Transcript
tion. The input return loss is better than 13 dB for frequencies upto 10 GHz.The downconversion micromixer demonstrated in this work has fully utilized the advantages of both SiGe HBT and MOStransistors. The isolations are very good when compared with otherpublished results [3, 4] because the SiGe HBT devices used in thiswork naturally possess better device matches and the deep trenchisolation technology, which eliminates the signal coupling throughthe silicon substrate, helps to improve the port-to-port isolation. ACKNOWLEDGMENTS This work is supported by National Science Council of Taiwan,Republic of China under contract numbers NSC 94–2752-E-009–001-PAE, NSC 94–2219-E-009–014 and by the Ministry of Eco-nomic Affairs of Taiwan, Republic of China under contract num-ber 94-EC-17-A-05-S1–020. The authors also thank the ChipImplementation Center (CIC) for its support. REFERENCES 1. K.-Y. Yeh, S.-S. Lu, and Y.-S. Lin, Monolithic InGaP-GaAs HBTreceiver front-end with 6 mW DC power consumption for 5 GHz BandWLAN applications, IEE Electron Lett 40, (2004).2. B. Gilbert, The MICROMIXER: A highly linear variant of the Gilbertmixer using a bisymmetric Class-AB input stage, IEEE J Solid-StateCircuits 32 (1997), 1412–1423.3. T.K. Johansen, J. Vidkjr, and V. Krozer, Analysis and design of wide-band SiGe HBT active mixers, IEEE Trans Microwave Theory Tech 53(2005), 2389–2397.4. C. Viallon, J. Graffeuil, and T. Parra, High performance K-band activemixer using BiCMOS SiGe process, Electron Lett 41 (2005).© 2006 Wiley Periodicals, Inc. NOVEL APPROACH TO OPTIMIZING A BROADBAND RIGHT-ANGLE COAXIAL-TO-MICROSTRIP TRANSITION Jin-Eep Roh, 1 Jun-Wen Li, 1 Bierng-Chearl Ahn, 1 Chan-Sik Park, 1 and Eun-Jong Cha 2 1 School of Electrical and Computer Engineering, Chungbuk NationalUniversity, #12, Gae-Shin Dong, Cheong-Ju City, Chungbuk 361–763, South Korea 2 Department of Medicine, Chungbuk National University, #12, Gae-Shin Dong, Cheong-Ju City, Chungbuk 361–763, South Korea Received 5 July 2006 ABSTRACT: A new approach to the optimization of a broadband right-angle coaxial-to-microstrip transition is presented. The right-angletransition finds many applications where printed circuits need to be fed from behind the ground plane using coaxial connectors. To obtain lowreflections over the whole operating frequency range of the transition,dimensional parameters, such as ground aperture and probe diameters,ground aperture offset, and microstrip stub length are optimized using a Figure2 Photograph of the high isolation SiGe BiCMOS common modefeedback downconversion micromixer. [Color figure can be viewed in theonline issue, which is available at www.interscience.wiley.com] Figure 3 Measured LO-IF, LO-RF, and RF-IF isolations of the highport-to-port isolation SiGe BiCMOS common mode feedback downcon-version micromixer Figure 4 Two-tone power measurement of the high isolation SiGeBiCMOS common mode feedback downconversion micromixer DOI 10.1002/mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 2, February 2007 451 commercial electromagnetic simulation software. Results are presented for the optimum right-angle transition from an SMA connector to a50- microstrip line on common reinforced PTFE substrates. Measure-ments of a fabricated transition on a 31-mil thick substrate show that itsreflection coefficient is less than 20 dB and the insertion loss is lessthan 0.5 dB over 0.05–20 GHz. © 2006 Wiley Periodicals, Inc.Microwave Opt Technol Lett 49: 451–456, 2007; Published online inWiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.22151 Key words: right-angle coaxial-to-microstrip transition; broadband transition; discontinuity compensation 1. INTRODUCTION A coaxial-to-microstrip transition is widely used in printedantenna and microstrip-based circuit applications. The coaxial-to-microstrip transition is implemented either in a straight form or ina right angle style. The former is accomplished by aligning the axisof the coaxial connector with the end of the microstrip line, and thelatter is made by inserting the coaxial connector perpendicular tothe bottom of the substrate and connecting the probe to the mi-crostrip line.The straight transition structures have been previously studied[1–5] for the extension of transition’s bandwidth and reduction of the return loss. Most of the previous