Uncertainties in extracting radiation parameters for an ... · UncertaintiesinExtracting...
Transcript of Uncertainties in extracting radiation parameters for an ... · UncertaintiesinExtracting...
co.
VNAT'L INST. OF STAND & TECH
A111DS TbfiVflE
\
NBS TECHNICAL NOTE 1064
U.S. DEPARTMENT OF COMMERCE / National Bureau of Standards
Uncertainties in Extracting
Radiation Parameters for anUnknown Interference SourceBased on Power and Phase
Measurements
NATIONAL BUREAU OF STANDARDS
The National Bureau of Standards' was established by an act of Congress on March 3, 1901.
The Bureau's overall goal is to strengthen and advance the Nation's science and technology
and facilitate their effective application for public benefit. To this end, the Bureau conducts
research and provides: (1) a basis for the Nation's physical measurement system, (2) scientific
and technological services for industry and government, (3) a technical basis for equity in
trade, and (4) technical services to promote public safety. The Bureau's technical work is per-
formed by the National Measurement Laboratory, the National Engineering Laboratory, and
the Institute for Computer Sciences and Technology.
THE NATIONAL MEASUREMENT LABORATORY provides the national system of
physical and chemical and materials measurement; coordinates the system with measurement
systems of other nations and furnishes essential services leading to accurate and uniform
physical and chemical measurement throughout the Nation's scientific community, industry,
and commerce; conducts materials research leading to improved methods of measurement,
standards, and data on the properties of materials needed by industry, commerce, educational
institutions, and Government; provides advisory and research services to other Government
agencies; develops, produces, and distributes Standard Reference Materials; and provides
calibration services. The Laboratory consists of the following centers:
Absolute Physical Quantities 2 — Radiation Research — Chemical Physics —Analytical Chemistry — Materials Science
THE NATIONAL ENGINEERING LABORATORY provides technology and technical ser-
vices to the public and private sectors to address national needs and to solve national
problems; conducts research in engineering and applied science in support of these efforts;
builds and maintains competence in the necessary disciplines required to carry out this
research and technical service; develops engineering data and measurement capabilities;
provides engineering measurement traceability services; develops test methods and proposes
engineering standards and code changes; develops and proposes new engineering practices;
and develops and improves mechanisms to transfer results of its research to the ultimate user.
The Laboratory consists of the following centers:
Applied Mathematics — Electronics and Electrical Engineering2 — Manufacturing
Engineering — Building Technology — Fire Research — Chemical Engineering2
THE INSTITUTE FOR COMPUTER SCIENCES AND TECHNOLOGY conducts
research and provides scientific and technical services to aid Federal agencies in the selection,
acquisition, application, and use of computer technology to improve effectiveness and
economy in Government operations in accordance with Public Law 89-306 (40 U.S.C. 759),
relevant Executive Orders, and other directives; carries out this mission by managing the
Federal Information Processing Standards Program, developing Federal ADP standards
guidelines, and managing Federal participation in ADP voluntary standardization activities;
provides scientific and technological advisory services and assistance to Federal agencies; and
provides the technical foundation for computer-related policies of the Federal Government.
The Institute consists of the following centers:
Programming Science and Technology — Computer Systems Engineering.
'Headquarters and Laboratories at Gaithersburg, MD, unless otherwise noted;
mailing address Washington, DC 20234.
'Some divisions within the center are located at Boulder, CO 80303.
Uncertainties in Extracting
Radiation Parameters for anUnknown Interference SourceBased on Power and PhaseMeasurements
N3SRESEARCH
WONce;
QcaOS
\1<3
Mark T. MaGalen H. Koepke
Electromagentic Fields Division
National Engineering LaboratoryNational Bureau of StandardsBoulder, Colorado 80303
&*-**
\e
U.S. DEPARTMENT OF COMMERCE, Malcolm Baldrige, Secretary
NATIONAL BUREAU OF STANDARDS, Ernest Ambler, Director
Issued June 1983
National Bureau of Standards Technical Note 1064
Natl. Bur. Stand. (U.S.), Tech. Note 1064, 48 pages (June 1983)
CODEN: NBTNAE
U.S. GOVERNMENT PRINTING OFFICEWASHINGTON: 1983
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 20402
Price $3.75
(Add 25 percent for other than U.S. mailing)
CONTENTS
Page
1. Introduction 1
2. Derivation of Potential Errors 9
3. A Simulated Example with Assumed Unbiased Measurement Errors 18
4. Another Simulated Example with Assumed Biased Measurement Errors.. 26
5. Sensitivity of Extracted Parameters with Respect to MeasurementAccuracy 32
6. Concluding Remarks 35
7
.
References 37
m
Uncertainties in Extracting Radiation Parameters for an
Unknown Interference Source Based on Power and Phase Measurements
Mark T. Ma and Galen H. Koepke
Electromagnetic Fields DivisionNational Bureau of Standards
Boulder, Colorado 80303
A method for determining the radiation characteristics of a
leaking interference source has been reported in a previous publi-cation [1], in which the unintentional, electrically small leakagesource was modeled by two vectors representing a combination ofequivalent electric and magnetic dipole moments. An experimentalsetup, measurement procedures, and the necessary theoretical basiswere all described therein to explain how the relevant sourceparameters can be extracted from the measurement data of outputpowers and phases taken when the interference source is placedinside a transverse electromagnetic (TEM) cell. Simulated exam-ples were also given to show that the equivalent source parametersof unknown vector dipole moments and, thus, the detailed radiationpattern and the total power radiated by the source in free spacecould be uniquely determined by the method if the measurement datawere not contaminated by noise. This report presents the mathe-matical analysis of the uncertainties in the final, extractedresults when the experimental data are degraded by the backgroundnoise and measurement inaccuracies.
Key words: dipole moments, electrically small source, erroranalysis, interference sources, phase measurements, power measure-ments, radiation pattern, TEM cell, total radiated power, uncer-tainties.
1. Introduction
It was shown in previous reports [1,2] that a general unknown leakage
interference source may be characterized by a composite set of parameters
representing both the equivalent vector electric and magnetic dipole moments
as displayed in figure 1. Necessary theoretical justifications were also
given therein to demonstrate that these parameters of twelve unknowns (six
dipole moment amplitudes and six dipole moment phases) may be determined, with
the experimental setup shown in figure 2 when the unknown source is placed
inside a TEM cell, by taking the output sum and difference powers and the
relative phases between the sum and difference ports for six different source
orientations. The corresponding radiation pattern and the total power radi-
ated by the source in free space can also be computed. Naturally, when the
experimental power and phase readings are not contaminated by noise or other
inaccuracies, the source parameters so extracted and the radiation character-
istics so computed are accurate and unique. In the practical world, however,
the experimental data are always degraded somewhat by the background noise,
equipment limitations, and the reading accuracy. The question is then: how
will these measurement imperfections affect the final results of source param-
eters? This report is intended to answer this question by giving necessary
mathematical derivations for, and performing the analysis of, these uncertain-
ties.
For the convenience of presenting the material, all the equations
required for this report are reproduced as follows with the detailed deriva-
tions and related measurement procedures referred to the previous report [1]:
"4 = ( psl
+ Ps2 " P
s3 - Ps4
+ Ps5
+ Ps6>/<2q
2) U>
4y - (Psl
+ Ps2
+ Ps3
+ Ps4 " P
s5 " Ps6>/<
2<2
> < 2 >
4 " (" Psl " P
s2+ P
s3+ P
s4+ P
s5+ Ps6>/f 2q
2) < 3 >
mexmeycose
el= (P
sl- P
s2 )/(2q2
) (4)
meymez
cosee2
= (Ps3 " p
S4)/(2q2
) (5)
m.,2
ezmex
cosee3 = {P
s5 " Ps6^ (2(^V (6)
eel ~ ^ex " ^ey »
ee2E
^ey " ^ez ee3=
*ez " ^ex > {7)
"4 = < Pdl+ Pd2 - Pd3 " P
d4+ Pd5
+ Pd6>/< 2kV> (8)
"my " < Pdl+ P
d2+ P
d3+ P
d4 " Pd5 " Pd6>/< 2kV) W
mmz = <"Pdl " p
d2+ P
d3+ P
d4+ P
d5+ P
d6>/< 2kV> (10)
"W%y cos0ml = (P
d2 " pdl>/( 2kV> UD
mmymmz cos9m2 = (p d4 " pd3
)/(2k2cl2 > (12)
"WW cosem3 " (P
d6 " pd5>/< 2k2^ < 13 >
eml ~ ^mx " V 9m2
E V " ^mz >em3
E%\z " *mx
* (14)
Here mex , m , and m
ez are the x-, y-, and z-components of the amplitude of
the equivalent vector electric dipole moment; cj;
ex , ^ey>and (\>ez
are the
associated phases of the electric dipole moment; mmx , n^ , and mmz are the x-,
y- 5 and z-components of the amplitude of the equivalent vector magnetic dipole
moment; (j^mx , <\> , and 4>mz are the associated phases of the magnetic dipole
moment; PS1-, i = 1, 2, . . .,6 are the measured sum powers at the required
six different source orientations; Pdi , i = 1, 2, . . ., 6 are the measured
difference powers at the same six source orientations; q is the normalized
amplitude of the vertical electric field which would exist at the center of an
empty TEM cell when it is operated in a receiving mode and is excited by an
input power of one watt; and k is the wave number.
