Radiation from a PCB with Coupling between a Low Frequency and a Digital Signal Traces Naoto Oka Chiharu Miyazaki Shuichi Nitta* MITSUBISHI ELECTRIC Corp. *Faculty of Technology Information Technology R&D Center Tokyo University of Agriculture & Technology 5-l-l Ofuna,Kamakura,Kanagawa,247,Japan 2-24-16 Nakacho,Koganei,Tokyo,184,Japan Abstract: Radiation from a PCB with coupling between a low frequency signal and a digital signal traces is studied in this paper. EM1 increasing by coupling between these traces on a digital PCB is shown experimentally and theoretically. These traces are routed on the same and the different signal planes in the multilayer PCB. Radiated EM1 is evaluated with calculation of crosstalk. It is shown that evaluation of EM1 level by crosstaik is useful to decide PCB s structure for EM1 reduction of a high-density assembled PCB. It is effective for EM1 reduction to separate a low frequency signal trace from a high-speed digital signal by ground plane. IN7230Duc~0N Downsized electronic products require high-density assemblies of the printed circuit boards (PCBs). These PCBs usually have a ground plane, a power plane and signal planes. Digital signal traces (clock and data traces) and low hequency signal traces (analog signal traces, reference voltage traces and traces for analog power supply) are placed on signal planes. These traces are closely routed to each other on the high-density assembled PCB. If digital signal traces couple to low frequency signal traces and their RF energy couples to I/O circuits, this coupling results in radiated EM1 [ 11. In the case that these traces were routed on the same signal plane in a PCB with wide ground plane, a study on EM1 level was reported [2]. Calculation results of radiation from a cable connected to a PCB were shown but a measurement result of radiation from the PCB was not shown in the above paper. In many cases of the high-density assembled PCB, traces are routed in parallel to each other on the different several signal planes in the PCB with various size of ground plane. In this paper, coupling between traces routed on the same and the different signal planes is studied. It is shown by measurement results that a coupled signal line causes drastic increase of EM1 horn a digital PCB. Radiated EM1 is evaluated with calculation of crosstalk. Width of PCB s ground plane is taken into account in this calculation. It is shown that evaluation of EM1 level by crosstalk is useful to decide PCB s structure for EM1 reduction of a high-density assembled PCB. From the viewpoint of practical application, it is effective for EM1 reduction to separate a low eequency signal trace from a high-speed digital signal by ground plane. EPERWENTALMODEL An example of the PCB studied in this paper is.shown in Figure 1 (1). Digital circuits and low frequency circuits are mounted on the PCB (a). Digital circuits have high-speed clock and data signals. Low frequency circuits are analog signal, analog power supply, reference voltage, control signal circuits and so on. A cable assembly connects signal traces &om low frequency circuits to the PCB(b). Low frequency signal traces and digital signal traces are closely routed to each other on the high-density assembled PCB. RF energy of the digital signal couples to the low frequency signal
traces. A simple model of this PCB is shown in Figure l(2). This board has two signal traces and a ground plane. One trace is a source trace (a model of a digital signal trace). The other trace is a victim trace (a model of a low frequency signal trace). The source trace is connected to a digital IC at the one end and terminated with a resister at the other end. The victim trace is terminated with a resister at the one end and is extended to the outside of the PCB as a cable section. These traces run in parallel to each other and make a coupling section. Cable PCB (2): The victim trace is routed just above the source trace. PCB (3): The victim trace is routed just above the source trace. PCB (4): The source trace is routed just above the victim trace. PCB (5): The source trace and the victim trace are on the same signal plane. 155mm 1 1 Oscjllator(25MHz) 4 B (sourcei330, (I ) (Viz:e Frequency circuit PCB(a) (1) An example of the P CB PCB(b) Victim Trace 74ALso4 f I I 4 4 50 ;r 100 100 (1) Top view of the experimental model Dielectric Source Victim Material Trace / Trace _ k Source Trace (2) A simple model of the PCB Cable Section Coupling Section Figure 1. An example of the PCB and a simple model. L Ground Plane PCB(l) PCB(2) PCB(3) - :Source Trace n :Victim Trace PCB(4) PCB(6) (2) Cross section of coupling section Experimental models of this PCB are shown in Figure 2. PdB (1) is used in order to measure radiation from a PCB without a victim trace and a cable section. Therefore this PCB has only a digital signal trace of 1OOmm length. PCBs (2)-(5) are used in order to measure radiation from a PCB with a victim trace and a cable section. Lengths of the coupling section and the cable section are 50mm and loomm, respectively. PCBs (2)-(5) have different cross-sectional structure of the coupling section. Figure 2. Experimental model of the PCB. In each of these PCBs, digital IC is 74ALSO4 logic IC and oscillator signal is fed to an input of this IC. The oscillator signal is a periodic trapezoidal wave of 25MHz fundamental frequency, that is, the signal of 25MHz-clock frequency. Width of signal traces is equal to 0.5mm. The PCB s dielectric material has specific permittivity E r =4.7. 636
