Application Case of Electromagnetic Simulation in EMC Design

---

foreword

"With the development of CEM (Computational Electromagnetics) and the improvement of computer computing power, more and more R&D technology companies choose to use simulation to guide the EMC design of products. With the blessing of simulation technology, the traditional EMC industry will develop What kind of new flower? This article uses several CST simulation cases to talk about the application of electromagnetic simulation in EMC design."

Keywords: EMC simulation, automotive Ethernet, switching power supply, motor drive controller

---

1 Significance of EMC simulation

Tell me why you want to do EMC simulation? Please look at the "X" curve below: In the EMC development cycle of a product, the closer to the end of product development, the fewer measures engineers can take to solve EMC problems, and the higher the corresponding cost. The meaning or purpose of EMC simulation can be summed up simply in two words - optimization. This "optimization" includes: shortening the development cycle, replacing low-cost solutions, imitating trend verification, providing theoretical support for design, and comparing iterations of solutions without hardware wait.

Regarding the substantive significance of EMC simulation, quite a few practitioners think that the efficiency of simulation is too low, and the imitation is not accurate, so it is better to directly test it. "Three tricks" - shielding, filtering, and grounding once again, the problem may be closed! And simulation, maybe still in the model processing stage?

Indeed, the computing power of early computers was relatively limited, and the application of electromagnetic simulation to guide system-level EMC design could not meet the requirements of efficiency and accuracy. For a simple PCB with a structural part, a single modeling calculation takes ten days and half a month, and the simulation results are not necessarily correct. I have been busy for a long time, but I can't get any guiding significance in engineering. Those who tried the simulation did not dare to make a decision on whether the plan would work, which resulted in very few "crab eaters" in the early stage.

But now, with the improvement of computer hardware (CPU, GPU computing power), the application of HPC technology, and the deep cultivation of major commercial and software giants in the field of computational electromagnetics, the accuracy and efficiency of simulation are not the same as they were more than ten years ago. . Leading R&D enterprises in the technology industry are gradually forming the awareness of simulation first in product R&D, using definite simulation calculations to guide product design (not just in the EMC field), and continuously accumulating various simulation databases and forming technology barrier.

Through the front-end Designing simulation, the defects in the EMC design can be found in advance, and the problems can be optimized or avoided in advance, and even the success of the first version can be achieved. Through the back-end Troubleshooting simulation, it helps developers understand the basic principles and optimization directions of electromagnetic problems, thereby greatly reducing and shortening the product development cost and cycle.

Therefore, in recent years, everyone has found that EMC simulation crabs are getting more and more delicious. If you don’t believe me, open the recruitment APP on your mobile phone and search for EMC-related jobs. Many companies will add a line in the JD: Those who master ANSYS/CST/FEKO and other electromagnetic simulation software are preferred.

2 Simulation Case of Automotive Ethernet

The electrification, intelligence, and networking of the automotive industry have brought about an increasing number and types of on-board electronic components. Data connection and communication are required between various vehicle electronic components, so that the vehicle bus must have higher data transmission capabilities.

Ethernet was put into commercial use in 1980. It has the advantages of low price, stability and reliability, and high communication rate, and is widely used in various industrial scenarios. But in the automotive field, it cannot be popularized due to its size and weight. Traditional Ethernet cables are usually composed of 4 pairs of twisted pairs (Twisted Pair, TP). According to whether there is a shielding layer, twisted pairs can be divided into shielded (STP) and unshielded (UTP).

In order to develop automotive Ethernet, the physical interface is simplified to a UTP, and the other components of the bus are kept unchanged, which can take into account the advantages of light weight and cost-effectiveness. BroadR-Reach is a point-to-point Ethernet physical layer standard for automotive applications. It allows two devices to perform full-duplex communication at a speed of 100Mb/s through UTP, and further increases to 1000Mb/s. The high-speed requirements make it compatible with BroadR-Reach Ethernet Network and automotive EMC standards become a highly challenging task.

