Docstoc

Template Name

Document Sample
Template Name Powered By Docstoc
					                                 EMI FILTER DESIGN

                                    DESIGN TEAM 15




        Executive Summary:

        Team 15, under the guidance of BOSCH, has been assigned the task of designing an
        active electromagnetic filter to reduce two kinds of conducted noise (Differential and
        Common mode) from the power system bus of an automobile. The filter will be
        designed to handle a 13.8 V DC potential as well as a current of up to 20 A.
        Parameters such as cost, size and safety will be optimized while maintaining the
        stringent regulations specified by the FCC as well as BOSCH. This paper will not
        only explore the design process and benefits of an active (vs. passive) filter, but also
        the components (LISN, PWM) designed in order to test the active filter.




81308f86-000d-460b-974a-5901c5874ed3.doc   Page 1 of 15                      Printed March 11, 2010
TABLE OF CONTENTS
1      Introduction ..................................................................................................... 3
2      Background ..................................................................................................... 4
3     Design Specifications .................................................................................... 5
    3.1   Frequency Response ................................................................................ 6
    3.2   DC Parameters ......................................................................................... 6
    3.3   Noise Suppression .................................................................................... 6
    3.4   Desirable Specifications ........................................................................... 6
    3.5   Test Setup Specifications ......................................................................... 6
    3.6   Line Impedance Stabilization Network ..................................................... 7
    3.7   PWM Motor Driver..................................................................................... 7
4     Conceptual Design ......................................................................................... 7
    4.1  LISN: Line Impedance Stabilization Network .......................................... 8
    4.2  Pulse Width Modulator (PWM) Circuit Design: ...................................... 10
    4.3  Active Filter Design ................................................................................. 12
5      Risk Analysis ................................................................................................ 14




List of Figures
Figure 3: Test Setup for Measuring Conducted Emissions ...................................................7
Figure 4: LISN ..........................................................................................................................8
Figure 5: LISN Behavior at higher frequencies......................................................................9
Figure 7: AStable PWM Circuit ............................................................................................11
Figure 8: Monostable Circuit Design ....................................................................................11
Figure 9: Active Filter Design ...............................................................................................12
Figure 10: Basic Inductor Model ...........................................................................................14
Figure 11: Impedance of Inductor over frequency range .....................................................14
Figure 12: Capacitor Model ...................................................................................................15
Figure 13: Impedance of Capacitor over frequency range..................................................15




81308f86-000d-460b-974a-5901c5874ed3.doc                     Page 2 of 15                                      Printed March 11, 2010
1 Introduction
        With the advance of automotive engineering a variety of critical (anti-lock breaks/, air
        bags etc.) and luxury (onboard GPS) electrical loads have been implemented. The
        number of loads a vehicle employs is increasing and appears to continue this growing
        trend. The issue here deals with the limited amount of space for which these loads can
        be laid out amongst the vehicle. It is crucial that all loads located within the vehicle
        function as intended and do not interfere with other loads or itself, while in an
        increasingly congested electromagnetic environment.

        Electromagnetic compatibility did not take priority until the early 20th century, when
        the International Special Committee on Radio Interference (CISPR) created a
        document regarding equipment and procedures for testing these electromagnetic
        interferences. The growing speed and minimization of the overall size of micro-
        electronics in the late 70‟s led the Federal Communications Commission (FCC) to
        step in and begin regulating the electromagnetic energy a given device could emit.

        In modern times automotive EMC standards are anything but static.
        Manufacturers/designers must ensure not only the functionality of the product but its
        reliability given circumstance, to guarantee the safety of the customer.

        Not too long ago certain gas stations decided to spare their costumers a few minutes at
        the pump with the introduction of the „Speed Pass‟, a small cylinder that conveniently
        fits on a persons key chains. This device, in the eyes of many, seemed to be a time
        saver. Simply pull up to the pump, put gasoline into the vehicle, swipe and drive
        away…if you could. A select few who happened to own the „wrong‟ car were able to
        follow the previously mentioned procedure upon entering the gas station, with the
        exception of the final step, driving away.

