CAN Bus: How Modules Talk, and What Happens When They Stop

Before networking, a value needed by three modules required wiring to all three. A vehicle with twenty modules and hundreds of shared values would need a harness of impractical size, weight and cost.

CAN solves this by broadcasting. All modules connect to the same two-wire pair. A module that has a value — wheel speed, engine RPM, throttle position, ambient temperature — puts it on the bus in a message. Every module on that bus sees every message and takes what it needs. Nothing is addressed to a specific recipient; the message carries an identifier describing what the data is, and receivers decide whether they care.

The practical consequences follow directly:

• A single value can serve any number of modules at no extra wiring cost.
• Adding a feature that consumes existing data may need no new wiring at all.
• Every module on a bus depends on the health of that shared pair. One physical fault affects everyone.

That last point is the whole reason CAN faults present so strangely.

The pair is CAN High and CAN Low, and they carry a *differential* signal — the data is in the voltage difference between them, not in either wire's voltage against ground.

At rest (recessive state) both lines sit at roughly the same voltage, near 2.5 volts on a typical high-speed CAN bus. To transmit a dominant bit, the transmitting node drives CAN High up and CAN Low down, commonly toward about 3.5 V and 1.5 V. The receiver looks only at the difference. This is what makes the bus robust: electrical noise picked up from an ignition system or an alternator tends to hit both wires equally, so it cancels out in the difference and the data survives. The two wires are twisted together specifically to make that noise coupling as equal as possible.

High-speed CAN (ISO 11898-2) is terminated with a 120-ohm resistor at each end of the bus. Two terminators in parallel means the resistance measured across CAN High and CAN Low, with the vehicle fully powered down, should be approximately 60 ohms. For a properly terminated high-speed CAN segment this is one of the few genuinely standard values in vehicle networking, and it is the most useful single measurement in diagnosing a high-speed CAN fault.

• Roughly 60 ohms: both terminators present, bus wiring likely intact.
• Roughly 120 ohms: one terminator missing — an open in the bus, or a terminating module disconnected or internally failed.
• Near 0 ohms: the two lines are shorted together.
• Very high or open: the bus is broken between your measurement point and the terminators.

One caveat before you interpret any of that. On many vehicles built in the last fifteen-plus years, the DLC's CAN pins (6 and 14) route through a central gateway module rather than sitting directly across a terminated segment. A reading taken at the DLC on such a vehicle can be something other than ~60 ohms with nothing wrong, or can read a healthy ~60 ohms while the segment you actually care about is broken on the far side of the gateway. Confirm from the wiring diagram which segment your probe points are really on, and where in the harness to measure to see both terminators.

Where the terminating resistors live is vehicle-specific — often inside two modules at opposite ends of the bus, sometimes as discrete resistors in the harness. Do not guess; use the service information for the specific vehicle.

Not every network in the vehicle follows these numbers. Vehicles run multiple buses at different speeds — a fast powertrain bus, a slower body or comfort bus, and often others — and low-speed fault-tolerant CAN uses different termination and different voltage levels entirely. LIN, FlexRay and automotive Ethernet coexist with CAN on many modern vehicles. Confirm which network you are probing before you interpret a reading.

The signature of a network fault is multiple unrelated systems misbehaving simultaneously, often with a cluster full of warning lights that make no sense together.

A vehicle with a compromised bus can show, at once:

• ABS and stability control lights
• Airbag light
• Check engine light
• Speedometer or tachometer reading zero or erratic
• Transmission stuck in a default gear
• Power steering assist reduced or absent
• HVAC or lighting behaving oddly

Important: those systems are not merely reporting — they are disabled. ABS and stability control are not intervening, so a panic stop or a slippery corner behaves like an old car with no electronic help. The restraint system may not deploy. Reduced or absent steering assist changes steering effort, and it can change without warning while you are turning. The root cause is often something cheap, but that has no bearing on the state of the vehicle in the meantime: it should be driven minimally or trailered until the fault is found, and driven as though none of those systems exist.

None of those modules failed internally. They stopped receiving data they depend on. A stability control module that cannot see engine torque or steering angle disables itself and reports a fault, correctly.

The corresponding codes are U-codes — the network communication family. A U0100 (lost communication with ECM/PCM) stored in the transmission module means that module stopped hearing the engine module. It does not mean the engine module is bad. It means the message did not arrive, which could be the sender, the receiver, the wiring between them, power or ground to either, or the bus itself.

