Emi Papaer

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 15, NO.
2, MARCH 2000 399
Analysis and Spectral Characteristics of a

Spread-Spectrum Technique for Conducted EMI
Suppression
K. K. Tse, Member, IEEE,, Henry Shu-Hung Chung, Member, IEEE,, S. Y. (Ron) Hui, Senior Member, IEEE,, and
H. C. So, Member, IEEE
Abstract—Frequency modulation (FM) and random switching switch voltage will appear on and . As the device's voltage
methods have been used for reducing conducted electromagnetic consists of the fast switching edges, harmonics up to several
interference (EMI) in power converters. Limited theoretical megahertz may appear on and , and hence on the supply
studies and comparisons of these schemes, however, are available.
In this paper, a detailed analysis and the spectral characteristics network. The input current is generally pulsating and thus is rich
of a random carrier-frequency (RCF) technique for suppressing in harmonic components. Most of the differential-mode noise
conducted EMI in an offline switched-mode power supply are pre- can in principle be bypassed by the filter capacitors and .
sented. The analysis provides a theoretical platform for studying Their capacitance values are dependent on the maximum peak
the characteristics of this random switching scheme. The level of the harmonic components. However, the practical filtering
of randomness is defined for the RCF scheme and varied in the
converter example so that its effects on the power spectra can be effects of capacitors are not perfect. It is necessary to consider
demonstrated. Theoretical predictions of the spectral characteris- other alternatives so that the conducted EMI can be minimized.
tics of this scheme are confirmed with measurements. The RCF The most effective way of suppressing conducted EMI is to
scheme has been compared with the standard constant-frequency act on the noise source. One obvious approach is to deal with the
pulsewidth modulation (PWM) scheme and the FM scheme. switching methods of the power device. Frequency modulation
Comparisons of their spectral performance show that the RCF
scheme has better conducted EMI suppression than the FM and (FM) has been incorporated into the pulse-width-modulation
standard PWM schemes. (PWM) of the power converter [2] in order to reduce the input
discrete harmonics. This approach is an important step forward
Index Terms—AC-DC power conversion, pulsewidth modula-
tion, random switching technique, switching circuit. in dealing with the conducted EMI issue because it offers a
simple and yet effective method for modifying a standard PWM
scheme to meet EMC standards. Recently, random switching
I. INTRODUCTION technique has been recognized as an emerging technology
for power converters [3]. Various random switching schemes,
N OWADAYS, switched-mode power supplies have to be
designed not only to provide the required electrical func-
tions, but also to meet international electromagnetic compati-
which are originated from statistical communication theory
[11], have been reported for dc/ac and dc/dc power conversion
bility (EMC) standards. Inherently, switched-mode power sup- [3]–[10]. The basic principle of introducing randomness into
plies are electrically noisy. A typical configuration with the con- standard PWM scheme is to spread out the harmonic power so
ducted electromagnetic interference (EMI) test setup is shown that no harmonic of significant magnitude exists. As a result,
in Fig. 1 [1]. Two conflicting filter design constraints have to discrete harmonics are significantly reduced and the harmonic
be considered. Firstly, in order to comply with the international power is spread over the spectrum as ‘noise’ (continuous spec-
EMI regulations, a differential and common-mode EMI filter trum). Although some papers such as [12] have comparisons
is generally put in series with the line input. Secondly, for sat- on PWM inverters, there is no reported comparison on these
isfying the safety requirements, the maximum values of the de- two approaches for offline switched-mode power supplies.
coupling capacitors and (which are usually less than 0.01 It has been pointed out in [10] that the continuous noise spec-
µF and are connected between the supply lines and ground) are trum within the pass band of the converter's output low-pass
limited. These conflicting requirements make proper design of filter could lead to noise-induced low-frequency voltage ripple
the noise filter not straightforward. in the converter's output. This undesirable feature may prohibit
The switching device in a switched-mode power supply is the use of random modulation in dc/dc converters, which re-
the principle source of EMI. Decoupling capacitors and quire tight voltage regulation. In [10], this problem has been
the parasitic capacitors – form a closed loop. Part of the highlighted in the random pulse width modulation (RPWM) and
random pulse position modulation (RPPM) schemes. For the
RPWM method, the duty cycle of the switch is continuously
Manuscript received August 25, 1998; revised July 7, 1999. This work was changing in every switching cycle although the average duty
supported by the Hong Kong Research Council. Recommended by Associate cycle is theoretically equal to the nominal duty cycle. This duty
Editor, J. Van Wyk. cycle variation worsens the voltage ripple problem. In the RPPM
The authors are with the Department of Electronic Engineering, City Univer-
sity of Hong Kong, Hong Kong (e-mail: eeshc@cityu.edu.hk). scheme, there is also some possibility that two adjacent pulses
Publisher Item Identifier S 0885-8993(00)02335-8. of fixed duty cycle may appear together. Rigorous analyses of
0885–8993/00$10.00 © 2000 IEEE
Authorized licensed use limited to: Motilal Nehru National Institute of Technology. Downloaded on June 06,2022 at 06:49:23 UTC from IEEE Xplore. Restrictions apply.
400 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 15, NO. 2, MARCH 2000
Fig. 1. Typical configuration of an offline switched-mode power supply with conducted EMI test setup.
TABLE I
COMPONENT VALUES OF A PRACTICAL
OFFLINE FLYBACK CONVERTER
duty cycle operation in the dc–dc converter. The variation of the

