determining the antenna factors of EMC testing antennas using open area test sites. The technique works well for dipoles and dipole-like antennas including ...
Measurement
of the Antenna Factor of Diverse Antennas Using a Modified Standard Site Method James McLean and Gentry Crook Tactical Systems Research, Inc. 3207 Yellowpine Austin, TX 78757 Zhong Chen EMC Test Systems, L.P. P.O. Box 80589 Austin, Texas, 78708 Heinrich Foltz The University of Texas-Pan American Department of Electrical Engineering Edinburg, TX 78539
Abs&act: The standard site method given in ANSI C63.5-1988 is a convenient, widely used technique for determining the antennafactors of EMC testing antennas using open area test sites. The technique works well for dipoles and dipole-like antennasincluding broadbandbiconical antennas. However, for some types of antennas including somelog-periodic dipole arraysas well as some new types of compact,broadbandradiators, this technique can result in significant measurementerror as given since the gain and phasepatterns of the antennasare not taken into account. The method can give accurateresults if it is modified to include both the gain and phasepatternsof the antennas. Here we describe this new technique and then demonstratethe effect it has on the accuracy of antenna factor measurements. INTRODUCTION
The antennafactor [l] has hecomea standarddescriptorof EMC metrology antennas. The ability to determine freespaceantennafactor from measurementsconductedon an openareatest site (OATS) with a conducting ground plane is crucial becauseanechoic chamberswith operating frequency rangesextending down to 20 MHz (and below) are prohibitively expensive. Accurate measurementscan be made using an OATS covered with broadband absorber. However, to obtain accurate results, the antennas (both transmit and receive) must be elevatedhigh enough off of the absorberso that their reactivenear fields do not extend
0-7803-5057-X/99/$10.00 © 1999 IEEE
down to the absorber.Thus at 20 MHz, an absoluteminimum acceptableheight would be & = 2.38 metersabove the top of the absorber. However, even this arrangement is questionableas the reflectivity of most absorberat extremely oblique (grazing) anglesis large. Thus, in order to obtain accuratedirect measurementsof free spacequantities using an OATS, large antenna heights are required even with broadbandabsorbersin place. In any case,it is very desirableto be able to obtain accurateantennafactor data without going to such lengths. A technique for extracting free spacedata from measurementsmade over a conducting ground plane was first given by Smith 12,31 and later incorporated in the ANSI [4] and CISPR [5] standards. The technique has since been scrutinized extensively [6, 7, 8, 9, lo]. Much effort has been invested in examining the effectsof ground plane proximity on the antennasas well as mutual impedancecoupling between the antennasdue to near field interaction. The problem of ground plane proximity effect on the input impedance and hence reflection loss of the antennascan be amehorated to somedegreeby simply increasing the heights of the transmit and receive antennas. Nevertheless,perfect correlation between antenna factor data extracted from measurementsmade in OATS environments and free spaceantennafactors hasbeendifficult to obtain for some types of antennas. Experimentation has shown that this problem occurs with antennaswhich exhibit radiation patterns significantly different from iso-
595
lated dipole antennas. This includes log-periodic dipole arrays (LJ?DAs)as well as several new compact radiating elements such as the one described in another paper submitted to this conference [ I1 1. Essentially all of the previous work in this areahasfocusedon dipoles (electric or magnetic) and dipole-like antennas(such as biconical antennas). STANDARDSITEMETHOD A crucial part of the determination of antennafactor from measurementsmade on an OATS is the extraction of the effects of the ground-reflected wave from the measured data. The standard site method incorporated in ANSI StandardC63.51988 and describedin detail by Smith [3] involves three site attenuation measurementsin which the receiving antennais scannedover a rangeof heights above the ground plane. The geometry of this method is illustrated in Figure 1. Direct Ray (length dl)
computed in a straightforward manner using geometrical optics (GO). It hasbeenpointed out that this is only valid when the separationdistanceplaces the antennasin each other’s far field region. This is generally the case for a spacingof 10 metersfor frequenciesof about 30 MHz and above. However, it is not for the commonly-usedtest distanceof 3 meters(at 30 MHz). Analytical expressionsfor Eom,~,which are accuratein the nearfield will be derived in this paper. In general, the geometrical optics 2-ray model gives the following expressionfor the electric field generatedby a simple source situated over a ground plane as shown in Figure 1: -jPd2
E= j/3-
d@%+
-jPdl + IPl dqq
,j$
d2
where
Observation Point
Transmit Antenna
dl
=
dR2t-
42
=
j/iqiz$
(hl
-h2)2,
Prti is the radiated power, G is the gain of the antenna, and pL$ is the complex reflection coefficient of the ground plane.
