2168 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 51, NO. 9, SEPTEMBER 2003

A Dual-Band Dual-Polarized Nested Vivaldi SlotArray With Multilevel Ground Plane

Hung Loui, Student Member, IEEE, Jan Peeters Weem, Student Member, IEEE, and Zoya Popovic´, Fellow, IEEE

Abstract—In this paper, we present a systematic approach tothe design, optimization and characterization of a broadband 5:1bandwidth (0.8 to 4.0 GHz) antenna subarray. The array elementis an optimized-taper antipodal Vivaldi slot with a bandwidth of2.5:1. Two such elements of different sizes and with 0.4 GHz (10%of the highest frequency) overlapping bandwidths are arrayed in anested lattice above a multilevel ground plane that shields the feedsand electronics. Return loss, radiation patterns, cross-polarizationand mutual coupling are measured from 0.5–5.0 GHz. This arraydemonstrates E plane patterns with 50 and 45 3-dB beamwidthsin the lower and upper frequency bands, respectively. The couplingbetween the elements is characterized for different relative antennapositions in all three dimensions.

Index Terms—Broad-band array, coupling.

I. INTRODUCTION

M ULTIBAND electronically steerable broad-band arrayscan find applications in a variety of communication and

radar systems. The work presented here is motivated by a non-cryogenic phased array radio-telescope with a very large col-lection aperture (SKA) [1]–[3]. In this case, a 10:1 bandwidth(0.2 to 2 GHz) needs to be accommodated by a dual-polarizedantenna array, while radiation pattern, polarization, return loss,and noise coupling conditions are maintained. In this paper, weaddress the first three antenna parameters. Since each antennaelement will ultimately contain a low-noise amplifier (LNA),the noise coupling in a scanned broad-band array becomes a lim-iting factor. A detailed analysis presented in [4] shows conclu-sively that a broad-band array with elements spaced closer than

at the lowest operating frequency can have very high noisecoupling between element LNAs. This results in noise-domi-nated received signals for many scan angles (“blind” angles).The solution to this problem is to nest arrays that cover sub-bands of the desired frequency range, with each array spacingat or above at the lowest sub-band edge. While the lowersub-band frequency is determined by noise due to mutual cou-pling, the upper sub-band edge is limited by grating lobes asthe array pattern is scanned. In the sections that follow, we

Manuscript received August 30, 2002; revised November 14, 2002. This workwas supported in part by the Netherlands Foundation for Research in Astronomy(NFRA), by the Office of Naval Research and in part by the Office of Secre-tary of Defense under Multidisciplinary Research Initiative Program (MURI)N00014-97-1-1006.

H. Loui and Z. Popovic´ are with the Department of Electrical and ComputerEngineering, University of Colorado, Boulder, CO 80309-0425 USA (e-mail:zoya.popovic@colorado.edu).

J. P. Weem was with the University of Colorado at Boulder, and is now withOptical Communications, Intel, Oakland, CA 94601 USA.

Digital Object Identifier 10.1109/TAP.2003.816336

Fig. 1. Sketch of a 20-element dual-band dual-polarized multilevel nestedantipodal Vivaldi slot antenna array. Each of the subarrays is above its ownground plane. The conducting ground plane height-adjustment rods andmount allow mutual coupling characterization as a function of relative verticalposition of the subarrays.

present a prototype dual-band array, with two bands referred toas lower-band (LB) (0.8 to 2.0 GHz) and upper-band (UB) (1.6to 4.0 GHz). The lower band is covered by dual-polarizationcrossed Vivaldi slot elements referred to as lower-band subarray(LBSA), and the upper band is covered by a scaled upper-bandsubarray (UBSA). The subarray ground plane levels need to alsobe scaled and a multilevel periodic ground plane is introduced,as suggested and theoretically analyzed in [5].

The dual-band dual-polarized nested antipodal Vivaldi arrayis shown in Fig. 1 together with the mounts used to vary UBSAsheights. In an attempt to describe the design and characterizationof this array, the paper has the following outline.