output voltage is not as significant as the RPWM and RPPM
schemes and therefore allows simple feedback control design.
At the same time, the power spectra of the converter input cur-
rent and the switch voltage will spread over a wide frequency
range so that no harmonic of significant magnitude exists. Ef-
fectively, the envelopes of the power spectra of the differen-
tial-mode and common-mode conducted EMI can in principle
be reduced. These EMI aspects have been verified experimen-
tally in [7]–[9]. However, the lack of theory of such scheme
makes it difficult to decide how much randomness should be
introduced. In this paper, a mathematical analysis of the above
phenomena of the RCF method is given. The scope of this paper
is not to answer all questions about the RCF scheme. Instead, it
aims at providing a platform for understanding the spectral per-
Fig. 2. Waveforms of flyback converter operating in discontinuous conduction formance and the effect of the variation of the level of random-
mode with RCF scheme. (a) Sawtooth and reference voltage waveforms. (b) ness. The model of an offline switched-mode power supply and
Gate signal. (c) Input current i . (d) Switch voltage v .
mathematical derivations of the frequency spectra of the input
current and the switch voltage waveform with and without RCF
these two random schemes for dc–dc converters have been pre- scheme are presented in Section II. Practical measurements of
sented in [13], [14]. the conducted EMI of a 58-W, 220-V/24-V, 50-Hz offline fly-
In this paper, the random carrier frequency (RCF) scheme back converter are given in Section III, together with analytical
for dc–dc converter is examined and compared with the stan- predictions. The results are compared with the standard constant
dard PWM scheme and the FM scheme. One attractive fea- switching frequency scheme and the FM [2] scheme. The con-
ture of the RCF scheme is that it inherently ensures constant clusion follows in Section IV.
TSE et al.: ANALYSIS AND SPECTRAL CHARACTERISTICS OF A SPREAD-SPECTRUM TECHNIQUE FOR CONDUCTED EMI SUPPRESSION 401
(a)
(b)
Fig. 3. Synthesis of RCF switching signal. (a) Feedback circuit. (b) Noise generator.
II. MODELS AND MATHEMATICAL DERIVATIONS For the input side, since , and form a potential di-
vider with the simulated 50- supply line resistance in the
A. Source of EMI LISN, part of appears on . Thus, in order to reduce the
As shown in Fig. 1, a standard line-impedance stabilization voltage across , and can be increased. However,
network (LISN) is connected between the power supply and this will substantially decrease the voltage supplying to the con-
the supply lines. As the supply source impedance is high, verter circuit, due to the increase in the voltage drop across
a common-mode harmonic current will flow through the , and .
ground plane, parasitic capacitance – , and . In practice, the value of and hence are slightly difficult
If denotes the effective parasitic capacitance between the to be determined. The above derivations are mainly for illus-
drain voltage of the switch (i.e., ) and the ground plane, trating the major EMI source. In order to simplify the analysis, it
the spectral magnitude of at frequency , can be is assumed that the switch voltage is not affected by the parasitic
obtained by components. The validity of this assumption is ensured by the
close agreement between the experimental measurements and
(1) the theoretical predictions.
In this study, the RCF scheme is applied to the PWM
where is the spectral magnitude of . Thus, the spectral switching of the main switch . Fig. 2(a) shows the randomized
magnitude of the voltage across and , can be sawtooth waveform that is compared with a reference signal
approximated by to generate the gate signal [Fig. 2(b)]. The duty cycle of
is fixed in the respective cycle, although the switching period
(stochastic variable) is varied. Thus, the differential-mode
noise (due to the input current) and the common-mode noise
(2) (due to the switch voltage) can be spread over a certain
frequency range. Spreading the harmonic power of the input
current and the switch voltage signal is discussed separately
if . in the following sub-sections. The mathematical derivations
(a)
(b)
Fig. 4. Experimental waveforms of the converter. (a) < = 0. (b) < = 0:2. [Channel 1: Output voltage (1 V/div, with ac coupling). Channel 2: Gate signal (10
V/div). Channel 3: Input current (1 A/div). Channel 4: Switch voltage (200 V/div). Timebase: 10 µs/div.]
are based on a flyback converter operating in discontinuous where is the input voltage and is the inductance of the fly-
conduction mode (DCM), which is a commonly used config- back transformer. is a randomized switching period resulting
uration in many low-power applications. The attraction of the from the RCF. The general expression of is
flyback converter is its single power conversion step, where
the ‘flyback’ transformer provides the features of electrical (4)
isolation and as an energy storage medium.
The auto-correlation of is defined as
B. Spectral Characteristics of the Input Current Waveform
Fig. 2(c) shows the waveform of the input current of the (5)
converter. For a generic switching cycle equals
can be expressed as where is the expected value of the quantity inside the
bracket. is the observation interval, containing expected
for value of . That is
(3)
elsewhere (6)
(a) (b)
(c) (d)
Fig. 5. Experimental power spectra. (a) Input current with < = 0. (b) Input current with < = 0 2. (c) Switch voltage with < = 0. (d) Switch voltage with
:
< = 0 2. :
By substituting (4) and (6) into (5) Equation (8) can be expressed as
(7)
If denotes the Fourier transform of and de-

