Sunteți pe pagina 1din 6
A Designer’s Guide To Microstrip Line Equations, data and conclusions from many sources are compiled to define transmission characteristics. Discussions center on loss, circuit Q, dispersion, dimensional ratios, and moding. Mis is not an announcement of a revolutionary new idea for microstrip cireuit design. Itis, however, a compendium of the best ideas presented over the past several years on design considera tions for this ubiquitous transmission medium. Microstrip technology is quite mature, offering a superior blend of performance characteristics! to the designer of microwave integrated circuits (Table 1). Nearly 50 refer- ences at the end of this article attest to the number of {investigators who have published design formulas for micro- strip. So today, the problem is not that the eireuit designer lacks information concerning this transmission medium, bbut that too much information is available, seattered ‘throughout too many journals. What follows is an attempt to review the most useful formulas and conclusions, and arrange them in a logical, easy-to-follow order. But first, understand the difference between microstrip and other forms of MIC transmission line that are often erroneously referred to as “microstrip.” By definition, ‘microstrip transmission line consists of a strip conductor and a ground plane separated hy a dielectrie medium (Fig. 12). The dielectric material serves as a structural substrate uupon which the thin-film metal conduetors are deposited Conduetors are usually gold or copper. Since field lines between the strip and the ground plane are not contained entirely in the substrate (Fig. 1b), the Propagating mode along the strip is not purely transverse electromagnetic (TEM) but quasi-TEM. Assuming the Table 1: Comparison of popular transmission media ‘rarcee ieee ie se bar ‘ah Une 0 bow ow Power coibily, kw tow Ison between poor fr nereoun eects Bann ‘ave Moiauesion xl une and weight sal Falaton 0 very sy passe eee Inertia Wh ch cevces ie ‘a very gone ret edu rer vey good lage ver aso wey ay lige oy peor lige toe Indian Institute of Technology, Adve Electronic Systems, Department of Electrical Engineers Ing, Kanpur-208016, India 174 1. Microstrip is separated from @ ground plane by dielectric substrate (a). Since not ail Held lines pa through the substrate (b), a quasi-7EM analysis Is usel quasi-TEM mode of propagation, the phase velocity microstrip i given by a where ¢ is the velocity of light, and cy is the effete Gieleetrie constant of the substrate material. The effet dieleetrie constant is lower than the relative delet constant, <, of the substrate, and takes into accoa external fields The wavelength, A,, in microstrip line is given by where V, is given by Eq. 1 and f is frequeney. ‘The characteristic impedance of the transmission lines sven by where C is the capacitance per unit length of the line ‘The analysis for the evaluation of ye and C based at guasi-TEM mode is fairly accurate at lower microvate frequencies. At higher frequencies, the ratio of longitudind- to-transverse electric field components becomes significa and the propagating mode ean no longer be considered qussk ‘TEM. Analysis of this “hybrid mode” is far more rigorow Closed-torm expressions are vital Early attempts to charactercize the performance of miete strip line were based on the quasi-TEM model. Vario electrostatic approximations such as conformal map MICROWAVES * Moy, 1977 4, relaxation methods', variation techniques,*" the od of Green's function**. and the moment method'" generally used. During the last few years, a number jnpers using a hybrid-mode model of microstrip line have ed, but these involve extensive computations. Closed- n expressions are absolutely necessary for optimization computer-aided design of a microstrip circuit. The closed form expressions for Z, and gy have been by Wheeler’, Schneider’ and -Hammerstad.t Wh in terms of Z, and g. For a practical range of, iostrip lines (0.05 < W/h < 30 and «, < 16) Hammerstad is expressions are more accurate than earlier and fall within =1 per cent of Wheeler's numerical lis, His expressions, which are based on the work of iveler and Schneider, include useful relationships defin- eww sown + a [a + 12 B/W) + 0.041-W/HY* 15) 120 #/ Vea Wh + 1.993 + 0.467 In (W/h+1.444) (6) (eww 2 Hammerstad notes that the maximum relative error in 5 per cent and 0.8 per cent, respective! < 16. His lapressions for W/h in terms of Z, and ¢ we Bor W/h < 2, wh ~ Sex A) expla} For W/h > 2, Wh = 2 (Brings) + { saayaae - ar at be [ES at = 000 2 2 Oe These expressions provide the same accuracy as Eqs. 4,5, Sai 7, ‘The results diseussed above assume a two-dimensional strip conductor. But in practice, the strip is three-dimen- sional—its thickness (t) must be considered. However, when Uh < 0.005, 2 < «, < 10, and 0.1 < W/h <5, the agreement ftween experimental and theoretical (t/h=0) results is foellent."