DUE to their low mass, possible conformity, and simple

Transcription

1 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 45, NO. 10, OCTOBER An Experimental Study on 2 2 Sequential-Rotation Arrays with Circularly Polarized Microstrip Radiators Ulrich R. Kraft, Senior Member, IEEE Abstract The radiation properties of sequential-rotation arrays with single-feed circularly polarized patch radiators are investigated experimentally with the goal to maximize the bandwidth for an XPD 30 db. Based on a particular crosspolarization discrimination (XPD) description for such arrays, the impacts of element polarization, mutual coupling, limited symmetry, and excitation errors on the array cross polarization are presented. It is shown that compact configurations can be obtained, which provide an on-axis XPD of 30 db over up to 2.0% bandwidth and of 35 db over up to 1.0%. These arrays exhibit copolarized sidelobes below 018 db and off-axis peak cross-polarization levels below 020 db. Index Terms Microstrip arrays. Fig. 1. Geometry of the SRA and its microstrip patch radiator. I. INTRODUCTION DUE to their low mass, possible conformity, and simple manufacturing, microstrip patch antennas are interesting candidates for a variety of applications whereby their main disadvantage is a rather limited bandwidth, which is particularly critical for circular polarization with severe cross-polarization requirements. Depending on the acceptable complexity of the antenna system, different solutions to this problem exist that implement an optimization of the cross-polarization suppression at different levels of the antenna. If moderate axial ratios and single radiators or very small arrays are required, an attractive solution are dual-linear polarized patches on thick low substrates with one feed network per element. In this case, axial ratios of 2 db have been obtained over an excellent bandwidth of 30% using an electromagnetically coupled -slot patch and a feed network with isolating, i.e., matched Wilkinson power dividers [1]. For larger arrays, the element and feed-network complexity becomes critical, i.e., the suppression of cross-polarization has to be performed either by the elements themselves or by an optimized feeding at lower antenna levels. Here, singlefeed circularly polarized patches on thin substrates permit the lowest complexity but provide only % bandwidth for an axial ratio of 2 db [2]. This performance is improved by the introduction of pairs or other sequential-rotation configurations that are excited by one simple corporate-type feed network per group and that provide bandwidths of 2% 3% for axial ratios down to 1 db [3] [5]. Such configurations are still fairly simple but suffer from multiple reflections, feednetwork radiation, and feed-phase errors, which deteriorate Manuscript received October 7, 1996; revised April 17, The author is with Daimler-Benz Aerospace AG-Dornier Satellite Systems, Munich, D Germany. Publisher Item Identifier S X(97)07378-X. Fig. 2. Measured radiation pattern of an isolated patch at f = 2:945 GHz. fully optimized version (Type A, c =2:0mm) moderately optimized version (Type B, c =1:5mm). the cross-polarization suppression and bandwidth [5]. If a cross-polarization discrimination (XPD) of 30 db (i.e., an axial ratio of 0.5 db) is required over a considerable bandwidth, a 2 2 sequential-rotation array (SRA) is an interesting solution since it offers an intrinsic suppression of higher order modes [5], [6] and permits an addition of individual XPD contributions from the elements and the feed network provided that circularly polarized radiators are used [7]. For this case, single-feed circularly polarized patches again minimize the complexity and avoid the increased off-axis cross-polarization and gain degradation caused by linearly polarized elements [6] though they are more sensitive to mutual coupling than dual- or linearly polarized alternatives. In order to avoid bandlimiting effects, however, the 2 2 SRA needs a separated feed network based on broad-band isolating (matched) power dividers such as Wilkinson splitters or hybrid couplers [1], [7] X/97$ IEEE