With a radiated power of one picowatt and half-wave dipolegain of 1.64the expressionfor ED~(= for horizontal dipoles becomes: lx102 Image 3/
where IphI & is the complex reflection coefficient for horizontal polar&&ion impinging on an imperfect ground Figure 1: StandardSite Method for Determination of An- with relative permittivity &Rand conductivity CT: tennaFactor -7
Ph
=
k’hl
@h
If the three measuredsite attentuationsare AI, AZ, and As, the antennafactors are given by AFl
AF2
AF3
=
10logJn - 24.46
+
;[E;;UX+&+&-As]
=
10logfm - 24.46
+
;[E~+&+&-AZ]
(2)
=
lOlogf,-24.46
+
$[Egm+A2+As-A~]
where lpVlL@”is the complex reflection coefficient for vertical polarization impinging on an imperfect ground (3) with relative permittivity &Rand conductivity CT:
For vertical dipoles (l)
Ef;.vwi=&i%R2
. ~~+d~lpvlz+2d:~lp~l~o~(O~--BId2--ci~I)]1’2 pv,m t6j
where all quantities are in dB. Here EEK is the maximum electric field strength in the receiving antennaheight scan range for an ideal losslessdipole. This parametercan be
44
Pv =
596
=
IPJ Lb
(&R-j6Ob)Siny(&R- j6Ob) Siny+
(&R~-j60bCOS2y)1'2 (&R- j6Ohc - COS2 y) ‘I2 .
(4)
It is important to note that these expressionsfor Er apply strictly only to dipoles and dipole-like antennas(antennaswith dipolar fields). When the method is usedwith other antenna types, these expressionsimplicitly assume that the relative contributions of the line-of-sight (LOS) and ground-reflected rays are the same as for a dipole. Smith noted that with some high gain antennas(such as horns), it might be necessaryto incorporate the far-field gain into the derivation of these expressions. He further noted that it did not appearnecessaryto do so with moderategain antennassuch asLPDAs. We disagreewith this point. Moreover, the crucial information that apparently has been left out is the phase diKerence of the two rays (the line-of-sight and the ground bouncerays) in the geometrical optics model. This phasedifference arises from the phasevariation of the farfield electric field of someantennaswith elevation or azimuth. Of course, for a dipole (regardlessof whether it is horizontally or vertically polarized), the phaseis constant with elevation angle, 8 (see Figure 1). This is definitely not the case for many other types of antennas. LL%IITATIONSOFTHE GEOMETRICALOPTICSAPPROACH: NONLINEARRETARDATIONOFFHASEINNEAR FIELD The geometrical optics approach used to obtain Equation 4 models the field retardation of the dipole as a linear function of distance. In the nearfield of any antenna, the phaseretardation with distance deviates significantly from a linear function. For dipoles one-half wavelength and shorter for which a sinusoidal current distribution can be assumedthe induced EMF technique provides an elegant, analytical solution which includes the effectsof deviation from linear phasein the nearfield [12, 13, 141.For an x-directed short wire dipole of length w centeredat the origin, the current distribution can be assumedto be of the form: Z(x) = I,sin(P(w/2-
Ix]).
(7)
This equation is exact in the sensethat, in its derivation, the convolution of the assumedcurrent with the Green’s function is obtained without approximation. The only assumption is the sinusoidal shapeof the current distribution; this is known to be quite accurate for electricallyshort dipoles. This solution may be used to compute the electric field due to a horizontal dipole situatedover a conducting ground plane by invoking an image as was done in references[2,3] using GO fields. The modeling of the effect of the ground plane with an image dipole is exact evenin the near field. Thus, the electric field due to a horizontal dipole situated over a conducting ground plane is given by: E=
pw e-iWl
2eP4
.
