CHARACTERISTICS OF MULTI-GAP LOOP MAR smhhhhlohll … · afwl-tr-82-82 afwl-tr-*82-82 measured...
Transcript of CHARACTERISTICS OF MULTI-GAP LOOP MAR smhhhhlohll … · afwl-tr-82-82 afwl-tr-*82-82 measured...
R127 817 MEASURED CHARACTERISTICS OF MULTI-GAP LOOP AND 1/2ASYMPTOTIC CONICAL DIPOLE..(U) MICHIGAN UNIY ANN ARBORV V LIEPA ET AL. MAR 83 t786-i-F AFWL-TR-82-82
UNCLASSIFIED F2968i-78-C-6882 F/G 9/i NL
smhhhhlohllEhhmhhhhhhllIl/ll/llllllllEEIIIIIIIIIIIIEIIIhIIIIhIIIElllllllflflll.l
*L 6
. . .. . . . . . . . . . . . . . . . . . . .
m
I 2
ilIuilna
1.25m
MICROCOPY RESOLUTION TEST CHARTNATIONAL BUREAU OF STANDARDS-1963-A
_ _ __'
_ _ _ _1
. . . . . .
AFWL-TR-82-82 AFWL-TR-*82-82
MEASURED CHARACTERISTICS OF MULTI-GAP* LOOP AND ASYMPTOTIC CONICAL DIPOLE
ELECTROMAGNETIC FIELD SENSORS
iV. V. LiepaT. B. A. Senior
The University of MichiganINO Ann Arbor MI 48109
"* March 1983
'14 Final Report , )T. C5 (M A*Y 19830
' Approved for public release; distribution unlimited.
r4.°.
IJ AIR FORCE WEAPONS LABORATORYAir Force Systems CommandKirtland Air Force Base, NM 87117
,8 05 u6-009
AFWL-TR-82-82
This final report was prepared by the University of Michigan, Ann Arbor,Michigan, under Contract F29601-78-C-0082, Job Order 37630132 with the AirForce Weapons Laboratory, Kirtland Air Force Base, New Mexico. Mr. William D.Prather (NTAA) was the Laboratory Project Officer-in-Charge.
When Government drawings, specifications, or other data are used for anypurpose other than in connection with a definitely Government-related procure-ment, the United States Government incurs no responsibility or any obligationwhatsoever. The fact that the Government may have formulated or in any waysupplied the said drawings, specifications, or other data, is not to beregarded by implication, or otherwise in any manner construed, as licensingthe holder, or any other person or corporation; or as conveying any rightsor permission to manufacture, use, or sell any patented invention that mayin any way be related thereto.
This report has been authored by a contractor of the United StatesGovernment. Accordingly, the United States Government retains a nonexclusive,royalty-free license to publish or reproduce the material contained herein,or allow others to do so, for the United States Government purposes.
This report has been reviewed by the Public Affairs Office and isreleasable to the National Technical Information Service (NTIS). At NTIS,it will be available to the general public, including foreign nations.
If your address has changed, if you wish to be removed from our mailinglist, or if your organization no longer em~ioys the addressee, please notifyAFWL/NTAA, Kirtland AFB, NM 87117 to help us maintain a current mailing list.
This technical report has been reviewed and is approved for publication.
WILLIAM D. PRATHERProject Officer
FOR THE COMMANDER
DAVID W. GARRISON RO CASE, JR.Lt Colonel, USAF Lt Colonel, USAFChief, Applications Branch Chief, Aircraft & Missiles Division
* DO NOT RETURN COPIES OF THIS REPORT UNLESS CONTRACTUAL OBLIGATIONS OR NOTICEON A SPECIFIC DOCUMENT REQUIRES THAT IT BE RETURNED.
-- -- . .. .7
UNCLASSI FIEDSECURITY CLASSIFICATION OF THIS PAGE (Mhn Date Entered)
PAGE READ INSTRUCTIONSREPORT DOCUMENTATION PBEFORE COMPLETING FORM
. REPORT NUMBER 2.GOVT ACESSIUtiO. 3. RECIPIENT'S CATALOG NUMBER
AFWL-TR-82-82 __ _
4. TITLE (and Subtitle) 5. TYPE OF REPORT & PERIOD COVERED
MEASURED'CHARACTERISTICS OF MULTI-CAP LOOP AND Final ReportASYMPTOTIC CONICAL DIPOLE ELECTROMAGNETIC FIELDSENSORS S. PERFORMING ORG. REPORT NUMBER
017816-1-F, EMPTD-6-UM-0017. AUTHOR(@) S. CONTRACTOR GRANT NUMUER(s)
V. V. LiepaT. B. A. Senior F29601-78-C-0082
S. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASK
AREA & WORK UNIT NUMBERS
The University of MichiganAnn Arbor, MI 48109 64711 F/37630132
II. CONTROLLING OFFICE NAME AND ADDRESS 12. REPORT DATEMarch 1983
Air Force Weapons Laboratory (NTAA) 13. NUMBER OF PAGESKirtland Air Force Base, NM 87117 158
14. MONITORING AGENCY NAME & ADDRESS(if different from Controlling Olfie) IS. SECURITY CLASS. (of this report)
UnclassifiedISs. DECL ASSI FIC ATION/ DOWN GRADING
SCHEDULE
16. DISTRIBUTION STATEMENT (of thls Report)
Approved for public release; distribution unlimited.
17. DISTRIBUTION STATEMENT (of the abstract entered In Block 20, It difierent from Report)
II. SUPPLEMENTARY NOTES
IS. KEY WORDS (Continua on reverse side If necessary and Identify by block number)
Asymptotic Conical Dipole (ACD)Multi-Gap Loop (MGL)Electromagnetic FieldCal IbrationSensors
_ 0. ABSTRACT (Continue on reverse aid* it necessar aind identify by block number)
The frequency responses of the MGL-2D(a), MGL-6A(A) and ACD-4A(R) free-spacesensors have been measured and analyzed. The measurements were made over thefrequency range 118 to 4400 MHz whose upper limit far exceeds the 3-dB roll-offfrequencies for the sensors. All three sensors were evaluated in terms of thelinearity of the frequency response, the angular behavior vis-a-vis a dipolepattern, and the rotational symmetry (MGL sensors only). Whereas the linearitypersists only slightly beyond the roll-off frequencies, the dipole and symmetry
(Ovpr)
DDIj 1473 EDITION OF I NOV 65IS OBSOLETEDSR JAN IY UNCLASSIFIEDSECURITY CL.ASSIFICATION OF THIS PACE (*len Datem Entered)
UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAOE(Whan Daea Entered)
20. ABSTRACT (Continued)
properties extend to frequencies two or three times greater.
U In addition, some preliminary data for the MGL-58(R) and MGL-7A(R) groundplane sensors are presented.
UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGIEr(b~f Date Entered)
L'4. ACKNOWLEDGEMENT
The authors are indebted to H. Yoon, of the University of Michigan, for
making the measurements, sometimes two or three times over; to C. Bickley
and T. M. Willis, III, of the University of Michigan, for developing the data
processing techniques and writing the programs; and to W. Rasey for typing
the manuscript. We are especially grateful to Dr. C. Baum of the Air Force
Weapons Laboratory for his suggestions in the area of data analysis and
presentation.
Aecoss21of ForIITIS QrA&I1
Dtic U" 1
D1istibuition
i.
.-... . - - - - - - - - - - - - - -
i. :Dist~u specia
.. .
TABLE OF CONTENTS
Section Page
1. INTRODUCTION 1
2. GENERAL FORMULAS 3
2.1 Theoretical Considerations 32.2 Practical Considerations 8
3. EXPERIMENTAL FACILITY 12
4. GROUND PLANE SENSOR EVALUATION 16
4.1 Ground Plane 164.2 Measurements and Calibrations 174.3 Sensor Responses 274.4 Discussion 35
5. FREE-SPACE SENSOR EVALUATION 46
5.1 The Sensors 465.2 Experimental Techniques 495.3 Angular Response Measurements 54
5.3.1 ACD-4A(R) Sensor 545.3.2 MGL-2D(A) and MGL-6D(A) 56
5.4 Angular Response Data Analysis 595.5 Frequency Response Data Analysis 81
6. SUMMARY 87
REFERENCES 90
APPENDIX A: ACD-4A(R) DIPOLE RESPONSE DATA (RAW) 91
APPENDIX B: MGL-2D(A) AND MGL-6A(A) DIPOLE RESPONSE DATA (RAW) 107
APPENDIX C: MGL-2D(A) AND MGL-6A(A) ROTATIONAL RESPONSEDATA (RAW) 136
" ill
- " : . - . . . .- .- _ , . -.. . , -, ,-. --. . .. . . . .. ._ _ __ _ _ _ _. _.. ..__ .. ..
ILLUSTRATIONS
Figure Page
I Coordinate systems used for theoretical analysis. 6
2 Block diagram of the facility. 13
3 Resistively loaded ground plane used in themeasurements. 18
4 Ground plane sensors used in the evaluation. 20
5 Calibration of the 20-dB attenuator, deduced frommeasurements numbers 5 and 7. 24
6 Calibration of the 6-dB attenuator, deduced frommeasurements numbers 6 and 7. 25
7 Calibration for the 0.020-inch coax, deduced frommeasurements numbers 9 and 10. 26
8 Computed calibration for the 0.020-inch coax. 28
9 Calibration for the 0.030-inch coax, deduced frommeasurements numbers 7 and 8. 29
10 Computed calibration for the 0.030-inch coax. 30
11 Response of the MGL-7 sensor referenced to that ofthe MGL-8. 31
12 Response of the MGL-7 sensor with MGL-8 used asa standard. 33
13 Response of the MGL-7 sensor referenced to that ofthe MGL-8. 34
14 Response of the MGL-5 sensor with the MGL-8 usedas a standard. 36
15 Amplitude of the MGL-7 response compared with theSharpe-Roussi simulation. The dots show thefrequencies corresponding to the dominant polesin the 10-pole simulation. 38
16 Phase of the MGL-7 response compared with theSharpe-Roussi simulation. The dots show thefrequencies corresponding to the dominant polesin the 10-pole simulation. 39
iv
Figure Page
17 Amplitude of the MGL-5 response compared with theSharpe-Roussi simulation. The dots show thefrequencies corresponding to the 18 poles used inthe simulation. 42
18 Phase of the MGL-5 response compared with theSharpe-Roussi simulation. The dots show thefrequencies corresponding to the 18 poles usedin the simulation. 43
19 Amplitude of a portion of the MGL-5 compared withthe Sharpe-Roussi simulation using 14 poles. 44
20 Phase of a portion of the MGL-5 compared with theShapre-Roussi simulation using 14 poles. 45
21 Sensors used in the study. 48
22 Modified MGL-9(R). (The coaxial cables were bent4.75 inches from the loop to adapt for H-verticalpolarization. In the original design the coaxwas straight.) 50
a 23 Twinaxial to dual-coaxial adapters. 53
24 Measurement geometry for the ACD sensors. (The•-variation produces the angular (dipole) response.) 55
25 Measurement geometry for MGL-series sensors. (Thea-variation produces angular (dipole) response and*-variation produces azimuthal (constant) response.) 57
26 Styrofoam jig used for support and rotation ofMGL-2 and MGL-6 sensors. (The incident field isfrom the right-hand side.) 58
27 Measured angular response of MGL-2D(A) at 150 MHz. 60
28 A (f) for ACD-4A(R). 62n
29 E2(f) for ACD-4A(R). 63
30 im (f) for ACD-4A(R), f61 < 600. 63
i 31 M(f) for ACD-4A(R), lel < 300. 64max
32 mx(f) for ACD-4A(R), e = 00. 64max
. . vi.
%V
1
Figure Page
33 An(f) for MGL-2D(A), gap No. 1 at € = 90. 65n34 C2 (f) for MGL-2D(A), gap No. 1 at € = 900. 66
35 -M(f) for MGL-2D(A), lel < 600, gap No. 1 atmax 66=900.
