CAN bus failing? How to find faults with an oscilloscope

What is the CAN bus and why does it fail in industry

The CAN bus, or Controller Area Network, is a robust communication system used to connect controllers, PLCs, and sensors in industrial machinery. In practice, it works like an organized conversation where multiple devices share the same wires without talking over each other. However, the industrial environment is extremely harsh for electrical signals, filled with heavy motors, frequency drives, and switching power supplies shooting invisible electromagnetic sparks. When the bus starts to fail, machines stop abruptly, false alarms trigger, and overall production suffers mysterious drops that challenge the maintenance crew.

For automation professionals, diagnosing these issues requires looking beyond supervisory software. When a communication error occurs, the system only warns that a device went offline, but it does not tell the physical reason. This is exactly where the bench oscilloscope comes in, a graphical instrument that shows the actual shape of the electrical wave traveling through the cables. If you have ever had to deal with complex electrical faults, you know that understanding hardware behavior is just as important as fixing cold solder joints in electronic boards when equipment stops responding on the workbench.

How differential signal reading works in CAN networks

Unlike simple systems that use a single wire and ground, the CAN bus uses two wires called CAN_H (High) and CAN_L (Low). In practice, the signal is not read by its voltage value relative to earth ground, but rather by the voltage difference between these two cables. When both wires suffer an external interference, such as the magnetic field from a nearby motor, the noise affects both in the exact same way. Since the system measures only the difference between them, the interference cancels itself out, ensuring communication that is surprisingly resistant to noisy environments.

On the oscilloscope, we configure the math measurement to subtract the CAN_L signal from the CAN_H signal. The expected result is a waveform resembling a square wave that alternates between well-defined levels. When everything is correct, the resting voltage sits around two and a half volts, and the dominant signal raises the difference to about two volts. Any deformation in this perfect geometry indicates something is wrong with the electrical health of the network, whether due to excessive cable capacitance or severe issues in the physical assembly of connectors.

Identifying termination problems with the oscilloscope

One of the most common mistakes when installing CAN networks is neglecting termination resistors. In practice, the bus acts like a racetrack where signals travel at high speed. If the end of the track is open, the signal hits the wall and bounces back, creating an electrical echo called wave reflection that confuses receivers. To prevent this, a one-hundred-twenty-ohm resistor is mandatory at each of the two physical ends of the main cable, never in a star topology or with long branches.

When we connect the oscilloscope to a network without proper termination, we notice rounded corners, exaggerated overshoots, and unwanted oscillations called ringing at the waveform edges. If both resistors are missing, total resistance drops by half or climbs too high, destroying signal amplitude and preventing transceiver chips from recognizing sent bits. Measuring resistance with a multimeter while the network is powered down is a good start, but only the oscilloscope shows the dynamic impact of these reflections during real-time data packet transmission.

To ensure the system operates with maximum stability, follow these basic physical verification procedures on the workbench and in the field:

1. Completely power down the electrical panel and use a multimeter to measure the ohmic resistance between CAN_H and CAN_L to confirm the sixty-ohm value on the terminated network.
2. Connect the oscilloscope probe with an isolated ground to the bus terminals and configure the coupling to DC mode.
3. Run the signal sweep monitoring the logical level transition during active message transmission by connected sensors.

Detecting electrical noise and grounding failures

Electromagnetic noise is the great silent villain of industrial networks. High-power induction motors act as radiofrequency interference generators that couple into communication cables if they are not properly shielded. On the oscilloscope, noise appears visually as an exaggerated thickness in the waveform line or as high-frequency spikes jumping over logical levels. If these spikes exceed the circuit immunity margin, the receiver interprets noise as valid data, causing checksum errors and intermittent lockups that are hard to track down.

The main cause of this behavior is usually incorrect grounding or the notorious multi-point ground loop. The shielding mesh of the CAN cable must be grounded at only a single point, preferably in the main cabinet, to drain parasitic currents without creating high-energy alternative paths. If your industrial plant needs to handle delicate concurrent analog signal calibrations, remember that calibrating industrial 4 to 20mA sensors also directly relies on eliminating these same reference noise issues in the field. To explore these architectural concepts further in production, see our related technical analysis on How to secure REST APIs against traffic abuse with Redis.

Final Considerations

Diagnosing faults in industrial CAN buses using a bench oscilloscope transforms maintenance from a purely trial-and-error process into an exact science grounded in physical evidence. Understanding how signal integrity reflects the condition of cables, terminations, and shielding avoids unnecessary replacement of expensive components and drastically reduces machine downtime.

Investing time in correctly interpreting waveforms separates reactive professionals from truly prepared engineering teams. With a methodical approach of differential analysis and rigorous attention to the physical installation details, any challenging industrial network can return to operating with total reliability and deterministic performance.