In obtaining (1) through (13), we have assumed that the TEM cell is made
of perfectly conducting material, symmetric, and impedance-matched at 50 Q so
that there exists no horizontal component of the electric field at the center
of a TEM cell. The frequency of the unknown interference source to be
detected by the spectrum analyzer included in the measurement instrumentation
as shown in figure 2 is assumed to allow only propagation of the dominant mode
inside a given TEM cell chosen for this study [2]. The size of the interfer-
ence source to be tested is also required to be small relative to the test
volume of the cell in order to minimize the potential perturbation to the
field distribution inside the cell [3].
Under the above conditions, the amplitudes of all the dipole moment com-
ponents, and the relative phases among the dipole moment components of the
same kind can be extracted based on the power (sum and difference) measure-
ments alone [2]. The relative phases among the dipole moment components of
the mixed kinds are still undetermined. This is the reason why measurements
of relative phases between the sum and difference signals are needed.
Before summarizing the equations for phase information, it is instructive
to examine the dependence of the power pattern, P(0,<t>), and the total power,
P-r, radiated by the source in free space on the source parameters. The
expressions for computing P(e,<{>) and Pj are [1, 2]:
P(9,<t>) = -£-£ [(m2x
+ k2 m2
x ) (cos 9 cos <j> + sin 2 <j>)
+ (m2 + k
2 m2 ) (cos e sin2* + cos *) + (m
2z
+ k2 m2
z ) sin2 e
- 2{mex
mey cosUex - <\>ey ) + k2mmxmmy cos(c|^
x- c|^
y)} sin
2G sin* cos*
- 2{meymez cos(<j,ey
- i\,
ez ) + k2mmymmz cosUmy - <|>mz )} sine cose sin*
- 2{mezmex cos(cpez
- <\>ex ) + k2mmzmmx cos((^
z- c}^
x )} sine cose cos<|>
+ 2k{mexmmy sin(cj,ex
- ipmy ) - meymmx s1n(cpey
- <^x )> cose
+ 2k{meymmz sinUey - <\^z ) - mez
n^y
sin(<pez- <\^y
)} sine cos*
+ 2k{mezmmx sin(cp
ez- cpmx ) - m
exmmz s1nUex - c|,mz )} sine sin*] . (15)
PT = J P(e,*)dQ
T^ (mex
+ <y + 4 + k * « mmx+"my
+ 4» (16)
where (0,<t>) are the spherical coordinates measured from the source center, and
\ is the wavelength.
Note that if one is interested only in determining the total power radi-
ated by the source, which may be the case in practice, all we need to extract
are m^- and m2^-, i = 1, 2, 3 from (1), (2), (3), (8), (9), and (10) based on
the measured powers. It is the radiation pattern for which the relative
phases are required. In particular, for the last three lines in (15), we also
need the relative phases among the mixed components, i|>ex- c^ , <j>e
- <j^x, <j>ey
- 4>mz , <\>
ez- 4> , (j,ez
- <j;mx , and cj;
ex-
<|>mz , which are related to some of the
quantities given in (1) through (14) and some of the measured phases <|>«, i =
1, 2, . . ., 6 between the sum and difference ports as follows:
*ex - V =*1
+ al
" h < 17 >
^ey " *mx=
*ex " % " eel " e
ml »(18)
*ey " V =*3
+ a3 " ?3 < 19 >
^ez " V =^ey " ^mz " e
e2 " em2 >(20)
*ez " ^mx=
^5+ a
5 " ^5 (21)
^ex " *mz=
^ez " *mx " ee3 " e
m3 '(22)
where
. m sine1
. mmv cosem1 - mmwpi
= tan (Jb^' ; (24)mx ml
. m sine9
°3 tan Cr t\ die ) (25)ey ez eZ
., mmw cosem9 - mm,
P3
"^\^ * •
and
1m sine ~
"5= ta" (iZ I mp> cose, J (27)
ez ex e3
. m cose , - mmvp 5 f-'i-Viwr) •
(28)mz m3
Alternatively, eqs (17), (19), and (21) may be replaced respectively by:
^ex " ^my=
^2+ a
2 ' p2 » * 29 )
*ey " ^mz=
^4+ a
4 " ^4 ' ^ 30 ^
and
*ez " *mx=
<>6+ a
6 " ^6 • (31)
where
. -m sine .
p = tan-1
!
mxm
™L W) , (33)
2 "V sineml
and
1-m sine
aA = tan_i
(- ^ -||—) , (34)4 ro_w - m «, cose ~ y
ey ez e2
. -m cose „ - m„ + , -If my mZ mz>, /01- x
P4= tan
( -mmv sine. ){35)
my m2
- -m sine ,as = tan (= ^ ^H—) . (36)6 ^m
ez- m
excose
e3^
- -m cose ,, - ma - -i-„r,-lr mz m3 mx> ,~,
x
h " tanC -m sine. )
' (37)
mz mi
It should be noted that, in principle, either (17) or (29) is required
for computing the mixed phase difference <j;ex- 4^ . They both should probably
be computed for the benefit of checking data consistency. The same note
applies for (19), (30), (21), and (31).
In addition, the possible sign ambiguity for eei
- and emi
- , i = 1, 2, 3
obtained from (4), (5), (6), (11), (12), and (13) may be resolved by the
obvious constraints,
3 3
I e . =, I e . = (38)
i=iei
i=imi
and the following special relationships:
sinem
. = M./N. , sinee
. = m./n. (39)
where
Ml
" 2(mmx - m
my ) mexmey
sin9el
+ (mex " m
ey} (m
mx " mmy } tan(
*l" *2 }
(40)
Nl
=" 2(m
ex " mey
} %mmy
+ 4mexmeyVm
mytan(
*l " V sin9el •
{41)
ml
= 2(mex " m
ey) mmx% sin9
ml+ {m
ex " mey
} (mmx " m
my ) tan(*l
" V(42)
nl
=" 2(V " % ] m
exmey
+ 4mex
meyVmmy
tan(*l " V sin9
ml •(43)
M2
= 2(mL " mmz ) m
eymez
sin9e2
+ (mey " m
ez} (m
my " mmz
} tan(*3 " *4 }
•
(44)
2 2N = - 2(m -m)mm +4mmmm tan(d> - a.) sine , (45)2 ey ez my mz ey ez my mz \3 M e2 '
m2 ' 2(n,
ey" m
ez) %V sine
m2+ (ln
ey " mez>
imly " mL' tan(
*3 " »4> '
(46)
n2
=" 2(m
my " 4' mey
ffl
ez+ 4m
eymezmn,yV tan(
*3 " V sin(W '(47)
M3
= 2(mL " mmx>
mezmex
sinee3
+ (mez " mL' (n4 ' m
mx>tan(
*5 " +6>
(48)
N3 - - 2(>4 - m^) Vm
mx VAlx tan( *5 " »6> sinee3 •
(49)
ffl
3= 2(m
ez - mex> Vm
mxs1n9
m3+ (l4 " m
ex>(mmz " "4' tan ( *5 " *6 ]
and
(50)
2 2n~ = - 2(ni - m) ni ni + 4 ni m m m tan (<j)c - d>c ) sine . (51)3 mz mx ez ex ez ex mz mx y
5y6 m3
In summary, eqs (17) through (28) constitute the complete set of expres-
sions for extracting the necessary relative phases required for computing the
detailed radiation pattern. Equations (29) through (37) serve as alternatives
for checking or averaging purposes. Equations (38) through (51) are necessary
for resolving the sign-ambiguity problem for eei
- and emi
-, i = 1, 2, and 3.
2. Derivation of Potential Errors
Examination of (1) reveals that when all the six sum-power readings are
not contaminated by measurement imperfections, the quantity, mex , so computed
contains no error. On the other hand, when the measurements are not perfect
due to the background noise, equipment imperfections, measurement environment,
or even human errors, the result of mexwill also be in error. To perform a
quantitative analysis of potential errors, we denote the measured values and
the parameters computed therefrom with the primes such as
Psi
= Psi
+ dP$i , i = 1, 2, . . ., 6 (52)
where PS1
- represents the true value, and dP$1
- is the small error contributed
by the measurement system.
For the general case, we also assume that the actual vertical electric
field, q, inside the TEM cell be changed to q' by a perturbation dq due to the
presence of source equipment placed inside the cell, and possible construction
defect and imperfect impedance matching of the cell,
q' = q + dq . (53)
We then have, in accordance with (1),
mexx " < P sl
+ Ps2 " P
s3 - Ps4
+ Ps5
+ ps6^ 2 <l'2
>
a m2x
+ (dPsl
+ dPs2
- dPs3
- dPs4
+ dPs5
+ dPs6
)/(2q'2
)
- {Psl
+ Ps2 " P
s3 " Ps4
+ Ps5
+ Ps6 ) W<? > (54)
or
dmex .
dPsl
+ dPs2 - dP
s3- dP
s4+ dP
s5+ dP
s6 do
^e7~=
2(Psl
+ Ps2 " P
s3 " Ps4
+ Ps5
+ Ps6>
~^'
m;x> , (55)
which gives the percentage error in the source parameter m'excomputed directly
from the measurement data of sum powers P^- involving imperfections dP$1-, i =
X) u j • • • j O •
In (54) and (55), we have neglected the second-order terms such as
(dmex ) , (dq)(dmex ), etc., for the simplifying purpose. This practice is
certainly justified if the first-order errors, dmex , dq, etc., are small.
When the situation warrants, the second-order terms can be included to improve
the error analysis.
