Voltage drop is the calculation most likely to decide your cable size. Current carrying capacity picks the minimum conductor that survives; voltage drop is what forces it bigger, especially on long runs. The method is simple, the limits are strict, and there is one correction hiding in Appendix 4 that most designers never apply, which means most designs are quietly more conservative than they need to be.
The limits
Appendix 4, section 6.4 sets them. For an installation supplied from the public network: 3 percent for lighting, 5 percent for everything else. From a private supply: 6 and 8 percent. At 230 V those public limits are 6.9 V and 11.5 V, and they apply from the origin of the installation to the point of use, so a final circuit does not get the whole allowance if the sub main feeding it has already spent some.
The method
The tables give a millivolt per amp per metre figure for each conductor size, and the arithmetic is one line:
Voltage drop in volts = (mV/A/m × Ib × length) ÷ 1000
Worked example. A 20 A single phase radial in 2.5 mm² two core 70 degree thermoplastic cable, 25 m long. Table 4D2B gives 18 mV/A/m.
Vd = 18 × 20 × 25 ÷ 1000 = 9.0 V, which is 3.9 percent.
That circuit passes for power at 5 percent and fails for lighting at 3 percent, on the same cable, at the same length. The load type decides the verdict, which is why "it is only lighting" is precisely the circuit to check rather than skip.
Two complications the tables hand you
Three phase values are line to line. The three phase mV/A/m columns give drop between lines. Divide by the square root of three when you need the line to neutral figure, for instance when checking the actual voltage at a point of use.
Large conductors split into resistance and reactance. From 25 mm² the tables stop giving one number and give three: r, x, and z. At these sizes reactance stops being negligible, and at low power factor the honest drop is r times cos phi plus x times sin phi, not the z column. A 70 mm² three phase cable is 0.55 r against 0.14 x, so a poor power factor load genuinely shifts the answer.
The correction most designers skip
The tabulated mV/A/m values assume the conductor is at its maximum operating temperature, 70 or 90 degrees, because resistance rises with heat. But a cable only reaches that temperature at full rated current. A cable loaded to half its corrected capacity runs cooler, its resistance is lower, and its true voltage drop is smaller than the table says.
Appendix 4 provides the correction, the factor Ct, and it is a gift that runs entirely in the designer's favour: applying it never makes a compliant design fail, it makes marginal designs pass honestly. On a lightly loaded buried cable the overstatement can run well past 10 percent, which at the margin is the difference between the cable you have and the size above it.
Almost nobody applies it by hand, because the formula wants the load ratio, the correction factors, and the operating temperature all at once, and that is exactly the kind of arithmetic that belongs in software rather than a margin note.
A detail from implementing this
Ct is also where CurrentProof caught its own most instructive defect. Our implementation mirrored a formula that had been built by pattern matching an earlier function, and a post release audit flagged it as the highest risk item in the codebase for precisely that reason: it had been copied, not independently checked. Checked against the IET Design Guide's own equation, it was wrong. Two of the four squared correction terms were missing from the bracket.
The error only surfaced for buried cables, where the missing terms are not equal to one, and it ran in the conservative direction, overstating drop by up to 15 percent in the tested cases. So no unsafe design ever left the tool. But conservative and correct are not the same thing, and the fix is live in every calculator. The lesson travels: any formula that was copied from somewhere, in software or in a spreadsheet, has inherited its source's errors plus whatever the copying added.