1.Background of the Problem

A product developed over nearly a year encountered occasional reset and restart issues just before its market launch. Due to the instability of the phenomenon, it took a long time to investigate, revealing that the MCU was experiencing code runaway, leading to watchdog resets. After numerous checks of the code with the software engineer, we spent a week sleeplessly searching for clues without success. Eventually, we discovered that bringing the onboard wireless communication antenna close to the MCU significantly increased the restart frequency. Could it be that the MCU was crashing due to wireless interference?

2.Problem Localization

This project is a wireless communication product, featuring a 2W wireless communication module on the board. Further experiments revealed that positioning the wireless antenna near the 8M crystal oscillator and the MCU made it very prone to crashing, leading to the suspicion that the crystal oscillator frequency was being interfered with. The following tests were conducted to localize the issue.

• Real-time measurement of the external crystal oscillator frequency through an internal channel showed that when the antenna was close, the oscillator frequency jumped from 8M to a maximum of 250M.

• When the MCU used the internal clock source, there were no crash or restart issues when the antenna was nearby;

• Analyzing the clock architecture of the MCU, the external 8M crystal oscillator is multiplied by PLL 20 to reach 160M. Therefore, when the external crystal oscillator is interfered with to reach 250M, after passing through the PLL 20 multiplier, the output clock far exceeds the MCU‘s maximum frequency limit, causing the program code to run away.

3.Problem Resolution

Although the cause of the problem was clear, the resolution process was fraught with challenges.

• Using the MCU‘s internal clock: Due to strict accuracy requirements for the clock source in the product, the relatively low precision of the MCU‘s internal clock made it unusable;

• Shielding: While shielding could indeed solve the problem, it required custom shielding covers, which would severely impact the product’s cost, structure, and time to market, thus it was rejected;

• PCB Optimization: Shortening the traces of the crystal oscillator and grounding them did not resolve the issue after testing;

• Switching to a four-layer board and placing the crystal oscillator traces on the inner layer: After modifying the board, testing confirmed that the issue was resolved, and it was no longer affected by wireless interference. This solution is currently being shipped.

4.Postscript

The inner layer routing of the crystal oscillator has a similar effect to adding a shielding cover, and it also indicates that the interference was coupled into the crystal oscillator traces, causing variations in the oscillator’s oscillation frequency. For someone like me who has obsessive-compulsive tendencies, although the four-layer board solved the problem, the crystal oscillator traces need to go through vias to the inner layer, and the parasitic parameters of the vias can affect the oscillator’s startup and stability. I will continue to research and seek the perfect solution.

Previous Articles:

Which are the electric field interferences and which are the magnetic field interferences in EMC testing?

Can you distinguish between electric field interference and magnetic field interference?

EMC Tools: Are you using “Conductive Tape” and “Copper Foil” correctly?

Correct Understanding of MCU “5V Tolerant” Pins

How is the maximum 4V for MCU VDD determined?

Factors Determining the Maximum Operating Frequency of MCU: HVT/SVT/LVT

What is Nano Electrical Measurement?

Why is the sample size in HTOL testing 77 pieces? (Revised)

How to interpret MTTF values after HTOL life testing?

Chip Decap Process and Steps

Is the internal VSS of the MCU directly connected to VSSA or separated? Neither!

What is the typical yield rate of the MCU you are using?

Introduction to MCU Internal Wire Bonding

Analysis of a Bit Failure Issue in MCU SRAMWhy do different MCUs have different system ESD levels in the same environment?Is your MCU using Spi Flash or eFlash, and how to choose?Dynamic Voltage Stress Test (DVST)Early Failure Screening of ChipsChip Failure Analysis: Principles, Functions, and Differences between EMMI and OBIRCHDoes X-RAY cause data loss in MCU Flash?High-frequency signals are not necessarily high-speed signals, and high-speed signals are not necessarily high-frequency signals; please distinguish!ESD HBM Level of Chips and Its Related Failure RatesDetailed Explanation: How to Improve the Accuracy of ADCs within MCUsDifferences between System-Level ESD and Chip ESDAnalysis of Causes and Countermeasures for Chip Early FailuresWhen the MCU’s main frequency skyrockets, is a 0.1μF decoupling capacitor still sufficient?The Comeback of Mid-Range MCUs: The Rise of Domestic Chips, the “Invisible Champions” of the Era of IoTESD Case 1: Static Discharge Issues Caused by Internal Wiring in an Industrial ProductAnalysis and Solutions for MCU POR and PDR Dead Zone IssuesAnalysis of MCU Power Architecture: Exploring Multi-Voltage Domain Design with Several Mainstream MCUs as Examples2024 Domestic MCU Chip Company Sales RankingMust-Read for Engineers! Basics of MCU Electrical Overstress (EOS)