notes the time integral in (7) (10)
Hence, the autocorrelation of can be expressed as
(11)
Based on the Wiener-Khinchin theorem [11], the power spec-

trum of a signal is the Fourier transform of its autocorrela-
(8) tion function . Conversely, the autocorrelation can be given
Referring to [11] by the inverse Fourier transform of the power spectrum. That is
(9) (12)
(a) (b)
(c) (d)
Fig. 6. Theoretical predictions of the power spectra. (a) Input current with < = 0. (b) Input current with < = 0 2. (c) Switch voltage with < = 0. (d) Switch
:
voltage with < = 0:2.
and For
(13) (16)
It should be noted that (15) is a general expression for RCF

By this relationship, the power spectrum of over the range of switching schemes under consideration if is substituted
positive frequency, , is easily observed from (11), i.e., by the Fourier transform of a cycle of the considered signal. In
Section II-C, the derivation of power spectrum of switch voltage
is based on (15).
1) Standard Constant Switching Frequency (Standard)
(14) Scheme: If the converter operates with standard scheme, (14)
is modified with
The expected value of the summation in (14) forms a geometric
series. Thus (17)
and
(18)
where is a constant. The power spectrum can be shown to be

for (15)
(19)
Detailed proofs are shown in the appendix.
Using the Poisson identity, the sum of exponential terms in (19)

is defined as the sum of delta functions over the spectrum. That
is
(20)
where .
Thus, the power spectrum of in the standard scheme is
(21)
(28)
By (3), the Fourier transform of a cycle of is
(22)
Thus, the squared absolute value is

