Short answer: when SPE wins at the sensor, and when 4-pair still does
Pick Single Pair Ethernet when the field device is bandwidth-light (10BASE-T1L class), power-limited, and routed through tight or moving geometry where a smaller OD changes the build. Keep M12 X-coded 4-pair when the device needs higher throughput, familiar PoE practice, or a mature multi-vendor connector ecosystem. Two of those three answers come from datasheets, not marketing pages.
That is the whole decision in one paragraph. The rest of this article is about the work between reading it and releasing a BOM — because the SPE-versus-4-pair question rarely stalls on data rate. It stalls on a power budget measured at maximum ambient, and on whether the connectors at both ends of the link will actually mate.
Decision table: the criteria that actually change the BOM
Most comparison content stops at "SPE is smaller." Size is real, but it is not the row that kills a project. The table below is organised by where each choice commits you during validation — and by who carries the risk if the answer turns out wrong.
| Criterion | Single Pair Ethernet (10BASE-T1L class) | 4-pair Ethernet (M12 X-coded) | Who owns the risk |
|---|---|---|---|
| Conductor count | One twisted pair | Four pairs | Cable supplier confirms construction; device vendor confirms PHY |
| Data rate at field length | Per device and switch datasheet — confirm reach at your actual run length, not the headline rate | Per device and switch datasheet — confirm category and reach for the speed you need | Switch/PHY vendor |
| Connector family / mating face | SPE mating faces within the IEC 63171 series; M8 and M12 quick-connect variants exist | M12 X-coded, widely standardised in practice | Shared — must be confirmed across every vendor in the link |
| Outside diameter | Per supplier datasheet | Per supplier datasheet | Cable supplier |
| Minimum bend radius (fixed / flexing) | Per supplier datasheet — flexing radius is the figure that matters in a drag chain | Per supplier datasheet — larger OD usually drives a larger radius | Cable supplier, but only against a routing drawing you provide |
| Power delivery | PoDL (IEEE 802.3bu) — class and per-port budget to be confirmed in writing | Commonly PoE or a separate field supply, depending on the device | Switch/device vendor; cable supplier confirms conductor gauge and resistance only |
| IP rating path | Achieved at the terminated assembly and panel feed-through, not by the cable alone | Same principle | Assembly supplier, with the mating connector specified |
| Termination method | Factory overmoulded or field-assembly connector — decide who is accountable | Same decision, more mature field-assembly tooling in the market | You, at the point you choose the supply model |
| Commissioning test | Channel test with SPE-capable tester and adapters; PoDL voltage under load | Standard industrial Ethernet channel test; PoE or supply voltage check | You — and it is a schedule item, not just a procedure |
Numeric cells are deliberately left to the datasheet of the part you are specifying. Any article that hands you a generic outside diameter or bend radius for "SPE cable" is describing one supplier's construction, not the category.
Geometry: where the smaller cable pays for itself — and where it does not
Cable size becomes a design constraint in four situations: dense sensor clusters inside small enclosures, moving gantries, drag-chain fill weight, and robot wrist service loops. In each of them, a smaller OD does more than save space — it can change the minimum bend radius you must respect, the strain-relief hardware you need, and how much of the chain's fill weight is consumed by copper.
The size and weight benefit is real but is often repeated without attribution. The igus explainer on Single Pair Ethernet (published 12 February 2024, toolbox.igus.com) states that SPE cables can reduce installation size by 25% and weight by 35% over standard Ethernet cabling, and frames that benefit around automotive and railway applications where weight savings are critical. Two limits belong with that figure: the page reports no test method or measurement conditions in the passage, and the claim describes cable-level comparison in those industries — not a guarantee for your assembly, your jacket material, or your connector. Treat 25/35 as a directional planning input from a single publisher, then get the actual OD and weight per metre from the datasheet of the part you intend to buy.
Where the smaller cable does not pay for itself: high-flex duty cycles combined with heavy connectors, or short runs inside a cabinet where routing space was never the constraint. In those cases the geometry saving is invisible and you are taking on a new PHY, new connectors and new test equipment for no build-level return.
The concrete check: take the tightest real bend in the assembly drawing — including the service loop and the point where the cable exits the gland — and compare it against the flexing radius on the datasheet, not the fixed-installation radius. Static radii are frequently quoted because they are friendlier numbers. A drag-chain application lives in the flexing column.