Note from the first expression in (54) that m^x
may be negative even
though the true value m2x
is always nonnegative. This can occur only when
dps3
+ dps4
1S relatively large and mex
almost vanishes. Under this condi-
tion, we just assume that dmex/nig X= 1, giving m
ex= 0.
9 9 9 9 9The above analysis also applies for m|
y, m|
z, m^
x , m^ , and mmz
respectively in (2), (3), (8), (9), and (10). For mj^, i = 1, 2, 3, we may
reasonably assume dk = implying no perturbation in frequency. Therefore, we
also have
10
dm«w dP*-i
+ dP oO + dP .-3+ dP
./i " dP .c ~ dP .c Aey si s2 s3 s4 s5 s6 dq, cc \W 27P
-1—T~P—TT1—TT1—^"P 1—
- P' ) 7r >lbb 'm
ey^ lf
slKs2
Ks3
Ks4
Ks5
Ks6' q
dmrt , - dp. - dP e9 + dP - + dP . + dP . + dP c .
ez si s2 s3 s4 s5 s6 dq / c -,v
W 2(- P'—^~P"
1—+T1—+~P~1—+~P~I—+ P' \ cT »
lb/ 'mez
^ l Ksl
Ks2
Ks3
Ks4
Ks5
Ks6
] q
dmmx
dPdl
+ dPd2 - dP
d3 - dPd4
+ dPd5 + dP
d6 dq
™r~ =
2(P' + P* - P' - P' + P' + P' ) " q7"
mx dl d2 d3 d4 d5 d6 M
and
dmmz
;
dPdl
- dPd2
+ dPd3
+ dPd4
+ dPd5 * dP
d6 dq
^T" Z( - pdl- pd2
+Pd3
+Pd4
+Pd5
+Pd6
) "^
(58)
dm dP ., + dP , + dP .- + dP .. - dP ._ - dP ,cmy dl d2 d3 d4 d5 d6 dq, cr.,W~ =
2(p' + p' + P' + P' - P' - P' ) q* (59)my K
dlKd2 d3 d4 d5 d6 H
(60)
The predicted values for mei
- and m^- , i = x, y, z, then become
mei
= m^- - dmei
= m^(l - dm^/m^)
and (61)
"mi=
"4l " dmmi = "^i^ 1 " dm
mi/n
\!ii) '
The above implies that if we have known a priori (or have some basis) to esti-
mate dP$i , dP
dl-, and dq, the percentage errors (dm
ei-/nig.. and dm^/n^) can be
evaluated to predict approximately the true values. Or, at least, these per-
centage errors can be computed as a function of assumed values of dP$1
- , dPdl-,
and dq.
If, however, there is strong belief that dPsi , dP
dl- , and dq are all truly
random, the problem is much simpler. Then, the average values of them, and
thus the average values of dmei
and dmml
- in accordance with (55) through (60)
all vanish. That is,
11
Ws
. = Wd
. = -3q = , (62)
and
mei
=^\n1
=° * (63)
We conclude that
nf". = m . and W. = m . . (64)ei ei mi mi
This implies that, if repeated sets of power measurements are taken, we can
compute the corresponding sets for m^ and m^^ . The average values should
approach the true values when all errors involved in the measurement are
random in nature.
Let us now consider an expression involving the relative phase such as
(4). Using the same notation convention and neglecting the second-order
terms, we obtain
dm dm dP , - dP
where
cose .
ei Ci"'ex
mey
r
si" ' s2
cose;i
(psi
- p; 2
)/2^2
m;x
mv •(66)
which can readily be computed based on the power measurements alone, and the
remaining terms in (65) can also be computed if the values for dP$1
- and dq are
either known or assumed. Taking the inverse cosines of (65) and (66) will
yield eel
and 9^ with the appropriate sign determined by (38) and (39). The
difference, deel
= 9^ - 9el , represents the error incurred in the relative
phase deduced from the power measurements.
Similarly, we have
dm dm dP - dP . .
12
and
dni dm dP c - dP ,. .
cosee3
. cose
'
3 ( 1 ^ + ^* . ?5_ p
,
s6 M), (68)
ej ej mez ex
Ks5
Ks6 q
dmmv *Lw dp ^o ~ dp ^i o^«rn ,
fl = rn^fi' M + .
mx + .
my d2 dl 2dq>, ,
fi<ncose _ coseml [1 + -r-_ + -p-
p, _
p, + -^J , (69)
mx my d2 dl ^
dm dm dP .. - dP ,~ .
cose. ~= cose' (i + "0. + "L - <* ^3 +2dq
}(70)
m2 m2 % mmz
Pd4 "
Pd3 q
dm dm dP .c - dP ... ,
a i r 1 mz . mx do db Zdq, /tincose . = cose. (1 + 7^1— + 7^ p' - p '
_+ -£) {71)
m3 " """rn3 ^ m' m' PT,. - P' q'
mz mx do d5
where
cose;2
(p;3 - p;4
>/< 2«'2
m;y
m^ >(72)
cose;3
(p;5 - p;6
>/< 2«'2
*;zm;*> • (73)
cose;i = (p
d2 - pai>/
(2k^'2mm v •
and
74)
cose;2
(pd4 - Pds'/'*
2"'
2m»f •
(75)
cos9;3
(pd6 - ^b^ 2*
2"'
2m;2
m;x ' •
(76 '
Note from (66) that the situation Icose^l > 1 may arise if dPgl
- dP$2
is relatively large and eel
nearly vanishes. In this case, we just set cose^
a 1 (or e^ = 0). Then we have
13
dP 1- dP„ dm dm « . . ._
d9el
s "el
~=^C p?
1
. pf2
- gi^ - i^ - ^)1/2
• (77)el el p
slps2
mex
mey q
Similarly, we have
d9 - e- g f
"Ps3 - dP
s4dm
ey *ez 24, ,1/2 „.,d9e2
9e2 - ^ ( P;
3- P;
4 ^ ^ "^ '' '
rffl fl- ^7 f
dPs5 - dP
s6dm
ezdm
ex 2^ 1/2, 70 .
dee3 " 9
e3 - ^ t p ' - P' ' HT~ ' HT ?]'
(79 >
so so ez ex ^
dP AO - dP .. dni drn OA . /0dV - V - "hfvrf "i?f -ifif "^ •
,80)d2 dl mx my
dP ., - dP .-, dm dm «. . /0
ha -ft
- /T fd4
.
d3 W .mz 2dqa/2 , ftndGm2 " " 9
m2 " /2t P' - P' ' <T " ^7 " "^j ' (81)
d4 do my mz n
and
^PjG ~ ^^C ^m, dmmv O,4o 1 IOde. - . § . /? (
d6 d5- "« - ™* - m.) 1/z
. (82)m3 m3 P
d6 " Pd5
mmz
mmx q
Thus, eqs (65) through (76) provide the necessary means for estimating
deei
- and demi
- when e^- and e^. computed from the power measurements are not
near zero. Equations (77) through (82) serve the same purpose when eei
- =
and 9^. = 0, with the values for dP$1
- , dPdl
- , and dq known or assumed. On the
other hand, if dPsl-, dP^ and dq are random, we see that He .
a '^RLi = 0>
yielding IT. = e . and el. = 9 . .3ei ei mi mi
Corresponding error expressions for the relative mixed phases, (17)
through (22), can also be derived. Since they involve taking the derivatives
14
of inverse tangent functions, the final forms are quite complicated. For this
reason, they are not presented here. One alternative but simple method of
achieving these is to (i) compute the primed quantities based on the measure-
ment data, (ii) compute the unprimed quantities based on the predicted ampli-
tudes of the component dipole moments and the predicted relative phases among
the same kind, which have already been obtained, and then (iii) take the dif-
ference between the results derived above.
Consider (17) first. Initially, we have the computed value for ^ x- <j^y
derived from the measurements,
*ex " *my = *[+ \ ' Pj * < 83 >
where <j>' represents the actual measured relative phase between the sum and
difference signals for one of the measurement positions of the interference
source, and a1
and p' can be calculated respectively in accordance with (23)
and (24),
. m' sine'
"'. tan (< 'Vcoe') • < 84 >
1 ex ey el
. m' cose1
,- m'
,
o;"--'(-^ sis -
*) •
(85)
1 mx ml
Of course, all the primed quantities in (84) and (85) are based on the
measured data, thus, containing some measurement errors.
The error involved in <j,g X- <\>^y
may be expressed as
d((|;ex " V )=
d<J)l
+ dal " df3
l* (86)
where d<h represents the actual measurement error in the relative phase for
the first source position, do^ = a' - c^ , dp^ = p' -pj_
with a^ and p^ to be
15
computed also from (23) and (24) but with the unprimed quantities deduced in
(61), (65), and (69).
Note that the phase error d<j>^ may be estimated (depending on the type of
equipment involved in the measurement system) or assumed for computational
purpose. The predicted true value for <\>ex- <j> then becomes
^ex " V =^ex " ^my " d{(
^ex " W *{87)
Substituting (87) into (18), we also have <j>ey- (|>mx
.