        The „Speed Pass‟ is classified as a RFID (Radio Frequency Identification Devices).
        When the „Speed Pass‟ is placed within range of a device capable of reading the
        information on it, radio transmitted signals issued by the reader energizes the chip
        within the „Speed Pass‟ allowing the information to swap. Unfortunately for a select
        few, the device the „Speed Pass‟ was coming into contact with was the ignition of their
        vehicle. Radiated EMI (electromagnetic interference) was preventing the user‟s car
        from starting, but whose fault is it? The customer for using the convenient tool while
        unaware of the risk, the gas station for trying to help their customers or the designer
        for not realizing every possible device the „Speed Pass‟ was to come in contact with.
        The point being made here is that the blame at least to its full extent does not belong to
        one individual or set of individuals. What will most likely happen is the following: the
        designer will redesign the „Speed Pass‟ by means of shielding and or adding an
        electromagnetic filter to the circuit in hopes of reducing the radiated emissions. The
        project contained in this document focuses mainly on conducted emissions (versus
        Radiated), and that stories of cars not starting (as above) or blowing up at the end of an
        assembly line (due to a combination of improper grounding and fast fueling) are all
        results of EMC and should not be taken lightly.

81308f86-000d-460b-974a-5901c5874ed3.doc   Page 3 of 15                        Printed March 11, 2010
        In order to prevent these issues from occurring, it is critical that the designer of a
        product use proper shielding and or filtration techniques. This is where team 15
        comes in, with a proposed and soon to be implemented design and test setup an Active
        Electromagnetic Filter for a DC automotive cooling fan motor.

2 Background
        It was mentioned that an active filter was to be designed vs. passive. The goal here is
        to remove the large passive filter commonly used in manufacturing today, and replace
        it with active op-amp circuitry and a smaller set of passive components, essentially
        forming a hybrid.

        Filters in general have two common tasks: 1.) Removing unwanted signals and 2.)
        Enhancing wanted signals. A passive filter is essentially a filter consisting entirely of
        passive components (resistors, capacitors and inductors). These electric components
        do share a common need when performing their intended functions, the need to
        consume energy. Depending on the complexity of the circuit the power required
        could become costly. This is the primary reason why active filters are now making
        their way into circuit design.

        Active filters are a type of Analog electronic filter. Analog often refers to a
        continuous signal in which small fluctuations do matter. Active filters consist of one
        or more active components which can be used to provide gain rather then consuming.
        Active filters are now being used for two primary reasons. First, the Op-amp circuitry
        within the filter can aid in shaping the filter‟s response. In order for a passive filter to
        gain this same control one must place additional inductors into the design, which often
        leads to more electromagnetic interference (going against the purpose of the filter).
        Second, the amplifier used could essentially buffer the filter from the electrical
        components it is driving.

        The focus of this paper is to suppress conducted emissions (vs. Radiated). Radiated
        emission often propagate through free-space or any non-conductive material. In order
        for the coupling to take place the two objects need only be on an order of a couple
        wave lengths apart. A prime example of this is the „Speed Pass‟ from the introduction.
        The pass itself was never inserted into the ignition, yet placed close enough for
        electromagnetic waves to interact. Conductive emissions (our concern) on the other
        hand occur over a direct or conductive materialized path between the two devices
        (power cords, antenna terminals or even a metal casing placed around the electronic
        device). It is also fairly common for ac power cords of a given length, to act as an
        antenna, either bringing in or sending unwanted signals which could alter the output or
        functionality of the device itself or other devices in the area. The main objective is to
        suppress the (conducted) noise generated from a DC automotive cooling fan motor
        when it is turned off/on (as commonly performed in the automotive industry when
        testing the speed of a motor).