* ‘The zero-thickness (t/h=0) formulas given above ean also tye modified to consider the thickness of the strip!» when the strip width, W, is replaced by an effective strip width, WICROWAVES * May, 1977 ‘We, Expressions for W, are: For Wh s 1/25, We Wy 4 (asin booth t [Additional restrictions for applying Eqs. 10 and 11 are t < hn and t < W/2. Typical strip thickness varies from 0.0002 to-0.0005 inch for metalized alumina substrate, and from 1.001 to 0.003 inch for low-dieleetric substrates. ‘Most microwave integrated cireuit applications require ‘a metallie enclosure for hermetic sealing, strength, eleetro- magnetic shielding and ease of handling. The effect of the top and side walls on the microstrip characteristics has been studied using numerical methods. *: ' The conclusion: packaging tends to lower impedance and effective dielectric constant. This is because the fringing flux lines are pre- maturely terminated, which inereases the density of field lines in air. But when the ratio of the distance between the lower and upper walls to substrate thickness is larger than five, and the side wall spacing is five times the strip width, te effect of enclosure is negligible on microstrip character istics. Consider dispersion at high ‘The formulas for characteristic impedanee and effective dielectric constant presented’thus far have been based on ‘a quasi-TEM mode of propagation. At lower frequencies, this is a good statie approximation of a dynamic structure. However, at higher frequencies the effective dielectric constant and characteristic impedance of a microstrip line begins to change as frequency, increases making the trans- mission line dispersive."% '* This dispersive characteristic is due to propagation of hybrid modes, eontinned om p. 176) 400 =) an) frequencies 2. Dispersive effects raise the effecive dielectric constant (ug) slightly as frequency is Increased. 175 ‘A DESIGNER'S GUIDE TO MICROSTRIP: ‘The frequency dependence of the effective dielectric ‘constant describes the influence of dispersion on the phase velocity, whereas the frequency dependence of the effective width describes the influence of the dispersion on the characteristic impedance." The phase velocity in microstrip line decreases with increasing frequency, hence ‘i in- fereases with frequency. The characteristic impedance of microstrip line increases with frequeney, so the effective width must decrease with frequency. Fortunately, changes in «and % with frequency are very small. However, the frequeney below which dispersion effects may be neglected is given by the relation"™ f, (Gi) = a8, / hve Equation 12 shows that f, is higher for high-impedance lines, fn thin substrates, ‘The numerical analysis for dispersion in shielded as well 8 open microstrip transmission lines has been treated extensively.*-# These analyses, however, require extensive computations and fail to provide insight into the dominant physical phenomenon at work. For these reasons, the ‘numerical approach is not convenient for microstrip circuit esign and therefore, not diseussed in this paper. Both experimental and empirical attempts to deseribe microstrip dispersion have also been reported.!