Similarly, from (19) and (21), we obtain
*ey " C = *3+ a
3 " ^3 (88)
and
^ez " C =4>5
+ a5 - ?5 •
< 89 >
where <|>g and <$>§ are the actual measured phases between the sum and difference
signals for two other measurement positions of the source,
. m' sine'
"3= ^ (ST-tV^%) . (90)
ey ez e2
. m' cose' - m'
my m2
. m' sine'^
•sta"
~\ 7?» < cole; ) •
(92)
ez ex e3
and
16
. m' cose' - m'
h ' tant m - sine - )
• (93)mz m3
The errors involved in these phases are:
d(^ey " W = d(})
3+ da
3 " dh » ( 94 >
d(*ez ' W =d,|)
5+ da
5 " d^5 ' (95)
da3
= a3
- a3
, d£3
= P3- P3 , (96)
and
da5
= a5 " a
5 ' % =^5 " ^5 ' (97 ^
where d$3
and d<t>g represent the phase errors in the measurements, and a3 , p3 ,
ag, P5 may be computed from (25) through (28) with the predicted values for
mei
-, m^ , 9e2,
and em2 (unprimed) deduced from (61), (67), (68), (70), and
(71).
The predicted true values for the mixed phases being considered then
become
^ey " *mz=
^ey " ^mz " d((^ey " W (98)
and
*ez - ^mx = *ez " C " d(<^ez " ^mx ) (99)
The phases deduced above can then be used in (20) and (22) to obtain <j>ez- (j,
and ^ex " *mz*
17
Alternatively, eqs (87) (98), and (99) can also be derived by starting
from (29), (30), and (31) based on the measured phases $£, <^, and <j>
6. Either
the mixed phases obtained here or those obtained in (87), (98), and (99), or
the average values between them may be inserted in (15) to compute the pre-
dicted power pattern, which may be compared with that computed based on the
primed values to reveal the final error in power pattern.
Of course, we may also conclude that, when repeated measurements of P ' •
,
P^. , and <j>| are available and dP$i
, dPdi , and d^ are believed to be random,
dT1
- <\>' = <\> - (\> , 4J1
- <b' = cb -ib (100)vex my
Yex
vmy' Tey mtix
yey mtix ^ '
471
- <b' = d, - <b ,dT - d/ = ib - ib (101)v
eyYmz Yey Ymz '
Tez Tmy Yez Miiyu '
and
<y - V = <\> - <\> » <K - V = <b - <b . (102)ez mx ez mtix ^ex vmz Yex Ymz v
Under this condition, the average power pattern will approach the true power
pattern.
3. A Simulated Example with Assumed Unbiased Measurement Errors
For the purpose of checking the formulation and derivation, an ideal
simulated example was given in the previous report [1], where an unknown
arbitrary interference source was assumed to be characterized by the equiva-
lent electric and magnetic dipole moments with the following values:
mex= 1.4, m
ey=1.8, m
ez= 1.6, (b
ex= 0, ^ey = 80 > *ez = 60,
mmx= °' 8 >
"to= °* 6 ' %z = °- 4 > *mx " "80 > % = -60, *mz = -45 '
(103)
where the normalized amplitude of the electric dipole moment is in meters,
that of the magnetic dipole moment is in meters-squared, and the phase is in
degrees. The value used for the vertical electric field existing inside the
18
particular TEM cell considered therein was q = 11.83 V/m, and the frequency
was 30 MHz. With the source parameters assumed in (103), we computed in [1]
that the sum and difference powers in watts and the relative sum-to-difference
phases in degrees should be:
si
's4
425.107856
27.106038
P s2= 302.626424 ,
Ps5
= 473.027282 ,
S3
's6
784.597583 ,
159.541746 ;
'dl
*d4
2.704333 ,
27.172985 ,
'd2
*d5
52.545279
7.617338
'd3
*d6
1.556813 ,
36.582351 ; (104)
<t> 1= -103.0261
d)/i = 105.5593
<|>2 = -77.0300
<t)5= 48.116
4>3 = -113.5502
<j>6= 91.9132
In the previous report [1], we also demonstrated that if the values in (104)
were the actual measurement results, we would recover uniquely the source
parameters exactly as those given in (103). In this case, of course, there
would be no error involved.
Since this is an example aimed for analyzing the final errors in the
extracted source parameters for given errors in the simulated "measurement
data," we treat the problem here in a deterministic manner. The averaging
technique, by taking repeated measurements, is useful only when we are sure
that the measurement errors are indeed random for the equipment involved in
the measurement system.
Now, let us consider the problem from the practical point of view.
Suppose that the hardware in the measurement system is not capable of taking
perfect readings of the outputs accurate to the fourth or sixth digit after
the decimal point as demanded in (104). Instead, all the readings are only
accurate to the second digit after the decimal point. In addition, suppose
that the combined effect of background noise, equipment imperfections, and
human factors also degrades the readings somewjiat such that the actual
measured powers in watts and phases in degrees are as follows:
19
psl
p s2=
ps3
=
PU =
p s5"
p s6=
pdl
=
p d2=
Pd3=
Pd4
=
Pd5
=
pd6
=
i<t>2
=
*3=
*5
446.36
287.50
745.37
28.46
496.68
151.56
2.57
49.92
1.63
28.53
8.00
34.75
- 92.72
69.33
-124.91
95.00
43.30
(increasing
(decreasing
(decreasing
(increasing
(increasing
(decreasing
(decreasing
(decreasing
(increasing
(increasing
(increasing
(decreasing
(increasing
(increasing
(decreasing
(decreasing
(decreasing
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 5%)
approximately by 10%
approximately by 10%
approximately by 10%
approximately by 10%
approximately by 10%
and
(j>g = 101.10 (increasing approximately by 10%)
(105)
For simplifying the problem, we also assume that there is no perturbation
in the vertical electric field inside the TEM cell. That is, dq = 0, q' = q =
11.83 V/m. We then obtain
m
m
m
ex
2ey,2ez,2
2mym
2.1732 ,
3.0706 ,
2.4588 ,
0.5890 ,
0.3611 ,
0.1848 ,
mex= 1.474 ,
mey
= 1.752 ,
mez
= 1.568 ,
Kx = 0.767 ,
v - 0.601 ,
<z = 0.430 ;
20
dmex/nig
X= 0.050 (meaning an error of +5% in m^
xdeduced above),
dmey/
mey= -°- 026 (meaning an error of -2.6% in m' deduced above),
dmez/nig
Z= -0.019 (meaning an error of -1.9% in m^
zdeduced above),
dmmx/mlJ1x= -0.042 (meaning an error of -4.2% in m
fJlxdeduced above),
dmmy/m'= 0.003 (meaning an error of +0.3% in m^ deduced above),
dmmz/mIJ1z= 0.068 (meaning an error of +6.8% in m^ deduced above);
mex (predicted true value) = m^x (l
- dmex
/nigX )
= 1.400 (practically no error compared with the true value of 1.4),
mey= 1.798 (small error compared with the true value of 1.8),
mez= 1.598 (small error compared with the true value of 1.6), (106)
mmx= 0.799 (small error compared with the true value of 0.8),
n^ = 0.599 (small error compared with the true value of 0.6),
and
mmz= 0.401 (small error compared with the true value of 0.4).
Thus, the method presented does predict very good true values for the
amplitudes of the component dipole moments as long as the arbitrary errors in
power measurements are within 5% of the actual readings. Results with other
percent errors in power readings are presented in figure 3a, b, where the
manner with which the errors occur is assumed to remain the same as that in
(105), and the second-order correction terms are also excluded. From figure
3a, we see that, even with the power measurement error as large as 40% and
with the second-order correction terms neglected, the predicted value for mey
is still about 1.68 vs. 1.80 for the true value, a decrease of only 6.67%.
From (66), (72) through (76), and the constraints given in (38) and (39),
we obtain approximately,
cose^ = 0.219778 , 9^ = -77.30°,
cos9g2
= 0.932363 , 9g2= 21.19°
,
cose^3
= 0.533491 , 9^ = 57.76°,
(107)
21
cose^ = 0.929589 , 9^ = -21.63°,
cose^ = 0.941998 , 9^ = -19.61°,
cosg^ = 0.734008 , 9^3
= 42.78° .
Since none of the above values is near zero, we use (65) and (67) through (71)
to yield
cos9el
= 0.174289 ,9el
= -79.96° (increased by 0.05%) ,
cos9e2
= 0.940726 , 9e2= 19.83° (decreased by 0.85%) ,
cos9e3
= 0.499926 , 9e3
= 60.00° (completely recovered);
(108)
cos9ml
= 0.939814 , 9ml
= -19.98° (increased by 0.1%) ,
cos9m2
= 0.961780 , 9m2
= -15.89° (decreased by 5.93%) ,
cos9m3
= 0.811744 , 9m3
= 35.73° (increased by 2.09%) .
Thus, the final errors in eei
« and 9mi
- in degrees are
d9el
= 9^ - 6el
= 2.66 , d9ml
= 9^ - eml
= -1.65
dee2
= ee2 " e
e2 = l ' 36 dem2 = em2 " em2 = " 3 ' 72 ( 109 >
dee3 = e
e3 " ee3 = " 2 ' 24
*de
m3 = em3 " e
m3 = 7 ' 05'
We, next analyze the mixed phases. From (84), (85), and (83), we have
the following results in degrees,
ol[ = -42.59, $[ = 158.39 , <|^
x- <\^ = 66.30 . (110)
Then, from (23), (24), and (86), we have (all results in degrees):
ax
= -45.94, pj_
= 150.91
giving
dax
= a1
- aj_ = 3.35 , dpj_ = pj -pj_
= 7.48
d((J;ex " W = d<1)l
+ dal " dp
l= 6 * 18
'
22
which, in turn, yields
^ex " V =^ex " 4% " d((
^ex " V }
= 60.12 (increased by 0.2%) , (111)
and
^ey " ^mx=
^ex " % " eel " eml
= 160.06 (increased by 0.04%) .
