162 IEEE TRISSACTIONS ON SSTENNAS .*XD PROPAGATION, VOL. AP-17, NO. 2, JIARCH 1969

Large 22-MHz Array for Radio Astronomy CA4RXAN H. COSTAIN, J. DAYID LAIC,EY, MEMBER, IEEE, ASD ROBERT S. ROGER

Abstract-The relative merits of T- and cross-type radio tele- scopes are discussed. A large array used for radio astronomical studies a t a frequency of 22.25 MHz (X = 13.5 m) is described. It is built in the form of a T with dimensions 96h X 2.5h east-west and 32.5X X 4h north-south. The 624 full-wave dipoles are mounted X / 8 above a 65 000 m? reflecting screen. The instrument has a pencil-beam response of 1.1 X 1.7 degrees at the zenith. A time- sharing technique is used to provide simultaneous observations of 5 adjacent declinations spaced at quarter-beamwidth intervals.

Observations, commenced in 1965, should provide flux density measures for 400-500 radio sources down to a limiting flux density of about 30 x 10-26 W .m-?Hr*. A map of the galactic background radiation from declination +90 to -20 degrees is being prepared.

I I. IXTRODTXT~OX

SSTRUMESTAL development in radio nstrouomy has been directed primarily toward the achievement of

systems of high resolving power. With t.he great improve- ments in high-frequency receivers and antennas. the t,rend a t most. observatories has been toward the use of higher and higher frequencies where a.dequate resolving power may be obtained with st,ruct.ures of moderate size. Apart. from t.he pioneer work of Shain, Komesaroff, and Higgins [l]. [a]. there has until recent,ly been a relative neglect, of the low- frequency end of the radio astronomy spectrum.

Observations at. low frequencies are, of course, required to extend our knowledge of the radio spect,ra of all sources of radio emission. Some st,udies however, such as the exam- ination of the complex st.ructure of the galact.ic cont,inuum radiation, are best carried out at the longer wavelengt.hs where the intensit.ies are very high. The lowfrequency radio telescope is the most sensitive detect,or of t.he diffuse clouds of ionized hydrogen which appear in absorption against the bright, galactic background (see Fig. 11).

When t.he present, instrument was planned, the largest low-frequency army in the northern henlisphere was t.he 38-XHz “moving-T” at the Mullard Radio Astronomy Observatory, Cambridge, England [3]. An operat,ing fre- quency near 20 MHz x m therefore desira.ble to extend spectral measurements about an oct,ave. The choice of 22.25 MHz (X = 13.5 m) was made after a careful study of the incidence of radio interference in nearby bands.

Bec.ause of the almost complete absence of previous measurements a.t. this frequency, i t was decided to build a general purpose telescope similar to the JIiUs cross [4]. In t.hese systems, the voltage outputs of tmo long thin orthog- onal arrays are multiplied together to produce the desired “pencil-beam” response. The 22-3fHz instrument is, in

Contributions are from the Dominion Observatory, KO. 255.

tory, Penticton, B. C., Canada.

Manuscript received July 26, 1968; revised November 4, 1968.

The authors are with the Dominion Radio Ast.rophysica.l Observa-

fact., a T-shaped array. The 96X x 2.5X east-west. arm and the 3 2 . 3 X 4X north-south arm combine to provide a response of 1.1 X 1.7 degrees at the zenith.

Several large interferomet.er systems, as opposed to pencil-beam instruments, are now being used to study small diameter radio sources at low frequencies: the 26.3-XHz compound grat.ing interferomet.er a t Clarke Lake, Calif. [5] and the broadband arrays (1&25 MHz, 20-40 MHz) in the U.S.S.R. [GI, [i]. The extremely favorable obserting conditione encountered during early observations with parts of the 22-XHz array prompted the decision, late in 1963, to build a similar instrument to operate a t a fre- quency near 10 MHz [8]. These arrays at the Dominion Radio Astrophysical Observatory, Penticton, B.C., Can- a,da, are, a.t present, unique facilit.ies in this frequency range in that t.hey also permit observations of the es- tended features of t.he galaxy.

This paper is devoted to a discussion of T-sha.ped antenna systems a.nd a det,ailed description of the design and performance of the 22.25-3.IHz array. Brief descrip- t.ions of the instrument. have appea,red elsewhere [9]; [lo].

11. T COXFIGURATION

The equivalence of “perfect?’ cross and T systems has been discussed in debail elsewhere [11]-[13]. It can be easily demonstrated by reference to Fig. 1 (a).

Since the out,put of a T or cross is the product. of the outputs of the two ort,hogonal arrays, t,he only relevant spacial components are those between the elements of one array and t.hose of t.he ot,her. It may be seen that all ele- ment spacings and orientations present in the uniform aperture are contained in the cross and T, with each com- ponent a.ppearing twice in the cross configuration [cf. t4he arrows in Fig. l(a) 1. The L configuration, where a further arm of the cross has been removed, does not contain all necessary element. orientations. With a.n appropriate weighting of t.he elements, t,he cross or T can precisely duplicate the response beam of t,he uniform a.pert,ure. Hon-- ever, rat,her tha.n discard half of one arm of the cross n-it,h the result,ant loss in sensitivity, the alterna.t.ive arra,nge- ment, shown in Fig. l(b), where t.he elements of the south arm are placed to provide a nort,h-south arm of tvrice the widt.11, may be used and is identical in both resolving power and sensitivity to the cross. It is t.hk T configuration which mill be considered in subsequent discussion.

