Upload
others
View
33
Download
0
Embed Size (px)
Citation preview
Progress In Electromagnetics Research, PIER 100, 327–349, 2010
INVESTIGATION INTO TIME- AND FREQUENCY-DOMAIN EMI-INDUCED NOISE IN BISTABLE MULTI-VIBRATOR
H.-C. Tsai
Department of Electronic EngineeringCheng-Shiu UniversityNo. 840, Chengcing Rd., Neausong Township, Kaohsiung CountyTaiwan 83305, R.O.C.
Abstract—Electromagnetic interference (EMI) has a negative effectupon the performance of circuit communication systems. The presentstudy considers the case of EMI induced in a conducting wire, andderives equations to establish the effect of the EMI on a bistablemultivibrator. The validity of the equations is verified experimentally.The results indicate that the degree of influence of the EMI on thebistable oscillator depends on the interference power, the interferencefrequency, the induced power, the output resistance of the circuit,and the parasitic capacitance. Moreover, it is shown that theharmonic noise increases with an increasing interference amplitude andfrequency. The theoretical results are found to be in good agreementwith the experimental data.
1. INTRODUCTION
Since digital circuits can only exist in one of two different discretebinary states, i.e., “on and off” or “high and low”, they are morereliable than their analogue counterparts, whose characteristics andstates are continuous. As a result, digital IC devices are widely usedin many fields throughout engineering and industry. In developingIC circuits, the sensitivity of the device performance to externalfactors such as the component values, component aging, the ambienttemperature, interference signals, EMI and shielding has attractedparticular attention. With the prevalence of Radio Frequency (RF)applications [1, 2] nowadays, such as TVs and radios, amateur radio,
Corresponding author: H.-C. Tsai ([email protected]).
328 Tsai
wireless “walkie-talkies”, cell phones, wireless remote controllers, BlueTooth communication systems [3, 4], and so on, the environmentis rapidly becoming saturated with electromagnetic waves. Someof these electromagnetic waves, termed as “signals”, are useful andnecessary in that they fulfill a specific design purpose. However,others are undesirable since they interfere with the performance ofother electrical or electronic devices. For example, it is well knownthat a cell phone used in close proximity to a computer may cause adistortion of the image on the screen or a static blast from the speakers.Similarly, cell phones may disrupt the delicate flight instrumentationonboard an aircraft and are therefore banned in flight. Finally, in acommunications/broadcasting context, interference may be manifestedin the form of “cross talk”, a phenomenon in which the conversationsof others are detected when speaking over the cell phone, or mayresult in the reception of two different radio stations at the samefrequency. Whilst in some cases, the effects of EMI are no morethan a minor irritant (e.g., the screen distortion caused by the useof a wireless emitting device near a TV), in others, they may bepotentially lethal (e.g., the use of a cell phone onboard an aircraft). Asa result, it is essential that the effects of EMI on common electroniccomponents are thoroughly understood [5–19]. Accordingly, thepresent study considers the EMI response of a bistable multivibrator,a component with wide-ranging applications in the electronics field,including Ref. [20–22]. In a normal indoor environment, the intensityof the low frequency electromagnetic waves is greater than that ofthe high frequency waves, i.e., 0.03µT and 0.0067µT, respectively(as measured by the current authors using CA40/43 instrumentation(Chauvin Arnoux, France)). Therefore, in analyzing the effects of EMIon the bistable multivibrator, the present analyses focus specifically onthe effects of low frequency EMI.
2. EXPERIMENTAL SETUP
Figure 1 illustrates the experimental setup used in this study tomeasure the interference noise spectrum induced in the bistablemultivibrator. The arrangement has the advantage that both theamplitude and the frequency of the EMI noise can be easily controlled.As shown, the experimental system was shielded within a metal caseto suppress the effects of external noise, and the amplifier and currentsource were both battery operated. In the experiments, EMI witha frequency ranging from 200 ∼ 900Hz and an amplitude between0.2V and 0.9V was induced in a conducting wire (CW) positioned inthe air gap of a ferromagnetic toroid wrapped with a current-carrying
Progress In Electromagnetics Research, PIER 100, 2010 329
Bistable
circuit
A
B
Shielding
Low NoiseAmplifier
Personal
Computer
Printer
Oscilloscope or Signal Analyzer
R
CW
C C
RS
L
Figure 1. Experimental setup for noise measurement in bistablemultivibrator.
coil, where CW is a circular wire with aluminum alloy material, thediameter is 0.001 m. The EMI interference was detected initially asradiated noise, amplified by a 73.98 dB low-noise pre-amplifier suchthat it could be detected using an oscilloscope, and then coupled inseries with the bistable multivibrator in the form of conducted noise.The EMI signals were transmitted to the oscilloscope via a circuitcoupling and were then passed to a spectrum analyzer (model HPE4440A) controlled by a PC via an IEEE-488 bus in order to obtain thetime domain and frequency spectrum plots of the interference signals.
3. THEORETICAL ANALYSIS
Multivibrators are two-state, digital circuits which employ positivefeed-back during state changes. Three different classes of multivibratorexist, namely bistable, monostable, and astable. The bistablemultivibrator, also known as a flip-flop, latch or binary circuit,is characterized by two stable states, which persist indefinitelyunless deliberately changed. Of all the multivibrators, the bistablemultivibrator is one of the most commonly used (typically forapplications such as single bit information storage), and is thereforechosen for analysis purposes in the present study.