--2cos(2)d'1 {
4
where Rrd is the radiation resistanceof the dipole. Note that, in order to couch the equation in terms of radiated power as was done in ref. [2, 31, the radiation resistance (however, not the reactance)must be d&et-mined. For an electrically short dipole, the radiation resistanceis to a very good approximation: R,.(&= BOB
Sk
(11)
However, the radiation resistance of a dipole in close proximity to a conducting ground plane deviates significantly from this. Nevertheless,analytical expressionsfor this quantity can be obtained [ 141: Rrad= w3~)2 -
30(/%42 sin(2/%)
. [~- 2i3h
sin(2j3h) -2flcos(2/3h)
n
(2Ph)31
(12)
On the other hand, the fields of an electrically-short dipole are known to be thoseof the lowest-orderTM-to-R spherical mode [ 16, 171Thus, we can write:
The electric field everywhereis then given by:
. [j~e-‘pdl(j&-+&+&)
Ex (x, y, z) = -j301,
,-.Nh .(
,--jpR,
-+-
Rl
(8)
RI
where
RI =Jx-,v/2)Z+yZ+zZ R2 = I
=
(x+w/2)2+y~+22 @GG?
+
jpe-jPd21pl$jLP
1 - 1 8d2 (AW2' (AW3 . (-1+7
(13)
This expression should yield more accurate predictions of electric field for dipoles shorter than one-half wavelength when the source and observationpoint are electrically close. However,the expressiongiven by Smith is ex(9) act for horizontally polarized dipoles in the farfield limit
597
and actually yields more accurate results for horizontal polarization when half-wave dipoles are used. In Figure 2, the electric field of a 1.0 meter long horizontal wire dipole operating at 25 MHz and situated 1.5 meters over a conducting ground plane is plotted as a function of height at a distanceof 3 metersfrom the dipole on the dipole’s principal axis. The electric held was computed using three approaches: first, the familiar GO expression given in Equation 4, second,using Equation 10 derived from the assumptionof a sinusoidal current distribution, third, using Equation 13 derived using the lowestorder TM spherical mode, and finally a numerical computation using the ‘WE” (nearl?eld)card in NEC-4. The NEC ‘WE” card computeselectric field using a numerical convolution of the computedcurrent distribution with the Green’s function. From Figure 2, it can be seen that the GO approximation agreeswith the NECY-4predictions only to within about 20 %. Equation 10 gives essentially identical results to thosecomputedwith NEC-4. It is quite interesting to note that Equation 13 also givesnearly identical results as the numerical simulation. Thus, one can conclude that the nonlinear phase terms in Equation 13 are the source of the deviation from the GO predictions. Finally, we note that this discrepancy occurs only when the source/receiverdistance is small. However, we note that a 3 meter distance does separatethe source and the receiver by one-quarter wavelength and that the reactive near field of a short dipole only extendsto about one-sixth of a wavelength.
r
Figure 2: Comparison of Electric Field Intensity Computed With Four Different Techniques In summary, either of the two expressionsfor electric field can be used with electrically-short dipoles to give more accuratepredictions of ~~~~ than the expressions given by reference [2, 31 or the ANSI standard. This is true for any separaliondistanceor height although the improvement in accuracy is primarily for closely separated
dipoles with low heights. Finally, we note that the analytical approach to obtaining the electric field given in Equation 10 can be extended to obtain results which are accuratefor half-wave dipoles. However, this is beyond the scope of this paper. Here we have only shown that ignoring nonlinear phaseretardation in the nearheld is a source of error in the ANSI standard(even for measurement distancesspecifiedby the standard). ANTENNASWHICHDONOTEXBIBITDIPOLAR RADIATIONPATTERNS
Of the group of antennaswhich do not exhibit a dipolar radiation pattern, it is useful to divide the group into two categories:antennaswith a well-defined phasecenter,and antennaswithout a well-defined phasecenter. The phase center as defined in IEBE Standardsfor Antennas [ 151is: “The location of a point associatedwith an antennasuch that, if taken as the center of a sphere whose radius extends into the far field, the phaseof a given field component over the surfaceof the radiation sphereis essentially constant,at least over the portion of the spherewhere radiation is significant.” Thus, the radiation pattern of an antennawith a well-defined phasecenteris completely specified with only magnitudeinformation. On the other hand, the radiation pattern of an antennawithout a well-defined phase center exhibits a radiation pattern which must be specifiedusing amplitude and phaseinformation. Unfortunately, in this case,the farfield phasepattern is strongly dependentupon the origin chosenfor the sphericalcoordinate representationof the radiation pattern. Log-periodic