2 1460 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 45, NO. 10, OCTOBER 1997 (c) Fig. 3. Measured structure-factor (XPD st )of222sra s plus on-axis XPD of corresponding isolated microstrip patches. Type A patches d = 0:4 0:6. Type A patches d =0:7 0:8plus isolated Type A patch. (c) Type B patches d =0:7plus isolated Type B patch 2 XPD of single patch: XPD st average of SRA; XPD st measurement with shorts at the elements two and four; XPD st measurement with shorts at the elements one and three. Within this contribution, such 2 2 SRA s are investigated experimentally with the goal to maximize the bandwidth for a XPD db. Starting with a brief summary of general SRA polarization properties in Section II and considerations on the individual patch radiators in Section III, the XPD of different SRA configurations is described in Sections IV and V. Finally, the radiation pattern are discussed in Section VI. II. SUMMARY OF GENERAL SRA POLARIZATION PROPERTIES Due to their special symmetry, all 2 2 sequential-rotation arrays exhibit some radiator-independent polarization properties, which lead to the particular XPD description used in the following sections. Though this description has been presented already in [7], its essentials and limits are briefly summarized in this section to support the understanding of the subsequently reported results. The geometry of the investigated 2 2 SRA and its microstrip patch radiator is shown in Fig. 1, together with the applied coordinate system. The SRA consists of four identical patch radiators that are sequentially rotated by multiples of 90 about the axis and are fed at the indicated feed points by four complex feed voltages to, generated by the associated feed network in presence of all possible coupling and mismatch effects. Because of its symmetry, any SRA provides perfect left-hand or right-hand circular polarization (LHCP, RHCP) in on-axis direction (i.e., ) ifitis excited by a perfect feed-vector or, respectively. Assuming LHCP as the copolarized component, an arbitrary excitation caused by amplitude and phase errors of the feed network, coupling between network and radiators, mismatch effects, etc., leads to an on-axis XPD of the SRA, which is given by with and XPD db XPD db XPD db (1) XPD db log XPD db log (3) Herein, XPD, XPD, and denote the XPD contributions of the feed network and the radiating structure and the left- and right-hand components of the radiated electrical field, respectively. Since the individual XPD contributions add (2)

3 KRAFT: EXPERIMENTAL STUDY ON 2 2 SEQUENTIAL-ROTATION ARRAYS 1461 optimized patch (Type A), which provides a maximum XPD on the order of 35 db, but is rather sensitive to manufacturing tolerances and deteriorations due to mutual coupling or to apply a more robust moderately optimized version with a maximum XPD of about 15 db (Type B). Whereas the Type B patch would be worthless as a single radiator or an element of conventional arrays, it can be used in the SRA because of the XPD-addition effect whereby its XPD for off-resonance frequencies can be similar to the Type A case depending on the required bandwidth. For reasons of comparison, both patch types have been realized and tested and have been used subsequently for the investigations on array performances. As shown in Fig. 1, the geometry of the patch radiators has been based on a circular disk resonator with a diameter and two symmetrical cut-out perturbations with the width and the depth, which excite two orthogonal modes with the desired 90 phasing [2]. The patches have been realized with mm and mm using a Duroid 5870 substrate with an of 2.33 and a thickness of mm and have been excited by a 1.4-mm semi-rigid coaxial cable at a distance a from the center of the disk and an angle of 45 from the perturbations. The parameters and have been optimized to obtain a good input match and XPD for a center frequency of GHz. The conditions for an optimized XPD are given in [2] to with (4) Fig. 4. Evaluation of XPD st measurements for the SRA s with Type A patches and spacings of d = 0:4 0:8. Shift of measured optimum SRA frequency w.r.t. optimum element frequency (i.e., optimum frequency of ideal SRA without mutual coupling). Measured maximum XPD st of SRA at corresponding optimum frequency. directly, fairly high XPD figures can be obtained over a larger bandwidth because a decreasing XPD contribution from the narrow-band patches at the band edges can be compensated by a corresponding contribution from the more wide-band feed network. Here, the only limiting factor is the limited symmetry of a practical SRA given by inevitable manufacturing tolerances, which ultimately determines the maximum achievable XPD since the cross-polarization suppression effect depends on a cancellation of cross-polarized components from the individual radiators. XPD can be determined by XPD measurements of a symmetrical four-probe horn connected to the SRA feed network, whereas XPD can be measured as the XPD of the SRA when two opposite radiators (e.g., 2 and 4) are shorted to the ground plane and the other two (1 and 3) are fed in antiphase [7]. III. CONSIDERATIONS ON THE INDIVIDUAL PATCH RADIATORS To optimize the overall XPD performance of the SRA, its radiating structure should provide a large structure factor XPD over the desired bandwidth, i.e., the XPD of the radiators in presence of coupling effects should be maximized over that band. Here, the basic options are to use a fully wherein,, and denote the area and factor of the disk resonator and the total area of the perturbation segments, respectively. To evaluate these conditions, the factor of an equivalent unperturbed resonator has been measured to about 34 to 36 using the relation wherein and denote the resonance frequency and the bandwidth for an input power reflection coefficient of db [8]. This result implies an optimum cut-out depth of about mm, which has been used for the Type A patch version. The radiation pattern of a Type A patch with mm and mm measured at GHz is shown in Fig. 2. Due to the full optimization, an on-axis XPD of about 40 db and an input power reflection of 19.8 db are obtained. The corresponding bandwidths are 0.1%, 0.2%, and 0.27% for on-axis XPD figures better than 35.0 db, 30.8 db, and 24.8 db and 2.6% for input power reflections better than 15 db. These results imply a general suitability of the radiator for SRA bandwidths up to 2.6% without a gain degradation exceeding 0.15 db, whereas the patch XPD decreases already to about 8 10 db at the edges of a 2% band. The corresponding radiation pattern of a Type B patch with mm and mm is also presented in Fig. 2. Due to the moderate optimization of, the on-axis XPD is limited to 15.9 db whereas the input power reflection is still 18.9 db with a bandwidth of 2.0% for figures better than 15 db. This implies that the patch can be used for SRA bandwidths up to (5)

4 1462 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 45, NO. 10, OCTOBER 1997 (c) (d) Fig. 5. Measured on-axis XPD of the SRA s plus XPD prediction according to (1). Type A patches d = 0:4 0:6. Type A patches d =0:7. (c) Type A patches d =0:8. (d) Type B patches d =0:7: measured XPD; predicted XPD. 2% whereby its XPD decreases to 7 9 db at the band edges. In comparison, both patch types provide a similar worst-case XPD performance if 2% bandwidth are required whereas Type A becomes rapidly superior for smaller bands. IV. XPD PROPERTIES OF THE SRA RADIATING STRUCTURE To study the impact of mutual coupling and manufacturing tolerances on the polarization of the radiators and the electrical symmetry of the SRA, a number of arrays with Type A patches and element spacings of mm ( at GHz) as well as one array with Type B patches and mm (0.7 ) have been manufactured on square Duroid 5870 substrates with mm and tested over a frequency range of GHz. Here, the impact of mutual coupling in the array can be determined from a comparison of measured XPD figures (which include such effects) with the XPD of a corresponding single radiator. Unsymmetries can be tested by a comparison of two subsequent XPD measurements at each frequency that are performed for alternative shortings of the elements and [7]. The results of these measurements are shown in Fig. 3 and for SRA s with Type A patches and Fig. 3(c) for the Type B version, whereby measured XPD figures of the corresponding isolated patches have been added to Fig. 3 and (c) for reasons of comparison. It is evident from the figures that mutual coupling significantly changes both the optimum frequency and the achieved maximum XPD of the Type A version, whereas its impact is less severe for the Type B array, as expected. This effect is due to a change of the mode balance on the patches, which causes a detuning and prevents the optimum balance of the initial Type A patch. In case of the Type B version, such optimum conditions do not exist even for the isolated radiator, leading to no noticable change in the optimum frequency and to a deterioration of XPD versus the initial radiator XPD in the same order as reported earlier for a SRA with helix radiators [7]. It is also evident from the figures, that manufacturing tolerances have a significant impact on the electrical symmetry of the Type A arrays (which is most critical close to the optimum frequency). Here, minimum variations of 0.3 db, typical figures of 0.8 db (0.7 )to 1.0 db (0.4 ), and maximum cases of 1.5 db (0.6 ) to 1.9 db (0.8 ) occur. Though these variations are smaller ( 0.5 db) for the less-critical Type B array, they are always considerably higher than the 0.2 db previously measured for nonresonant helix radiators [7], which can lead to significant limitations for the achievable overall XPD of a microstrip SRA. An evaluation of the frequency shift and the achieved maximum XPD is shown in Fig. 4 and for the Type A array together with reference figures given by an ideal SRA

