The Time-Domain Electromagnetic Interference Measurement System

Transcription

1 330 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 45, NO. 2, MAY 2003 The Time-Domain Electromagnetic Interference Measurement System Florian Krug, Student Member, IEEE, and Peter Russer, Fellow, IEEE Abstract A novel real-time broad-band time-domain electromagnetic interference (TDEMI) measurement system for the MHz frequency range is described. The signals from the antenna are sampled, analog-to-digital (A/D) converted and digitally processed. The fast-fourier transform (FFT), the Welchand Bartlett periodograms are digitally computed. Compared with state-of-the-art EMI measurement systems, the novel described TDEMI system samples the complete phase and amplitude information of the EMI simultaneously over the whole frequency band under consideration. With the presented time domain measurement system the measurement time can be reduced by a factor of 10. The digital processing of EMI measurements allows emulation in real-time of the various modes of conventional analogous equipment, e.g., peak, average, rms and quasi-peak detector and also introduces new concepts of analysis, e.g., phase spectra, short-time spectra, statistical evaluation and FFT-based time-frequency analysis methods. Index Terms Digital signal processing, discrete Fourier transforms (DFTs), electromagnetic interference (EMI), fast Fourier transform (FFT), measurement, spectral analysis, time-domain measurements. I. INTRODUCTION DUE to the rapid development of new electronic products and due to emerging new technologies the ability to achieve and to improve electromagnetic compatibility is a major challenge in development of electronic products. EMC and EMI measurement equipment which allows extraction of extensive and accurate information within short measurement times will allow the reduction of the costs and to improve the quality in circuit and system development. In the past and currently, radio noise and electromagnetic interference (EMI) are measured and characterized using superheterodyne radio receivers. The disadvantage of this method is the rleatively long measurement time of typically 30 min for a frequency band from 30 MHz to 1 GHz [1]. Since such a long measurement time results in high test costs, it is important to look up for possibilities to reduce the measurement time without loss of quality. Since conventional measurement systems are not evaluating the phase information of the measured EMI signal, important information is lost. The digital processing of time-domain EMI (TDEMI) measurements using Fourier transform allows the decomposition Manuscript received August 27, 2002; revised February 17, This work was supported in part by Rohde&Schwarz GmbH and Co. KG and in part by the Albatross Projects GmbH. The authors are with the Institute for High-Frequency Engineering, Technische Universität München, D Munich, Germany ( fkrug@ieee.org). Digital Object Identifier /TEMC TABLE I STATE OF THE ART of the measured signal into its spectral components. The use of Fourier techniques has grown rapidly in recent years because of the economy of programs using the fast Fourier transform (FFT). In general, the digital processing of EMI measurements allows emulation in real-time of the various modes of conventional analogous equipment, e.g., peak, average, rms and quasi-peak detector and also introduces new concepts of analysis, e.g., phase spectra, short-time spectra, statistical evaluation and FFT-based time-frequency analysis methods. Beyond this, time-domain techniques exhibit additional advantages. Since time-domain techniques allow processing all the amplitude and phase information over the whole signal spectrum in parallel, the measurement time may be reduced by at least one order of magnitude and the information obtained goes far beyond the information obtained with conventional analogue measurement systems. In this paper, a novel real-time broad-band TDEMI measurement system for the MHz frequency range is described, and new techniques and algorithms for TDEMI measurements are presented. Because the TDEMI system is based on digital processing of sampled EMI signals, one of the