dipole arrays fall into this class of antennas. A remedy hasbeenproposedfor LPDAs in the form of a phasecenter which moves as a function of frequency. In Figure 3, the simulated variation of the farfield gain and phase(arbitrary reference) of a typical LPDA (z = -88, d = -05) is given as a function of co-latitude angle. In the simulation, the phasecenter of the antenna was taken to be at the center of the dipole which was resonant at the simulation frequency, thus the phasecenter was taken to be at the center of the so-called active region of the LPDA. This is in keeping with the assumptionthat the phasecenter tracks the center of the active region in an WDA. As can be seen,over 90 degreesof co-latitude angle, the gain varies about 4 dB and thus the magnitude of the electric field varies by about a factor of 1.5. However, the phase also varies significantly (more than 20 degrees)over this range of co-latitude angle. As can be seenthe assumption that the phasecenterof an LPDA is the centerof the active region only partially compensatesfor the phasevariation with co-latitude angle. This variation arisesfrom the fact that more than one dipole contributes to the radiation at any frequency within the operating frequencyrangeof the antenna.Thus the radiation at any given frequencyis arising from severalspatially distinct antennas.
598
active region of the LPDA. if this were not done {as is generahy the casewhen a single phasecenter is arbitrarily chosen for the entire LPDA), the agreementbetween results obtainedfrom the different approacheswould have beenquite poor. 7e-06
Co-latitude Angle (Degrees)
Figure 3: Gain variation of LPDA with co-latitude angle
2
6c-o.5
sgI
5c-o6
0 1
4.506
2 .g!
3%06
B Ei
2c-06
MODIFIEDAPPROACH
k-06
0
0.5
1
15
2
2.5
3
35
4
A more rigorous approachto computing EF is to begin with the ray model (using the appropriatereflection coef- Figure 4: Comparison of Electric Field Intensity Comficient for horizontal or vertical polarization) and derive puted With Four Different Techniques (instead of Equation 4) E=&i
APPLICATIONTOPXMANTENNAS
.
&qiqp@ i
+
djqqejLqeZ)
,
where F(8) is the complex directivity pattern of the transmit antenna and 81 and 92 are the co-latitude angles of the direct (LOS) and ground bouncerays. This approach, while not including the nearfield nonlinear retardation of phase with distance, should, in principle yield very accurate results. That is, the method is exact in the in the farfield limit. In Figure 4, the electric field as a function of height abovethe ground phtne is plotted for the LPDA in Figure 3. The LPDA is horizontally polarized and situated 1.5 meters above a perfectly conducting ground plane. The horizontal distance from the center of the active region to the observationpoint is 3 metersand the frequency is 453 MHz (The LPDA is designed to cover from 150 m to 2 GHz so this frequency is not near either the upper or the lower operating frequency limits). The electric field is computed first using the NIX-4 “NE” card which can be assumedto be accurate,secondwith GO assuming a dipolar field (Equation 4, but using the boresight gain of the LPDA), third with GO but adjusting the gain to account for changesin gain as a function of elevation angle, and fourth using Equation 14. The agreementbetweenall four approachesis reasonably good but the inclusion of phase information does provide enhancedaccuracy. Finally, it is important to note that the phasecenter implicitly used in Equation 4, is taken to be at the center of the
The P x M antenna,a detailed description of which is presentedin another paper at this conference [l 11,is a particular antenna for which the radiation pattern gain characteristics must be included in the expressionfor q if an accurateassessmentof the antennafactor is to be obtained using a standardsite method. This is true in spite of the fact that the P x M antennais a relatively low gain antenna (4.77 dBi maximum gain) and the fact that the P x M antennahas a well defined phase center and thus no variation of phaseover the far field radiation sphere. An analytical expressionfor ET of the P x M antenna can be derived easily by noting that the farfield electric field in free spacevaries as: E
=
G
[l +cos(0)]
where t3is the latitude angle measuredfrom the boresight axis as shown in Figure 1. In the presenceof a conducting ground plane, the imagesof the electric and magnetic dipole can he usedto obtain the following expressionfor far-held electric field generatedby a P x M antennaradiating 1 pW of power:
599
where {p+l L&,” is the complex reflection coefficient for horizontal or vertical polarization impinging on an imperfect ground with relative permittivity ERand conductivity u: The factor of 2a reflects the gain of 3.0 for the P x M antenna (as opposedto 1.64 for the half-wave dipole antenna).