• Section II describes the design, simulations and measure-ments of a taper-optimized antipodal Vivaldi slot element,as well as coupling between two cross-polarized elementsas a function of their relative position.

• Section III describes design and performance of a dual-polarized sub-band subarray.

• Section IV presents performance of the dual-band array,including mutual coupling characterization as a functionof relative ground plane height.

• Section V presents a discussion on nested multiband ar-rays, based on obtained experimental results.

0018-926X/03$17.00 © 2003 IEEE

LOUI et al.: DUAL-BAND DUAL-POLARIZED NESTED VIVALDI SLOT ARRAY 2169

Fig. 2. Sketch of a taper-optimized antipodal Vivaldi antenna that operatesfrom 1.6–4.0 GHz. The solid line is the outline of the metallization of the frontside of the substrate while the backside is shaded. The dashed lines representthe rotated taper. All dimensions are in millimeters.

Fig. 3. Simulated return loss of a taper-rotated small antipodal Vivaldi antenna(Fig. 2 without the microstrip to printed-twinline balun) as a function of bothfrequency and exponential taper parameter�. The dotted line surrounds the�10-dB match. The maximum�10-dB bandwidth is achieved at� = 400

(solid line).

II. A NTIPODAL VIVALDI SLOT ELEMENT

Continuously scaled traveling-wave antennas such as theexponentially-tapered Vivaldi slot have been demonstrated toexhibit broad-band operation in both return loss and radiationpatterns [6]. In an array of such elements, bandwidth perfor-

Fig. 4. Simulated versus measured return loss of the optimized-taper� = 400

small antipodal Vivaldi antenna with the microstrip to printed-twinline balun.The�10 dB bandwidth of the antenna increases from 2.0–4.0 GHz to 1.6–4.0GHz when the rotation of the metallization given by (2) is implemented.

mance is limited by the element spacing. Theoretical analysisof tapered slot radiation characteristics on dielectric substratescan be found in [7]–[9]. Variants of the Vivaldi element havebeen documented in [10]–[12]. Commonly, printed Vivaldi an-tennas feature metallization on only one side of the substrate,which requires a high relative permittivity of the dielectricsubstrate to achieve the desired 50-impedance at the feedpoint. In large arrays where the cost of high permittivity di-electric substrates is an issue (such as the case for SKA),lower permittivity substrates can be substituted by adoptinga Vivaldi variant using an antipodal feed [11].

The antipodal Vivaldi element shown in Fig. 2 is built on aDuroid RT/5880 substrate with a relative dielectric constant of2.2 and thickness of 0.508 mm. The radiation arm tapers aredetermined by

(1)

and shifted to the left/right by half the microstrip line width withrespect to the origin. The tapers are then rotated by

(2)

so that the feed line near the origin has infinite slope. This ro-tation effectively extends the 10-dB matching to lower fre-quencies. The backside metallization is the mirror image of thefront with the exception of the ground taper. The parametercontrols the extent of the taper whilerepresents the height ofthe inner or outer taper. Zeland’s method-of-moments(MoM) CAD tool IE3D was used to optimize the return loss ofthe element as a function of in (1) with and in meters.The results are shown in Fig. 3 whereis varied from 35–2000,corresponding to nearly linear and practically right angle tapers,respectively. The dashed line in Fig. 3 indicates a range of fre-quencies and tapers for which the return loss is10 dB or better.

This optimization allows us to choose as the op-timal in terms of 2:1 VSWR between 1.6 and 3.5 GHz. The re-sulting element shape and its measured versus simulated returnloss is shown in Figs. 2 and 4. A computer-controlled anechoic

2170 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 51, NO. 9, SEPTEMBER 2003

Fig. 5. Measured E plane (a) and H plane (b) normalized power radiationpatterns of the upper-band Vivaldi antenna from 1.5–3.5 GHz. The black contourdenotes the 3-dB beamwidth.

chamber was used to measure the E and H plane radiation pat-terns, and the measured copolarized patterns are shown in Fig. 5.Shallow nulls in the H plane at a few frequency points agree withsimulations using IE3D.