Both syst,ems are, of course, unfilled apertures and as such are less sensitive than t,he uniform aperture of equiv- alent, resolving power. Ryle and Hewish [12] have ex- pressed this effect in t e r m of an efficiency paramet.er E, which is the ratio of the effective collecting area of the system d4efi to that of t,he equivalent uniform aperture A.

COST-US et a/ . : L-IRGE 2 ' 7 - u ~ ~ A R R . ~ Y FOR RADIO ASTROSONY

(bi Fig. 1. (,ai Aperatures of equal resolving poxer. (b) T configura- Fig. 2. Declination polar diagram of T configuration xsrith dif-

tion equivalent to the cros configuration in both resolving power ferential phase errors of 0 degreez;, indicated by solid h e : and and sensitivity. 20 degree.*, indicated by dashed line.

For the T configuration shown in Fig. l (b) , t,his is given by

where T , is t.he antenna t.emperat>ure (neglecting losses) obt,ained when t.he system is used to observe a region of uniform brightness Tb. The major design paramet.ers, therefore, for a given resolution, are t.he array widths dl and cl, with the optimum values being those which ensure that. t.he system has sufficient sensit,ivity to reach t.he confusion limit, for point, source observations.

Most, practical syst,enls employing the T configuration have used it, as an interim phase in the development of a cross syst,em [15] or have used the apert,ure synt,hesis technique 1111: [3], [14]. Discussion has centered largely on the sensit.ivity of the T to errors in the excitations of t.he array elements [13]. However, the T offers some unique advanhges over t.he cross which merit serious considera- tion. The T configura,tion of Fig. l(b) is superior to t.he cross in three respects.

1) The number of phasing switches! mult,ibeam ele- ments, preamplifielx! cables, etc. needed for t,he north- sout,h arm is reduced by half.

2 ) The extent. in hour angle over which the sidelobes of the east-west array are significant is reduced by half because of the inerased x7idt.h of the north-s0ut.h array.

3) The maximum north-south dlmension required at. the antenna. sit.e, for a given beamwidth, is also reduced by half.

Perhaps t.he major adrant.age of the T configuration lies in the economies that result from halving t.he feeder and phasing arrangements of the north-south army. As men- tioned previously, for reasons of sensit,ivity, the area of the nort,h-south array and, t,herefore, the cost of the basic struct.ure are likely to be similar in the T and cross systenx. The phasing and multibeam arrangements neces- sary to provide beam scanning in declination, however, remain a major cost factor and perhaps the most difficult engineering problem in t,he design of large arrays. For any large imtmment. it seems likely that a monitor system

to maintain the accuracy of the excit.at.ions of the north- sout,h array call be illst,alled for a small fmction of the cost. of duplicating the entire north-south feeder system.

The sidelobes of unfilled apertures such as t>he T or cross are always troublesome bemuse these instruments are used to examine a universe populated with many bright. radio sources. The sidelobes in hour angle are reduced to negligi- ble values at angles beyond the first, zero of the east-west responx of the n0rt.h-south array. The reduced angular extent of the hour angle sidelobes mitigates, to a large extent, the dightl; larger declination sidelobes discussed in the following.

The re.<:pect in which t.he T is inferior to t.he cross is in its greater sensitivity to errors in the excitations of t.he ele- ments of the north-sout,h array. These errors produce spurious sidelobe responses in declinat,ion and ma,y be divided into two classes9 random and systemat.ic, whose effects are considered sepa.rately.

Errors in measurement, in the adjustment. of the excita- tions will result in ra.ndom errors in a.mplitude and phase. I n general. rhese errors will have a mean value of zero and rnx values dependent. to a large extent on t.he time and effort. expended to reduce them. For a given rnls error, t,he T sy tem would be expected to have a declination side- lobe level \'? larger simply because only half the number of independent e1ement.e a.re used [4]. -1 more serious effect results from a systematic differ-

ent.ial phase shift between t,he nort>h-south and east.-west arrays. This is illustrated in Fig. 2 which shows the asymmetric sidelobes and t,he 0.14 beamwidth collima- tion error clue to a gross different,ial phase error of 20 degree?. _%voidance of t.hk type of error requires some sort of system to monitor the different.ia1 phase. This is easily accomplished by t.he injection of a test signal into the overlap region conmon to both arrays.

