Electromagnetic interference (EMI) filters are commonly used in automotive 13.8 VDC power supply networks to reduce high-frequency noise that is conducted from the printed circuit board (PCB) and causes EMI problems. Due to parasitic effects related to the filter itself or PCB layout, filter performance is difficult to predict and is often affected at high frequencies. Power line filters using surface mount technology ferrite and multilayer ceramic capacitors are attractive solutions for reducing RF noise in high-density automotive PCBs. A lumped element SPICE model is introduced to optimize the π filter design, including a frequency-dependent ferrite component model. Provides an overview of the PCB implementation of EMI filters to achieve the best filter performance.
© 2010 IEEE Reprinted with permission from the Proceedings of the 2010 IEEE International Electromagnetic Compatibility Symposium.
Switching power supplies (SMPS) are widely used in automotive electronic modules due to their excellent efficiency. SMPS almost no longer use linear regulator power supply. However, their inherent noise and generated ripple require filters to meet strict EMC regulations. Automotive EMC requirements [1, 2, 3, 4] are at least 54 dB more stringent than FCC-15/JB category limits [5]. The use of EMI filters as a noise suppression technique must be carefully considered, because cost and space in a highly crowded PCB are the two main determinants. Discrete π filters in small packages are a common design practice in the automotive design world. However, EMI filter design and PCB implementation are not easy tasks. Therefore, when the EMI filter does not meet the requirements and cannot provide the required noise suppression, this is a major problem [9, 10].
A π filter that works well in theory using traditional circuit theory or simulation tools may not meet the strict requirements of its actual requirements (CE mitigation). When a well-designed EMI filter fails to work properly and cannot provide any noise reduction function, this is usually considered a magical art [11, 12].
Therefore, it is necessary to consider and adopt experimental methods to identify "technical errors" that may cause filter failure. Under the influence of complex input/output impedance changes, the electrical behavior of the EMI filter installed on the actual product may be quite different from its basic electrical schematic. In order to gain insight into the actual behavior of EMI filters, consider using non-ideal electrical models for filter components. However, this can also be limited to provide a solution. In addition, under the actual product development budget, if not impossible, PCB parasitic effects due to single copper traces or component interactions should be considered. In each product development timeline, engineers are looking for realistic, fast and meaningful solutions. There are a large number of EMI filter design guidelines and best practices available to the engineering community. The product design team must provide an EMI robust module, which must meet strict EMI requirements, be cost-effective, and have almost no opportunity for design iteration. In the cost-effective, mass-produced automotive world, avoiding EMI failures and reducing design iterations are absolutely necessary. For the engineer, getting it right in his first attempt is an extremely important aspect of the automotive world. However, the EMI filter design is considered here, and its components do deviate from the ideal behavior in a linear and non-linear manner. Non-linear behavior is rarely understood or understood, more specifically, when ferrite is used in distribution networks [13, 14]. SMT ferrite suppliers provide the relationship between impedance characteristics and frequency. In addition, they also provide several lumped circuit models to help design engineers perform SPICE simulations [6, 15].
In an ideal situation, the π filter provides a perfect solution for the noise generated by EMI; more specifically, it is used to reduce the noise caused by the conducted radio frequency radiation generated by the switch mode power supply. The main concern is the MW-AM band radio (0.52 MHz – 1.73 MHz). Currently SMPS can operate in a switching frequency range of 100 kHz to 2 MHz. The main problem is the first few harmonics of the SMPS switching frequency, in these harmonics they may cause EMI inconsistency.
This section introduces the inspection of basic problems encountered during the EMI compliance assessment of automotive safety restraint control, which failed to meet the conducted emission requirements.
A typical airbag restraint module is designed on a printed circuit board composed of four layers of copper and fabricated on an FR4 substrate to accommodate more than 500 components. One layer of PCB is dedicated to the return (ground) plane. The remaining inner and outer layers are connected to a large number of low-speed signal traces, power distribution network, high-speed communication network, PWM-based sensor communication and HS-CAN bus communication. Unused gaps are filled with copper traces or copper islands. The "ground fill" gap is "stitched" to the return plane using a large number of through holes distributed less than 1 cm in distance.