researches on the coaxial-to-microstrip transition have been mainly applicable to basic straight-transition geometries. There are many practical applications wherethe right-angle transition is preferable to the straight one for easyhandling and convenient connector placement [6, 7]. Morgan andWeinreb have presented the optimization of the right-angle tran-sition where the offset of the ground plane aperture is employedfor wideband discontinuity compensation [8]. Their approach has,however, limited success since only one of many structural pa-rameters is employed for optimization. Recently, a right-anglecoaxial-to-microstrip transition for multilayer substrates has beenproposed [9], where part of the ground plane around the circularaperture is incrementally removed from each metal layer, and atthe same time the probe of the coaxial connector is graduallyshifted off-center toward the microstrip line. The transition pro-posed in Ref. 9 shows a reflection coefficient less than 20 dBover 1.8–2.7 GHz.In this article, a new approach to the optimization of theright-angle coaxial-to-microstrip is presented where additionalstructural parameters such as ground aperture and probe diameters,the offset of the ground aperture, and the microstrip stub length areemployed for an improved performance. 2. TRANSITION DESIGN The structure of an optimized right-angle transition is shown inFigure 1. The ground plane of the printed circuit board is electri-cally connected to the outer conductor of the coaxial connector.The printed circuit board has a via-hole into which the machinedcoaxial probe is inserted. The connector probe extends through theground aperture and the substrate, and then is soldered to themicrostrip line on top side of the substrate. Figure 1 Structure of the right-angle coaxial-to-microstrip transition.[Color figure can be viewed in the online issue, which is available atwww.interscience.wiley.com] Figure 2 Top view (a) and bottom view (b) of the transition 452 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 2, February 2007 DOI 10.1002/mop Top and bottom views of the transition are shown in Figure 2.Design parameters of the transition are the via-hole diameter D V ,which is same as the probe diameter, the ground aperture diameter D GA , the ground aperture offset O GA , and the microstrip stublength L S . These design parameters are optimized using the Mi-crowave studio (MWS TM ), a general electromagnetic simulationsoftware by Computer Simulation Technologies.Discontinuity effects in an uncompensated transition can berepresented using the parasitic inductance and capacitance. Figure3 shows electric field lines in the transition region. The diameter of the ground aperture is normally set to be same as the outerconductor radius b of the connector. As shown in Figure 3,however, a better performance can be obtained by reducing theground aperture diameter ( D GA ) and by offsetting the groundaperture in the direction opposite to the microstrip line. In this casethe path length of electrical current flow from the probe to themicrostrip line is decreased, which leads to a reduced discontinuityinductance. The parasitic reactance can be further reduced bydecreasing the probe diameter. Additionally, the length of the stubopposite to the microstrip line can be chosen for the lowestreflection coefficient over a broad frequency range.Effects of design parameters D GA , D V , O GA , and L S on thereflection and transmission performances of the transition aresimulated using the MWS TM . At frequencies up to the Ku-band,the SMA coaxial connector is widely used because they are ruggedand cheap. In this article, we designed SMA right-angle transitionsfor use with a 50- microstrip line on 10, 20, and 31-mil thick substrates (Rogers’s RT5880®, r 2.2, tan 0.0009 at 10GHz). In the simulation, the detailed inner structure of the SMAconnector is simplified into a uniform lossless 50- coaxial line of the same size. Figure 3 Electric field distributions in the transition region: (a) bottomview; (b) A-A cross sectional view Figure 4 Simulated reflection coefficients of the transition between an SMA connector and a 50- microstrip line on a 31-mil thick substrate versus: (a)ground aperture diameter; (b) via-hole diameter; (c) ground aperture offset; (d) microstrip stub length DOI 