advantages of this system is that the performance of the system may be improved via software. The TDEMI measurement system allows emulation of the modes of operation of conventional analogous EMI measurement systems. This is required in order to characterize systems under test with respect to the International CISPR 16 standard [2], [3]. Furthermore, the additional signal characterizations mentioned above also may be performed. A. State of the Art Many authors have investigated TDEMI measurement techniques in the lower frequency range. Table I summarizes the results of several recent studies dating from 1989 to All recent published TDEMI measurement techniques are based on the FFT. Some of the described time-domain measurement systems need an additional EMI receiver to get a correction curve /03$ IEEE

2 KRUG AND RUSSER: TDEMI MEASUREMENT SYSTEM 331 TABLE II CHARACTER OF THE DISTURBANCE TABLE III TDEMI MEASUREMENT SETUP Fig. 1. TDEMI measurement setup. containing all adjustment of the measurement setup, e.g., antenna gain and cable losses. B. Classification of Interferences The EMI originating from the equipment under test (EUT) depends on frequency, time, and geometry of the test setup (position, distance, and direction). The interferences may be classified on the basis of the receiver and interference bandwidth as follows [11] [13] as shown in Table II. Furthermore, EMI signals may be classified on the basis of their statistical behavior as random or deterministic signals. The random signals can be further subdivided in stationary and nonstationary signals [14]. The statistical properties of nonstationary random signals may change considerably over the observation time. The deterministic signals may be either periodic, quasi-periodic, nonperiodic, or a combination of these signal types. Periodic and quasi-periodic signals exhibit line spectra. Transients are nonperiodic signals. Nonperiodic signals exhibit continuous spectra. Finally, signals can be combinations of two or more of the above classes. II. TDEMI MEASUREMENT SYSTEM A. Measurement Equipment The block diagram of the time-domain measurement setup consisting of the TDEMI measurement system and a conventional EMI receiver is depicted in Fig. 1. The EMI receiver is used for validation of the results obtained with the TDEMI measurement system. Table III summarizes the components of the time-domain measurement setup. The TDEMI measurement system consists of a broad-band antenna, a switching unit, an amplifier, a low-pass filter, an analog-to-digital converter (ADC), and a personal computer. The broad-band antenna combines the characteristics of a biconical and a log-periodic antenna to facilitate measurements in the frequency range from 30 to 1000 MHz. The anti-aliasing filter limits the signal bandwidth according to the requirement of the sampling theorem. The ADC has an analog bandwidth of 1.5 GHz. The data are transmitted via a general-purpose interface bus (GPIB) to the personal computer. Fig. 2. Measurement scenario (distance between laptop and antenna is 1 m). B. Measurement Scenario As the EUT a commercial laptop with a 200-MHz clock frequency has been chosen. The measurements on the laptop are performed in its power-on mode, when the laptop is supplied from the internal battery. All measurements were performed in an anechoic chamber with hybrid absorbers. The distance between the EUT and the vertically polarized antenna was 1 m. The measurement scenario is shown in Fig. 2. III. SIGNAL PROCESSING A. Data Acquisition The data acquisition process for the time-domain measurement is shown in Fig. 3. The sampled EMI data are transferred from the main memory of the oscilloscope via the GPIB bus to the personal computer. Then, the amplitude spectra via FFT is digitally computed. The errors due to the frequency characteristics of antenna, transmission line, amplifier, anti-aliasing filter and ADC are corrected by signal processing. Afterwards, a valuation with the peak-, rms- or average-detector mode is made. An additional noise-floor adjustment for comparison with a conventional EMI receiver has to be taken into account. A representative measured TDEMI signal of the EUT is shown in Fig. 4.