Figure 1 illustrates the noise spectrum content, where deviations are recorded at the harmonics of 390 kHz (SMPS switching frequency).
The diagnostic work of identifying, identifying and isolating the noise source and its conduction path can easily be reduced to the input port of the boost converter operating at 390 kHz. It is recognized that the unit under test may need a simple third-order EMI filter with 60 dB insertion loss (MW-AM band) to meet CE requirements. The simplest form of EMI filter is an impedance discontinuous network inserted into the input port of a 13.8 V DC battery or ignition wire. The EMI filter provides the required attenuation for noise signals that may cause the conducted emission test to fail.
The π filter topology is a common design practice in most automotive noise suppression design techniques. The π filter network uses 0603 SMT ferrite beads and two identical 4.7/μF, 1210 multilayer ceramic capacitors (MLCC) designs to provide the required insertion loss for the frequency band of interest. However, the behavior of the EMI filter seems to be quite unpredictable, which is an unacceptable situation. Design engineers with little EMC training may resort to incomprehensible statements, and for most engineers, participating in the ghost hunting scene is frustrating.
In order to eliminate or reduce the inconvenience caused by EMI identification of poor filter designs for noise reduction, a systematic method for evaluating the success of the design is outlined. Since multiple iterations attempted by replacing the filter assembly can be problematic and costly, it is insightful to seek technical guidance to achieve a high success rate. The questions that need to be answered are:
SMT ferrite beads have the potential to solve the problem of large-scale interference as a frequency-selective nonlinear device. Unlike magnetic metals, ferrite is a magnetic dielectric that allows electromagnetic waves to penetrate the ferrite, thereby allowing interaction between the electromagnetic waves and the magnetization in the ferrite. Ferrite is a non-conductive ceramic material made by sintering iron oxide and zinc oxide or a mixture of nickel and zinc oxide. Figure 2 illustrates the impedance characteristics of 0603 SMT ferrite beads selected for π filter applications.
For this application, the impedance of the SMT ferrite bead is 181 Ω at 100 MHz, as shown in Figure 2. Obviously, at lower frequencies (less than 25 MHz) the inductance of the magnetic beads is the main item. In the frequency range between 25 MHz and 400 MHz, frequency-dependent resistive dissipative elements dominate the impedance characteristics. At higher frequencies (above 400 MHz), the parasitic capacitance of the SMT ferrite structure is the main term.
Figure 2: SMT 0603 chip ferrite bead impedance-frequency characteristics
Figure 3 is an illustration of a simple electrical equivalent circuit lumped model provided by a ferrite bead manufacturer. It is assumed that the value provided for the equivalent circuit model is only valid at a frequency of 100 MHz. In fact, this model is a very simple representation of actual electrical behavior and does not represent frequency-related terms. It should be used with this restriction. However, the usual practice is to check the ferrite at 100 MHz.
Two 4.7/μF 25 V SMT X7R (1210, 3.2 mm × 2.5 mm) MLC shunt capacitors [7] complete the structure of the π filter. Figure 4 is the equivalent circuit model diagram of the capacitor used, and Table 1 provides the values of each component at 2.29 MHz (SRF frequency). Figure 5 illustrates the impedance-frequency characteristics of a 4.7/μF MLC capacitor.
Figure 4: MLCC (4.7/μF) equivalent circuit model
Table 1: MLCC 1210 (4.7 μ/F) model component values
Figure 5: MLCC 1210, 4.7/μF impedance characteristics
Using the electrical model and component values provided in the figure above, you can use PSPICE to develop a simple π filter circuit simulation to evaluate the effectiveness of the EMI filter in the conducted emission frequency range. In fact, one such simulation shows that the insertion loss in the entire AM band reaches an incredible 80 dB. The realistic goal requires us to demonstrate a practical application in which an insertion loss of 80 dB can be achieved. If not, what are the restrictions? To investigate this further, a 2-sided test PCB (67 mm × 52 mm) was constructed as shown in Figure 6. The underside of the PCB is dedicated to the solid copper return plane. As shown in Figure 6, a PCB trace with a width of 2.8 mm and a length of 30 mm is realized and terminated to the SMA port where the filter component is connected.