That asymmetry is the most important thing about U-codes: the module that stores the code is the one that noticed, not the one at fault. Reading which modules report loss of which other modules is a mapping exercise, and the pattern usually points at the physical location of the problem.

A second signature is intermittent, position-dependent behavior — faults that appear over bumps, in rain, on hard turns, or when a specific accessory is used. That points at a physical connection: a chafed harness, a corroded connector, water intrusion, a poor ground.

The order matters, and it starts with the least invasive tests.

Scan every module and record which are present and which are missing. A tool that performs an all-module scan will list modules that failed to respond. If the missing modules share a bus segment, you have localized the fault before touching a wire. If a single module is missing while everything else responds, suspect that module's own power, ground, or branch wiring first.

Check the obvious power and ground. A weak or failing battery, or a poor main ground, produces network chaos that mimics a wiring fault. Modules browning out and resetting drop off the bus and return. Verify battery condition and charging system before anything else — this step resolves a meaningful share of "random network faults" and costs nothing.

Measure termination resistance, key off and battery disconnected as appropriate for the vehicle. Across CAN High and CAN Low on a high-speed bus, expect approximately 60 ohms as described above. Before you act on the number, apply the gateway caveat from the previous section: confirm from the wiring diagram that the pins you are probing land directly on the segment you are chasing rather than on the gateway's DLC side. Measured on the right segment, this one reading separates open-circuit, short-circuit and healthy-wiring cases immediately.

Measure bus voltages with the network live. Compare CAN High and CAN Low against ground and against each other. A line shorted to ground, shorted to battery voltage, or shorted to its partner shows up clearly here.

Scope the bus if the tools allow. An oscilloscope on both lines shows waveform quality that a multimeter averages away — reflections from a missing terminator, one line collapsed, a node babbling. A shorted or continuously-transmitting node can dominate a bus in a way that only a scope makes obvious.

Isolate by disconnection, carefully. Progressively disconnecting modules while watching resistance and bus behavior localizes a shorted node. Do this with the manufacturer's guidance — disconnecting some modules on some vehicles has consequences, and restraint system modules should be handled per the manufacturer's safety procedure without exception, including the specified battery-disconnect and stand-down time before any connector is touched.

Network diagnosis is time-intensive. One to three hours is common, and $150 to $400 in diagnostic labor at an independent shop is a realistic range. It is genuinely cheaper than replacing modules speculatively, which frequently requires programming and is often non-returnable.

Do not replace a module because it stored a U-code. The storing module is the reporter. Replacing it is the single most common and most expensive error in this area, and on most vehicles a replacement module requires programming and configuration to the specific vehicle — meaning the money is gone whether or not it helped.

Do not tap into the bus for accessories casually. Aftermarket installations that splice into CAN wiring, add poorly-designed devices to the bus, or draw power from network circuits are a frequent cause of intermittent network faults. If a vehicle developed communication problems after an aftermarket alarm, remote start, stereo, or telematics install, that installation is a prime suspect and should be inspected before anything else.

Do not repair CAN wiring with ordinary splices. The twisted pair geometry is functional, not cosmetic — it is what makes noise cancel. Repairs should maintain the twist rate and use appropriate methods per the manufacturer's repair procedure. A long untwisted section or a poorly executed splice can produce a bus that works on the bench and fails in traffic.

Do not assume all buses behave alike. Voltage levels, termination and speed differ between the networks in one vehicle. Applying high-speed CAN expectations to a low-speed or single-wire network yields readings that look like faults and aren't.

Do not clear codes before mapping them. The pattern of which module lost which other module is the map. Erase it and you start over — and on an intermittent fault, you may not get a second map.

Do not hand the vehicle back "to monitor" with the network fault still present. Whatever the root cause turns out to be, the driver is operating without ABS, without stability control, possibly without restraint deployment, and possibly with altered steering effort.

A dashboard lighting up with six unrelated warnings looks catastrophic and is often a single connector, ground, or battery — though the safety systems really are offline until it is found, so the cheapness of the fix says nothing about how safe the car is to drive in the meantime. Understanding that the modules are reporting missing data rather than internal failure changes the whole approach — from replacing whichever module holds a code to finding the one physical problem that starved several modules at once. The 60-ohm termination check, taken on a segment you have confirmed from the wiring diagram, takes minutes and separates the most common cases, which is a very good return for a multimeter.