From Figs. 4 and 5, we conclude that the taper-optimized an-tipodal Vivaldi slot antenna is suitable in both return loss andradiation pattern for the frequency range of 1.5 to 3.5 GHz. Inorder to design a dual-polarized array of Vivaldi elements [13],the copolarized (CP) and cross-polarized (XP) coupling versuselement separation was measured using two identical antennasand the coordinate system shown in Fig. 6. The mutual cou-pling was measured using a 4.0-GHz, 0-dBm source connectedto Antenna 1 and a power meter connected to Antenna 2. Theresulting parameter as a function of separation distances in

and is shown in Fig. 7. The measurements indicate that thecoupling is at most 20 dB at 4 GHz where the two antennasare parallel or perpendicular to each other and their substratestouching. For separations greater than 20 mm, coupling is not a

Fig. 6. Measurement configurations for XP (a) and CP (b) coupling betweentwo identical Vivaldi elements. Antenna 2 was translated in they and x

dimensions, respectively. All dimensions are in millimeters.

Fig. 7. Measured coupling coefficient between two Vivaldi antennas at 4 GHzas a function of distance. The red surface represents CP coupling and the bluesurface depicts XP coupling. Point A indicates the relative element positionchosen in the implementation.

strong function of position and the XP coupling is much lowerthan CP coupling. Therefore, in the design of the dual-polar-ized sub-band subarray, four elements, two for each orthogonalpolarization are arranged in a cross (Fig. 8), with coupling cor-responding to point A labeled in Fig. 7.

III. D UAL-POLARIZED SUB-BAND SUB-ARRAY

Fig. 8 shows the top view of a dual-polarized subarray. Theports of the four feeds are labeled as they are referred to insubsequent measurement results. In a nested array (Fig. 1), thesubarray can be thought of as the fundamental nesting struc-ture. Following similar procedures as outlined in Section II, adual-polarized LBSA element scaled by a factor of 2 is designedand operates in a lower frequency band. The return losses ofboth elements are shown in Fig. 9. This figure shows that thereturn loss of a single Vivaldi element nested in its relevant sub-array is below 10 dB within its specified bandwidth of op-eration. Due to the arrangement of elements in the subarrays,there will be CP coupling (coupling between elements on the

LOUI et al.: DUAL-BAND DUAL-POLARIZED NESTED VIVALDI SLOT ARRAY 2171

Fig. 8. Sketch of a dual-polarized four-element antipodal Vivaldi upper-bandsubarray (top view). All dimensions are in millimeters.

Fig. 9. MeasuredjS j of a single element in a LBSA or low frequencyantenna and UBSA or high frequency antenna, respectively.

same line) and XP coupling (coupling between elements per-pendicular to each other). Coupling measurements are made byterminating two of the four ports with matching loads while ob-serving the transmission characteristics of CP ports 1–3 and XPports 1–2. The results are shown in Fig. 10.

The coupling between LBSA elements is in general less thanthat of UBSA especially at mid to high frequencies. This is dueto the fact that at higher frequencies, the field distribution islocalized near the feeds, which are electrically further apart inthe LBSA. Also evident is the general trend of decreasing cou-pling as a function of frequency observed in both the LBSA andUBSA shown in Fig. 10. In either case, the CP and XP couplinglevels are of the same order of magnitude and at most9 dB inmagnitude.

Fig. 10. Separately measured copolarized and cross-polarized coupling in aUBSA and a LBSA.

IV. DUAL-BAND ARRAY WITH DUAL-POLARIZATION AND

MULTILEVEL GROUND

A. Array Layout

Parametric studies and simulations of arrays of iden-tical Vivaldi elements arranged in a row-by-row andcolumn-by-column egg-crate layout have been documented in[14]–[16].