(29)
By substituting (26)–(29) into (15), in the RCF scheme

can be found. As will be shown in Section III, will be-
(23) come close to a continuous spectrum as increases.
For the standard scheme, is determined by substituting C. Spectral Characteristics of the Switch Voltage Waveform
(23) into (21).
The waveform of the drain-source voltage of in
2) RCF Scheme: Assume that the randomness of is sub-
Fig. 2(d) is expressed as
ject to a probability density function , which has the uni-
form distribution with upper limit and lower limit (i.e., for
). Let denote the period deviation and de- for
note the level of randomness on for
(30)
(24a) where is the turns ratio of the flyback transformer and is
(24b) the output voltage, which is equal to
Hence, the probability density function of is of the form (31)
(25) where is the output load resistance. is the expected value

otherwise of the duty cycle in RCF scheme. It is the nominal value of the
duty cycle in standard scheme.
For a non-zero , the expected values of the terms in (15) are
found as follows:
(32)
(26) (33)
Following the similar approach for , the power spectrum

can be expressed as
(27)
for (34)
For
(35)
1) Standard Constant Switching Frequency (Standard)

Scheme: For the standard scheme, is a constant and
(36)
(41)
The Fourier transform of a cycle of is
Similar to the characteristics of the input current will be
changed from a discrete harmonic spectrum to a continuous
spectrum as the level of randomness in increases.
III. EXPERIMENTAL VERIFICATIONS

Fig. 3(a) shows the synthesis of RCF switching signal in the
(37) feedback network for the offline flyback converter shown in
Fig. 1. The noise generator circuit is shown in Fig. 3(b). The
and hence, its squared absolute value is given as transistor is used as a noise diode that gives constant noise signal
strength over wide frequency band. The ripple of noise output
and its DC value are adjustable by changing VR1
and VR2, respectively. The component values of the converter
are tabulated in Table I. The output power of the converter is
58 W. The output of the error amplifier is compared to a ran-
(38) domized sawtooth signal. A composite random signal, which
contains a fixed dc signal and a noise with maximum am-
2) RCF Scheme: In order to calculate , (34) is used. plitude , generates the randomized sawtooth signal. The
With the same assumptions of randomness of in (25), the level of randomness in the RCF scheme can also be defined
expected values of the terms in (34) can be expressed as follows, as
(42)
At the end of every switching cycle, the composite random

signal is sampled and then fed to a voltage controlled oscillator
(VCO) to generate the next sawtooth cycle. The nominal
switching frequency is set at 50 kHz. corresponds to
the standard PWM scheme and the switching frequency is 50
kHz. When , the switching frequency is uniformly
randomized within the frequency range from 45 kHz to 55 kHz.
Fig. 4 shows the experimental voltage and current waveforms
of the converter's main switch when and ,
respectively. The gate signal is measured before the driver in
(39)
Fig. 3(a). The output voltage under two cases is maintained
at 24 V with ripple voltage of about 400 mV. In Fig. 4(a),
the period of the input current pulses is fixed at 20 µs while
the one in Fig. 4(b) is changed in every cycle. Although RCF
can generate low-frequency noise within the pass band of the
output filter, the output can be regulated to an acceptable level
since the dynamics of the voltage output will be much slower
than the crossover frequency of the feedback path. Hence the
output voltage can be effectively stabilized as shown in Fig. 4.
Experimental measurements of the power spectra of [i.e.,
] and [i.e., ] are shown in Fig. 5. The
results were taken from a signal analyzer HP89410A with the
use of the Hanning window, 4096 time samples, and a sampling
(40) rate of 2.5 MHz.
(a)
(a)
(b)
(b) Fig. 8. Measured conducted emission. (a) < = 0. (b) < = 0 2.

