Probe Based Radiation Pattern Measurements for Highly Integrated Millimeter-Wave Antennas

Transcription

1 Probe Based Radiation Pattern Measurements for Highly Integrated Millimeter-Wave Antennas Stefan Beer, Thomas Zwick Karlsruhe Institute of Technology (KIT) Institut fuer Hochfrequenztechnik und Elektronik (IHE) Kaiserstr. 12, Karlsruhe, Germany Abstract This paper presents the existing measurement solutions for integrated millimeter-wave antennas as well as the novel measurement assembly at the authors institute. This setup allows probe based radiation pattern, gain and return loss measurements of antennas in the frequency range between 50 and 110 GHz. In the setup a far field distance of 60 cm is achieved that allows the characterization of antenna arrays as well as complete packaging solutions that include a dielectric lens or a parabolic reflector. The use of a VNA and a custom made wafer chuck makes it possible to measure the antenna s return loss in the same setup. A custom made bended waveguide and a horn antenna can be used to achieve a proper gain calibration. Using two rotary stages a receive antenna can be rotated around the antenna under test to measure its 3D radiation pattern. I. INTRODUCTION Usually antennas are measured in an anechoic chamber on a rotating tower. In that case the AUT (Antenna Under Test) is connected to the measurement system by a coaxial cable or a waveguide connector and then rotated around its axis to measure the radiation pattern. For printed mmw (millimeterwave) antennas it becomes more and more difficult to contact these with a coaxial plug or a waveguide connector as their size is in the range or even below the size of the plugs. Additionally, as antennas are more frequently integrated directly onto the semiconductor chip or into the chip package, the measurement of these integrated antennas would be distorted if a plug was used. In that case it becomes favorable to contact the AUT with a RF (radio frequency) probe while measuring it. Thus it becomes possible to characterize the AUT exactly at the point where it will be connected to the chip later and the difficult attachment of a plug or connector that is not used in the packaging solution is avoided. II. PROBE BASED MEASUREMENT SOLUTIONS In [1] antennas are for the first time contacted by a RF probe. The system is calibrated to the probe tips to allow a correct measurement of the antenna impedance. Two identical antennas are aligned in a known distance R to measure the gain in direction of the main lobe. The antennas are positioned on an expanded foam block instead of a metallic probe station to avoid reflections. A probe based radiation pattern measurement is for the first time presented in [2]. The AUT is positioned on a usual probe station that is covered with absorbing material. An open ended waveguide used as the measurement antenna is attached to a plexiglass arm that is rotated by a step motor. Thus the radiation pattern is measured in one sectional half plane. Due to the fact that not all metal parts of the probe station can be covered with absorbing material measurement errors occur. A similar setup is proposed in [3] for antennas in the frequency range of 2 GHz to 40 GHz. In that case the AUT is positioned above a cavity that is filled with absorbing material to avoid reflections. Here two sectional half planes can be measured. The arm to which the measurement antenna is attached can be shifted in length to reach a far field distance of up to 1.5 m. A setup based on a near-field far-field transformation is presented in [4]. A near-field measurement probe is scanning the antenna s near-field above the AUT in two dimensions. Then the far-field pattern in the half-plane above the AUT can be calculated. The setup is dimensioned for antennas around 24 GHz. The necessary maximum distance (a fifth wavelength) between the AUT and the near-field probe however impedes the measurement of antennas above 24 GHz. One of the authors for the first time presented a setup in which the AUT is measured quasi in air, completely without a probe station [5]. The AUT is clipped to a sample holder, which is attached to a long arm. Another arm is attached to a probe head and holds the RF probe. Around the AUT a waveguide arm is rotated to which a measurement horn antenna is connected. The distance from the horn to the AUT is 38 cm. Depending on the plane which has to be measured the step motor is attached to the floor or to the wall. Thus 3 sectoral planes