First published on the CurrentProof blog. CurrentProof is a browser based BS 7671 calculation platform built by an electrical designer. This article reflects BS 7671:2018+A4:2026.
A sub main that fails its voltage drop check by half a volt on paper usually gets a size up. On a 100 m run of four core SWA that is the difference between 70 mm² and 95 mm², and at current copper prices nobody needs me to price that out. What is less well known is that the number which failed was never the real voltage drop. BS 7671 says so itself, in terms: direct use of the tabulated values "may lead to pessimistically high calculated values of voltage drop or, in other words, to unnecessarily low values of permitted circuit lengths". That sentence sits at the top of Appendix 4, Section 6, and the same section then hands you 2 corrections that sharpen the figure. Most designers never use them. This article is about what they are, when they apply, and how much they are worth.
What the tabulated mV/A/m actually assumes
Appendix 4, Section 6 states the basis of the voltage drop tables plainly. The tabulated values are per ampere per metre of route, they represent the drop in all the circuit conductors together, and for three phase circuits they relate to the line voltage with balanced conditions assumed. The AC values apply between 49 and 61 Hz, and for single core armoured cables they assume the armour is bonded to earth at both ends.
The assumption that matters here is thermal: "the values of voltage drop assume that the conductors are at their maximum permitted normal operating temperature". For a 70 °C thermoplastic cable the table is priced at 70 °C, full stop. But a conductor only reaches its maximum permitted temperature when it is carrying its full effective current carrying capacity. A circuit whose design current is comfortably below that runs cooler, and a cooler copper conductor has a lower resistance. The table does not know your loading. You do.
Why large cables get a second column
For conductors of 16 mm² or less the inductance can be ignored and a single resistive mV/A/m value is tabulated. Above 16 mm² the tables split into 3 columns: the resistive component (mV/A/m)r, the reactive component (mV/A/m)x, and the impedance value (mV/A/m)z.
The reason is visible in the numbers. In Table 4D4B, the three phase columns for multicore armoured 70 °C thermoplastic cable, the resistive component falls from 1.50 at 25 mm² to 0.100 at 400 mm², a factor of 15. The reactive component across the same range barely moves, from 0.145 to 0.125. Resistance scales with cross sectional area and reactance mostly does not, so somewhere in the hundreds of square millimetres the cable stops being a resistor and starts being an inductor. Section 6.1 even tells you when to stop bothering with the thermal correction: where x/r is 3 or more, the resistive part is so small that correcting it is not worth the arithmetic.
Correction 1: your conductor is not at 70 °C
Section 6.1 gives a factor Ct for the operating temperature:
Ct = [230 + tp − (Ca²Cg²Cs²Cd² − Ib²/It²)(tp − 30)] / (230 + tp)
where tp is the maximum permitted normal operating temperature and the C factors are the rating factors you already applied for ambient temperature, grouping, soil resistivity and depth of burial. The term Ib²/It² is the one doing the work: the further your design current sits below the tabulated capacity, the cooler the conductor and the smaller Ct gets. The note under the equation records the physics, an approximate resistance temperature coefficient of 0.004 per °C at 20 °C, which holds for both copper and aluminium.
Two conditions are attached in one sentence, and they matter. The equation applies only where the overcurrent protective device is other than a BS 3036 fuse, and where the actual ambient temperature is equal to or greater than 30 °C. And for conductors above 16 mm², Ct is applied to the resistive component only, because temperature moves resistance and not reactance.
Correction 2: your load is not resistive
The tabulated impedance value z is the vector magnitude of r and x. Using it directly amounts to assuming the worst case power factor, the one that lines the load angle up with the cable's own impedance angle. Section 6 notes that for the larger sizes this "leads to a calculated value of the voltage drop higher than the actual value".
Section 6.2 gives the sharper figure. At or below 16 mm² you simply multiply the tabulated value by cos Ø. Above 16 mm² the design value becomes:
cos Ø (tabulated (mV/A/m)r) + sin Ø (tabulated (mV/A/m)x)
One caveat is printed alongside: for single core cables in flat formation the tabulated values apply to the outer cables, and may underestimate the drop between an outer and the centre cable above 240 mm² at power factors over 0.8.
Section 6.3 then combines both corrections. For the large sizes the design figure is Ct cos Ø (mV/A/m)r + sin Ø (mV/A/m)x.
The worked example, and what the corrections are worth
Take a 400 V three phase sub main: 70 mm² four core SWA, 70 °C thermoplastic, on perforated tray, which is Reference Method E. Table 4D4A gives It = 207 A. Design current Ib = 150 A at power factor 0.8, route length 100 m, no grouping, 30 °C ambient. The designer has allowed 2% of nominal, 8 V, for the sub main, leaving the rest of the Table 4Ab allowance for the final circuits.
The naive method takes z = 0.57 from Table 4D4B: 0.57 × 150 × 100 / 1000 = 8.55 V. That is 2.14% and the check fails. The next size up is 95 mm².
Now apply Section 6.3. Power factor first: 0.8 × 0.55 + 0.6 × 0.140 = 0.524 mV/A/m. Then temperature: Ct = [300 − (1 − 150²/207²)(40)] / 300 = 0.937. The design figure becomes 0.937 × 0.8 × 0.55 + 0.6 × 0.140 = 0.496 mV/A/m, and the drop is 0.496 × 150 × 100 / 1000 = 7.44 V, which is 1.86%. The same cable passes with margin. The corrections took 13% off the computed drop, and they move the maximum compliant run at this loading from 94 m to 107 m. Nothing was fudged: every step is printed in the appendix.
Where the limit itself comes from
The limit you are checking against is Regulation 525. Its structure is worth knowing. Regulation 525.1 requires the voltage at the terminals of fixed current using equipment to stay above the lower limit of the relevant product standard, and 525.202 deems that satisfied if the drop from the origin to the socket outlet or equipment terminals does not exceed Appendix 4, Section 6.4. Table 4Ab there gives 3% for lighting and 5% for other uses where the installation is supplied directly from a public LV network, and 6% and 8% where it is supplied from a private supply. For wiring systems longer than 100 m, the limits may be increased by 0.005% per metre beyond 100 m, up to an extra 0.5%. The calculated drop should include any effects of harmonic currents, and 525.203 permits a greater drop during motor starting and for high inrush equipment, provided the voltage stays within the equipment standard's limits.
Note what the 2% in my example was not: it was a design allocation, not a regulation. Table 4Ab limits the whole path from origin to load point, and splitting it between sub main and final circuit is the designer's call.
A detail from implementing this
CurrentProof's engine applies the Section 6.1 temperature correction, and building it taught me that the formula is the easy half. The two conditions attached to Appendix 4 Equation 6 are the part that bites. Because Ct only ever reduces the computed drop, applying it somewhere the standard excludes it is an error in the unsafe direction, and a review of our own implementation found exactly that: one calculation path could apply the correction with a BS 3036 fuse selected, because the gate was driven by a separate control rather than by the selected device. The fix was wiring the condition to the device itself. The general lesson holds for spreadsheets too. If your voltage drop sheet multiplies by Ct, check it also knows what fuse you picked and what ambient you declared, because the sharper method is only compliant inside the conditions the appendix attaches to it.
If you want the tabulated and corrected figures side by side without building the sheet, the free calculators at currentproof.co.uk do this, and pricing for the full platform is £59 per seat at currentproof.co.uk/pricing.