10.1002/mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 2, February 2007 453 Optimum values of transition design parameters are obtained inthe following way. Firstly, a possible range of variation and theincrement of variation are determined for each design parameter.Secondly, the performances of the transition are analyzed usingMWS TM for all possible combinations of parameter values. Fromthese analyses, we obtain the range of near-optimum values foreach design parameter. Thirdly, we successively adjust each de-sign parameter for the best reflection and transmission coefficientsfrom DC to the highest operating frequency (20 GHz). Repeatingthis fine tuning procedure gives us a set of optimum parametervalues.In the simulation, D GA , D V , O GA , and L S are varied in the rangeof 2.4–4.1 mm, 0.3–1.3 mm, 0.7 to 0.0 mm, and 0.0–0.5 mm,respectively for 10-, 20-, and 31-mil-thick substrates. Figure 4shows the reflection coefficient of an SMA transition on the 31-milthick substrate for near-final values of design parameters. For thetransition on a 31-mil thick substrate, optimum values of D GA , D V , O GA , and L S are found to be 2.8, 0.4, 0.4, and 0.2 mm, respec-tively. For reference, uncompensated values D GA , D V , and O GA are 4.1, 1.3, and 0 mm, respectively.Figure 5 shows the reflection and transmission coefficients of the optimized SMA transition on a 31-mil substrate. The reflectioncoefficient of the optimum transition is less than 24 dB overDC-20 GHz, whereas that of the uncompensated transition is aslarge as 8 dB at 20 GHz. The transmission coefficient of thedesigned transition is greater than 0.45 dB over DC-20 GHz,whereas that of the uncompensated transition is as small as 2.7dB at 20 GHz. A further improvement in the performance of thecompensated transition can be obtained by rounding the rectangu-lar edges of the microstrip stub as shown in Figure 2. In this case,the reflection coefficient of the microstrip port ( S 22 ) is decreasedby about 4 dB at 20 GHz.Table 1 shows the optimum values of design parameters of theSMA transition for 10, 20, and 31-mil thick substrates. From Table1, one can observe that different values of design parametersshould be used for different substrate thicknesses. 3. MEASUREMENTS AND DISCUSSIONS To validate the proposed design method of the optimum right-angletransition, two practically identical transitions are fabricated usingSMA connectors and a 50- microstrip line on a 31-mil thick Rog-ers’s RT5880 ® substrate as shown in Figure 6. The length of themicrostrip line is 20 mm. We used two SMA connectors of the samedesign and made by the same manufacturer. Scattering parameters S 11 , S 21 , and S 22 of the transition are measured using the HP 8510Cnetwork analyzer over 0.05–25 GHz, where the port 1 is at the coaxialinput of the SMA connector.Figure 7 shows the measured reflection and transmission coef-ficients of the structure shown in Figure 6. The peaks and nulls inthe reflection coefficient are caused by the in-phase addition andout-of-phase subtraction of reflections at two transitions separatedby the 20-mm long microstrip line. For the simulation, a section of an ideal coaxial transmission line is assumed for each SMAconnector. We observe good agreements between measured andsimulated results.Assuming that each set of the SMA connector–transition com-bination has same characteristics and subtracting out the theoret-ical microstrip line loss, one can obtain the scattering parametersof a single SMA connector–transition combination from measuredscattering parameters of the structure shown in Figure 6 using themicrowave network theory [10].To accurately obtain the characteristics of the right-angle co-axial-to-microstrip transition itself, one has to remove the imper-fect characteristics of the real SMA connector used in the mea-surement, which has an insertion loss of about 0.2 dB and a returnloss of 20 dB at 20 GHz according to the manufacturer’s specifi-cations. We applied two correction methods to remove the effect of the SMA connector.Firstly, we constructed a back-to-back connection (Fig. 8) of two SMA connectors used in the construction of the structureshown in Figure 6. We then measured its scattering parameters, Figure 5 Reflection and transmission coefficients of the optimized andnonoptimized transitions between an SMA connector and