3 332 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 45, NO. 2, MAY 2003 Fig. 5. Antenna factor H (f ) in decibels (1/m), amplifier gain H (f ) in decibels. Fig. 3. Process of data acquisition. Fig. 6. Filter frequency response H (f ), Cable losses H (f ), Voltage attenuation ADC H (f ). Fig. 4. Measured TDEMI signal of the EUT. B. Error Correction of the Measurement System In order to compute the spectrum accurately from the time-domain measurements the frequency characteristics of the time-domain measurement system has to be compensated. In Fig. 5 the antenna factor and the amplifier gain are shown. The antenna factor depends on the effective antenna length, the antenna impedance and the input impedance of the amplifier. In Fig. 6, the measured filter frequency response, the cable losses and the voltage attenuation of the ADC are shown. The calculated spectrum from the time-domain data is corrected considering the total transfer function of TDEMI system IV. MODEL OF THE EMI RECEIVER For comparison of the time-domain measurements with the EMI receiver measurements, the IF filter characteristic and the detector type of the EMI receiver have to be taken into account. The measured IF filter characteristics of the EMI receiver is considered in the spectrum calculation via FFT or periodogram. In Fig. 7, the measured IF filter characteristics of the EMI receiver for the bandwidth 200 Hz, 9 khz, 120 khz, and 1 MHz is shown. For using peak, average or rms detectors a time-domain modeling of the detector circuit has to be implemented in the validation of the EMI spectrum. To develop an equivalent system behavior for the TDEMI measurement system, an accurate system model for the EMI receiver is required. The block diagram of a conventional EMI receiver is shown in Fig. 8. Table IV summarizes the components of the EMI receiver. The mixer output signal spectrum in Fig. 8 is given by (2) This yields the output signal spectrum of the IF filter (1) (3)

4 KRUG AND RUSSER: TDEMI MEASUREMENT SYSTEM 333 where is the pulse response of the analog integrator. For a RC integrator the pulse response is given by The RC integrator yields an exponentially weighted mean value. In that case may be considered as an equivalent time of integration. Linear rectification in connection with peak value detection over an interval from to yields (8) (9) Fig. 7. Measured IF filter characteristics. In the case of analog systems the description of the peak detector has to be modified in the following way: (10) Fig. 8. EMI receiver block diagram. TABLE IV EMI RECEIVER V. COMPARISON OF THE TDEMI MEASUREMENT SYSTEM WITH AN EMI RECEIVER To analyze the TDEMI measurement system performance, the measurements performed with the TDEMI measurement system are compared with the results obtained with a conventional EMI receiver in the peak detector mode and the average detector mode. The FFT, Bartlett and Welch periodogram signal processing algorithms have been used for processing of the data measured with the TDEMI system. The results obtained thereby have been compared with the results obtained with a commercial EMI receiver using the peak detector and average detector. The heterodyne receiver may determine either the mean value or the peak value of the rectified signal. The rectifier may be a linear rectifier or a square law rectifier. In the case of a linear rectifier and mean value detection the detector output signal is given by (4) A. The Detector Modes The digital signal processing in the TDEMI measurement system allows emulation in real-time of the various modes of conventional analogs EMI measurement systems. The detector output signal for peak detection is given by The detector output signal by (11) for average detection is given (12) The mean value is formed over the time interval from to. For square law rectification and mean value detection the detector output signal is given by (5) The detector output signal by for rms detection is given (13) With measurement systems based on analog signal processing time averaging usually is performed via RC circuits. In this case (4) and (5) have to be replaced by (6) (7) B. Signal Processing With FFT The fourier Transform for discrete signals is carried out via discrete Fourier transform (DFT) [15] (14)

5 334 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 45, NO. 2, MAY 2003 tional EMI receiver in average-detection mode for amplitudes db V. Fig. 9. Comparison: FFT (black line) and EMI receiver (gray line). C. Signal Processing With Periodogram The EMI spectrum is computed from the time-domain signal using the Bartlett- and Welch- periodograms. Both methods are based on the averaging of the spectra obtained by FFT from time segments of the measured signal. The methods are called nonparametric since no assumptions about the data structure are made. Let be the time-discrete signal to be processed. In the Bartlett method [16] the time domain sequence is subdivided into nonoverlapping segments, where each segment has length. For each segment the periodogram is computed and the Bartlett power spectral estimate is obtained by averaging the periodograms for the segments. The frequency spectrum calculated by the Bartlett periodogram [15] is given by (15) By this averaging of the spectrum the variance of the spectrum estimation is reduced by a factor, however at the expense of a reduction of the frequency resolution by the same factor [17]. Welch [18] has modified Bartletts method by using windowed data segments overlapping in time. The windowing is applied to reduce the spectral leakage associated with finite observation intervals. The overlapping time windows yield a further reduction of the periodogram