Interference-pattern intermediate-distanceantenna measurement technique

Indexing terms:

K.M. Keen

Antenna radiation problems, Communications, Computing, Microwave antennas, Micro-wave measurement

Abstract: A new antenna-pattern measurement technique for antennas with large-field distances is described.The method is applicable to conventional ranges with 2-axis positioners and receivers that work in amplitudeonly. Basically, the technique is to make pattern measurements on a conventional finite-length range in thenormal way but with a third antenna or an auxiliary circuit that causes interference. Interference patternsand finite-distance amplitude patterns are recorded and the finite-distance phase distribution extracted bydata processing. This finite-distance data is then the input to a near-field/far-field transformation computerprogram that gives the far-field patterns.

1 Introduction

The far-field characteristics of the majority of microwaveantennas can be determined on conventional measurementranges in the normal way, i.e. by rotating the antenna andmonitoring its radiation pattern with a remote antennasituated a sufficient distance away. However, there arecases - either because the Rayleigh criterion does notapply, or because the antenna aperture is large in terms ofthe operating wavelength - in which the need for very largerange distances precludes conventional methods and someform of near-field measurement method may be necessary.A recent example was a shaped-beam satellite antenna forearth coverage,1 for which an 'intermediate-range distance'facility was set up near Chelmsford, England.2 On thisrange both phase and amplitude are monitored, and thequasi near-field information is transformed by computerprocessing to give the far-field radiation patterns. This typeof range has additional usefulness in that it can be operatedin the conventional mode for measuring the majority ofantennas.

From the reverse point of view, existing conventionalranges could be adapted to this system for the measurementof antennas with large far-field distances; the onlydrawback being that the above mentioned range uses asophisticated (and expensive) phase and amplitude receiver,which most ranges do not have. In an attempt to extend thecapabilities of the antenna measurement facility at theEuropean Space Research and Technology Centre withoutrecourse to a phase and amplitude receiver, a newintermediate-range technique has been devised whereby thephase information is extracted from interference patterns,but in a more direct way from that accomplished by theholographic method developed at the University ofSheffield.3

the new technique was first simulated by a series ofcomputer programs to establish its viability and give some'feel' for the extent of the data processing that would beinvolved. This was followed by a practical trial withencouraging results, and the technique is now considered asa generally available feature of the facility at which it was

Paper T198 M, first received 3rd April and in revised form 17th May1978Mr. Keen is with the RF Technology Centre, Electrical ResearchAssociation, Cleeve Road, Leatherhead, Surrey KT22 7SA, England

MICROWAVES, OPTICS AND ACOUSTICS, JULY 1978, Vol. 2, No. 4

developed.4 Aspects of the development have been reportedpreviously,5"7 but this paper gives a comprehensive reviewof the technique and the investigation and implementationof it.

2 Description of the technique

In Fig. 1, antenna A is the antenna to be measured and issituated on a 2-axis mount, B is a static reference antennaand C is the remote antenna. The diagram shows themeasurement hardware in an anechoic chamber, but thesystem is the same for indoor or outdoor ranges. Fig. 2shows the feeding circuit, which includes a phase shifterable to switch from some arbitrary zero to +90° and

Fig. 1 Measurement-method geometry

r.f. source

Fig. 2 Feeding circuit - method 1

113

0308-6976/78/198M-0113 $1-50/0

+ 180° relative to that zero. The purpose of antenna B isto act as a fixed reference for interference patterns set upat antenna C by interference between radiation from A andB. Fig. 3 shows an alternative circuit for causing the desiredinterference and this will be described later, but for themoment the technique is best explained by reference toFig. 2.

