Troubleshooting CAN Bus Errors and HMI Screen Blackout

Client Context

The client is an existing customer with whom we have an established working relationship. In previous engagements (see this case study), we supported them in resolving conducted and radiated emissions issues, enabling timely market release of their large power systems.

Through these successful collaborations, we have become a trusted EMC partner and a go-to resource for solving complex electromagnetic interference challenges.

Problem Description

In this case, the issue was not related to compliance testing, but rather a real-world EMI problem observed in the field.

The unit under investigation was a large industrial cabinet (approximately 2 m × 3 m × 2 m), containing:

• High-power bi-directional switched-mode power supplies

• A three-phase power system

The system exhibited:

• Intermittent CAN communication dropouts

• Occasional HMI screen blackout

These symptoms strongly indicated EMI-related issues. However, the non-repeatable nature of the problem made troubleshooting particularly challenging.  From the early stage review, we believed that the noise was generated internally within the system rather than from external sources.

Investigation Approach

Given the two distinct issues, the investigation was divided into:

1. CAN Bus Communication Errors

The CAN bus utilised a twisted pair within a screened cable, making screen termination strategy a key focus.

To assess noise coupling, two matched RF current probes were used, one on the inner conductors (CAN lines), one on the cable screen, this allowed direct comparison of noise inside vs. outside the screen

In this case, the cable screen is terminated via a pigtail. While this is not ideal from an EMC perspective, the dominant noise source within the cabinet is from relatively low-frequency, high-power switching circuits. As such, the pigtail termination is unlikely to be the limiting factor in this scenario and is considered acceptable as a baseline.

By using matched current probes, we are able to directly compare the noise measured on the screen and within the inner conductors. This allows us to assess the effectiveness of the termination and develop an optimised scheme to address the CAN communication issue.

2. HMI Screen Blackout

The HMI was driven by an off-the-shelf small computer module.

A reciprocal approach was adopted:

• Susceptibility testing → to reproduce the failure

• Emission scanning → to identify the noise source and coupling paths

Troubleshooting Details

CAN Bus Analysis

On the CAN side, improving the screen termination reduced the noise measured inside the cable. This was visible both in time domain and frequency domain. Nothing unexpected, but it confirms that the termination is doing its job when done properly.

Screen Blackout Investigation

For the screen blackout issue, the troubleshooting details are as follows:

First, we used a noise generator near the small computer module to see if we could reproduce the failure mode observed by the client in the field. This was achieved using a Langer E1 Set. The Langer SGZ 21 Burst Generator was used, as it can generate a high level of burst pulses, with two selectable waveform rise times.

In terms of coupling noise into the system, there are several options. In this case, we used the near-field probe supplied with the E1 set. We observed that when the magnetic field probe (BS 02) was positioned at the side of the small computer module, the failure mode could be triggered.

To couple noise into the cables connected to the small computer module, we used a simple approach by effectively shorting the two terminals of the noise generator. This forces the generated pulses to flow through a wire. This wire was then wrapped around the cables under test, forming strong magnetic coupling between the generator wires and the cables under test. This allowed us to test individual cables one by one to determine which were most susceptible to noise.

It was found that the power leads are robust, whereas the Ethernet cables are weak points. This aligns with our experience in the field.

So far, the method is focused on susceptibility. However, the question remains: why is this small computer module prone to noise in the system? To answer this, we need to understand the noise coupling path in order to develop a practical solution.

We then used our emissions troubleshooting kit. In this case, a Langer LF1 Set was selected. This set is specifically designed for low-frequency noise pickup and is highly sensitive. This makes it suitable for use in this system, where probing in high-voltage regions requires strict safety practices. A sensitive probe is therefore a good choice.

As shown, the LF-R 400 probe was taped to a safety-insulated rescue stick, allowing safe probing of the areas of interest. The results are convincing: the area where the small computer module is located is very noisy. Measurements were taken in continuous mode, and transient noise was also observed during start-up and shut-down.

Proposed Actions

• Move the small computer module to a quieter zone inside the cabinet (the proposed location was measured to confirm relatively low noise levels).

• Add a 31 ferrite material to the Ethernet cable to improve the robustness of the module.

Conclusion

Using appropriate test equipment and practical field experience, we were able to identify the noise coupling mechanisms and help the client mitigate noise issues affecting system performance.