Measurement of Integrated Circuit Conducted Emissions by Using a

622

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 43, NO. 4, NOVEMBER 2001

Measurement of Integrated Circuit Conducted Emissions by Using a Transverse Electromagnetic Mode (TEM) Cell Franco Fiori, Member, IEEE, and Francesco Musolino, Student Member, IEEE

Abstract—This paper presents a new technique for the measurement of integrated circuit (IC) conducted emissions. In particular, the spectrum of interfering current flowing through an IC port is detected by using a transverse electromagnetic mode (TEM) cell. A structure composed of a matched TEM cell with inside a transmission line is considered. The structure is excited by an interfering source connected to one end of the transmission line. The relationship between the current spectrum of the source and the spectrum of the RF power delivered to the TEM mode of the cell is derived. This relationship is evaluated for one specific structure and the experimental validation is shown. Results of conducted emission measurement performed by using such a technique are shown as well and compared with those derived by using the magnetic probe method. Index Terms—Conducted emission measurement, electromagnetic compatibility (EMC), electromagnetic emission (EME), integrated circuit (IC), TEM cell, transmission line.

I. INTRODUCTION

O

PERATIONS of digital integrated circuits (ICs) drive the electromagnetic emissions of electronic systems, hence ICs like microprocessors and microcontrollers can be considered as primary sources of electromagnetic emissions (EMEs). Steep currents and voltage glitches at the power supply and ground pins of an IC and proper signals at its input/output (I/O) pins are considered IC conducted emissions since they drive EMEs of antennas composed of printed circuit board (PCB) traces and/or interconnecting cables of the electronic system, which the IC is part of. Furthermore, pulsed currents flowing through IC package leads and circuits routed at silicon level excite IC direct radiations of interfering electromagnetic fields. IC radiated and conducted emissions heavily influence the electromagnetic compatibility (EMC) features of electronic apparatuses, hence the right selection of ICs, from the EME point of view, implies a reduced number of EMI filters at PCB and system level, necessary to be compliant with system level EMC requirements [1]. For this reason, in past years the interest on IC EME measurement methods has grown: several studies and measurement techniques have been presented in the literature [2]–[7] and some of Manuscript received April 26, 2001; revised June 13, 2001. This work was supported in part by the CNR-progetto Finalizzato MADESS II and the MEDEA A509 (MESDIE) European Project. The authors are with the Dipartimento di Elettronica, Politecnico di Torino, 24-I-10129, Torino, Italy (e-mail: [email protected]). Publisher Item Identifier S 0018-9375(01)10017-7.

them are standardized [8] or are up on the way to become international standards [9]. Concerning conducted emissions, the chapters 4 and 6 of the IEC document [9] describe, respectively, Direct Coupling Method and the Magnetic Probe the Method. In particular, the 1 Direct Coupling Method makes possible the measurement of conducted emissions related to IC power supply pulsed-currents. A 1 resistor is inserted into the IC power supply net (over IC bypass capacitors) and the IC power supply current spectrum is obtained by measuring the voltage spectrum across the 1 resistor. Such a technique is useful also in the case of ICs with multiple ground pins. The 150 Direct Coupling Method is proposed for the evaluation of conducted emissions of long external wiring and long PCB traces driven by I/O pins. In particular, this method is aimed to evaluate the contribution of a single pin to the EME of an IC and consists in the measurement of the voltage spectrum at an I/O pin connected to a 150 load. Alternatively, IC conducted emission measurements are performed by the Magnetic Probe Method. The measurement of pulsed current spectrum flowing through an IC pin is obtained by means of a contact-less miniature magnetic probe. In practice, the magnetic field strength over a microstrip line connected to a power supply pin (or an output driver pin) of the IC under test is measured and then converted by calculation into the RF current [4], [9]. In this paper, the spectrum of pulsed current flowing through an IC pin is obtained by using a transverse electromagnetic mode (TEM) cell. A structure composed of a microstrip line inserted in a matched TEM cell is considered. The microstrip line is driven at one end by the interfering source (i.e., an IC pin) and current flowing through the microstrip line excites RF energy propagating into the TEM cell toward its terminations. Elaborating the spectra of RF power delivered to TEM cell terminations the current spectrum of the interfering source is derived. The paper has the following structure. In Section II, the structure composed of the TEM cell and a microstrip line is considered. The relationship between the microstrip line feeding current and the power delivered to the TEM mode of the cell is derived. In Section III, the innovative measurement method is presented in detail and it is employed in two different cases, measuring conducted emissions excited by an output driver operation and those excited by an IC power supply disturbances. Results of such a measuring techniques are compared with those obtained by the magnetic probe method. Section IV draws concluding remarks.