3.1. Basic Parameters of Bistable Multivibrator
Figure 2 presents the basic circuit of a bistable multivibrator. Let thelogical output state of the multivibrator be denoted as Q. When Q = 1,the inverter has a complementary logical output, i.e., Q=0. This state
330 Tsai
: Interfered point
RC1R2R1
V CC
RC2
C2C1
Q2Q1
V OUT
R5 R6
D1 D2
C3 C4R3 R4
D3 D4V IN
C C
RL
Figure 2. Bistable multivibrator circuit.
persists until an excitation signal is applied, at which point it flip-flopsto the complementary value. This new state is then maintained untila further excitation signal is applied, and so on. As shown in Fig. 2,the multivibrator circuit is symmetrical, i.e., RC1 = RC2, C1 = C2,R1 = R2, R3 = R4, R5 = R6, and C3 = C4, Q1 and Q2 are thesame type, but which is impossible to have the same characteristics.Furthermore, transistors Q1 and Q2 are of same type, but have differentcharacteristics such that when one is “on”, the other is “off”.
3.1.1. Circuit Design
Assume that when the power is initially turned on, Q1 is “on” and Q2
is “off”. Assume also that VCE1(sat) = 0.2V and RC1 = RC2 = 1K .From the circuit, it can be shown that
IC1(sat) =VCC − 0.2V
RC1=
1.4− 0.21K
= 1.2mA (1)
Check
βIB1 > IC1(sat) (2)
IB1(sat·min) ≥IC1(sat)
β=
1.2mA157
= 7.64µA (3)
Select VCE2(off ) = 1.05V
IC2(off ) =VCC − VCE2(off )
RC2=
1.4− 1.051K
= 0.35mA (4)
Progress In Electromagnetics Research, PIER 100, 2010 331
Obtain
R2 =VCE2(off ) − VBE1(sat)
IC2(off )=
1.05V − 0.7V
0.35mA= 1K (5)
Select R5 = 2.2K
IR5 =VBE1
R5=
0.72.2K
= 0.318 mA (6)
Obtain
IB1 = IC2(off ) − IR5 = 0.35 mA− 0.318mA= 32 µA > IB1(sat·min) = 7.64µA (7)
From Kirchhoff’s voltage law, it follows that
RC1 ¿ R1 + R6, RC2 ¿ R2 + R5 (8)
SelectVCE2(off ) = 1.05V ,
ImpliesVRC2 = 1.4V − 1.05V = 0.35V
Check
VRC2 = 1.4V 1K
1K + 1K + 2.2K= 0.33V ∼= 0.35V
Differential Circuit:Select
R3 = 12RC1 = 12K = R4, f = 20K , T = 50µs,
TakeC3 = C4 = 0.001µf
3.2. Electromagnetic Interference
In the present study, the interference source has the form of a current-carrying coil wound on a ferromagnetic toroid with an air gap. Whena current is passed through the coil, an EMI voltage of magnitude Veff
is induced in a conducting wire (CW) positioned within the air gap.The magnetic field density and flux density in the air gap are givenrespectively by [23]
Hg =µNI0
µ0 (2πr0 − lg) + µlg(9)
PrB =12
∫µ0H
2g dv =
12
∫HgBgdv (10)
332 Tsai
where µ0 is the permeability of free space, µ is the permeability of theferromagnetic material, I0 is the current within the coil, r0 is the meanradius of the toroid, and lg is the width of the air gap. The value ofPrB represents the average energy per second of the interference on theCW, and can be calculated from Eq. (10) or measured directly.
In general, any periodic signal can be represented by the followingcomplex Fourier series:
V (t) =∞∑
n=−∞Cnejnωt = C0 +
∞∑
n=1
2 |Cn| cos (nωt + ∠Cn) (11)
where
Cn =1T
∫ T2
−T2
V (t) ejnωtdt (12)
In the present experiments, the input signal of the ferromagnetic toroidhas the form of a periodic square wave, as shown in Fig. 3. In this case,the following Fourier transform applies:
Cn =1T
∫ t0
0V0 (t) e−jnωtdt =
V0t0T
sin 12nωt0
12nωt0
e−j 12nωt0 (13)
and
Kn =V0t0T
sin 12nωt0
12nωt0
(14)
∠Cn =12nωt0 (15)
Substituting Eqs. (14) and (15) into Eq. (11) gives
V (t) =V0
2+
∞∑
n=1
Kn cos(nωt0 +
nπ
2
), for t0 =
T
2(16)
Figure 4 shows the equivalent circuit of the ferromagnetic toroid.The following admittance can be obtained:
Y =(Rω + jnωL) + 1
jnωCω
(Rω + jnωL) 1jnωCω
=1
R2ω+n2ω2L2
[Rω+j
(nωR2
ωCω−nωL+n3ω3L2Cω
)]=|Y |∠θ1 (17)
I (jnω) = Y V (t) = C0Y0 +∞∑
n=1
2Kn |Y | cos (nωt + ∠Cn + θ1) (18)
Progress In Electromagnetics Research, PIER 100, 2010 333
From Eq. (9), it can be shown that
Veff = − 1dt
∫
s
~B · d~s = − 1dt
[µ0µNI (jnω)
µ0 (2πr0 − lg) + µlg·∆zd
]
=µ0µN (∆zd)
µ0 (2πr0 − lg) + µlg
∞∑
n=1
2Kn |Y |nω sin (nωt + ∠Cn + θ1)
=∞∑
n=1
2KKn |Y |nω sin (nωt + ∠Cn + θ1) (19)
K =µ0µN∆zd
µ0 (2πr0 − lg) + µlg(20)
Figure 5 shows the equivalent circuit of the experimental measurementsystem.
