Sizing the DC cable between PV modules and the inverter: the parts of IEC 62548 that get left out

A PV array is the one place in an installation where the cable calculation is done twice by different people and comes out differently both times. The electrical side is simple — direct current, no reactance, no harmonics, no neutral — and that simplicity is exactly what hides the five steps below. Each of them is written in IEC 62548:2016, each is skipped often enough to be worth naming, and each one moves the answer by a whole cable size or more.

The rule that decides everything

IEC 62548:2016, 7.3.7.1.1 sets out the criteria for a PV string cable, a PV sub-array cable and a PV array main cable:

• a) overcurrent protection ratings where in use,
• b) the minimum current rating, referring to Table 5,
• c) the voltage drop and prospective fault current.

Then one sentence: the largest cable size obtained from these criteria shall be applied. Not the average of them, not the one the engineer looked at first. Three separate calculations, and the worst governs.

That sentence is the reason the rest of this matters. A criterion that is skipped cannot govern, and a criterion that is computed with the wrong temperature or the wrong conductor class governs at the wrong number.

1. The cable near the modules is not at ambient temperature

The note under Table 5 is the single most consequential line in the clause:

The operating temperature of PV modules and consequently their associated wiring can be significantly higher than the ambient temperature. A minimum operating temperature equal to the maximum expected ambient temperature + 40 °C should be considered for cables installed near or in contact with PV modules.

Clause 7.3.7.2 says the same thing from the insulation side: PV modules frequently operate at temperatures of the order of 40 °C above ambient, and cable insulation of wiring installed in contact with or near PV modules shall be rated accordingly.

Forty kelvin. On a 45 °C site, a string cable clipped to a module frame is rated at 85 °C, not at 45 °C.

What that does to the current rating is not a rounding error. The ambient correction factor of IEC 60364-5-52 Table B.52.14 follows the form the table is built on,

f = √( (T_max − T_ambient) / (T_max − 30) )

which reproduces every printed value of the PVC and XLPE columns to within 0,005. For a 90 °C XLPE cable:

The rating falls to a third. A 6 mm² cable that carried 58 A on the table carries 17 A where it touches the module, and a string of four parallel unprotected strings needs 52 A. That is not a cable that runs warm; it is a cable that is three sizes short.

This is also the reason a 90 °C cable is the wrong product for a string. At a 120 °C conductor rating the same 85 °C gives f = 0,62 instead of 0,29 — the whole purpose of a PV-specific cable. Note that IEC 60364-5-52 Table B.52.14 carries no 120 °C column at all, so a 120 °C cable has to be rated either from the manufacturer's own tables or from the form above, stated as such.

The trap inside the trap: the +40 K applies to the part of the run that is near the modules, and a string cable is usually near them for its whole length while an array main is not. Sizing the whole route at ambient is wrong at the string; sizing the whole route at ambient + 40 K is merely expensive at the main. The route has to be split, and IEC 60364-5-52, 523.8 already says the conductor is selected for the part of the route with the most adverse conditions.

2. Table 5 has six cases, not one

Table 5 does not give "1,25 × I_SC". It gives a different rating for each combination of circuit level and protection, and the differences are large.

For a PV string with no string overcurrent protection:

• a single-string array: 1,25 × I_SC_MOD
• everything else: I_n + 1,25 × I_SC_MOD × (N_PO − 1)

where I_n is the rating of the nearest downstream overcurrent protection device and N_PO is the number of parallel connected strings protected by it. Note (b) is the case people miss: when no overcurrent protection is used anywhere in the array, I_n is replaced by zero and N_PO is every parallel string in the complete array.

Take a 13,9 A module and four parallel strings with no protection anywhere:

0 + 1,25 × 13,9 × (4 − 1) = 52,1 A

against the 17,4 A that 1,25 × I_SC_MOD would have given. A factor of three, and it comes from the physics rather than from a margin: a fault at one string is fed by every other string in parallel with it.

With a string fuse fitted, Table 5 rates the cable on the fuse, at I_n, and the 52 A disappears. Which is the engineering point of a string fuse — it is not there to protect the module, it is there to bound what the cable can ever be asked to carry.

For a PV sub-array with no sub-array protection, the rating is the greater of two things, and the first of them is the one that surprises people:

• a) I_n(array) + 1,25 × Σ I_SC of all the OTHER sub-arrays
• b) 1,25 × I_SC S-ARRAY of this sub-array

Branch (a) is current flowing into this sub-array from its neighbours and from whatever the array device lets through. With a 250 A array device and two other sub-arrays of 166,8 A each, branch (a) gives 667 A against branch (b)'s 208,5 A. The sub-array cable is sized on what the rest of the array can push into it, not on what it generates.

And a case that is entirely absent from most spreadsheets: where the modules are connected through DC conditioning units — optimisers, rapid-shutdown devices — Table 5 sends the reader to 5.1.5.2, and the rating becomes the greater of the DCU's own maximum output and 1,25 × I_SC_MOD.