can be measured in total or nearly in total. A gain calibration can be fulfilled by attaching a reference horn with well known gain exactly at the point where the AUT would be positioned. The return loss is measured in the same setup as a VNA is used. A similar system that is also based on a waveguide arm that rotates around the AUT in one of three possible cut planes is presented in [6]. This assembly is placed in a small anechoic chamber to reduce reflections. A spectrum analyzer is used to measure the transmission loss in every rotation angle and thus to calculate the radiation pattern. Using a GSGSG probe the symmetrical antenna is measured without using a balun on the substrate, however a waveguide balun is used between probe and measurement receiver. During the time when the authors where building the new setup at their institute another setup was presented that for the first time allowed a probe based 3-dimensional radiation

2 pattern measurement [7]. This is achieved by using two stepper motors and two arms that can rotate a measurement antenna on the surface of a sphere around the AUT. In that case the AUT is also measured quasi in air to avoid reflections. A gain calibration is achieved by connecting the signal source directly to the spectrum analyzer that is used on the receiver side. The calculated free-space loss is then subtracted from the measurement results. Thus the 3-D radiation pattern and the gain can be calculated. III. MEASUREMENT ASSEMBLY AT IHE Based on the works mentioned and the work previously fulfilled by the authors a new measurement setup was realized at the Institut fuer Hochfrequenztechnik und Elektronik (IHE). The aim was to be able to measure the 3-D radiation pattern in both polarizations, the gain and the return loss of printed mmw antennas as well as arrays or complete packaging solutions that include a reflector or a lens. Through this configuration and through the use of counterweights it became possible to achieve a larger far field distance of 60 cm without overloading the stepper motors. The two rotary stages which are consistent with the azimuth and elevation angles allow it to either measure nearly the complete 3D-radiation pattern or three sectional planes without circumstantially rearranging the setup like in [5], [6]. The attachment of the measurement horn is rotatable around 90 to realize the measurement of both polarizations. The rotating angle of the arms is partly blocked by the table such that the horizontal plane can only be measured within 270 and one vertical plane only within 255. The AUT sample holder is attached to a plastic arm whose other end is attached to the measurement table, as can be seen in Fig. 2. Directly above, the probe is attached to another plastic arm. This arm also holds the signal source and the waveguide couplers and is attached to a 3-D probe head. Thus the probe can be aligned and landed on the AUT as it is common on a usual probe station. The alignment of the probe can be observed using a digital microscope that is connected to a PC and provides a 200x magnification. The additional arm that holds the microscope can be turned out of the measurement range during measurement. Fig. 2 Probe head for accurate probe alignment The AUT sample holder (Fig. 3) is made of dielectric foam with a low permittivity (ε r < 1.1) and low losses (tanδ < 0.001). Thus it resembles air in terms of electromagnetic properties. The AUT substrate is attached to the foam such that the antenna itself is positioned in air while the probe pads are above the foam. Therefore the antenna is measured quasi-inair while the dielectric microscope allows probing the pads. Fig. 1 Measurement setup A. Mechanical Assembly Realized was a setup mechanically similar to [7] in which two step motors and two arms are used to rotate a receive horn around the AUT, as can be seen in Fig. 1. In contrast to [7] the bigger rotary stage was attached onto the floor, exactly in the axis below the AUT and thus the second and smaller rotary stage is positioned in the horizontal plane of the AUT. B. Measurement Equipment The used measurement equipment is the Agilent V/W Millimeter Subsystem, based on 2 signal generators, a pre-amplifier, a millimeter-wave source, waveguide couplers, harmonic mixers, diplexers and a receiver, as depicted in Fig. 4. Measurements can either be performed in V- or W- Band, as the mmw source and the harmonic mixers have to be interchanged. An advantage of the system is that the two test ports comprise a diplexer. One of these diplexers is positioned

