Solar Cable Installation: Bend Radius, UV, MC4 & Test Rules

01Where a PV cable actually fails

Modules carry 25-year performance warranties. Inverters carry 10–12-year warranties that buyers routinely extend. The DC cable that ties them together is specified once, buried in a tray or clipped under a rail, and then expected to disappear for a quarter of a century. When it does not, the failure is rarely mysterious in hindsight: a bend that was too tight, a jacket that chalked and crazed, a crimp that was made with the wrong die, or a test that was never run because the schedule was tight.

What makes cable failures expensive is not the cable. It is the access cost of finding and fixing one — removing modules, opening a tray, and losing a string's output for weeks. That asymmetry is the whole argument for getting the installation details right the first time.

A word on scope. This article assumes single-conductor PV cable built to EN 50618 with the harmonised designation H1Z2Z2-K, or the international equivalent IEC 62930. If you are still deciding between the two standard families, that comparison has its own article on the IEC 62930 vs EN 50618 page. For the earlier TÜV 2 PfG 1169 / PV1-F generation, see H1Z2Z2-K vs PV1-F.

024mm solar cable bending radius: the number everyone quotes and nobody agrees on

Bend radius is the tightest curve a cable can take without damaging the insulation or distorting the conductor geometry. The damage is not visible at the time: a too-tight bend leaves the outer copper strands stretched, the inner strands compressed, and the insulation under permanent tensile stress at the outer face. Nothing trips. Nothing reads wrong on a commissioning megger test. The joint simply has a head start on 25 years of thermal cycling.

What the published datasheets actually say

Here we have to be honest about something the industry usually glosses over. There is no single agreed figure. Three distinct positions appear in circulation, and we found no standard text that settles it cleanly:

Sources listed in the verification log. Note that four of the five positions are manufacturer reproductions rather than standard text we could read directly.

Tracker runs and moving cable

Everything above concerns fixed installation. Where the cable articulates — single-axis trackers being the common case — the constraint is fatigue, not a single bend. The cable flexes through a full range twice a day for 25 years, which is roughly 18,000 cycles. Two practical consequences follow. First, the radius has to be measured through the full travel of the tracker, not at the rest position. Second, the flexing length should be a free, unsupported loop with strain relief at both ends; a loop that is clipped at both ends and short converts tracker rotation into tensile load on the conductor.

If the harness is long and repetitive, factory-terminated assemblies are worth pricing against field assembly — not because field crews cannot crimp, but because the failure mode moves from "operator-dependent" to "machine-verified". We cover that trade-off in section 05.

Cold-weather installation

PV cable is tested for low-temperature behaviour rather than given a cold-weather bend factor. EN 50618-referenced test lists include cold bending, cold elongation and cold impact at −40 °C to EN 60811-504, -505 and -506. Separately, manufacturers publish a minimum handling and installation temperature, typically −25 °C.

03UV protection: what EN 50618 actually requires

UV degrades polymers by cleaving chains at the surface, which shows up first as loss of elongation, then as surface crazing, then as cracking that reaches the conductor. The cable does not fail electrically the day the jacket cracks — it fails the first time water sits in that crack.

The test, and the number that matters

EN 50618 Annex E is normative, and it is specific. Specimens are prepared to EN 60811-501, exposed to UV light per EN 50289-4-17 method A for 720 hours (360 cycles), conditioned at ambient for at least 16 hours, then tested for tensile strength and elongation at break against five unexposed control specimens. The acceptance criterion: values after exposure must be at least 70% of the unexposed values.

Two related properties are worth specifying alongside it, because they are tested separately and fail separately: ozone resistance per EN 50396 type B, and the damp-heat test at 1,000 hours, 90 °C and 85% relative humidity per EN 60068-2-78. A jacket that passes UV but fails damp heat is a real possibility in tropical coastal sites.

What we could not verify

Carbon black is the standard UV stabiliser for black cable jackets, and it is common to see a specific content figure quoted. The previous version of this article cited 2.6% ± 0.25% from GB/T 15065-2009. We tried to verify that as a PV cable requirement and could not: GB/T 15065 covers black polyethylene compounds for wire and cable generally, not the cross-linked halogen-free polyolefin sheath used on H1Z2Z2-K, and we found no carbon-black content requirement in EN 50618 or IEC 62930. We have removed the figure rather than repeat it. If your specification needs a quantified stabiliser content, ask the cable maker for the compound formulation data — do not accept a number from a blog post, including this one.

04How to crimp PV cable: the MC4 connector, honestly

Connector failures are the most-discussed PV failure mode and the easiest to get wrong, so it is worth being precise about what the evidence supports. A peer-reviewed fault-tree analysis of rooftop PV fires — Mohd Nizam Ong and colleagues, Journal of Building Engineering 46:103752, 2022 — found that the connector was the prime contributor in 17% of PV-related fires, ahead of the module, isolator and inverter at component level, with 33% of incidents attributed to unknown or unrelated ignition sources.

A separate line of reporting puts the figure higher. An academic volume citing the IEA's 2014 report states that PV connectors alone may be responsible for up to one-third of PV module fire incidents. We are presenting both rather than choosing: 17% is the peer-reviewed figure, one-third is a single-source attribution.

Crimping

The crimp is the single step that decides the outcome, and it is the step most often done with whatever tool is on the truck. The published field thresholds are worth knowing because they turn a visual judgement into a measurement:

• Use the connector maker's tool and die. Generic crimpers and wrong die sets produce either an under-compressed crimp (high contact resistance, heat) or an over-compressed one (severed strands).
• Pull-out force. At least 310 N for a 4 mm² conductor is the figure published by one connector supplier; confirm against your connector's installation instructions, because this number is model-specific.
• Contact resistance. Above 0.5 mΩ is suspect; above 1 mΩ is a failed joint. For reference, Stäubli publishes a connector contact resistance of 0.25 mΩ for MC4 and 0.2 mΩ for MC4-Evo 2.
• Gland nut torque. Typically 2.5–3 N·m. Under-tightened glands are the usual path for water ingress; over-tightened ones deform the seal.

