Chapter 12 Digital Circuit Radiation Electromagnetic Compatibility Engineering by Henry W. Ott
Forward Emission control should be treated as a design problem from the start, it should receive the necessary engineering resources throughout the design process. This chapter provides a method for predicting the radiated emission as a function of the electrical characteristics of the signals and the physical properties of the system. Knowing the parameters that affect radiation helps to develop techniques to minimize it. JHLin, EMC; Digital Ckt Radiation 2
Forward Differential-mode radiation from PCB Although these signal loops are necessary for circuit operation, their size and area must be controlled during the design process to minimize the radiation. JHLin, EMC; Digital Ckt Radiation 3
Forward Common-mode radiation from system cables Common-mode radiation is often harder to understand and control. JHLin, EMC; Digital Ckt Radiation 4
Differential-Mode Radiation Differential-mode radiation can be modeled as a small loop antenna. E 131.6 10 f AI dm sin (V/m) r 16 2 1 JHLin, EMC; Digital Ckt Radiation 5
Differential-Mode Radiation For a loop whose perimeter equals a wavelength, the radiation pattern rotates by 90 so that the maximum radiation occurs normal to the plane of the loop. All small loops having equal area radiate the same regardless of the shape. Correcting for the ground reflection and assuming E 263 10 f AI dm (V/m) r 16 2 1 Differential-mode radiation can be controlled by 1. Reducing the magnitude of the current 90 2. Reducing the frequency or harmonic content of the current JHLin, EMC; Digital Ckt Radiation 6
Differential-Mode Radiation 3. Reducing the loop area Loop Area The primary way to control differential-mode radiation is to minimize the area enclosed by current flow. Placing signal leads and their associated ground-return leads close together. Loop Current The current is seldom known accurately; therefore it must be modeled, measured, or estimated. Fourier Series Because digital circuits use square waves, the harmonic content of the current must be known before calculating the emission. JHLin, EMC; Digital Ckt Radiation 7
Differential-Mode Radiation Assuming the rise time equals the fall time I n n tr sin( ) sin( n d) 2Id T n d n t r T For a 50% duty cycle (d=0.5), the first harmonic has an amplitude I and only odd harmonics are present. 1 0.64I JHLin, EMC; Digital Ckt Radiation 8
Differential-Mode Radiation Differential-mode radiated emission envelope vs. frequency To minimize the emission, it is desirable to slow down the rise time of the signal as much as functionally possible. JHLin, EMC; Digital Ckt Radiation 9
Differential-Mode Radiation The spectrum for a 6-MHz, 4-ns rise time, 35-mA clock signal in a 10- cm 2 loop. JHLin, EMC; Digital Ckt Radiation 10
Differential-Mode Radiation Radiated Emission Envelope It can easily be estimated if the frequency, peak current, and rise time of a signal as well as the loop area are known. Because the shape of the radiated emission envelope is known, only the amplitude of the radiation at the fundamental frequency needs to be calculated. JHLin, EMC; Digital Ckt Radiation 11
Controlling Differential-Mode Radiation Board Layout The most critical loops should be individually analyzed; however, most other noncritical loops can be controlled by just using good PCB layout practices. The most critical loops are those that operate at the highest frequency and where the signal is periodic. 12
Controlling Differential-Mode Radiation Clock signals should be routed first on a PCB. Route them in a manner that produces the absolute smallest loop area possible. 1. The length of the clock trace. The number of vias. 2. On a multilayer PCB,. Many other strobe or control signals are also periodic. In a microprocessor-based system, some critical, periodic signals are CLK, ALE, RAS, and CAS 3. The clock circuitry should be located away from the I/O cables or circuitry. 4. To minimize crosstalk, clock traces should not be run parallel to data bus or signal leads for long distances. Address buses and data buses are the second concern after clocks. JHLin, EMC; Digital Ckt Radiation 13
Controlling Differential-Mode Radiation 1. On multilayer PCBs,. 2. On double-sided boards, at least one signal return (ground) trace should be provided adjacent to each group of eight data or address leads. This return trace is best placed adjacent to the least significant bit. Line and bus drivers can also be offenders because they often carry high currents. However, they generate broadband noise. 1. Located close to the lines they drive. 2. Drivers for cables leaving the PCB should be located close to the connectors. 3. Line driver ICs used to drive off-board loads should not be also be used to drive other circuits on the board. JHLin, EMC; Digital Ckt Radiation 14
Controlling Differential-Mode Radiation Over the last 5 or 10 years, the emission problem has increased 100- fold, but our ability to deal with it had increased at most twofold. Other unconventional means to reduce the emission: 1. canceling loops; 2. spread-spectrum clocks. Canceling Loops The layout of Fig. 12-8B will radiate 20+ db less than that of Fig. 12-8A. JHLin, EMC; Digital Ckt Radiation 15
Controlling Differential-Mode Radiation Dithered Clocks Another approach to reducing the radiated emission is to spread the emission out in the frequency spectrum. The spreading of the emission can be accomplished by using a dithered, or spread-spectrum clock --- basically frequency modulating the clock. Using an optimum design, the emission can be reduced about 15 db by this method. The degree of reduction is a function of the dithering waveshape as well as the frequency deviation. Optimum modulating waveform for a spread-spectrum JHLin, EMC; Digital Ckt Radiation 16
Controlling Differential-Mode Radiation Most clock dithering circuits use a triangular waveform. The specification of a typical spreadspectrum clock might read as follows: 1. Modulating waveform: Triangular 2. Modulation frequency: 35 khz 3. Frequency deviation: 0.6 % 4. Modulation direction: Downward 5. Emission reduction: 12 db JHLin, EMC; Digital Ckt Radiation 17