The Vout signal supplied to the oscilloscope and signal analyzer isgiven by
Vout = VeffK1
RS· K3∠θ3
K2∠θ2· Ri
K4∠θ4(21)
Tt0
Vi (t)
t
V0
Figure 3. Experimental inputsignal of ferromagnetic toroid.
i C
L Rω
ω
Figure 4. Equivalent circuit offerromagnetic toroid.
C CRS
Signal
Analyzer
Cce
Rm RLV eff
Figure 5. Equivalent circuit of experimental measurement system.
334 Tsai
where K1 = RS//Rm, Rm = 1/hoe//Rc2//R2//R4 for a small valueof Cce.
Q2 = K1 − j1
ωC= K2∠θ2 (22)
K2 =
[K2
1 +(
1ωC
)2] 1
2
θ2 = tan−1 − 1ωC
K1(23)
Q3 = K2∠θ2//RL = K3∠θ3 = a + bj (24)
K3 =1
(K1 + R2)2 +
(1
ωC
)2
[K1RL(K1+RL)+
R2
ω2C2
]2
+[K1RL
ωC−(K1+RL)
RL
ωC
]21/2
θ3 = tan−1K1RL
ωC − (K1 + RL) RLωC
K1RL (K1 + RL) + RLω2C2
(25)
Q4 = (RS//Rm + XC) //RL + XC + Ri = Q3 + XC + Ri
= a + Ri + j
(b− 1
ωC
)= K4∠θ4 (26)
K4 =
[(a + Ri)
2 +(
b− 1ωC
)2] 1
2
(27)
θ4 = tan−1 b− 1ωC
a + Ri(28)
From the above, it can be shown that
Vout =∞∑
n=1
2KKn |Y |nωK1K3Risin(nωt+∠Cn+θ1+θ3−θ2−θ4)RsK2K4
Vout =∞∑
n=1
C32KKn |Y |nωK1K3Ri
RsK2K4sin(nωt+∠Cn+θ1+θ3−θ2−θ4)
(29)
where the constant C3 is the effective inducted coefficient. Eq. (29)shows that the magnitude of the EMI-induced noise is governed bythe pulse height, the output load, the parasitic capacitance, theinterference frequency and the interference amplitude.
3.3. Noise Analysis
Conducting wires render IC circuits highly susceptible to the effectsof electromagnetic interference and noise [24–29]. When a periodic
Progress In Electromagnetics Research, PIER 100, 2010 335
square wave is applied across points A and B in Fig. 1, the CW inthe air gap of the ferromagnetic torroid induces a pulse voltage inthe bistable multivibrator circuit, which in turn produces a responsenoise spectrum. Fig. 6 shows the typical response wave for a periodicpulse with a period T . The periodic pulse function can be analyzed bytaking the discrete Fourier transformation of function Sn, where Sn isa function of A1 and A2 (i.e., the pulse heights), i.e.,
Sn =1T
0∫
−T0
−A1e−a(T0+t)e−jωntdt +
T0∫
0
A2e−ate−jωntdt
=1T
a− jωn
a2 + ω2n2[E − jD] (30)
whereE = A1e
−aT0 + A2 −A2e−aT0 cosωnT0 −A1 cosωnT0 (31)
D = −A2e−aT0 sinωnT0 + A1 sinωnT0 (32)
S2n =
1T 2(a2 + ω2n2)
[E2 + D2
](33)
in which n is an integer, ω = 2πf = 2π/T , a is the attenuation factorof the exponential function, and A1 and A2 are the amplitudes of theupper-half and lower-half periods of the EMI induced by the CW,respectively.
0.000 0.001 0.002
-600.0 m
-400.0 m
-200.0 m
0.0
200.0 m
400.0 m
600.0 m
Vol
tage
(V
)
Time (Second)
Measurement
Simulation
Figure 6. Experimental and simulation results for typical periodicpulse wave generated by periodic EMI signal with period T .
336 Tsai
s
eff
R
VCce Ro= Rs //Rm //R i
Figure 7. Simplified representation of EMI output port in Fig. 1.
The amplitude spectrum of the EMI current can be obtained byplotting Sn against discrete frequencies, i.e., ωn. The square of Sn hasdimensions A2 and corresponds to the current power spectrum Siλ(fn)over 2T [30].
Figure 7 shows the Norton equivalent output circuit of the bistablemultivibrator, where Cce is the parasitic capacitance between the drainand the source. Let Vce = VC and Cce = C, then
in = iC + i0 = Cd∆VC
dt+
∆VC
R0⇒ d∆VC
dt+
∆VC
R0C=
inC
(34)
where R0 = Rs//Rm//Ri, and ∆Vc is the variation in the voltageacross the capacitor.
