What Is a T-Harness? Design, Pinouts, and CAN Bus Applications

A T-harness adds a device to an existing circuit without cutting a single wire, and its design turns on four constraints:

Key Takeaways

• A T-harness is a pass-through assembly: it inserts between a mated connector pair, preserves the original circuit, and adds a third branch for a new device.
• On SAE J1962 (OBD-II) connectors, CAN High is pin 6, CAN Low is pin 14, battery positive is pin 16, and grounds are pins 4 and 5.
• The tap branch is electrically a bus stub: SAE J1939-11 limits stub length to 1 m at 250 kbit/s, and faster CAN rates demand shorter branches.
• The most common failure is a third 120 Ω terminator — many telematics devices terminate internally, dropping a mid-bus network from 60 Ω to 40 Ω and corrupting differential levels.
• Each T-harness inserts two additional mated interfaces into a circuit that previously had one, so contact resistance, sealing and latch loading must all be re-verified.

Engineering rule of thumb: keep the tap branch as short as the installation allows, hold CAN High and CAN Low twisted to the crimps, and confirm the added device's internal terminator is disabled before the harness is built.

T-Harness, Y-Harness, or Splice: What the Names Actually Mean

A T-harness breaks into a circuit that is already mated. You unplug an existing connector pair, insert the T between the halves, and the original signal path continues unchanged while a third branch feeds a new device. Nothing is cut, and removing the T restores the machine to original configuration.

That is a different structure from the three assemblies it gets confused with — a Y-harness, an inline splice and a breakout box — which the table below separates on continuity, reversibility and added interfaces.

The distinction is commercial as much as semantic: OEM and fleet programs choose a T-harness because it leaves no trace in the base machine, which warranty terms, lease returns and resale value all depend on.

The Pass-Through Branch: Pin-for-Pin Integrity

Everything in the original circuit must survive the detour. The pass-through branch is pin-for-pin, gauge matched or larger, and keyed so it cannot be installed backwards. Because the T adds conductor length plus two mated interfaces, voltage drop on a power circuit is measurably worse than the original — on high-current loads, size the pass-through conductors up rather than matching the OEM gauge.

Sourcing the connector pair is usually the hard part. A buildable T-harness needs the housing family, contact pitch, keying variant, terminal part number and seal type identified on both ends. Off-highway equipment commonly mates to a Deutsch wire harness using DT, DTM or DTP housings; industrial machinery typically uses M12 circulars. An undersized contact will not retain in a sealed housing, and the wrong seal diameter for the wire OD defeats the ingress rating.

Declare workmanship class up front, because it governs inspection rather than sentiment. A custom wire harness built to IPC/WHMA-A-620 Class 3 carries tighter crimp and inspection criteria than Class 2, with crimp tensile strength verified against the standard's pull-force table for the gauge.

The Tap Branch Is a Bus Stub — Treat It Like One

This is where most T-harness installations fail, and it is rarely discussed. A high-speed CAN network per ISO 11898-2 is a linear bus terminated with 120 Ω at each physical end, with nodes hanging off it on short stubs. When a T-harness breaks CAN High and CAN Low out to a telematics box, that branch is a stub, and stub length is bounded by bit rate.

SAE J1939-11 gives concrete numbers for heavy-duty networks at 250 kbit/s: stubs up to 1 m, a backbone up to 40 m, and a maximum of 30 nodes. ISO 11898-2 publishes no universal figure because the limit scales with bit rate; at 500 kbit/s, practice commonly holds stubs near 0.3 m.

Two further rules apply to the branch. The differential pair must stay twisted through the T, since untwisted length degrades common-mode rejection and raises radiated emissions — separate the pair only at the crimps. And a T inserted at the physical end of a bus relocates where the network terminates, changing the topology rather than extending it. The same logic that governs CAN bus stub length and termination placement applies unchanged here.

Power, Ground and Fusing on the Tap

A tapped device needs a power strategy decided at design time. Constant power — SAE J1962 pin 16, or an unswitched machine circuit — keeps the device alive while the equipment sits, which is what fleet tracking requires and also what flattens batteries. Specify sleep current against the dormant period, or take an ignition-sense circuit so the device sleeps.

Fuse the tap branch close to the tap point, and rate the fuse to the branch conductor rather than the source circuit. An unfused branch means a fault in the added device opens an OEM circuit or overheats OEM wiring — the opposite of the non-invasive intent. SAE J1128 low-tension primary types cover general routing; cross-linked insulations are specified for under-hood temperatures.

Ground the added device at its signal reference, not at a convenient chassis point. Splitting power return and signal ground across two chassis locations creates a ground loop that surfaces as CAN error frames. The same single-point discipline governs retrofit taps in I/O and control cable assembly work.

Sealing and Mechanical Load: Two New Interfaces

A T-harness converts one mated interface into three, and each is a potential ingress path. If the original pair was rated IP67 per IEC 60529, both new interfaces must achieve the same rating or the assembly has downgraded the machine. That means seals matched to the actual wire outside diameter and cavity plugs in every unused position — an open cavity voids the rating as completely as a missing seal, as with any sealed IP67 cable assembly.

Mechanically, the T hangs mass off a latch designed to retain a harness, not a harness plus an adapter plus a branch. Secure the tap branch to structure close to the connector so vibration and cable weight load the tie-down rather than the latch.

T-Harness vs. Y-Harness vs. Splice vs. Breakout Box

SAE J1962 Pin Assignments Relevant to a CAN Tap

Common Questions About T-Harness Design

What is the difference between a T-harness and a Y-harness?

A T-harness passes an existing mated connector pair straight through and adds a third branch; a Y-harness splits one connector into two with no pass-through. A T intercepts a working circuit to observe or supplement it, while a Y divides a source between two loads.

How long can the branch on a CAN bus T-harness be?

Treat the branch as a bus stub and apply the protocol's stub limit — SAE J1939-11 allows up to 1 m at 250 kbit/s. Higher bit rates require shorter stubs, and 500 kbit/s networks are commonly held near 0.3 m. Measure from the backbone tap point to the device's transceiver, not to the connector shell.

Does a T-harness need its own 120 Ω termination resistor?

No — and adding one is the most common way a T-harness breaks a working bus. A correctly terminated CAN network already has 120 Ω at each end, presenting roughly 60 Ω to the drivers; a third terminator pulls that to about 40 Ω and the transceivers can no longer develop correct differential levels. Check whether the added device terminates internally, because many telematics units ship configured as end nodes.

Which OBD-II pins carry CAN on a J1962 connector?

CAN High is pin 6 and CAN Low is pin 14 on an SAE J1962 connector. Pin 16 carries permanent battery positive, pin 4 is chassis ground and pin 5 is signal ground. Several remaining pins are left to manufacturer discretion, so confirm against the specific platform.

Can you build a T-harness from just the machine's connector, with no drawing?

Yes — identifying the connector from a physical sample or clear photographs is a normal part of quoting. What is needed is the mating connector or good images, the circuit list for the tap branch, the target bit rate and stub length, and the environmental rating the assembly must hold. Samples are available for fit verification before a production release.

A T-harness is only as good as four decisions: pin-for-pin pass-through on conductors sized for the added length, a tap branch inside the protocol's stub limit with the pair twisted throughout, a fused and correctly referenced power strategy, and sealing and strain relief matching what the original interface provided. Verify the added device's internal termination before the harness is built — that one check prevents the most common field failure.