36 6m(f) for MGL-2D(A), gap No. I at 0 = 900, lel < 300. 67max
37 6m(f) for MGL-2D(A), gap No. 1 at 0 = 900,max 67e = 00.
38 A (f) for MGL-2D(A), gap No. 1 at 0 = 45°. 68n39 E 2 (f) for MGL-2D(A), gap No. 1 at 450. 69
40 max (f) for MGL-2D(A), gap No. 1 at 450, lel <600. 69
41 6e (f), for MGL-2D(A), gap No. 1 at 450, !e < 300. 70max
42 6' (f), for MGL-2D(A), gap No. 1 at 450, e =00. 70
43 B n(f) for MGL-20(A). 71
44 '2(f) for MGL-2D(A). 72
45 r (f) for MGL-2D(A). 72max
46 A n(f) for MGL-6A(A), gap No. 1 at 0 = 900. 73
47 2(f) for MGL-6(A), gap No. 1 at o = 900, 7448 6m(f) for MGL-6(A), gap No. 1 at € = 900,
emax 74lel <600.
49 dma(f) for MGL-6A(A), gap No. 1 at 900, lel <_ 30 75max
50 (fmaxf) for MGL-6A(A), gap No. 1 at 900, e = 00.
51 A (f) for MGL-6A(A), gap No. 1 at = 450. 76n52 0 2(f) for MGL-6A(A), gap No. I at = 450. 77
53 ax(f) for MGL-6A(A), gap No. 1 at 0 = 45,
lel <60. 77
vi
• .I• " . . , . . o . . • . . . , . . : . .
Figure Page
54 (f) for MGL-6A(A), gaps at = 450, jej 300. 78max55 ax(f) for MGL-6A(A), gaps at = 450, =00. 78
ma
56 B n(f) for MGL-6A(A). 79
57 e2(f) for MGL-6A(A). 80
58 6 r (f) for MGL-6A(A). 80I rmax
59 Response ratio of ACD-2/MGL-9. 83
60 Response ratio of ACD-4/ACD-2. 84
61 Response ratio of MGL-2/MGL-9. 85
62 Response ratio of MGL-6/MGL-9. 86
41
i
i.i
TABLES
Table Pg
1 Components Studied 21
2 Measurement Combinations 22
3 Poles and Residues for MGL-7 40
4 Identification of Free-Space Sensors Used 47
5 Summnary of Results 88
Al List of ACD-4 Dipole Data Plots 91
BI List of MGL-2 Dipole Data Plots 107
Cl List of MGL-2 and MGL-6 Rotational Data Plots 136
viii
1. INTRODUCTION
An electromagnetic field sensor is a special purpose receiving
antenna which converts a given electromagnetic field quantity into a
voltage or current according to a prescribed relation. The relation
should be as simple as possible, e.g., a constant of proportionality,
or a linear form involving frequency or time.
As is true with most instruments, however, there is a maximum
frequency (frequency domain) or minimum rise time (time domain) beyond
which the simple relation no longer holds. In general, the smaller
the sensor, the larger the frequency range over which the performance
can be prescribed. Since the output of a time derivative sensor is
inversely proportional to its size and directly proportional to the
frequency, the output of a particular sensor may not be large enough
for accurate measurements at low frequencies. In this case one could
choose to use a larger sensor to obtain a bigger output at the expense
of reducing the upper frequency limit for the prescribed behavior. If
used at frequencies beyond this limit (the roll-off frequency),
erroneous data will result; and though two sensors of different sizes
might seem to constitute a solution, their use would complicate the
data acquisition procedures. In addition, there may not be room on
the test object to mount more than one sensor.
For these and other reasons, it is desirable to obtain as much
information as possible about the behavior of the sensors
,4
.- ,
that are used. The purpose of the present study was to investigate
and analyze the responses of various EMP sensors over a frequency
range extending well beyond the so-called 3-dB roll-off frequencies.
The sensors which were studied were the MGL-2D(A), MGL-6A(A) and
ACD-4A(R) free-space sensors, and the MGL-5B(R) and MGL-7A(R)
ground plane versions. In each instance the responses were measured
as a function of frequency and, in addition, for the free-space
sensors, the dipole patterns and rotational symmetry were determined.
The mathematical formulas which define the responses and which
were used in analyzing the data are presented in Chapter 2, and this
.- is followed (Chapter 3) by a description of the experimental facility
* . used to make the measurements. Chapter 4 is concerned with the
ground plane sensors and includes a detailed presentation of the
data and the results of the analyses. The majority of the effort
was devoted to the free-space sensors, which are treated in Chapter 5.
The significant results are summarized in Chapter 6 in a manner such
that the performance of the various sensors can be compared.
-2-
2. GENERAL FORMULAS
"g Since the MGL and ACD electromagnetic field sensors were developed
to measure electromagnetic fields in the time domain, it is natural to
evaluate them in the time domain by measuring their impulse or step
responses. In general, however, such tests are not accurate enough to
quantify their detailed performance in the frequency domain. This is
particularly true at the higher frequencies where the response cannot
be deduced from the time domain behavior because of instrument and/or
recording limitations, and in the study reported here the frequency
response of the MGL and ACD sensors was measured directly in the
frequency domain,
2.1 Theoretical Considerations
The sensors that were tested were designed [1] to provide a simple
mathematical relation between a designated field quantity and the output
voltage over a wide range of frequencies. For the ACD sensors the
voltage response is [2]
V0(t) =R0 A - E(t) (time domain)
(1)or
Vo(f) = j27fco A E(f) (frequency domain)
-3-
where V0 is the sensor output (volts),
::'-iR is the sensor's characteristic load impedance
(usually 100 ohms),
A is the sensor's equivalent area (inm2 ),eq
(367r) " 10"9 (Farad/m),
E is the electric field intensity (volt/m),
f is the frequency (Hz),
and a time dependence exp(j27Tft) has been assumed. For the MGL
sensors,
V eq dH(t) (time domain)
or (2)
V 0 (f) = j2rfuoAeq H(f) (frequency domain)
where io = 47 x l0 (Henry/m) and
H is the magnetic field intensity (Ampere/m).
Each sensor is a calibrated device, and its equivalent area is
specified.
The above formulas can be obtained by retaining only the dipole
terms in the low frequency expansions of the responses, and are valid
for sufficiently low frequencies. As the frequency is increased, the
influence of higher order multipoles becomes significant, leading to
" -i a response which is no longer linear in frequency and does not have
the aspect dependence characteristic of a dipole, e.g., the dot
product (cosine) behavior as a function of angle.
-4-
The coordinate systems used in analyzing the measured data
for the responses of the ACD (electric dipole) and MGL (magnetic
dipole) free space sensors are shown in Fig. 1. In each case the z
axis is fixed by the sensor, and the relevant incident field vector
(E for the ACD sensor and H for the MGL sensor) is chosen to be in
the 0' direction of a spherical polar coordinate system (r,e',f). The
frequency domain responses (1) and (2) then become
V(f) = j27f Re A E(f) sin e' (ACD) (3)0 o eq
Vo(f) = j2rfmo AeqH(f) sin e' (MGL) (4)
and both can be written as
V(e',O;f) = A(f) F sin e' (5)
where A(f) is a function of the frequency alone and F denotes the
appropriate incident field quantity (E for the ACD sensor and H for
the MGL). Outside the frequency range where (1) and (2) are
applicable,
V(e',;f) = {A(f) sin e' + c(e',4;f)} F (6)
where e is an error term embodying, amongst other things, the effect
of the higher order multipoles.
-5-
° -
. - -A .01
z 0..
(b Manei loo eel'L
Fiue . Coriat yse se orteoeialaalss
-6
To determine A and e from V, the standard procedure is to
expand V(e',O;f) in zonal harmonics [3], so that
V(e',O;f) = A(f)Pl(cos ') + {terms in e',o,f}1
where V is the recorded signal normalized relative to the incident
field strength, i.e., V = V/F. We can now find A(f) by using the
orthogonality properties of the zonal harmonics. If V is assumed
independent of 0, multiplication by Pl(cos e') and integration over
4Tr steradians (do = sin e' do' do) gives
A(f) 3 V(e',O;f) sin 2 e' do' (7)
0
We note in passing that if the entire right-hand side of (6) is assumed
to be independent of 0 rather than just the first term, an alternative
approach is to expand V in a Fourier sine series. This leads to an
alternative weighting factor in (7) which gives more weight to the
angles at which the sensor is end-on, but is less defensible as a
general procedure.
From (7) the mean squared angular error is defined as
=fIv(e',O;f) - A(f)sin 0'12 do
f IA(f) sin 6'12 dQ
-7-
-I
i.e.,
'-. [A(f) 2 V'f
s2: 2(f) :) I fV(e',o;f) - A(f)sin e'12 sin 6' de'4..,
0 "(8)
and is a measure of the mean squared deviation of the angular response
data from the dipole behavior. The maximum error ax (f) is-. "max
likewise defined as
"maxV(e',f) A(f)sin o'""- . .'ma(f) maxi" = max
IA(f)sin e'
.max < e 'eax (9)
and indicates the maximum deviation from the dipole pattern over a
*m specified angular range.
2.2 Practical Considerations
In antenna work the angle defining the angular response or
pattern is generally measured from the direction of the peak response,
and this is the convention used in the sensor catalog [2] where,
-for example, data sheet 1118 cites the response of an ACD sensor
as
V - RA d or V - RA d Dcos e. (10)eq dt eq dt
Here, e is the angle between the direction of the vector equivalent
area Aeq and that of the appropriate field vector, so that
-8-
e - /2(11)
where e' is the spherical polar angle defined in the previous
section. For consistency with the literature, we shall henceforth
express all results in terms of e, where -r/2 < 8 < 7r/2.
Consider also the nature of the signal v(e,0;f) present at
the output of a sensor. In practice,
v(e,O;f) = V(e,o;f) F K (12)
where V is the sensor response to a unit incident field, F is an
incident field of unknown strength, and K is the response of the
chamber, cables, amplifiers, etc. In our measurements it is
postulated that F and K are time invariant, and thus, for example,
the ratio of measurements made for some value of e and for 8 = 0
(broadside) at two different times yields
Vn(O,¢;f) = V(o,;f) (13)n V(0,0;f)
where V is independent of F and K. V n(e,;f) is the response".- n
relative to that of the same sensor at broadside incidence, and
(7) through (9) are readily expressed in terms of this normalized
quantity:
-9.
*.. .. . .. - ° - : : -/ " I . . . . .
A3 Vn(O;f)cos2 e de (14)
7r/2
4(f) -f ( ;f) A(f)COS
-7/ (15)
IV (e,;f) - A (f)cos 91emax(f) = max , o < ( e m (16)
m IAn(f) cos el
In the derivation of (7) through (16) it was assumed that
the sensor response is independent of (see Fig. 1), and 0 has been
shown as one of the variables of Vn only for generality. For the
ACD sensors the response is € independent from symmetry, but this
is not true for an MGL sensor. The free-space version evaluated
in this study has four gaps which, at low frequencies, should be
"invisible." As shown later, however, at high frequencies the
response is affected by the orientation of the field with respect to
the gaps.
To assess the rotational response of the MGL sensors, an
average rotational response coefficient is defined as
n(f) = F Vn(O,0;f)d (17)
0
(cf. (14)). By analogy with (15), the mean squared rotational error
is
o"
-.. ~ - 10- . " . . ' . -
27r= B (f-2fJ(O;)-B (f)12 do 182w(f n~f I1 n fl~) (8
0
and the maximum rotational error is
IV n(0,0;f) B B(f)j6' (f) =max (19)
r maxIBn(f)I
The formulas in (14) through (19) were progranmmed in BASIC
for the HP9845 calculator, and results are presented in Chapter 5.