a 50- micros-trip line on a 31-mil thick substrate TABLE 1 Optimum Values of SMA-to-Microstrip Transition Parameters on 10- 20- and 31- mil Thick Substrates Thickness of Substrate(mils)Via-hole (Probe)Diameter (mm)Stub Length(mm)Ground ApertureOffset (mm)Ground ApertureDiameter (mm)10 (0.254) a 0.4 0 0.2 2.020 (0.508) 0.4 0.1 0.3 2.531 (0.787) 0.4 0.2 0.4 2.8 a Values in parenthesis are in mm Figure 6 Photograph of a back-to-back transition test block: (a) topview; (b) bottom view. [Color figure can be viewed in the online issue,which is available at www.interscience.wiley.com] 454 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 2, February 2007 DOI 10.1002/mop from which the scattering matrix of a single SMA connector iscalculated using the microwave network theory [10] as done in thecalculation of a single set of the SMA connector - transition of thestructure shown in Figure 6, assuming that two SMA connectorshave the same characteristics.Reflection and transmission coefficients of the cascaded con-nection of an SMA connector and a transition are given by M 11 E 11 S 11 E 21 E 12 1 E 22 S 11 E 11 S 11 E 21 E 12 (1) M 21 E 21 S 21 1 E 22 S 11 E 21 S 21 (2)where E 11 , E 21 , E 12 , and E 22 are the scattering parameters of theSMA connector, and S 11 and S 21 are the reflection and transmis-sion coefficients of the transition, where the Port 1 is located on theprobe side. In Eq. (1), 1 E 22 S 11 can be assumed to be 1 becausemagnitudes of E 22 and S 11 are about 20 dB. The maximum valueof M 11 magnitude is given by| M 11 | max | E 11 | | S 11 | (3)since the magnitude of E 21 E 12 is close to 1. The magnitude of M 21 is given by| M 21 | | E 21 || S 21 | (4)We employed Eqs. (3) and (4) to remove the effect of the SMAconnector using the first method.In the second method for removing the effect of the SMAconnector used in the measurement, we employed the time-domainwindow function provided by the HP 8510C vector network ana-lyzer. Figure 9 shows the time domain reflected signal of thestructure shown in Figure 6, which is obtained by measuring thereflection coefficient over 0.05–25 GHz and by using the time-domain band-pass impulse response function provided by thenetwork analyzer. The frequency-domain reflection coefficient of the transition 1 is obtained by removing time-domain reflectedsignals due to the connectors 1 and 2, and due to the transition 2and Fourier-transforming the resultant time-domain signal.Employing above correction methods for imperfect character-istics of the SMA connector used in the measurement, we finallyobtained the reflection coefficient S 11, cor (SM) and S 11, cor (TWF) ,and the transmission coefficient S 21, cor of the right-angle transitionalone as shown in Figure 10. S 11, cor (SM) in Figure 10 is thereflection coefficient of the transition obtained using the firstcorrection method where the measured scattering parameters of theSMA connector is employed, while S 11, cor (TWF) is the oneobtained using the second correction method where the time-domain window function is used. Two corrected reflection coef-ficients S 11, cor (SM) and S 11, cor (TWF) agree well with each other,showing that there are little differences between two correctionmethods. The S 11, sim and S 21, sim in Figure 10 are the reflectionand transmission coefficients of the transition predicted by theMWS TM software. They show average deviations of 5.4 dB and0.12 dB over 0.05–20 GHz from the measured reflection andtransmission coefficients S 11, cor (SM) and S 21, cor . The 5.4-dBdifference in the reflection coefficient is not significant since thereflection coefficient itself is a small value (i.e., less than 25 dB).The measured reflection coefficient is less than 20 dB and themeasured transmission coefficient is greater than 0.5 dB over0.05–20 GHz. The performance of the right-angle coaxial-to-micros- Figure 9 Measured time-domain reflected signal of the back-to-back transition test block Figure 7 Reflection and transmission coefficients of the back-to-back transition test block Figure 8 Schematic of a back-to-back test block used in measuring thecharacteristics of SMA connectors. [Color figure can be viewed in theonline issue, which is available at www.interscience.wiley.com] DOI 10.1002/mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 49, No. 2, February 2007 455