The radiation patterns of both right-hand circular polarization(RHCP) and left-hand circular polarization (LHCP) in BGANtransmit and receive band are shown in Figure 6. In Figure 6, it isclearly seen that this antenna can provide RHCP across the oper-ating bandwidth and very pure circular polarization radiation canbe produced within a broad angular range because of the more than15 dB cross-polarization suppression. The measured patterns areslightly asymmetric because that layer misalignment and mis-matched elements caused power redistribution in the feed networkand feeding phase error. But the measured patterns are in goodagreement with the computed patterns.

4. CONCLUSION

A wideband circular polarization antenna with low-cost foam andLDPE substrate is proposed to obtain wide bandwidth and highperformance. The proposed antenna is composed of the stackedcorner-truncated patches with sequential rotation oblique feed anda serial feed network. Having above 28% return loss (with voltagestanding wave ratio VSWR � 2) and 18% axial ratio (AR � 3 dB)bandwidths, the designed antenna can be suitable for application ofbroadband system using circular polarization. Good antenna char-acteristics are achieved over the BGAN operating bands. Theproposed antenna has many advantages such as low profile, lightweight, and simple structure. It seems to be a well-suited candidatefor the high-volume market arising with the BGAN satellite com-munication services.

ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foun-dation of China (No. 60771032) and the Science and TechnologyProjects Fund of DaLian City (No. 2007A10GX107). The authorsthank FuChang Ma for helping with the measurements.

REFERENCES

1. D.C. Morse and K.Griep, Next generation FANS Over inmarsat broad-band global area network (BGAN), 23rd DASC Conference, 2004,11.B., p. 4–1–13.

2. L.E. Garcia-Castillo, J.G. de la Fuente, G.G. Gentili and J.I.A. Montes,Receiving stacked patch array antenna for satellite mobile communica-tions in L-band, IEEE Electrotechnical Conference, 1996, pp. 1389–1392.

3. N.C. Karmakar and M.E. Bialkowski, Circularly polarized aperture-coupled circular microstrip patch antennas for L-band applications,IEEE Trans Antennas Propag 47 (1999), 933–940.

4. S.Q. Fu, S.J. Fang, and S.W. Lu, Low cost antenna for INMARSATmobile ground terminal, J Radio Sci (2008), 310–314.

5. K.L. Chung and A.S. Mohan, A systematic design method to obtainbroadband characteristics for singly-fed electromagnetically coupledpatch antennas for circular polarization, IEEE Transactions on Antennasand Propagation 51 (2003), pp. 3239–3248.

6. N. Herscovici, New considerations in the design of microstrip antennas,IEEE Transactions on Antennas and Propagation 46 (1998), pp. 807–812.

7. M.N. Jazi and M.N. Azarmanesh, Design and implementation of circu-larly polarised microstrip antenna array using a new serial feed sequen-tially rotated technique, IEE Proceedings Microwaves, Antennas andPropagation 54 (2006), pp. 133–140.

© 2009 Wiley Periodicals, Inc.

A NOVEL LTCC MULTILAYER SIWLINEAR PHASE FILTER

Wei Shen,1,2 Wen-Yan Yin,1 Xiao-Wei Sun,2 and Qi-Fu Wei11 Center for Microwave and RF Technologies, School of ElectronicInformation and Electrical Engineering, Shanghai Jiao Tong University,Shanghai 200240, People’s Republic of China; Correspondingauthor: weishen@sjtu.edu.cn2 Institute of Microsystem and Information Technology of ChineseAcademy of Sciences, Shanghai 200050, People’s Republic of China

Received 13 January 2009

ABSTRACT: In this article, a novel multilayer substrate-integratedwaveguide (SIW) linear phase filter is proposed. Four SIW cavities areplaced in two layers based on a standard low temperature cofired ce-ramic technology. Neural network is used to improve the efficiency ofdesign. One slot which is etched on the top metal layer of SIW cavity isadopted to suppress the first high mode. Flat group delay characteristicscan be achieved in the filter that has a rejection level of �25 dB in up-per stopband. The filter has a 55% area reduction in comparison withplanar structure. Good agreement is acquired between measured andsimulated results. © 2009 Wiley Periodicals, Inc. Microwave OptTechnol Lett 51: 2357–2360, 2009; Published online in Wiley Inter-Science (www.interscience.wiley.com). DOI 10.1002/mop.24595

Key words: linear phase filter; low temperature cofired ceramic; multi-layer; spurious mode; substrate-integrated waveguide

1. INTRODUCTION

Distortion of signal often takes place as a result of nonlinear phasefilter (LPF) transfer functions. In addition to the selectivity of abandpass filter, the flat group delay in passband is required toreduce the signal distortion [1]. Two common methods are used toachieve flat group delay in passband [2]. The approach commonlyused is to design a bandpass filter with an imposed linear phaserequirement in addition to the amplitude requirement [3]. As weknow, the simplest way to design a LPF is making cross-couplingand direct-coupling have the same kind of coupling, i.e., bothcouplings are either electric or magnetic or both capacitive orinductive [3].