3. EXPERIMENTAL FACILITY
The measurements were made in the University of Michigan's
surface (and near) field facility, a block diagram of which is
shown in Fig. 2. The system is a CW one in which the frequency is
swept (stepped) over a wide range, and a key part of the facility is
a tapered.anechoic chamber approximately 50 feet in length. The
rectangular test region is 18 feet wide and 12 feet high. The
rear wall is covered with 72-inch-high performance pyramidal absorber,
with 18-inch material used on the side walls, floor and ceiling.
The material in the tapered section (or throat) is 2-inch hairflex
absorber. The chamber can be thought of as a lossy wall horn antenna
terminated by the rear wall. The signal is launched from a single
exponentially tapered broadband antenna located at the apex of the
chamber. The antenna is fixed and since the radiated signal is
horizontally polarized, the pseudo plane wave in the center ('quiet
zone') portion of the test region is also horizontally polarized.
The instrumentation is centered around a Hewlett-Packard
8410B network analyzer, and is computer controlled. An HP9830A
calculator controls the frequency to be generated, switches in the
appropriate power amplifiers and low-pass filters, and reads and
stores the amplitude and phase of the signal picked up by the
sensor. During a run, the frequencies are typically stepped from
118 to 4400 MHz. Because of the limited memory size of the
calculator, this frequency range is recorded in four bands: 118 to
550 MHz (in 4.8 MHz steps), 550 to 1100 MHz (in 4.8 MHz steps),
-12-
CC,--
Al C. l
III
-0--
Cl 0L
Cc r- -
~~.0~~C -.----- -
CC m
i I
1100 to 2000 MHz (in 9.6 MHz steps), and 2000 to 4400 MHz (in 16 MHz
steps). The data from each band are stored by the HP9830A calculator
- on a cassette for later transfer to an HP9845B calculator which
processes and plots the data. If substantial processing or
computation is involved, or if a need exists to write the data on
standard computer tape, the data are transmitted to the central
University of Michigan AMDAHL/V8 (IBM compatible) computer.
The signal picked up by the sensor is a function of the
response not only of the sensor but also of the entire facility,
including chamber, antenna, amplifiers and cabling, and it would be
a virtually impossible task to separate out the contributions of each.
An alternative approach is to apply some form of calibration or
normalization whereby the facility response is, in principle,
eliminated. When measuring the surface fields on, say, a scale model
aircraft, two distinct measurements are made using the same probe
or sensor: one on the aircraft, and the other on a metallic sphere
or other object whose surface field is known. In conjunction with the
known sphere solution, the ratio of the two measurements gives the
aircraft response relative to the incident field.
In the present study, two different calibration procedures were
used. To determine the angular response characteristics of a sensor, the
data were normalized to the measured values of the maximum response
.0 (broadside incidence) for the same sensor. This eliminated the effects
of any cable mismatch. To obtain the frequency response, each set of
measurements was followed immediately by the analogous measurements
-14-
for a small sensor such as the ACD-2(R) or MGL-9(R), whose response
is believed uniform and predictable out to frequencies beyond
those of interest for the original sensor. The procedure worked wellfor the ground plane sensors (Chapter 4), but the results were less
satisfactory for the free-space sensors (Chapter 5). One reason
for this was the presence of currents induced on the sensor leads
by "stray" fields in the chamber that ultimately affect (especially
the ACD-type) sensor response, but a more severe problem was caused
by the mismatches in the twin-axial leads, particularly at the twin-
coax to twin-axial junctions.
- The details of the measurement procedures for each type of
sensor are described in the section devoted to that sensor.
-.1
1'-15-
- - -
4. GROUND PLANE SENSOR EVALUATION
Although the study was primarily directed at the MGL and ACDN! lines of free space sensors, it was felt prudent to start with MGL ground
plane sensors to gain experience in the simpler situation that results.
Because of the ground plane, the possibility of any lead interaction is
greatly reduced, and, in addition, the probes do not have the aspect
sensitivity that the free-space ones do.
,* In the following sections we describe the ground plane that was
used, the measurements that were carried out on three different MGL
sensors, the calibration of the various components involved, the deduced
frequency responses of two of the sensors referenced to the response of
the third, and the implications of the results obtained.
4.1 Ground Plane
Ideally the ground plane should be infinite in extent, and one
consequence of using a finite size is that the surface field departs from
its infinite plane value due mainly to the effect of the edge currents.
For normal incidence on a plane several wavelengths in dimension, the
departures are primarily due to the edges perpendicular to the incident
electric vector that set up a standing wave pattern across the plate.
The influence of the other edges is confined to a distance of a wavelength
(or less) from them.
For use on another project it was necessary to design and
fabricate a ground plane which was small enough to fit comfortably
-16-
I."
inside the anechoic chamber but which still simulated an infinite plane
over at least a central portion of the surface. To reduce the surface
" field perturbation created by the vertical edges, resistive sheets were
added as shown in Fig. 3. Each sheet was shaped like a quarter
cylinder of radius 25.5 in. and had a resistivity which increased
quadratically as a function of the surface distance from the metallic
*plane. The effect of different resistivity variations was determined
by numerical experiments carried out using codes based on resistive
strip formulations, and the resistivity that was finally selected started
with zero at the edge, increasing quadratically to about 1000 ohms/square
at the rear.
The sheets were fabricated by spraying thin layers of resistive
material on art paper. After each treatment the resistivity was
measured with an ohm meter, and the process repeated until the appropriate
resistivity profile was achieved. The final products had a resistivity
which started at about 10 ohms/square and increased to the required
value of 1000 ohms/square at the outer edges. Although the surface
fields on the metallic part of the resulting ground plane have not been
measured, it is inferred from the numerical experiments that over a
central portion of the plane at least, the fields are virtually
identical to those on an infinite plane.
4.2 Measurements and Calibrations
The MGL line are B-dot sensors with specified equivalent area Aeq
Three different size versions were available for this study:
MGL-5B(R), serial no. 14, having Aeq = l m" i 2 ; MGL-7A(R), serial no. 5,
eqq-. having Ae 10-4 m2; and MGL-8A(R), serial no. 1, having Ae 10-5 m2
-17-aI
.........
40 inch sheets
52 in.
B -dotsensor
j 48 in.
-. 1000 Q/sq ___________
0~0 2/sq.
48 in.
Figure 3. Resistively loaded ground plane used in the measurements.
-18-
• •• . . .V. - o • , ' • , . .
All were manufactured by EG&G Inc. and are described in their
catalog [2]. Figure 4 is a photograph of the three sensors.
Each sensor was mounted in turn at the center of the ground
plane in our surface field facility, and its output was measured as
the frequency was scanned (or digitally stepped) from 118 to 4400 MHz.
To connect a sensor to the system it was necessary to use adaptors
as indicated in the first three lines of Table 1. These adaptors are
henceforth regarded as part of the sensors, and no attempt has been
made to compensate for their presence.
Any quantitative measurement is, by its nature, a comparison
process in which the reference is either the response of the same
system to a known field, or of a calibrated system to the existing
(unknown) field. For the sensor measurements the incident field was
unknown, but reproducible over a sufficient length of time to enable
each sensor to be mounted in turn at the same location on the ground
plane and have its response to the same field recorded. Any one sensor
can therefore serve as the reference for the other two.
In practice, however, there is a difficulty. To connect a sensor
to the measurement lead requires a length of coaxial cable, and because
of the larger outputs from the larger sensors, attenuators must also
be inserted. To eliminate their effect, it is necessary to calibrate
both the cable and the attenuator. To do so, two cables and two
attenuators were used as indicated in Table 1, and measurements were
made on various combinations of these sensors, and the four components.
The measurements performed are listed in Table 2.
-19-
Figue 4 Phtogaphof the ground plane sensorsusdi th
r'iure'~ evaluation.un
-20-
TABLE 1. COMPONENTS STUDIED
Component Description
MGL-8A(R) B-dot sensor, Aeq = 10- m2, Ser. No. 1; with BRRM(F) *
OSSM(F), OSSM(M) - OSM(F) adaptors
,4 2
MGL-7A(R) B-dot sensor, A = lO- m2, Ser. No. 5; no adaptorseq.
MGL-5B(R) B-dot sensor, A = lO" m2, Ser. No. 14;h eq.
GR SMA(F) adaptors
20-dB attenuator Attenuator, Midwest Microwave, Model 444-20
6-dB attenuator Attenuator, Midwest Microwave, Model 444-6
0.020-inch coax 35.5 cm long coax with OSSM connectors at each end;
Uniform Tubes, UT20
0.034-inch coax 61.6 cm long coax with OSSM connectors at each end;
Uniform Tubes, UT34
-
.D."L I-21 -
6j
La.. LL..0
cmI 0 0 4-jLL. Q. Q E
-L CD LA- a
Ln to' IU LL. I LL. ia - a
0 4DCD I~ 09
0
4-))
% L) L)
to
4) CD CD a11 %EU j r. 1 . . %r4 %
0. CD CD0 CD CD CD. CD CD
EUUC..) EUj
Q- VU4)
to U0
LAJ0 0 0 0 0 0 0 0 0C
04m
0D 0D %0 w C 04w0 L
C*.J 0
CD (z C D D C CD CD CD C
4A W
4-)
4) 61 V)L
ILp 77 77 77rr r. . l *41 CL C CL CL U u
too4) 4)
C e -t Lc, w r, cc) r-. c-L
EU ~ ~ ~ -22-
From measurements numbers 5 and 7 using the MGL-7 sensor with
and without the 20-dB attenuator, the effect of the attenuator can be
found. The ratio of the measured outputs is the response of the
K attenuator, and its amplitude and phase are plotted in Fig. 5. Although
the amplitude is slightly noisy, the noise is almost certainly due to
mismatches in the system rather than to the attenuator itself. Indeed
the attenuator is a physically small device with a specified operating
range of 0 to 18 GHz, and should have a constant response up to and beyond
the highest frequency used in the measurements. This would seem to
justify taking an average value for both the amplitude and phases. From
Fig. 3 the average amplitude is 0.1019 (= -19.48 dB), and the average
phase is -0.0326667 degrees/MHz. The latter is equivalent to an effective
free space length of 2.722 cm.
Similarly, from measurements numbers 6 and 7 the response of the
6-dB attenuator can be found. The amplitude and phase are shown in
Fig. 6, and average 0.5183 (= -5.71 dB) and 2.387 cm of electrical
length, respectively.
The same procedure was also used to calibrate the coaxial cables
(see Table 2 for the data sets involved), and the results for the
0.020-inch coax are shown in Fig. 7. The noise is again attributed to
the entire system rather than to this single component, and since the
skin effect is the main source of cable loss in this frequency range,
the amplitude of the response should be exp(-ct/f) where a depends on the
length and characteristics of the cable, but is independent of the
7. frequency f. By eyeball fit, the (average) amplitude at 3000 MHz is
estimated to be 0.8, implying a = 7.438 x lO" per MHz. The resulting
-23-
- -
* --
" '%CRL r :jP 2 -0 1 R TT , .,
r 157 t lH
- I4". - -.
uIHI
k -U
1 U--C ..-... ... . ---
* e Il{.l_ iYL..C' Cn' i O4] C'C3C..,....
Figure 5. Calibration of the 20-dB attenuator, deduced from measurementsnumbers 5 and 7.
-24-
%-J
-. .'--
.-7L 0,. I. C .
.- 25
ILI
• i
I0 -- I - . .~.~
u ij-
i ''
21 Figure 6. Calibration of the 6-dB attenuator, deduced from measurements
.. numbers 6 and 7.
,'-' -25-
"" ,,A-:L P3 . 2 0 :.: 4 41
1.5
z U-
U11
-26
-I I *.
1. . ? ..1 I i . ....... ... .. ...
Ii I ,,u~ ;. IIH-
iinI
i, lilI 1 I.
I I
' I II \ l
" '.
