Wire count on an IO-Link cordset is set by the port class on the master and the current the device actually draws — not by the sensor brand, and not by what the last project used. Three conductors are the baseline the standard is built on; the fourth and fifth conductors exist for pin assignment and auxiliary power. Shielding is an EMC decision about the installation, not a default line on the drawing. And the maximum run length is something you verify against the current edition of the standard plus your master's datasheet — this page will not quote a distance figure it cannot trace to a primary source, and neither should your RFQ.
Most selection pages stop at connector families and a "use shielded cable" checklist line. The table below is the decision artifact: each conductor count mapped to the port class it serves, the device type it fits, the cost it adds and the failure it prevents.
The selection table: four parameters decide most IO-Link cordsets
| Cordset | Port class it fits | Typical device / use | Shield default | Cost driver | Failure if mis-specified |
|---|---|---|---|---|---|
| 3-wire | Class A, where the device is powered from the port and no auxiliary rail exists | Simple sensors and actuators on the standard 3-wire connection | No — unshielded is the baseline described by the standard | Lowest conductor count | Cannot carry auxiliary power; unusable on Class B devices |
| 4-wire / 4-pin | Class A (4-pin M8 or M12, single 24 V system supply) | Sensors and low-power devices; fourth conductor often a digital signal | No, unless the EMC conditions below apply | Baseline; the most common over-spec is adding a shield here | Fitted to a Class B port: actuator starves or drops out under load |
| 5-wire / 5-pin | Class B (5-pin M8 or M12 with auxiliary supply) | Valve islands, high-power actuators, devices needing a separate voltage | No by default; decide from the EMC conditions | Extra conductor plus, often, a larger cross-section | Fitted to a Class A port: auxiliary conductor carries nothing — paid for, unused, and a drawing error waiting to be copied |
Read it in this order: port class first, device current second, conductor count third, shield last. Reversing that order is how a 5-pin shielded assembly ends up on a 4-pin sensor port.
Scope note: M12 coding schemes (A/B/D/X) are a separate topic and are not re-taught here. This page stays on the electrical, distance and EMC constraints — the part that decides whether the machine commissions cleanly.
Wire count: what Class A and Class B ports actually demand
IO-Link is standardised under IEC 61131-9 and built on the established three-wire sensor and actuator connection, with no additional requirements on cable material, using M12, M8 or M5 connectors (IO-Link Consortium, io-link.com). That is the floor, not the whole picture — the port class on the master determines what the cordset has to carry.
- Class A: 4-pin M8 or M12 connectors, 24 V single system supply, roughly 200–500 mA current capacity through pins 1 and 3 (ifm technology page). The IO-Link Design Guideline likewise works from a 200 mA-per-port Class A supply assumption, and permits M5, M8 or M12 plugs for IP67-rated devices.
- Class B: 5-pin M8 or M12 connectors with approximately 2 A capacity and an auxiliary power supply, intended for high-power actuators such as valve islands (ifm).
Where the extra conductor goes is the single most common drawing error. Typical IO-Link master-to-device links are described as 3-wire or 5-wire unshielded cable with four- or five-pin connectors, and the remaining pins are vendor-dependent — used for digital signals or additional power supply depending on the device (KEB America). Two consequences follow:
| Configuration | What happens electrically | What the buyer sees |
|---|---|---|
| 5-pin cordset on a Class A port | Auxiliary conductor is unused; pin assignment may not match the device | Extra cost per drop, plus a pin-out mismatch found at commissioning |
| 4-pin cordset on a Class B port | No auxiliary rail reaches the actuator | Actuator starves or drops out under load, typically after a few cycles |
| 3-wire cordset on either port | Viable only where the device is fully powered from the port and no auxiliary rail or digital signal is needed | Works on the bench, fails once the device is loaded or a second function is enabled |
Citation trap worth naming: secondary sources sometimes attribute IO-Link pin assignment to an unrelated standard number. Take pin assignment from the IO-Link specification and the specific device datasheet — never from a blog, and never from the master's drawing if the device datasheet disagrees.
Buyer action before the conductor count is fixed: pull the master's port-class table and the device datasheet's current consumption, and reconcile them per drop. Do not inherit the previous project's cordset because it "worked".
Run length: verify your 20 m assumption instead of trusting it
The maximum cable length between a master port and its device is bounded by the IO-Link specification. Confirm the figure in the current edition of IEC 61131-9 — the standard IO-Link is published under (io-link.com) — and cross-check it against the specific master's datasheet before it goes on the drawing. Catalogue pages and distributor filters quote "standard" lengths; a catalogue length is not a specification limit.
This page deliberately does not assert a number. If a supplier quotes a maximum run without an edition of the standard or a master datasheet behind it, treat it as unverified.
The check you run yourself, per drop:
- Sum cordset length + any in-line extension + the device pigtail. The pigtail counts; buyers routinely leave it out.
- Compare the total against the port's stated limit in the master datasheet, not against the cable's "standard length" listing.
- If the total is close to the limit, re-check the conductor cross-section against the device's stated consumption — an undersized conductor turns a compliant run into a voltage-drop problem at the far end.
- Record the answer on the drawing, with the master datasheet revision you checked against.
Distinguish the failure modes at commissioning. A run beyond the verified limit typically shows up as intermittent communication or device dropouts that worsen under start-up load. An EMC problem tends to correlate with a drive or welder switching on, not with load cycles. Different fault, different fix — one needs an architecture change, the other a routing or bonding change.
