Daniël Janse van Rensburg & Chris Walker*
Nearfield Systems Inc, Suite 524, 223rd Street, Carson, CA,
Tel: (613) 270 9259
Fax: (613) 2709260
*European Antennas Ltd, Lambda House, Cheveley, Newmarket, Suffolk,
CB8 9RG, United Kingdom
Tel: +441 638 731888
Fax: +441 638 731999
Back projection techniques have been used extensively in planar near-field ranges and to a lesser degree in spherical near-field ranges. Recently a back projection technique allowing back projection from spherical near-field data onto a planar surface has been published and implemented. This paper explores this technique further through the presentation of measured data for a large microstrip array antenna. The results demonstrate how the technique can be used to investigate anomalies in the feed structure of the array.
Keywords: Antenna measurements, Holographic back projection, Measurement diagnostics, Spherical near-field.
Back projection techniques have been used extensively on planar near-field ranges for aperture diagnostics purposes and calibration of electronically steered arrays [1 - 3]. This technique has certain capabilities and limitations . In  a high-resolution back projection formulation was presented for spherical near-field systems and this technique promised better performance than the plane wave equivalent. Recently the spherical back projection formulation of  was generalized to allow for back projection onto arbitrary planes . The capabilities and limitations of this implementation are explored in this paper through the evaluation of an electrically large planar array.
The measured data presented are for a large fixed feed microstrip array and it will be shown how feed anomalies were identified through the back projected data and how the subsequent fix was evaluated.
The purpose of this paper is to provide guidelines for the application of near-field back projection in a spherical near-field range.
2 Overview Of Spherical Near-Field Back Projection
The extraction of back projected images from spherical near-field data was described in . This publication claimed higher resolution for this technique than what can be achieved from a plane wave expansion. The formulation presented in  was also limited to spherical back projection surfaces, although the potential for expanding that to arbitrary surfaces was pointed out. In  a back projection formulation was presented allowing for spherical near-field data to be used for extracting back projection images on an arbitrary planar surface within the measurement sphere. The high resolution capability claimed in  was also demonstrated in .
The formulation presented in  is that implemented in the NSI spherical near-field software  and was used to generate the back projection data sets presented below.
3 Measured Data
The antenna under consideration is a large microstrip array of dimensions 550mm x 550mm as shown in Figure 1, with detail shown in Figure 2.
Figure 1: Drawing of the 32 x 32 element X/Ku-band microstrip array.
The antenna shown in Figure 1 is center fed and a corporate feed network distributes power to the centers of the four quadrants from where power is again equally divided to the four sub-quadrants of each quadrant. This antenna was mounted in the spherical near-field facility as shown in Figure 3. In this configuration the antenna main beam was pointed nominally at the near-field probe for the azimuth angle q=0°.
Figure 2: Detail of the feed network for the large microstrip array shown in Figure 1.
Figure 3: Microstrip array mounted in the chamber. Array is facing the near-field probe.
For the antenna as shown mounted in Figure 3, a minimum radius sphere of radius 44cm is required to enclose all radiating parts of the antenna (plus some margin). At a frequency of 14.5 GHz an angular sampling interval of 1.2° is required to resolve all the spherical wave expansion coefficients and these were acquired over a q span of 240° and a f span of 180°. (Based on the high directivity of the antenna, the near-field data set was truncated beyond this.) From this near-field data set far-field radiation patterns as shown in Figure 4 can be extracted.
Figure 4: Co-polar (top) and cross-polar (bottom) principal radiation pattern cuts for the AUT shown in Figure 1 at 14.5 GHz.
The radiation patterns shown in Figure 4 displayed a higher than designed cross-polar level and a directivity of 30.5 dBi, which was also lower than expected. By extracting an aperture field distribution for the AUT, using the described holographic back projection technique, these performance limitations can be studied and is reported on in the following Section.
4 Initial Back Projected Data
For the far-field radiation patterns shown in Figure 4, back projected data can be extracted for two orthogonal polarization components on a planar surface close to the aperture of the AUT. This back projected data amplitude is shown in Figure 5 below.
Figure 5: Back projected amplitude for principal (top) and cross (bottom) polarization components close to the antenna aperture: Pre feed network redesign.