81308f86-000d-460b-974a-5901c5874ed3.doc    Page 4 of 15                        Printed March 11, 2010
        There is one final concept that must be touched upon before
        getting into the design specs. The noise from the DC motor
        exists in two modes, Differential and Common. Differential
        mode current flows in opposing directions in regards to
        ground and VDD as shown in the circuit above on the right.                   Figure 1: Differential
                                                                                            Current
                                              The common mode current will travel the same
                                              direction across wires, starting at the source and
                                              moving towards the load. This is shown in the
                                              circuit to the left. The referenced ground provides a
                                              path back to the noise source for the common mode
                                              current. The tricky part comes with the application
                                              of measurements and knowing how to isolate the
                  Figure 2: Common mode       two modes. A solution has been implemented and
                           Current            will be further explained below, in the Conceptual
                                              Design of a LISN, portion of the lab.



3 Design Specifications
The active filter design must meet or exceed a stringent set of test requirements. Because
these parameters must be precisely measured, the design specification shall be divided
into two segments; the first segment describes the active filter specifications, and the
second describes the testing setup specifications.


3.1     Active Filter Feasibility Specifications

The active filter design is targeted to reduce conducted emissions on the power system
bus of an automobile. Bosch provided the primary specifications that the active filter
design must meet. The parameters specify the frequency range that the filter must
operate on, the DC parameters, and the level of noise suppression that the filter must
achieve. Table 1 summarizes these requirements.

                                    Table 1: Active Filter Design Requirements
           Frequency Range                           150kHz – 108 MHz
           Input DC Voltage                          13.8 V
           Maximum Load Current                      20 A
           Common Mode Insertion Loss                60 dB or greater at 250kHz
           Differential Mode Insertion Loss          70 dB or greater at 250kHz
           Input Impedance Range                     0.1 Ohms – 100 Ohms
           Output Impedance Range                    0.1 Ohms – 100 Ohms


81308f86-000d-460b-974a-5901c5874ed3.doc       Page 5 of 15                       Printed March 11, 2010
3.2     Frequency Response
        The frequency response range of interest includes all international and domestic radio
        broadcast frequencies, from AM through FM broadcast bands. The noise suppression
        requirements are designed in part to protect automobile receivers from harmful
        interference. Satisfaction of this requirement is required for the design to be feasible.

3.3     DC Parameters
        The active filter is planned for use on the power supply bus of an automobile. Further,
        it will be suppressing noise generated by PWM motor/controller assemblies, like
        power steering pumps. Thus, the filter must operate on and pass (with little power
        loss) a 13.8 V DC potential, with currents as high as 20 Amps. This requirement must
        be satisfied by the filter design.

3.4     Noise Suppression

        The active filter must introduce an insertion loss for common mode noise of at least 60
        dB (measured at 250 kHz). Also, the filter must introduce an insertion loss for
        differential mode noise of at least 70 dB (again measured at 250 kHz). Significant
        insertion loss is also desired along the rest of the frequency range of interest (described
        above). The insertion loss shall be measured with an input and output impedance
        match of 50 ohms; however, the filter must also suppress noise when the input and
        output impedances are significantly mismatched (0.1 through 100 Ohms).

3.5     Desirable Specifications
        Aside from the primary requirements listed earlier, the active filter also should satisfy
        a variety of secondary design constraints. Because of the targeted application (in
        automobiles), the cost of the device should be minimized. Also, the device reliability
        should be considered, both in choosing components and circuit structure. Safety is an
        important consideration, as failure of the filter could, if not designed properly,
        endanger the automobile‟s occupants.

3.6     Test Setup Specifications

        To test the active filter for compliance with the feasibility requirements, a test setup
        will be required. Figure Y shows a test setup for measuring the conducted emissions
        of the active filter. A more extensive discussion of this test setup is provided in
        CISPR 25, a standard from the IEC. The Department of Engineering at MSU is
        providing access to a suitable spectrum analyzer. The 13.8V source will be a suitable
        automotive battery. The motor shall be a PMDC type, rated for 13.8V at 20 Amperes.
        The design team must fabricate the PWM driver for the motor, as well as the Line
        Impedance Stabilization Network, or LISN.
81308f86-000d-460b-974a-5901c5874ed3.doc   Page 6 of 15                        Printed March 11, 2010
                                           LISN

             13.8 VDC                                           PWM
                                                                                 MOT
              Source                                            Driver

