Ξemail: tholt@apelc.com

A MARX GENERATOR DRIVEN IMPULSE RADIATING ANTENNA

T. A. HoltΞ, M. G. Mayes, M. B. Lara, J. R. Mayes Applied Physical Electronics L C , PO Box 341149

Austin, TX, USA Abstract APELC has developed an Impulse Radiating Antenna (IRA) that consists of a TEM-horn-fed parabolic reflector that is directly driven by a 22-J, 400-kV Marx generator. The system is based on standard Marx generator designs offered by APELC. The Marx generator output couples directly to the TEM horn via a transition from a coaxial geometry that approximates a standard coaxial-to-parallel plate transition. Primary design considerations that facilitate achievement of high instantaneous radiated power include appropriate Marx generator rise time, transition design, and TEM horn focal point positioning. Data collected over the course of the system design is presented.

I. BACKGROUND Impulse Radiating Antennas (IRAs) have been extensively tested and reported on in open literature [1],[2]. Systems with direct feed schemes similar to that proposed herein, however, have not been reported

extensively. The advantage of the APELC IRA system is that the Marx generator driving the IRA is directly connected to the RF feed which eliminates the need for a low-impedance cable between the Marx and RF feed and also reduces the need for a complicated impedance-matching balun. The disadvantage of the proposed system is the aperture blockage caused by the Marx and Marx support structure.

II. SYSTEM DESIGN

The APELC impulse radiating antenna, shown in Figure 1, is comprised of a MG10-3C-940PF Marx generator, a 6-ft parabolic reflector, a custom built support frame, a transparent TEM horn, and a TEM horn feed section. The Marx generator directly drives an inline balun in a configuration referred to as an unbalanced feed which then flares out to form a TEM horn that terminates on the perimeter of the parabolic reflector.

(a)

(b)

Figure 1. (a) Major components of the APELC 6-ft IRA. (b) Major components comprising the RF transition

section, the parabolic reflector, and the Marx generator mounting hardware.

A. Marx Generator The Marx generator used in the system is a slight modification of the APELC MG15-3C-940PF, reported in [3], the only difference being that five stages were removed from the generator resulting in a lower impedance, 10-stage, Marx generator capable of a peak erected voltage of 400 kV. The electrical specifications of the MG10-3C-940PF used as the source for the IRA system are listed in Table 1. The stages that were removed allowed for the accommodation of a peaking circuit that was placed in-line between the Marx generator output and the TEM horn feed section.

Table 1. Electrical Specifications of the MG10-3C-940PF.

Parameter Value units

Number of stages 10 - Peak erected voltage 400 kV

Maximum charge voltage 40 kV Maximum energy per pulse 22 J

Capacitance per stage 2.82 nF Erected capacitance 282 pF Series inductance 250 nH Source impedance 30 Ohms

B. RF Specifications An inline balun is used to gradually transition from the coaxial Marx generator geometry to the parallel plate feed required for the TEM horn that sources the parabolic reflector. The gradual taper is a necessity in order to reduce geometric discontinuities that can cause impedance mismatches and consequent reflections of propagating waves. Abrupt geometric discontinuities ultimately lead to an inefficient RF system.

A traditional TEM horn could not be used to source the parabolic dish along the axis of the parabolic reflector due to the large aperture area (and subsequent aperture blockage) required to radiate the frequencies of interest. A semi-transparent TEM horn was used to source the reflector and was constructed using ¼” brass rods to guide the propagating wave onto the reflector.

The parabolic reflector has a diameter of 180 cm and a focal distance of 68.2 cm yielding an F/D ratio of 0.38. Due to the amorphous nature of the transition from the coaxial Marx generator geometry to the semi-transparent TEM horn, a precise definition of the phase center of the source was difficult. Consequently, the optimal position of the Marx with respect to the parabolic dish was determined by experiment and simulation rather than by analysis. Experiments with the system demonstrated that very small deviations in the position of the Marx with respect to the parabolic reflector significantly affected the radiated field amplitude and pulse shape. Deviations as small as ± 0.200” from the optimal location resulted in a

p

t2

tM

n

rr

w

bt

A

10% decrease 0.500” resultedfield. Similar r

C. System SpThe system

approximatelyportable and sa source of comthe Marx gene250 psi.