2/>4(0,0) (0,0)]

i I

powerdivider

r.f. source

phaseshifter

matched load

directional.coupler

toRx

Fig. 3 Feeding circuit - method 2

Let the field distribution at the chosen intermediatedistance (i.e. somewhere between, say, the Rayleigh dis-tance and the close proximities of conventional near-fieldprobing methods of measurement) and in one knownpolarisation be represented by

and let the corresponding far-field distribution be

(1)

(2)

where Ax and A2 are amplitudes and i//, and \p2 arephases. The desired description of the radiating charac-teristics of the antenna is y42(^, 0), usually expressed indecibels, together with its orthogonal polarisation counter-part (o.p.c). These are the far-field radiation distributions,and often only certain 'cuts' i.e. radiation patterns, arerequired. From electromagnetic theory it is known thatA2(Q, 0) and its o.p.c. can be found from A\(Q, 0) and' / ' I ^ J 0) plus their o.p.c.s, and there are a number of near-field to far-field transformation computer programs inexistence that do this. The interference method provides away of determining the quasi near-field phase, and theprocedure is as follows.

With antenna B switched out, the quasi near-field power-level distribution P\{6, 0) is recorded. This determinesA\(6, 0), since, ignoring decibel scaling and a constant ofproportionality,

P1(0,4>) = /4i(0,0) (3)

Then, with the reference antenna switched in and the phaseshifter set to 0°, + 180° and + 90°, three more power leveldistributions are recorded. These are P2(d, 0), ^3(0,0) and-^4(0, 0), respectively. The phase distribution can then beextracted from the four power-level distributions using thefollowing relationships:

(0,4>)[Pt(0,0) + P3(0,0) ,0)](4)

and

,0)[P2(0,0) + P3(0,0) - 2/\(0,0)]}(5)

The constant 6 has no effect on the subsequent trans-formation and can be ignored; it is actually the real valueof the arbitrary zero of the phase shifter. To extract4/1(6, 0) from eqns. 4 and 5 it is only necessary to evaluateone equation and find the arithmetic sign of the other.

The above outlines the basis of the measurementtechnique; four power-level distributions are recorded, andfrom these a phase distribution and an amplitude dis-tribution are extracted by data processing. If this is done intwo orthogonal polarisations, the resulting phase andamplitude information can be processed by a near-field tofar-field transformation computer program. The datasampling about antenna A requires a 2-axis positioner, butmany conventional ranges have these, as sampling in thisfashion is carried out for the measurement of directivity.8'9

The spherical envelope about the antenna can be monitoredby either great circle (constant 0) or conical (constant 0)rotations.

3 Computer simulation

Before trying the technique in practice, a computersimulation was made with a mathematical model of anarray of 17 halfwave dipoles with interelement spacings of1-2 wavelengths at a frequency of 2-4 GHz. In the array,the dipoles were axially aligned and equally fed in ampli-tude and phase, giving a far-field radiation pattern with amain beam of 2-5° half-power beamwidth in the arrayplane, two prominent sidelobes and many minor lobes. TheRayleigh distance for the array is 97 m. The near-fieldsampling was made in Ee and EQ components, and tosimplify the simulation the array was placed along the0 = 0° axis of the measurement co-ordinate system, so thatthe EQ field was always zero. The near-field range lengthwas set at 5 m, and antenna B was placed this same distancebelow A and inclined 45° to the horizontal. The parametersof the feeding circuit were chosen so that, with the main-beam peak of the array directed toward C, the fieldamplitudes from antennas A and B were equal at C.Radiation-distribution samplings were made at intervals of2-25° in 0 and 90° in 0. These spacings were dictated bythe computer program subsequently used for trans-formation, and represent the minimum number of samplesnecessary for the matching of spherical modes for this sizeof antenna.

Fig. 4 shows the quasi near-field power-level pattern^1(0) of the array, and the continuous line curve of Fig. 5shows the directly computed far-field pattern. In the

30 60 90 1209. degrees

150 180

Sin [0,(0, 0)-5]Fig. 4 Power-level pattern Px (8) of the 17-dipole array at 5 mdistance

114 MICROWAVES, OPTICS AND ACOUSTICS, JULY 1978, Vol. 2, No. 4

measurement simulation, after the amplitude and phasedata were extracted from the sets of four near-field dis-tributions, they were transformed to the far field using thecomputer program SNIFT.10 This is a spherical-modematching program that makes use of fast-Fourier-transformtechniques. Output from this is shown as dots on Fig. 5,and it can be seen that there is good agreement with thedirectly computed far-field pattern.