Power at the device: a PoDL budget worksheet
SPE can carry power over the data line: Power over Data Line is specified in IEEE 802.3bu. What that standard does not do is tell you whether your specific sensor will receive enough volts at the end of your specific run at your specific cabinet temperature. This article cannot publish PoDL power classes or per-port budgets as conclusions for your build, because they are properties of the switch/PHY and the device, not of the cable category.
Instead, run the budget as a worksheet and put the numbers in the project file:
- Device current draw — from the device datasheet, at the operating point you actually run (worst case, not typical).
- Loop resistance — conductor resistance per metre × run length × 2, from the cable datasheet at the relevant temperature.
- Voltage drop — current × loop resistance. This is the line the cable supplier can genuinely help you calculate.
- Ambient at the enclosure — copper resistance rises with temperature, so a budget calculated at 20 °C and deployed at 60 °C is a different budget.
- Resulting voltage at the device input — compare against the device's minimum input voltage, including any brown-out margin the controls engineer wants.
- Port budget — confirm the switch's per-port power class and total budget in writing with the switch vendor.
Two things follow from that list. First, the cable supplier can confirm conductor gauge and resistance and nothing more; the power class is a switch-vendor and device-vendor statement. Second, a "powered over one pair" architecture can remove a separate 24 V field cable from some designs — which changes spare-parts strategy and how maintenance troubleshoots a dead sensor. That is a maintenance-acceptance decision, not an electrical one, and it should be signed off before it is designed in.
Connectors and IP67 sealing: the compatibility trap
The 4-pair side of this decision is comfortable: M12 X-coded is a mature interface with broad multi-vendor availability, and most machine builders already have the crimp tooling, test adapters and spares on the shelf. The SPE side is newer, and the mating faces sit within the IEC 63171 series, with M8 and M12 quick-connect variants appearing at field level.
The procurement risk is this: two connectors can both be legitimately described as SPE and still not be interchangeable across vendors. Whether a given plug, socket and panel feed-through combination actually mates and holds its IP rating is an interoperability statement that must be confirmed in writing, vendor by vendor across the whole link — switch side, device side and any bulkhead in between. Make that confirmation an RFQ line item rather than an assumption at commissioning.
The IP67 path deserves the same treatment. Ingress protection on a terminated assembly is a property of the mated connector pair, the sealing at the panel feed-through and the torque applied at installation — not of the cable jacket. Whether you buy overmoulded cordsets or field-assembly connectors, name the connector part number in the IP67 claim and state who is accountable for the seal. Note also that a smaller cable changes gland and strain-relief selection: hardware sized for a 4-pair jacket may not clamp a single-pair cable correctly, and an over-clamped or under-clamped gland is a leak path.
Finally, quick-connect cordsets eliminate cable glands, housing disassembly and screw terminals. That is a labour and rework argument as much as an electrical one, and it often matters more to the build schedule than the per-metre cable price.
Standards to name in the RFQ — and what they do not cover
- IEEE 802.3cg — the amendment covering 10BASE-T1L, the long-reach single-pair PHY most relevant to industrial field devices.
- IEEE 802.3bu — Power over Data Line.
- IEC 63171 series — the mating-face standards for single-pair connectors.
- IEC 61918 — the installation standard for industrial automation cabling, which has been amended to address 1-pair cabling.
None of these is a product certificate. Compliance with a PHY standard does not certify a complete channel end to end; a connector standard does not certify an assembly's IP rating; and an installation standard describes how cabling should be deployed, not how your specific cordset was built. Ask each supplier for the declaration of conformity and the exact standard revision they are working to, rather than accepting a logo on a datasheet.
Field assembly, commissioning and test method
A single-pair link has no spare conductors. Damage the pair and the link is dead — there is no degraded-but-working fallback. That raises the stakes on termination quality and test discipline compared with 4-pair, where a marginal pair sometimes limps along.
Decide termination ownership explicitly: factory-overmoulded assemblies put the sealing and the channel result on the assembly supplier, while field-assembly connectors put it on your technician and your tooling. Both are legitimate; a mix of the two without a named owner is where IP67 claims and test failures end up in dispute.
A bench and commissioning sequence a machine builder can actually run:
- Continuity and pinout check on every cordset before installation.
- Channel/performance test with a tester and adapter set matched to the interface being terminated.
- PoDL voltage measurement at the device input, under load, at the ambient the enclosure will see.
- Bend-cycle trial on the service loop for any moving axis, before the cable is dressed into the chain.