As mentiontransmitted byMarx output plimit the highlower limit wdiameter, whenot propagatapproaches threflector desigreflector diamfrequency tranwhich was sign

The system bursts. Subseqthe IRA sysexceeding 200Additional sys

A conservastrength can besimplifying asapproximatelyOhm load anconservative eof 1 meter can

Table 2.

Pa

MaximumMaximumMaximum

Focal Length Diameter of

Far fie

from peak radid in a 13% deresults were fou

pecifications m’s maximumy 10 ft. X 6 ft.ingle-man opempressed breaterator as an insu

ned before, ty the system ispulse rise time.hest frequencywould be set re frequencieste well beche reflector dgn guidelines meter at 3 wnsmitted or rnificantly stret

is capable of quent revisions stem have de0 Hz for burststem parameter

ative estimatee made using tssumptions. Asy 200 kV (i.e. nd a system estimate of 490 be derived as

APELC

arameter

m system heighm system widtm system lengtof parabolic re

f parabolic refleeld of system

iated E-field. Aecrease from pund via simula

m volumetric. X 7.75 ft anderational. The sthable dry air wulating gas at p

the frequency inversely prop The realized r

y content radiby the para

lower than 10cause the hdiameter. Stan

actually suggwavelengths received (a detched for the A

50 Hz operatto the Marx ge

emonstrated rt lengths of sers are given in

e of radiated the radar equatssuming the M

800 MW) coantenna gain

0 kV/m normalshown in Equa

C IRA system

Va

ht 9th 7th 12

eflector 6ector 1

A deviation of ±eak radiated E

ation.

c footprint id it is two-masystem requirewhich is used ipressures below

y of radiatioportional to thrise time wouliated while thabolic reflecto00 MHz shoulhalf-wavelengtndard paraboligest sizing thof the lowesesign guidelin

APELC IRA).

ion for 10 shoeometry used irepetition rateeveral secondsTable 2.

electric fieltion and severa

Marx voltage oouples to a 50

of 10 dB, lized to a rangation 1.

(1)

parameters.

alue units

92 in 72 in 20 in 82 mm80 cm 5 m

± E-

is an es in w

n he ld he or ld th ic he st

ne

ot in es s.

ld al of 0-a

ge

I

A. Preli The peprior to itfitted to tone-way tand the lowere testegeometriethe insulasystem reamplitude2.

A cabpeaking ctransitioncable (RGpurpose sseal betwenvironmdesign coinches inMarx ouConsequethe pulse shown inmore grasection. Pto incorpoin place.

Figure

B. Simu Severalincludingparabolic

III. SYSTP

iminary Expererformance of ts integration ithe output of transit time isooad. Several died to reduce thes tested used hating gas. The educed the Me to 80% from

ble adapter scircuit and ser

n from the coaxG 220) used forserved by the cween the Marxment. The mech

ontradict eachn length) highutput and the ently it is impo

applied to the Figure 2 beca

adual impedanPlans to re-engorate diagnosti

e 2. Exafrom testi

ulations of the l simulations a

g an excursion reflector to d

TEM DESIPERFORMAriments

the Marx geninto the systemthe generator

olation betweenifferent peakin

he Marx generahigh-pressure geometry impl

Marx rise time 2 ns to 520 ps

ection immedrves two purpoxial Marx genr transit time i

cable adapter isx pressure veshanical and eleh other resultih impedance s

beginning ofossible to infere RF transitionause the RF tr

nce profile thaineer the cableics on the RF t

ample Marx geing with a cable

System assisted the dof source pos

etermine the d

GN AND ANCE

nerator was vem. A cable loa

and provided n the Marx genng circuit geomator rise time, adry breathablelemented in thfrom 20% of

s, as shown in F

diately followoses. The firsterator to the csolation. The ss to create a prssel and the octrical aspects ing in a smasection betweef the coaxial r that the rise tin is the same aransition has a an the cable ae adapter sectiotransition secti

enerator wavefoe load.