Traditional waveguide filters are characterized by high-qualityfactor and low inset loss, but are difficult to integrate with mono-lithic circuits due to their weight and area. In the past few years,special attention has been paid to the design of filters usingdifferent substrate-integrated waveguide (SIW) which are nor-mally fabricated by introducing metallic via holes in single as wellas multilayer printed circuit board or low temperature cofired(LTCC). Many planar SIW passive components were demon-strated in [4, 5]. In this study, four SIW cavities are placed in twolayers based on LTCC to construct a LPF. Three direct and onecross-couplings are both magnetic.

In this article, a novel multilayer LPF is proposed and fabri-cated using a standard LTCC. Neural network (NN) is adopted toimprove the efficiency of design. A slot on the top metal layer ofSIW cavity is used to suppress the harmonic mode. Flat groupdelay in passband is also obtained while the filter has a highrejection level in upper stopband. Good agreement is acquiredbetween measurement and simulated results.

2. FILTER ANALYSIS AND DESIGN

2.1. Topology StructureThe topology of the LPF is shown in Figure 1(a), where the solidlines denote positive or negative couplings. Based on Figure 1(a),

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 10, October 2009 2357

a multilayer LPF is realized as presented in Figure 1(b). Thecoupling between electrically nonadjacent and adjacent resonatorsis referred to as cross- and direct-coupling, respectively. Thecross-coupling has the same phase as the direct-coupling, so thatboth couplings are magnetic. The M12 and M34 are carried out byapertures etched on the middle metal layer which is betweenelectrically adjacent cavities and M14 and M23 are achieved by irisbetween horizontally adjacent cavities.

2.2. Filter DesignOur filter is center at 10 GHz and fractional bandwidth is 5%, thereturn loss in the passband is less than �20 dB, we can obtain thecoupling matrix and quality factor by method in [6, 7]

M � �0 0.05437 0 0.0111

0.05437 0 0.03592 00 0.03592 0 0.05437

0.0111 0 0.05437 0�,

Q � 15.54 (1)

The group delay �g can be calculated by following equation

�g � ���/�� (2)

where �� and �� are discrete interval of phase-angle of S-parameters and frequency, respectively.

The loop currents, which are grouped in the vector [I], aregoverned by the following matrix equation [6]:

���U� � j�R� � �M�� � �I� � �A��I� � �j�e� (3)

where [U] is the identity matrix, [R] is a matrix whose onlynonzero entries are R11 � Rin and Rnn � Rout. The excitationvector [e] is given by �e�f � [1,0,0· · ·,0], � is low-pass prototypein the frequency variable.

The S-parameters are determined by:

S21 � � 2j�RinRout�A�1�n1, S11 � 12jRin�A

�1�11 (4)

The initial size of the SIW cavity is determined by equationsshown in [4]. We used single cavity to determine the externalcoupling and two cavities are used to determine the internalcoupling. Then, the initial model of the filter can be constructed inhigh frequency structure simulator. The model is tuned by adjust-ing the geometries of the coupling mechanisms until ideal re-sponses are obtained.

Here, to improve the efficiency of design, NN method is usedin fitting the coupling coefficient versus length of the aperture andwidth of the iris between two adjacent cavities. The results of NNand simulated are plotted in Figure 2.

2.3. Spurious SuppressionTo suppress the first higher mode, a novel slot which is etched onthe top metal layer of SIW cavity is used to truncate the surfacecurrent of the first higher mode. The electric field distribution of

Figure 1 (a) Topology and (b) three-dimensional view of the linearphase filter. [Color figure can be viewed in the online issue, which isavailable at www.interscience.wiley.com]

Figure 2 Coupling coefficient versus (a) the length of the aperture and(b) the width of the iris between adjacent cavities. [Color figure can beviewed in the online issue, which is available at www.interscience.wiley.com]

2358 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 10, October 2009 DOI 10.1002/mop

TE102 and its surface current are shown in Figure 3. In principle,the slot must locate at the center of the TE102 mode where thesurface current achieves maximum. The width and length of theslot may affect the performance of the spurious suppression. In thisdesign, the location of the slot deviates slightly to the center of thecavity.

