EMC rectification case sharing

zidonghua01

EMC rectification case sharing [Copy link]

A certain vehicle-mounted electronic device needs to pass the vehicle standard CISRP25 LEVE3 test. The conduction current method exceeds the standard, and the 31MHZ frequency exceeds the standard by 11.5DB. The test current clamp includes a 24V power supply line and a CAN communication line. The specific test data is as follows: 1. Problem location The vehicle-mounted device is mainly used for battery heating of new energy vehicles. The circuit architecture is mainly divided into two parts, the low-voltage control part and the high-voltage heating part. The low-voltage control part is mainly realized by a 64-bit single-chip microcomputer, and the high-voltage heating is realized by IGBT switching. IGBT has no PWM control signal. From the functional analysis, the high-voltage circuit has almost no effect on the test. This is also confirmed in the actual test process. When the high-voltage power supply is cut off and the high-voltage power line is removed, the test data basically does not change. Therefore, the problem is located in the low-voltage control part. The low-voltage control circuit can be simply divided into three parts, DC-DC, single-chip control circuit and CAN communication circuit. From previous rectification experience, these three parts are likely to generate noise. When the conduction current method is tested, only the four lines of 24V, GND and CAN communication are tested. There are two test positions, 5cm and 75cm away from the device under test. The test found that the test results at the two positions are not much different, so the possibility of the device under test being coupled to the current clamp through space radiation is ruled out. The differential pair signal line of CAN communication is pulled out from the current clamp, and the test waveform has no obvious change, so the CAN signal is ruled out as a noise source. The problem is located on the two lines of 24V and GND. Now that the coupling path has been found, the only thing left is to find the source of the noise. DC-DC is 24V to 5V, 5V is used to power the microcontroller and CAN chip, 5V is cut off from the motherboard, cement resistors are used to replace the microcontroller and CAN chip as simulated loads. 31MHZ is still over the limit. Finally, we located the problem on the 24V to 5V step-down chip. 2. Processing methods This vehicle-mounted PCB is a 4-layer board. The light-on device in the figure below is the problematic DC-DC. It can be seen from the figure that the DC-DC is only routed on the top and bottom layers, and the middle layer does not have ground or power (red is the TOP layer, blue is the BOTTOM layer, and light blue is the GND layer). In addition, the DC-DC (interference source) is too close to the edge of the board. To sum up, two changes were made to PCB:①A complete GND was laid under the DC-DC. ②Move the DC-DC position inside the board, away from the edge of the board. The PCB after rectification is as follows: Test data after rectification: 3. Summary In the process of EMC rectification, problem location is the key, and this link often consumes most of our time. Only by accurately locating the problem and combining it with EMC theoretical knowledge and PCB drawing board principles can rectification be more effective. Source: Paperless Recorder yunrun.com.cn/product/

骑猪逛大街

Thank you for sharing. What is the "conduction current method test" and how does it work? Thank you