Taking the Fourier series expansions of ∆Vc and iλ gives
inC
=1C
∞∑n=−∞
αn exp(jωnt) (35)
d(∆Vcn)dt
+∆Vcn
R0C=
1C
αn exp(jωnt) (36)
From which it can be shown that
∆Vcn = βn exp(jωnt) (37)
whereβn =
αnR0
1 + jωnR0C(38)
Therefore, the noise power spectrum, S∆V c(fn), of the bistablemultivibrator induced by the EMI is given by
S∆V c(fn) = 2Tβnβ∗n (39)
S∆V c(fn) = Siλ(fn)R2
0
1 + (ω1nR0C)2(40)
whereSiλ(fn) = 2Tαnα∗n = 2TS2
n (41)
Progress In Electromagnetics Research, PIER 100, 2010 337
RC2
RO2
VBM
RS
RO1
Veff
(a) (b)
Figure 8. Output equivalent circuits of (a) EMI signal and (b) bistablemultibibrator signal.
From Eqs. (5), (12) and (13), it can be shown that
S∆VC(fn) = 2T
R20
1 + (ω1nR0C)2
[1
T 2(a2 + ω2n2)(E2 + D2)
](42)
The total noise power induced in the bistable multivibratorcan be obtained by summing S∆V c(fn) over all possible integers,n. In identifying the relative magnitudes of the various harmoniccomponents within the power spectrum, the present analysis firstfinds the value of A (the pulse height) from the measured powerspectral intensity of the fundamental harmonic and then evaluates thepower spectral intensities of the higher-order harmonics. Adopting thisapproach, the typical experimental response wave shown in Fig. 6 canbe transformed into the noise spectrum shown in Fig. 9 and the valueof the EMI then quantified directly.
3.4. Effects of Basic Bistable Multivibrator Frequency onOutput Power
The basic output frequency of a bistable multivibrator has the formof a square wave. Generally speaking, the effects of the basic outputfrequency on the noise power spectrum of the multivibrator can bemitigated using an impedance-matching method. Fig. 8(a) shows theoutput equivalent circuit of the EMI signal supplied to the oscilloscopeor signal analyzer, where RO1 = 1/hoe//RC2//R2//R4//RL//Ri =45.21Ω. Meanwhile, Fig. 8(b) shows the equivalent circuit of thebistable multivibrator square wave output signal, where RO2 =1/hoe//RS//R2//R4//RL//Ri = 24.31Ω. The output powers of the
338 Tsai
EMI and multivibrator signals can be obtained respectively as
PEMI =(
Veff
50 + 45.21
)2
45.21 = 0.005(Veff )2 (43)
PBM =(
VBM
1k + 24.31
)2
24.31 = 2.32× 10−5(VBM )2 (44)
Here, PBM/PEMI = 0.0046. In other words, the output power ofthe bistable multivibrator is negligible compared to that of the EMIinduced signal and can therefore be neglected.
4. RESULTS AND DISCUSSION
Figure 6 presents the experimental and simulation results for thetypical periodic pulse function generated by a periodic EMI signalwith a period T . Note that in obtaining the simulation results usingEq. (29), the following parameter settings are assumed: V0 = 0.56V(at time t = 0), V0 = 0.45V (at time t = T/2), µ0 = 4πE − 7H/m,µr = 4000, N = 500, F = 500 Hz, Rω = 0.52Ω, r0 = 0.09m,lg = 0.005 m, Cω = 6.558E − 10F, 1/hoe = 30 kΩ, Rc1 = Rc2 = 1 kΩ,R1 = R2 = 1kΩ, R3 = R4 = 12 kΩ, C = 100E − 6F, ∆z = 0.01m,d = 0.001 m, Rs = 50 Ω, C3 = 0.081, XC = 1/(2πfC), RL = 220 kΩ,L = 1E − 4H, ω = 2πf and Ri = 50 Ω. Fig. 6 confirms that agood agreement exists between the simulated pulse function and theexperimental function. In the experimental setup shown in Fig. 1,Cce = 12.77E − 8F, RL = 220 kΩ, and T0 = 0.4990T . Utilizing avalue of a = 13622 in Eq. (42), Fig. 9 compares the experimentaland simulation results for the harmonic wave component of the noisespectrum induced in the bistable multivibrator by EMI with anamplitude of VP = 0.4V and a frequency of f = 500Hz. Again, agood agreement is observed between the two sets of results. To furtherinvestigate the effect of EMI on the noise spectrum induced in thebistable multivibrator, the amplitude of the AC interference signal wasvaried between 0.2 V and 0.9 V while the frequency was maintained at aconstant f = 500 Hz. Fig. 10 presents the corresponding experimentaland simulation results for the odd-order harmonics of the inducednoise spectrum. Note that in obtaining the simulation results usingEq. (42), A1 ranges from 5364 ∼ 33006µA while A2 ranges from 5514∼ 33930µA. It is observed that a good agreement is obtained betweenthe experimental and simulation results at all values of Vp.