EMI INCREASEDBY A COUPLED SIGNAL~LINE(V ICTIM) Radiation from the PCBs (l)-(5) is measured according to the EMI Sm-measurement method in the EM1 anechoic chamber. This measurement method is based on the EM1 measurement standards. The PCB is placed on the table of 0.3m height and the measurement results express maximum electric field strength in consideration of height pattern and directivity. Electric field of horizontal polarization is studied in this paper. Measurement results of radiation from the PCBs are shown in Figure 3. -o-pcb(1) -+PCB(2) +PCB(3) *PCB(4) +-PCB(5) 60, I Frequency [#Hr]. Figure 3. Measurement results of radiation fkom the PCBs. INFLUENCE OF RADIATION FROM A CABLE SECTION A calculated result of radiation from the cable section on the PCB (2) and a measurement result of radiation ikom this PCB are shown in Figure 4. First step of in this calculation, radiation from current on the cable section was calculated by equation (2). This equation is derived from equation (1)., Ex,_ 2@- 1o-71,1 r Ex : Electric field strength [v/m]. k : Wave number in free-space [radlm]. r : Radial distance from origin [ml. I : Signal line length [ml. I(x) : Current distribution on the cable section, located on the x axis [A]. f : Frequency [Hz] I; :Current measured at the center of the cable section [A]. Id -c- Calculated --c Measured 3 60, I (2) Maximum level of PCB (1) is, ;33mV/m (225MHz). On the other hand, maximum levels of PCBs.(2)-(5) are from 49 to 54dBuVlm (35OMHz). From the above discussion, it can be understood that a coupled signal trace with a cable section causes drastic increase of EM1 from a digital PCB. In the case that a victim trace was routed just above a source trace (PCB (2)), maximum level of EM1 was observed. The above phenomena were observed in the case that coupling and cable sections were short (50mm and loomm, respectively). It is predicted that the longer coupling or cable section increases radiation from the PCB furthermore. Q Frequency [MHz] Figure 4. Calculated and measured radiati:? the PCB (2)... @om -. The PCB and a cable section were placed in rectangular coordinates as shown in Figure 5. Next, the height pattern in -the EM1 3m measurement method, that is, the influence of the anechoic
chamber s ground plane was corrected by using image theory. Current I, was measured by using current transformer at the center of the cable section. terminal voltage of the source trace (output of the IC) and V4 is a voltage on the victim trace at the end of the coupling section (opposite side of the termination). In this paper, crosstalk is defined by V4N1, that is, far-end crosstalk. Mcrostrip Line Coupling Section Vl v4 zc v2 Figure 6. Crosstalk calculation model of the PCB. Figure 5. Position of the PCB in rectangular coordinates. Current distribution on the cable section contributes to the increase of EM1 from the PCB. Coupling between the source and the victim traces causes this current distribution. Therefore crosstalk from the source trace to the victim trace is evaluated in the next section. The measurement result is higher than this calculated result because of contribution of current distribution on the ground plane. Magnetic field measured on the ground plane of PCB (2) was 20dB higher than that of PCB (1) at 350MHz. This result implies that large ground plane increases radiation. (Study of this is the future EMI EVATJJATION issue.) BY CROSSTALK Radiation Tom the PCB model is able to evaluate by applying numerical analysis. In the case of microstrip structure PCB with thin dielectric substrate, many short segments are required and therefore long time is necessary for computer calculation. In practice, some of PCBs are of O.lmm thickness of dielectric substrate. In this study, radiation from the PCB is evaluated by using crosstalk from the source trace to the victim trace. The circuit model of the PCB for this crosstalk calculation is shown in Figure 6. Vl is an input Zc is an impedance of termination of the victim trace, that is, input impedance of the cable section. This Zc is calculated by the moment method with wire grid model. A model for calculation of calculated Zc are shown in Figure 7. /+Tzr+ F200.z 160 100 50 (1)Model for calculation Voltage Skrce and result 1o Segmen 0-4000 400 600 800 1000 Frequency[MHz] (2) Calculated Zc (R+jx). Figure 7. A model for calculation of Zc (input impedance of the cable section) and calculated result.