:
Fig. 7. Variation of power spectrum with <. (a) Input current. (b) Switch
voltage. (43) associated with the solutions of (15), (21), (34) and (36), re-
spectively. Both analytical and experimental results are in close
It is important to note that the true power spectrum derived in agreement. It can be observed that the RCF technique substan-
Section II is based on infinite time-records of the RCF switching tially reduces the discrete switching frequency harmonics. The
signal. However, as pointed out in [15], power spectra obtained peak harmonic power of the RCF scheme is much lower than
from digital signal processing technique are strictly approxima- that of the standard scheme. Using the theory developed, a three-
tions because of the finite number of time records involved in the dimensional representation of the variations of the power spec-
calculation. In order to make a better comparison of the theoret- trum with respect to the value of are generated and shown in
ical power spectrum and the experimental ones ,a Fig. 7. When equals zero, the operation is same as the standard
mathematical compensation for the analytical power spectrum PWM scheme. Discrete harmonics at multiples of the switching
is performed by convoluting with the window func- frequency can be observed. It is important to note that the power
tion . That is spectrum changes gradually from discrete harmonics spectrum
to continuous noise spectrum with an increasing . The overall
envelope of the power spectrum also decreases as increases.
(43) For the measurement of the conducted EMI of the converter,
a LISN Farnell LSN30 and an EMC analyzer HP8591EM are
where is one of discrete frequency points given by FFT anal- used. Quasi-peak detector is used for all the testing. Fig. 8 shows
ysis, and the characteristic of the window function is subject the measured conducted emission under the above two cases
to the following factor in FFT-based processing. and the limits of CISPR Publication 22 Class A devices. When
1) Type of windowing function. , some discrete harmonics exist and exceed the limit. With
2) Number of sampled time-record. , the harmonics are spread over. The spectrum is within
3) Sampling rate. the limit. Moreover, there is an approximately 10 dB reduction
By considering the above factors, the modified analytical solu- in the envelope of the spectrum with RCF. This can demonstrate
tion is compared to the experimental from the the effectiveness of using RCF in suppression of EMI of the
FFT-based spectrum analyzer, using the same parameters as offline switched-mode power supply.
HP89410A. The RCF scheme is also compared with the FM scheme. Like
The analytical prediction of and are shown the RCF scheme, the switching frequency of the FM scheme is
in Fig. 6. All results are determined by taking the convolution allowed to vary within the frequency range from 45 kHz to 55
(a) (b)
(c)
Fig. 9. Power spectra of the conducted emissions. (a) Standard PWM. (b) FM of 2 kHz on standard PWM. (c) RCF scheme.
kHz. The FM scheme is modulated at 2 kHz. Fig. 9 shows a converter as a wide-spectrum noise. It has the effect of reducing
comparison of the conducted EMI when the converter is oper- conducted EMI emissions as demonstrated in an experimental
ated with prototype. The RCF technique involves the randomization of
1) standard PWM [Fig. 9(a)]; the carrier frequency, but has the inherent feature of keeping
2) FM of 2 kHz on standard PWM [2] [Fig. 9(b)]; duty cycle constant. The effects of the level of randomness
3) RCF scheme [Fig. 9(c)]. on the degree of spectrum spreading are studied. Analytical
In order to investigate the details of the spectral characteristics, prediction is verified with the practical measurements. The
a frequency span of 1 MHz–1.1 MHz is studied for improving implementation here is simple and only involves a slight modi-
the frequency resolution. It can be seen that the FM scheme fication on existing circuits using standard PWM technique. In
manages to reduce the discrete harmonic components to some conclusion, the RCF scheme offers a simple solution to reduce
extent when compared with the standard PWM scheme. The conducted EMI in power converters. In terms of conducted
discrete harmonics are formed by the modulating frequency EMI suppression, this study demonstrates that the RCF scheme
and the switching frequency [2]. Compared to the standard is better than the FM scheme which has been incorporated into