Table 1 compares the measured and simulated peak values of theharmonic components of the maximum noise power spectral intensityfor EMI with amplitudes in the range VP = 0.2 ∼ 0.9V and a constant
Progress In Electromagnetics Research, PIER 100, 2010 339
0 2000 4000 6000 8000 10000
-80
-60
-40
-20
0
20
40
SVc
(fn )
(DB
mA
/Hz^
1/2)
Frequency (Hz)
:Measurement :Simulation
Figure 9. Harmonic wave component of bistable multivibrator noisespectrum response for VP = 0.4V and f = 500 Hz. (Note that inobtaining simulation results using Eq. (42), A1 and A2 are specified as12208µA and 12549µA, respectively.)
0 2000 4000 6000 8000 100000
10
20
30
40
50
SVc
(fn)
(DB
mA
/Hz^
1/2)
Frequency(Hz)
All f=500 Hz1.Vp=0.9 V Measurement
Simulation2.Vp=0.8 V Measurement
Simulation3.Vp=0.7 V Measurement
Simulation4.Vp=0.6 V Measurement
Simulation5.Vp=0.5 V Measurement
Simulation6.Vp=0.4 V Measurement
Simulation7.Vp=0.3 V Measurement
Simulation8.Vp=0.2 V Measurement
Simulation
Figure 10. Experimental and simulation results for odd-orderharmonics of noise spectrum induced by EMI interference withfrequency of 500 Hz and amplitude of Vp = 0.2 ∼ 0.9V. (Simulationresults computed using Eq. (42) shown using filled symbols).
340 Tsai
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 100000
10
20
30
40
SVc
(fn )(
DB
mA
/Hz^
1/2)
Frequency(Hz)
All Vp=1 V
1.F=700 Hz Measurement Simulation
2.F=600 Hz Measurement Simulation
3.F=500 Hz Measurement Simulation
4.F=400 Hz Measurement Simulation
5.F=300 Hz Measurement Simulation
Figure 11. Experimental and simulation results for odd-orderharmonics of noise spectrum induced by EMI interference frequency of300 ∼ 700Hz and amplitude of Vp = 1 V. (Simulation results computedusing Eq. (42) shown using filled symbols.)
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 100000
10
20
30
40
SVc
(fn )(
DB
mA
/Hz^
1/2)
Frequency(Hz)
All Vp=1 V
1.F=700 Hz Measurement Simulation
2.F=600 Hz Measurement Simulation
3.F=500 Hz Measurement Simulation
4.F=400 Hz Measurement Simulation
5.F=300 Hz Measurement Simulation
Figure 12. Experimental and simulation results for odd-orderharmonics of noise spectrum induced by EMI interference frequencyof 300 ∼ 700Hz and variable amplitude parameters A1 and A2 (seeTable 3). (Simulation results computed using Eq. (42) shown usingfilled symbols.)
Progress In Electromagnetics Research, PIER 100, 2010 341
Table 1. Comparison of experimental and theoretical results formaximum noise power spectral intensity for various values of Vp.
Frequency fixed, Amplitude variable (First harmonic wave)
Vp (V) 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Frequency
(Hz)500 500 500 500 500 500 500 500
Measurement
(dBmA/Hz1/2)21.1 24.6 28 30.2 32.1 33.8 35.3 36.6
Simulation
(dBmA/Hz1/2)20.90 24.6 28.07 30.18 32.15 33.82 35.47 36.71
Table 2. Comparison of experimental and theoretical results formaximum noise power spectral intensity for various values of frequency.
Amplitude fixed, Frequency variable (First harmonic wave)
Frequency (Hz) 200 300 400 500 600 700 800 900
Measurement
(dBmA/Hz1/2)25.6 27.7 29.1 30.2 31 31.8 32.3 32.7
Simulation
(dBmA/Hz1/2)26.39 28.1 29.29 30.18 30.87 31.43 31.89 32.27
frequency of f = 500Hz, i.e., the harmonic value varies only as afunction of the interference amplitude while the remaining parametersare constant. Table 2 compares the experimental and simulation resultsfor the maximum noise power spectral intensity for EMI frequenciesin the range 200 ∼ 900Hz and a constant interference amplitude ofA1 = 15565µA and A2 = 16001µA, i.e., the noise power spectralintensity varies only as a function of the AC interference frequencywhile the other parameters remain constant. Fig. 11 presents theexperimental and simulation results for the odd-order harmonics ofthe noise spectrum induced by EMI interference with a frequencyof 300 ∼ 700Hz and an amplitude of Vp = 1 V. Although a −2 dBdifference is observed between the two sets of results, it is clear thatthe EMI frequency has a significant effect upon the noise response ofthe multivibrator. By tuning the amplitude parameters in recognitionof the fact that a variable EMI frequency affects both the attenuationfactor and the induced wave type, the same degree of simulationaccuracy as that shown in Fig. 10 can be obtained, as shown inFig. 12 and Table 3. In other words, the EMI frequency is a function
342 Tsai
of the EMI amplitude, as proven in Eqs. (29) and (42). AlthoughFig. 11 shows that the frequency has a significant effect on theodd-order harmonic waves, i.e., Eq. (42) denotes dω/dA1 6= 0 anddω/dA2 6= 0 for the nonlinear function S∆V c(fn), Fig. 9 shows thatthe difference between the experimental and simulation results for thevariable amplitude case is less than that for the variable frequency case.In other words, the amplitude of the EMI has a greater effect on thenoise response of the bistable multivibrator than the EMI frequency.In practice, this finding implies that the parasitic capacitance andthe dynamic input resistance of the multivibrator vary with a varyingfrequency.