Let's take the vehicle Ethernet as the research object and talk about how to carry out the EMC simulation of the vehicle Ethernet.

▲Vehicle Ethernet application scenarios

2.1 EMC Theoretical Basis of Automotive Ethernet

When doing EMC simulation, you must not rush to model blindly. Model simplification and grid division are important, but the first step is to understand the system architecture of the research object and the background of EMC problems, so it is necessary to fully consult the information.

Vehicle Ethernet adopts differential transmission signal, so that the cable has low interference and high anti-interference ability: (1) Since the signal line pairs are adjacent to each other, the magnetic fields generated by the twisted signal currents per unit length cancel each other out, Therefore, it is difficult for noise to be emitted outward; (2) For external interference, noise usually enters the two differential transmission signal lines in the same way, so it does not affect the waveform of the final useful signal.

▲The differential transmission line with twisted pair structure can make the magnetic field generated by the current within the unit length of the cable cancel each other, and effectively suppress EMI problems such as radiation emission

▲The clutter signal in the space generally enters the differential transmission line pair in the same way, and the actual effective received signal is the difference between the two complementary signals, so it can improve its EMS electromagnetic anti-interference ability

2.2 Causes of Common Mode Noise in Automotive Ethernet

As mentioned earlier, Ethernet using twisted-pair differential line transmission has the advantages of low EMI interference and strong EMS anti-interference, so why is there still EMC problems in applications?

Because the previous analysis is under ideal conditions, the differential twisted pair can indeed effectively solve the EMC problem. But if the signals between the differential lines are skewed (time difference) or have different amplitudes, the signal balance between the two lines will be disrupted, resulting in common mode noise.

▲The principle of common mode noise generated by differential lines

In the actual application process, there are the following main factors that will cause the EMC problem of the vehicle Ethernet: (1) The internal influence of the chip: the layout and wiring of the SiP substrate, the PINMAP design, the isolation between the power Ball and the signal Ball, etc. Factors (generally companies buy finished chips, this part is not involved, so the focus of the design is on the latter two cases); (2) The near-field inductance or capacitance of on-board devices around the signal link (such as switching power supply circuits) (3) The differential transmission link is unbalanced (such as wiring length mismatch, port connector impedance difference, etc.), the higher the speed of the scene, any small difference will be amplified.

▲Three main factors that cause Ethernet EMC problems

From the Maxwell equation and the electric dipole radiation theory, the maximum field strength of the common mode radiation (that is, the electric dipole radiation) at the far field can be calculated.

Among them, f is the signal frequency, L is the length of the radiation cable, ICM is the common mode current, and r is the test distance.

▲Electric dipole radiation model

From the theoretical formula, to reduce the common-mode radiation of Cable, we can start by reducing the signal frequency f, shortening the cable length L, reducing the common-mode current ICM, and increasing the test distance r. But in fact, only the common-mode current and cable length can be dealt with, because the signal frequency determines the transmission capability, and the test distance is required by EMC regulations and standards, which are difficult to promote changes.

2.3 EMC modeling method of automotive Ethernet

Field-circuit co-simulation is the most common simulation method in EMC simulation. In this example, the 3D field simulation chooses the frequency domain solver of the finite element method (FEM), and performs AC frequency domain analysis on the PCB and cable models within the frequency range of 200MHz. simulation.

Using NXP's vehicle-mounted Ethernet driver IC (TJA1100), the following figure shows the application circuit and PCB of the IC, including discrete ports and lumped components of chips, passive components, wiring and connectors.

▲Vehicle Ethernet TJA1100 application_PCB 3D field simulation model

▲Vehicle Ethernet TJA1100 application_ field road joint simulation model

2.4 Solutions for Automotive Ethernet CE

After the model is built, what can we get from the simulation? Assuming that the VCE test item of the vehicle Ethernet EMC exceeds the standard, what is the solution to the common mode noise problem?

▲Original VCE simulation results

2.4.1 Effect of VCE test arrangement on results

When some old drivers are doing EMC testing and rectification, they can achieve rejuvenation without adding a capacitor, relying on "professional" placement. So in this case, how would the cycloid arrangement improve?