Capacitance and inductance matrices in the Calculation results of crosstalk by easier method coupling section are calculated by the two are shown in Figure 9. In these results, by dimensional boundary element method. This substituting a resistance (330 Q) for impedance Zc, method can treat various width of the ground plane calculation can be done simply. The results in and various cross-sectional structure of the PCB. Figure 9 are similar to the results in Figure 8. So, The voltage V4 is calculated by a circuit simulation EM1 reduction effect is simply evaluated by this too1 [3]. method. Absolute level of radiation from the PCB model can be predicted with the voltage V4 and the moment EMI REDUCTION method without microstrip structure of the PCB model shown in Figure 7. As the above-mentioned, Configuration of layer in the PCB in order to fist of all, radiation from the PCB is evaluated by evaluate effect of EM1 reduction is shown in Figure using crosstalk. 10. In the PCB (6), the source and the victim traces Calculation results of crosstalk on the experimental are separated by ground plane. In the PCB (T), model are shown in Figure 8. From the results in these traces are separated by conductive plane Figure 2 (PCB (2-5)) and Figure 8, it is recognized (100*95mm: power plane or ground plane for analog that EM1 reduction effect can be evaluated by circuit is available in practice). In the PCB (B), the crosstalk. distance between these traces is Smm. 0 ';;i -5 g -10 j -15 g -20 CJ -25-30 t PCB(2) + PCB(3) * PCB(4) -e- PCB(5) o 100 200 300 400 500 600 700 800 Frequency [MHz] Figure 8. Calculation results of crosstalk on the experimental model (1). ; -5 4pCB(2) +PCB(3) *PCB(4) -PcB(5) g -10 3-15 j -20-25 -30 ' Frequenoy[MHzl Figure 9. Calculation results of crosstalk on the experimental model (2). Conductive Plane I o :Source Trace :Victim Trace PCB(6) PW7) PCB(8) Figure 10. Cross section of the PCBs for evaluation of EM1 reduction. Calculation results of crosstalk with the method used in Figure 9 are shown in Figure 11. These.PCBs have good performance for EM1 reduction. Measurement results of radiation horn these PCBs are shown in Figure 12. These PCBs shown in Figure 10 also have good performance in these results. Radiation from these PCBs with the cable section is almost equal to that from the PCB without the coupling and the cable sections. From these results, it is recognized that EM1 reduction effect is simply evaluated by crosstalk. For the PCB (S), it is difticult to obtain enough space in the highdensity assembly because of small PCB s area. Configurations of PCB (6) and (7) are suitable for
the high-density assembled PCB. EMI increase that the PCB (7) shows at 750 MHz is caused by resonance with conductive and ground planes. +PCB(P) -a~-pcb(~) *PCB(7) +pcb(8) Frequency[dB] Figure 11. Calculation results of crosstalk of the PCBs. tpcb(z) -m-pcb(0) *PCB(7) -P'S(8) 2 60 I I experimentally and theoretically. A coupled signal trace with a cable section causes drastic increase of EM1 from a digital PCB. Maximum level of the PCB with the coupling and the cable sections is 20dB higher than the PCB without these sections. In case that a victim trace is routed just above a source trace, maximum level of EMI is generated. This structure must to be avoided on the PCB. From the viewpoint of practical application, it is effective for EM1 reduction to separate a low &equency signal trace from a high-speed digital signal by ground plane. Radiation from this type of the PCB with the cable section is almost equal that from the PCB without the coupling and the cable sections. EM1 increasing by coupling between signal traces is evaluated by calculation of crosstalk. Moreover, it is shown that evaluation of EM1 level by crosstalk is useful to decide PCB s structure of a high-density assembled PCB. for EM1 reduction Future issue: Influence of various sizes of a ground plane and a thinner dielectric substrate of a PCB on EMI. REFERENCES Frequency [MHii Figure 12. Measurement results of radiation from the PCBs. CONCLUSION The conclusion is as follows: EM1 increasing by coupling between a low hequency and a digital signal traces was shown [l] M.I.Montrose, Printed Circuit Board Design Techniques for EMC Compliance, IEEE PRESS, 1996 [2] W.Cui,H.Shi,X.Luo,F.Sha,J.L.Drewniak,T.P.Van Doren and TAnderson Lumped-element Sections for Modeling Coupling Between High-Speed Digital and I/O Lines IEEE 1997 Int. Symp. on EMC, 1997, ~~260-265 [3] PCB Greenfield User s manual, Quantic Laboratory Inc., 1991!