111. CENTRAL OVERLAP REGIOS One of the design problems in T a.nd cross systems is

encountered in the region where t,he two arms intersect. I n t.he original llills cross [4], the central dipoles of the east-west. arm were removed. This results in a negative

164 IEEE TRANSACTIONS ON ANTENNAS AND PROPAG.~TION, MARCH 1969

system response, elongated in t.he east-west. plane. A por- tion of the output of the north-south a.rm, which had a. B E I M D T H

similar though far from identical response, was added as I?I x 1 3

compensation. Although t,his met.hod worked well for a REFLECTOR uIE*

uniform background or point sources, it. was subject to 65000 m'

large errors when the system was used to observe inter- L mediate broad components such as those encount,ered near the galactic plane.

2 . 5 4 ~

f c ~ ~ ~ ~ ~~~~A. 1~~ ~ .~ II ~. . - ~ ~

-

A different t,echnique, which offers a much more satis- factory solution, positmiom the elements in the overlap region so that. they may be considered to be part of either array. Their outputs are divided with hybrid networks and are fed, with t,he appropriate weighting, into the feeder systems of both arrays. The output of the system then cont.ains spacial Fourier components down to zero spacing and, therefore, precisely reproduces :I pencil-beam re- sponse.

This method \-cas used with the 1031Hz and ??-MHz array:: a t the Dominion Radio Xstrophysicnl Observatory.

A . General Descriptiw

The 2.2.?5->1Hz array is located near Penticton, B.C., Canada, at the Dominion Radio Alstrophgsicnl Observatory (longitude = 119"37'08"W, latitude = 49"19'14"S). It is a T-shaped array of 624 full-wtve dipoles; 368 in the east- west arm. 240 in the nortll-south arm, and 16 common to both arms. The dipoles are polarized in t,he east.-west direction. -1 full reflecting screen (xrea .- 63 000 m2) is mounted h/8 below t.he dipoles, and t.he entire structure is supported on a grid of 1698 wooden poles. The plane of the array was adjust.ed for t,he best. average fit to the terrain. It. was, therefore, necessary to skew the axis of the array slightly from the east-west direction so t.hat the axial plane of the east-west arm passes through the north celest,ial pole. The dimensions of t.he system are shown in Fig. 3, and Fig. 4 shows the array from t.he east. -At. t.he zenith, the instrument has an elliptical beam with half- power widths of 1.1 degrees in hour angle and 1.7 degrees in declination.

The outputs of the east-west and north-s0ut.h arms a.re carried by coaxial ca.bla to preamplifiers located at. t,he center of each arm and from t,here to t,he main observatory building. The signals are then amplified and correlat.ed using t,he standard phase-sm-it.ching technique. They are then integrated, digitized, and punched on cards.

The array ca.n be phased n0rt.h and south of the zenith along the meridian. This phasing to any declination is accomplished by means of remotely controlled switches operat.ed from the main observatory building. X ra.pid- phasing network, operating on a time-sharing basis, pro- vides quasi-simultaneom observat,ions of five adjacent, declinat,ions.

-At 22 MHz, the mean sky brightness temperat.ure of the order of 30 000°K completely donlimtes the receiver noise. Losses of 10 dB ma,y be tolerated wit.h negligible deteriora- tion in signal-to-noise ratio, permitt.ing the use of inex-

Fig. 3. 39.25-11Hz ana?-.

Fig. 4. 22.25-NHz array from the ea-t.

pensive coaxial a b l e for the bulk of the feeder system and the use of resistive attenuatow for the grading of t.he aperture.

The array has an efficiency, as defined by (l), of approxi- mat,ely 10 percent., a value considerably higher than t,hat usually employed [I], [4]. The effect.ire collecting area of the inst,rument is therefore

This efficiency permits observation of areas of low surface bright.ness away from the galactic plane. The collect,ing area is more than adequate for the det,ection of sources with flux densities near t,ke confusion limit of S = 33 X 10-266\T. n1r2Hz-' (1 source per 20 beam areas). Under ideal observing conditions, the system is capable of provid- ing flux density measures for approximately 600 radio sources.

B. Basic Eleme,lt The basic element. in the array consists of tn-o full-wave

dipoles h/8 above the reflecting screen (see Fig. 5 ) . A broadband dipole (impedance 2000 ohms resistivej

was constructed from two loops of wire. d vertical twin line couples the dipole to the feeders below the reflecting screen. A h/4 short-circuited stub a,nd a X 4 transformer permit adjust,ment of t,he impedance to 400 OhIli~. Sections

COST.IIX ef a [ . : L.\RGE '22-NHZ ARR.%Y FOR L I D 1 0 .ASTRONOUT 165

Fig.

1 b- WOODEN POLES -

E - W ARRAY - N - S ARRAY

x w -

0 " ' " " " ' " * * OX 16X 32X 48X

ELEMENT DISTANCE

Fig. 6. Excitation function for eaxt-rret iE-11-j and north-south iN-S) arrays.

Fig. '7. Feeder system of nort,h-south array. *-main phazing switch=, 0-rapid phasing switches, X-grading abtenuatorsj.

of 400-ohm line couple the t-mo adjacent dipoles, a.nd the resulta.nt, impedance of 200 ohms is converted to 50 ohnls n-it.h a coaxial balun transformer. Slightly different. match- ing transformers were required for t.he dipoles in the in- terior and exterior rows of each array in order to com- pensat.e for the different mutual impedance effects.