                                           LISN


                                                                   Ground Plane
                              Spectrum                        Faraday
                              Analyzer                        Cage

                              Figure 3: Test Setup for Measuring Conducted Emissions




3.7     Line Impedance Stabilization Network

        The LISN devices used for the test setup shall be fabricated to comply with the
        standards outlined in the IEC standards document CISPR 25. Briefly, the device shall
        couple the DC source to the PWM driver with a low impedance, and decouple the
        spectrum analyzer from the spectrum analyzer at DC and low frequencies. In the
        frequency range of interest, the LISN shall effectively disconnect the battery from the
        PWM driver, and present the spectrum analyzer with a 50 ohm impedance suitable for
        measuring conducted emissions. Important design considerations for the LISN are
        cost and compliance with the CISPR 25 specifications.

3.8     PWM Motor Driver

        The PWM motor driver, along with the motor, shall provide the noise source for the
        conducted emissions test setup. The PWM driver shall switch at a frequency of
        10kHz, and provide for variable duty cycle and rise/fall time (to change the generated
        noise harmonics). The output driver must be capable of switching 20 A currents at
        13.8 VDC, with an additional safety margin to prevent failures. Cost, reliability, and
        safety, along with proper noise generation, are primary considerations for this design.



4 Conceptual Design


81308f86-000d-460b-974a-5901c5874ed3.doc       Page 7 of 15                            Printed March 11, 2010
4.1     LISN: Line Impedance Stabilization Network

        A LISN is simply a low pass filter placed between the power supply and the PUT
        (product(s) under testing). Generally speaking the LISN must maintain characteristic
        impedance to the PUT and isolate the PUT from unwanted Radio Frequency signals
        (AC and DC) while allowing the desired voltage and current to propagate towards the
        PUT.

        The LISN is capable of performing any of the following four tasks:

                Provide the product(s) under testing with an adequate power supply.

                Must focus and feed noise disturbances through the measurement points.

                Prevent external noise from near by (connected) devices which may take part
                 in the testing setup.

                Provide constant impedance with respect to frequency which allows for
                 consistent measurements from location to location along the circuitry.

        The circuits shown below (Figures 3, 4 and 5), prior to simplification, taken from
        module 11 of the course notes written by a Michigan State EM research team under
        the supervision of the National Science Foundation GOALI program, appears to be
        (with minor alterations) suited for our design specifications




                                             Figure 4: LISN
        Looking first at the inputs/outputs one can count a total of three. For the purpose of
        the filter testing only two inputs/outputs are needed (in regards to the LISN). The
        green wire will be removed being as we are only concerned with DC. Let us begin by
        focusing on the 50uH inductors placed on the phase and neutral wires. At higher
81308f86-000d-460b-974a-5901c5874ed3.doc   Page 8 of 15                      Printed March 11, 2010
        frequencies these passive components will act as open circuits and prevent any
        external noise produced by the power supply from interfering with the circuitry (task
        #3 of the 4 listed above). The 1uF capacitors are short circuited at high frequency and
        will act as a path, for noise already existent in the circuitry, diverting it from the
        measurements. Next come the 1kΩ resistor and .1uF capacitor. In the case a DC
        surge is present within in the schematic the .1uF capacitor will absorb it, preventing an
        build up at the input of the test receiver. The 1kΩ resistor will remain a failsafe for the
        .1uF capacitor in the event it was to discharge and the 50Ω resistor was removed. The
        50Ω is essentially the impedance seen by spectrum analyzer when looking into the
        cable input of the LISN. So the spectrum analyzer will be hooked up to the LISN
        where the 50Ω dummy loads are placed, and will measure the spectral composition of
        the circuit. Both an AC and DC spectrum analyzer are provided in the laboratory at
        Michigan State, and do not need to be designed.

        Lets take a look at the LISN circuit over a range of high frequencies (capacitors
        becomes shorts and inductors become opens).