Figure 7. Calibration for the 0.020-inch coax, deduced from measurementsnumbers 9 and 10.
.. -26-
ad
amplitude curve is shown in Fig. 8 and is seen to provide a good
fit to the measured data. The slope of the phase curve is likewise
-0.61463 degrees/MHz, corresponding to an effective free space length
of 51.216 cm, and the computed phase curve is given in Fig. 8.
The analogous results for the 0.034-in. coax are shown in Figs. 9
and 10. With the amplitude chosen to be 0.808 at 3000 MHz, the deduced
is 7.106 x l0" per MHz, and the slope of the phase curve is -1.0556
degrees/MHz, corresponding to an effective free-space length of 87.970 cm.
4.3 Sensor Responses
Measurements numbered 3 and 4 (see Table 2) were done with the
MGL-7 and MGL-8 sensors respectively, using the same cable and attenuator,
and the ratios of the measured data are shown in Fig. 11. At low
frequencies the ratio is 9.33, compared with the value 10 obtained from
the equivalent areas. The discrepancy is believed due to the small size
of the MGL-8 sensor whose construction is inevitably difficult. The phase
is zero at low frequencies, but deviates at higher frequencies, due not
only to the effect of the sensors individually, but also to the inclusion
of the adaptors whose presence has not been compensated for.
The MGL-8 is the smallest sensor of the three, and its response
should be linear out to (and beyond) the highest frequency used.
According to the quasi-static approximation, its output voltage should
therefore be
dV wIIHA eq
where w is the radian frequency, p is the permeability of free space
"( 1.2566 x 0"6), H is the amplitude of the magnetic field, and
-27-
-Jk::
'~~I . .. . . .. ... . .. . .. . . . .... .
I ti
"I
kI'
I'
11,0 f1
Q I
I I
-28--
-'lF g r 8 . Co p t d c l b i on f r t e O 0 0 i o xI Io.
I .L b;..c W..; %
Li
C-29
r., ...
%I ,
F Fr
"S.a"- V .. .. .. .. . . . ..... . . . .. .
, i ' ' L V 'U r I , i-4z
Fiue9 arto o h 034ic c, deue from measuement* number 7 and 8
I-29
C'
il" 1 ! n , l ]2
low. -0
-rJ
13 I ( 4
LI C '-!H
I Figure 10. Computed calibration for the 0.034-inch coax.
-30-
- -___-- __ _ -
-. ,31
I cI:- .- -- -- . ....-- . _.... -.. . --
0 !@ O~ L00 ":c1 40%8 - :,:
* ,
C-4LI
310
I
0 I0 00 2000 i0 4 0L0 .
F'..Ey 'C r arIHz,
Figure 11. Response of the MGL-7 sensor referenced to that of the MGL-8.
-31-S
5 2
Aeq I0-s m2 On multiplying the data in Fig. 11, by jwuA , we
obtain the response of the MGL-7. The behavior is not unexpected, but
the droop in the amplitude curve at frequencies above 4000 MHz was not
present in some of the earlier data, and may be due to an equipment
mal functi on.
The ratio of the measured responses of the MGL-5 and MGL-8
sensors is shown in Fig. 13. At low frequencies the ratio is 90.0,
compared with the value 100 predicted by the equivalent areas of the
two sensors. At a frequency of about 500 MHz, there is a sharp notch
in the curves, the origin of which is not clear at this time. It
" . occurs in all data that have been obtained with the MGL-5 sensor, and
is therefore associated with this sensor rather than the MGL-8.
In an attempt to locate its source, the measurements were
repeated with the front of the sensor retaped to the ground plane. We
also taped the back of the sensor assembly where it protrudes through
the ground plane to eliminate any possibility of resonances in a cavity
formed by the plane and the plate on which the sensor is mounted. There
was no change in the data, and it would therefore appear that the
- -notch is a feature of this particular MGL-5 sensor.
To test this, it was susggested [4] that the sensor be examined
electrically using a time domain reflectometer, and visually as well.
It was found that the two 100-ohm lines from the gaps have an
electrical length of about 15 cm, but one is about 1 cm shorter than
the other. Because of the design of the sensor, the two lines and the
-32-
F~ ~ ~~1: HI.-~gr~;F-z
I Oct
I. -"F .': r"
/ r
--- 3
Fiue 2 esoseooheML-o eso ih h ,L8usdasasanad
r-33
I0 Ot -------------- -___ ____________
M-L- :L -
4 0
-J 40
'IC-10.. -:0 1c:
Figur 13C r -7 70 1000-0 . .. C
Figur 13.Response of the MGL-5 sensor referenced to that of the MGL-8.
7 .-34-.
loop can form a high Q resonant circuit for excitation in the
anti-symmetrical mode, and if the 100-ohm lines are not of equal
length, the resulting signal could create the notch at 500 MHz.
In presenting the responses of the MGL-5 and MGL-7 sensors, the
MGL-8 has been used as a reference because of the expected linearity of
its output over a frequency range exceeding the highest frequency used
in the measurements. On the other hand, its small size increases the
probability that its specified equivalent area is less accurate than
those of the larger sensors. To judge from the low frequency limits
in Figs. 11 and 13 it would appear that the actual value of Aeq for the
MGL-8 exceeds lO m2, and a more probable value is (say) 1/2{(0.933) "'
+ (0.966)'1}0 " m 2, i.e., Aeq = 1.054 x l0" m2. The resulting
recalibration of the MGL-5 and MGL-7 responses would uniformly increase
the amplitudes in Figs. 12 and 14 by 1.054.
4.4 Discussion
The amplitude and phase of the frequency response of the MGL-7
sensor are shown in Fig. 12. At low frequencies, the response increases
linearly with f according to the formula
e q ' (20)
with A = 0.933 x 10-4 M2 . It rises above the value predicted by (20)
eq
starting at f = 800 MHz, indicative of the growing influence of the
multiple contributions, and then recrosses the linear curve at about
1800 MHz, where the response has its (initial) peak value. Thereafter,
4 the amplitude falls increasingly below the linear curve, and is more
than 3 dB below for all f > 2300 MHz.
-35-
6
o1r-1& riGL- :R Z
II
-IS
Figure 14.=_ Repos of, thG eso ihteM L- use as. a... adard
-J j ,- /
,I-- . A /II - -,".
. . v,-*.,j ". 7 - -
Rt'-£ -LIE-IC
V I
•1 , [, /'I":J I
, /
a . ' ---
- INy. .- - - l-4 l'J- rl- I ,
S J~'- Cxu zCO~-4lLLb
* F't :i ! M z
.iue1. Repne fte G- sesrwtIh G- sda tnad
u-36-
6,
Some additional information can be obtained by applying the
Sharpe-Roussi program [5] to give a partial fraction representation of
a real (as a function of s = jw) rational function approximation to the
ro response. For this purpose it is convenient to write the response as a
function of w in units of l09 radians/sec. Using ten poles, the fit
to the measured amplitude and phase is shown in Figs. 15 and 16, and
we note that even the small amplitude dip near 4000 MHz (w = 25) was
faithfully reproduced. The poles and residues are given in Table 3.
The resonant frequencies are therefore 12.19, 18.10, 25.58 and 26.23
G rad/sec, but because of its small residue, the one at 25.58 can be
ignored. The other three are indicated in Fig. 15. The fundamental
corresponds to that of a loop of radius 2.496 cm, and the next two
values are then consistent with the first anti-resonance and the second
resonance, respectively. It is not known whether the loop radius can
be identified as the dimension of a specific structural element of
the sensor.
For w << 12, the partial fraction representation of V can be
expanded as a power series in w, giving
V = -0.0216 + jw 0.1174 + w2 0.001678 + jW3 0.0001399 + 0(W4)
(21)
The non-zero constant is physically unrealistic, and is due to noise
in the measured data and/or the approximations inherent in the
numerical simulation. The dipole contribution (proportional to jw)
implies (see eq. 1) A' = 0.934 x 10 i 2 in excellent agreement witheq
-37-
• .U. _ . . " . - ', _ . . . . . .. • . . , . . . . - .
0 C
CD
CL
4-)
*- 0_
A 4-C.Q00
CL- 4
(A-
0 CA
-J 0.4-J
a 0) 0
CY 4-0
* *L0.
r. 3-
0 0a
-J
CIL
CD~
'0041 C
)0.
Im
Es- * . 0
-39-
TABLE 3. POLES AND RESIDUES FOR MGL-7
Poles wi Residues R.
12.19 + j4.452 1.521 - j2.839
-12.19 + j4. 452 -1.521 - j2.839
18.10 + jO.9277 0.1528 -jO.008782
-18.10 + j0.9277 -0.1528 -jO.008782
25.58 + jO.4500 -0.01499 + jO.07588
-25.58 + jO.4500 0.01499 + jO.07588
26.23 + j2.384 -2.703 - l.759
-26.23 + j2.394 2.703 - jl.759
j83.96 j277.7
j93.29 j311.5
-40-
the value deduced from Fig. 11. The other terms in (21) are then the
next higher order multipole contributions, and bearing in mind that
- (21) requires w << 12, their coefficients are remarkably small. Although
S-these terms are responsible for the initial departure from the linear
behavior with frequency, it is evident that a study of the quadrupole
contributions to the sensor response would provide little in the way
of an improved representation of the output.
For the MGL-5 sensor the amplitude and phase of the response
(see Fig. 14), replotted as a function of w in G rad/sec, are shown
in Figs. 17 and 18, along with the simulation provided by the Sharpe-
Roussi program using 18 poles. The frequencies appropriate to the pole
locations are indicated, and we observe that these do not always
correspond to the peaks and nulls in the response. Nevertheless, the
overall fit is excellent except at frequencies in the vicinity of
= 3.7 where the notch occurs, and where the simulation averages
through the increased response at the lower frequencies and the
subsequent deep null. Because of this, the low frequency behavior
of the simulated voltage differs from the known behavior of the sensor.
Indeed, by expansion of the partial fractions we obtain
V = 0.0607 + jw 1.135 + O(W2)
= 0.903 -3 2implying A'q 0.903 x 10 m which is still seven percent less than
the value obtained from Fig. 13.
By applying the Sharpe-Roussi program to the low frequency
portion of the curve up to w = 10, the notch at w = 3.6 is faithfully
V reproduced using 14 poles, as shown in Figs. 19 and 20.
-41-
L. . ---. _.
-J
c
s-
01.
•E 0
o -
I -
- * IE
S).-
4-
(4- c
CL W
U.- -V
, o 4
4Cfl
o -4 2 -
I
• 4o(..)
• -42-
-&n 4
U7)
4.)
:4 4I-.-
C3,
0
q) 4-)
I It)
z (A
3SV~d-~C 4.WK *v-c-
cm0
-43-~
"W~~-~ .... .... .... .... .WWI
41
Loo
I-!-
tic
x 0L
LC
Zi,
U-
-44-)
lop)
00
CC
*1u
it-j
*n 0.
L
-(40'
0 00
IL.)
-45
5. FREE-SPACE SENSOR EVALUATION
Apart from its intrinsic value, the preceding study of ground
plane sensors was helpful in enabling us to develop the experimental
techniques necessary to tackle the more difficult problem of
evaluating the free-space sensors.
5.1 The Sensors
The free-space sensors studied were the MGL-2D(A), MGL-6A(A)
and ACD-4A(R), but in addition the smaller size MGL-9(R) and ACD-2(R)
were used as references for the frequency variation study since their
regions of linearity extend well beyond the largest frequency used
in the measurements.
. Table 4 identifies the particular sensors, and we note that
"- two MGL-6's are listed. Because of the poor performance of the
first model (serial No. 2), it was felt that the sensor was defective,
K; and a second model (serial No. 1) was requested from the Air Force.