0

5

-10

-15

-20

-25

-30

-35

-400 15 30 45 60 75 90 105 120 135 150 165 180

9 , degrees

Fig. 5 Far-field pattern of the 17-dipole array

Continuous line is the calculated pattern at infinity, dots are fromthe simulated near-field measurement

4 Trial measurements using the technique

Fig. 6 shows the first intermediate-distance measurementarrangement although, as will be described, this was sub-sequently modified. In the photograph the antenna under-going measurement is a 1 -3 m paraboloid and feed. This sizeof antenna was convenient as a test case, as its Rayleighdistance was small enough for conventional far-fieldmeasurements to be easily made and used as a comparison.

Fig. 6 Initial measurement arrangement

MICROWA VES, OPTICS AND ACOUSTICS, JULY 1978, Vol. 2, No. 4

The feed was a linearly polarised log-periodic antenna. Noattempt was made to optimise its position for the bestsecondary patterns, as poor patterns would not have beendetrimental to the investigation. In fact, high sidelobesseen with the //-plane secondary pattern and a deep nulljust off the main beam in the £-plane secondary pattern(caused by a feed strut in this plane) were considered usefulfor monitoring the accuracy of the technique at differentlevels and angular positions. The remote antenna was acrossed log-periodic dipole array, as was the interference-wave antenna. Measurements were made at 1200 MHz,ultimately at 3 m range length (the photograph shows a5 m separation) and also at 20 m range length on a con-ventional elevated range for comparison. The Rayleighdistance of the paraboloid antenna is 13-5 m.

The purpose of the interference-wave antenna is toprovide a constant reference wave for interference-patternformation at the remote antenna, and on the originalmeasurement set-up the ability to do this was tested bytransmitting with the interference antenna alone, receivingwith the remote antenna and rotating the test antenna.When this was done, the interference wave was seen to befar from constant because of scattering from the testantenna and its tower and metal base. The obvious solutionto this problem would be to use a more directive inter-ference antenna, but an alternative scheme was proposed atthis time and adopted for the trial measurements. With thisscheme, shown in Fig. 3, instead of forming the interferencepattern in air from two waves propagated by antennas, theintereference-wave antenna is removed, a cable link takesthe reference wave to a point beyond the receive antenna,and interference is caused by combination of the two wavesin a directional coupler. Fig. 7 shows how a directionalcoupler can be used in this way.

The upper diagram of Fig. 7 shows a directional couplerin normal use; if a wave of voltage Fis applied at the inputport, then KXV and K2V are seen at the output ports,where Kx and K2 are complex constants each composed ofan insertion phase and an amplitude reduction term. If thecoupler is used to combine two waves of voltage VB and Vc

(lower diagram) applied at ports B and C, then a certainamount of the input energy goes to the load, and an outputKXVB + K2VC is seen at port A. The phase constant

K,V

load1 1

1 1! c

fload

J — * - K , V 8 . K 2 V c

A

Fig. 7 Use of a directional coupler for a 2-channel combinera Coupler in normal useb Coupler combining the reference- and direct-channel signals

115

introduced by K\ and K7 is unimportant, as all measure-ments are in relative phase anyway. The amplitude con-stants are also unimportant as they have the same effect aschanging the amplitude of the reference signal, and this canbe set to anything convenient — the only requirement beingthat it stays constant throughout data recording.

A practical point that has arisen from the use of acoupler in this way is that extra cable must be put into thedirect circuit to approximately equalise the direct andreference path lengths to minimise dispersion.