                                   Figure 5: LISN Behavior at higher frequencies



        As mentioned before the green wire shall be ignored. Shown above is the
        directionality of the common/differential mode currents (along with the phase/neutral
        currents) which will need to be suppressed. In order for the suppression of these two
        noises, they must both be isolated and measured one at a time. This can easily be done
        using two LISNS. One will be connected from the positive terminal of the power
        supply (13.8V battery) to ground, and the other LISN from the negative terminal of
81308f86-000d-460b-974a-5901c5874ed3.doc       Page 9 of 15                        Printed March 11, 2010
        the battery to ground. This will provide the total current on each of the two wires
        (phase and neutral). If focusing just on the common and differential currents, the
        following algebraic maneuvers can be performed to get either one of the two currents.
        For the understanding of these maneuvers take note of following: the common mode
        current is moving in the same direction with respect to the two wires, and the different
        mode current runs in opposition amongst the two wires. Adding the current on both
        wires will force Id (differential mode) to cancel out leaving 2*Ic (common mode).
        Simply divide by two and you have now achieved the measurement for the common
        mode current. Subtracting the two wire‟s currents in the same fashion will yield Id
        (differential mode).

        Common mode choke coils are also a common way of separating the two currents. In
        simplicity it is a wire wound about a ferrite core. When common mode currents enter
        the wire a magnetic field is produced causing impedance which blocks the currents
        path (inductor). On the other hand, if differential currents runs through the wire, the
        magnetic field produced by each current will cancel out (Id1 moves in opposing
        direction then Id2) essentially equaling a wire. Although this technique is straight
        forward it can become costly (depending on the number of chokes needed) not to
        mention the inductances produced by the coiled wire could drastically effect the
                                                     measurements.

                                                          The schematic seen on the left appears to
                                                          be the LISN design that will be used to test
                                                          the active filter, when the time comes. It is
                                                          slightly more simple then the above model,
                                                          and not does utilize the green wire. As of
                                                          now the only component not existent in the
                                                          picture is a possible metal casing that will
                        Figure 6: LISN Design             exist around the LISN, providing a path for
                                                      the common mode current.

4.2     Pulse Width Modulator (PWM) Circuit Design:
        Pulse Width Modulation deals with the modulation of duty cycle to control the amount
        of power delivered to the load. According to the specifications provided by the
        sponsor, a 10 kHz PWM circuit is needed to operate the DC motor.

        In order to design the PWM circuit, 555 timer IC will be used. To design a PWM
        circuit with variable duty cycle, potentiometer to vary the duty cycle and the frequency
        of the pulse generated.

        An astable circuit will be used to control the frequency of the PWM. The design of the
        circuit is shown in the Figure 4 below. Resistor RA will be replaced with a
        potentiometer to obtain the desired variable frequency circuit.




81308f86-000d-460b-974a-5901c5874ed3.doc        Page 10 of 15                       Printed March 11, 2010
                                           Figure 7: AStable PWM Circuit



        A monostable circuit will be designed to achieve the variable duty cycle circuit. The
        input of the monostable circuit will be connected to the output of the astable circuit.
        The design of the monostable circuit is shown in the Figure 8 below




                                       Figure 8: Monostable Circuit Design
        By Connecting the two circuits (astable and monostable), we can not only achieve a
        variable duty cycle, but also a variable frequency range. With the current design in
        mind, it is expected to attain a frequency range of 7.5 kHz to 13.5 kHz and duty cycle
        of 33% to 91%.

81308f86-000d-460b-974a-5901c5874ed3.doc       Page 11 of 15                 Printed March 11, 2010
4.3     Active Filter Design

        A couple different designs for the active electromagnetic filter are currently being
        analyzed. One based on the topology of a Low-pass filter and the other a Band-pass
        filter. It was noted in the design specifications that a frequency range, 150 kHz-
        108MHz, is needed which is why a band-pass filter is being considered (upper and
        lower end cut-off frequency). The band-pass filter would require a higher
        mathematical understanding and could become extremely tedious. However at the
        same time offer a much higher precision due having multiple cut off frequencies. A
        low-pass filter would be a lot easier to implement however would end up letting in
        signals corresponding to frequencies lower then 150 kHz. Regardless of difficulty
        team 15 will utilize which ever design yields the most promising results.