It turned out that this was no better, and we eventually decided to
employ the second one only. Figure 21 is a photograph of the five
sensors used in the study, with the ACD-4 and ACD-2 on top and the
.* MGL-2, MGL-6 and MGL-9 below. The dual coax leads on the MGL-9
have been bent through 90 degrees to make the sensor compatible with
the other two. Since the incident electric field in the chamber is
horizontai, it was necessary for us to bend the original straight
leads to enable the sensor to receive the (vertical) magnetic field
-46-
4- E
r_ 0-.O4-1-4--
U((A
C 4-)CJ .(A u
U= 4-) 04- .0 C~
0 to-( a - . 0 U(x -0 W (U1
tz 0 4 V) toaJ4-- ~ ~ ~ . 3_.1 ... , )
(1 4- Aa
aI. 44)* c(UU sJ0J 0
u4- 4-) Ln4nfl go O c cu
C -)n X -L = -u
Li.)
LO Iui
A1
0)0e
- 0 01-J 0 . -I
- (ULi.) r-.LJ C) CDW C)LJ 1
4- 1oa e DO. r- *a -~ a CS aLL. ~ ~ ~ ~ I s-C C )CD mC M0tio ~ ~ r- r-= R* L-. .J ( U
LLUG
OC E~ CD a-C ODC>~~~~ Ua OQDCDaCco C 0 CD
I- - * * LO .0 *M * *X, CDr-% O CD Mc Or- Or- Or
0 ui
4JI
V~ 0 x x -CL aj 41 4 tZ .) o 4 to 4to 4
ea I 3c 1 3L) co CL CO L36
-47-
ACD-4A(R) ACD-2(R)
MGL-2D(A) MGL 60(A) MGL-9(R)
Figure 21. Sensors used in the study.
-48-
with the leads vertical. Figure 22 is a close-up of the region near
the bend. The coaxial cables were unsoldered and separated up to
about 4 inches from the loop, bent with a radius of about 3/4 inches,
and then resoldered to avoid the possibility of any loop resonances
on the leads.
5.2 Experimental Techniques
The chamber and its instrumentation were described in
Chapter 3 and, in principle at least, measuring the frequency response
of the free space sensors is straightforward. Each sensor is
positioned and supported inside the chamber arid itfi output recorded
over the frequency range from 118 to 4400 MHz. The sensor is then
replaced by the smaller sized calibration sensor (MGL-9 or ACD-2)
and the measurement repeated. From a knowledge of the latter's
response, the ratio of the two outputs gives the normalized response
of the original sensor.
This procedure worked well for the ground plane sensors. The
same signal lead could be connected directly to the base of each
sensor and, in addition, all the cabling was beneath the ground plane
where it was invisible to the incident field. The situation is very
different with the free space sensors. For example, each of the
MGL-2, MGL-6 and ACD-4 sensors has a pair of coaxial cables extending
out about 18 inches from the sensor and terminating in a connector.
The connector is a so-called twinaxial where the transition is made
from the pair of coaxial cables to a twinaxial cable. At the far end
.of the twinax, a transition is again necessary to go to the 50-ohm
-49-
0
Pair SMIA (jack) connectors
Pair 0.085 in. dian.semiricid coax
H
4.75"
KFigure 22. Modified MGL-9(R). The coaxial cables werebent 4.75 inches from the loop to adapt forH-vertical polarization. In the originaldesign the coax was straight.
-50-
, k . '. . . - .- . L- -
- -, . _ . ii'
. ." - ." " . . .. . . . - . - . .. .-
coaxial geometry required by the network analyzer. Although
balanced to unbalanced transformers are available to fit the
connector (EG&G, DLT-96 series), they are usable up to 130 MHz only,
and therefore inappropriate to the frequency range of concern to us.
An alternative is to change from the twinaxial transmission line
geometry to a twin coaxial one, and to record each of the two outputs
individually, combining them digitally using the calculator. A device
called a twinaxial connector (EG&G, TCT) is available, but
unfortunately it is quite bulky. To avoid interference with the field
inside the chamber, it is necessary to place it outside, which then
requires a long length of twinaxial cable from the sensor. Because
of the discontinuities produced at the two connectors, standing waves
are set up on the twinaxial cable which show up as oscillations
in the measured data. Though we tried to reduce the oscillations using
a 6-dB twin-line attenuator which we designed and built for insertion
in the twinax, the resulting data were still unacceptable.
We remark in passing that the removal of the twinaxial connector
at the sensor would give direct access to each of the coaxial leads
from the sensor, and eliminate most of our problems. However, this
would modify the sensor, and since our results would not then be
appropriate to the sensors as built, this simple solution was
unacceptable.
By this time we had concluded that the most practical approach
was to separate the twinax into two coaxes as close to the sensor as
possible, thereby minimizing oscillations due to cable mismatches.
4
-51-
I.
A miniature twinaxial to coaxial adaptor was constructed using
twinaxial and General Radio coaxial connector parts, and this is
shown in Fig. 23 along with the TCT-lA connector. The adaptor is
no larger than the twinaxial connector and has two 0.141-inch-diam.
semi-rigid coaxial cables with SMA connectors coming out. Because
of its small size, the adaptor can be attached directly to the
sensor and the signal can then be taken out of the chamber dith a
pair of semi-rigid cables; and since these cables and the connectors
have low VSWR, they produce no appreciable mismatches to affect the
data. The mismatches that remain are at the sensor, twinaxial
connector and the adaptor. These are all close together, and the
oscillations that they produce in the frequency data are relatively
slow. Such oscillations are of no concern if a sensor response is
calibrated against the response of the same sensor at a different
aspect, provided the associated cabling.is in no way disturbed.
The above problems do not occur with the small calibration
sensor. The MGL-9 and ACD-2 do not have twinaxial lines. A pair
of semi-rigid coaxial cables lead directly from the sensor to a
connector of SMA type, and since there are no discontinuities, there
should be no oscillations in the frequency response data. An
unfortunate consequence is that the oscillations in the data for the
larger sensors are not removed when the data are calibrated against
data for these smaller sensors.
Compared to the ground plane sensor study, the measurement of
the free-space sensor responses was difficult and frustrating. In
spite of our best endeavors, all of the procedures tried led to noisy
i -52-
"v .
i.4
Ca) Miniature (b) rcT-lA
Fiqure 23. Twinaxial to dual-coaxial adapters.
ao
-3
C
- .I. .
data. Whenever a cable was changed or even moved in the chamber, the
oscillations were affected to such a degree that they were no longer
eliminated by the calibration. To obtain data which are, as much
as possible, independent of cable mismatches, and require minimal
disturbance of the sensor in the chamber, it was therefore decided
to separate the measurements into two parts:
(i) angular response measurements,
(ii) frequency response measurements.
In (i) the procedure is to record the data for a variety of e and/or
p, and to normalize these data against the maximum response for that
sensor, e.g., the data for broadside incidence for the ACD sensor.
In (ii) the response is measured only under the maximum signal
condition and normalized with respect to the corresponding response
of the appropriate calibration sensor to determine the frequency
response of the test sensor.
5.3 Angular Response Measurements
The difficulties described above ate up considerable time,
and many measurements were made before usable data were obtained.
5.3.1 ACD-4A(R) Sensor
This was the easiest to measure and was the one studied first.
The sensor rested on a styrofoam pedestal on which a piece of polar
paper was placed to show the sensor orientation, and the two semi-
rigid coaxial leads went up vertically through the roof of the
chamber. Figure 24 shows how the angle e was defined relative to the
-54-
aj
H
0E
Figure 24 Mesrmn- emtyfr h esr. (h
/-aito rdcsteanua dpl)rsos.
-55
• _" -, -,.. - , . - .- . . . . . • -. -
direction of the incident electric field, and measurements were
made for e = -90(15)90 degrees. The data were normalized relative
to those for e = 0 and are presented in Appendix A. Thus, for
0 = 0, the normalized amplitude is unity and the phase zero.
5.3.2 MGL-2D(A) and MGL-6D(A)
The geometry is shown in Fig. 25. In contrast to the ACD
sensor, the MGL are not azimuthally independent, and it was therefore
necessary to measure the aximuthal (@) dependence as well as the
e dependence appropriate to the angular or dipole response.
To measure the , variation, the sensor was placed on a
styrofoam pedestal to which a piece of polar paper was attached. The
output leads were taken vertically up through the roof. With gap
No. 1 used as the zero reference for the aximuthal angle 0, data were
recorded for 0 = -90(15)90 degrees with e = 0, and normalized with
respect to the measured data for 0 = 0. The normalized data are
presented in Appendix C.
To measure the angular response, the sensor was rotated forward
through an angle e (see Fig. 25), and a special jig was constructed
to accurately control the tilt angle e. Figure 26 is a photograph
- of the jig with a sensor in place. To permit the bending, a matched
pair of coaxial cables was inserted above the sensor handle (see
Fig. 26), and a braided shield was added to avoid leakage and resonances.
For each sensor, measurements were made for e = 0(15)90 degrees with
= 45 degrees, corresponding to incidence midway between gaps 1 and 2,
*and = 90 degrees, corresponding to incidence on gap 2. The data were
-56-
=-#1
k k 7
GAP
• 4
Figure 25. Measurement geometry for MGL-serles sensors. (The e-variationproduces angular (dipole) response and p-variation producesazimuthal (constant) response.)
I
-57
Figure 26. Styrofoam jig used for supportand rotation of MGL-2 and MGL-6sensors. (The incident field isfrom the right-hand side.)
-58-
normalized with respect to the measured data for the sensor in
question when e = 0, and are presented in Appendix B.
5.4 Angular Response Data Analysis
The measured data were processed according to the formulas
in Chapter 2. For the angular (dipole) response, the quantities
computed are
(i) the dipole response coefficients A (f), from (14);n
(ii) the mean squared angular errorC 2(f), from (15);
(iii) the maximum error ''max(f), from (16).
For the MGL sensors, the following additional quantities were
computed to determine the azimuthal variation of the response:
(iv) the average rotational response coefficient Bn(f), from (17);
(v) the mean squared rotational error 4(f), from (18);
(vi) the maximum rotational error ( maxf), from (19).
Two programs were written, one for each type of calculation, and the
processing and graphics were done on the HP9845 calculator.
Figure 27 shows the angular (dipole) dependence of the response
of the MGL-2D(A) sensor as a function of e with 0 = 45 degrees and
f = 150 MHz. These data were recorded at this one frequency during
the "pre-measurement" stage of the study, and are presented here to
help in understanding the meaning of An(f),0 2(f) and 8 max(f). Thus,
An(f) is the coefficient of the best fit cosine curve, e(f) measures
how closely the curve fits the data, and max(f) is a measure of the
maximum deviation of the data from the curve. The fit in Fig. 27 is
* very good but as the frequency increases, so do the deviations. This
-59-
L.
- HI
- >1
-J 4
0 U
0 --
'4-
0 L.
.-n #A
C4
-600
-J - - - - -l
is evident from the subsequent figures where the various quantities
are presented as functions of frequency:
Figures 28 through 32: dipole response for ACD-4A(R);
Figures 33 through 37: dipole response for MGL-2D(A) with
gap No. 1 at =900 ;
Figures 38 through 42: dipole response for MGL-2D(A) with
gap No. 1 at 0 = 450;
Figures 43 through 45: rotational response for MGL-2D(A);
Figures 46 through 50: dipole response for MGL-6A(A) with
gap No. 1 at 0 = 900;
Figures 51 through 55: dipole response for MGL-6A(A) with
gap No. 1 at 0 = 450;
Figures 56 through 58: rotational response for MGL-6A(A).
The implications of the 3-dB frequencies shown in these plots are
discussed in Chapter 6.
-61-
A, D-4 R iR) D ipc Rrq P - C111
p 1.5i
15 PIR7 12z Li
tt5 gi UP,
500 I~2...c 1 C I)L'UErJI: e MHz .. .
Fgure 28. A n(f) for ACD-4A(R).