0018–9375/01$10.00 © 2001 IEEE

FIORI AND MUSOLINO: MEASUREMENT OF IC CONDUCTED EMISSIONS

623

Fig. 2. TEM cell x-y cross section.

Fig. 1. (a) y -z cross section of a TEM cell with inside a microstrip line of length L. The current source I feeds the structure at port 1. Port 2 is loaded by Z impedance and ports 3, 4 are matched and (b) top view of the same structure.

II. EMISSION OF A TRANSMISSION LINE IN A TEM CELL The structure under analysis is reported in Fig. 1; it shows a and a transmission TEM cell with characteristic impedance line of length and characteristic impedance . The conductor A of the transmission line is a metallic trace over an isolating layer (dielectric constant ). The trace is positioned along the -axis just in the middle of the cell. The conductor B of the transmission line is the upper wall of the TEM cell just over the isolating layer. The two terminals of the transmission line are labeled as ports 1 and 2, while the TEM cell terminations are labeled as port 3 and port 4. The structure is driven at the port 1 by a current source and it , while port is loaded at the port 2 by the lumped impedance 3 and 4 are connected to matched loads. In this section the relationship between the power delivered to the TEM mode of the cell and the feeding current ( ) at the port 1 is derived. Such a relationship is obtained in three steps: the electric and magnetic field distribution related to the TEM is computed by operating the method of moment, the current distribution along the transmission line is computed by using the transmission line theory. Finally, forward and backward power traveling waves, excited by the microstrip current distribution, are determined by an application of the Lorenz reciprocity formula [10], hence powers delivered to the TEM cell terminations (ports 3 and 4) are computed as a function of the current at the port 1. A. Electric and Magnetic Field Distribution in a TEM Cell The TEM cell or Crawford cell is a rectangular-strip transmission line composed of a central conductor (septum) surrounded by an outer conductor with rectangular-shape.

The useful frequency range of a TEM cell is limited by the cutoff frequency of the higher order TE and TM modes and the appearance of resonances due to reflections at the tapered regions. Usually, TEM cells employed in EMC tests are designed to have spurious resonances at frequencies higher than the first higher mode cutoff frequency. If a TEM cell with these features is operated below its cutoff frequency (i.e., the frequency at which the first higher order mode begins to propagate [11], [12]) and it is terminated on its characteristic impedance, a propagating TEM field is present into the cell, which approximates a plane wave and, consequently, uniform field exists over the central portion of the cell. In what follows to this paper, a symmetric TEM cell with the features described as above is considered. Furthermore, it is assumed that the interference source ( in Fig. 1) excites only the TEM mode of cell and its spectral components become negligible at frequencies higher than the cell cutoff frequency. With these assumptions the and field distribution within the TEM cell is computed by using the numerical technique described in [13]. In particular, the integral equation of the potential is formulated in terms of the line-charge density over the transverse conand of the cell (see Fig. 2). This integral equation is tour solved by the method of moment for the line-charge density and the -field distribution is subsequently calculated. The -field distribution is derived from that of -field by using the intrinsic impedance of free space, i.e., 377 [13]. This technique is employed for the computation of the TEM mode and field distributions in a symmetric cell of dimencm, cm, cm and cm sions (see Fig. 2). A voltage (peak value of 1 V) applied between TEM cell septum and the outer conductor excites the electromagnetic field into the cell. The first higher order mode of such GHz hence a cell is the TE 01 with cutoff frequency the field distributions reported in Fig. 3 are valid up to this frequency. Furthermore, the -field distribution, derived from the -field distribution by the free space impedance, is valid for frequencies higher than 2 kHz [13], hence computations derived in the following of this paper are valid in the frequency range 2 kHz–1 GHz. -field components Curves in the Fig. 3 represent the and into the cell. In particular, continuous lines show the distribu-

624


magnetic vector transversal to the direction of propagation along the positive axis; axial electric vector; axial magnetic vector. In (1)–(4), is a general summation index and implies a summation over all possible modes. Since the source current spectrum is considered to be below the TEM cell cutoff frequency only the TEM mode is excited and (1)–(4) become

(5) where

is the TEM mode propagation constant

Fig. 3. Continuous lines represent the distribution along the y -axis of the electric field vertical component (E ) for three different y values. Dashed lines represent the distribution along the x-axis of the electric field horizontal component (E ) for the same y values. In particular, curves A, B, C refer to y 5 cm, y = 6:5 cm and y = 8 cm respectively.