PWM, some discrete harmonics that do not exist in the standard the PWM scheme of many commercial power converters.
PWM scheme are generated with the FM. However, the RCF
scheme offers the best conducted EMI suppression among the APPENDIX
three schemes under consideration. denotes the double-summation of the expected term in
(14); thus, (14) becomes
IV. CONCLUSION
An analysis on the random carrier frequency PWM method
has been presented. The theory provides a mathematical (A.1)
platform for studying the spectral characteristic of this random
PWM scheme. Spread spectrum switching technique disperses (A.2)
the discrete switching harmonics in standard PWM dc/dc
By substituting (A.11) into (A.6):
(A.3)
It should be noted that is dependent on if
(A.4)
(A.12)
and is dependent on if
(A.5) Finally, (A.1) becomes
Hence, (A.3) can be shown to be
(A.6)
(A.13)
where
REFERENCES
[1] K. H. Billings, Ed., Switchmode Power Supply Handbook. New York:
McGraw Hill, 1989.
[2] F. Lin and D. Y. Chen, “Reduction of power supply EMI emission by
switching frequency modulation,” in Proc. IEEE Power Electron. Spec.
Conf., 1993, pp. 127–133.
[3] M. P. Kazmierkowski and F. Blaabjerg, “Impact of emerging technolo-
gies on PWM control of power electronic converters,” IEEE Ind. Elec-
tron. Soc. Newsletter, pp. 9–13, Dec. 1995.
[4] A. M. Stankovic, G. C. Verghese, and D. J. Perreault, “Analysis and syn-
thesis of randomised modulation schemes for power converters,” IEEE
Trans. Power Electron., vol. 10, no. 6, pp. 680–693, Nov. 1995.
[5] S. Y. R. Hui, I. Oppermann, and S. Sathiakumar, “Microprocessor based
(A.7) random PWM schemes for DC-AC power conversion,” IEEE Trans.
Power Electron., vol. 12, no. 2, pp. 253–260, Mar. 1997.
[6] T. Tanaka, T. Ninomiya, and K. Harada, “Random switching control in
dc–dc converters,” in Proc. IEEE Power Electron. Spec. Conf., 1989, pp.
Consider 500–507.
[7] D. S. Stone and B. Chambers, “The effect of carrier frequency modula-
(A.8) tion of PWM waveforms on conducted EMC problems in switched mode
power supplies,” EPE J., vol. 5, no. 3, pp. 32–37, Jan. 1996.
and [8] , “Effect of spread-spectrum modulation of switched mode power
converter PWM carrier frequencies on conducted EMI,” Electron. Lett.,
vol. 31, no. 10, pp. 769–770, May 1995.
for [9] , “Easing problems in switched mode power converters by random
(A.9)
for modulation of the PWM carrier frequency,” in Proc. Appl. Power Elec-
tron. Spec. Conf., 1996, pp. 327–332.
[10] Y. Shrivastava, S. Y. R. Hui, S. Sathiakumar, H. Chung, and K. K. Tse,
Thus, (A.7) becomes a real number as follow: “Effects of continuous noise in randomised switching dc–dc converters,”
Electron. Lett., vol. 33, no. 11, pp. 919–921, May 1997.
[11] D. Middleton, An Introduction to Statistical Communication
Theory. New York: McGraw-Hill, 1988.
[12] J. T. Boys, “Theoretical spectra for narrow-band random PWM wave-
forms,” Proc. Inst. Elect. Eng. B, vol. 140, no. 6, Nov. 1993.
(A.10) [13] Y. Shrivastava, S. Y. R. Hui, S. Sathiakumar, H. Chung, and K. K. Tse,
“Harmonic analysis of non-deterministic switching methods for dc–dc
power converters,” IEEE Trans. Circuits Syst.—Part I. , to be published.
Since , (A.10) is simplified as [14] Y. Shrivastava, S. Y. R. Hui, S. Sathiakumar, K. K. Tse, and H. Chung,
“A comparison of deterministic and non-deterministic switching
methods for dc–dc converters,” IEEE Trans. Power Electron., vol. 13,
pp. 1046–1055, Nov. 1998.
[15] M. M. Bech, J. K. Pedersen, F. Blaabjerg, and A. M. Trzynadlowski, “A
methodology for true comparison of analytical and measured frequency
(A.11) domain spectra in random PWM converters,” in Proc. IEEE Power Elec-
tron. Spec. Conf., 1998, pp. 36–43.
K. K. Tse (S'95) received the B.Eng. degree in S. Y. (Ron) Hui (SM'94) was born in Hong Kong
electrical engineering (with honors) from the Hong in 1961. He received the B.Sc. degree (with honors
Kong Polytechnic University, Hong Kong, in 1995 from the University of Birmingham, U.K. in 1984,
and the Ph.D. degree from the Department of and the D.I.C. and Ph.D. degrees from the Imperial
Electronic Engineering, City University of Hong College of Science, Technology, and Medicine,
Kong, Hong Kong, in 2000. London, U.K., in 1987.
He is a Research Fellow with the Department of He was a Lecturer in power electronics at the Uni-
Electronic Engineering, City University of Hong versity of Nottingham, U.K. from 1987 to 1990. In