The top view of the dual-band array with dual polarizationis shown in Fig. 11. Without the LBSA nested in the center(Fig. 11), the return loss and antenna patterns of the four UBSAwould be similar to those already in the literatures above and aretherefore not repeated. Here we examine the effect of nestingtwo arrays designed for different frequency bands. In specific,the element return loss and array radiation patterns across theentire operating bandwidth are of interest.

The half wavelength spacing (18 cm) for the LBSA is at 833MHz, with the corresponding return loss of10 dB (Fig. 9)and in-band XP coupling of 20 dB (Fig. 10). The half wave-length spacing (8 cm) for the UBSA is at 1.875 GHz, with thecorresponding return loss of18 dB (Fig. 9) and in-band XPcoupling of 15 dB (Fig. 10). The two arrays together cover abandwidth of 5:1, from 0.8–4 GHz. At the highest frequency,the electrical spacing between upper-band elements is 1.07.

B. Cross-Band Coupling

In Section II, the coupling between two identical elementswas quantified as a function of the horizontal spacing (an ex-ample is given in Fig. 7). In a dual-band array, an additionalparameter that can affect coupling is the relative distance in thevertical direction. Copolarized (1) and cross-polarized (2) cou-pling between closest elements of the LBSA and UBSA as afunction of frequency and relative height were measured usingthe HP8719ES network analyzer with a coaxial calibration. Re-ferring to Fig. 11, the measurements were made at ports 3 and 5for case (1) and at ports 4 and 5 for case (2). In these measure-ments, only one of the UBSAs is raised while the other three arekept at the lowest vertical position.

2172 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 51, NO. 9, SEPTEMBER 2003

Fig. 11. Sketch of top view of a 20-element dual-polarized antipodal fedVivaldi antenna array. All dimensions are in millimeters.

(a)

(b)

Fig. 12. (a)jS j of a LBSA element and (b)jS j of a UBSA element versusUBSAs height (cm) above LBSAs ground plane. The black contours surround�10-dB match. The horizontal line represents a relative height of 12 cm betweenthe ground plane of the UBSA and that of the LBSA.

Fig. 13. (a) CP couplingjS j and (b) XP couplingjS j between closestUBSA and LBSA elements versus UBSA height (cm) above LBSAs groundplane.

The return loss of the lower-frequency element as a functionof vertical position of the higher-frequency element is shown inFig. 12(a). As can be seen from the measurements, the returnloss of the LBSA is not a strong function of vertical po-sition until the UBSAs ground plane is completely above thehighest points of the LBSA (about 24 cm), acting as a scattererin the near field. Fig. 12(b) shows the measured return loss ofthe upper-frequency element as the UBSAs vertical position isvaried. The dependence on height is stronger than in the pre-vious case as the LBSA acts as a finite reflector in the near fieldof the UBSA. In this case, in the operating band of the UBSA,the optimal height above the main ground seems to be the middleheight range where is not a strong function of frequency.Therefore, a compromise in height of 12 cm is chosen for thefinal topology and pattern measurements.

The measured CP coupling between ports 3 and 5 shown inFig. 13(a) indicates that the coupling is always smaller than

13 dB for the chosen height of 12 cm; it is below20 dBfor most of the frequency range. In general, below 15 cm, CPcoupling is not a strong function of height and improves with in-creasing frequency. The cross-polarized coupling between ports4 and 5 shown in Fig. 13(b) is more dependent on height but stillbelow 15 dB over the entire range.

LOUI et al.: DUAL-BAND DUAL-POLARIZED NESTED VIVALDI SLOT ARRAY 2173

After determining the useful range of relative ground planeheights, all four of the UBSAs were raised simultaneously from9–18 cm in order to see the effect of relative position of severalUBSAs on the relevant parameters. The measurements showthat characterization of the array performance as a function ofsingle UBSA height is sufficient. As an example, the measuredreturn loss of the UBSA element and coupling coefficient

are shown in Fig. 14.