C'. Rejecting Screen

The ground plane of t.he array is accurately defined by a full reflecting screen which extends A/2 beyond the dipoles in t,he north-south direction. It consists of fine galvanized steel wires aligned parallel t.0 the dipoles a.nd spaced h/22 apa,rt.. The area of the reflect,ing screen is approximately 65 000 m2.

D. East-JVest A nay

In the easbw-mt, arm, t.hc elements are a.rranged in 4 rows. Ea.ch row has 3 groups of 32 full-wave dipoles. The rows a.re spaced A,'? apart? and to permit phasing in tahe llorth-south direction, each row has its own feeder syst.em. To preserve t,he bandwidt,h of the syst.en1, a binary branch- ing (Christmas tree) feeder network is used t,hroughout. This provides an ident,ical cable path of about. E X , from each dipole to the cent.ral junct,ion. Of this length, Soh, is composed of pha.se-stable coaxial cable. At each junction, a coa,xial matching section converts t,he impedance to 50 ohms. The coupling of the 4 rows in t.he north-south plane is also done with a branching network 1vit.h phasing switches (see Section IV-F) installed at. each junct.ion.

Systems such as the T or cross require a severe apodiza- tion of their apertures to reduce sidelobe responses. The escitation functions used a.re shown in Fig. 6. For con-

venience, a conlmon grading attenuator is used for each group of 4 dipoles. A uniform illunlina,t.ion is used in t.he north-south plane. The grat.ing responses t.hat would norma.lly result from this staircse function fall on the zeros of the response of the 4 h aperture of the north-south array. This excitation function gives the east<-west arm a voltage response with a. half-width of 1.1 degrees in hour angle, a negative first sidelobe of 1 percent, and remaining sidelobe of less t.han 1 percent..

When the east-west arm is used by itself in a tota.1 power system, i t has a half-power beamwidth of 0.8 X 26 degrees at t,he zenit,h.

E. ,$'osth,-S&k A I T ~ Y

The north-sout,h arm consists of 6-3 rows spaced A/2 apart,, each wit,h 4 full-v-ave dipoles. The 4 dipoles are coupled together? and each row is then connected into the nort,h-south feeder system shown in Fig. 7. The total cable path to ea,ch dipole is made up of the same length and t.ype of cable used in t.he east,-west a.rray to eliminate differentia,l temperature effects.

In order to point the hea.m of R I ~ array employing a branching feeder system, it is necessary to alter the rela- tive phases a t each 2: 1 junct,ion. The phasing system is discussed in detail in the following section. The phasing of adjacent row^ however, ubo imposes impedance mat.ching problems because of the large mutual coupling at the X;;?2 spacing.

When the relative phaw of two adjacent rows was rotated through 350 degrees, t.hp junction impedance traced out. an approximate cardioid pattern 011 a Smith impedance chart. By ad juhng t.he dipole matching trans-

166 IEEE T R 1 S S I C T I O X S OS AXTEXXAS AXD PROPAGITION, XARCH 1969

formers and t.he lengths of cable between the dipoles and t,he junction it was possible to arrange for the pattern to lie very close to the unit circle for all phasings except those corresponding to extreme zenit.h angles. A single reactive stub for each phase increment, therefore, was sufficient to make t.he junction impedance resist,ive n7it.h a nlaximum VSWR of 1.3. Because of circuit losses, further concessions to mut.ual effects at. succeeding junct,ions %\*ere not. re- quired.

The aperture illumination of t.he north-south arm is uni- form in the east-west. plane and follows the function given in Fig. 6 in the north-sout,h plane. Kote t,hat tjhe zero- order component from t,he overlap region of a T configura- tion must appear wit.h half the weight of the higher order spacial components in order to reproduce the component weighting of a uniform apert.ure. The computed volt,age response for this function has a half-width of 1.7 see 2 de- grees, where Z is the zenit,h a.ngle. The first sidelobe is - l percent and all remaining side lobes are less than 1 percent,.

When used alone as a total power array, t,he north- s0ut.h arm has a response with a beamwidt.h in declina- tion of 2.1 see Z degrees and broad low-level wings. The east-west, bea.mwidth is 14 de, mrees.

F. Phascng System

As mentioned previously, the phasing of a branching feeder network requires the installation of a phasing switch at each junct,ion. The main phasing system of t,he %-MHz array employs 3 switches in t.he east-west arm and 63 switches in t.he north-south arm.

The phasing switches have 16 positions, each position corresponding to the insertion of an independent length of ca.ble. The use of independent 1engt.hs of cable permits the introduct,ion of a unique matching stub for each position of the switches used to phase the adjacent rows of dipoles. The ca.ble lengths vary from 0 to 15X/16 in increments of X116 or 22.5 degrees. The phase of any row can therefore be adjusted with a maximum error of *11 degrees. For survey work, t.he beam positions can be chosen so that. t.he phasing required in the last three st.ages of the feeder sys- tem is an exact multiple of X/lG. The systemat,ic errors introduced in the first two stages due to the incremental nat,ure of t,he phase adjustment produce a phme modula- t.ion along the apert.ure with periods of 1 X and ?X. The corresponding grating responses lie on t.he zeros of the north-sout,h beam of the east-west array and are rejected.