Similarly, from (90, 91, 88), (92, 93, 89), (25, 26), and (27, 28), we obtain
the following in degrees:
«3
a5
a3
i _
and
Ctc "
10.00 , p3' = 145.98 , *ey " ^mz
= 99.11 ;
27.90 , P5 = -57.10 , ^ez " ^mx= 128.30 ;
9.33 , (3 3= 133.12 ;
27.82 , p 5= -63.68 . (112)
Then,
dao = a\ - «3
~ 0.67 dp3
= p3' - p 3
= 12.86
da5
= a£ - a5
= 0.08, dp5
= p 5- p5
= 6.58 ;
d ^ey " W = d4)3
+ da3 " % = "23 - 55
>
and
d(4>ez ^mx ]
= d<J>5
+ da5
- dp5
= -11.32
yield
*ey " *mz=
^ey " ^mz " d(c^ey " W
= 122.66 (decreased by 1.87%) ,
23
^ez " ^mx=
^ez " ^mx " d((^ez " W
= 139.62 (decreased by 0.27%) , (113)
^ez " ^my=
^ey " ^mz " ee2 " em2
= 118.72 (decreased by 1.07%) ,
and
^ex " ^mz = ^ez " ^mx " ee3 " e
m3
= 43.89 (decreased by 2.47%) .
The above results in (111) and (113) are based on the measured phases <j>'
,
<t>2, 4)5 and the assumed values for d<^, d^, and d^. Alternatively, the
results may also be derived from the measured 4>2 , <j>
4 , <|>g and the assumed
values for d<{>2
, d<j>4 , and d<j>g in accordance with (29) through (37):
a£ = 57.50 , P2'
=" 77,86
' ^ex " ^my= 66,03
a^J= -62.90 , P4
' = - 78.55 , <\,' - 4^z= 110.65
a6
= -57.91 , p£ = -105.10 , <^2- <|^x
= 148.29 ;
a2
= 58.46 , p2= " 78.57
a4= -61.47 , P4 = - 80.47
a6
= -53.47 , p6= -101.76 ;
da2
= a2' - a
2= -0.96 , dp
2= p
2- p2
= 0.71
da4
= a^ - a4
= -1.43 , dp4
= pj - p4= 1.92
da6
= a6 " a
6= "4 * 44
'dp
6=
^6 ' ^6= ~3 * 34 ;
d( ^ex " ^ =d<t>
2+ da
2 " dp2
= 6 * 03
d(<J>ey- c^
mz ) = d<|>4
+ da4
- dj34
= -13.91
d(*ez " V * d*6
+ da6 - <% = 8 ' 09
!
and
^ex " V =*ex " *my " d( ^ex " V }
= 60.00 (completely recovered)
24
^ey mz ^ey " ^mz - d(i>ey- ^mz )
= 124.56 (decreased by 0.35%) (114)
^ez mx=
^ez " ^mx " d((^ez " V
= 140.20 (increased by 0.14%)
The results obtained above by the alternative method may be used either to
compare with those obtained in (111) and (113) to check the consistency of the
measured data, or to be averaged with those in (111) and (113) to minimize the
overall errors in the deduced results, because, in reality, the actual
measurement accuracy is unknown. If the averaged values are desirable, we
have:
((^ex " %{ ^ey - ^mz
((^ez " V
( ^ey " ^mx
(+ez
av
av
av
= 60.06 (vs. the true value of 60.00) ,
= 123.61 (vs. the true value of 125.00) ,
= 139.91 (vs. the true value of 140.00) ,
= ((^ex " Vav " e
el " eml = 160 ' 00
<exact
> >
(115)
Vav =^ey " Vav " e
e2 " 9m2
= 119 ' 67
(vs. the true value of 120.00) ,
and
(^ " i,)^ = <*« " iv)ex 'mz'av wez " vmx'av " ne3 " °m3
(vs. the true value of 45.00) .
= 44.18
The above predicted results together with the predicted relative phases
obtained in (108) and the amplitudes obtained in (106) may be used in (15) to
compute the predicted radiation pattern, which may then be compared with that
computed by the primed values (based on the measured results directly) for any
9 and <|>. This comparison should reveal the error caused by the measurement
inaccuracy assumed in (105), and the contribution by the predicting process
discussed here. Typical graphical results are presented in figure 4, where
the original ideal case without involving measurement errors is also included.
25
The predicted amplitudes of the component dipole moments (given in (106))
alone may be used in (16) to predict the total power radiated in free space by
the interference source being investigated. That is,
PT= 0.4 it
2[m|
x+ m|
y+ m|z + k
2 (n4 + m^ + m*x ) ]
= 32.39 W (vs. the true value of 32.44 W)
.
Should the primed values based on direct measurements be used, we would have
Pj = 32.17 W (decreased by 0.83% compared to the true value).
Thus, from the anlaysis made so far for this particular simulated example, we
may conclude that the final predicted results are not substantially affected
by the measurement errors (e.g., + 5% in power measurements and + 10% in phase
measurements) arbitrarily assumed in (105). Selected results with other com-
binations of percentage errors in power and phase measurements, with the same
manner of error occurrence as assumed in (105), are presented in figure 5.
From these results, one may make a quantitative assessment as to how much per-
centage errors in measurement can be tolerated in order to achieve a certain
accuracy in the final predicted source characteristics.
4. Another Simulated Example with Assumed Biased Measurement Errors
Sometimes, we experience biased errors due to the measurement system. In
this case, the measurement readings are consistently higher than the actual
status. For simulating this situation, we assume that dPsi
= dPdl
- = 20 W and
d<j>,- = 20° for the example considered in the previous section. Thus, the
measured powers in watts and the measured phases in degrees become
P^ = 445.11 , P^2
= 322.63 , P^3
= 804.60 ,
P^4
= 47.11 , P^5
= 493.03 , P's6
= 179.54 ;
P^ = 22.70 , P^2
= 72.55 , P^ = 21.56 ,
P^4
= 47.17 , P^5
= 27.62 , Pjfi
= 56.58 ; (116)
26
${ = -83.03
d>; = 125.56
4>2 = -57.03 ,
<t>5 = 68.11 ,
<J>5= -93.55 ,
4>6= 111.91 .
Note that in the above, the errors in P^, P^, and P^ are quite enormous as
compared to the true values given in (104). Consequently, we should expect
large errors to appear in the final deduced source parameters related to these
difference powers. Here, we once again assume dq = for simplicity. Then,
from (54) through (61), we obtain
mex= 2.1029 nig
X= 1.450 , dmex/mex
= 0.034
mey= 3.3829 m^
y= 1.839 , dmey/nigy
= 0.021
mel= 2.7029 m^
z= 1.644 , dm
ez/mez
= 0.026
mmx= 1.0020 m^ = 1.001 , dmmx/mmx
= 0.181
<y = 0.7220 m^y = 0.850 , dmmy/mmy= 0.251
m'2mmz= 0.5220 m^
z= 0.722 , dm
mz/mmz
= 0.347
mex
(the pred icted \/alue) = mgX(l - dm
ex/m;
x )
= 1.401 (vs. 1 .40, +0.07%) ,
mey
= 1.800 !vs. 1 .80, no error) ,
mez= 1.601 (vs. 1 .60, +0.06%) ,
"\nx= 0.820 [vs. .80, +2.50%) ,
mmy= 0.637 (vs. .60, +6.17%) ,
(117)
and
mmz= 0.471 (vs. 0.40, +17.75%) ,
where, for simplicity, the plus sign in parenthesis denotes increasing while
the minus sign denotes decreasing.
From (66), (72) through (76), and the constraint requirements in (38) and
(39), we also obtain the relative phases in degrees among the same kind of
dipole components directly from the measured data,
27
cose^ = 0.164103 , 9^ = -80.55
cos9g2
= 0.895146 , 9g 2= 26.47
cos9g3
= 0.469845 , 9g3
= 61.98 ;
(118)
cose^ = 0.530216 , 9^ = -57.98
cos9^2
= 0.377654 , 9^2= -67.81
cos9^3
= 0.362634 , 9^3= 68.74 .
Equations (65) and (67) through (71) yield the predicted relative phases in
degrees:
cos9el
= 0.173129 , 9el
= -80.03 (vs. -80.0, -0.04%)
cos9e2
= 0.937218 , 9e2
= 20.41 (vs. 20.0, +2.05%)
9e3
= 60.13 (vs. 60.0, +0.22%) ;
(119)
9ml
= -40.60 (vs. -20.0, -103.00%)
9m2= -52.88 (vs. -15.0, -252.53%)
9m3= 56.35 (vs. 35.0, +61.00%) .
Note that the 9^-s obtained above still deviate substantially from the
constraint requirement in (38) while the 9^s so deduced are very
satisfactory. Note further that only the first-order correction terms are
retained in (69) to (71). This will yield a reasonable predicted result only
when the percentage errors involved are small. Since the actual percent
errors for dm^/m^, i = x, y, z are all relatively large in view of the
measurement errors assumed for P^, P^3
, and P^, it is not surprising to see
that the results for 9mi
- predicted in (119) are what they are. Should the
second-order correction terms also be included in the formulation, the
predicted results for 9ml
- would be more reasonable.