Table 3. Comparison of experimental and theoretical results formaximum noise power spectral intensity for various values of frequencyand amplitude.
Amplitude, Frequency variable (First harmonic wave)Frequency (Hz) 300 400 500 600 700
InducedCurrentA1 (µA)
13165 14065 15565 16065 16865
InducedCurrentA2 (µA)
13533 14459 16001 16515 17337
Measurement(dBmA/Hz1/2)
27.7 29.1 30.2 31 31.8
Simulation(dBmA/Hz1/2)
26.65 28.41 30.18 31.24 32.13
Table 4. Comparison the simulations for various values of frequency,amplitude and attenuation factor with Vp = 0.2V, f = 500 Hz.
Parameters a A1 (µA) A2 (µA)Remarks
(Shown in)Curve a. 13622 5364 5514 Figs. 10, 13Curve b. 11376 4624 4753 Fig. 13Curve c. 8622 3642 3742 Fig. 13
Progress In Electromagnetics Research, PIER 100, 2010 343
Table 5. Relative SNR of current induced by EMI.
VP (V) 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2
Induced
Current
A1 (µA)
33006 28643 23661 19534 15565 12208 8193 5364
Induced
Current
A2 (µA)
33930 29445 24323 20081 16001 12549 8422 5514
Induced
Peak
Current
Ni (µA)
6.6936 5.8088 4.7984 3.9615 3.1566 2.4757 1.6615 1.0878
Relative
SNR
(dBµA)
3.487∗ 4.718∗ 6.378∗ 8.043∗ 10.016∗ 12.126∗ 15.590∗ 19.269∗
*where SNR = 20 log(Si/Ni), Si = 10 µA,
Ni = ((A1 + A2)/2)/5000 and f = 500Hz.
If we want to probe the more accuracy of Fig. 10, the attenuationfactor a must be considered as a variable of power spectral intensityfunction. We take a curve of VP = 0.2V in Fig. 10 to illustrate theresult. Table 4 shows the noise power spectral intensity varies as afunction of the AC interference frequency, amplitude and attenuationfactor a, we can find that the simulation of curve b and c in Fig. 13reveal the more accuracy than curve a. In general, the simulation andexperimental results presented above indicate that the magnitude ofthe EMI induced by the CW is affected by the amplitude, frequencyand period of the interference signal, the drain resistance and parasiticcapacitance of the multivibrator, the output load, and parametera. Table 5 indicates the relative signal-to-noise (SNR) ratio of themultivibrator for interference amplitudes in the range VP = 0.2 ∼0.9V, a constant frequency of f = 500Hz and a reference signalSi of 10µA. The CW used in the present EMI experiments has alength of 1 cm. Thus, if the CW in a practical electrical circuit hasa length of 1 m (say), the EMI effects will potentially be magnified 100times compared to that observed in the present study. For example,assuming an induced current of 3.157µA, a CW of length 1 cm, anda load resistance of 45 Ω, the induced voltage is 142µV. This valuemay then be amplified by up to 1000 times in a practical system, andthus a significant EMI effect occurs. According to CISPR 11 (The
344 Tsai
International Special Committee on Radio Interference), EMI shouldbe limited to no more than 110 dBµV in the interference frequencyrange 0.009 ∼ 0.050MHz, equivalent to a voltage of 316227µV. As aresult, a useful design guideline of 2.226 cm for the maximum lengthof the CW in bistable multivibrator circuits can be inferred. Thebasic theorem presented in this paper is applicable to different circuits.However, the actual derived formulae are specific to the particularcircuit. If the EMI perturbs the Wien bridge oscillator (WBO) asFig. 14, the same results as those obtained in the present discussions.Fig. 15 compares the experimental and simulation results obtainedfor the harmonic components in the noise spectrum of a WBO forVP = 0.3V ∼ 1.0V. Comparing these results with those presented inFig. 10 for the bistable multivibrator, it can be seen that the resultsare broadly similar. However, the function level is somewhat higherdue to the different parameters of the two circuits.
The preceding discussions have considered the effects of lowfrequency interference. However, as shown in the following, the basictheorems developed in this study can also be applied to the case ofhigh frequency interference. Fig. 16 shows that a good agreement isobtained between the experimental results and the simulation resultsfor the noise spectrum responses induced by EMI with a frequency of500 kHz with Vp = 0.4 ∼ 1.6 V of MOSFET circuit.
0 2000 4000 6000 8000 100000
10
20
30
40
50
SV
c (f
n)(
DB
mA
/Hz^1
/2)
Frequency(Hz)
All f=500Hz
1.Vp=0.2V Measurement
a. Simulation
b. Simulation
c. Simulation
Figure 13. The noise power spectral intensity varies as a function ofthe AC interference frequency, amplitude and the attenuation factor awith Table 4 data.