Since the switches at. any one level in the feeder system require a. common phase increment, they are coupled to drive one anot,her in series. By setting t,he two master switches of t.he east-west. array and the six of the nort,h- south array to tabulated values, it is possible to rapidly point. the beam of the instrument to any desired zenith angle in the n0rt.h-south plane.

To permit, faster sky coverage for mapping observa- tions and position measurements for point sources, a. rapid- phasing network has recentJy been added. The system operates on a time-sharing ba.sis and provides effectively simultaneous coverage of five adjacent declinations.

E - w mmr N - S ARR4V

r- I

I I

I I I I I I I I I L- -_ - - - _ -

Fig. S. 2?.2.5-lIHz receiving system. Solid lines indicate signa p a t h , dazhed line. il1dicat.e control pat.hs.

Seven fast-acting crystal diode switches, placed as shown in Fig. '7. *en-e to shift the bea.m t,hrough angles of 0, +0.163. and = t O . Z C i beamwidths (0, = k O . G , =t0.90 degree at the zenith). The beam is pointed to each position four t.imes in ench integration period, permitting the observing intervals for each beam position to be arranged sym- metrically around the same mea.n obseming time.

The utilization o f broadband dipoles and branching feeder systemsn-a? intended to ensureadequate bandwidth. The presence of man-made radio interference in adjacent ba,nds, hon-ever. usually restricts the useful bandwidth of the system to :lpprosimately 1 percent of the cent>er fre- quency. -1 receiver bandwidt.11 of 300 kHz has therefore been used for all observations. The maximum usable bandwidth governed by impedance effects alone has not. been determined.

The phasing syitches insert lengt,hs of cable from 0 to 1.3;lCi. nnd phase path differences of ~ z h are ignored. When a finite brtndu-idth is used, this results in a progrmsive decrease i n system gain with increasing zenith angle. At a zenith nngle of 60 degrees, the gain reduction is 3 percent for ;I. Ix~ndn-idth of 300 k H z and would fall to about one half ior a bandwidt.h of 1.6 NHz. If required, this effect could be virtually elimina.t,ed by t,he provision for the insertion of hnndn-idt,h cable a t a few of the major junctions of the north-south feeder system.

I-. RECEIVISG ASD AUXILIARY EQGIPNEST

The recekels and associat,ed equipment used wit.h the Z-AIHz telescope are illustrated in Fig. 8. The preampli- fiers are located at the center of ea.ch arm. The remaining units sho~s-n are all in the control room of t.he main ob- servatory building.

COSTMN et al.: LARGE 2 2 4 4 ~ 2 ARRAY FOR RADIO ASTROSOXY 167

The receiven used are ext,remely gain-st.able stagger- t,uned amplifiers. They have a flat passband of 300 kHz centered on 22.25 MHz with a gain that may be varied from 60 t.0 120 dB. The correlat,ion is done using the stan- dard phase-switching technique developed by Ryle [16]. The use of receivew 1 and 2 t,o detect both the in-phase and ant,iphase components improves the signal-to-noise rat.io by fi and greatly a.ids in the suppression of int,er- ference incident on only one of the arms [17]. Receiver 3 is used to monitor t,he total power outputs of the txo arns.

The ent.ire system is controlled by the programmer which permits unattended automatic operat.ion. The in- temit.ies for t.he 5 beam posit,ions, the tot.al power from t.he two arrays, and the riometer out.put are recorded digitally. Provision is made for 30-second (6 < 60 degrees) or 60- second (6 > 60 degrees) integrat.ion periods and single- or mult,ibeam operation.

The X,/2 switch reverses the phase of the correlat.ed signal for half of each int.egration period. -\ synchronous reversal of the integrator input emure that only the correlat,ed component, is recorded, completely independent of any dc drifts that might. develop in the system. The A;4 switch is used only for testing; it provides a null indication of equal phase paths for the calibration and for t,he arrays when a test signal is injected into the overlap region (cf. Sect.ion 11).

A diode noise genemtor inserts a stable correlated cali- brat.ion signal int.0 the syst,em during every second int,egra- tion period. This provides a continuous measure of sensi- tivit.y and ensures t.he long-t,erm continuity of the ob, qerva- tions.

A 22.25-MHz riometer, connected to a sepamte crossed- dipole antenna.? continuously monitors t,he ionospheric absorption.

TI. CALIBRATION AND PERFORhlBKCE

.in absolut,e ca.libration is part.icularly import,ant for a large radio astronomy a.rray operat,ing at. a frequency where few previous measurements have been made. -4 method which relates the syst,enl gain to t>ha.t of a st,andard dipole Kas developed by Little [18] for calibration of the original 31ills cross. X standard dipole has been construct.ed for t.he 22-MHz array, but only preliminary measurements have been made.