Controlling Differential-Mode Radiation Some systems cannot tolerate a spread-spectrum clock, but most can. Many PCs and printers use spread-spectrum clocking. If absolute real-time accuracy is required, a spread-spectrum clock may be a problem. Most phase locked loops, however, work just fine with a spread-spectrum clock. Two basic approach to clock dithering: center spread and downward spread. The reduction in emission is the same in either case. The downward spread is less likely to cause timing margin problems, because the clock frequency is not increased. JHLin, EMC; Digital Ckt Radiation 18
Controlling Differential-Mode Radiation Frequency spectrum of the third harmonic of a 60-MHz clock with and without dithering Reducing the loop area, or providing canceling loops, only controls the differential-mode emission and has no effect on the commonmode emission. JHLin, EMC; Digital Ckt Radiation 19
Controlling Differential-Mode Radiation Dithering the clock frequency reduces both radiation modes. JHLin, EMC; Digital Ckt Radiation 20
Common-Mode Radiation For a short dipole antenna: 4 10 ( )sin 7 f I E cm (V/m) r For a real dipole antenna, the current goes to zero at the open ends of the wire. In practice, a more uniform current distribution can be achieved if capacitor loaded or top-hat antenna: This configuration is approximated when the antenna (cable) connects to another piece of equipment. 21
Common-Mode Radiation The common-mode radiation can be controlled by: 1. Reducing the magnitude of the common-mode current 2. Reducing the frequency or harmonic content of the current 3. Reducing the antenna (cable) length The primary method of minimizing the common-mode radiation is to limit the common-mode current. Common-mode radiated emission envelope vs. frequency JHLin, EMC; Digital Ckt Radiation 22
Common-Mode Radiation The common-mode emission is more likely to be a problem at low frequencies, and differential-mode emission is more likely to be a problem at high frequencies. Idm I cm 48 10 f A 6 A common-mode current of a few As can cause the same amount of radiation as a few mas of differential-mode current. JHLin, EMC; Digital Ckt Radiation 23
Common-Mode Radiation For a long cable ( 4), Eq. 12-7 can be corrected for by using the /4 prediction at all frequencies above where the cable is a /4 long. JHLin, EMC; Digital Ckt Radiation 24
Controlling Common-Mode Radiation The common-mode current can be thought of as a control knob on the radiated emission. The net common-mode current on a cable can be controlled by the following techniques: 1. Minimizing the common-mode source voltage, normally the ground potential. 2. Providing a large common-mode impedance (choke) in series with the cable. 3. Shunting the current off the cable. 4. Shielding the cable. 5. Isolating the cable from the PCB ground, for example, with a transformer or optical coupler. JHLin, EMC; Digital Ckt Radiation 25
Controlling Common-Mode Radiation Today, many types of I/O signals are at frequencies as high as or in some cases actually higher than the clock frequency [e.g., USB or Ethernet], thus considerably complicating the requirement that the common-mode suppression technique not interfere with the desired signal. Common-Mode Voltage Also, the importance of avoiding slots in ground planes cannot be overemphasized. The farther away that the circuit-to-chassis ground connection is from the point where cables terminate on the board, the more likely it is that there is a large noise voltage between the points. The reference or return plane for common-mode currents on the external cables is the enclosure. JHLin, EMC; Digital Ckt Radiation 26
Controlling Common-Mode Radiation The circuit ground in the I/O area of the PCB should be at the same potential as the enclosure. The two grounds must be connected together in this area. To be effective, the impedance (inductance) of this connection must be extremely low across the complete frequency range of interest, which usually requires multiple connections. Even when the ground voltage is minimized, it is usually not sufficient to control the common-mode radiated emission. Cable Filtering and Shielding How the cable shield termination affects the shielding effectiveness of a cable is shown in Fig. 12-13. JHLin, EMC; Digital Ckt Radiation 27
Controlling Common-Mode Radiation A B C D JHLin, EMC; Digital Ckt Radiation 28
Controlling Common-Mode Radiation In Fig. 12-13D, the higher frequency harmonics will choose the path that involves the outside surface of the shield, and they will radiate, whereas the lower frequency harmonics will choose the path that involves the pigtail and will not radiate. In Fig. 12-13E, the shield on a shielded cable is unshielded. PCB mounted connector backshells must make 360 contact with the enclosure. E 29
Controlling Common-Mode Radiation Filtering of the I/O cables can be accomplished by adding a high impedance in series with the common-mode noise (e.g., a common-mode choke or ferrite core), or by providing a lowimpedance shunt (a capacitor) to divert the common-mode noise to ground. Separate I/O Grounds The I/O ground plane should have multiple connections to the enclosure to minimize its inductance and provide a low impedance connection. 30
Controlling Common-Mode Radiation The bridge should be wide enough to accommodate the required number of traces plus 20 times the height of the trace above the ground plane on each side. The I/O ground should be thought of as an extension of the enclosure. The PCB s power plane should not be allowed to extend into the I/O ground area. In large systems, the I/O ground might even be a separate PCB located at the cable entrance and containing only connectors and I/O cable filter capacitors. The clean I/O ground should be located at the point where the cables leave/enter the system. The effectiveness of the I/O cable filter capacitor depends on the common-mode source impedance of the driving circuits. 31
Controlling Common-Mode Radiation Dealing with Common-Mode Radiation Issues It is a good practice at the beginning of a new design to make a list of all the cables, including power cables, and document what technique will be used to reduce or eliminate the common-mode current on that cable. During the prototype stage, the common-mode currents on all cables can be measured using the technique in Sec. 18.3, and the results compared with the appropriate limit determined by Eq. 12-8. JHLin, EMC; Digital Ckt Radiation 32