▲ VCE simulation results under different cable spacings

▲Under extreme conditions, the VCE simulation results of Cable vertical extension

In summary, except for the last vertically extended pendulum method, which has improved by a few dB, there is no improvement in other cases. Although there is no clear requirement for the wiring harness layout of the VCE test in the EMC regulations and standards, it is also a strange practice to stretch the power cord so long during the test. To solve the problem, we still have to start from the essence, crooked ways are not the fundamental solution.

2.4.2 Effect of Ethernet twisted pair length on results

▲ VCE simulation results under different twisted pair lengths

The three spectrum comparison results: the noise from the twisted pair to the connector is lower, which is also in line with the actual situation. When doing rectification, many people usually twist the power line or signal line, and will especially require the quality of the twisted pair at both ends of the Connector to Connector.

2.4.3 Worst Case Design

▲By separating the power line (increasing the loop area), the signal line and the power line can be fully coupled

▲Worst Case Design（PCB+Connector+Power Cable）

Among the many results, Worst-Case is more than 40dB worse than the original design. Therefore, it is very important to master the underlying logic for top-level design. Instead of tinkering at the back end, it’s better to choose the right direction from the beginning, isn’t it more effective?

2.4.4 Magnetic ring on the power cord

Adding a magnetic ring is also the favorite of many old drivers, especially on the power line. During the test, all kinds of loads are put on first.

▲The magnetic ring is located at the head and tail of the cable

Comparing the two cases of adding magnetic rings at the head and tail of the cable, it is found that the results of the two are very similar, the peak value remains unchanged, but the resonance frequency is much earlier than the original case. According to 1/2πsqrt(LC), the magnetic ring increases the inductance L on the power line, and then f decreases, which makes the envelope peak advance (here is forced to analyze based on the simulation results).

From this point of view, it is beneficial to cover the power cord with a magnetic ring in this case, because in the VCE standard, the limit line of the low frequency band is generally looser than that of the high frequency band. Sometimes it is only a few dB difference, the envelope peak moves a little forward, and the test passes.

2.4.5 Cable shielding

▲Cable complete shielding and incomplete shielding comparison

The shielding layer of the cable is replaced by PEC. It can be seen from the simulation that the shielding of the cable is very effective, and the shielding layer near the connector is very important. This is why, no matter what kind of cable has a shielding layer, everyone will attach great importance to the grounding of the shielding layer at the cable termination position.

▲ EMC installation requirements of a certain system (360° grounding of cable shielding clips)

Comparison of common mode current paths between shielded and unshielded wires:

▲Common mode current path of shielded wire

▲Common mode current path of unshielded wire

2.4.6 Common Mode Inductance

The previous part is about optimizing the cables, and then we will talk about whether the single board can do something. According to the previous electric dipole radiation theoretical formula, in the case of constant common mode excitation voltage, reducing common mode noise can be achieved by increasing the common mode impedance of the link, that is, increasing the CMC (common mode inductance, also called common mode inductance). die choke).

CMC is made by winding two wires in opposite directions on the same magnetic core. The differential-mode current magnetic flux generated by the two wires cancels each other out, so it has no effect on the differential-mode current. The magnetic flux generated by the two lines for the common-mode current strengthens each other, so it plays the role of increasing the common-mode inductance and attenuating the common-mode noise without affecting the differential signal.

▲Filtering principle of common mode inductor

▲Common mode inductance port diagram and its calculation formula

▲Frequency characteristics of common mode inductors

Regarding the selection of common mode inductors, its frequency characteristics need to be considered. It is required that Sdd21 should preferably be low loss, and Scc21 and Scd21 should preferably be broadband and high attenuation. In actual application, it can be determined according to the frequency spectrum of common mode noise. For example, the frequency components contained in the differential mode signals of 100Base-T1 and 1000Base-T1 are different, and the suitable CMC mode conversion characteristics are also different.

