You are on page 1of 3

Progress In Electromagnetics Research Symposium 2007, Prague, Czech Republic, August 27-30

467

The Calculation of the V-shape Microstrip Line Impedance by the Conformal Mapping Method
a Vclav Sdek, Pavel Fiala, and Michal Hadinec a Faculty of Electrical Engineering and Communication, Brno University of Technology Kolejni 2906/4, 612 00 Brno, Czech Republic

Abstract The boundary element method (BEM) is used for the strip centered coaxial line
(SCCL). The common microstrip line has one disadvantage a lot of electromagnetic eld is spread outside the dielectric substrate. This eld moves er times faster than eld under the microstrip inside the dielectric substrate. This deformation of the eld (HEM wave) complicates the application of the microstrip line on frequencies over c. 20 GHz. Described complication can be eliminated in the structure, which cumulates major portmon of power density of the EM eld in dielectric substrate to the detriment of free space above the strip. 1. INTRODUCTION

A lot of dierent electronic equipments have to work together. Unfortunately the power levels among them are over 200 dB very often. Coaxial structures are widely used because of their good shielding eect, which suppress the elds around strong distortion sources (e.g., transmitting antenna feeder) and protect sensitive parts of receivers, measurements etc. [1, 2]. Whereas coaxial line (two concentric cylindrical electrodes) is widely known, strip-centered coaxial line (SCCL, Fig. 1.) is mentioned rarely (founded only in very special literature like [3]).The SCCL structure also oers a very attractive occasion of matching to microstrip line, coplanar waveguide, etc.
1

2z

Figure 1: Normalized strip-centered coaxial line.

The sequence of three conformal mappings (used also in [1, 2]) faces to the formula for the characteristic wave impedance of the V-shape microstrip line Z0 = where = h sin

30 2 r,ef arctanh

2 +

(1)

b tan

and =

h sin

b sin

(2)

The eective relative permittivity is given as r,ef = r (r 1) arctanh 2 + 2 arctanh


2 +

(3)

The result (1) was compared to numerical solution in sc-toolbox in MATLABTM [3]. The conformal mapping is not absolutely exact; there is an approximation in the method, which causes the

468

PIERS Proceedings, August 27-30, Prague, Czech Republic, 2007

error. This error depends on h/b ratio and in a mild way also on the angle . and the result (1) is always bigger than numerical result. The worst case is for small h/b ratio (for h/b = 1.5 is error approximately 15%), for higher ratios increase (for h/b = 5 is only 10% and for h/b = 50 less then 5%). The error of calculated eective relative permittivity is hard to determine, but inuence of this error would be very small, because of square root in (1).
2. BOUNDARY ELEMENT METHOD

The boundary element method (BEM) [5] is based on integral Maxwells equations (Laplaces equation is solved along boundary of the structure only). We need values of the eld intensity just along PECs for characteristic impedance determination according formula below: Z0 = Edy U = 120 z4 , Hdx Ey dx
0

(4)

where U is voltage between conductors and Ey is electric eld y-component along central electrode. (The integrals in rst fraction are only informative; they both are not displayed in exact form). Two dierent strategies have been tested. First of them is based on constant number of boundary elements (n = 50) per one basic element. The at inner electrode is divided onto n = 50 elements, the round outer electrode is divided also onto n = 50 elements and both the magnetic walls are also divided each into n = 50 pieces. That means total sum of all boundary elements is 4n = 200. The lengths of element are not the same, especially in case of z4 = 0.5. The second strategy is based on equidistant division of the border (except outer electrode). The inner electrode is divided to z4 times n pieces, the continuing magnetic wall into (1 z4 ) times n pieces, the outer electrode into n elements and last magnetic wall also n elements. On the ground of calculation in wide range of z4 , the number of elements was n = 100.
Table 1: Results of analysis (error compared to Wadell [3]). Z0 [] / error [%] z4 0,01 0,05 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 0,95 0,99 Wadell [3] 317,89 221,33 179,74 138,16 113,83 96,57 82,63 71,19 60,98 51,22 40,70 33,98 24,73 scCtoolbox 317,86 / 0,01 221,31 / 0,02 179,71 / 0,02 138,11 / 0,03 113,73 / 0,08 96,34 / 0,24 82,66 / 0,04 71,17 / 0,03 60,94 / 0,06 51,17 / 0,08 40,64 / 0,15 33,89 / 0,25 24,57 / 0,64 BEM-1 651,5 / 105 273,3 / 23,5 206,8 / 15,1 152,7 / 10,5 122,8 / 7,52 96,5 / 0,03 82,9 / 0,32 71,42 / 0,32 60,98 / 0,38 51,48 / 0,50 41,03 / 0,81 34,44 / 1,35 26,14 / 5,71 BEM-2 323,4 / 1,73 223,2 / 0,85 180,8 / 0,55 138,6 / 0,35 114,1 / 0,25 96,63 / 0,06 82,9 / 0,34 71,39 / 0,28 61,15 / 0,28 51,38 / 0,32 40,89 / 0,48 34,24 / 0,76 25,39 / 2,68

3. RESULTS

Results of all mentioned methods are displayed in Table 1. All results are compared to relevant results of method mentioned in Wadells book [3], however he doesnt inform about the accuracy of the mentioned formula. He only cites his sources Bongianni [6] and Hilberg [7]. Bongianni made an experiment which conrms this method (the accuracy is in range of measurement errors), Hilberg is the author of this method, but he only describes it, not compare to any other method.For

Progress In Electromagnetics Research Symposium 2007, Prague, Czech Republic, August 27-30

469

each value of z4 has been also numerical Schwarz-Christoel method used [4]. The result of this method is in very good agreement with Wadells results [3].
4. CONCLUSION

One method founded in literature [3] is easy, but with the chicken-to-egg problem of choosing one of two formulae according to their results. The error of designed method compared to [3] is very small, grows for both methods for z4 very close to 0 or 1, but for middle values of z4 (range 0.4 to 0.9 in 1st variant and 0.05 to 0.95 in 2nd one) is the error under the level 1% There is a wide space for next work on this thema. First of them is a suppression of the error, next dierent variants of the dielectric substrate (multiple layers, segments, gaps . . .). Also dierent numerical models should be calculated for better verication of results. Also experimental results will be heplful for next work.
REFERENCES

1. Svana, J., Electromagnetic Compatibility Lectures, Textbook of Brno University of Techc nology, FEKT VUT Brno, 2002. 2. Armstrong, K., Design Techniques for EMC, http://www.complianceclub.com/keith armstrong.asp. 3. Wadell, B. C., Transmission Line Design Handbook, Artech House, Boston/London, 1991. 4. Driscoll, T. A., SC-Toolbox, for MATLAB, http://www.math.edu/ driscoll/SC/. 5. Paris, F. and J. Canas, Boundary Element Method, Oxford University Press, Oxford, 1997. 6. Bongianni, W. L., Fabrication and performance of strip-centered microminiature coaxial cable, Proceeding of the IEEE, Vol. 72, No. 12, 18101811, December 1984. 7. Hilberg, W., Electric characteristics of Transmission Lines, 122, Artech House Nooks, Dedham, MA, 1979.

You might also like