An alternative method which has been used to provide an interim calibration, relates all measurement8 to the flux density of t,he radio source Cygnus A (S = 29 000 X 1WZ6 T1'.m-2 Hz-'). This is t<he only source for which a 22-MHz value of sufficient. accuracy is available [19], and this pro- vides a measure of the instrument sensitivity near t,he ze- nith. To extend this calibration t.0 other zenith angles, one may use flux densities obtained by extrapolation for those sources which have well defined spectra down to 38 MHz. For the present work, t,he spectra have been determined from the 38- and 178-MHz flux densities of the Mullard Observatory [20]-[22] a,nd the 750- and 1400-AIHz values of t,he Na.tional Radio Astronomy Observatory [23]. The mtio of flux densities, measured in terns of the zenith

0 00 f 30' 60' 90'

ZENITH ANGLE

Fig. 9. Relative gain of the 22.25-MHz array as a funct,ion of zenith angle. Solid line indicates isolated element, .-Cygnus 9 !3C405), 0-mean of weaker sources in zenith angle zones 0-20 degree.y 20-40 degrees, 40-60 degrem.

calibmtion, to the extrapolated flux densities was cal- culated for 98 sources. The results have been grouped for three zenith angle zones: 2 = 0 to 20 degrees, 20 to 40 degrees. and 40 to 60 degrees. The mea.n ratios are shown as the circles in Fig. 9 toget,her with the point for Cygnus A . The error ba,rs shown are probable errors and are ap- proximately 3 percent,.

The response of an isolat,ed element, X/8 above a reflect- ing screen. as a function of zenith angle, i.e.,

is show1 by t,he solid line in Fig. 9. The excellent agreement of t,he results from the fainter

sources near t.he zenith with t,he Cygnus A calibration sug- gests that. t,he method is relia.ble. There is no evidence for departures from the isolated element. response given by (3) for zenith angles less than 60 degrees.

A possible departure for very large zenit.h angles, shown as the dashed curves in Fig. 9, is suggested by observat,ions of a few southern sources. Some deviation of this sort is to be expected when the mutual impedance effects a.re la.rge enough to dominate the reflecting screen factor.

The brightness temperature scale, which depends on beam shape as well as forward gain, may be derived from the flus density calibration using the relation

where Q = Qo/cos 2 is t.he effect.ive solid angle of t.he in- strument. corrected for foreshortening. Tb is the brightness temperature of an extended source averaged over the system response, and S is the flux density of t,he equivalent point source.

In general, the 22-MHz system has performed very much a s expected. The bea.m shape, as indicat,ed by observations of point sources, conforms very closely to t.he theoretical pattern. Fig. 10 shows t.he comput,er reduction of 3 scans, where the galactic background has been removed and t,he bheoretical beam shape fit,ted: Fig. 10(a) 3C 3-1.8 (S = 2690 X 10-26W.m-zHz-1) under good observing conditions, Fig. lO(b) 3C 348 showing severe ionospheric scint,illat,ion,

16s IEEE TFUKKKTIONS o s AXTENNAS ASD PROPAGATIOK, MARCH 1969

x

(C)

Fig. 10. Computer-type plots of three scms over point sources

beam shape (0 0 0) fitted (coincidence of x aud 0 shown as -). ( X X X ) wit.h galactic background removed and bheoretical

(a) 3C 348 S = 2690 X 10-26FV.m-?Hz-1. (b) 3C 348 severe ionospheric scintillation. (c) 3C 190 S = 50 X 10-?6W.m-ZHz-1.

and Fig. lO(c) 3C 190 (X = 50 X 10-26W*m-2Hz-1), a weaker source near the confusion limit.

The sidelobe responses are confined mainly t.0 the two principal planes of t.he instrument.. In hour angle, the first sidelobe is typically -3 to -4 percent? a,nd the remainder a.re all less than 1 percent, falling below the limit of detec- t.ion (0.1 percent.) a t angles greater than 10 degrees. In declinatioI1, the fiwt sidelobe is approximately -4 percent, and other nearby sidelobes are of the order of 2 percent or less. An examination of all available suwey records for spurious responses in declination from the intense sources, Cassiopeia -4 m d T’irgo A, revealed no values greater than 1 percent for angles nore t>han 18 degrees away from the main beam and a nlea,tl absolute value of less than03 per- cent.

The response of’ the syst.en1 to an extended source is illustrated in Fig. 11. This shows a scan across the galactic plane at 6 = -12.2 degrees (2 = 61.0 degrees). The com- parison scan (dashed line) a t 6 = 12.9 degrees was taken with a 5O-MHz array a t Jica.marca, Peru? direct.ed to the zenith [24]. This array is a, simple square aperture with a. beamn-idt.h of i .1 degrees. The intensit.>- has been scaled to provide the best fit. The agreement. is excellent except, for regions near a = ljm and a = l B h 20m, where

22 20” 18‘ 16n RIGHT ASCENSION

Fig. 11. Scans across galactic plane. Solid line indicat.es 22.25 MHz, dashed line indicates 50 MHz [24], dotted line indicates 4.7 JIHz [25] (not tQ scale).

absorption by int.erstellar ionized hydrogen is clearly in- dicxted. This absorption is also evident on the 4.7-MHz observat.ions (dotted line) of Ellis and Handton [E]. Similar comparisons mith the 178-MHz observations of Turtle and Baldwin [2G] a,lso shorn a very high degree of corrclation. The out,put of the %MHz array: therefore, corresponds very closely to a t.me pencil-beam response.