- A documented pass/fail record per run, filed against the assembly part number.
Flag one schedule risk early: test adapters for SPE interfaces are not yet standard kit in every technician's case. Confirm adapter availability — and who buys them — before the commissioning window, not during it.
Cost and lifecycle levers beyond cable price
Comparing per-metre cable prices answers the wrong question. The cost of the channel includes the number of home-run cables per device, switch port density and single-pair switch availability, any media converters or gateways needed to reach existing infrastructure, spare-part strategy, and technician training on a new interface.
Halving the cable count per device changes panel space and cabinet wiring time — a labour line item that procurement often misses when comparing metres. And the honest framing runs both ways: the cheaper cable is not always the cheaper channel once converters and new test gear are counted, just as the familiar 4-pair option is not automatically the lower-risk one where geometry is the binding constraint.
Pre-BOM validation checklist
Hand this to suppliers before you release the BOM, with a named sign-off for each line.
| # | Item to confirm in writing | Sign-off |
|---|---|---|
| 1 | Required data rate confirmed at the real field run length, from device and switch datasheets | Controls engineering |
| 2 | PoDL power class and per-port budget confirmed by the switch/PHY vendor; device minimum input voltage confirmed by the device vendor | Controls engineering |
| 3 | Voltage-drop calculation completed at maximum ambient at the enclosure, with conductor resistance from the cable datasheet | Controls engineering |
| 4 | Mating-face interoperability documented across every vendor in the link, including bulkheads and panel feed-throughs | Procurement |
| 5 | Published flexing bend radius compared against the tightest real bend on the assembly drawing | Controls engineering |
| 6 | IP67 claim tied to a named connector part number and stated installation torque | Procurement |
| 7 | Termination ownership agreed: factory overmoulded versus field assembly, with accountability for sealing and channel result | Procurement / maintenance |
| 8 | Standard revision and declaration of conformity received from each supplier | Procurement |
| 9 | Test method, tester and adapter availability confirmed, including who purchases adapters | Maintenance |
| 10 | Maintenance acceptance of any loss of a separate 24 V field cable, and revised spare-parts strategy | Maintenance |
Next step: samples and a build-to-print quote
If the checklist above points to SPE, the fastest way to de-risk it is to test a physical sample against your real routing and your real device load. If it points to M12 X-coded 4-pair, you still need a datasheet that states bend radius in the flexing column for your specific part number.
Request a quote or sample for SPE / M12 X-coded sensor cable assemblies. For a build-to-print quote on custom sensor and industrial Ethernet cable assemblies — connectors, overmoulding, IP67 sealing and per-part datasheets — send the following and you will get a specification back rather than a price list: the device list and required data rate, run lengths and the number of home runs, connector types at both ends, flex duty (static or drag chain, with cycle expectations), ambient temperature at the enclosure, required IP rating, and the tightest bend in the assembly drawing.
If your problem turns out to be link integrity rather than media choice — intermittent drops on an existing GigE Vision link, for example — the cabling question is secondary and the diagnosis should start elsewhere. Say so in your enquiry and we will point you at the right starting point.
Frequently asked questions
Can Single Pair Ethernet replace 4-pair Ethernet for all sensors on a machine?
No. SPE fits low-bandwidth, power-limited field devices on long runs where geometry matters. Devices needing higher throughput, familiar PoE practice, or connectors from a broad multi-vendor ecosystem usually stay on 4-pair M12 X-coded. Mixed architectures are common and normally the pragmatic answer.
How much power can PoDL deliver to a field sensor, and how do I confirm it for my device?
The power class and per-port budget are statements from the switch/PHY vendor and the device vendor — this article does not publish them as fixed values, because they depend on the specific hardware and on ambient temperature. Confirm both in writing, then calculate voltage drop from the device's worst-case current and the cable's conductor resistance over your actual run length.
Are SPE M12 connectors from different vendors interoperable?
Not automatically. Mating faces are addressed within the IEC 63171 series, but whether a specific plug, socket and feed-through combination from different vendors mates and holds its IP rating must be confirmed in writing for every component in the link.
What minimum bend radius should I specify for sensor cable in a drag chain?
Specify against the published flexing radius for the exact part number, and compare it with the tightest real bend in the routing — including the service loop and the cable exit at the gland. Do not use the fixed-installation radius as a proxy; it is usually smaller and will not survive continuous movement.