For the images presented in Figure 5 it should be noted that both images are normalized to the global maximum and color scales are shown from 0dB to –20dB. The top image is for the principal polarization component and displays radiation principally from the center feed point of the array. Radiation is also visible from the center of each quadrant power divider. The lower image of Figure 5 shows the back projected amplitude for the orthogonal polarization component and the relative high level of radiation from the feed point is again evident and hints to a source of cross-polar radiation. Comparing the two images shows the edge of the array in the orthogonal polarization image, but not in the principal polarization image.
5 Antenna Redesign
Based on the aperture field distribution information shown in Figure 5, a redesign of the antenna feed network was completed. This redesign was performed chiefly to reduce the radiation from the feed network which had been detected through the back projection process. The impedance of the primary feeds driving the two halves of the antenna was increased, as the back projection had shown a significant amount of radiation from along the length of this feed. This primary feed is the vertical feed shown in Figures 6 & 7. The back projection process had indicated a very high level of cross-polar radiation from the tail connecting the primary feeds, so it was eliminated. This change can be seen in the centers of Figures 6 & 7. Finally, a grid was added to the radome to reduce the cross polar signal further.
Figure 6: Initial feed network.
Figure 7: Redesigned feed network.
6 Final Back Projected Data
Upon completion of the redesign of the feed network, the spherical near-field testing and the back projection were repeated. This back projected data amplitude is shown in Figure 8 below. Once again both images are normalized to the global maximum and color scales are shown from 0dB to –20dB.
The top image is for the principal polarization component and radiation from the center feed point is still visible. However, it is important to notice that the radiation from the rest of the array is now at a much higher level than before and the edges of the array are clearly visible. The lower image of Figure 8 shows the back projected amplitude for the orthogonal polarization component and the significant lower level of radiation from the entire aperture is evident.
Figure 8: Back projected amplitude for principal (top) and cross (bottom) polarization components close to the antenna aperture: Post feed network redesign.
Comparing far-field radiation patterns of the two cases demonstrates the increase in antenna gain due to the redesign. This comparison is presented in Figure 9 (patterns are shown as gain calibrated patterns) with azimuth patterns at the top and elevation patterns at the bottom. Roughly 4dBi improvement in gain is evident.
Figure 10 shows the improvement in AUT cross-polarization performance, based on the feed network redesign. Patterns are shown as gain calibrated patterns with azimuth patterns at the top and elevation patter5ns at the bottom. A 5 to 15 dB improvement in cross-polarization level is evident.
Figure 9: Antenna gain improvement after feed network re-design. Azimuth patterns are shown at the top and elevation patterns at the bottom.
Figure 10: Cross-polar performance improvement after feed network re-design. Azimuth patterns are shown at the top and elevation patterns at the bottom.
A practical example demonstrating the application of holographic back projection
on a planer surface, using spherical near-field data has been presented. In
this particular case a 32 x 32 element microstrip array was considered and the
back projected data allowed for the analysis of the array feed network. A redesign
of the feed network proved to be successful and the holographic imaging was
again used to verify the performance.
The results presented demonstrate the high resolution capability of this type of spherical near-field back projection formulation. It also demonstrates its unique applicability to antenna design, for a case which would otherwise have been difficult to resolve.
 D. Garneski, CSUN Antenna Measurement Course Notes, Northridge, CA, June 2001.
 C. A. Rose, “Accuracy estimation of microwave holography from planar near-field measurements”, AMTA Proceedings 2000, p. 12, Philadelphia, PA, Oct 2000.
 M. H. A. Paquay, “Locating Defective Array Modules Using Planar Near-Field Measurements”, AMTA Proceedings 1998, Montreal, QC, Canada, Oct 1998.
 D. Janse van Rensburg, “Limitations of Near-field Back Projection For Phased Array Tuning Applications”, AMTA Proceedings 2001, Denver, CO, Oct 2001.
 M. G. Guler, E. B. Joy, "High resolution spherical microwave holography", IEEE Trans. APS, Vol. 43, May 1995, pp. 464 -472.
 A. C. Newell, B. Schlüper & R. J. Davis, “Holographic Projection To An Arbitrary Plane From Spherical Near-Field Measurements”, AMTA Proceedings 2001, Denver, CO, Oct 2001.
 NSI 2000 Software Users Manual, Nearfield Systems Inc, Carson, CA, USA, 2001.