Figure 6: π filter test circuit printed circuit board geometry
Use a calibrated network analyzer (HP8753D) to measure the voltage transfer function (|S21|) of the π filter structure, as in the test PCB.
The SMT ferrite bead is removed and replaced with a 2.7 nH ceramic core inductor, and the voltage transfer function (|S21|) is captured. In addition, a large ferrite core inductor of 4.7/μH is used instead of 0603 ferrite beads, and the voltage transfer function (|S21|) is recorded. Figure 7 illustrates the post-processed data.
In addition, the quality factor of 2.7 nH ceramic core inductors and ferrite beads was measured at 800 MHz: Q ~ 63 of 2.7 nH ceramic core 0805 inductors is considered "high Q", while BLM18PG181SN1 ferrite SMT The Q of 0603 is ~ 0.68, which is considered "low Q". This is a basic EMI filter design guideline to avoid the use of high-Q components. Therefore, ceramic core inductors must be avoided.
Obviously, the large and expensive 4.7/μH inductor has superior performance at lower frequencies (below 1 MHz) compared to ferrite beads, as expected. However, it is important to note that for frequencies greater than 1 MHz, ferrite beads are as effective as large inductors, and in the frequency range (1 MHz – 10 MHz), its performance is 3 dB higher than large inductors . Obviously, small 2.7 nH ceramic core inductors are not valid in the MW-AM frequency band and should be avoided. The measurement data in the "experimental PCB" is useful for design engineers to evaluate the effectiveness of the filter and provide design guidelines.
Figure 7: Voltage transfer function of π filter
This section explains the product PCB implementation of the π filter. The implementation rules stipulate that automotive electronic products need to operate reliably under harsh environmental conditions. In some applications, printed circuit boards are treated with conformal coatings to protect the PCB and its electronic components. Due to the conformal coating process, several important manufacturing requirements arise. For example, an important rule requires PCB design engineers to avoid placing vias under electrolytic capacitors. Electrolytic capacitors have failure modes related to chemical leakage. An incorrect electrolyte formulation in a malfunctioning capacitor can cause hydrogen to be generated, causing the capacitor casing to swell or deform, and eventually cause electrolyte leakage. In the event of a capacitor failure, ground fill islands and vias to the PCB ground plane can cause catastrophic electrical shorts. In the electronic restraint control module, a large electrolytic capacitor (6.7 mF/25 VDC) is required to deploy the airbag when the input power is lost. If the vehicle's power supply system is interrupted, the airbag deployment function will not be affected. In high-density printed circuit boards, the 6.7 mF/25 VDC electrolytic energy storage capacitor is indeed the largest physical component.
The integration of the π filter is shown in Figure 8, where the positions of C1, C2 and SMT ferrite components are indicated. The π filter assembly is placed close to the power management ASIC and chassis ground. The energy storage capacitor (6.7 mF aluminum electrolytic capacitor) is installed on the other side of the PCB. The energy storage capacitor is grounded as shown in the PCB picture. It is important to note that there are no vias in the "copper island" related to the ground connection of the energy storage capacitor and the ground connection of the EMI filters C1 and C2. The basic EMC design guidelines require the use of multiple vias to stitch the ground-filled copper island to the ground plane. However, this requirement cannot be implemented here because, as mentioned earlier in this paragraph, all conformal coating printed circuit board manufacturing requirements must avoid the placement of through holes.
Figure 8: Printed circuit board (board #1)
For EMI compliance, products need to be tested according to regulations. The conducted radio frequency emission test device meets the requirements of CISPR 25 [8], as shown in Figure 9. It should be noted that the safety restraint control module is electrically connected to the body through the floor shell during the production process. Therefore, the DUT is placed directly on the CISPR 25 ground plane and joined to the ground plane as shown in the setup picture. The test harness is 1700 (300, -0) mm long and is routed 50 mm above the ground plane.
Figure 9: CISPR 25 conducted emission (voltage) test setup
The electrical schematic diagram of the automotive line impedance stabilization network (LISN) is shown in Figure 10.
