New Microstrip-to-CPS Transition for Millimeter-wave Application Kyu Hwan Han 1,, Benjamin Lacroix, John Papapolymerou and Madhavan Swaminathan 1, 1 Interconnect and Packaging Center (IPC), SRC Center of Excellence @ GT School of Electrical and Computer Engineering, Georgia Institute of Technology 66 Ferst Drive, Atlanta, GA 3033 USA Email: khan7@gatech.edu Abstract This paper presents new design guidelines for microstripto-cps transition for millimeter-wave applications. A circular path transition is proposed with an advantage of reducing design time as well as improving bandwidth by reducing the coupling effect of the structure especially in the 60 GHz frequency band. This new transition method is studied using 3D full-wave EM simulation tool, microwave studio, and measured experimentally up to 67 GHz with Agilent PNA E8361C. Two structures of microstrip-to-cps are presented in this paper. One approach uses an asymmetric T-junction and the other approach uses a symmetric T-junction. For a backto-back microstrip-to-cps transition experimental measurement, insertion loss of less than 3 db was measured over a 14% bandwidth with a center frequency of 6 GHz for the asymmetric structure and insertion loss better than.8 db is obtained over a % bandwidth for the symmetric structure around the target frequency of 60 GHz. 1. Introduction In recent years, increasing demand of high speed wireless communication systems has brought a lot of attention to the millimeter-wave band around 57~64 GHz which is used for high capacity and short distance communication. In this frequency band, signals are severely attenuated in air. For this reason, highly directive and high gain antennas are necessary to increase efficiency of the communication throughput [1]. In order to achieve low cost and easy fabrication, as well as high directivity, endfire microstrip planar antennas such as the quasi-yagi antenna have been widely implemented []. This kind of dipole antenna requires a balun which converts unbalanced input to balanced output and vice-versa for connecting the antenna to the system as well as testing and measuring the performance of the antenna. Coplanar stripline (CPS) which is a uniplanar transmission line structure is a well-known structure for feeding the printed dipole antennas and mounting solid state devices without via-holes [5]. Therefore, a microstrip-to-cps transition with good insertion loss as well as good return loss with broadband characteristic is required. Various methods in the open literature of designing CPSto-microstrip transitions have been introduced as shown in Fig. 1. In [1] and [4], an asymmetric T-junction has been used for signal dividing and combining in CPS, and in [3], a symmetric T-junction has been used for improving the efficiency of the transition compared to [4]. Due to the very small wavelength in the 60 GHz band, the square shape used in [1], [3] and [4] is not optimal because paths in transition are too close to each other, resulting in parasitic coupling effect. This coupling effect introduces complexity for designing baluns because each design parameter is dependent on others, which increases the design cycle. On the other hand, in [], [5] and [6], radial stubs, artificial transmission lines and viaholes have been used for microstrip-to-cps transition respectively. These approaches require complicated ground alignment with the signal path or drilling vias that result in high fabrication cost especially at the 60 GHz frequency. Besides, there are no guidelines available on designing the microstrip-to-cps transition to excite the odd mode at the coupled microstrip line except that the two paths need to be 180 degree out of phase. (c) Fig. 1 Conventional microstrip-to-cps transition square loop [1], [3] and [4], with via [6], (c) radial stub [], (d) artificial transmission line [5] This paper presents a simple design guideline for microstrip-to-cps transition for 60 GHz frequency band. A back-to-back structure (microstrip-cps-microstrip) has been fabricated for measurement with network analyzer. Both (d) MS-to-CPS Transition 978-1-6184-498-5/11/$6.00 011 IEEE 105 011 Electronic Components and Technology Conference
asymmetric and symmetric T-junctions have been designed and fabricated. Both structures have been studied theoretically and experimentally using 3D full-wave EM simulations and a network analyzer, which can measure S-parameters up to 67 GHz. The microstrip-to-cps transition in this paper is designed on liquid crystal polymer (LCP) substrate, which is an organic, lightweight and flexible material with excellent loss characteristics up to 110 GHz as well as very good mechanical stability [1]. This paper is organized as follows. Section explains theory on designing balun and section 3 discusses the design guidelines for both asymmetric and symmetric structures. Section 4 demonstrates S-parameters from simulation and measurement, and finally section 5 summarizes the conclusion.. Theory Microstrip-to-CPS transition is one kind of method to design a balun which is a type of electrical transformer that can convert electrical signals that are balanced to signals that are unbalanced and vice versa. As shown in Fig., the microstrip line is unbalanced and the CPS is balanced. Since planar dipole