VII. DISCESSIOX

The performance of the a,rray shon-s that the T config- uration may be used to produce a pract,ical pencil-beam inst,rument.. The declina.tion sidelobes are small, a.nd t,he extra width of the north-sout.11 array has proved an effec- t,ive meam of suppressing the far out sidelobes in hour angle. The sharing of t.he dipoles in the overlap region has resulted in an accurate total power response.

I t should be emphasized t1la.t the spurious responses of the system may be either positive or negat.ive and have a mean value very near zero. The instrument is, therefore, largely free of the systematic bias that is present in the out- put of a conventional a.perture where the sidelobe contribu- tion is always positive.

The 22-MHz a.rray has been in full operation since mid- 1965 and is used prinlal-ily for t,wo Binds of observations. First., drift scans of the type shown in Fig. 11 have been made at half-beanm-idt.h intervals and are being used to prepare a complete map of the northern sky from declina- t,ion +90 to -20 degrees. This will permit st,udies of the distlibution of the thermal a,nd nont.herma.1 radiation from the gala.xy. Second, several short scans of the type shown in Fig. 10 have been made over the positions of ea.ch of about 400 ra.dio sources. These results, coupled n-it.h t.he survey obsewat.ions, Kill provide 22-11Hz flux density measures for 400 to 500 sources down to a. 1imit.ing flux demitry of 30 X 10-‘6W.m+Hz-1.

ACKNOWLEDGMEST

The a.ut.lom wish to thank Drs. J. L. Locke and J. A. Galt for their help and encouragement- during the design and construction of the array. J. H. Dawson assist.ed ably m-it.h the construction and maintena.nce of the receiver system.

IEEE TRAXSACTIOSS OS AXTESSAS .LSD PROPIGATIOS, VOL. AP-17, NO. 2, XIARCH 1969 169

REFEREKCEG

[ l ] C. A . Shain, "The Sydney 19.7-mc radio telescope," Proc. IRE, vol. 46, pp. 85-88, January 19.58.

[2] C. A. Shain, M. M . Komesaroff, and C. S. Higgins, "A high

V C J ~ . 14, 110. 1, pp. .508-514, 1961. resolution galactic survey at 19.7 Mr/s," AustraZ. J. Ph.ys.,

131 C. H. Costain and F. G. Smith, "The radio t.elesc?,pe for 7.9 meters wavelengt.h at the hlullard Ob*ervat.ory, Xonthly

[4 ] B. T. 1,lills: A. G. Little, K. V. Sheridan, and 0. €3. Slee, "A Solires Roy. -4stron. SOC., vol. 121, no. 4, pp. 105412, 1960:.

rol. 46, pp. 67-84, January 19.38. high resolution radio telwcope for use at 3.5 m," Proc. IRE,

[ 5 ! IV. C. Erickwn, "The decametric arra.ys at. t.he Clark Lake Eadio Observatory," IEEE Trans. dnfennas and Propagalion,

[6] L. L. Bazelyan, 8. Ya. Braude, V. I-. 1-aisberg, V. 1'. Krymkin, vol. AP-IS, pp. S22-127, hIay 1963.

-1. 1'. >Ieu*? and L. G. Sodin, "Study of the spectra of discrete sourre.< of cosmic radio emission at freqltencies below 40 1lc/s," Sorict .4str-on-AJ, vol. 9, pp. 471-479, Sovember/Decemher

[ i ] Y I I . 11. BIYJlIk, h-. Yu. Goncharov, 9 . V. Men', L. G. Sodin, 196%

and S. K. Sharykin, "T-form radio telescope with electrical beam scanning in 10-25 Mc/s range" (in Rusian), Radiofizika, rol. 10, no. 5, pp. 608419, 196i.

[SI J. A. Galt, C. R. Purton, and P. A. G. Scheuer, "-4 large 10 1 l f I z array for radio ast.ronomy," Pub[. Dominion Obs., vol. 2.5, no. 10, pp. 295-304, 1967.

[Q] C. H. Costain, "The new 23 JIc,:s radio tF1escope of the Domin- ion Radio liztrophysical Observatory,' Publ. Astron. SOC.

[lo] J. A . Galt and C. H. Costain, "Low frequency radio astron- Pnc<'c, vol. i 4 : p. 400, October 1962.

only:" Trans. Roy. SOC. Canada, sec. 4, vol. 3, pp. 119-430, June 196.3.

[ l l ] J. €1. Blythe, ''A new type of pencil-beam aerial for radio at;tronomy," Monthly :\ofices Roy. dstron. Soc., vol. 117, no. 6, pp. 6-1.1431, 1937.