esign of the ssition relative dish focal poin

erified ad was

30-ns nerator metries and all e air as e final f peak Figure

ws the t is to

coaxial second ressure outside

of the all (~4 en the cable.

ime of as that much

adapter on and ion are

form

system to the

nt. The

FM

p

bw

p

t

r

abridged resulFigure 3. ThMicrowave SSolidWorks. Tsystem gain hposition as wafrom standardbehavior betwwere similarcomputational precisely moConsequently,the parabolic optimal distanc

Figure 3. Marx to par

frequencies odistance be

Following treflector sepa

Figure 4. A side on view

lts of a focal phe simulationsStudio using The simulatiohad the same as observed in ed antenna th

ween simulatio, the simula

requirementodel the ph the optimal lodish varied

ce of 46 inches

The depenrabolic reflectoof interest. Theetween the sour

the determinatiaration distanc

(a) A top dw of a near fiel

point excursions were perfosolid models

on results demsensitivity to experiment andheory. While on and experation space ts were too hysical hardw

ocation of the Mby several ins found through

ndence of relator on system ga legend indicatrce and the par

ion of the optce, simulation

down view of ald cross-section

n are shown iormed in CSTs generated imonstrated tharelative sourc

d as is expectethe observe

rimental resultand resultinextensive t

ware createdMarx relative tnches from thh simulation.

tive position ofain at several tes the relativerabolic dish.

timal source tns of the RF

a near field cron of the IRA sy

field gain fro

in T in at ce ed ed ts

ng to d. to he

f

to F

transitionparabolic sourced b15 ns fallallow theresults ofdown viedivergingconformageometrysphericallbecause illuminatishows a caused bresults in on view ilinear scanote the sline transi

C. Low TDR asystem byconstant indicated (between arms) incgradual imthe 33-Oh377 Ohmvariation for the ne

ss-section of thystem excludinom the IRA sys

n region, the sereflector were

by a unipolar il time, and a te simulated sf the simulationew shown in Fg (i.e. convex) al to the sphey of the parabolly diverging wthe RF transion distance. Tsomewhat br

by an asymmebehavior as a

in Figure 4 (cale and showsside lobe emergition.

Voltage Diagnand VNA meay replacing th

impedance that the impthe Marx gen

creased graduampedance variahm source imp

ms, it is unknois optimal. A

ext revision of t

he IRA systemng the Marx genstem.

emi-transparente performed. Thimpulse with aotal excitation

system to reacns are shown iFigure 4 (a) rwavefront whierically convelic reflector. T

wave with the rsition is locatThe side on vroken sphericaetrical unipola

a half TEM fee) illustrates tha 15 deg mai

ging from the c

nostics surements wer

he Marx generfeed. The Tpedance in thnerator output ally from 60 toation is desirab

pedance to freeown at this timAdditional simu

the APELC IR

m excluding thenerator. (c) Sid

t TEM horn, ahe RF transitioa 600 ps rise tn interval of 40ch quiescencein Figure 4. Threveals a spherich is approxim

erging (or conhe conformity reflector is expted at the op

view in Figureal wavefront, ar excitation ed antenna. Thhe far field gaiin lobe skew. coax-to-transm

re conducted orator with a coTDR measurehe transition r

and the TEMo 90 Ohms. Wble in order to e space impedame if the impeulations are pl

RA.

e Marx generatode on view of t

and the on was ime, a

0 ns to e. The he top rically mately ncave) of the

pected ptimal 4 (b) likely which

he side in in a Also

mission

on the oaxial, ements region

M horn While a

match ance of edance lanned

or. (b) the far

ppwM

D

Figu

VNA measu360 MHz, 42somewhat copresented in present betwewas also founMHz and 220

D. RF MeasuThe entire s

diagnostics angenerator. RF

ure 5.