Kong. He has authored over eighteen technical 1990, he went to Australia and took up a lectureship
papers in his research interests, which include at the University of Technology, Sydney, Australia,
computer-aided simulation technique, numerical where he became a Senior Lecturer in 1991. Later,
modeling methods, EMI reduction, and random switching scheme for dc–dc he joined the University of Sydney and was promoted to Reader of Electrical
converters. Engineering and Director of Power Electronics and the Drives Research Group
Dr. Tse received First Prize in the 1998 IEEE Postgraduate Student Paper in 1996. He is now a Chair Professor of Electronic Engineering at the City Uni-
Contest, IEEE Hong Kong Section, and third prize in 1998 IEEE Region 10 versity of Hong Kong. He has published over 130 technical papers, including
Postgraduate student paper contest. over 70 refereed journal publications. His research interests include all aspects
of power electronics.
Dr. Hui is a Fellow of the IEE, the IEAust and the HKIE. He is an Associate
Editor of the IEEE TRANSACTIONS ON POWER ELECTRONICS.
Henry Shu-Hung Chung (S'92–M'95) received the
B.Eng. degree (with first class honors) in electrical
engineering and the Ph.D. degree from The Hong
Kong Polytechnic University, Hong Kong, in 1991
and 1994, respectively.
Since 1995, he has been with the City University
of Hong Kong, where he is currently an Associate
Professor in the Department of Electronic Engi-
neering. His research interests include time- and H. C. So (M'95) was born in Hong Kong in 1968. He
frequency-domain analysis of power electronic received the B.Eng. degree in electronic engineering
circuits, switched-capacitor-based converters, from the City Polytechnic of Hong Kong, Hong
random-switching techniques, digital audio amplifiers, fuzzy-logic control, and Kong, in 1990 and the Ph.D. degree in electronic
soft-switching converters. He has authored over 105 technical papers including engineering from the Chinese University of Hong
over 47 journal papers. Kong, Hong Kong.
Dr. Chung received the China Light and Power Prize and was awarded the From 1990 to 1991, he was an Electronic Engineer
Scholarship and Fellowship of the Sir Edward Youde Memorial Fund, in 1991 with the Research and Development Division, Everex
and 1993, respectively. He is currently Chairman of the Council of the Sir Systems Engineering, Ltd., Hong Kong. From 1995
Edward Youde Scholar's Association and IEEE student branch counselor. He to 1996, he worked as a Post-Doctoral Fellow at the
was track chair of the technical committee on power electronics circuits and Chinese University of Hong Kong. He is currently a
power systems of IEEE Circuits and Systems Society in 1997–1998. He is Research Assistant Professor in the Department of Electronic Engineering, City
currently an Associate Editor of the IEEE TRANSACTIONS ON C IRCUITS AND University of Hong Kong. His research interests include adaptive signal pro-
SYSTEMS—PART I. He has also been listed in Marquis Who's Who in the World. cessing, detection and estimation, source localization, and wavelet transform.

Emi Papaer

Uploaded by

Copyright:

Available Formats

Emi Papaer

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Emi Papaer

Uploaded by

Copyright:

Available Formats

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 15, NO.

2, MARCH 2000 399

Analysis and Spectral Characteristics of a

duty cycle operation in the dc–dc converter. The variation of the

If denotes the Fourier transform of and de-

Hence, the autocorrelation of can be expressed as

Based on the Wiener-Khinchin theorem [11], the power spec-

It should be noted that (15) is a general expression for RCF

where is a constant. The power spectrum can be shown to be

Using the Poisson identity, the sum of exponential terms in (19)

Thus, the squared absolute value is

By substituting (26)–(29) into (15), in the RCF scheme

Hence, the probability density function of is of the form (31)

(25) where is the output load resistance. is the expected value

Following the similar approach for , the power spectrum

1) Standard Constant Switching Frequency (Standard)

III. EXPERIMENTAL VERIFICATIONS

At the end of every switching cycle, the composite random

(b) Fig. 8. Measured conducted emission. (a) < = 0. (b) < = 0 2.

By substituting (A.11) into (A.6):

It should be noted that is dependent on if

(A.5) Finally, (A.1) becomes

Hence, (A.3) can be shown to be

You might also like