5 KRAFT: EXPERIMENTAL STUDY ON 2 2 SEQUENTIAL-ROTATION ARRAYS 1463 Fig. 6. Measured radiation pattern of the SRA s in one pricinpal ( = 0 ) and one diagonal ( =45 ) plane. SRA s are operated at the corresponding frequency for optimum on-axis XPD. Type A patches d =0:4,f=3:000 GHz. Type A patches d =0:6,f=2:930 GHz: LHCP; RHCP. without mutual coupling, i.e., a corresponding single patch. It is evident from Fig. 4 that the presence of coupling causes a significant, nonmonotonic frequency shift that can easily be in the same order as the expected overall bandwidth of 2% and is always considerably larger than the 30-dB XPD bandwidth of a single patch. Whereas a maximum of 1.9% occurs for the smallest spacing of (as expected), the shift exhibits a minimum around 0.5 followed by a negative increase to 0.7% at 0.6 and a slow decrease to 0.50% at 0.7 and 0.35% at 0.8. This implies the need for an appropriate frequency correction of the patch design, which is most critical for small spacings but still necessary even for figures of In a similar way, the maximum XPD figures are affected, where a very strong coupling reduces the ideal XPD of 40 db to only 6.3 db for the smallest spacing of 0.4 (i.e., the individual patches generate almost linear polarization in this case). This figure improves slowly and nonmonotonic to 12.1 db (0.5 ), 23.0 db (0.6 ), 18.6 db (0.7 ), and 25.9 db (0.8 ) but does not achieve the order of magnitude given by the ideal case. In principle, the strong dependence of the array on mutual coupling, which needs to be considered during the array design and limits the benefit of a fully optimized patch could be reduced if dual-linear patches with feed networks on element-level or single-linear patches would be used. This change would imply, however, a significant increase in complexity for the former case and a missing XPD contribution from the elements plus gain loss and increased off-axis cross polarization for the latter type, which reduces the benefits of such alternatives considerably. V. OVERALL ON-AXIS XPD OR THE COMPLETE 2 2 ARRAYS Though mutual coupling can significantly affect optimum frequency and maximum XPD figures (as shown in the previous section) the minimum XPD achieved over 2% bandwidth is still about 8 10 db for Type A arrays with spacings of and about 5 db for the Type B version. This implies the need for a feed network with XPD db in the former and XPD db in the latter case, if an overall XPD db should be maintained over the full bandwidth [cf. (1)]. Corresponding networks based on commercially available components have been built and tested, whereby a 90 stripline hybrid followed by two 180 hybrids and semi-rigid coaxial connection lines have been used to obtain the desired element excitations. Using the comparison method described in Section II, the feed-network factor of the configuration has been measured to db over GHz, which is sufficient for the Type A arrays. For measurements of the Type B version, the network configuration has been upgraded by variable line extender that

6 1464 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 45, NO. 10, OCTOBER 1997 (c) (d) Fig. 6. (Continued.) Measured radiation pattern of the SRA s in one pricinpal ( = 0 ) and one diagonal ( =45 ) plane. SRA s are operated at the corresponding frequency for optimum on-axis XPD. (c) Type A patches d =0:8, f=2:935 GHz. (d) Type B patches d =0:7, f = 2:945 GHz: LHCP; RHCP. partly compensated phase errors and improved XPD to db over GHz. The results of corresponding overall XPD measurements are shown in Fig. 5 (c) for the Type A arrays and the basic network and in Fig. 5(d) for the Type B array and the upgraded version. For reasons of comparison, the sum of separately measured XPD and XPD figures according to (1) is also shown. As expected, the larger unsymmetries of the Type A arrays with, and lead to larger deviations between prediction and measurement since the assumption of perfect symmetry is clearly not valid for these cases. Hereby, the deviation can be a deterioration (0.6, 0.8 ) as well as an improvement (0.4 ) of the measured XPD compared to the prediction. The smaller unsymmetries of all other arrays lead to a fairly good agreement of prediction and measurement, which implies that electrical unsymmetries must be kept below 1.0 db, if the advantages of an optimized radiating structure should be utilized in practice. The obtained results confirm the fairly high sensitivity of fully optimized