variance. The frequency spectrum calculated by the Welch periodogram [15] is given by Fig. 10. Difference between the amplitude spectrum measured with the TDEMI system (signal processing with FFT) and the EMI receiver for js[k]j > 40 dbv. where is the input data vector, is the input signal, is the sampling interval, is an integer, is the length of the input data vector, and gives the complex frequency-domain output data vector. In the following, we denominate a continuous signal with and the corresponding sampled signal with where the integer argument denominates samples of the continuous signal at times. The discrete spectrum corresponds to the sampled continuous spectrum with. The DFT can be evaluated with high efficiency using FFT. In Fig. 9 the comparison between the spectrum based on the FFT and the EMI receiver in peak-detector mode is shown. Fig. 9 show the match of the amplitude spectrum measured with the TDEMI system and the conventional EMI receiver for narrow-band harmonic signals, the mean deviation for stationary signal is typically less than 0.5 db. Slight differences appear in the noise floor. These are caused by the differing noise behavior of the TDEMI system and the conventional EMI receiver. Fig. 10 show the difference of the amplitude spectrum measured with the TDEMI system and the conven- (16) is the discrete-time window energy of the used window function as defined as follows: (17) Figs. 11 and 12 show the comparison between the Bartlett periodogram and the EMI receiver in average-detector mode, and the Welch periodogram and the EMI receiver in average-detector mode, respectively. In this case the Welch periodogram yielded a reduction of the variance of the noise floor by a factor 8. The amplitude spectra measured with the TDEMI system and the conventional EMI receiver in Figs. 11 and 12 show a match and the effect of noise variance reduction with the periodogram is shown. The average deviation between the spectrum calculated from the measured time domain data and spectrum measured with the EMI receiver is less than 3 db within a frequency range of MHz. The measurement error depends on the statistical properties of the interferences. One reason for this deviation is the nonstationarity of the measured EMI spectrum. D. Measurement Time The main advantage of an TDEMI measurement method is the reduced measurement time. In Table V, the contributions to

6 KRUG AND RUSSER: TDEMI MEASUREMENT SYSTEM 335 TABLE V MEASUREMENT AND PROCESSING TIME This SNR increase is caused by a spectral spreading of the quantization noise power as the sampling frequency increases. For the 8-bit ADC used in this work (18) yields a typical SNR value of db at for sampling at the Nyquist rate. Fig. 11. Comparison: Bartlett periodogram (black line) and EMI receiver (gray line). Fig. 12. line). Comparison: Welch periodogram (black line) and EMI receiver (gray the measurement time are listed for the TDEMI system and for the analog EMI receiver. According to the CISPR 16 standards [2], [3] the EMI receiver operates in the peak-detector mode and has a dwell time of 100 ms/step. The measured frequency range is MHz. E. Dynamic Range The dynamic range of the TDEMI measurement system depends on the dynamic range of the used ADC. For an ADC driven by a harmonic input signal with an amplitude equal to the ADC full-scale input, the maximum signal-to-noise ratio (SNR) [19] in decibels is where is the maximum input signal frequency, is the sampling frequency of the ADC, and is the number of bits of the ADC. From (18), it follows that for the SNR increases with increasing sampling frequency. VI. NEW CONCEPTS OF ANALYSIS The time-domain signal processing techniques include further advantages in addition to the above mentioned ones. Since time-domain techniques allow processing the amplitude and phase information of the whole signal spectrum in parallel, the measurement time may be reduced by at least one order of magnitude and the information obtained goes far beyond the information obtained with conventional analogue measurement systems. A. Phase Spectra Computation for Broad-Band EMI Signals via FFT For analyzing the pulse-response amplitude and phase spectra of the EMI signal are required. With a time-domain measurement the complete signal information (amplitude and phase of the EMI signal) is measurable. Based upon the complete spectral information the propagation of the EMI pulses and the influence of propagation and scattering on the pulse shape may be analyzed. This gives valuable additional information for the optimization of the systems under test. FFT-based signal processing allows rapid analysis of the scattering characteristics of RF absorbers [20] [22]. Also, novel correlation techniques for reduction of ambient noise in EMI measurement [23] are possible. In Fig. 13, the phase spectrum calculated via FFT for ten independent time-domain measurements of the EUT is shown. The TDEMI system gives the possibility for new broad-band correlation methods in time-domain. B. Broad-Band FFT-Based Time-Frequency Analysis of EMI Signals The EMI originating from the EUT depends on frequency, time and the geometry of the test setup (position, distance and direction). In general measured EMI signals contain transient parts. The time window of length considered in one FFT usually will generally not cover the length of a transient under consideration. To analyze the characteristic of a transient signal the short-time Fourier transform (STFT) [24] is used. The STFT represents a compromise between the time- and frequency-based view of the EMI signal. The method provides information about both when and at what frequencies a signal event occurs.