3 on the rotating arm (opposite to the receive antenna to counterweight) and thus only one coaxial cable has to be connected through the rotary stages using single port rotary joints. Directly down-converting the signal at the measurement horn also improves the dynamic range in contrast to the systems in [5], [6]. The use of a VNA instead of a spectrum analyser [6], [7] allows return loss measurements, a proper gain calibration, as well as gating in time domain. correct reflection coefficient raut can be calculated out of the measured reflection coefficient s11,m using raut = s11,m e00. e11 (s11,m e00 ) + e10e01 (1) Fig. 5 Impedance Calibration using a calibration substrate D. Gain Calibration Fig. 3 AUT sample holder and coaxial RF probe To determine the gain of the AUT, an additional calibration measurement has to be performed to identify the free space loss, as well as all system losses. Therefore a horn antenna with well known gain GHorn is attached to the system instead of the probe and the AUT using a bended waveguide (Fig. 6). Thus using the measurement result s21,horn of the gain calibration all system losses (including the free space loss) can be calculated as 2 GSystem s21,horn =. GHorn (2) Fig. 4 Block diagram of measurement equipment C. Impedance Calibration The impedance calibration is a two step process that is also needed for the gain calibration. The full calibration process is described in more detail in chapter 17 of [8]. First, the transmit path of the system is calibrated to the output of the waveguide couplers. Then the influence of the probe and the 1 mm coaxial cable is calculated by performing an OSL calibration to the probe tips using a custom made wafer chuck and a standard calibration substrate, as can be seen in Fig. 5. The wafer chuck is attached to the system instead of the AUT sample holder. Based on the two port error model the error terms e00, e11 and e10e01 can be determined by measuring the three calibration standards open, short and load. Thus the Fig. 6 Gain calibration using a reference horn As the losses of the probe and the 1 mm cable have been determined in the previous calibration step as GProbe= e10e01, the proper gain of the AUT can then be calculated by 2 GAUT s21,m, = GSystem GPr obe (3) where s21,m is the measurement result of the AUT transmission measurement for every angular direction needed. Favourably, the gain calibration is performed at first, because more modifications have to be done at the setup. Also,

4 this calibration does not need phase measurement and thus is not as sensible to changes in temperature or drifts of the measurement equipment. generated by reflections from the metallic probe which are hard to suppress. IV. MEASUREMENT RESULTS Different 77 GHz antennas were designed on a mm thick Alumina substrate and measured with the described setup in the W-Band. A. Omni-directional Antenna First, an omni-directional antenna has been measured, a bow-tie slot antenna, as shown in Fig. 7. The bow-tie slot antenna is known to yield wide bandwidth and high efficiency while offering an easy way to influence the impedance for a given resonance frequency by changing the dimensions w1 and w2. The resonance frequency is given by the slot length l, which is 1.6 mm in our case. An optimized bandwidth was found for the slot widths w1 = 0.32 mm, w2 = 1.5 mm. Fig. 9. Gain of the bow-tie slot antenna: H-Plane. E-Plane B. End-fire Antenna The measured end-fire antenna is a Vivaldi antenna with the dimensions given in Fig. 10. The measured return loss does not match the simulation results perfectly but tallies with the fact that the matching is well below 10 db up to 100 GHz. Fig. 7 Omni-directional bow-tie slot antenna The dual bow-tie antenna has a return loss of better 10 db from 75 to 87 GHz (Fig. 8). The measured and simulated input return loss show good agreement. The H-Plane is perpendicular to the feeding lines and can be measured completely. The E-Plane is parallel to the feeding lines and can be measured in a range of 255, while the rest of the plane is blocked by the table. In both planes 0 is the direction upwards and ±180 is the direction downwards through the antenna substrate. The measured radiation patterns at 80 GHz also confirm the simulation results and verify the omnidirectional behaviour (Fig. 9). Fig. 10 Vivaldi antenna Fig. 11 Return Loss of the Vivaldi antenna Fig. 8. Return Loss of the bow-tie slot antenna The peaks in the E-plane around 90 have been observed for different antennas and it is assumed that these are The antenna s H-plane is perpendicular to the antenna substrate while the E-plane coincides with the substrate surface. The measured gain is slightly below the simulation results, but the radiation patterns agree well, as can be seen in Fig. 12. As the measurement table is blocking some parts of the measurement range both planes could not be measured completely. To still be able to evaluate the backlobe, the same Vivaldi antenna has been manufactured twice, once with straight feeding lines and once with a 90 bend in the feeding