-62-
.. 1
.iC-,l0l8 :F.FqEr
t.- 9 ,.,U
p . r G
1798 MHz
Figure 9. mxf* for ACD-4A(R). 60
"-63
• 02 ,51 .4..,t:: .. , '...L
500 1000_ 1502 jH 5 0 "_ 0 00O ;-s D oo , -.-,i
Figure 9. max f) for ACD-4A(R).e <60
i'
Al-:.R P'' dIij- Err -'' C
.4I
2 .q 2402 Mliz
- ~ ~ ~ i .17 R 9Z UMi
0 ~~~~j -10 2~u FJ Q.U 05C 3~0 5040FREKU-'ENC~ r MHzI
Fgre 31. 'mx~ for ACD-4A(R), H< 300.
. 3 3dB................................2402 MHz
0 clct2 Iso 1500 e 2500 000 ?SOO 4 0 S '.:'
Figure 32. a x()for ACD-4A(R), e=00.
-64-
NGML-?DLA,ps9;Dipol* An .sp;rILIJ11
.5
12 PIRF7 92 UPI0
S OOJ "'50 100E017 1250 1500 1150 il00FREOUE14CY 'MHz)
45
12 TR~ 92 UPI
0 250 Soo 7's0 1000 1250I 15 00 1750FPEOUENC ( MHz)
oFigure 33. An (f) for MGL-2D(A), gap No. 1 at =900.
-65-
I-I
.04
-" P P -- Uri
0 a 5O 00 - 1 011ci 12a i10 oe2CIPFPUA I EtJCY H
Figure 34. 42(f) for MGL-2D(A), gap No. I at 900.
Err .- O.&:- L 1:
II
.2~ . 3dB4932 M4Hz
1 4 1 P I 'r , I
a'.- .250 500] ?5O IO'r¢ _ 12f 1: :,m C.O' 15' *-O'0,-.
Figuur35 34.m (f) for MGL-2D(A), 1eJ 600, gap No. 1 at, g 9oo.r 3. mx
-66-[
, . ' -V.. -.. . -- - ,
Pis Ef 32 t
•,
" .-J . .
.4-"
3 dB,, , 1215 MHz
"-L3 215ff C, C '50 1 C C 12 15 f C IO0 .0,2
F' ,uE ,: Y MH-z
Figure 36. (f) for MGL-2D(A), gap No. 1 at : 900, IJ < 30'max
rIGL-2ntR--; GEro P .I r r.0: MCLC15
.4
3 dB
.23 1215 MHz
.C 250 50 COO', 1 .O .+ I .. --.
Figure 37. max(f) for MGL-2D(A), gap No. 1 at + : 900, a 00.
-67-
1-i?1 7~
0
°-...
• .-
LI.H
0I- D5 A0 J5 1000 250 P,--~ I 1Hr
454
WW
-45-.
0 2 Q 0 so 100 1250 1+0 . .50
FRE)UPE: MHz
g0
Figure 38. A(f) for MGL-2D(A), gap No. 1 at 45
-68-
.03
-J P1J
-92 Uri
3 dB
Fiur f F C o10 1250i 1~J c ic; 1u:5
FRE-DC,:, tl'Hz -
Figure 40.(f) frMGL-2D() , gap No. 1 at 45',O a ..60
-69
D, N., s E -
.4
A
.2 I'
I , , -" 3 dB.1 .,IIF,...,.'.,i
rji1397 MHz
0 250 5 ' L 5o IH0 I C1250 1.QX1 -cl
f -uH': Hz
Figure 41. 6m (f), for MGL-2D(A), gap No. 1 at 450, < < 300
max
i ' PIL-2D.} 4I;sr41- Err, 3MC L20
.4.."7"
-j
3 dB.1 ~1397 MHz
1 0 1 P sl U1
a 250 5O 0' ,, I "CECl 12d5B0 1 3 1.0I:'F;;'UOIrJC 'r' ,'Mz,
Figure 42. Cmax (f), for MGL-2D(A), gap No. 1 at 450, e = 00.
-70-
~1-D' Pn F'ct a tt a ri s, r M iLO 1
1.5
Ld
e 1 -----. ,
- I"
C
0 I I I 15 PIR US PI
1-45
0 250 S0 C, E0 1000c 120 1 0 1750CC, C C1-, "--
I,7
I0 IIII I
. r l~~~tGL-2DLFJ ;P,-t~t i.-ar-,al F'esp: M"GLC'?--:
--
I:. ,/r . .... . .. ... . ......
6 n
"i. -71 -
.~ 04
Figure 44. Cr orML-2DA
NGL-f) MGL-MO(A).
13172-I
, °..
. cL -RF I-. GF;90.EIII:.:I. A4n: R's £pj r1LOu 1
1.5
•l I l 1 0 i0 CI 'S
F*o E. ~j: P
.
1~~~A :1 L.s- &..~
45
.5
12 PIP a-- UPI
*~~~~ ' 50 10010200200 0 ;SD 4 CC 4SCJK
Kiur 46. A (fIo G-AA a o t 90
0., "'" 'C o '.- -,
-, ------- 4I I I
,- '3 50'-) l 0 150.) 2000C 250 00 :co "-"c0o 4oo3 42C- .JC 7
n
-73-
E lL _ I ... . .. .IIn , w i. -. .• • pr: E' Ol. L J . , I , _
Cie
.04
LC"
J-1. -; s 0 I- -I
J --,
A Ol
SI
3 dB
2580 MHz
,,, ,, I ; . --- ,'-12 P4IRC G 2 Pqt
-00 1 il 2 C -000 2 500 .O: 00 -5P 0 4': 4 ,'
F" Ekl,:Ei, r Hz
•.. Figure 4. (f) for MGL-6(A), gap No. 1 at . 900 ,Ie < 600.,max
-74-
L: S90; Mi - r
- - I 1
_j .4
-J , 4 4
4 3 dB. v. 3307 MHz
1:1 5 00 06-ifD 15 0i' _Q -1 2500 . D:0 ce ":E 0 .- 3:
FE!LIEr.r': Y ( Hz
Figure 49. 4max(f) for MGL-6A(A), gap No. 1 at 9o°,lel < 30" .
4 , rl i -
.4 I
S3 dB-j 3307 MHz
a 50 I C10D I .CO 20010 2500 3000 5 0 4 13O% -45;: S7
Figure 50. a(f) for MGL-6A(A), qap No. 1 at 90°, e : 00
-75-
P.' ;PIiOii
0 ~ -1- tIFF 92 U"t
S' 0U Q D! s 15-30 0002U 2500 1:1L0 -500 4 000 4%
PREOUENCY 'MlHz
45j
I
-451
Soo0 I0 C0Ci ISOO, 2 LIOC 2500 0 00 SSDO 400 'F"RUUNCY rHz
Figure 51. An f for MGL-6A(A), gap No. 1 at 4~450
-76-
.1
-Ao -
* LIII
* .4
F~~:2880 M Iz
Figure 5 .em x(f) for MGL-6A(A), gap No. 1 at 4509 60
-77-
m~ ~~~~ - -- - -- - -- .r - . . - • . .
7"6; i -I ,f-s t , - r'!,; L I
1,2 "I2
92 i'
C)C l 0 1' 0 0 4' C I
.8 , *1
N"L.
Fiur 54. ('f fa
--"",4 :N a- Ert . L I !~ i
.4
S3 dB
., 3439 MHz"' al l' i V"f ' r .
J- If l,
12 p-1l; u n 1
a
:.," 1.- ,F G- UP, I
3 -013 1O' 1500 ~030 '500I~ 0 00 : 5 DO 4 0C' 47-1
Figure 54. 6'max(f) for MGL-6A(A), gaps at : 45 lel < 30°
.- 71
: " " I
I /
-:. -i .4 ]
r )
~ ~ .. ,..~ .~.r....>\~j' A', 1 , 3454 M~z
'3-- _ __ __,__ _ _______ _
-0 L i l ,' 1 5 0 2 O0 3' -: S O O :O .'-:5 D O 4 Q,:1 ' ~.).'.
*- Figure 55. m (f) for MGL-6A(A), gaps at , 45, e) = 00.
::: "ax"
.'.. -78-
2 1
*71.5 I
CL
Cl 50G 1000Cl 1500 26,00 2500 0oo :1500 4000'2FREc'UEIN':y r Hz
4-5
c'i
Ap RP92 L-11
0 5o0 I0. 113 I 00 2500Q C0i l 4000'- 4E577F;El'LIENI: Y MHz
* Figure 56. Bn(f) for IGL-6A(A).
-79-
r1GL-G fA 1; H~r. Er ri; M.LL02
.02
.02
Figure 57. 2 (f) for MGL-6A(A).
.4
3 dB3256 MHz
CI 51 0~1~ C-RO 25002 345 0 CI SO0 4 ClQ,0 4~
Figure 58. (F.axf for MGL-6A(A).
-80-
5.5 Frequency Response Data Analysis
The normalization of the data in Section 5.4 eliminated the
effects of many of the cable mismatches, but also removed the true
variation with frequency. We now seek this information for the MGL-2,
MGL-4 and ACD-4 sensors.
With the sensor oriented for maximum signal the response was
recorded for each of the above three sensors as well as for the two
calibration sensors. To minimize any errors resulting from changes in
the equipment and/or chamber, the measurements were carried out in as
short an interval of time as possible. We include here four curves
of more than 60 measurements that were performed for the frequency
response analyses, most of them unacceptable due to the bad oscillations
in data caused by cable mismatches. Once these problems were isolated
and partly circumvented (c.f. Section 5.2) data were recorded and
where appropriate we have averaged (or combined) two sets of data to
produce the curves presented.
Figure 59 shows the response of the ACD-2 sensor normalized with
respect to the MGL-9 response. Both are miniature sensors with 3-dB
roll-off frequencies of approximately 7.5 And 10 GHz, respectively, and
the ratio of their responses should therefore be constant over the
measured range of frequencies. It is not. At low frequenices the ratio
oscillates around the constant value of approximately 1.25 (compared
with the theoretical value of 1.33), but for frequencies exceeding
about 1600 MHz the oscillation is about a linearly increasing function
of frequency. We have no explanation for this increase, and since the
effect of cable losses was eliminated by referencing the responses to the
output terminals at the sensors, the increase cannot be attributed to this.
'" -81-
For the other (ACD-4, MGL-2 and MGL-4) sensors, the voltages
were referenced to the twinaxial connectors and the phase data
adjusted by adding a constant multiple of the frequency (equivalent
to a delay) to produce the required constant phase at low frequencies.INFigure 60 shows the ratio of the ACD-4 response to that of the
ACD-2. If the latter is presumed linear out to 2000 MHz, the curves
serve to define the frequency response of the ACD-4. At low
. frequencies the measured ratio averages 120 compared with the theoretical
value 100, but decreases starting at about 800 MHz. The 3-dB roll-off
frequency is 1096 MHz, which is consistent with the manufacturer's
specification of >750 MHz. The response of the MGL-2 sensor
normalized to that of the MGL-9 is shown in Fig. 61. The low frequency
ratio is estimated to be 400 and the 3-dB roll-off frequency is 482
MHz. The corresponding values listed by the manufacturer are 500 and
>300 MHz.
Finally, Fig. 62 shows the ratio of the MGL-6 and MGL-9 responses,
and illustrates the type of difficulty encountered throughout this study.
.- The fact is that a twinaxial system (cables, connectors, baluns,
etc.) is not adequate above 100 MHz. The oscillations that are seen
are attributable [6-9] to the MGL-6 geometry, including the handle, and
the other model (serial No. 2) of this sensor showed the same
behavior. The low frequency ratio is estimated to be 40 compared with
the theoretical value 50, and the measured 3-dB roll-off frequency is
1940 (c.f. >1800 MHz).p-2
i, -82-
,..- A..~.. N,-_.:~~A D- 2 P 1 "FI'L-1- j; ,, .. "
r--i
CL
3 dB2425 MHz
Iia PiR7 r. LJMI
0 10 1,0D 1 000 2EIC .E OI' 0 00 4 OCCt 4s
FRI E J_-'r' (JlC.