7 336 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 45, NO. 2, MAY 2003 Fig. 15. Spectrum of a EMI signal of class C. Fig. 13. Phase spectra: 10 time-domain measurements. Fig. 16. Comparison between 30 FFT measurements. Fig. 14. Spectrogram of a EMI signal. The STFT uses a sliding window. It is defined as (19) where the DFT is of size, is the signal to be analyzed and is a time-domain window with zero value outside the interval [0, ]. The spectrogram of a sequence is the magnitude of the time-dependent Fourier transform versus the time. The spectrogram of a measured EMI signal is shown in Fig. 14. In Fig. 14, stationary signal parts are shown as horizontal lines (e.g., CPU clock MHz) and transient signal parts are shown as vertical lines (e.g., memory access ms) in the spectrogram. The spectrogram give the possibility to analyze the character of the EMI signal. The TDEMI system gives the possibility to analyze all classes of EMI signals. With the STFT the behavior of class C signals can be analyzed in detail. In Fig. 15, the spectrum of a measured EMI signal from a drill machine is shown. C. Statistical Analysis and Minimization of Uncertainty in Broad-Band EMI Measurement To evaluate the EMI of the performance of communication systems the information of the statistical properties of the ambient noise, e.g., amplitude probability distribution, crossing rate distribution, pulse spacing distribution and pulse duration distribution [25] are necessary. In such nonstationary interferences the time domain measurement system is capable to give information about the nonstationary behavior of the EUT, within a short measurement time. In Fig. 16, the FFT calculated spectrum for 30 independent measurements is shown. At 200 MHz (= clock frequency of the EUT) a steady field amplitude over 30 measurements is detectable. In the frequency range from 400 to 500 MHz a strong variation of the short-time spectrum with time can be observed. In Fig. 17, the deviation of the field spectrum a 200 MHz and 521 MHz for 30 independent measurements is shown. The problem of measurement uncertainty meets with growing interest and may be considered in future standards for EMC/EMI measurements [26], [27]. Probability distribution and standard deviation of the measured EMC/EMI signal have to be considered. Because of the long measurement time of the conventional EMI receiver, statistical analysis is not possible. With the TDEMI system,

8 KRUG AND RUSSER: TDEMI MEASUREMENT SYSTEM 337 Fig. 17. Deviation of the E field at 200 and 521 MHz. Fig. 19. Standard deviation: 400 time-domain measurements of a nonharmonic oscillator. frequency of the oscillator (stationary signal, e.g., 40, 80, 120 MHz, etc.) the standard deviation of the time-domain measurement is below 0.1 db. Fig. 18. Standard deviation: 400 time-domain measurements. the possibility to analyze the standard deviation is given, so the character of the measurement uncertainty can be researched. The standard deviation of a data vector is defined as where (20) (21) and is the number of elements in the sample. In Fig. 18, the standard deviation (calculated via FFT) for 400 independent measurements of a laptop under repeatability conditions is shown. A standard deviation of approximately 6 db of the radiated emission is shown. The statistical analysis gives information about the emission uncertainty of the EUT under repeatability conditions. Fig. 19 shows the calculated standard deviation of 400 measured EMI amplitude spectra of a nonharmonic 40-MHz oscillator with a stationary emission behavior. The statistical analysis gives an answer about the measurement uncertainty caused by the TDEMI system. At the resonance VII. CONCLUSION A novel TDEMI measurement system has been presented. The TDEMI measurement system allows emulation of the modes of operation of conventional analog EMI measurement systems, e.g., peak, average, rms, and quasi-peak detection. In addition to these