Comparing the results in (118) and (119), we have the errors in degrees
for the relative phases among the same kind of dipole components for the
example considered herein,
cos9e3
— 0.498036 »
COS9ml
= 0.759269 »
cos9m2= 0.603491 »
cosem3
= 0.554105 j
28
deel " e
el " eel = "°- 52
>d9
ml = eml " U ' " 17 ' 38
dee2
= e^ - ee2
= 6.06 , dem2
= e^ - em2= -14.93 (120)
dee3 = e
e3 " ee3 = l ' 85
>de
m3 = em3 " e
m3 = 12 ' 39
Now, we treat the mixed phases. In accordance with (83, 84, 85) and (17,
23, 24), we obtain the following in degrees:
aj = - 46.00 , aj_ = - 45.99 , dax
= - 0.01
p{ = -159.39 , Pj = -178.45 , dpj_ = 19.06
^ex " ^my=
*i+ a
l" P{ = 30 - 36
d((j,ex- 4>my ) = d<j>
1+ da
x- dp
x= 0.93
*ex " V =*ex " ^y " d( *ex " V = 29 * 43
'(121)
The above relative phase can also be deduced from <j>
2. That is,
cxo = 57.67 , <x2
= 58.43 , da2
= - 0.76
p2= - 58.42 , p 2
= - 67.04 , dp2
= 8.62
^ex " Ky =*2
+ a2 " ?2
= 59 ' 06
d(<pex- 4>my ) = d<(>2 + da
2- dp
2= 10.62
*ex " *my = *ex " ^y " d{ *ex " V = 48 * 44 (122)
which may be used to average with that in (121), yielding
(c|>ex - <l>my ) av= 38 - 94 < vs - 60 -°> -35.11%) , (123)
(4-ey" Wav =
{ ^ex ' Wav " eel " eml
= 159.57 (vs. 160.0, -0.27%) . (124)
Similarly, we also have the following in degrees:
ct3 = 12.48 , a3
= 9.60 , da3
= 2.88
p 3' = -153.00 , (3 3
= -170.33 , dp3
= 17.33
29
*ey " *mz=
*3+ a
3 " 6h = 71 ' 93
d( *ey " W = d<t>3 + da3
- dp3
= 5.55
^ey " ^mz=
^ey " *iz " d( ^ey " +nu)= 66.38 ;
a^ = - 63.38 , a^ = - 61.79 , da4
= - 1.59
pj = - 52.96 p4= - 59.30 , dp
4= 6.34
^ey " ^mz=
<t>4 + «4 - P4 = 115.14
d^ey " ^mz^= d<))
4+ da
4" = 12.07
^ey " *mz=^y " *mz " d( *ey " W = 103 ' 07 >
(<|>ey-
<l>mz ) av= 84.73 (vs. 125.0, -32.22%) ,
(125)
(126)
(127)
( *ez " V } av = ((^ey " Wav " ee2 " e
m2= 117.20 (vs. 120.0, -2.33%) ; (128)
ocr - 28.83 ,
p£ = - 47.69 ,
a5
= 27.86 ,
0r = - 54.95 ,
da5
= 0.97
dp5
= 7.26
*ez " *mx = *5+ a
5 " P5 = 144 ' 63
d(<|>ez- <\>m ) = d<t> 5
+ da5
- dp5
= 13.71
*ez " *mx = *ez " «ix " d^ez " W = 130 - 92>
(129)
ag = - 53.05 ,
pg = -118.05 ,
ag = - 53.37 ,
p6= -109.94 ,
dag = 0.32
dpfi
= - 8.11
*ez " Kx =*6
+ a6 " ?6 = 176 ' 91
d( ^ez " W = <% + da6 " <% = 28 ' 43
*ez - *mx=
*ez " C " d( *ez " W = 148 ' 48>
Uez- 4>mx ) av
= 139.70 (vs. 140.0, -0.21%),
(130)
(131)
( *ex " Wav = ((^ ' *my)ez ^mx' "e3 " m3
= 23.22 (vs. 45.0, -48.40%) . (132)
30
The predicted amplitudes of the dipole moment obtained in (117) may be
used in (16) to compute the predicted total power radiated by the source
emitter under study. That is,
PT
= 32.69 W (vs. the true value of 32.44 W, +0.76%) , (133)
which is still very reasonable in view of the large errors incurred in the
magnetic dipole moments. If the primed source parameters are used to compute
the total radiated power, we would get
Pf = 35.82 W (vs. 32.44 W, +10.42%) . (134)
The above result implies that the measured errors, dP • = dPd
- = 20 W and d<t>.
= 20° assumed in this example, will contribute more than 10% of error toward
the derived radiated power if no application of the predicting process as
demonstrated is made. Once the predicting process is used, the overall error
in the deduced total radiated power amounts only to less than 1%.
In addition, the predicted phases among the same kind of dipole
components obtained in (119) and the predicted average mixed phases obtained
in (123, 124, 127, 128, 131, 132), together with the predicted amplitudes just
considered can be used in (15) to evaluate the detailed normalized free-space
radiation pattern. This pattern may then be compared with that computed by
the primed source parameters to reveal the error caused by the measurement
inaccuracy and the correction contributed by the process so presented.
Thus, both the example considered in this section and that given in the
previous section show that the predicting process outlined in this report
(i.e., obtaining the necessary unprimed source parameters from the primed ones
computed directly from the experimental data containing measurement errors) is
indeed useful to extract the source parameters and, therefore, to predict the
radiation characteristics of an unknown inteference source, provided that the
approximate measurement error is known or can be estimated.
31
5. Sensitivity of Extracted Parameters with Respect to Measurement Accuracy
Another useful concept for error analysis is via the sensitivity func-
tion. Mathematically, the sensitivity of a parameter A with repect to a
quantity B may be defined as
CA dA/A „. B 5A , 1or ,
bB
=WB~ = A 9B"
(135)
giving the percentage change in A for a given percentage change in B, or
indicating the sensitivity of dependence of A on B.
Refering to (1), if we let A = mex , and B = PS1-, we can derive an
expression for the percentage error incurred in the deduced source parameter
mex due to a given percentage error in the measured quantity P$1
-
:
c
mex
Psi ^ex _ .
Psi r+ for i = 1,2,5,6 n , fi
,
Psi
=
% * Psi
" "4m
p2xq
2 L f0r 1 " 3 ' 4'
(136)
Similarly, we obtain
m D
c ey _ . si r+ for i = 1,2,3,4 MrnbP .
" * ~2 2 [ - for 1 = 5,6
[unS1 4m
eyq
m P<: ez - 4. si r+ for i = 3,4,5,6 M ,
ft»
si 4mezq2
m Pc mx _ di r+ for i = 1,2,5,6 mq \
P " ± —2—2~2 *- for i = 3 4di 4m
mxk q
32
S"W „ + /d1 (+ for 1 = 1,2,3,4 . .
m P
Smz = + dl {+ for 1 = 3,4,5,6 , .
PJ 2 2 2*- for i=1? 1 141
J
mz ^
P P . 5PT OOO Pc-i
SP
T =p^Lap-L = (20it
Z/\V) -p^- (142)
and
P P
Sp
T= (207i
2/\
2q2
) p^- . (143)Kdi
KT
The corresponding expressions for A = eei
-, 6ml
-, or <p ^- 4, • (i, j =
x,y,z, i * j) and B = P . , Pdl
- , or <|>j are much more involved, and thus not
presented here.
For the simulated example (103) and (104) considered in section 3, we
have
m mftY mOYSD
ex = 0.387 , S pex = 0.276 , S D
ex = -0.715 ,Ksl
Ks2 Ks3
mOY mOY mOYS p
ex = -0.025 , S pex = 0.431 , Sp
ex = 0.145 ;ps4
ps5
ps6
nuw mov/ mQX/S p
ey = 0.234 , Spey = 0.167 , Sp
ey = 0.433 ,Ksl
Ks2Ks3
mftw nipy mpvSp
ey = 0.015 , S pey = -0.261 , S p
ey= -0.088 ;p
s4ps5
Ks6
33
Sp" = -0.297 , Sp" = -0.211 , Spez = 0.547 ,rsl K s2 K s3
mP7 mQVS p
ez= 0.330 , Sp
ez = 0.111 ;
Sp™J= 0.372 , S
pmx = -0.011 ,
Sp™ = 0.054 , Spmx = 0.259 ;
Sp^ = 0.660 , Sp^ = 0.020 ,
SpV = -0.096 , S^ = -0.460 ;
S^J* = -1.486 , Sp1"12 = 0.044 ,
cmez .Sps4
_ 0.019»
cmmx _Spdl
" 0.019 5
P d4-0.192
S
pdl
0.034 »
yA4
0.342 J
Spdl"
-0.076>
Spd4
" 0.768 3
Ksl
0.185 9
Ps4
0.012 >
d2 rd3
!
mz = 0.215 , Spmz
d5 p d6
mm-7 mm ,Spmz = 0.215 , Sp
mz = 1.035 ,
(144)
Sp1 = 0.132 , Sp
1
= 0.341 ,Ks2
ps3
PT PTSp 1 = 0.206 , Sp
' = 0.069 ;r s5 r s6
>
T PT Pt>' = 0.001 , Sp
1
= 0.023 , S p' = 0.001 ,
dlpd2
pd3
*T PT PT,' = 0.012 , Sp 1 = 0.003 , Sp 1 = 0.016 .
d4 pd5 pd6
From the above numbers for this particular example, we may interpret the
results as follows:
1) The accuracy in measuring P s3has the most effect on m
ex. In partic-
ular, the deduced source parameter mex
decreases from the true value
by approximately 0.7% whenever the measured PS 3
is caused to increase
from its true value by 1% due to the measurement error. The accuracy
34
of the measured Ps3
has also relatively large effect on mey and
mez . This observaton is, of course, obvious in view of the fact that
P s3is the largest reading in (104). The implication is that more
care should be exercised for taking the stronger data.