The conducted noise measurement of PCB #1 is shown in Figure 1. Obviously, the implementation of PCB #1 exceeded the CE compliance requirements for SMPS switching frequency and its harmonics. Many changes to the value or type of EMI components have failed to mitigate CE noise. Placing the π filter component at the power input pin and electrically connecting it to the chassis ground through a low impedance connection is a basic EMI design guideline. Obviously, the filter components are strategically positioned in accordance with the requirements of the EMI guidelines. In order to study the effectiveness of the EMI filter, it was decided to remove the filter components and replace the ferrite beads with 20 mn resistors. The CE noise spectral content and its amplitude remain the same in both cases. In fact, the presence or absence of EMI filters led to the same CE data, revealing serious design errors. Someone suggested that implementing an EMI filter on PCB #1 is completely different from the test board (Figure 6) or simulation results. Therefore, the electrical behavior of EMI filters is difficult to predict and cannot be explained using traditional electrical network theory.
In the last attempt, the EMI filter components were removed and placed on a small (1 cm × 1 cm) double-sided PCB with the lower side dedicated to the return plane. The modified small EMI filter PCB is placed on the main product PCB and configured between the ASIC power input pins and the module power input pins. The EMI filter PCB is connected to the main PCB ground structure through a short wire at the "chassis ground" to reduce the connection inductance. The CE measurement result with this modification produces a 40 dB suppression of the SMPS noise component. Therefore, the main analysis of the EMI filter implementation of PCB #1 is required.
The PCB #2 shown in Figure 11 was developed due to the CE failure observed in the previous PCB, and Figure 12 illustrates the CE measurements of two printed circuit boards. Obviously, PCB #2 meets the EMI requirements and provides the expected performance for electrical simulation and measurement in the test PCB.
Figure 11: Printed circuit board (board #2)
Figure 12: Comparison of conducted radiation (voltage) of #1 and #2 boards; measurement in the MW-AM frequency band
Close inspection of Figures 8 and 11 can reveal the most critical differences between the two different printed circuit boards. It should be noted that the grounding of the π filter capacitor C1 plays an important role in the effectiveness of the EMI filter. In Figure 8, referring to PCB #1, C1 and a large energy storage capacitor are connected to the same "grounded copper island". The existence of the high frequency charging current (490 kHz and its harmonics) of the energy storage capacitor is the main problem of the "grounded copper island". Further investigation of C1 using the Agilent 85024A high-frequency current probe revealed the presence of SMPS noise currents. When the EMI filter is referenced to a non-RF reference plane (that is, the chassis or "quiet ground"), a reference plane that is relatively unaffected by high-frequency RF currents is required. Obviously, "grounded copper island" cannot be called ideal grounding or chassis grounding. In order to verify the filter grounding concept, several short electrical connections were introduced between the "grounded copper island" and the chassis ground. The measured CE data reveals an amazing improvement in filter performance. As a result of the report investigation, the design of PCB #2 and the generated CE data are shown in Figure 12. Obviously, PCB #2 provides a noise rejection requirement of 40 dB. It is important to pay attention to the installation strategy and ground connection of the EMI filter capacitor.
This research examines the basic problems encountered in the implementation of simple EMI filters on printed circuit boards. Engineers face the challenge of designing products for automotive applications under harsh environmental conditions with strict EMI requirements. Analysis tools that meet the guidelines required by engineers to develop automotive products with multiple requirements may be restricted, and these requirements often conflict with each other. In this report, we outline the errors in the implementation of the π filter PCB due to manufacturing requirements that exceed best EMI practices. Engineers' adherence to low-cost technology and the ability to solve heat, EMI, manufacturing constraints, and several other major requirements determine the success of automotive products.
Cyrous Rostamzadeh, Robert Bosch LLC, Automotive Group, Cyrous.Rostamzadeh@us.bosch.com
Flavio Canavero, Politecnico di Torino, flavio.canavero@polito.it
Feraydune Kashefi, Havalan Institute of Technology, Fred.Kashefi@gmail.com
In Compliance is the main source of news, information, education and inspiration for electrical and electronic engineering professionals.
Aerospace Automobile Communication Consumer Electronics Education Energy and Power Industry Information Technology Medical Military and National Defense