C. Radiation Patterns

Several radiation pattern measurements were made of thenested Vivaldi array. The feed network was assembled usingequal-length cables and matched power dividers. First, patternswere measured for a single two-element linear polarizedupper-band array (UBA), and compared to the patterns ofthe same elements embedded in a dual-polarized UBSA, andsubsequently placed in the complete dual-band dual-polarizedarray. From Fig. 15, it can be seen that the UBA pattern is notsubstantially affected by the other array elements in the chosenarray design. The asymmetry in the sidelobe levels and thesquint are the results of the unbalanced tapered feed transition.The 3-dB beamwidth is about 45in the E plane and 60inthe H plane. The H plane pattern of the UBA is broader thanits E plane pattern, therefore when placed in the full array thecoupling due to other elements in the H plane is stronger anda noticeable improvement in directivity can be observed inFig. 15(b).

The LBSA patterns were also measured at the frequency cor-responding to the element spacing of , for two differentUBA heights, Fig. 16. The measurements demonstrate that theLBSAs E plane pattern is not sensitive to the height displace-ment of the UBSAs.

D. Effect of the Feed

At lower frequencies, there is a tilt in the E plane pattern. Inorder to understand this effect, the radiation patterns for severalVivaldi slots with identical tapers were simulated: a) a Vivaldiprinted on one side of the substrate with no balun feed; (b) anantipodal (two-sided) Vivaldi with no balun; and c) an antipodalVivaldi with a balun as depicted in Fig. 2.

The simulation results in Fig. 17 show that the main-beam tiltis due to the balun feed only, and is therefore not a fundamentalproperty of the antipodal Vivaldi slot. At higher frequencies theground taper of the microstrip to twinline balun is electricallylong; this facilitates the gradual transition to the radiation armstherefore provides a better feed. Other feed configurations suchas ones in [12], [13] can also be used to avoid the slight patterntilt at lower frequencies.

V. DISCUSSION ANDCONCLUSION

This paper presents the design and detailed experimentalcharacterization of a dual-band dual-polarized nested an-tipodal Vivaldi slot antenna array. We showed that the nestedsub-band subarray approach could be successfully optimizedfor element-to-element coupling while preserving the subarray

Fig. 14. (a)jS j of a UBSA element and (b) CP couplingjS j versus UBSAheight in centimeters above LBSAs ground plane. All four UBSAs were raisedtogether from 9–18 cm.

pattern and input impedance in meeting broad-band require-ments of next generation broad-band phased arrays. The mainconclusions can be summarized as follows:

• A systematical approach to the design, optimization andcharacterization of a dual-band dual-polarized multi-level nested antipodal Vivaldi antenna array has beendeveloped.

• A 5:1 bandwidth in and pattern can be achieved withtwo scaled Vivaldi elements placed in a nested cross con-figuration above two-level ground.

• The array performance depends strongly on couplingbetween elements in the same band and in the twosub-bands. The coupling can be kept at a low level(below 20 dB) by optimizing the relative positions ofthe elements in all three dimensions while maintaininga generally good return loss match in both sub-bandsfor both polarizations.

• Coupling characterization of a single subarray at itsnesting level with respect to its immediate neighboringdiverse sub-band subarray is an adequate and efficientway of quantifying the overall coupling behavior due toall subarrays at the same nesting level.

2174 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 51, NO. 9, SEPTEMBER 2003

Fig. 15. Measured normalized (a) E plane and (b) H plane patterns of astand-alone two-element linear-polarized upper-band array (UBA) and itssubsequent settings in an UBSA and LBSA at 1.875 GHz. The axis of rotationis at the center of the UBA. The measured cross-polarized patterns for a singleelement are added only for comparison.

Fig. 16. Measured normalized E plane patterns of the LBSA in the presenceof four UBSAs at 833 MHz. All four UBSAs were raised to 1 cm and 12 cmabove the LBSAs ground plane, respectively. The axis of rotation is at the centerof the LBSA.

Fig. 17. Simulated E plane directivity patterns at (a) 2 GHz and (b) 4 GHz.The solid lines correspond to a symmetrical one-sided Vivaldi slot. The dashedlines correspond to identical tapered antipodal antennas with and without themicrostrip-to-twinline balun.