=

tion of the -component electric field magnitude ( ) along the -axis, while dashed lines show the distribution of the electric field -component magnitude ( ) along the -axis. Such field distributions are obtained for three different distances from the central septum. Considering these field distributions it can be (plane , ) is null observed that close to the plane hence electromagnetic field approximate plane wave propagation in free space.

and the coefficients are computed by the following expressions: (6) (7) where [10]:

B. Evaluation of the Overall Radiated Power The expression of the overall power radiated by a transmission line in a TEM cell (see the structure in Fig. 1) is derived with the port 1 driven by an ideal current source and port 2 . loaded by an impedance In this paper, we will assume that the source current spectrum is below the TEM cell cutoff frequency hence, only the TEM mode is excited. In addition, we consider the TEM cell ). perfectly matched at both ends (ports 3 and 4 loaded by In this specific condition, the cell behaves like a waveguide of infinite length and the current source is located in finite volume and (see Fig. 1). In general, the field radiated by between this source may be expressed as an infinite sum of guided modes (1) (2) (3) (4) where unknown coefficients; field propagation constant; electric vector transversal to the direction of propagation along the positive axis;

is a normalization constant defined as follows

(8) in (8) is the surface bounded by the transverse contour and of the cell (see the cross section in Fig. 2). The current density distribution, i.e., the current density along the transmission line, can be written with Dirac functions:

(9) where ,

coordinates of the microstrip line in the - plane; coordinate of the microstrip ends corresponding to port 1; coordinate of the microstrip ends corresponding to port 2. Referring to the structure reported in Fig. 1(a) and (b), and , and . In the previous

expressions , , are unit vectors. Substituting (9) in (6) and (7), it appears that the current along the microstrip line in the direction is orthogonal to the trans, hence it does not excite any electromagnetic verse mode field into the cell. In practice, the field into the TEM cell is excited by the vertical ( ) current sources since they are coupled with the TEM mode.


625

Assuming that the surrounding TEM cell does not influence the current distribution along the transmission line and as, the transmission line theory can be employed suming as a function of (i.e., the current at to derive the port 1). In fact (10) is the reflection coefficient related to the impedance where (connected to the port 2), referred to the transmission line is the transmission line propcharacteristic impedance . agation constant. In case of a perfect matching at the port 2 ), the two currents , at the microstrip ( ports are equal in amplitude but differ in phase. A signal applied at the port 1 propagates on the microstrip line and reaches the port 2 after a time delay hence a phase shift between the two currents exists. , of By substituting (10) in (9) the coefficients (6) and (7) can be expressed as follows:

where power becomes

is the free space impedance. Hence, the total

(12) By substituting (11) in (12) the relation between the overall radiated power and the square of the current magnitude is derived

Finally, we define the radiation resistance ( crostrip line inserted into the TEM cell as

) of the mi-

(13) hence

(14) (11) Finally, the overall power radiated by a transmission line into a TEM cell (see Fig. 2) is obtained by computing the real part of the Poynting vector related with the TEM mode evaluated on surface in Fig. 2). the TEM cell cross section ( In particular

Electric and magnetic field of the TEM mode are related by

Substituting (8) in (14), (15) (shown at the bottom of the page) holds. The radiation resistance depends on the TEM cell dimensions, the microstrip line geometry and the relative position of the microstrip line respect to the cell. Furthermore, the knowledge of the microstrip loading condition is required ( at port 2). ) versus freFig. 4 shows the radiation resistance ( cm, characteristic quency of a microstrip line of length , matched at the port 2 ( ) impedance of and inserted in a matched TEM cell. Continuous line is obtained by using (15) while dotted line is derived from measurements. In particular, the structure composed of the microstrip line and the TEM cell has been characterized in terms of scattering parameters. In particular, the 3 3 -parameter matrix at the ports 1, 3, 4 has been obtained by using the network analyzer HP 8753ES [14] and executing two sets of measurements: -parameter matrix at the ports 1, 3 has been measured with ports 2 and 4 matched and similarly the -parameter matrix at the port