capabilities, the TDEMI measurement system facilitates the extraction of phase spectra and short-time spectra from the measurements. Furthermore, statistical signal valuation and FFT-based time-frequency analysis of the EMC/EMI signals is possible. The system performance has been evaluated experimentally. Compared with conventional analog EMI measurement equipment the measurement time is reduced by one order of magnitude. ACKNOWLEDGMENT The authors would like to thank T. Hermann and D. Mueller for carrying out measurements and Matlab simulations; and the Deutsche Forschungsgemeinschaft for the support for acquisition of the EMC measurement chamber and measurement systems. REFERENCES [1] C. Christopoulos, Principles and Techniques of Elektromagnetic Compatibility. Boca Raton, FL: CRC, [2] Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods Part 1: Radio Disturbance and Immunity Measuring Apparatus, International Electrotechnical Commission, Geneva, Switzerland, CISPR16 1, [3] Specification for Radio Disturbance and Immunity Measuring Apparatus and Methods Part 2: Methods of Measurement of Disturbances and Immunity, International Electrotechnical Commission, Geneva, Switzerland, CISPR16 2, [4] C. Keller and K. Feser, Fast emission measurement in time domain, in Proc. 14th Int. Symp. Electromagnetic Compatibility, vol. 2, Zurich, Switzerland, 22 22, 2001, Paper 70K7. [5] M. Stecher, Timing analysis A necessary improvment of EMI emission tests, Dig. Int. Symp. Electromagnetic Compatibility Digest, Sept

9 338 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 45, NO. 2, MAY 2003 [6] U. Reinhardt, K. Feser, and K. Feurer, Vergleich von EMV-messungen im frequenz- und zeitbereich anhand praktischer beispiele aus der fahrzeugtechnik, in Proc. Int. Fachmesse und Kongress für Elektromagnetische Verträglichkeit, vol. 2, Düsseldorf, Germany, 20 22, 1996, pp [7] P. A. Sikora, An EMI receiver design using modern digital techniques, in Proc. 11th Int. Symp. Electromagnetic Compatibility, vol. 3, Zurich, Switzerland, 7 9, 1995, 86N2, pp [8] A. Schütte and H. C. Kärner, Comparison of time domain and frequency domain electromagnetic susceptibility testing, Dig. IEEE Int. Symp. Electromagnetic Compatibility, pp , Aug [9] E. L. Bronaugh and J. D. M. Osburn, New ideas in EMC instrumentation and measurement, in Proc. 10th Int. Symp. Electromagnetic Compatibility, Zurich, Switzerland, 1993, Paper 58J1, pp [10] E. L. Bronaugh, An advanced electromagnetic interference meter for the twenty-first century, in Proc. 8th Int. Symp. Electromagnetic Compatibility, Zurich, Switzerland, 1989, Paper 42H5, pp [11] W. P. Kodali, Engineering Electromagnetic Compatability: Principles, Measurements, Technologies, and Computer Models. New York: Wiley, [12] D. Middleton, Statistical-physical models of electromagnetic interference, IEEE Trans. Electromagn. Compat., vol. 19, pp , Aug [13] M. Uchino, O. Tagiri, and T. Shinozuka, Real-time measurment of noise statistics, IEEE Trans. Electromagn. Compat., vol. 43, pp , Nov [14] W. B. Davenport and W. L. Root, An Introduction to the Theory of Random Signals and Noise. New York: Wiley, [15] S. L. Marple, Digital Spectral Analysis with Applications. Englewood Cliffs, NJ: Prentice-Hall, [16] M. S. Bartlett, Smoothing periodograms from time series with continous spectra, Nature, vol. 161, pp , [17] H. C. So, Y. T. Chan, Q. Ma, and P. C. Ching, Comparison of various periodograms for sinusoid detection and frequency estimation, IEEE Trans. Aerospace Elektron. Syst., vol. 35, pp , July [18] P. D. Welch, The use of fast fourier transform for the estimation of power spectra: A method based on time averaging over short, modified periodograms, IEEE Trans. Electromagn. Compat., vol. 19, pp , Aug [19] R. M. Gray, Quantization noise spectra, IEEE Trans. Inform. Theory, vol. 36, pp , Nov [20] Corona and G. Ferrara, A spectral approach for the determination of the reverberating chamber quality factor, IEEE