3. EXPERIMENT RESULTS AND DISCUSSION

The LPFs were fabricated using eight-layer LTCC substrate Ferro-A6, with the relative permittivity� � 5.9, the loss tangent tan�

� 0.0015, and the thickness of each layer 0.096 mm. All metallicvias used to build SIW have the same diameter 0.18 mm. Thegeometrical parameters are shown Figures 4(a) and 4(b) andTable 1.

The photograph of the filters is shown in Figure 4(c) and theresults are shown in Figures 5 and 6. The measured in-band returnand insertion losses are below �19 dB and about 1.1 dB, respec-tively. Figure 6(b) points out that the stopband of the filter ex-tended to 20 GHz with the rejection level of �25 dB. We wouldlike to say that good agreement is obtained between the simulatedand measured S-parameters in Figure 5, except a little discrepancyin the in-band insert loss presented. This is mainly caused by theinaccuracy in the characterization of loss of test fixture. A smallfrequency shift may be contributed to the bigger LTCC shrinkage;the cavities are smaller than expected, so the central frequency

Figure 3 (a) The electric field distribution and (b) the surface current ofthe first high mode in the filter. [Color figure can be viewed in the onlineissue, which is available at www.interscience.wiley.com]

Figure 4 Geometry parameters of filter (a) top view of the filter, (b)bottom view of the filter, (c) photograph of the filters. [Color figure can beviewed in the online issue, which is available at www.interscience.wiley.com]

TABLE 1 Dimensions of the Proposed SIE Filter

SYMBOL VALUE (MM) SYMBOL VALUE (MM)

L1 8.63 Lx1 3.59L2 8.38 pc1 1.48L3 8.38 pc2 2.33L4 8.63 ls1 1.79wid 8.44 ws1 0.5w1 0.6 Ls 6.0w2 0.3 Ws 0.15

DOI 10.1002/mop MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 10, October 2009 2359

drifts up slightly. The variation of group delay of S21 -parameter issmaller than 1% over 50% of the passband. For the second filter,flat group delay and better spurious suppression are both achievedsimultaneously, as shown in Figure 7. The concave of the groupdelay or a convex part of the S21 -parameter is shown in Figure6(b), which locates around 10.4 GHz, this may be caused byfabrication error.

4. CONCLUSION

A novel multilayer LPF that fabricated with a standard LTCCtechnology is proposed in this article. A slot which is etched on thetop metal layer is adopted to suppress the first high mode. Flatgroup delay in the passband is also achieved while the filter witha high rejection level in upper stopband. The area has a 55%reduction in comparison with planar structure. It is more suitablefor millimeter wave integrated circuits.

ACKNOWLEDGMENT

This work was supported by the National Basic Research Programunder Grant 2009CB320204, and by the NSF under Grant90607011 and 60771058 of China.

REFERENCES

1. G. Matthaei, L. Young, and E.M.T. Jones, Microwave filters, imped-ance-matching networks, and coupling structures, Artech House, Bos-ton, MA, 1980.

2. J.S. Hong and M.J. Lancaster, Microstrip filters for RF/microwaveapplications, Wiley, New York, 2001.

3. L. Harle and L.P.B. Katehi, A silicon micromachined four-pole linearphase filter, IEEE Trans Microwave Theory Tech 52 (2004), 1598–1607.

4. X.P. Chen and K. Wu, Substrate integrated waveguide cross-coupledfilter with negative coupling structure, IEEE Trans Microwave TheoryTech 56 (2008), 142–149.

5. D. Dealandes and K. Wu, Single-substrate integration techniques forplanar circuits and waveguide filters, IEEE Trans Microwave TheoryTech 51 (2003), 593–596.

6. S. Amari, Synthesis of cross-coupled resonator filters using an analyt-ical gradient-based optimization technique, IEEE Trans MicrowaveTheory Tech 48 (2000), 1559–1564.

7. R. Levy, Filters with single transmission zeros at real and imaginaryfrequencies, IEEE Trans Microwave Theory Tech 24 (1976), 172–181.

© 2009 Wiley Periodicals, Inc.

Figure 5 S-parameter responses and group delay of the filter. [Colorfigure can be viewed in the online issue, which is available at www.interscience.wiley.com]

Figure 6 (a) Wideband S-parameter responses and (b) spurious suppres-sion responses of the filter. [Color figure can be viewed in the online issue,which is available at www.interscience.wiley.com]

Figure 7 The group delay of the filter with spurious suppression

2360 MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 51, No. 10, October 2009 DOI 10.1002/mop