Aircon cable size: the decision buried behind the wall

Cable size follows the circuit, not the machine. The figure that ends up on a drawing is the output of several separate questions, and the aircon answers only one of them. How much current the load draws is the starting point. How the cable is installed, what else runs beside it, how warm that space gets, and how far the run travels all move the answer afterwards.

Singapore settles this in one document, and it is worth naming correctly. SS 638 is the Singapore Standard that governs electrical installations. Its own cover carries the line [Formerly CP 5], because SS 638:2018 is the third revision of CP 5 and was re-designated at that revision. The current edition is SS 638:2018+C1:2020+A1:2022. A specification still citing CP 5 is citing a designation retired in 2018.

That code splits the cable question the way a designer splits it. The inputs sit in its opening principles, named one at a time: nature of demand, environmental conditions, type of wiring and method of installation, and cross-sectional area of conductors. Wiring systems get a chapter of their own. So does protection against overcurrent. The cable and the switch that guards it are settled against the same circuit.

A code can be voluntary and still bind, which is where owners lose the thread. The standard carries a note saying a Singapore Standard is voluntary unless a regulatory authority makes it mandatory. EMA is that authority here. It names SS 638 as the standard it adopts for electrical installations, under the Electricity (Electrical Installations) Regulations. So the code is not a trade preference. It is the yardstick the work gets measured against.

The honest reply to what size cable an aircon needs is that the question is incomplete. Two identical machines in two flats can properly land on different conductors, because the route and the conditions differ. Anyone who answers from a model number alone has skipped every input except the first.

Copper warms as it carries current. Whether that warmth can escape is the design question. A cable clipped to a surface in moving air sheds it easily. The same cable buried in a chase, sleeved in conduit, or lying against thermal insulation sheds it slowly. Nothing about the conductor has changed. What changed is how much it can carry before its own insulation begins to age.

Bundling makes that worse in a way nobody looks for. Several circuits run together in one trunking each warm the space the others occupy. The group as a whole then carries less than any single member of it would alone. A renovation that consolidates cable routes for neatness can quietly reduce what every run in that bundle is good for.

Ambient heat is the input Singapore supplies for free. A ceiling void under a flat roof, a service riser, a run along an unshaded ledge: each of those sits warmer than the room does. Warmer surroundings leave the copper less margin before it reaches the same limit. A cable specified against a temperate assumption has been specified against the wrong climate.

Equipment manuals know none of this, and they say so plainly. Daikin's own installation manual for the R32 split series covers models FTXM35K3V1B, FTXM42K3V1B and FTXM50K3V1B. It requires that electrical work be performed in accordance with relevant local and national regulations, and with the manual's own instructions. It also requires a dedicated power supply circuit, and that the specified wires are used. What no manual can supply is the route, the grouping and the ambient conditions in one particular flat.

A cable can be thick enough to avoid overheating and still be too thin to do its job. Copper has resistance. Current pushed through resistance loses voltage on the way, so the machine at the far end sees less than the board does. That loss grows with the length of the run and with the current being drawn.

Thermal capacity and delivered voltage are separate tests, and a run can pass the first while failing the second. The code treats them that way. SS 638's cable appendix is titled current-carrying capacity and voltage drop for cables, holding both properties of one conductor in a single place because both have to be satisfied. Sizing up a long run is therefore not generosity. It is the second test doing its work.

A compressor feels this more than any other household load. It draws hardest in the instant it starts, which is precisely when a sagging supply has the least to give. Whether a soft starter belongs on that machine gets settled elsewhere, on its own evidence. What belongs here is simpler: a long, marginal run can make a healthy system behave badly, and the system gets blamed for it.

The limit itself is a number, and it belongs in the code and not on this page. SS 638 is a paid standard. Its front matter and contents are public, which is where the appendix title above comes from. The provisions are not public, and any figure quoted without that document open has come from memory or from somebody else's blog.

An undersized conductor performs on installation day. It carries the current, the room cools, and the handover shows that the system runs. What the handover cannot show is anything about the cable, because nothing on that day asks the cable a hard question.

The evidence that would settle it sits out of the household's reach. The conductor sits behind plaster, inside trunking, or above a ceiling nobody opens. Its temperature is invisible. No appliance reports it, no bill separates it, and no symptom in the room names it. A homeowner has no instrument here, which is a fact of the installation and not a failing.

What surfaces later is a pattern instead of an event. Trips that cluster in the hottest stretch of the year and stay away through the rest of it. Behaviour that follows the weather more closely than it follows the machine. Those reports get read as an ageing aircon, because the aircon is the visible thing, so the money goes there first.

Nothing observed from inside the room separates the two candidates. A machine near the end of its life and a machine on a marginal supply produce reports that read alike from a sofa. Settling it means measuring at the board and at the equipment, with the load actually running. So the useful thing to hand over is a note of when the trouble appears, not a theory about which part failed.

One shortcut then turns a hidden weakness into a live one. The conductor and its protective device were chosen as a pair, so the aircon breaker rating describes the cable sitting behind it. Fitting a larger device after repeated trips leaves the conductor exactly where it was and raises the level at which anything intervenes at all.

Aircon systems get replaced and circuits do not. A replacement lands on a conductor chosen for the machine that came out, under conditions that may have moved since. More indoor units on one condenser, a larger outdoor unit, or another circuit added to the same trunking in a later renovation all change the arithmetic without anyone touching the cable.

Replacement day is the natural moment to re-ask the question. It is also the one that gets skipped. The old system worked, so the supply is assumed to have been adequate, and the new one goes onto the same run. That assumption is reasonable and it remains an assumption. It carries forward whatever was decided the first time, including a decision that was never really made.

What an owner can establish is narrower than a size and more useful than one. Not what the conductor ought to be, but whether anybody worked it out. A circuit that was designed leaves evidence behind it, because someone specified it, tested it and put a name to it. A circuit that was assumed leaves a working aircon and nothing else.

Singapore makes that evidence easy to ask for. EMA tells consumers that electrical work, at home or in the office, should be carried out or supervised by a licensed electrical worker. Installation, repair or modification of wiring falls inside that. The work then has to be tested and certified safe for use, by a licence holder of the appropriate class. EMA also says to check the worker's card before any of it starts.

This page will not produce a size, and that is deliberate. Sizing a circuit is design work, done against the code with the route and the conditions in front of the person doing it. A contractor who names a cable size from a photograph and a model number is guessing at four of the five inputs. The question worth putting to a quotation is not what size, but who worked it out and against what.