5 lines. Thus it becomes possible to measure the antenna s backlobe, as depicted in Fig. 13. Both radiation patterns agree well within the overlapping range and a backlobe at around 1 dbi can be observed which is around 3 db higher in regards to the simulation results. deviation is mainly caused by the antenna itself and not by the measurement setup. Fig. 15 Gain of the Vivaldi antenna with reflector: H-Plane. E-Plane Fig. 12 Gain of the Vivaldi antenna: H-Plane. E-Plane Fig. 13 Vivaldi antenna backlobe measurement: H-Plane. E-Plane C. End-fire Antenna and Parabolic Reflector The Vivaldi antenna has also been combined with a cylindrical parabolic reflector to validate an antenna concept for automotive radars [9]. To be able to measure this antenna/reflector combination with the probe based setup, it is very important that a far field distance of 60 cm is achieved, which allows the measurement of apertures as large as 3.3 cm at 80 GHz. Fig. 14 Vivaldi antenna combined with a parabolic reflector The antenna / reflector combination is depicted in Fig. 14 and the gain and radiation pattern measurement results in Fig. 15. It can be observed that the shape of both E-plane and H-plane beams comply with the simulation results and that also the sidelobes in the H-plane are proved. The gain V. CONCLUSIONS The described setup allows probe based radiation pattern, gain and return loss measurements of antennas in the frequency range between 50 and 110 GHz. In the setup a far field distance of 60 cm is achieved that even allows to measure an antenna that is combined with a parabolic reflector. The use of a VNA and a custom made wafer chuck makes it possible to measure the antenna s return loss in the same setup. Additionally measurement results can be gated in time domain to erase potential reflections at surrounding walls. A gain calibration that even de-embeds the influence of the RF probe is achieved by using a two step calibration process. The setup allows 3D-radiation pattern measurements or the measurement of different sectoral planes without mechanical alterations. REFERENCES [1] R.N. Simons and R.Q. Lee, On-wafer characterization of millimeterwave antennas for wireless applications, Microwave Theory and Techniques, IEEE Transactions on, 47(1):92 96, Jan [2] R. Simons, Novel on-wafer radiation pattern measurement technique for mems actuator based reconfigurable patch antennas, Antenna Measurement Techniques Association Meeting and Symposium, [3] K. Van Caekenberghe, K.M. Brakora, W. Hong, K. Jumani, DaHan Liao, M. Rangwala, Y.-Z. Wee, X. Zhu, and K. Sarabandi, A 2 to 40 GHz probe station based setup for on-wafer antenna measurements. Antennas and Propagation, IEEE Transactions on, 56(10): , Oct [4] A. Shamim, L. Roy, N. Fong, and N.G. Tarr, 24 GHz on-chip antennas and balun on bulk si for air transmission, Antennas and Propagation, IEEE Transactions on, 56(2): , Feb [5] T. Zwick, C. Baks, U.R. Pfeiffer, Duixian Liu, and B.P. Gaucher, Probe based mmw antenna measurement setup, Antennas and Propagation Society International Symposium, IEEE, 1: Vol.1, June [6] Ito, T.; Tsutsumi, Y.; Obayashi, S.; Shoki, H.; Morooka, T., "Radiation pattern measurement system for millimeter-wave antenna fed by contact probe," Microwave Conference, EuMC European, vol., no., pp , Sept Oct [7] S. Ranvier, M. Kyrö, C. Icheln, C. Luxey, R. Staraj, and Vainikainen P., Compact 3-d on-wafer radiation pattern measurement system for 60 GHz antennas, European Cooperation in the Field of Scientific and Technical Research, [8] D. Liu, U. Pfeiffer, J. Grzyb, B. Gaucher, Advanced Millimeter-Wave Technologies: Antennas, Packaging and Circuits, Wiley, 2009 [9] Beer, S.; Adamiuk, G.; Zwick, T., "Novel Antenna Concept for Compact Millimeter-Wave Automotive Radar Sensors," Antennas and Wireless Propagation Letters, IEEE, vol.8, no., pp , 2009