patches, whereas elements with a XPD below 20 db seem to be rather uncritical, as shown by Fig. 5(d) and [7]. The achieved bandwidths of the Type A arrays are 0.9% (0.4 ), 1.7% (0.5 ), 1.9% (0.6 ), and 2.0% (0.7, 0.8 ) for an overall XPD 30 db, which are some of the best results reported for comparable configurations and frequencies. In addition, a spacing of at least 0.7 permits a XPD 35 db over 0.7% bandwidth, whereas the single patch itself would provide less then 0.1% for such XPD levels. Similarly, the Type B array achieves a XPD 30 db over 1.9% bandwidth and 35 db over 1.0%, which proves that a comparable performance can be obtained by the combination of a less optimized radiating structure with an improved feed network, as predicted by (1). Consequently, almost any desired XPD bandwidth can be achieved within the matching and gain limits of the radiators if a sufficiently optimized feed network is used which compensates a poor performance of the radiators. Whereas such high-performing feed networks can lead to excellent bandwidths for moderate XPD levels using even linearly polarized radiators [1], [6], the requirements for their amplitude and phase accuracy in the presence of coupling and matching effects increase rapidly with increasing XPD requirements and decreasing element performance, which increases the associated implementation effort accordingly. VI. SRA PATTERN CHARACTERISTICS Apart from the polarization properties for the on-axis direction, the radiation pattern of the investigated arrays are of interest for most applications since they describe the antenna

7 KRAFT: EXPERIMENTAL STUDY ON 2 2 SEQUENTIAL-ROTATION ARRAYS 1465 Fig. 8. Evaluation of measured peak levels for cross polarization (XPOL) in the diagonal planes versus XPD st for different SRA s with spacings of d =0:4 0:9: Type A patches d =0:4; Type A patches d =0:5; Type A patches d =0:6; Type A patches d =0:7; Type A patches d =0:8; 4 Type B patches d =0:7; 2 linearly polarized elements d =0:87 (from [6]); approximated curve fit. Fig. 7. Evaluation of pattern measurements for the SRA s with Type A patches and spacings of d = 0:4 0:8: Peak levels for cross polarization (XPOL) in the diagonal planes and copolarized sidelobes in the principal planes. Half-power beamwidth (HPBW): individual measurements; average. performance over a larger angular range. Examples of such pattern are presented in Fig. 6 (c) for Type A arrays with spacings of 0.4, 0.6, and 0.8 and in Fig. 6(d) for the Type B array, whereby cuts in one principal and one diagonal plane measured at the corresponding optimum frequencies are shown. Basically, these figures illustrate the typical pattern of 2 2 SRA s with low-gain radiators, which are characterized by a suppression of cross polarization within the main-beam, peak cross-polarization levels in the diagonal planes, and possible copolarized sidelobes in the principal planes (c.f. [6]). In order to obtain design data, the measured pattern of all investigated arrays have been evaluated in terms of half-power beamwidth (HPBW), peak cross-polarization levels, and sidelobes. This evaluation is summarized in Fig. 7 and for the Type A versions, whereas the Type B array provides similar figures. It is evident that higher peak cross-polarization levels occur only for the smallest spacing, where the elements radiate almost linear polarization, whereas larger spacings lead to peak levels around 20 db without larger variations. Copolarized sidelobed occur for and increase to 14 db for 0.8, whereas the HPBW decreases fairly regularly from 46 for 0.4 to 29.5 for 0.8. Some kind of optimum is achieved for spacings of and an HPBW of about where the peak cross-polarization levels and the copolarized sidelobes are kept below 20 and 18 db, respectively. Whereas sidelobes and HPBW depend directly on the element spacing, as usual for an array, the decrease of peak cross polarization for increased spacings occurs because mutual coupling decreases accordingly, which improves the in situ polarization purity of the radiators and decreases, therefore, the totally radiated cross polarization. In order to obtain a direct relation between the in situ radiator XPD described by XPD and the corresponding peak cross-polarization levels of the array, these figures are plotted in Fig. 8, which also includes the result for linearly polarized patches reported in [6]. Whereas true linearly polarized patches (XPD db) provide figures of 0.7 db, which improve to 10.1 db for XPD db, all configurations with XPD db obtain figures below 20 db. This implies a lower limit of db for XPD, which is required to avoid larger offaxis