Where the distance genuinely cannot be met, change the architecture rather than the cordset: relocate the remote I/O hub closer to the device, or use IO-Link Wireless, which removes the physical cable while maintaining real-time communication (io-link.com). A longer cordset is not the answer.
Shielding: three conditions where it earns its cost, two where it does not
Start from the standard's own position: IO-Link's point-to-point communication is based on the long-established three-wire sensor and actuator connection without additional requirements for the cable material (io-link.com), and the typical master-to-device link is described as unshielded (KEB America). A shield is therefore a response to an EMC condition, not a compliance requirement.
Specify a shield when:
- The cordset runs parallel to VFD motor leads or servo drive output cables in a shared duct or tray for any meaningful distance. This is a routing decision as much as a cable decision — separation often removes the need entirely.
- The installation sits in a welding cell or near drives switched at high dv/dt inside the same enclosure.
- The duty is drag-chain or continuous flex, where jacket abrasion and shield flex-fatigue become the reliability driver rather than EMC.
Where it does not pay:
- Short point-to-point runs to a sensor in a clean cell, with no adjacent power cabling and a panel that has already been EMC-verified. The unshielded cordset described by the standard is the lower-cost, tighter-bend-radius option.
- Anywhere the shield cannot be bonded correctly. A pigtail-bonded or unbonded shield is a worse antenna than no shield at all. A shield adds cost, capacitance and bend radius, and returns nothing if the termination is wrong.
Treat this as an installation question you answer in the design review, with the routing drawing in front of you — not a checkbox applied to every drop.
Connector checklist: M12 A-coded as the default, M8/M5 where space forces it
M12 A-coded is the standard device-side connector across the ecosystem; the standard relies on M12, M8 or M5 connectors (io-link.com), and the Design Guideline permits M5, M8 or M12 plugs for IP67-rated devices (IO-Link Design Guideline).
Do not conflate the three cable roles in one master-to-device installation: L-coded M12 carries power to the master, D-coded M12 carries Ethernet communication, and A-coded M12 or M8 connects IO-Link devices and standard sensors/actuators (Automation World). Three roles, three specifications — a single line item for "IO-Link cables" on the BOM is how mismatches get ordered.
M8 and M5 suit compact sensors and tight manifold spacing, but the smaller body limits conductor cross-section and therefore current. M8 is a poor fit for Class B auxiliary power.
Drawing fields to complete before the RFQ goes out: connector family and coding; gender; pin count; pin assignment printed on the drawing (typically pin 1 = 24 V, pin 3 = 0 V, pin 4 = communication/switching — confirm against the device, not the master); keying and orientation; coupling nut material; moulded versus field-attachable; length; shield yes/no with a stated justification.
Mechanical and environmental specs that fail after the warranty conversation
- IP rating chain: an assembly is only as good as its weakest joint. An IP67 plug on an IP65 receptacle does not deliver an IP67 circuit. Require a rating statement per mated pair.
- Flex and torsion: ask for the test basis the vendor claims — bend radius, cycle count, torsion angle — and request the test report rather than accepting the datasheet claim.
- Jacket material: PUR where coolant, oil mist or continuous flex is present; PVC for static indoor runs only. Washdown and food-processing zones deserve a separate review.
- Temperature and weld-spatter exposure: confirm at the mated-pair level; the connector body often derates before the cable does.
Where over-specification quietly costs you money
Shielded 5-pin assemblies bought by default for 4-pin Class A sensor ports are the two most common over-specs — and both add cost on every drop, multiplied across a machine.
| Cost driver | What moves the price | When it is justified |
|---|---|---|
| Conductor count | Extra conductor, sometimes a larger cross-section | Only when the port class or a vendor-specific digital function requires it |
| Shield | Foil/braid plus a drain and correct bonding | EMC condition present and bonding achievable |
| Jacket material | PUR versus PVC | Coolant, oil mist or continuous flex |
| Moulded vs field-attachable | Assembly labour and repeatability | Field-attachable only where length must be set on site |
| Custom vs catalogue length | Cutting, terminating, testing | When slack loops or extensions would otherwise be used |
The counter-cost matters in the design review: one intermittent communication fault at commissioning consumes more engineering hours than the per-unit saving across an entire machine's cordsets. That is the argument to make — not a claim of measured savings. Specify to the port and the environment first, then standardise across the machine to cut SKU count. The real saving sits in that standardisation, not in shaving a conductor.
Pre-RFQ confirmation checklist
- Master port class recorded per drop (Class A or Class B).
- Device current consumption taken from the device datasheet, not assumed.
- Pin assignment confirmed against the device, printed on the drawing.
- Conductor count derived from 1–3, not inherited from a previous project.
- Shield decision recorded as yes/no with the EMC justification written down.
- Run length totalled (cordset + extension + pigtail) and verified against the current standard edition and the master datasheet.
- Connector family, coding, gender and required IP rating per mated pair.
- Flex or static duty cycle, with bend radius and cycle count if flexing.
- Approvals required for the destination market and the panel's certification path.
Then order one assembly in the longest length and the worst-bend drop, install it, and verify communication before releasing the volume order.
Our position, stated plainly: we build to your drawing and pin-out. We do not certify a run length for your installation, and we will tell you when a specification needs a test we cannot perform in-house rather than sign it off.
Request an IO-Link cable sample or quote with pin-out and length confirmed. Send the port class per drop, the pin assignment, the total length and whether a shield is required — we will return a configuration and a sample built to that drawing.