Sizing a cable is two separate tests and the larger answer wins. The first is current-carrying capacity: can the cable carry the load without its insulation exceeding its rated temperature. The second is voltage drop, and on any long run it is the one that decides the size — 20 A over 40 m at 230 V within a 3 per cent limit needs 3.99 mm², so the next standard size up is 4 mm².
Voltage drop rarely trips anything. It just quietly degrades everything downstream, which is why it is the test people skip.
What does the drop formula say?
For single phase, drop is 2 × ρ × L × I ÷ A, where ρ is resistivity — 0.0172 Ω·mm²/m for copper — L is the one-way length, I the current and A the conductor area.
Sixteen amps over 30 m of 2.5 mm² copper drops 6.6 V, which is 2.9 per cent of a 230 V supply and just inside the usual limit. Run the same circuit 45 m and it is 4.3 per cent, which is not.
The length is doubled because the current goes out and comes back: the drop happens in both conductors, so the resistance that matters is twice the one-way figure. Three-phase uses √3 rather than 2, because the return path is shared between phases.
What are the limits and why?
Three per cent for lighting and five per cent for power are the usual figures, and they exist because equipment is designed to tolerate roughly that much and no more.
- Lamps run dim, and on filament and some LED drivers the colour shifts as well.
- Motors run hot and draw more current to make the same torque, which increases the drop further.
- Heating elements deliver less than their rating, because power falls with the square of voltage — a 5 per cent drop costs about 10 per cent of the output.
- Electronics mostly cope, until the supply reaches the bottom of its input range and starts resetting under load.
On low-voltage systems the percentage is what bites. Two volts lost on a 230 V supply is under 1 per cent; the same two volts on 12 V is 17 per cent, and the load simply does not work properly.
Why is my calculated size not a real cable?
Because conductors come in a preferred-number series and the formula produces a continuous answer. A result of 3.99 mm² means 4 mm², and the 4 mm² cable then drops 2.99 per cent — just inside the limit rather than exactly on it.
Always round up. Rounding down to the nearest available size is how a design that calculated as compliant becomes an installation that is not.
How do AWG and metric sizes compare?
They interleave rather than match, because AWG is a geometric series and metric sizes are a preferred-number one. AWG diameter is 0.127 × 92^((36−n)/39) millimetres, which makes the scale run backwards — a larger number is a thinner wire — and every three gauges roughly doubles or halves the area.
| AWG | Diameter | Area | Nearest metric |
|---|---|---|---|
| 16 | 1.291 mm | 1.31 mm² | 1.5 mm² |
| 14 | 1.628 mm | 2.08 mm² | 2.5 mm² |
| 12 | 2.053 mm | 3.31 mm² | 4 mm² |
| 10 | 2.588 mm | 5.26 mm² | 6 mm² |
The metric size in the last column is always the larger of the pair, which makes substituting metric for AWG the safe direction and the reverse a downgrade. A calculated result of "AWG 15.5" is real information: the metric cable sits between two gauges and neither is an exact replacement.
Does the same thinking apply on a circuit board?
Yes, with a different formula. IPC-2221 sizes a trace from its cross-sectional area: A = (I ÷ (k × ΔT^0.44))^(1/0.725), where k is 0.048 for an external trace and 0.024 for an internal one. Two amps on an external 1 oz layer at a 10 °C rise needs about 33 mil, or 0.84 mm.
Internal traces need to be roughly twice as wide for the same current, because they are surrounded by board material rather than air and cannot shed heat as easily. The temperature rise is a design choice rather than a limit — 20 °C above ambient is entirely normal — and what the formula does not capture is context: a trace running past a heat-sensitive part wants a lower rise than the number suggests.
Questions people ask
Does voltage drop waste energy? Yes. The lost voltage times the current is power dissipated as heat in the cable, which is why an undersized run is both a performance and a running-cost problem.
Is aluminium worse than copper? Higher resistivity, so an aluminium conductor needs roughly 1.6 times the area for the same drop. It is lighter and cheaper, which is why it appears in large feeders and service cables.
Does a longer run always need thicker cable? For the drop test, yes — drop is linear in length. The capacity test does not change with length at all, which is why the two tests diverge as runs get longer.
Can I use this for fixed wiring in my house? As a design check, yes. As the basis for an installation, no — capacity derating for grouping, ambient temperature and installation method is a substantial part of the real calculation, and fixed wiring is regulated work.
Run both tests and take the bigger cable. The voltage drop calculator works the drop for a given size, the cable size calculator inverts it to find the minimum area, the wire gauge calculator, AWG to mm and mm to AWG handle the two size systems, and the PCB trace width calculator applies the same idea to copper on a board.