Mating, and the one thing not to do

Stäubli's own instructions are unusually explicit: mate until you hear and feel the click, then verify by pulling gently on the connector with a maximum force of 20 N. That is the whole check. A connector that does not click has not latched and can separate under thermal cycling.

The one thing not to do is mix brands. IEC 62852 defines test requirements for a connector tested against its own mating counterpart; it does not make different manufacturers' products interchangeable. Pin diameter, contact spring force and seal geometry differ by fractions of a millimetre between brands, which is enough to reduce contact force and create a resistance hotspot at operating current. Specify one brand for the whole project, and do not substitute on partial shipments. More on this is in the MC4 connector compatibility guide.

Values from Stäubli product documentation for the stated models. Connector ratings are model-specific and change between product generations — the datasheet for the exact part number you are buying is the controlling document.

Factory-terminated versus field-crimped

The trade-off is real and it is not only about labour. Factory termination moves crimp force control from operator-dependent to machine-verified, and it makes 100% pull-out and insulation-resistance testing economical in a way that is simply not practical on a roof. Field assembly wins on flexibility: you cut to the length you discover on the day, and you are not exposed to a measurement error made during design.

Our position: for fixed-geometry residential strings where the run lengths are known before mobilisation, factory-terminated harnesses remove the dominant failure mode. For trackers, retrofits and any array where runs are determined on site, field assembly is right — but then the crimp tool, the pull test and the documented result are not optional.

05IEC 62446-1 test sequence: the corrected clause map

This is the section where the earlier version of this article was most wrong, and the correction matters because people copy clause numbers into commissioning documents. IEC 62446-1:2016 Clause 6 sets out the Category 1 test procedures for grid-connected systems. The actual clause order is:

Clause titles from the BSI and MADCAD tables of contents for BS EN 62446-1:2016+A1:2018. Clause 7 covers Category 2 testing (string I-V curves, infrared inspection); Clause 8 covers additional tests including the wet insulation resistance test at 8.3.

The two tolerance figures worth memorising

For strings, the standard's own language is "typically within 5%": measured V_oc should match the expected value within about 5% (Clause 6.4), and under stable irradiance individual string currents should agree with each other within about 5% of the average (Clause 6.5.2). Both are fault-detection thresholds, not performance guarantees — the standard says so explicitly, and points to I-V curve testing under Category 2 when you want performance.

Insulation resistance: test voltage and minimum value

This is the other place where the old article was wrong in a way that could damage equipment. The test voltage is set by system voltage, and it does not scale the way most people assume:

Rows 1–3 from Fluke and InsulationTesting, which agree. The 1,500 V row for systems above 1,000 V appears in Fluke's guidance only — graded [B]. Never apply a test voltage above the maximum system voltage on the module datasheet.

More detail on this specific test, including instrument requirements, is on the insulation resistance testing page.

06How long can solar cable be?

This is one of the most-searched questions in the category and the honest answer is: as long as your voltage drop allows. There is no length limit in the cable standard. The limit comes from the percentage of the string voltage you are willing to lose to the conductors, which is a function of cross-section, current and one-way distance.

For a typical 10 kW array at 13–14 A per string, the practical ceiling is around 60–80 m on 6 mm² and 100–120 m on 10 mm² before you exceed a 2% drop. Rather than carry a rule of thumb, use the calculation: the worked tables for 20 m through 100 m are in our companion article on 10kW solar cable sizing and voltage drop, and the general method for 600 V and 1500 V systems is in the DC cable sizing guide. If the run is buried or in conduit, check the burial and conduit requirements too.

07Pre-installation and pre-energisation checklist

Before you pull

• Confirm the cable designation. H1Z2Z2-K or PV1-F, correct cross-section, and a sheath marking that carries the standard reference.
• Get the OD and the bend radius off the datasheet and tell the crew the number in millimetres, not as a multiple. "30 mm minimum" is enforceable on a roof; "4 × OD" is not.
• Inspect the jacket before pulling. Transport damage is the cheapest defect to find and the most expensive to find later.
• Use rollers or a feed tube at roof edges, tray transitions and conduit entries — the three places where a bend gets tightened by the installation method rather than the design.
• Check the reel temperature against the manufacturer's minimum installation temperature, typically −25 °C.

Before you energise

• Confirm the crimp tool matches the connector and the die matches the conductor size in use.
• Pull-test every crimped connector and record it. Sample testing is not enough when the failure mode is operator-dependent.
• Verify the click on every mated pair with a 20 N pull check.
• Confirm one connector brand throughout — no substitutions across partial shipments.
• Run the IEC 62446-1 Category 1 sequence in the clause order in Table 3, and record actual values, not pass/fail ticks.
• Label both ends of every string cable before routing, and photograph the tray and termination work before it is covered.

08What this article does not claim

We have tried to be specific about the limits of what we know. To be explicit:

• We did not find a single authoritative minimum bend radius for PV cable. Three published positions are presented side by side and we did not pick a winner.
• We could not verify a quantified carbon-black content requirement in EN 50618 or IEC 62930, and removed the figure the previous version carried.
• The crimp pull-force, contact-resistance and gland-torque figures come from one industry source, not from a standard text we could read. They are labelled as such.
• The 310 N, 0.5/1 mΩ and 2.5–3 N·m thresholds are connector-model-specific. Your connector maker's instructions override this article.
• None of this is legal or compliance advice. The authority having jurisdiction, the module datasheet and the inverter manual outrank any checklist published here.