▲100Base-T1 and 1000Base-T1 differential mode signal spectrum

According to the measured data of some CMC devices provided by Murata, the degree to which Scc21 suppresses common-mode noise in the low frequency band and the degree to which Scd12 suppresses common-mode conversion in the high-frequency band will affect the overall measurement results of the final conducted emission.

Let's talk about how to simulate the common mode inductor in CST. There is no common mode inductor model in the software, so it can only be imported by its TouchStone or Spice model. If the supplier does not provide it, the S4P file of the CMC can also be measured by the network score. The previous article on the import method has been published, and you can refer to " Component Model Import and Pre-simulation ".

In order to facilitate the later maintenance of the simulation model (for example, to compare the influence of different CMCs on the simulation results), it is recommended to add discrete ports in 3D first, and then import and connect the CMC model in Schematic.

▲In the simulation, set the delay for the signals of the two Ports

▲Compare the situation with or without CMC and with or without delay

It can be seen that the suppression effect of CMC on the common mode noise caused by time delay is very obvious, and the peak part suppression exceeds 20dB. Here, you can also compare the results by importing different types of common-mode inductor models to verify the suppression effect of Scd21 on common-mode noise mentioned above.

If further research is carried out, the winding structure of the common mode inductor can also be designed and optimized. It is mentioned in the following TDK patent that the reduction of the mode conversion characteristic Scd in CMC is largely related to the distributed capacitance of the inter-turn windings where the two sets of windings are different from each other. This can be obtained through the (quasi) electrostatic field simulation of CST EM Studio, so as long as you want to make details, any factor can become a variable for the comparison of the final noise results.

▲US Patent：US9659701B2，TDK

2.5 Automotive Ethernet RE Simulation Method

To perform RE simulation, it is necessary to add Field Probe in the 3D field to obtain the field strength at this position, and automatically draw the envelope of the highest noise of all Probes through the macro command that comes with CST.

▲RE simulation settings and post-processing methods

▲Add field probe

▲ EMC macro instruction result post-processing

▲Simulation results of 1m field of vehicle Ethernet

Similarly, the above CE simulation items can be compared, and details are not repeated here.

2.6 BCI/CCE Simulation Method for Automotive Ethernet

For BCI simulation, a current clamp model is required. Make 3D modeling of the current clamp used for your test, and then perform injection (mA) at the feed point port to simulate the voltage and current waveforms on the PIN pin of the chip.

▲BCI modeling process

▲BCI injection, get time domain waveform on IC

On the contrary, if CCE simulation is to be performed, an excitation source is added at the PIN pin of the chip, and the induced current noise (dBuA) is obtained at the current clamp.

3 Simulation case of switching power supply

Switching power supply is the hardest hit area of ​​electromagnetic interference in various large and small systems. There are many related academic papers on the research of power supply EMC simulation, and the application in engineering is relatively mature.

Miniaturization and high power density are one of the core goals pursued by power supply technology. The most effective way to reduce the volume of power supply is to increase the switching frequency, because increasing dt can greatly reduce the volume of transformers and inductors. After the third-generation wide bandgap semiconductor devices (GaN, SiC) are widely used, higher dV/dt also brings more EMC problems. The following is a brief introduction to an EMC simulation case of a BUCK power supply.

3.1 Power supply EMC modeling method

The core of EMC modeling of switching power supply is to establish accurate models of power devices and magnetic devices. Affected by the length of this article, we only introduce the situation of modeling with ideal switches for the time being. Welcome to continue to pay attention to the follow-up tweets of this account.

▲ BUCK power supply_PCB 3D field simulation model

▲ BUCK power supply_field-road co-simulation model

3.2 Power supply EMC simulation results

Here we compare the CE noise and electric field distribution of the power supply with and without the filter module. The CE noise peak is distributed at the multiplier of the switching frequency, and the noise reduction with the filter is about 30dB compared with that without the filter. From the perspective of electric field distribution, with filtering, the overall field strength on the PCB surface is much weaker.

