Diagnosing intermittent CAN faults with oscilloscope captures

Intermittent CAN bus faults are the most frustrating diagnostics in heavy equipment. The machine runs fine for hours, then a phantom fault code appears, a sensor drops out, or an actuator misbehaves—only to return to normal before you can probe anything. Unlike hard failures, these transient faults don't leave a persistent code, and the network often recovers before the ECU logs an event. The cost isn't just the repair; it's the repeated downtime and the slow erosion of trust in every networked module. The solution is a methodical oscilloscope capture strategy that turns a random glitch into a reproducible waveform you can analyze.

Why intermittent faults are a different class of problem

A hard failure—a severed wire or a direct short—creates a persistent, scannable fault code. An intermittent CAN fault is a transient breach in the communication protocol. The network may recover, but the corrupted data has already been sent, leading to erroneous actuator commands or lost sensor signals. The culprit is almost invariably physical degradation that only manifests under specific environmental stressors: resonant vibration, thermal expansion during a hot soak, humidity ingress after a washdown, or long-term corrosive buildup. The issue is rarely in the software; it's in the hardware's electromechanical connection. That's why random probing fails—you're not reproducing the stress condition that triggers the fault.

Step 1: Establish a baseline before touching anything

Your first move is to record the network's undisturbed state. Connect a dual-channel automotive oscilloscope across CAN_H and CAN_L at the most accessible point on the suspect network segment. Critically, do not unplug any connectors at this stage. You're capturing the network's native 'voice'—the normal waveform pattern that you'll compare against when the fault occurs. A properly functioning CAN bus shows a clean differential signal with a ~2.5V average when idle, and the two lines should mirror each other during active frames. Save this baseline capture for reference.

Step 2: Use mask testing to automate fault detection

Manual monitoring for an intermittent fault can take hours. Oscilloscope mask testing automates the process. A mask is a defined boundary—often a six-point polygon based on CAN physical-layer specifications—that your waveform must stay within. The oscilloscope continuously compares incoming waveforms against the mask. When a signal drifts outside the mask limits due to noise, trigger jitter, or a genuine fault, the scope flags a failure. Modern scopes can perform actions on failure: stop the acquisition, save the waveform, or even send an email alert. This turns a long, boring watch into an automated capture event. For CAN specifically, eye-diagram mask testing randomly captures and overlays every differential bit of every CAN frame, giving you a composite view of worst-case timing and amplitude variations. This is particularly useful for catching marginal signals that haven't failed outright yet.

Step 3: Set up triggers to catch the glitch

Mask testing is powerful, but you also need targeted triggers. Set your oscilloscope to trigger on the specific anomaly you suspect—a missing frame, a voltage drop below the recessive level, or an extra pulse. For example, if you suspect a short to ground, trigger on a CAN_H signal that falls below its normal low level. If you suspect an open circuit, trigger on a missing acknowledge bit or a frame that abruptly stops. The key is to define the trigger condition based on your baseline and the symptoms you're chasing. Many scopes allow you to trigger on a specific CAN identifier or data value, which helps isolate a particular node's communication.

Step 4: Reproduce the fault under controlled stress

Intermittent faults often appear only under specific conditions. Once your scope is armed with mask testing and triggers, you need to reproduce those conditions. Common techniques include:

• Vibration: Gently tap or wiggle the suspected harness sections while watching the scope. A fault that appears during this test points to a loose connector or broken wire.
• Thermal cycling: Use a heat gun or cold spray on the suspect area to simulate hot soak or cold start. Watch for waveform changes as temperature shifts.
• Load changes: Operate the machine through its normal cycles—lifting, steering, or engaging hydraulics—to stress the network.
• Washdown simulation: Spray water near connectors if humidity ingress is suspected, but be careful to avoid shorting other components.

Each time the fault occurs, the scope should capture the waveform. Save every capture with a timestamp and note the conditions. This evidence is your roadmap to the root cause.

Step 5: Analyze the capture to isolate the cause

Once you have a fault capture, compare it to your baseline. Look for specific anomalies:

• Noise or distortion: Indicates electromagnetic interference or a failing transceiver.
• Missing frames or gaps: Suggests an open circuit or a node that's dropping off the bus.
• Voltage levels outside spec: A CAN_H stuck low or CAN_L stuck high points to a short to ground or power.
• Termination issues: Measure resistance across CAN_H and CAN_L. You should see approximately 60 ohms with both terminating resistors in place (two 120-ohm resistors in parallel). A reading of 120 ohms means one terminator is missing or open.

Use the waveform to narrow down the physical location. For example, if the fault appears only when you wiggle a specific connector, that's your smoking gun. If the waveform shows a slow edge or ringing, you may have a capacitance issue from water ingress or a damaged cable.

Practical checklist for your next intermittent CAN fault

• Connect scope across CAN_H and CAN_L at an accessible point without disconnecting anything.
• Capture a baseline waveform and save it.
• Set up mask testing with a CAN-specific mask (or create your own based on the spec).
• Configure the scope to save or alert on mask failure.
• Add a trigger for the specific anomaly you suspect (e.g., low voltage, missing frame).
• Reproduce the fault under controlled stress (vibration, heat, load, moisture).
• Save all fault captures with timestamps and conditions.
• Compare fault captures to baseline and identify the anomaly type.
• Measure termination resistance (expect ~60 ohms) to rule out open terminators.
• Isolate the fault to a specific connector, wire, or node based on the evidence.

Quiz: Test your knowledge

What is the expected resistance reading across CAN_H and CAN_L on a healthy high-speed CAN bus with both terminating resistors in place?

• 120 ohms
• 60 ohms
• 0 ohms

How the Featured Expert Can Help

If you're facing a stubborn intermittent CAN fault or need expert repair for your control circuit boards, joysticks, or sensors, RE-CLICK offers nationwide support from certified technicians. Their focus is on reducing downtime and extending equipment life through fast, reliable electronic control repair and diagnostics. Call them to get your machine back to work.