2) The accuracy in measuring Pd2
and Pdg
has more effect on mmi , i =
x,y,z than that in measuring Pdl , P
d3 , Pd4 , and Pd5 .
3) The total radiated power, Py, by the interference source is not yery
sensitive to the accuracy in measuring the sum and difference powers,
because S < 0.4, a = P cn- and PH ,-.a SI ul
6. Concluding Remarks
Based on the previous formulation for quantifying the radiation charac-
teristics of an unknown interference source from power and phase measurements
with the aid of a TEM cell, we have derived in this report the deviation
expressions, due to possible errors in measuring these powers and phases, for
all those source parameters required for computing the radiation pattern and
the total power radiated by the subject source in free space. From these
deviation expressions, we can make corrections to those source parameters
deduced directly from the measured data in order to obtain results closer to
the true values, when the measurement errors are provided either by a
manufacturer's estimation or by assumption. If, on the other hand, the
measurement errors are believed to be random in nature, the derivations con-
tained herein indicate that repeated measurements are required, from which a
set of source parameters may be directly computed. The averages of these
respective computed source parameters should approach the true values.
Two numerical examples based on two sets of measurement errors—one is
arbitrarily assumed and the other is based on a systematic biased error—have
been given to illustrate the general procedure of making error analysis.
Utilization of the sensitivity function to yield a percentage change in one of
the deduced source parameters for a given percentage error in one of the
measurement data has also been applied.
35
In practice, the accuracy in power and phase measurements may be esti-
mated in terms of worst-case bounds such as + x% in power measurements and
+ y% in phase measurements (or in an absolute term such as + x watts and + y
degrees respectively). Under this situation, the formulation outlined in this
report is still applicable by considering all the possible combinations of
power and phase readings to derive an upper and lower bounds for the deduced
source parameters and, consequently, for the radiation characteristics of the
unknown source. When one of the source parameters so computed based on the
measurement data is dominant, more attention can then be paid to the bounds of
this particular parameter, because its accuracy should have more impact on the
accuracy in the final result. For example, if me2
is the largest among the
source parameters based on the direct sum power measurements P'-, i = 1, 2,
. . . ,6, then, in accordance with the first expression given in (54), we
easily obtain the upper and lower bounds for m2x
as follows:
m2J = [P!i(l + x%) + Plpd + x%) - Plo(l - x%)e
' upper -
1 *c b0o
- P^4(l - x£) + P;
5(l + x%) + P^
6(l + x%)] /(2q
z)
(145)
"4l lower= [
p sl^ " *%) + Ps2^ - ^ ' Ps3d + ^- P^
4(l + x%) + P;
5(l - x%) + P^
6(l - x%)] /(2q
2)
where dq = is also assumed.
7 7Based on the same sequence of assumed errors as above, nig and m^
zcan
then be computed. After applying a similar consideration for m^ , the upper
and lower bounds for the total radiated power P-j- may be estimated by (16).
Strictly speaking, the formulation outlined in this report applies only
to the symmetric TEM cell, where the center septum is exactly midway between
the top and bottom outer conductors. For an asymmetric cell designed to pro-
vide more test zone inside either top or bottom half of the cell [4], the
electric fields at the centers of the two halves are no longer identical.
When the vertical component of the electric field in either half, qa , is known
(by theoretical computation or by actual measurement) and the corresponding
36
horizontal component is relatively insignificant--as it should be for a well-
designed cell—the formulation and analysis presented herein can still be
applicable by merely replacing the parameter q in (1) through (13) by qa.
7. References
[1] Ma, M. T.; Koepke, G. H. A new method to quantify the radiation charac-
teristics of an unknown interference source. Nat. Bur. Stand. (U.S.)
Tech. Note 1059; 1982,
[2] Sreenivasiah, I.; Chang, D. C; Ma, M. T. Characteristics of electri-
cally small radiating sources by tests inside a transmission line cell.
Nat. Bur. Stand. (U.S.) Tech. Note 1017; 1980.
[3] Crawford, M. L. Generation of standard EM fields using TEM transmission
cells. IEEE Trans. EMC-16_: 189 - 195; 1974.
[4] Crawford, M. L.; Workman, J. L. Asymmetric versus symmetric TEM cells
for EMI measurements. Proc. IEEE International EMC Symposium, Atlanta:
204 - 210; 1978 June 20 - 22.
37
mi±ex/ ex
mt,ey/ ey %y/ my
Figure 1. Unknown equipment under test (EUT) is made of
equivalent three orthogonal electric and three orthogonal
magnetic dipoles.
38
TEM CELL
n>INSTRUMENTATION
FOR POWER AND
PHASE MEASUREMENT
l±> COMPUTER SYSTEM
Figure 2. Emissions testing measurement system
39
CD
_ ojm g3
CO
01 fi
h- •H
™ ^% u 5-1
OOJ 2- 5-1
U 5-1
q:Z) 4->
CO
o5-1
OJ 2- OJ
Pu
o: CO
Ld 3s CO
cu
>
Z CO
4-1
l-H fi0)
££ gcog
u
e
a)HOex
i- •H
z nd
(XJu
u
a•H5-1
4-1
oD_ CDH
a)
TJ00 a)
4-»
O•HT3CU
5-i
P-i
^3-
cO
mCU
U3M
G) fe
cu en OD rN- co in co oj
ui ' S1N3N0N 310dIQ 31^10313
40
e
^r
uqj
3co
a.ro
QJ
aCCD
S-i
OJ
OJ M-4
4-1m •H
LD Hh-
J-i
03 LdO
OJ H Uu QJ
a: +J
Z> cinCE
QJ
aJ-4^r
Ld ajOJ
21
en
LY =3 ON
Ld co 1—
1
XOJ P-
G>O >
OJ CO ,~s, j-i m
Z. C c(—1 0) i—
1
LY& C
COOLY QJ
•H
LY rH 13
LdOa.
QJ
•H 3h- TD en
Z OCO
cti
OJ LdU
•H4-1 CO
LYLd
QJ
Ctoo
cd
LY CT3 CO
e 4-1
-d a)
0) e03 u a)
a u•H 3TJ CO
QJ cO
S-i QJ
Pm e
sTCO
QJ
00
J I tspn
en tr 1^ CD LD ro OJ S
,m ' S1N3N0N mOdlLI DI13NDUW
41
_
X)-
•> CU—
cO •-s 4-> —+j MO O _co o a _T3 i-H CU
T3 ^ -
4-1 <r«
a a a) o "
0) •H CO H —
6 CO ^^ _§ T3 ou <U pi
3 d H •H™"
CO •H CO —
CO CO cu p —cu 4-J -o a) __e o
H >•H
-
cu a) 00CO CO ^ —
co M 4-1 CO—
"^ rC CU 4-1 _
<^\. ^ 4-1
CU
o4-1
CO -
^^St ^W Pi co iH (U^~
\ \ co U CO CO—
CO O CO _/ / U a •H rd _S/ cu
cu
4-1
cPL _
^sCs^ o a cu T3>^*r Cu }-i T3 P -
3 H cd —X) o __cu CO +J u _4-1 CO QJ
CO T3 o 5a CU i O —
•H 4-> iH &.g O <: ^^ _CO 'H s_^ CU iH4-1 T3 > cfl
_
c CU • u QJ"
o . 5-1 ^-N 3 T3O /-^ puLO a H -
LO rHcu o CU r-l tj a _d iH ,d s—
^
•H ,cu ^ 4J
T3 O4J
~
s a CJ Pi CO P _
o •H o CO
(Uc —
nd a T3 •-N rC TZcu cu CU co 4-1 a "1
CO > CO o c/"
CO •H CO ^H >. n -
Q M rd J=l ,£
-h c
- -C )
-
-h
ss
r
-
—
, \./K i , i < I.I.|
1| |
l . l . "
<s
GO
inincu
Q_
0)
r\i U(n U
LL(JLdnrrN i-
CD u*—
i
XH
<SOJ
s inoo in.
Q_
SOJ OJ to Q ^r ao OJ CO Q ^r 00
1 1»—
i
1 l 1
OJ1
OJ1
rol l
coi
ap UJL3^^T3d UOX^ETpFJ p9ZT-[12Ul.ION
42
/v u-
/ / CU-
If '""' 3 •"
II ^ CU 3 —II "^ CO CO —
P T ^ CH e _| cu x.