.~ - it-i :-s% ,4 , ***.-..V ~.. . ....1
-45
"r 10 PIRR 92 UPI
a S~ C .u; Q C'.J 2O3O 31 2000) :-: 00400
I Figure 59. Response ratio of ACD-2/MGL-9.
-83-
,U
RCn-4R P "ACD--i -
180 t.4"r'lA"
3 dB1096 MHz
-1 3- 1 P R P
F-;t.'EJEC 'r' , lk z ,
90RCfl-4F I F' R "CD-2t 'F N_-i '
I.- I
jI ,1
-. 0sy~ p ' '. t, ,. ' \1I'',Qf
. , ,,}\ ,, ' ,'
-45
i'' - ,I -- _ _ _ _ _ _:. _' _ •., _ _.',
I. 00 1000 1500 20C1 0
FRECLCIF:V ' Hz
Figure 60. Response ratio of ACD-4/ACD-2.
-84-
il -
.U L . .......
M L-2 Rj -'lr;L- J t- -
IO0 3 dB482 MHz -
, _ ___- . - - . )
1001
P' E,;uC'1C ' 1 , 1.Hz.
~C
-85
M iL-2DiAi -'_ - - ,-; r1,- .:-!
L.. -
, 10..
L.J
." "1
: Figure 61. Response ratio of MGL-2/MGL-9.
°
',.;:-85-
RD-Ai27 8i7 MEASURED CHRACTERISTICS OF MULTI-GRP LOOP AND 2/2ASYMPTOTIC CONICAL DIPOLE..(U) MICHIGAN UNIV ANN ARBOR7.i78 V V LIEPA ET AL. MAR 83 Wi?86-i-F RFWL-TR-82-82
UNCLASSIFIED F2968i-8-C-802 F/G 9/i N
b.9
Ii
11111gig 1L3 28
NATIONAL BUREAU Of STANDARDS-1963-A
- _-_
III:
i
P .. ; : ' .---..' -- . -, 2 " ' , -.-i ' , ... .- .-.-. . -
. ..- _'.
.' -. - . .
.-
. .-
.-.. . . .- - . -. ' ." . ---.' ".- -' --. ' . . " . -
.r l... .. ........... ...R. A I P
".'
3
194050 M,~
" 45
lm
;L
I-.I...N
u-' ,J ,\) o
S0 Q I ON I : 3 dB l 'I . l lO 1940 tIlz '.
'I 7'
,-.:II)I~i I
k*" " 1~* lti 'E P
,'-" Figure 62. Response ratio of MGL-6/MGL-9.-86
La -86',,I
6. SUMMARY
U Table 5 summarizes the main conclusions of the study. The first
two columns identify the sensors, and the third lists the minimum 3-dB
roll-off frequencies as estimated by the manufacturer. The remaining
columns contain the information obtained from our measurements. The
fourth column shows the measured 3-dB roll-off frequencies, and in each
instance the value exceeds that given by EG&G [2]. The next three
columns (Columns 5,6 and 7) relate to the angular behavior of the sensors
and list the frequencies at which the patterns deviate from the ideal
(cos e) dipole pattern fitted to angular data in the least squares sense.
The eighth column gives the frequencies at which the azimuthal (rotational)
symmetry breaks down and is applicable only for the MGL free space
sensors. The last column gives the deduced figures of merit as defined
by Baum et al,
'.:": e -"Z] I 2 - -- and 2 f h !0 ZA° I/
0.707 = eq 0.707 Z Aeq ]for electric and magnetic sensors respectively. The 0.707 subscript
indicates that the value is based on the cut-off frequency f of the
sensor (Column 4). In the formulae Zc is the sensor load impedance
(100 ohms) for both the free space and ground plane sensors, Zo is the
free space characteristic impedance, A is the equivalent area as giveneqin the EG&G catalog but doubled for the ground plane sensors, and c
is the velocity of light.
In essence Table 5 shows the frequencies up to which each sensor
. could be used. As already noted, the measured 3-dB roll-off frequency
exceeds the manufacturer's (minimum) specification in each case, but
-87-
CC
1mJ.
c N f- N 0M
IL N
- 0 *
LL
I. m 6n -- v; 0-,r v
LU i U% .
Lii ~ ~ ~ 0 0U' %L -
MJ. N NmC- NNLnN~
Ef, -5 -.
-c mA LAO
N . sU N ~ U -
14' N NNN N A
ca 0 ~~u 00 S. % A0
IL. .0 . v ?
M,. 'au. f"i -z N
ea'
vi w IvE -j4L1
(J0-88-
it is also substantially less than the frequencies derived from the
measured angular responses. This suggests that a sensor could, in fact,
be used with confidence at frequencies up to two or three times the
manufacturer's specification, provided the responses were corrected for
the frequency roll-off. In practice, the maximum frequency will depend
on the particular sensor used, and to see this, consider the data for
the MGL-2D(A). The measured 3-dB roll-off frequency is 482 MHz, but
the angular response frequencies are considerably larger. Suppose one
wishes to receive a signal over ±60 degrees range of angle. The maximum
frequency is then 932 MHz, and since this is less than the value in the
last column, the rotational symmetry would still exist. Nevertheless,
to use the sensor at frequencies up to 932 MHz it would be necessary to
correct for the frequency roll-off either with a specially designed
(analog) compensating network or by correcting the measured values when
processing the data.
The study was a classic example of one which is theoretically and
conceptually straightforward, but difficult to accomplish in practice.
The most severe difficulties were encountered with the free-space
sensors that use twinaxial cable systems. Such systems are not effective
above about 100 MHz, and generated oscillation that degraded the accuracy
of our measured data. The high (> 100 MHz) frequency performance of
the free-space sensors merits further study, and we remark that the
performance could be improved if the sensors were designed with twin
coaxial rather than twinaxial output lines.
We understand [10] that the latest versions of the MGL-2 and
MGL-6 ACD sensors have improved impedance matching and extended
frequency range, respectively.
-89-
7. REFERENCES
[1] Baum, C. E., E. L. Breen, J. G. Giles, J. O'Neill, and G. Sower,
"Sensors for electromagnetic pulse measurement both inside and
away from nuclear source regions," IEEE Trans. Antennas Propagat.,
•I .AP-26, 22-35, 1978.
[2) EG&G, Inc. Catalog, "Standard EMP Instrumentation," Albuquerque,
NM, 87106, 1979.
[3] Stratton, J. A., "Electromagnetic Theory," McGraw-Hill Book Co.,
Inc., New York, 1941.
[4] Baum, C. E. and E. Breen, personal communication, AFWL,
Albuquerque, NM, 87117, 1980.
[5] Senior, T.B.A. and J. Pond, "Pole extraction in the frequency
domain," Radiation Laboratory Final Report No. 017815-1-F,
AFWL Interaction Note 408, December 1981.
[6] Edge], W. R., "MGL-6 B-dot sensor development," EG&G Report No.
AL-1lOl, Albuquerque, NM, 87106, 1974.
[7) Edgel, W. R., "MGL-S7A B-dot sensor development," EG&G Report
No. Al-1104, Albuquerque, NM, 87106, 1974.
[8) Mory, R., P. Anderson, J. Kraemer, and C. Murphy, "Development
and production of multi-gap loop (MGL) Series EMP B-dot
sensors," AFWL-TR-70-153, February 1971.
[9] Olson, S. L. "MGL-S8(R) B-dot sensor development," EG&G Report
No. AL-1187, AFWL-TR-75-252, September 1975.
[10) Prather, W. D., Private communication, AFWL/NTAAT, January 1983.
-90-
APPENDIX A: ACD-4A(R) DIPOLE RESPONSE DATA (RAW)
Measured data are presented for the ACD-4 sensor as functions
of the frequency for the rotation angles e = -90(15)90 degrees
(see Fig. 1). The data are normalized to the measured values for
e = 0, producing unity plots for e = 0 (see Plot A8). Data for
e = ±85 degrees are also included, but were not used in the analyses
described in Section 5.4.
TABLE Al. LIST OF ACD-4 DIPOLE DATA PLOTS
e (degrees) Plot No. File No.
-90 Al AS 6501
-US A2 AS 6301
*. -75 A3 AS 6309
-60 A4 AS 6317
-45 AS AS 6325
-30 A6 AS 6333
-15 A7 AS 6341
0 A8 AS 6349
15 A9 AS 6357
30 A1O AS 6365
45 All AS 637360 A12 AS 6441
-A12
75 A13 AS 6449mA13
85 A14 AS 6457
90 A15 AS 6473
-91-
A1.... "A p C~;
1 II-
I I iZ
C U LI O~
Plot Al
-92-
ZLL
-93
.,,-_,_.., - t .. . . +.. . -. -. . -. .-._- . -. .- , . ,+.... ,- -,.- ...- . ,-.,./ ++ i+ + : .- ..
. L.--+r
'7m
Aii
4-I94
S- . .-' . - ..
..... ::I..+
!-94
-. . . . . . . .,
3
2
12 rr2aiu2
Im-
-200
Plot A4
-95-
PC D4-A-P 4--EC.-------rAC3
LLT-..4
a!u
FP %)',j .=M z)
U
fll
1J 2 1 m J.11
1 OW 2000 303l 4130
F EOUENCY ( MHz)
1:::' I"
Plot -5-96-
'°: Plot A5
• ' -96-
=..
r . * .r.* -
744-.
"- RCD-4R (R) , -32Cr., VR-VP; R 333
3
2
10 M'i 12 UM
0 I I
0 1000 2000 3000 4000 5000FREQUENCY (MHz)
10ACD-48 ( W, -30fEG, VR-V 1 iRSGI3
50
|U
-50
10 MRY 02 UM
0 1800 2080 2000 4000 500oFREQUENCY (MHz)
Plot A6
-97-
S *o..
-A :D4( .(P i SDEC.. VAi-VE/; AGG 4 1
-Jt
-PvEC 4 W
12 rPl 91 um~
FFR.uENC Y (Mi-z)
-- 98
- ~~ ~ P D.~;, OD G. VR--'-S RS6 4----7 '-.-..
0 1000 2000i 2000 4 0 05CEFPEQUENCt' (MHz)
RCD4A (P, , GDEC.,VR'.8 \ FIS-V 49
I:j
0 -LaJ
LT
12 rrcs a1 UPI
0 low0 203000 400 s--
FRE0UE(i*r (M~z)
Plot A8
-99-
13 FEB ni Uti
FREGiLIEN'CY IkMI-z)
PIC 04 (P I .V A-"VP;ASSS 57
-25
-5
8 00 2 000 3000 4000 5 3
Plot A9
-100-
4
24
FRE(U~t Ul(Mz)
RC04R), 30DG, VR-VR; RSc -5]
-50
13 rca 81 up
0 1000 2000 3000 4000
FREQL$Ef4Cr tMI-z
Plot A10
6
,:-101-
* V ,
Plot All
-102-
7.1
rr.
-103-
*~~~~~~~7 -r ,. r ij T- 7. --
"...."
-- ..°.. 1 LY ka
,- I-
.., F,,-
. , *-"
Plot A1 3
-104-
C.O
,% - , -t -
Plt A14 ,
- 1 05 "-*.4 j'"p ip4 VIV . ' i :'p
- ICC
- 1-105-
- * . - ... r- . - ... . " " " Y " " "* " ' ...........................- ,........-.-.............*...
Il E.: -.
4.j
-5'2 43IC
Plot A15
-106-
APPENDIX B: MGL-2D(A) AND MGL-6A(A) DIPOLE RESPONSE DATA (RAW)
Measured data are presented for the MGL-2 and MGL-6 sensors as
functions of frequency for the rotation angles a = 0(15)90 degrees
in the dipole response plane. For each sensor, data are shown for
gap No. 1 at = 45 and 90 degrees (see Fig. 25), and all data are
normalized to the measured values for e = 0.