• Nested arrays that cover sub-bands with about 2:1 band-widths can effectively mitigate noise coupling in scannedactive arrays with LNAs in each element [4].

• This array design includes a ground plane that shieldsthe feeds and electronics. This is important for adaptivephased arrays such as the square kilometer array (SKA)[1], which was the main motivation of this work.

The array bandwidth is much larger than the currently avail-able digital signal processor (DSP) electronics bandwidths, andonly relatively slow adaptive algorithms are possible. Althoughthis is adequate for radio astronomy where integration timesare long, this antenna array approach can also be applied tosmart antennas utilizing broad-band analog processing. An ex-ample of such a principle component analysis (PCA) systemwith multi-GHz signal bandwidth capability is described in [17].The integration of a multiband array with a broad-band dynamicholographic processor is the topic of continued work.

ACKNOWLEDGMENT

The authors would like to thank Dr. A. Van Ardenne fromthe Netherlands Foundation for Radio Astronomy and Prof. D.

LOUI et al.: DUAL-BAND DUAL-POLARIZED NESTED VIVALDI SLOT ARRAY 2175

Filipovic at the University of Colorado for their helpful inputsand useful discussions.

REFERENCES

[1] A. V. Ardenne, “The technology challenge for the next generation radiotelescopes,”Perspectives on Radio Astronomy: Technologies for LargeAntenna Array, pp. ix–xviii, 1999.

[2] P. V. Genderen, “State-of-the-art and trends in phased array radar,”Per-spectives on Radio Astronomy: Technologies for Large Antenna Array,pp. 1–10, 1999.

[3] J. P. Weem and Z. Popovic, “Vivaldi antenna arrays for SKA,” inProc.Antennas and Propagation Soc. Int. Symp., vol. 1, 2000, pp. 174–177.

[4] , “A method for determining noise coupling in a phased array an-tenna,” inProc. Microwave Symp. Dig., IEEE MTT-S Int., vol. 1, 2001,pp. 271–274.

[5] B. M. Notaros, B. D. Popovic, J. P. Weem, R. A. Brown, and Z. Popovic,“Efficient large-domain Mom solutions to electrically large practical EMproblems,”IEEE Trans. Microwave Theory Tech., vol. 49, pp. 151–159,Jan. 2001.

[6] P. J. Gibson, “The Vivaldi aerial,” inProc. 9th Eur. Microwave Conf.,Brighton, U.K., Sept. 1979, pp. 101–105.

[7] R. Janaswamy and D. H. Schaubert, “Analysis of the tapered slot an-tenna,”IEEE Trans. Antennas Propagat., vol. AP-35, pp. 1058–1064,Sept. 1987.

[8] B. Stockbroeckx and A. V. Vorst, “Copolar and cross-polar radiation ofVivaldi antenna on dielectric substrate,”IEEE Trans. Antennas Prop-agat., vol. 48, pp. 19–25, Jan. 2000.

[9] P. Jia, L. Y. Chen, N. S. Cheng, and R. A. York, “Design of waveguideFinline arrays for spatial power combining,”IEEE Trans. MicrowaveTheory Tech., vol. 49, pp. 609–614, Apr. 2001.

[10] D. H. Schaubert and T.-H.Tan-Huat Choi, “Wideband Vivaldi arrays forlarge aperture antennas,”Perspectives on Radio Astronomy: Technolo-gies for Large Antenna Array, pp. 49–58, 1999.

[11] E.Ehud Gazit, “Improved design of the Vivaldi antenna,”Proc. Inst.Elect. Eng., pt. H, vol. 135, no. 2, pp. 89–92, Apr. 1988.

[12] J. D. S. Langley, P. S. Hall, and P. Newham, “Balanced antipodal Vivaldiantenna for wide bandwidth phased arrays,” inProc. IEEE AntennasPropagation, vol. 143, Apr. 1996, pp. 97–102.