(15)

626


mismatches at the port 2 due to the electrical interconnection of the microstrip line to the load (see Fig. 1). Finally, the square magnitude current spectrum of an interfering source connected to the port 1 (see Fig. 1) can be computed by the following expression (19) In the next section an IC is characterized in terms of conduced emissions by using both the new measurement technique and the magnetic probe method. Experimental results are compared. III. THE MEASUREMENT METHOD AND ITS APPLICATIONS

=

Fig. 4. Radiation resistance (R ) of a microstrip line of length L 6 cm, h = 1:5 mm, w = 3:1 mm, " = 3:97 inserted in a 1 GHz TEM cell along the z -axis, as shown is the Fig. 1.

1 and 4 has been measured with port 2 and 3 matched. Hence, ) is derived from such a measurethe radiation resistance ( ment results as follows:

where , are vectors containing respectively forward and backward power waves at the ports and power waves are defined as:

(16) where , are, respectively, the voltage and the current at the is the reference resistance associated to each port of port . the structure. contains scattering parameters of the The 3 3 matrix is defined as 3-ports structure. For instance (17) and it is obtained by the measurement of and with ports 2 and 3 matched on . Hence, the radiation resistance of the microstrip line inserted in the TEM cell can be derived from -parameters by using (16) and (17). In fact (18) Differences between the radiation resistance computed by (15) and that obtained experimentally on an actual structure (see Fig. 4) are due to mismatches in the TEM cell tapered regions, losses in the microstrip line dielectric and misalignment of the microstrip line with the longitudinal ( -axis) direction of the TEM cell. In addition, the radiation resistance derived from -parameters measurement includes the microstrip line

The structure presented in the following is useful for the evaluation of a single pin conducted emissions. An IC pin is connected to the port 1 of the test setup (see Fig. 1) hence pulsed current ( ) flowing in such a port drives the electromagnetic is connected to the emissions of the microstrip line. A load port 2. The square magnitude of the current spectra is computed is computed by (19). In particular, the radiation resistance by using (15) or it is derived by experimental test results (18) is obtained summing and the spectrum of the total power the power spectra measured at the TEM cell terminations (ports 3 and 4). The evaluation of the interfering source current spectrum by this technique is effective since the influence of the TEM cell structure on the current flowing in the microstrip line is negligible. In facts, the microstrip line and the TEM cell are weakly coupled. This fact has been pointed out by the measurement of the microstrip line input impedance (port 1) in two different cases: 1) the microstrip line placed into a 1-GHz TEM cell (as shown in Fig. 1); and 2) placed in a fully anechoic chamber. Measurement results have shown that variations of such an impedance all over the frequency range 150 kHz–1 GHz [9] are negligible hence the interference source loading condition is not influenced by the presence of the TEM cell. In the following of this section two different applications of this method are presented: the first concerns the measurement of an output driver conducted emissions, while the second is on the measurement of an IC power supply conducted emissions. Current spectra obtained by this new measurement method are compared with those obtained by the magnetic probe method [4], [9]. A. Measurement of Output Driver Conducted Emissions An output driver of the device 74HC04 drives the port 1 of loads the port 2. the microstrip line and the impedance is the series of the capacitor and the resistor Impedance . The schematic view of the test board is reported in Fig. 5. cm and characteristic The microstrip line has length . A substrate with permittivity impedance and height mm is employed, hence, the microstrip mm. A TEM cell with cutoff frequency of 1 width is GHz (compliant with the document IEC 61 967-2) is employed and dimensions of the PCB are suitable with those of the up-side aperture of the TEM cell.


627

Fig. 5. Schematic view of the printed circuit board employed for the measurement of a 74HC04 output driver conducted emissions. In this circuit V 5 V, R = 50 , C = 100 pF, C = 100 nF and the microstrip line has length L = 6 cm and characteristic impedance Z = 50 .

=

Fig. 6. Envelopes of 74HC04 output driver current spectra. Continuous line is obtained by operating the new measurement method presented in this paper, while dashed line is obtained by the magnetic probe technique.