cross-polarized radiation but is fairly simple to achieve even without a full optimization of the patches. VII. SUMMARY AND CONCLUSIONS The radiation properties of 2 2 sequential-rotation arrays with single-feed circularly polarized patches and feed networks based on wide-band isolating (matched) power dividers have been investigated experimentally with the goal to maximize the bandwidth for a XPD db. Hereby, a separate description of XPD contributions from the feed network and the radiating structure (XPD, XPD ) has been used to study the impact of element polarization, mutual coupling, electrical symmetry, and excitation errors. It has been shown that properly designed arrays provide bandwidths of up to 2% for a XPD db and up to 1% for a XPD db, whereby both fully optimized patches with an initial XPD db as well as moderately optimized versions with db can be used. Whereas the former type minimizes the accuracy requirements for the feed network but is more sensitive to mutual coupling and manufacturing tolerances, the latter type

8 1466 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 45, NO. 10, OCTOBER 1997 is more robust but requires an improvement of the network performance XPD by about 3 5 db. Mutual coupling has been identified as a critical issue for the patch design since it can easily change the optimum frequency by up to 1.9% and deteriorates the in situ radiator XPD by up to 30 db, depending on the element spacing and the initial figure. For a spacing of and a radiator contribution XPD of at least db, a compromise in terms of array size, impact of mutual coupling, and pattern characteristics is obtained, which provides beamwidths of 33 38, sidelobe levels below 18 db and peak cross-polarization levels below 20 db. [4] T. Teshirogi, M. Tanaka, and W. Chujo, Wide-band circularly polarized array antenna with sequential rotations and phase shift of elements, in Int. Symp. Antennas Propagat. ISAP, Tokyo, Japan, Aug. 1985, pp [5] P. S. Hall, Application of sequential feeding to wide bandwidth, circularly polarized microstrip patch arrays, Proc. Inst. Elect. Eng., vol. 136, pt. H, pp , May [6] J. Huang, A technique for an array to generate circular polarization with linearly polarized elements, IEEE Trans. Antennas Propagat., vol. AP-34, pp , Sept [7] U. R. Kraft, Main-beam polarization properties of four-element sequential-rotation arrays with arbitrary radiators, IEEE Trans. Antennas Propagat., vol. 44, pp , Apr [8] J. R. James, P. S. Hall, and C. Wood, Microstrip Antenna Theory and Design.Stevenage, U.K.: Peregrinus, 1981, pp ACKNOWLEDGMENT The author would like to thank Prof. G. Mönich from the Technical University, Berlin, Germany, for numerous valuable discussions on this subject and D. Blaschke, L. Jensen, and H. Wolf from Dornier Satellite Systems, Munich, Germany, for their support during the preparation of the paper. REFERENCES [1] S. D. Targonsky and D. M. Pozar, Design of wide-band circularly polarized aperture-coupled microstrip antennas, IEEE Trans. Antennas Propagat., vol. 41, pp , Feb [2] M. Haneishi, T. Nambara, and S. Yoshida, Study on ellipticity properties of single-feed-type circularly polarized microstrip antennas, Electron. Lett., vol. 18, no. 5, pp , [3] M. Haneishi, Y. Hakura, S. Saito, and T. Hasegawa, A new circularly polarized planar antenna fed by electromagnetical coupling and its subarray, in Proc. 18th Eur. Microwave Conf., Stockholm, Sweden, 1988, pp Ulrich R. Kraft (M 89 SM 97) received the Dipl. and Ph.D. degrees in electrical engineering (both summa cum laude) from the Technical University of Berlin, Germany, in 1986 and 1989, respectively. From 1985 to 1989, he worked as a Research Associate at the Microwave Institute of the Technical University of Berlin, where he was involved in research projects covering the basic aspects of the description and optimization of circular polarization as well as the optimization of helix and spiral antennas, microstrip patches, and small arrays. In 1989 he joined Daimler-Benz Aerospace/Domier Satellite Systems, Munich, Germany, where he worked on several technology and development projects in the areas of submillimeter-wave atmospheric sensors, synthetic aperture interferometric radiometers, and active satellite antennas. Since 1995 he has headed the Projects Department of that company s Product Center for Antennas and Payload Components, which involves satellite and groundstation antennas, antenna measurement facilities, and special microwave equipment. He is currently serving as the editor of the MTT/AP German Newsletter. Dr. Kraft received the Karl Ramsauer Research Award in 1989.