Progress In Electromagnetics Research, PIER 100, 2010 345
+
R
C
C=0.01
R=1K
R2=10K
Y
R1=1K
X
+ 10 f
RB1
=1M
RC1
=5K
+
RE1
=1K
D
RB2
=1M
RC2
=5K
+
VCC=10V
RE2
=1K
+
Oscilloscope
or Signal
Analyzer
C C
RL2
D is the interfered
point of EMI signal
− − −
−
µ
10 f µ10 f µ
µ
10 f µ
Figure 14. The circuit of WBO.
0 2000 4000 6000 8000 10000 12000 140000
10
20
30
40
50
SV
c (f
n)(
dB
mA
/Hz^1
/2)
Frequency(Hz)
All f=500Hz
1.Vp=1.0V Measurement
Simulation
2.Vp=0.9V Measurement
Simulation
3.Vp=0.8V Measurement
Simulation
4.Vp=0.7V Measurement
Simulation
5.Vp=0.6V Measurement
Simulation
6.Vp=0.5V Measurement
Simulation
7.Vp=0.4V Measurement
Simulation
8.Vp=0.3V Measurement
Simulation
Figure 15. Experimental and simulated results for odd-orderharmonics at 500Hz with Vp = 0.3V to 1.0V of WBO.
346 Tsai
0 1000000 2000000 3000000 4000000 5000000-50
-40
-30
-20
-10
0
10
S?
Vc (f
n)(
DB
mA
)
Frequency(Hz)
All f=500 kHz
1.Vp=1.6V Measurement
Simulation
2.Vp=1.4V Measurement
Simulation
3.Vp=1.2V Measurement
Simulation
4.Vp=1.0V Measurement
Simulation
5.Vp=0.8V Measurement
Simulation
6.Vp=0.6V Measurement
Simulation
7.Vp=0.4V Measurement
Simulation
Figure 16. Experimental and simulated results at 500 kHz withVp = 0.4V to 1.6 V of MOSFET circuit.
5. CONCLUSIONS
Experimental and theoretical methods have been used to characterizethe noise spectrum of a bistable multivibrator subject to periodic EMIinduced via a CW. A good agreement has been observed between theexperimental and simulation results for the noise spectral intensity ofthe bistable multivibrator in both the time domain and the frequencydomain. In general, the results have shown that the noise responseof the multivibrator is significantly affected by EMI. Specifically, thedegree of the EMI effect on the bistable oscillator is determined bythe radiated power and the following circuit parameters: f , V0, µ0,µr, N , Rω, r0, lg, Rs, Rc1, Rc2, R1, R2, R3, R4, Cω, Cce, Ri, ∆z,d, L, A1, A2, a, T0, RL and C. Furthermore, it has been shownthat the magnitude of the induced interference current increases withan increasing interference frequency and an increasing interferenceamplitude. Of these two characteristics, the amplitude has a greatereffect on the noise response of the multivibrator than the frequency.However, a variable EMI frequency influences both the attenuationfactor and the induced wave type. In accordance with CISPR andEN norms, the results presented in this study imply that the CWsused in practical bistable multivibrator circuits should not exceed alength of 2.226 cm. Whilst the methods outlined in this study havefocused specifically on the case of bistable multivibrators, they areequally applicable to the EMI analysis of all general wavelength-basedelectronic devices.
Progress In Electromagnetics Research, PIER 100, 2010 347
ACKNOWLEDGMENT
The author wishes to acknowledge the invaluable assistance providedby Haw-Jiun Liang in the course of this study.
REFERENCES
1. Masya, S., S. Takehiko, and N. Hideya, “Numerical simulation ofa potassium-seeded turbulent RF inductively coupled plasma withparticles,” Thin Solid Films, Vol. 435, No. 1–2, 5–12, July 1, 2003.
2. Kwang, J. K., H. K. Jong, and H. K. Ji, “Structural and opticalcharacterization of Cu3N films prepared by reactive RF magnetronsputtering,” Journal of Crystal Growth, Vol. 222, No. 4, 767–772,February, 2001.
3. Xue, J. X. and T. D. Todd, “Basestation collaboration inbluetooth voice networks,” Computer Networks, Vol. 41, No. 3,289–301, February 21, 2003.
4. Dai, D., “Bluetooth,” Network Security, Vol. 2002, No. 4, 11–12,April 1, 2002.
5. Khah, S. K., T. Chakravarty, and P. Balamurali, “Analysis ofan electromagnetically coupled microstrip ring antenna usingan extended feedline,” Journal of Electromagnetic Waves andApplications, Vol. 23, No. 2–3, 369–376, 2009.
6. Hong, J.-I., S.-M. Hwang, and C.-S. Huh, “Susceptibility ofmicrocontroller devices due to coupling effects under narrow-band high power electromagnetic waves by magnetron,” Journalof Electromagnetic Waves and Applications, Vol. 22, No. 17–18,2451–2462, 2009.
7. Tsai, H. C. and K. C. Wang, “Investigation of EMI-inducednoise spectrum on an enhancement-type MOSFET,” Solid StateElectronics, Vol. 52, No. 8, 1207–1216, 2008.