▲Simulation comparison results with and without filters

4 Simulation case of motor drive controller

The essence of the EMC simulation of the motor drive controller is the same type of simulation as the switching power supply. After all, the switching power supply, the motor body, and the motor control all belong to different research directions of the power electronics discipline. Since around 2015, there have been many research papers on topics related to electric drive EMC. This may be mainly due to the rise of new energy vehicles, because the core of trams is the three-electric system: battery, motor, and electronic control.

The original plan of this chapter is to introduce the EMC simulation of the electric drive system, but the introduction may not be clear in a few words. Here is a foreshadowing for the follow-up article, including the EMC simulation of the switching power supply part in the previous section 3, which can also be simulated by ANSYS software, which can be very detailed, and there will be a chance to write in detail later.

Here is an extra word: When we study electric drive EMC simulation, we focus more on the controller part, but in fact the design of the motor body is equally important. For example, the optimized design of the motor winding and cogging structure can reduce the distribution parameters such as the leakage inductance of the winding, and directly lay a good foundation for the EMC/EMI of the motor and its drive control system. In addition, in the part of the control algorithm, since the control strategy of the tram is different in different states, the algorithm part also has an impact on EMC, and it can also be studied in combination with Matlab/Simulink.

5 Conclusion

At the end of the article, I would like to talk about the imitation inaccuracy. EMC electromagnetic simulation belongs to the simulation of unintentional emission and weak signal. Compared with SI/PI, EMC/EMI simulation is more of a process. Before the simulation, it is necessary to fully understand the system architecture and details of the research object. The high-frequency parameters of each device on the single board, and the screws of every gap and lap joint on the shell are very important. The simulated envelope does not even match the trend of the actual measurement results, and it must be that the details are not detailed enough. Just like hardware development pursues the success of the first version, simulation can also pursue the accuracy of the first version.

The simulation results of the aforementioned cases in this paper are not absolute value simulations. It means that there is still a large gap between the simulation results and the measured results. To achieve a complete absolute value simulation, a series of data processing and parameter correction operations are required. The most important thing before doing absolute value simulation is to evaluate whether it is necessary. On the other hand, there is more than one noise source in a complex system. Absolute value simulation means that you need to perform full model simulation, and you need to consider issues such as vector superposition of noise. This not only sounds complicated, but it is also difficult to do. You can write a high-level master's thesis.

In the past two years, the Martian Musk has become popular with a term that is both philosophical and scientific, called first principles. Originated from a point of view put forward by Aristotle, it roughly talks about the need to return to the essence of things, split them into various elements for deconstructive analysis, and find the optimal path to achieve the goal. Looking back at the previous cases, the first principles are especially suitable for EMC simulation. For practitioners in the EMC industry, the patched barefoot doctor-like method of seeing a doctor may one day be replaced by electromagnetic simulation, returning to one of the greatest formulas of classical physics - Maxwell's equations.

6 References

[1] Ethernet, Wikipedia.

[2] BroadR-Reach® Physical Layer Transceiver Specification For Automotive Applications, Broadcom Corporation.

[3] TJA1100 datasheet, NXP Semiconductors.

[4] EMC Simulation for Automotive Ethernet, Tamara Monti.

[5] Bulk Current Injection Application Note, CST Studio Suite.

[6] Common Mode Filter Basics, Panasonic.

[7] Common Mode Filter, US9659701B2, TDK Crop.

[8] An Introduction to Noise Suppression in Onboard Ethernet 100Base-T1, Murata.

[9] Automotive Ethernet for ADAS Noise suppression measures for 1000Base-T1, Murata.

[10] Emissions Simulation for Power Electronics Printed Circuit Boards, Patrick DeRoy.
[11] Mechanism Diagnosis and Characteristic Estimation of Circuit Radiation Interference, Zhao Yang, Journal of Electrotechnical Society.

---

Pay attention to "Electromagnetic Society" and make electromagnetic simulation no longer complicated!

• Follow Reply -

[Information] Obtaining learning materials

[Join group] Add technical exchange group

【Consultation】Product Solution