[13] J. P. Weem, B. M. Notaros, and Z. Popovic, “Broadband element arrayconsiderations for SKA,”Perspectives on Radio Astronomy: Technolo-gies for Large Antenna Array, pp. 59–67, 1999.

[14] J. Shin and D. H. Schaubert, “A parameter study of stripline-fed Vi-valdi notch-antenna arrays,”IEEE Trans. Antennas Propagat., vol. 47,pp. 879–886, May 1999.

[15] E. Thiele and A. Taflove, “FD-TD analysis of Vivaldi flared hornantennas and arrays,”IEEE Trans. Antennas Propagat., vol. 42, pp.633–641, May 1994.

[16] H. Holter, T.-H.Tan-Huat Chio, and D. H. Schaubert, “Experimental re-sults of 144-element dual-polarized endfire tapered-slot phased arrays,”IEEE Trans. Antennas Propagat., vol. 48, pp. 1707–1718, Nov. 2000.

[17] E. Fotheringham, S. Romisch, P. C. Smith, D. Popovic, D. Z. Anderson,and Z. Popovic, “A lens antenna array with adaptive optical processing,”IEEE Trans. Antennas Propagat., vol. 50, pp. 607–617, May 2002.

Hung Loui (S’02) received the B.S.E.E. and B.M.(piano performance) degrees from the University ofColorado at Boulder, in 2001, and is currently pur-suing the Ph.D. degree in electromagnetics.

From 1997 to 2001, he did undergraduate researchat the Laboratory for Atmospheric and Space Physics(LASP) where he was responsible for the design, pro-duction and characterization of the instrumental sci-entific data processing (DSP) boards aboard the SolarRadiation and Climate Experiment (SORCE) satel-lite. While at LASP, he was also involved in the de-

sign and integration of the (CHAMP) microscope camera with the prototypeK9 Mar’s rover and participated in the dust particle collisions in planetary rings(COLLIDE) project. In music, he has performed classical piano concertos withboth the Grand Junction Symphony, Grand Junction, CO, and the university or-chestra.

Jan Peeters Weem(S’99) received the B.S. degree in mathematics from OregonState University, Corvallis, in 1994, the M.S. degree in applied mathematics,and the Ph.D. degree in applied electromagnetics both from the University ofColorado, Boulder, in 1996 and 2001, respectively. His Ph.D. dissertation wason noise coupling in broad-band receive phased arrays for radio astronomy.

From 1996 to 1998, he was involved with numerical solutions for EM prob-lems at the University of Colorado. In 2001, he joined Intel, Oakland, CA, wherehe is now working on high-speed optical communication link components.

Zoya Popovic (S’86–M’90–SM’99–F’02) receivedthe Dipl.Ing. degree from the University of Bel-grade, Yugoslavia, in 1985 and the M.S. and Ph.D.degrees from the California Institute of Technology,Pasadena, in 1986 and 1990, respectively.

She is currently a Professor of Electrical andComputer Engineering at the University of Colorado,Boulder. She coauthoredIntroductory Electromag-netics (Englewood Cliffs, NJ: Prentice-Hall, 1999)and coedited Active and Quasi-Optical arraysfor Solid-state Power Combining(New York:

Wiley, 1997). Her research interests include microwave and millimeter-wavequasi-optical techniques and active antenna arrays, high-efficiency microwavecircuits, RF photonics, and antennas and receivers for radio astronomy.

Dr. Popovicwas a recipient of the 1996 URSI International Issac Koga GoldMedal and was a 1993 NSF Presidential Faculty Fellow. She was the recipientof the 1993 IEEE Microwave Theory and Techniques Society (IEEE MTT-S)Microwave Prize for pioneering work in quasi-optical grid oscillators. She wasnamed the Eta Kappa Nu Professor of the year by her University of Coloradostudents. In 2000, she was the recipient of a Humboldt Research Award fromthe German Alexander von Humboldt Stiftung. In 2001, she received HP/ASEEFrederick Emmons Terman Award for simultaneous achievements in teachingand research.