[4J ANSI C63.5-1988: American National Standard for electromagnetic compatibility-radiated emission measurementsin electromagnetic interference WI) control-calibration of antennas,IEEE, New York 1988. [S] CISPR/mGl (Heirman) 4, August, 1990: Determination of ant&a factors:CentraiOffice of the IEC, rue de VarembB,Geneva,Switzerland.
161Z. Chen and M. D. Foegelle, “A Numerical Investigation of Ground Plane Effects on Biconical Antenna Factor,” 1998 IEEE Symposium Record, August 1998,pp. 802-806 171 H. F. Garn, W. Miillener, and H. Kremser, ‘A Crit-
ical Evaluation of UncertantiesAssociated with the ANSI C63.5 Antenna Calibration Method and a Proposal for Improvements,” 1992 DEEEEMC Symposium Digest.
I81 J. DeMarinis, ‘Antenna Calibration as a Function of Height,” 1987IEEE EMC Symposium,pp. 107-l 14. Figure 5: Predicted electric field for P x M antennaover ground plane
191 H. Iida, S. Ishigami, I. Yokoshima, and T. Iwasaki,
‘Measurement of Antenna Factor of Dipole Antennas on a Ground Plane by 3-Antenna Method,” IEICE Trans. Commum., Vol. E78-B, No. 2, pp. 260267 Feb. 1995.
The predicted electric field for a canonical P x M antenna is plotted in Figure 5 as a function of height above the ground plane. flO1 A. Sugiura, T. Morikawa, T. Tejima, and H. Masuzawa, “‘EMI Dipole Antenna Factors,” IEKE CONCLUSION Tms. Commum., Vol. E78-33,No. 2, pp. 134-139 Feb. 1995. The limitations of the standard site technique for nonJ. S. McLean and G. E. Crook, “P x M Antennas for dipole antennas have been demonstrated. A modifica- r.111 Immunity Testing and other Field GenerationApplition to the technique in which the maximum receivedfield cations,” presentedat IEEE 1999EMC Symposium. is calculated using the complex dire&v&y pattern of the transmit antenna is proposed. Finally, an analytical ex- cm Robert S. Elliot, Antenna Theory and Design, pression for this maximum field for a P x M antenna is Prentice-Hall, Englewood Cliffs, NJ., 1981, pp. given. 329-332. 1131 S. A. Schelkunoff, Antennas Theory and Practice,
John Wiley & Sons,New York 1952,pp 368-370.
References I31 C. R. Paul, Introduction lo Electromagnetic Cornparibility, Wiley Interscience,New York, 1992.
I141 R. W. P. King, meory of Linear AntennaT, Harvard University Press,Cambridge,Massachusett,1956.
PI A. A. Smith, Jr., R. F. German,and 3.3. Pate,“Cal-
culation of Site Attenuation From Antenna Factors,” IEEE Trans. Ekcrromagnetic Comp., Vol. EMC-24,
No. 3, pp 301-315, August 1982. [31 A. A. Smith, Jr., “Standard Site Method for Determining Antenna Factors,” IEEE Trans. Electromagnetic Comp., Vol. EMC-24, No. 3, pp 3 16-322 Au-
gust 1982.
600
“IEEE Standard Definitions For Antennas,” IEEE Tms. Antemw and Propagation, Vol. AP-31, No. 6, Nov. 1983. J. S. McLean, “A Re-examination of the Fundamental Limits of the Radiation Q of Electrically-small Antennas”, IEEE Trans.Ant. Prop., April 1994. R. F. Harrington, Time Harmonic Electromagnetic Fields, McGraw-Hill 1961.