One more line, printed under the table and almost never applied: where an inverter or other power conversion equipment can provide backfeed current into the array under fault conditions, that current shall be taken into account in all calculations of circuit current ratings, and in some circumstances it has to be added to the Table 5 rating. The backfeed current is a rating the inverter declares under IEC 62109-1. It is on the data sheet. It is rarely in the calculation.

3. The conductor is class 5, and class 5 is more resistive

Clause 7.3.7.2: where movement of the cable is expected, the conductor of the cable shall be flexible — class 5 of IEC 60228 — and the clause names the cases: string cables, trackers, and cables connected using plugs and sockets. In other words, most of the DC side of an array.

A class 5 conductor is made of finer wires than a class 2 stranded conductor, and IEC 60228 allows it a higher maximum resistance for the same nominal section:

A PV cable is tinned, so the figure that applies is 3,39 — 10 % above the class 2 value a generic cable table would give. On a 40 m string run that is 10 % more voltage drop than the calculation said, every year, for the life of the plant.

4. R₂₀ is a purchase specification, not an operating value

IEC 60228 gives a maximum resistance at 20 °C. The conductor of a string cable in an array field does not sit at 20 °C — it sits near its maximum operating temperature, which is the whole subject of point 1 above.

R_θ = R₂₀ × (1 + α × (θ − 20)) α(Cu) = 0,00393 /K

For the 6 mm² class 5 tinned conductor at 90 °C:

3,39 × (1 + 0,00393 × 70) = 3,39 × 1,2751 = 4,3226 Ω/km

That is 27,5 % more resistance than the catalogue number. Combined with point 3, a calculation done on class 2 at 20 °C understates the drop of a tinned class 5 conductor at 90 °C by a factor of 1,40.

5. A DC circuit is two conductors

There is no return through an earthed neutral. The positive and the negative conductor both carry the full current and both have resistance, so the loop is twice the route length:

ΔU = 2 × L × R_θ × I

The same 40 m string, 6 mm² class 5 tinned at 90 °C, carrying I_MP = 13,1 A on a 620 V string:

ΔU = 2 × 40/1000 × 4,3226 × 13,1 = 4,53 V = 0,73 %

Compute it as a single conductor at R₂₀ on class 2 figures and the same run reads 0,26 %. Same cable, same route, and a number that is almost three times too small.

Two smaller notes on the same subject:

• The current in the voltage drop is I_MP, not I_SC. The array operates at

the maximum power point; it does not operate in short circuit. Using I_SC here is conservative but it is not the operating condition, and on a long main it buys a cable size nobody needed.

• IEC 62548 sets no voltage drop limit. Clause 5.1.10 lists voltage drop in

cables among the factors that affect array performance, and stops there; the performance clause does not quantify it either. Whatever percentage a project adopts is the designer's decision, and a report that prints "compliant with IEC 62548, ΔU < 1 %" is claiming something the standard does not say.

What else the clause asks for, and what it does not

Two checks that belong to the cable and are cheap to make:

The voltage rating. 7.3.7.2 requires a voltage rating at or above the PV array maximum voltage determined in 7.2 — which is U_OC ARRAY corrected for the lowest expected operating temperature, either from the module manufacturer's coefficient or from the factors of Table 4. A 1000 V cable on an array whose corrected open-circuit voltage is 1150 V is a defect that no amount of current rating fixes.

The prospective fault current. 7.3.7.1.1 c) names it, and the reason it almost never governs is worth understanding rather than assuming. A PV array with no battery, no second source and no inverter backfeed is current limited: it cannot deliver much more than its own short-circuit current, so the adiabatic check passes at a section far below anything the other criteria allow. Where a battery is connected, or another array section can feed the fault, the check becomes real — and the factor k should then be computed for a live conductor starting at its operating temperature, from the normative formula of IEC 60364-5-54:2011 Annex A, and not taken from the tabulated protective- conductor values of Table A.54.2, which start at 30 °C. For copper going 90 → 250 °C the formula gives k = 143, where the table's 30 → 250 °C row prints

1. Using 176 for a live conductor understates the required section by 23 %.

And three things that are out of scope of any current calculation and still have to be done: the mechanical requirements of 7.3.7.3 (support against wind and snow fatigue, protection from sharp edges, the manufacturer's minimum bending radius, UV-resistant conduit and cable ties that outlive the plant), the segregation of AC and DC circuits of 7.3.8, and the connector requirements of 7.3.9 — which include, easy to forget, a current rating at or above the Table 5 rating of the circuit the connector is fitted to.

The worked case, end to end

Four parallel strings, no overcurrent protection anywhere, 13,9 A modules, 6 mm² class 5 tinned copper clipped to the module frames, two circuits bunched together, 45 °C site, 40 m run.

The same array with the cable routed clear of the modules needs 10 mm². The +40 K note is worth two cable sizes on this run, and it is one sentence in a footnote.

---

The calculator for this is on the site: it takes the module from a library of 14 694 records, applies each of the three criteria of 7.3.7.1.1 separately, shows the substitution for every number, and names which criterion governed.