® 3-1” The empirieal expressions reported for dispersion are limited in terms of applicability and suffer from inadequate theo- retical foundation. Analytieal formulas for dispersion which agree closely with experimental and numerical results have appeared just recently. ‘These analytical expressions, by Getsinger"* and Carlin”, fare very similar. But the results given by Getsinger are loser to experimental as well as numerieal results. The dispersion in ee is given by ¢h in em) (12) edd) 1+ GEA, 1. "oan G = 06 + 0,009 7, Here, frequency, f, is in GHz and substrate thickness, h, in em. It ean be seen from Bq. 13 that for >, aa (0) «a In other words, high-impedance lines on thin substrates are less dispersive, Experimental and theoretical results for various mierostrip geometries are compared in Fig. 2. There is a close agreement between the calculated values (Eq. 18) and experimental values Although many researchers have attempted to describe the effect of dispersion on cg, there are fewer analyses hich deal with frequency dependent behavior of Z,. Krage ‘and Haddad and Knorr and Tufekcioglu®, for example, describe the increase in characteristic impedance with frequency by numerical analyses. The agreement between the results given by these two analyses is reasonably good. Recently, closed-form expressions for Z, (f) based on a paralel-plate model of microstrip ine have been reported.** ' These expressions are: Tip = —— it he ial) Veg ‘The effeetive width, Wy (D, is given by Wa (0) = W 1+ iF where Way (0) is obtained from Eq. 14 when f = ‘The variation in characteristic impedance with freq is shown in Fig. 3. The solid curve is arrived at using Ba 14 and 15, while the dotted curve is the one reported b Knorr and Tufekeioglu. The increase in 2 (f) (for ¢* 10 and W/h = 1) is only 4 per cent from DC to 10 Gli, which is quite small. This change cannot be eonfimd experimentally sinee, at 10 GHz, transitions pose a am siderable problem in accurate measurement. Therefore, effect of dispersion on Z, ean he generally neglected Wey (9 = W + Two mechanisms contribute to loss Attenuation constant, a, is one of the most importa characteristics of any transmission line. There are tm sources of dissipative losses in a microstrip cireuit:eondax tor loss and substrate dielectric loss. Assuming a uniform current distribution across st width and ground plane, conductor loss may be appr imated as: a= SSR, aByem ao) w The surfece resistivity, R, for the conductor is gives i R= ve Fale where is the free’ space permesbi conductivity of the microstrip material Tt should be noted that thi tor loss is valid only for very wide strip widths (W/hs} There isa very large discrepancy between the experimen data and this expression at practical values of W/h Hag ever, Eq. 16 can be brought closer to reality by consdeag nonuniform current distribution. Expressions for the oat ductor loss derived by Pucel™ in this manner are wey accurate Por Wh < V2, = SER pe 2a Zh Ww, + ets mt (Continued on 43. Characteristic impedance increases somewhat frequency. But note that there is only a 4 per cent change from DO to 10 GHz. MICROWAVES * May, For V2 < Wh < 608 R a> SBR peg 2+ Zh For Wh = 3 68 ah aa} of where: or a fired characterise impedance, conductor les de Cress versely wth subnets thickness an nceases ‘Th the auare rato requnc Ie anes fas in mie Tiss importa Gilneas sikcorare aretis Tojeng Gaal oe Xevution ae considered Welch and Prats and Scneder™ ried the spresica for in sueutoe eto or a flere with an anges tanh vee blo a= m3 i tam Goa . aB/em where , is the free space wavelength Foro ¥ 0, ay = 434 Se} _ # aB/em Vee 6 is the free space permittivity (as) where Dielectric losses are normally very small compared with conductor losses for dielectrie substrate, Dieleetrie losses in silicon substrates, however, are usually in the same order as, or larger than conductor losses. The reason for this is, ‘that resistivities higher than few hundred ohm-cm are sikemet 4. Losses in semiconductor substrates are substantially higher than those in ceramic or quartz material. Silicon introduces high losses due to the difficulty of growing hhigh resistivity material 178 ~ ‘A DESIGNERS GUIDE TO MICROSTRIP difficult to maintain for Si, However, higher resistivity au be maintained in GaAs, and hence the losses are less fi this material re 4 compares the total loss for 50-ohm micros lines on silicon, GaAs, alumina and quartz substrates. ‘obvious from the figure that silicon MICs are more las fads circuits will have more loss than ceramic circu because the semiconductor substrates are thinner. Lines quartz have the least loss Quality factor depends on substrate ‘The quality factor, Q, of « microstrip line ean be reid to the total loses in the line by® a where Qy is the total resonator Q, ap is the total