8. Kim, Y. J., U. Choi, and Y. S. Kim, “Screen filter designconsiderations for plasma display panels (PDP) to achieve a highbrightness with a minimal loss of EMI shielding effectiveness,”Journal of Electromagnetic Waves and Applications, Vol. 22,No. 5–6, 775–786, 2008.
9. Tsai, H. C., “Numerical and experimental analysis of EMI effectson circuits with MESFET devices,” Microelectronics Reliability,Vol. 48, No. 4, 537–546, 2008.
10. Wessels, P., M. van Swanenberg, H. Zwol, B. Krabbenborg,H. Boezen, M. Berkhout, et al., “Advanced BCD technology
348 Tsai
for automotive, audio and power applications,” Solid StateElectronics, Vol. 51, No. 2, 195–211, 2007.
11. Tsai, H. C., “An investigation on the influence of electromagneticinterference induced in conducting wire of universal LEDs,”Microelectronics Reliability, Vol. 47, No. 6, 959–966, 2007.
12. Tsai, H. C., “An investigation into EMI-induced noise in nanome-ter multi-quantum well InGaN LEDs,” Optics Communications,Vol. 273, No. 2, 311–319, 2007.
13. Michette, A. G., S. J. Pfauntsch, A. Erko, A. Firsov, andA. Svintsov, “Nanometer focusing of X-rays with modifiedreflection zone plates,” Optics Communications, Vol. 245, No. 1–6,249–253, 2005.
14. Busatto, G., L. Fratelli, C. Abbate, R. Mamzo, and F. Iannuzzo,“Analysis and optimisation through innovative driving strategyof high power IGBT performances/EMI reduction trade-offfor converter systems in railway applications,” MicroelectronicsReliability, Vol. 44, No. 9–11, 1443–1448, 2004.
15. Quach, H. P. and C. P. Chui Talso, “Low temperature magneticproperties of Metglas 2714A and its potential use as core materialfor EMI filters,” Cryogenics, Vol. 44, No. 6–8, 445–449, 2004.
16. Steinecke, T., H. Koehne, and M. Schmidt, “Behavioral EMImodels of complex digital VLSI circuits,” MicroelectronicsJournal, Vol. 35, No. 6, 547–555, 2004.
17. Vasa, P., B. P. Singh, P. Taneja, and P. Ayyub, “Antiresonant ringinterferometry as a sensitive technique for measuring nonlinearoptical properties of thin films,” Optics Communications, Vol. 233,No. 1–6, 297–304, 2004.
18. Tsai, H. C., “Reliable study of digital IC circuits with marginvoltage among variable DC power supply, electromagneticinterference and conducting wire antenna,” MicroelectronicsReliability, Vol. 43, No. 12, 2001–2009, 2003.
19. Tsai, H. C., “Impact of electromagnetic radiation induced byconducting wire antenna on optical fiber communication systems,”Optics Communications, Vol. 223, No. 1–3, 81–89, 2003.
20. Lee, T.-H. and H.-D. Kim, “Radiation hardness test ofpreamplifier circuits composed of commercial bipolar transistors,”Nuclear Inst. and Methods in Physics Research, A, Vol. 579, No. 1,260–263, 2007.
21. Zhao, Z., Y. M. Li, J. Cheng, and Y. F. Xu, “Current sensorutilizing giant magneto-impedance effect in amorphous ribbontoroidal core and CMOS inverter multivibrator,” Sensors &
Progress In Electromagnetics Research, PIER 100, 2010 349
Actuators: A. Physical, Vol. 137, No. 1, 64–67, 2007.22. Muselli, M., “Frequency analysis of binary oscillators triggered by
a random noise,” Physica D, Vol. 188, No. 1–2, 119–133, 2004.23. Cheng, D. K., Field and Wave Electromagnetics, 2nd edition, 252–
634, Addison-Wesley Publishing Company, USA, 1989.24. Bhatia, V. and B. Mulgrew, “Non-parametric likelihood based
channel estimator for Gaussian mixture noise,” Signal Processing,Vol. 87, No. 11, 2569–2586, 2007.
25. Chambers, J., D. Bullock, Y. Kahana, A. Kots, and A. Palmer,“Developments in active noise control sound systems for magneticresonance imaging,” Applied Acoustics, Vol. 68, No. 3, 281–295,2007.
26. Yuan, J., “Adaptive Laguerre filters for active noise control,”Applied Acoustics, Vol. 68, No. 1, 86–96, 2007.
27. Roelant, R., D. Constales, G. S. Yablonsky, R. Van Keer,M. A. Rude, and G. B. Marin, “Noise in temporal analysisof products (TAP) pulse responses,” Catalysis Today, Vol. 121,No. 3–4, 269–281, 2007.
28. Lisnanski, R. and A. J. Weiss, “Low complexity generalizedEM algorithm for blind channel estimation and data detectionin optical communication systems,” Signal Processing, Vol. 86,No. 11, 3393–3403, 2006.
29. Li, H. G. and G. Meng, “Detection of harmonic signals fromchaotic interference by empirical mode decomposition,” Chaos,Solitons and Fractals, Vol. 30, No. 4, 930–935, 2006.
30. Van Der Ziel, A., Noise in Solid State Devices and Circuits, 10–20,Wiley, New York, 1986.