antennas use a truncated ground plane as a reflector, the antenna feed must be excited with the odd mode. 3. Design Guidelines Fig. 3 and Fig. 4 show the two configurations of the proposed microstrip-to-cps transition. Both configurations consist of a large semi-circle and a combination of a small semi-circle and transmission lines. Ideally, the length difference between the two paths needs to be 0.5 λ g, where λ g is the guided wavelength, in order to obtain 180 degree phase offset between the two paths. No quarter-wavelength transformer is required in this proposed design because it would limit the bandwidth of the transition [5]. Fig. 3 Top view of the newly proposed microstrip-to-cps transition with asymmetric T-junction (Structure 1) Fig. Top view and cross-sectional view with electric field of the microstrip-to-cps transition The odd mode at the CPS is excited with 180 degree phase difference between the two coupled microstrip lines. Therefore, an electric field at the CPS section has a direction from one microstrip line to the other. Making one signal line half-wavelength longer than the other signal line in the microstrip-to-cps transition in Fig. can provide the phase delay. This paper shows a new method of making 180 degree phase delay at the CPS by using a semi-circular loop. Fig. 4 Top view of the transition with symmetric T-junction (Structure ) For the first design shown in Fig. 3, an asymmetric T- junction is used at point A. With a wide space between two transmission lines in the transition area, parasitic coupling will be reduced. The length of the long path is equal to πr, which is half the circumference of circle with radius R. The length of the short path is l+πr, which is the combination of half-circumference of the small circle with radius r and the sum of the two straight microstrips of length l. The radius of the large semi-circle is: 1053
w R r l (1) As described earlier, in order to achieve 180 degree phase difference to excite the odd mode at the CPS, the relationship between the long path and the short path can be expressed as: R (l r) w ( r l ) (l r) w r l l r () w l( ) w l 4 After simplification as shown in equation (), the r term disappears and the relationship between the two paths provides an equation for the parameter l. Therefore the radius for the small circular path turns out to be a free variable and only the length l is the main design parameter which is frequency dependent. Therefore by calculating this value, the balun can be designed at the desired frequency. In equation (), λ g is determined from the target frequency as well as the relative permittivity of the dielectric substrate. The value of the radius for the small half-circular loop can be selected by the designer and the limitation is that the size of the radius needs to be larger than the width of the CPS. The width of the microstrip is determined based on the thickness and dielectric constant of the substrate. As a result, all design parameters to choose the radius of the large semi-circular path in equation (1) are calculated and decided by the designer. These values of parameters provide a good starting point for designing the microstrip-to-cps transition without any ambiguity. For the second structure shown in Fig. 4, instead of a straight transmission line like the one presented in Fig. 3, two semi-circular paths are used at point A which are dividing and combining points. This symmetric T-junction will improve the efficiency of the transition. The length of the long path is πr, which is half-circumference of the circle with radius R, and the length of the short path is πr + l, which is the sum of the circumference of the circle with radius r and the length l of the straight microstrip. The radius of the large half-loop is: l R r (3) The relationship between the two paths in order to obtain 180 degree phase difference can be expressed as: R (r l) l (r ) r l l r r l (4) l( 1) l As shown in equation (4), the r term is eliminated and the relationship between the paths provides the formula for l as in the previous structure. Only the l term is frequency dependent and the designer can choose the value of the small loop radius with the same restrictions of the first structure. The gap between the CPS is not considered here because it is very small compared to the size of the circular path. Simulations show that subtracting half-size of the gap from both long and short paths would not affect the transition. 4. Simulation and Based on the guideline presented in the previous section, a balanced back-to-back microstrip-to-cps transition was designed for simulation and measurement as shown in Fig. 5. The specification for this transition is a bandwidth of 57 to 64 GHz with return loss better than 10 db and insertion loss better than 3 db. 1500 μm 1500 μm 800 μm 800 μm Fig. 5 Balanced back-to-back microstrip-to-cps transitions: structure 1 with asymmetric T-junction structure with symmetric T-junction The dielectric substrate is a -mil thick LCP with a dielectric constant of 3.16 and a loss tangent of 0.00 at 60 GHz. The dimensions of the first structure are: w = 60 μm, r = 100 μm, l = 960 μm, R = 1090 μm, gap = 0 μm and the dimensions of the second structure are: w = 60 μm, r = 100 μm, l = 1800 μm, R = 1100 μm, gap = 0 μm. 1054