ls the resonator and 3 = 2r/A,. Mierostrp line Qs are low than the Qs of coaxial or waveguide transmission live When the loses in a resonant line are considered (ut as 4/2 of A/4 resonators) another loss factor, «, dae raulation atthe discontinuities must also be considend ‘The corresponding radiation Q-factor is given bys 7, Q By © aah where fd®) + ak) — 1 ah) e+ Tei ‘Note thatthe effeet of dispersion is considered, as descr by Eq. 18. The total Q of the resonator ean be express by (eA P ety ty (2) ® @ ee Here, Q., Qsand Q. are the quality factors corresponding to eonduetor, dielectric and radiation losses, respective Finally, the eireuit quality factor, Q,, ean be defined Ady tke + a Qe ®& Vb The variation of Q,, Q, and Q with frequency fore quarter-wave resonator on GaAs, lumina end qed strateslsehown in Fig. 5. A quarter-wave 0-ohm rs on 25milthek lamina substrate has Q, of about 20 ‘2 GHz and 550 at 10 GHz, whereas Q, is 230 at 2 GHz and nearly 160 at 10 GH, Tiss de to the fact hat radian Tower are higher than eonductor and dielectric lone igher frequencies. A quarter-wave, S-ohn resonaa Tovmil Gras subotrate has Q, of about 82 at 2 GH td 160 at 10 GHiz, whereas Qy in 82 at 2 GHa and ner 1 at 10 GHz. This de to the fact that radaton lowes Smaller than conductor and dieectre loses for thin ok Strats at Higher frequencies. Thus, the commonly acc Tule for high-Q microstrip circuits using thick subta dove not apply ue to High ration late incurred un this condition (2) Moding limits high-frequency operation ‘Maximum frequency of operation in microstrip lini limited by the exeitation of spurious modes in the form d (Continued on 9. MICROWAVES * May, 1977 —————_——-A DESIGNER'S GUIDE To MICROSTAIP. surface waves and transverse resonances. Surface waves are TM and TE modes which propagate across a dielectric substrate with ground plane. The frequency at which significant coupling occurs between the quasi-TEM mode and the lowest-order surface wave mode is given by" For «,>10, Eq. 28 reduces to fr (GHe) ~ 108 (his in em) a hye Cutoff frequency, fr, decreases when either the substrate thickness or dielectric constant is increased, ‘Thus three limitations—maximum substrate thickness, minimum Q and surface wave excitation—define a region of useful microstrip line operation (see Fig. 6). From the saan) 3 usable region of Fig. 6, one obtains a range of substrate arene thickness which should be used for microstrip line cireuits For « = 9.7, this range is: 023 Shs 18 em @ 2 GHz 0.01 < h < 0.36 em @ 10 GHz 0.01 10 mil) require dimensional, tolerances in the order of 0.5 mil for satisfactory per formanee (i.e, VSWR= 1 + a 2/2, ~ 1.03). ‘Some final notes on power handling ‘The peak power handling capability of microstrip line is oor. Although it is limited primarily by the sharp edges of the strip conductor, in some cases, connectors or launchers decide the power handling capability of the microstrip line. One additional factor which may signifi cantly reduce the power handling capability is the effect of internal mismatches, Howe” has reported successful operation up to 10kW at S-band and 4kW at X-band, Peale power data is not readily available and difficult to ealeulate accurately. ‘The average power capability of microstrip line is. in- fluenced by the temperature rise of the strip conduetor and the supporting substrate. Iti, therefore, related to the loss tangent and thermal conductivity of the substrate (low loss tangent and large thermal conductivity will increase the 7. Fabrication tolerances become more important as the ‘aspect ratio grows larger. 192 A DESIGNER'S GUIDE TO MICROSTRIP. ‘average power capability of microstrip lines), It has b reported that a 50-ohm alumina line (50-mil) ean eal hhandle 100 W of CW power. # ‘eevee wi than Mr, ORE SE te hah tal msn 2 parte amon Sore, ‘Aperunn TEBE rs irr hs Mek Va Wee en file ae peu hte aa ae Mere Tier Pk ral 8 AGE ie Met a Cr ei ray al at SR ge i eset “a jt one ri Ene ge Fes eat ios meee se “Tae Ta HEE pt ihe roe ay Re ae phe teal St ate ee aaa ‘her Bi i ee ae Pre IESE, Vol. 6 yp 144146 BENQ My Digan Mon” ESE Tr, Merve ari Se Mo Per Marini. Stee Mire, amc Megane Ce, “ve Na pt th i SA, iar Fea Pee api cae Meir Patg a il Tics Sri ica wan he Bi MICROWAVES * May, 19 a ey o aa ee Orevra

S-ar putea să vă placă și