TABLE BI. LIST OF MGL-2 DIPOLE DATA PLOTS
Sap No. 1 at :45' Gap No. 1 at o =9V°
e (denrees) Plot ;!o. File No. Plot N!o. File "o.
0 a1 MG 9165 "C 9357
15 B2 MG 9173 B9 A!G 965
30 B3 MG 9201 B10 MG 9373
45 B4 MG 9209 BI1 MG 9401
60 B5 MG 9217 B12 MG 9409
75 B6 MG 9225 B13 MG 9417
90 B7 MG 9233 B14 MG 9425
TABLE B2. LIST OF MGL-6 DIPOLE DATA PLOTS
Gap io. 1 at = 45" Gap No. 1 at = 90°
e (degrees) Plot No. File No. Plot No. File No.
0 B15 MG 9033 B22 MG 9109
15 B16 MG 9041 B23 MG 9117
30 B17 MG 9049 B24 MG 9125
45 B18 MG 9057 B25 MG 9133
60 B19 MG 9065 B26 MG 9141
75 B20 MG 9073 B27 MG 9149
90 B21 MG 9101 B28 MG 9157
-107-
U
NMGL- 2DtR J .D E'2LJR;NG9 165
- 0 a
"%,.
10
p4 -
14 CaIC 12 UPI
%'ZO
a 500 18800 158I00
FR OEC -,'z
I,.-
:'.:..Plo Blo t 2 ,,
-1008
14 O! LI
FREUE'v'-:z
.I! -Plot'81L 'G - ,O E',J7 rG .
!1""108-
h77
MGL-20 1 *I.DE&.34 i'
1.5
2 I-. aI P
se 500 ~ Un
-1109
- .
+ ''
M GI L** -2 0 .R D E, E I R 1;**2
2 JUN 9 1 UKi
0 .----
5 01000 1500
.r0o
o 8
~.5
-10
rFEI+"rJ:
Plot B
.ii+.110-
- . . ~ - .- -. - - - - - - - -
-JC'L1rJ:'' M z
u-i
3'UN 0 1 UPI
Ir 200
100 00
F*P.4UE 1-: -
Plot -B4*-
1.5
LL 1
2 J~JsUN Q; P
F, .,
tIH
-100
2JUN I Ut
Plot B5
-112-
I .
. ~ ~~~ML-20'(,A ,.-E5DE,;., JRi ,- 2..
a A].
-JJ
I ...J.'
.5.-
(*1* ....
*4 . -i
.].Jut! al uti
S o -0•00 .. 12
' Pl ot B6
100 .- -I13-
....
IE ..-
-rw "":" ' -" "-:.-" :." .."*~ ** ,7 -. .. W , C, '. r:°" - -" r r"."r-' " 'A i - U" -. • . .
M i;L - 2 D 4C i .I 9 o " E & .2 F, h;,;ll . "- _-
1. 5
s- DO
5* . *---. 3-.-JN g a
-0 iSQ
. ~F 0 0, 100 0 1 10.:,J _
FP E,'J IC r : r IHz200 ' ' ' G 3 R jv ; 1
PI o B7
.100 . . . . .i , .; ",\ '
4i " I'
- 0 I0 0 0!5 0:" ."."5*il;
lI~,-t 'fll
'"; -114-
HiL- E, -' R. j. ODEC..- w 7P -
1. JUEEl 1~
5 JUN*E 19 1Uj?-o -- s--
13 100 0 1 Sa Q
FPE0ULj'C r MHZ
-10115-
MiLJ0 A IIDI.J;
-4 JUNT G 1Liti
FPELI;.J A' E EM1.7V, .
101
-I Oxi
4 3JUN 21 Ut'
Plot B9
-116-
'4
**-2 ' A . . ~ l. .
5 OI I
ML-20(,A20D~i.TAr!,Li?
1 AiN91UP
-20
0- 1 e ) O
F.~ '~R* -AI- 1 r I
Plo B1
M4L-2OCi A''.45DLI2L. Y;; f;:MZ402I
LJ
00 11ac
Ica
-101
5 2JN I LI ?I
O s CO 1000 15S00CI0F7RQEN~r: r C Ill-
Plot __ -11
L-10
MGL-2 [If A 7, F4;HL4C9
Cl a
-100
-119-
D A -LF-L.! , lEI4
I DO
-100
~~CiQ~~ LCIOI) I~)E
r -1-120i
.'-
~~M,.;L-2 [I , 1 . 0 DE,. .7A, H.I L9-4 2 - c
l I IO.
JN
EL U"
-7
-AJ
C
H
x"ii f, iTI 0
., , y
" JUILNC'1 ui q
Fg[.,>lJ ('. -'2 0 f--t--
Plot B14
-121-
15
-o -1
".<.9
1 <-J
iCj
U)
1 JLJ. a, Uri ,
"- 0 lOL :000200 300,] 4000 "- -.'
a ac
Plot B15
-122-
IT,! 1 -
C-1
U..--.
I 3UN 01 1,,
20
-:- . .=
0i 1000 - 000 ~ 00 Q 40 -
Plot B16
-123-
I.°
- I
* ° ,
, • -
P-11 i D E 1 I JUN t 1,710
1.5
o
L...-.
C .L
CIL L2 A 1 C0
'3 E E O0 +r -LC'C~F'RKE--: PlHz
I
S-. .L/ ,
%'" -1
-- T~~~el ~ JU . I - - . . . L
" i F , : . ........ 2CA C0 7 J4 ¢' -.. _
• -'E .b~t's'- r lPI z
~PlotB17
-124
R- I *SE YA.r~c
1 .5
*l ri C- 1 0 ... 0Cl
I-
IUL in-
rRL..j1H. t .IH
Plot B18
-125-
TCA
Ut
-AJ
00
I ILurA Q6 ut.
Vp,,,~~' r%4z
PlotILn- B19E. 1
1006
I U 9 1021G 'J-. DC ri92
L. -G I 7SD :,. -. ;
-)Jr
CtL
W.4
- 1 JN GJ U1 j0 0002 lc 082 lcl( SOO 400C:
-1271
IG- -I
I I, N 1) t
1001
-~ IU 2Ut 1 Utlj
~~00 -
I II 0 *c ,.3c I0i"
IdI
~1 00 128-
PIGL-C.l A -1I. D LEC ]. I. ;
L.5
_____ _____ _ _-I DD ti. Ilip
100
L.J
-100j
0 11 2 000 2000 400 0
Plot B22
-129-
7J i
SF;RE0uEr-j: e *i uP'
2r3iI-46zut2 1) el 0
-p C L Er K M
Plo 823
-130
i I C
w'.".
100
LJJ
F;: E ,
-1130
22PR.9 LPIOC OI34OFi: n~: ~
*1L a iJ-I J.DE.,.rA "".;. ."
4 Plot B2
-131
rI1GL-Rt Ai -I. 45flEG.;;rZil
1.5I. . 1
-100
22 FIR. 6: LI
I C1007 2 CI00 S1000 4t000
FF-E.LIjEIfl:Y OlI M
Plo0B2
10032-
r1.L',LH-I.60EDE. J. rH;M.4I
I-a
FREO2LIEr'r MIHz
-1130
I.T7
F-~EU~i: ~ MHz 4JC l~
...................
39 Pig. 91 um
* ~~~- ZOO--- --
C. 000 ~ 2 0 0t320FRE.'IJE+-r Hz
Plot B27
-134-
-- ~ 7. 7. -G -J -.
I- It
,_.5
I l;--h i- r
1000
I5
-,
Plot B28
-135-
"°a"
'
APPENDIX C: MGL-2D(A) AND MGL-6A(A) ROTATIONAL RESPONSE DATA (RAW)
The data presented here were obtained for 7 different angles of
rotation of the sensors about their axes, viz € = 0(15)90 degrees.
The angle p is measured from gap No. 2, implying that for o = 0 the
exciting signal is incident on this gap. The data are normalized to
the measured values for 0 = 45 degrees, corresponding to incidence
midway between the two gaps. Because of the symmetry of the sensor
designs, data were recorded over a single quadrant only. Ideally,
the responses should be independent of €, implying unity plots, but
* the results show that this is not the case at high frequencies.
TABLE CL. LIST OF MGL-2 AND MGL-6 ROTATIONAL DATA PLOTS
MGL-2 MIGL-6
(degrees) Plot No. File No. Plot No. File No.
0 Cl MG 9509 C8 MG 9601
15 C2 MG 9517 C9 MG 9609
30 C3 MG 9525 ClO MG 9617
45 C4 MG 9541 Cll MG 9625
60 C5 MG 9557 C12 MG 9633
75 C6 MG 9565 C13 MG 9641
90 C7 MG 9573 C14 MG 9649
-136-
: --;" '' . . . - , ,m ' , - ".
I I1L-2D if Fi. OIDE?.. JR t,9 ,--
S E Zu-4C10
IO
-:-OctH I - -
I~ ~ ~ -K 3 :
Plot C1
-10137-
- --- - - - -
1.5
-J~
2 FES a U
Fr. C.,1 I ,:' Hz
plo C
H 138- ~ rL};
IAim
3 FE -- UI
00 CD~ 15i 00 2 .21
3 FEE I UHif
1200
1 10001C 3ii2 O
Plot C3
-139-
'2L-:[ ). SD ,; R l;'
2Ji C P - 1
F~RKIJ'Utl-C Y 'Hzl
-1140
1
" i
FREO".E-CY MHz
lMt 1;L- .. IA p
L.5
-. E - -I:Y ,.H~
• 2~-00 '1i~
rtl
So 1C SOOU. rPLcUErF?! rIH
Plot C5
%':, -141-
,,4
,4
NMGL-2D1Fi .75DEC. F NR.q- 95
14 OCT 82W1
IMGL-2 0 A 5 DEi. IPIILS 5
3r
-100-
14 Oc r2 fe
Plot C6
'7-142-
p 1.5
M 1. L 1.'. C.0 D i i;.TR 0!
I OL'
LiA
C100
3 I 9, up
-~0-------- -- 1C
ilHz
Plot C7
-143-
- ~ ~ ~ ~ ~ G -~ A) -I. DDC. Jn; ..- 7-.- -
DD,; 1 j~qc.96D
I n-I
14 CCT 02 Ufl
a3 1000' 2 Oct 0 l0c00CFREQUENCY fflHz -
~~~t~~~~~cP -_ _ _ _ __ _ _ _ __ _ _ _ _ __ _ _ __8_ _ _ _
-144- ~ - D~.J, L6
1.71
4
101
1030L0
FREi;-UENC> (MHzI
Plot C9
-145-
-.7
.5
1I SiU a ] LJm
Z, 00
20M.;L-GAL P i- I ?DE,;. 7P: .~417
100
011)030 2000.100
Plot ClO
-146-
V7W'E' -F4 F11; -25-
IL!
.5
t, MHz
U 2003
100
'°i
LAJ
-100
IA IUU SX UM
, FR UEN':Y ,AiI44z,~ a 4~c
Plot Cl I
-147-
. . .. - . .. , , . . - . , ,
MC.L-GI At A I 6 :!DE,. 3I ;
IIc
2100 -
1 clO aI- '
Plot C12
-148-
....................................
iGL-G4 1 L-i, R "5 DEC TA NGGS-. 4 1
Z/I-I
1 °1
14 OCT 92 UI
00- 1000 21300 C 0 4000
I PI~~GL-G A Aj -i, 75DE,; R;NGSS@41
-A
14 OCT 12 UPI
0 1000 200 a0O0 4000FPCOUEN," Yr M Hz,
Plot C13
-149-
ILI
4i-
0 10i00 2000 aQ 4000i
-1150
'41'4. e-'4 y
Il
"Aw,
' 4V4
k4 'k
IRV
qy- V4$