G CX CU11 »H • e~~
V \ ^ C /~^ CO —
\ \ ^ •H m CO —
vv \ ^l o _
W Tt ^ >-, t-i CU
V< \ ^ O ^^< 4=^^ \ Vj 4J <r
"
^^ \ £ >-, 3 c^^S. \w ^ CU •H 3 ^H —^^^^v j-j •H * _^^S^N. M-i B^S T3
^^*wS« m <U T3 c^g\. k^ CN e CU •H
™
^^S^ ^ 3 c -
^""^^w 4-J TD CO •H c —^S^^vw C-J c CO CO (U
^^V^s^ ^ cd to 4J « >\S*^Nnnw ^ X <r •HVNy'v *h CD CO O bOT ^\. ^^s. ""^ P fO cu
—/ \ \ c CO u cc
X ) J ^ CU u M 3 4J _y/ y y <u e 0) CU M CO
v^ ./ ./ u CU c 4-) •H •X3>^ ^r^/^ 3 u 3 0) M-i
v^ yv/ ^ 3 CO £ a -
^d* >o<J CD CO B CO C Cf-
v^ ^^s00'^ ""^ a u •H cc ->^ ^^^ cu cu CO ^3
v^ ^S^r if) e e a. CU CXjt sp10̂
j-i CO >V^ y^^ QJ J-, CO cu u X -
>T j&r j_J CU o 3 c3T /TO Q s CU u o a: —
/ /y^ s o ,3 3 _/ >^^ ^ ex 4-1 o oc >-
/ >y^ ^ CO c a/ ^y ^ c X 13 •H 3/ ^^ Q, »H 4-1
CU T3 c ~~'
/ i/ H 4J d ex
t nP ^ Vj s U o _/ Jy ^ o •H ex 1
—
f V ^ Vj 4-1 13 CO cc
D U 3 CU CU a11 ^ CU ,o U u T3 -
\ \ ™ 6-5 „a
o•r- -
Y, Vj (L) m CO cu u a\ \ ^ T-\ 4-1 ^ X\ \ "^ 3 4-1 CU 4-
~
\ \ x. CU J=.-
w. ^ 4-1 e C 4-1 c\.^v o •H CU o c\^W 3 !-i S-i
^V^h ""O 3 X) o t:
^V^^. QJ rt CO CU 14-1 a~
N^/^S. CO 4-> CO CO Cf-
>v^S. nj cO CU cfl CO cc —
^\ "
ID 6
1
CO ^ -
"m T f-
Jj + o1
—
. I A./\ ,1,1,11
A1
1 1 J_1111111
ru \J CO Q ^r1 1
1 1
ao
Q 0)
V aI-,
3in O
COinCXJ CU
llCJ
Q M CCU
CD X >-l
Q. CU
<4-l
uCU
4J
cCS •H
OJ CU^-1 ^2
4J
<4-(
o
r3 u CO
cfD u i-i
q: CU
cj4-1
4J
u CO
p cx
CE ol\l h- •HLU u 4J
-H I cO
1- 13CO
>-,
CU
C3OJ
uCO
(X^H CO
1
CU
<u
V-i
Q inT3CU
CD [N- Nll •H
H i—
1
I CO
a.O3
C3'cr m
a)
Vj
360•H
QCXI CD G> ^r CDCXI
1
CXI
1
cni I I
gp 'uae^^Bd uot^htpfj; pazTXHUiaoN
43
NBS-114A (rev. 2-ao
U.S. DEPT. OF COMM.
BIBLIOGRAPHIC DATASHEET (See instructions)
1. PUBLICATION ORREPORT NO.
NBS TN-1064
2. Performing Organ. Report No 3. Publication Date
June 1983
4. TITLE AND SUBTITLE
Uncertainties in Extracting Radiation Parameters for an Unknown Interference SourceBased on Power and Phase Measurements
5. AUTHOR(S)
Mark T. Ma and Galen H. Koepke
6. PERFORMING ORGANIZATION (If joint or other than NBS, see instructions)
NATIONAL BUREAU OF STANDARDSDEPARTMENT OF COMMERCEWASHINGTON, D.C. 20234
7. Contract/Grant No.
8. Type of Report & Period Covered
9. SPONSORING ORGANIZATION NAME AND COMPLETE ADDRESS (Street. City, State. ZIP)
10. SUPPLEMENTARY NOTES
3 Document describes a computer program; SF-185, FIPS Software Summary, is attached.
11. ABSTRACT (A 200-word or less (actual summary of most significant information. If document includes a significant
bibliography or literature survey, mention it here)
A method for determining the radiation characteristics of a leaking interferencesource has been reported in a previous publication [l], in which the unintentional
electrically small leakage source was modeled by two vectors representing a
combination of equivalent electric and magnetic dipole moments. An experimental
setup, measurement procedures, and the necessary theoretical basis were all described
therein to explain how the relevant source parameters can be extracted from the
measurement data of output powers and phases taken when the interference source is
placed inside a transverse electromagnetic (TEM) cell. Simulated examples were
also given to show that the equivalent source parameters of unknown vector dipole
moments and thus the detailed radiation pattern and the total power radiated by
the source in free space could be uniquely determined by the method if the measure-
ment data were not contaminated by noise. This report Dresents the mathematical
analvsis of the uncertainties in the final, extracted results when the experimental
data are degraded by the background noise and measurement inaccuracies.
12. KEY WORDS (Six to twelve entries; alphabetical order; capitalize only proper names; and separate key words by semicolons)
dipole moments; electrically small source; error analysis; interference sources;phase measurements; power measurements; radiation pattern; TEM cell; total radiatedpower; uncertainties.
13. AVAILABILITY
[23 Unlimited
| |
For Official Distribution. Do Not Release to NTIS
[X] Order From Superintendent of Documents, U.S. Government Printing Office, Washington, D.C.20402.
[] Order From National Technical Information Service (NTIS), Springfield, VA. 22161
14. NO. OFPRINTED PAGES
48
15. Price '
$8.50
•frU.S. GOVERNMENT PRINTING OFFICE: 1983-679 901/346USCOMM-DC 6043-P80
NBS TECHNICAL PUBLICATIONS
PERIODICALS
JOURNAL OF RESEARCH—The Journal of Research of the
National Bureau of Standards reports NBS research and develop-
ment in those disciplines of the physical and engineering sciences in
which the Bureau is active. These include physics, chemistry,
engineering, mathematics, and computer sciences. Papers cover a
broad range of subjects, with major emphasis on measurement
methodology and the basic technology underlying standardization.
Also included from time to time are survey articles on topics
closely related to the Bureau's technical and scientific programs.
As a special service to subscribers each issue contains complete
citations to all recent Bureau publications in both NBS and non-
NBS media. Issued six times a year. Annual subscription: domestic
$18; foreign $22.50. Single copy, $4.25 domestic; $5.35 foreign.
NONPERIODICALS
Monographs—Major contributions to the technical literature on
various subjects related to the Bureau's scientific and technical ac-
tivities.
Handbooks—Recommended codes of engineering and industrial
practice (including safety codes) developed in cooperation with in-
terested industries, professional organizations, and regulatory
bodies.
Special Publications—Include proceedings of conferences spon-
sored by NBS, NBS annual reports, and other special publications
appropriate to this grouping such as wall charts, pocket cards, andbibliographies.
Applied Mathematics Series— Mathematical tables, manuals, andstudies of special interest to physicists, engineers, chemists,
biologists, mathematicians, computer programmers, and others
engaged in scientific and technical work.
National Standard Reference Data Series— Provides quantitative
data on the physical and chemical properties of materials, com-piled from the world's literature and critically evaluated.
Developed under a worldwide program coordinated by NBS under
the authority of the National Standard Data Act (Public Law90-396).
NOTE: The principal publication outlet for the foregoing data is
the Journal of Physical and Chemical Reference Data (JPCRD)published quarterly for NBS by the American Chemical Society
(ACS) and the American Institute of Physics (AIP). Subscriptions,
reprints, and supplements available from ACS, 1 155 Sixteenth St.,
NW, Washington, DC 20056.
Building Science Series— Disseminates technical information
developed at the Bureau on building materials, components,
systems, and whole structures. The series presents research results,
test methods, and performance criteria related to the structural and
environmental functions and the durability and safety charac-
teristics of building elements and systems.
Technical Notes—Studies or reports which are complete in them-
selves but restrictive in their treatment of a subject. Analogous to
monographs but not so comprehensive in scope or definitive in
treatment of the subject area. Often serve as a vehicle for final
reports of work performed at NBS under the sponsorship of other
government agencies.
Voluntary Product Standards— Developed under procedures
published by the Department of Commerce in Part 10, Title 15, of
the Code of Federal Regulations. The standards establish
nationally recognized requirements for products, and provide all
concerned interests with a basis for common understanding of the
characteristics of the products. NBS administers this program as a
supplement to the activities of the private sector standardizing
organizations.
Consumer Information Series— Practical information, based on
N BS research and experience, covering areas of interest to the con-
sumer. Easily understandable language and illustrations provide
useful background knowledge for shopping in today's tech-
nological marketplace.
Order the above NBS publications from: Superintendent of Docu-
ments, Government Printing Office, Washington, DC 20402.
Order the following NBS publications—FIPS and NBSIR's—fromthe National Technical Information Services, Springfield, VA 22161
.
Federal Information Processing Standards Publications (FIPS
PUB)— Publications in this series collectively constitute the
Federal Information Processing Standards Register. The Register
serves as the official source of information in the Federal Govern-
ment regarding standards issued by NBS pursuant to the Federal
Property and Administrative Services Act of 1949 as amended.
Public Law 89-306 (79 Stat. 1127), and as implemented by Ex-
ecutive Order 1 1717 (38 FR 12315, dated May II, 1973) and Part 6
of Title 15 CFR (Code of Federal Regulations).
NBS Interagency Reports (NBSIR)—A special series of interim or
final reports on work performed by NBS for outside sponsors
(both government and non-government). In general, initial dis-
tribution is handled by the sponsor; public distribution is by the
National Technical Information Services, Springfield, VA 22161,
in paper copy or microfiche form.
U.S. Department of CommerceNational Bureau of Standards
Washington, D.C. 20234Official Business
Penalty for Private Use $300
POSTAGE AND FEES PAID
U.S. DEPARTMENT OF COMMERCECOM-215
THIRD CLASSBULK RATE