[12] 11. Hyle and -4. Hewish, ' T h e synthesis of large radio tele- scopes," Monthly X o t l C ~ s Roy. dstrun. Soc., voi. 120, no. 3, pp. 220-2:30, 1960.

[13] B. Y. Xills, "Cross-type radio telescopes," Proc. IRE (hus-

1141 J. H. Croather and R. W . Clarke, ''A Dencil-beam radio tele- tralia), vol. 24, pp. 132-140, February 1963.

. . scope operating a t 178 kIc/s," &onihhJ il.'olices Roy. -4stron.

[l5] R. K. Bracewell and G. Snarup, 'The &anford microwave Soc., vol. 132, no. 4, 1966.

spect.roheliograph antenna, a microst,eradian pencil beam i n k -

pp. 22-30, January 1961. ferometer," IRE Trans. Aniennas a.nd Propagalion, vol. "-9,

[16] Lf. Kyle, ",4 new radio interferometer and its application to the observation of weak radio stars," Proc. Roy. SOC. (London), per. A , vol. 211, pp. 351-3175, 1952.

[l7] P. F. Scott, ?ti. Ryle, and A . Hewish, "First results of radio st.ar observat,ions using the method of aperture synthesis," Monthly aVotices Roy. -4stron. SOC., vol. 122, no. 2, pp. 95-111, 1961.

[18] X. ( i . Litt.le, "Gain measurementn of large radio telescope aerial..," d m t . J. Phys., vol. 11, no. 1, pp. '70-78, 1958.

[19] K. I. Kellermann, "A compilation of radio source flux densi- t,ie5,'' Publ. Owens Valley Radio Obs., vol. 1, no. 1, 196.4.

[20] P. J. S. IVilliams, 6. Kenderdine, and J. E. Baldwin, "A survey of radio sources and background radiation a t 38 >.€c/s," Mm.

[21] J. F. R., Gower, P. F. Scott., and D. Will$, ''A survey of radio Roy. Astron.. SOC., vol. 70, pp. 53-110, 1966.

sources 111 t.he declinat.ion ranges -07O to 20" and 40' to SO"," .llem. Hoy. Astron. Soc., vol. 71, pp. 19-144, 1967.

[22] J. 1). H. Pilkington and P. F. Scott, "-4 survey of radio source between declinations 20" and 40"," I f em . Ray. Astron. SOC.,

[23] I. I. K. Pauliny-Toth, C. >I. Wade, and D. S. Heeschen, "Positions and flux densitia of radio sourcq" dstrophys. J . (Suppl.) , vol. 13, pp. 63-124, May 1966.

[21] G. R. Ochs, "Synchrotron radiation measurements near the mametlc eouator." in Radiation Travved in he Earth's Mau-

5-01. 69, pp. 183-221, 1965.

n.di'c Field, *B. ai. McCormac, Ed. Dordrecht., The Nether- lands: L). Reidel Publ. Co., 1966, p. 703.

1251 G. R. 8 . Ellis and P. $. Hamilton, "Cosmic radio noise survey

' =

. . at 4.7 hIc/s," Astrophys. J., vol. 143, no. 1, pp. 22i-235, 1966.

[26] 9. J. Turtle and J. E. Baldwin, "A survey of galactic radiat.ion

6, pp. 1c50-476, 1962. a t 178 3Ic/s," Monthly Sotices Roy. Astron. SOC., vol. 124, no.

Scattering of a Plane Electromagnetic Wave by a Linear Array of Cylindrical Elements

OLIVER D. SLEDGE

A b s i r ~ c i - h approximate method for numerically calculating the scattered field of a linear array of center-loaded cylindrical elements illuminated by a plane wave incident in the H plane of the array is described. To illustrate the use of the approximation technique, the H-plane scattered field of an eight-element model array is calcu- lated for various conditions of impedance loading and illumination.

From extensive calculations, it is found that the H-plane scattered field of the array can be represented approximately by the product of the electric field scattered in the reflected direction and an inter- ference factor. The interference factor is the equivalent of the com- plex array factor of the model array when excited with uniform amplitude and an element-to-element phase progression propor- tional to the sine of the angle of incidence of the illumination.

Matillscript received July 1, 1968; revised October 25, 1968. The allt,hor is v7it.h t,he E. S. Kava1 Research Laboratory, Wash-

ington, D. C. 20390.

A ISTRODCCTIOX

NCMBER of papers have appeared in the last decade in the litemture which t.reat t.he backscatter-

ing of a center-loaded cylindrical a.ntenna [l], [ 2 ] . Chen and Liepa [3] st,udied the effect. of central loading on the current. induced on a t,hin cylinder by a plane electro- ma.gnet.ic n-ave.

The purpose of t.his paper is to investigate the scatter- ing propert,ies of a linear array of cylindrical elements. The approach used to solve the steadystate scattering problem is first to find an approsirnation to the currents induced in the array by the incident, illuminat,ion. Once the current dist,ribution in the array is dekrrnined, one can easily calculate the far-zone scattered field.