Trans. Electromagn. Compat., vol. 40, pp , May [21] R. T. Johnk, D. R. Novotny, C. M. Weil, M. Taylor, and T. J. O Hara, Efficient and accurate testing of an EMC compliance chamber using an ultrawideband measurement system, Dig. IEEE Int. Symp. Electromagnetic Compatibility, pp , Aug [22] D. Pommerenke, Directional coupling antennas applied to absorber lined chambers and OATS problems, Dig. IEEE Int. Symp. Electromagnetic Compatibility, pp , Aug [23] T. Shinozuka and A. Sugiura, Reduction of ambient noise in EMI measurement, Dig. IEEE Int. Symp. Electromagnetic Compatibility, pp , Sept [24] J. B. Allen, Short term spectral analysis, synthesis, and modification by discrete fourier transform, IEEE Trans. Acoustics, Speech, Signal Processing, vol. 25, no. 3, pp , June [25] Y. Yamanaka and T. Shinozuka, Statisical parameter measurement of unwanted emission from microwave ovens, Dig. Int. Symp. Electromagnetic Compatibility, pp , Aug [26] J. Perini, Measurement uncertainties in EMC instrumation, in Proc. 13th Int. Zurich Symp. Electromagnetic Compatibility, vol. 2, Zurich, Switzerland, 16 18, 1999, Paper W1, pp [27] J. J. Goedbloed, Uncertainties in standardized EMC compliance testing, in Proc. 13th Int. Zurich Symp. Electromagnetic Compatibility, vol. 2, Zurich, Switzerland, 16 18, 1999, Paper W3, pp Florian Krug (S 01) received the Dipl.-Ing. degree in electrical engineering and information technology from the Technische Universitaet München, Munich, Germany, in Since 2001, he has been working toward the Dr.-Ing. degree at tthe same university. His current research is focused on the analysis of electromagnetic compatibility problems using modern spectral estimation methods and classical frequency domain measurements. Mr. Krug received the Best Student Paper Award from the IEEE Electromagnetic Compatibility Society in He is a member of IEE and VDE. Peter Russer (SM 81 F 94) received the Dipl.-Ing. degree in 1967 and the Dr. Techn. degree in 1971, both in electrical engineering, from the Technische Universität Wien, Vienna, Austria. He was Assistant Professor at the same university from 1968 to In 1971, he joined the Research Institute of AEG-Telefunken, Ulm, Germany, where he worked on fiber-optic communication, broad-band solid-state electronic circuits, statistical noise analysis of microwave circuits, laser modulation, and fiber optic gyroscopes. Since 1981, he has been a Professor and Head of the Institute of High Frequency Engineering at the Technische Universität München, Munich, Germany. In 1990 he was Visiting Professor at the University of Ottawa, Ottawa, ON, Canada, and in 1993 he has been Visiting Professor at the University of Victoria, Victoria, BC, Canada. From October 1992 through to March 1995, he was Director of the Ferdinand-Braun-Institut für Höchstfrequenztechnik, Berlin, Germany. His current research interests are electromagnetic fields, integrated microwave and millimeter-wave circuits, statistical noise analysis of microwave circuits, and methods for computer-aided design of microwave circuits. In 1979, Dr. Russer was corecipient of the NTG Award for the publication Electronic Circuits for High Bit Rate Digital Fiber Optic Communication Systems. In 1994, he was elected to the grade of Fellow of the IEEE for fundamental contributions to noise analysis and low-noise optimization of linear electronic circuits with general topology. He has served as a Member of the Technical Program Committees and Steering Committees of various international conferences (IEEE MTT-S, European Microwave Conference) and as the member of the Editorial Board of several international journals inlcuding Electromagnetics and International Journal of Numerical Modeling. He is the Co-Chairman of U.R.S.I. Commission D. He is author of more than three hundred scientific papers in these areas. He is a Member of the German Informationstechnische Gesellschaft (ITG) and the German as well as the Austrian Physical Societies.