The driver supply voltage is V, the input signal V, V ( ) is a 1-MHz square waveform, and 50% duty cycle. The spectra of RF power delivered to the TEM cell terminations (ports 3 and 4) are measured by using the spectrum analyzer HP8591E [15] and the total power de) is computed by summing such livered to the TEM mode ( a power spectra. The current spectrum square magnitude of the interfering source ( ) is computed by (19), where both the total and the radiation resistance are those obtained power from measurements. Fig. 6 shows the current spectrum envelopes obtained by the new measurement technique proposed in this paper (continuous line) and the magnetic probe method (dashed line). B. Measurement of IC Conducted Emissions via Power Supply Pins The new measurement technique is employed in the measurement of IC power supply pin conducted emissions. In particular, pin of a 74HC04 is connected to the port 1 of the mithe crostrip line while port 2 is connected to a 5 V power supply. Filtering capacitors are connected to the ports 1 and 2 as specified in [9]. Fig. 7 shows the schematic view of the PCB employed for these experimental tests.

Fig. 7. Schematic view of the printed circuit board employed for the measurement of 74HC04 power supply conducted emissions, induced by commutations = 5 V, C = 200 nF, of six synchronous output driver. In this circuit V C = 47 pF, C = 100 nF and the microstrip line has length L = 6 cm and characteristic impedance Z = 50 .

Fig. 8. Envelopes of 74HC04 power supply current spectra. Continuous line is obtained by operating the new measurement method presented in this paper, while dashed line is obtained by the magnetic probe technique.

In particular, the six inverter of the 74HC04 are driven by a V, V and a duty 1-MHz square waveform, cycle of 50%. Due to output driver commutations, pulsed current is ab). A part of such a sorbed from the IC power supply pin ( and the rest is current is supplied by the bypass capacitor absorbed from the power supply net ( ). The power supply flow in the microstrip line, hence its spectrum can current be measured by using the method presented in this paper. In current spectrum is computed by (19), where particular the and the radiation resistance are both the total power those obtained from measurements. Fig. 8 shows the spectrum envelopes of the power supply current absorbed by six inverters, which switch synchronously. Continuous line has been obtained by operating the measurement method described in this paper, while dashed line in the same figure has been obtained by using the magnetic probe method. The good agreement of experimental data reported in Figs. 6 and 8 confirms the effectiveness of the measurement method proposed in this paper. However, it can be observed that the selected device under test has shown significant current spectra for frequencies lower than 500 MHz, hence the current spectrum (15) can be computed by using in (19) either the simulated (18). For interfering source with current or the measured

628


spectrum extending up to the TEM cell maximum operating freobquency (in our case, 1 GHz) the employment of the tained from measurements compensates those errors due to mismatches in the TEM cell tapered regions, losses in the dielectric of the microstrip line and misalignment of the microstrip line with the longitudinal ( -axis) direction of the TEM cell. IV. CONCLUSION In this paper a new technique for the measurement of integrated circuit conducted emissions has been shown. A structure composed of a microstrip line inserted into a matched TEM cell has been considered and the relationship between the current exciting one of the microstrip port and the overall power radiated into the cell has been derived (19). Such a relationship makes possible the computation of the interfering source current spectrum on the basis of the power spectra picked up at the TEM cell terminations. It includes the radiation resistance of the mi) inserted into the TEM cell, which has been crostrip line ( derived from the waveguide excitation theory [10]. Such a radiation resistance has been computed for a specific structure and it has been validated experimentally. Differences between the radiation resistance derived from computation and that obtained from measurements can be justified considering that hypothesis assumed to be valid for the computation of the radiation resistance are not longer valid at higher frequencies. In facts, mismatches in the TEM cell tapered regions, losses in the microstrip line dielectric, parasitic related to the actual realization of the test board and misalignment of the microstrip line with the longitudinal ( -axis) direction of the TEM cell are some aspects which are neglected in the computation on the radiation resistance. For these reasons the measurement technique presented in this paper requires a calibration procedure which consists in the identification of the radiation resistance all over the frequency range of interest. The new measurement technique and the magnetic probe method have been employed for the evaluation of the power supply current spectrum of a six integrated output drivers switching simultaneously at 1 MHz and the spectrum of the current flowing in an IC output pin fed by an output driver. The good agreement between data obtained by employing these two different measurement methods confirms the effectiveness of the technique proposed in this paper. ACKNOWLEDGMENT The authors wish to thank the anonymous reviewers for their valuable comments. REFERENCES [1] “Limits and Methods of Measurement of Radio Disturbance Characteristics of Information Technology Equipment,” CENELEC, Bruxelles, Belgium, EN 55022, 1994.