3D full-wave electro-magnetic simulation software microwave studio was used to simulate the response and the Agilent PNA E8361C network analyzer with SOLT (Short Open Load Thru) calibration was used to measure the insertion loss and the return loss for both configurations 1 and presented in Fig. 5. For 60 GHz frequency measurement, using GPPO connectors on the structure can produce unexpected response depending on how the connectors are mounted and soldered. Therefore, the structures were measured with GSG probes. In order to measure the performance of the proposed microstrip-to-cps transition with probes, via-less coplanar waveguide (CPW)-tomicrostrip transitions [7] were used, as shown in Fig. 6. It also simplifies the fabrication process. band. The minimum insertion loss achieved by this structure is. db at 60.3 GHz. Fig. 6 Via-less CPW-to-microstrip transitions: GW = 1000 μm, CPW_G = 70 μm, CPW_L = 00 μm Fig. 7 Photograph of the first configuration while measuring the transition Fig. 8 and Fig. 9 show the S-parameters of the balanced back-to-back configurations presented in Fig. 5 and 5 respectively. The overall agreement between the experimental and theoretical results is satisfactory. In both figures, the plain line represents the simulated response from while the line with x marks represent the experimental. As shown in Fig. 8, simulated return loss is better than 10 db from 58.75 GHz to 67.5 GHz and insertion loss is better than 1.53 db in the frequency band. For the experimental result, the return loss is better than 10 db from 58.5 GHz to 67 GHz and the insertion loss is better than 3 db in the same Fig. 8 S-parameter of the balanced back-to-back transition of structure 1 return loss and insertion loss For the second structure using a symmetric T-junction, as shown in Fig. 9, return loss from simulation has a bandwidth of % from 54.5 GHz to 68 GHz and insertion loss is better than 1.61 db. The measured insertion loss is better than.83 db and the measured -10 db bandwidth starts at the same frequency of 54.5 GHz as the simulated result. The minimum insertion loss achieved using this structure is 1.8 db at 58. GHz. 1055
baluns can be designed at the target frequency and this approach could reduce design cycle time. Circular path instead of square path reduces the coupling effect between the paths in transition and provides simplicity in designing as well as improves the bandwidth of the transition. Using symmetric T-junction at point A as shown in Fig. 4 gives wider bandwidth than structure 1. For structure 1, insertion loss better than 3 db was achieved over a 13% bandwidth with a minimum insertion loss of. db. For structure, the insertion loss was better than.81 db over a % bandwidth with a minimum insertion loss of 1.8 db Fig. 9 S-parameter of the balanced back-to-back transition of structure return loss and insertion loss As discussed earlier in this paper, the results of the second structure show that it has wider bandwidth than the first structure by using a symmetric T-junction at point A (Fig. 3). There is a shift of the resonant frequency as shown in Fig. 8 and Fig. 9. This discrepancy could be attributed to the chosen permittivity for the substrate, which is a crucial design factor at 60 GHz. As shown in Fig. 10, better correlation was shown with a smaller value of the dielectric constant. The difference in insertion loss between the simulation and measurement could be due to the reflection between the structure and the probes. The measured insertion loss (<3 db) still satisfies the specifications. Fig. 10 Return loss response for structure 1 with ε r =3 and structure with ε r =.9 5. Conclusions This paper presents a new design and measurement of microstrip-to-cps transitions along with design guidelines for the 60 GHz band. With one simple parameter calculation, 1056
References 1. A.L. Amadjikpe, D. Choudhury, G.E. Ponchak, J. Papapolymerou, High Gain Quasi-Yagi Planar Antenna Evaluation in Platform Material Environment for 60 GHz Wireless Application, Microwave Symposium Digest, 009. MTT '09. IEEE MTT-S International, June. 009, pp. 385-388.. D. Woo, Y. Kim, K. Kim, Y. Cho, A Simplified Design of Quasi-Yagi Antennas Using the New Microstrip-to- CPS Transitions, Antennas and Propagation Society International Symposium, 007 IEEE, June. 007, pp. 781-784. 3. Y. Qian and T. Itoh, A Broadband Uniplanar Microstripto-CPS Transition, Microwave Conference Proceedings, 1997. APMC '97, 1997 Asia-Pacific, Dec. 1997, pp. 609-61. 4. N.I. Dib, R.N. Simons, L.P.B. Katehi, New Uniplanar Transitions for Circuit and Antenna Applications, IEEE Trans. Microwave Theory Tech., vol 43, no. 1, Dec. 1995, pp. 868-873. 5. Y. Suh and K. Chang, A Wideband Coplanar Stripline to Microstrip Transition, IEEE Microw. Wireless Compon. Lett., vol. 11, Jan. 001, pp. 8-9. 6. T. Ma, C. Wang, R. Hua, J. Tsai, A Modified Quasi-Yagi Antenna With a New Compact Microstrip-to Coplanar Strip Trnasition using Artificial Transmission Lines, IEEE Trans. Antennas and Propagation, Vol. 57, No. 8, Aug. 009, pp. 469-474. 7. G. Zheng, J. Papapolymerou, M. M. Tentzeris, Wideband Coplanar Waveguide RF Probe Pad to Microstrip Transitions Without Via Holes, IEEE Microwave and Wireless Components Letters, Vol. 13, No. 1, Dec. 003, pp. 544-546. 1057