urements indic20 MHz, and orroborated bthe subseque

een the aforemnd that frequenMHz were not

urements ystem was testnd the cable measurements

Figur

Experimentally

cated strong rej500 MHz, all

by the RF ent section. Pmentioned freqncy componentt rejected.

ted following tload testing

s were conduc

re 6. Sp

y measured rad

jections arounl of which ar

measurementPassbands werquencies and ts between 15

the low-voltagof the Mar

cted on a 10-m

pectral content

diated electric

nd re ts re it 0

ge rx m

antenna rhorn (mod

The sysoptimal experimenand indicapproximof 1 metshown infrequency

of the radiated

field normalize

range using an del SAS-570) a

stem was testepeaking sw

ntally measurecates a peak ra

mately 220 kV/er. The spectr

n Figure 6 as ay of 210 MHz a

d signal shown

ed to a range o

A. H. Systemas the receivin

d at a charge vwitch pressured waveform isadiated electric/m when normral content of a normalized Fand a 50% ban

in Figure 5.

of 1 meter.

s double ridgeng antenna.

voltage of 40 kre. The ress shown in Figc field amplitu

malized to a dithe radiated s

FFT, shows a ndwidth of 80 M

guide

kV and sulting gure 5, ude of istance signal, center

MHz.

wF

tb

w

rr

Figure 7. Radiating

measured fieldmax

The antennawas measured Figure 8. Asantenna, the bthe axis of beamwidth ofexpected fromcould possiblwavelength ofis in the neighreflector, whicreflector to illumination ocould also caustudy is planne

Figure 8.

MeasurDish. The patt

d at each angulximum measure

a pattern of theabout the hor

s is typical wbeam is narrowthe parabolic

f 20° is perhm a system witly be explainf the center frehborhood of tch might advecollimate thef the reflector use the broaded prior to the n

The orientrespect to

red antenna pattern given is a lar location (in ed field in deci

e system, showrizontal plane awith any paraw and tightly c reflector. Thaps greater tth a parabolicned by the

equency of the the diameter orsely affect the beam. Theas shown thro

d antenna pattenext system re

tation of the fiethe system.

ttern for 6-ft ratio of the degrees) to th

ibels.

wn in Figure 7as illustrated iabolic reflectoconfined alon

The half-powethan would b reflector. Thifact that thradiated signa

of the parabolihe ability of the non-uniformough simulatioern. Additiona

evision.

eld pattern with

e

7, in or ng er be is

he al ic he m n al

h

APELCradiated enormalizeMHz. Thenearly haThe systsection totransparenfirst revisscheme ebelow thAdditionageneratorunderway

[1] I.D. Sa Parfor InternTech64.

[2] E.G. AnteWide131 -

[3] C. NRepeconfe

IV. CC has developeelectric field sted range of 1 e peak field str

alf of the valuetem uses a co mate a coant TEM horn dsion demonstraeven though thhe maximum pal developmentr peaking switcy.

V.Smith et al., "Draboloidal RefImpulse-like national Pulse

hnical Papers,

Farr et al., “nnas: Theoryeband Short-Pu- 144. Nunnally, et etition GW erence.

CONCLUSIed a Marx drirengths in excem and a cent

rength producee predicted by

coaxial-to-paralaxial Marx gedriven in an unated the feasibhe radiated fiepredicted by t on the feed sch are either p

REFERENDesign, Fabricaflector Antenna

Waveforms",ed Power ConVolume 1, 3-

“Multifunctiony and Experulse Electroma

al., "CompMarx Genera

IONS iven IRA capaess of 200 kV/er frequency o

ed by the systemy the radar equllel plate tranenerator to a nbalanced modility of a direceld strength wthe radar equcheme and theplanned or pre

NCES ation, and Testa and Pulser S, in Tenth nference, Dig6 Jul 1995, p

n Impulse Radriment”, in agnetics 4, 199

pact 200-Hz ator System,"

able of /m at a of 210 m was uation. nsition

semi-de. The ct feed

was far uation. e Marx esently

ting of System

IEEE est of p.56 -

diating Ultra-

98, pp.

Pulse " this