[2] J. P. Muccioli, T. M. North, and K. P. Slattery, “Investigation of the theoretical basis for using a 1GHz TEM cell to evaluate the radiated emissions from integrated circuits,” in Proc. IEEE Int. Symp. Electromagnetic Compatibility, Santa Clara, CA, 1996, pp. 63–67. [3] M. Coenen, “On-chip measures to achieve EMC,” in Proc. EMC Symp., Zurich, Switzerland, 1997, pp. 31–36. [4] N. Masuda, N. Tamaki, H. Wabuka, T. Watanabe, and K. Ishizaka, “RF current evaluation of ICs by MP-10L,” NEC R&D, vol. 40, no. 2, pp. 253–258, 1999. [5] M. J. Coenen, “EMC workbench: Testing methodology, module level testing and standardization,” Philips J. Res., vol. 48, no. 4, pp. 83–116, 1994. [6] W. R. Pfaff, “Application independent evaluation of electromagnetic emissions for integrated circuits by the measurement of conducted signals,” in Proc. IEEE Int. Symp. Electromagnetic Compatibility, Denver, CO, 1998, pp. 219–224. [7] R. De Smedt, S. Criel, F. Bonjean, G. Spildooren, B. Demoulin, and J. Baudet, “Emission of a line in free space and in a TEM cell,” in Proc. CEM COMPO, Toulouse, France, 2000, pp. 20–25. [8] “Electromagnetic Compatibility Measurement Procedures for Integrated Circuits—Integrated Circuit Radiated Emissions Measurement Procedure,” Society of Automotive Engineers (SAE), Warrendale, PA, 150 kHz to 1000 MHz, Surface Vehicle Recommended Practice J1752/3, 1995. [9] “Integrated Circuits, Measurements of Electromagnetic Emission 150 kHz–1GHz,” International Electrotechnical Commission, Geneva, Switzerland, IEC 61967, 2000. [10] R. E. Collin, Foundations for Microwave Engineering, 2nd ed. Singapore: McGraw-Hill, 1992, ch. 4. [11] D. A. Hill, “Bandwidth limitations of TEM cells due to resonances,” J. Microwave Power, vol. 18, no. 2, pp. 182–195, 1983. [12] C. M. Weil, W. T. Joines, and J. B. Kinn, “Frequency range of large-scale TEM mode rectangular strip lines,” Microwave J., vol. 24, no. 4, pp. 93–100, 1981. [13] R. J. Spiegel, W. T. Joines, C. F. Blackman, and A. W. Wood, “A method for calculating electric and magnetic fields in TEM cell at ELF,” IEEE Trans. Electromagn. Compat., vol. EMC-29, pp. 265–272, Nov. 1987. [14] HP8753ES Network Analyzer User’s Guide, Hewlett-Packard, 1999. [15] HP8591E Spectrum Analyzer User’s Guide, Hewlett-Packard, 1996.

Franco Fiori (M’01) was born in Sassari, Italy, in 1968. He received the degree of Laurea and the Ph.D. degree in electronic engineering from the Politecnico di Torino, Turin, Italy, in 1993, and 1997, respectively. From 1997 to 1998, he was with the R&D division of STMicroelectronics as Leader of the EMC group. Since 1999, he has been with the Department of Electronics of the Politecnico di Torino as Assistant Professor. His main research interests concern the design of analog ICs robust to RF interference and the design of digital ICs with reduced emission profile.

Francesco Musolino (S’01) was born in Torino, Italy, in 1972. He received the degree in electronic engineering from the Politecnico di Torino, Turin, Italy, in 1999. Currently, he is working toward the Ph.D degree in electronic engineering with the Department of Electronics of the same university. His research interests include electromagnetic compatibility measurement techniques, especially the measurement methods by which integrated circuit emission or susceptibility are evaluated.