor. Sourcing for older models is the usual constraint, and pcb repair vs replacement takes that decision apart. How a failing board announces itself is a separate question again, sorted under faulty pcb signs.

Protection is positioned at boundaries, and a boundary is wherever a service crosses from a more exposed zone into a less exposed one. IEC 62305-4 puts it as fitting appropriate devices at every such crossing. The first crossing is where the supply enters the building. The next is the distribution board. The last is the equipment itself.

No single position covers the job, and the standard is blunt about it. A heavy device at the service entrance takes the bulk of the energy, and the guidance states that such devices on their own give no effective protection against failure of sensitive electrical or electronic systems. Downstream devices then hold what is left down to something the equipment can survive. The set is described as coordinated, which means the positions have to be chosen together.

In a flat, only the last of those positions belongs to the household. The incoming service and the main switchboard of an HDB block or a condominium belong to the building, and nothing an owner arranges reaches them. What an owner controls starts at the consumer unit inside the flat, which is the db box on the wall.

That leaves a trap worth naming before anybody spends money on it. A plug-in protector guards whatever is plugged into it, and a wall aircon is wired to a dedicated aircon circuit with no plug anywhere in the path. Buying one for a television and assuming the aircon came along with it is a common and entirely understandable error. The aircon has its own circuit precisely so that it shares nothing.

Anything fitted inside a consumer unit is licensed work, covered under licensed electrical worker. An owner's part of the job is deciding whether to ask the question, never choosing the part.

Every standard in this area speaks of reducing risk. None of them offers to remove it. SS 555 Part 4 is the Singapore Standard covering protection of electrical and electronic systems. It describes its measures as reducing the risk of permanent failures caused by a lightning electromagnetic impulse. Part 2 of the same standard sets out a risk assessment procedure. That would have no purpose if some residual risk were not assumed.

A device limits, and limiting leaves something behind. What gets through is the let-through voltage, and the design question is whether that remainder falls below what the equipment can take. Where it does, the board survives an event it would not have survived bare. Where it does not, the device operated correctly and the board failed anyway.

Badly chosen positions can make matters worse instead of better. The guidance notes that poor coordination leaves downstream devices absorbing more energy than they were selected for, putting both the device and the equipment at risk. A protective part that was never sized for what reached it becomes a hazard sitting inside a switchboard.

A fitted device also has a service life, and is not a permanent alteration to the building. SS 555 Part 4 covers inspection, maintenance and testing alongside design and installation, which tells an owner the thing has to be looked at again. Installed and forgotten, it is an assumption, and the assumption stays untested until the next event arrives.

The useful record gets made on the night, and almost none of it survives the week. What was running, what stopped, what carried on, and roughly when. Those four facts are what tell a supply event apart from a fault that had been building quietly.

Whether other equipment in the flat was affected is the single most valuable line in that record. Something arriving on the supply reaches the whole circuit, not one appliance. Where the aircon board and a router and a set-top box all went in the same hour, the supply is implicated. Where the aircon alone stopped and everything else carried on, the case for a supply event weakens sharply.

There is a public record of lightning, though it is built to look forward. NEA runs a detection system covering cloud-to-ground lightning and lightning occurring inside storm clouds. The myENV app issues alerts to subscribers when detection or a thundery-shower forecast falls within a six-kilometre radius of a saved location. An alert received at the time is worth keeping. The service warns rather than archives, so it helps only where somebody noticed.

A storm and a dead board on the same evening make a sequence. Turning a sequence into a cause is somebody's job, and nobody can do it from memory alone. Boards give way on quiet nights too, and one already marginal will often pick the night the supply moves. The sequence earns its place in the record because it stays checkable later.

• •The date, and the nearest hour anyone can name
• •Whether the weather did anything unusual then, and whether an alert came through
• •Which unit stopped, and whether the other units in the flat kept running
• •What else on the same circuit or in the same flat behaved oddly afterwards
• •Whether anything at the db box had moved, and whether anyone reset it
• •Whether the unit was running, switched off at the remote, or off at the wall
• •What the display or indicator lights did immediately after, if anyone saw them