        Another design spec was designing the filter to vary with a 60 dB/dec slope. To
        produce such quality, multiple filters (three to be exact) must be placed in series. A
        physical representation of this is seen below in Figure 1.




                                                       Z4

                                                                                Z5

                                                                                    Vdd
                                                                                     8



           V1                                                               3
                                                                                +         1                   V2
                           Z1                          Z3                   2
                                                                                -
                                   Z2
                                                                                     4




                                           Figure 9: Active Filter Design
        This type of setup is commonly known as Multiple Feedback Filters. The circuit
        above as mentioned in the background would be a hybrid (passive/active). Z1-Z5
        represents the passive portion of the design, and consists of resistors and capacitors.
        The op-amp here will provide gain to the filter via feedback and will not consume
        power (as its passive equivalent had). The General Equation used to analyze the
        Multiple Feedback Filters is shown below.




81308f86-000d-460b-974a-5901c5874ed3.doc       Page 12 of 15                                  Printed March 11, 2010
         V2             Y1  Y3
            
         V1 (Y1  Y2  Y3  Y4 )Y5  Y3  Y4


        The values for the above equation will be placed into more detail once the team has
        decided on the filter to be used for the project (Low-pass / Band-pass). In the event
        the Band-pass filter is proven to be dominant the following transfer function can be
        used.



                        w 
                  H BP  O  s
                        Q 
         V2
                        O
         V1         w 
              s 2   O  s  wO
                                 2
                    Q 
                     O
        To equate the numerators, it is necessary to select either Z1 or Z3 to be a capacitor,
        because the power of “s” is still one. Pick Z3 as a capacitor and Z1 as a resistor, then
        the type of Z4 and Z5 becomes very obvious; Z4 is a capacitor, and Z5 has to be a
        resistor now. And Z2 could be either a resistor or a capacitor, let Z2 be a resistor.
        Then the transfer function is represented as shown Figure 3 below. And its simplified
        form let us determine values of impedances by comparing Figure 3 to Figure 4. It
        must again be stressed that these equations only account for a 20dB/dec slope and a
        series connection must be made to accommodate a 60dB/dec slope



                            1
                                sC3
         V2                 R1
             
         V1  1       1            1
                 R R  sC3  sC4  R  s C3C4
                                
                                           2

                 1    2           5
                               1
                           s
                             R1C4
         
                   1  1  1             1
            s 2   s    
                   R  C C 
                   5  4   3            R1R2 
                                          R R 
                                  C3C4 R5        
                                           1   2 



        A low-pass filter design would follow closely to this procedure however as mentioned
        prior, the mathematical computation would become less intense.



81308f86-000d-460b-974a-5901c5874ed3.doc   Page 13 of 15                     Printed March 11, 2010
5 Risk Analysis
        Since the EMI filter deals with the high frequency, special care will have to be
        undertaken while designing the PCB layout. This is due to the reason that at higher
        frequency, the capacitors and inductors change their impedance and their behavior
        which needs to be accounted for while designing the circuit.

        The passive components tend to have variable impedance over the wide range of
        frequencies. While simulating the active filter in PSpice, the actual models of resistors,
        capacitors and inductors need to be placed. The actual model of the inductor is given
        by




                                           Figure 10: Basic Inductor Model




                               Figure 11: Impedance of Inductor over frequency range




81308f86-000d-460b-974a-5901c5874ed3.doc        Page 14 of 15                          Printed March 11, 2010
        The actual model of the capacitor is given by




                                           Figure 12: Capacitor Model




                              Figure 13: Impedance of Capacitor over frequency range



        If ideal capacitors and inductors are used to simulate the behavior of the circuit, the
        results will be deviating from the actual circuit response. Hence, special care needs to
        be taken to ensure maximum circuit performance.




81308f86-000d-460b-974a-5901c5874ed3.doc      Page 15 of 15                            Printed March 11, 2010

				
DOCUMENT INFO