The direct and interference radiation distributions weresampled in orthogonal linear polarisations at intervals of5° in 8 and 6° in 0 over the spherical region about theparabolic antenna; the number of data points correspondingto this spacing were in accordance with the input require-,ments of the transformation program,10 and are theminimum number of samples necessary for spherical-modematching over this particular size of antenna. The sampledrelative power levels were recorded on 8-level punchedpaper tape to give a resolution of 0-25 dB, and then thisdata was transferred to a digital computer for processingwith two computer programs. The first of these extractedthe amplitudes and phases from the recorded power levelsaccording to eqns. 3, 4 and 5, and then these values weretreated by the near-field to far-field transformation programSNIFT to give far-field radiation patterns. The programrequires only relative phase values and is able to allow foran added phase variation due to the phase centre of theantenna under test being not precisely on the axes ofrotation.

Figs. 8 and 9 show how the far-field patterns obtainedby the interference technique compare with the con-ventionally measured ones. Fig. 8 shows //-plane patternsof the paraboloid antenna and Fig. 9 shows £"-planepatterns, and in each case the hard line represents theconventionally measured pattern and the dashed line showswhere the interference-technique pattern deviates from it.It can be seen that there is good agreement betweenpatterns measured by the two techniques down to lowlevels. There are, of course, some discrepancies, but it mustbe remembered that the comparison is being made betweenpatterns measured at 20 m (i.e. 1-5 times the Rayleighdistance of the antenna) and patterns in the true far field

-180 -120 -60 e 608. degrees

120 180

i.e. projected to infinity. Also, the conventional measure-ments were made in the presence of a ground reflectionlevel of the order of —30 to — 35 dB. In general, it isbelieved that the conventionally measured patterns are theless accurate of the two sets.

-180 -120' -60° 0° 60°9 . degrees

120° 180°

Fig. 8 H-plane patterns of parabolic antenna

Full line is from conventional pattern measurement at 20m, brokenline is pattern from interference method at 3 m

Fig. 9 E-plane patterns of parabolic antenna

Full line is from conventional pattern measurement at 20 m, brokenline is pattern from interference method at 3 m

5 Acknowledgments

The author would like to thank B.E. Westerman andW.B.S.M. Kneefel of the Christiaan HuygenslaboratoriumB.V., Noordwijk, Holland for carrying out measurementsand for suggesting the use of the directional couplercombiner. He would also like to thank J. Aasted, Head ofthe Antennas and Propagation Section, ESTEC, forencouragement, and D.J. Brain, also of ESTEC, for severalhelpful discussions.

6 References

1 COWAN, J.H.: 'A shaped beam antenna for a maritime com-munications satellite'. IEE Conf. Publ.128, 1975, pp. 101-106

2 WOOD, P.J. and LOCKETT, N.J.: 'A new type of test range forsatellite antenna polar diagram measurements', ibid., pp. 186—191

3 BENNETT, J.C., ANDERSON, A.P., McINNES, P.A. andWHITAKER, A.J.T.: 'Microwave holographic metrology of largereflector antennas', IEEE Trans., 1976, AP-24, pp. 295-303

4 KEEN, K.M.: 'The ESTEC antenna test range'. European SpaceAgency report, to be issued

5 KEEN, K.M.: 'Simulation of a proposed near-field to far-fieldantenna measurement system', Electron. Lett., 1977, 13, (8),pp. 225-226

6 KEEN, K.M.: 'An interference pattern intermediate distanceantenna measurement method'. ESTEC workshop on antennatest techniques: ESA preprint handbook SP127, June 1977

7 KEEN, K.M.: 'Antenna measurements on an intermediatedistance range by using an interference method', Electron. Lett.,1977,13, (13), pp. 375-376

8 HOLLIS, J.S., et al.: 'Microwave antenna measurements'(Scientific-Atlanta Inc., 1970)

9 KEEN, K.M.: 'A computer program system for evaluating partialand total antenna directivities from measured data', Proc. IEE,1977,124, (12), pp. 1117-1120

10 JENSEN, F.: 'SNIFT — computer program for spherical near-field far-field technique'. Appendix E to TICRA final reportS-45-01, under ESTEC contract 2478/75AK

116 MICROWAVES, OPTICS AND ACOUSTICS, JULY 1978, Vol. 2, No. 4