Scribd Logo
Upload

Welcome to Scribd!

  • 9 views

    Microstrip-SlotlineTranstion MTT88

    Uploaded by

    dfghjklvbhnm

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content

    You might also like

    Microstrip-SlotlineTranstion MTT88

    Uploaded by

    dfghjklvbhnm
    9 views11 pages

    Original Title

    Microstrip-SlotlineTranstion_MTT88

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    9 views11 pages

    Microstrip-SlotlineTranstion MTT88

    Uploaded by

    dfghjklvbhnm

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    of 11
    WHE TRANSACTIONS ON CHOWAN THEORY AND TECHN, VO, 36,80 6 AGUS 98S Microstrip /Slotline Transitions: Modeling and Experimental Investigation BERND SCHUPPERT Abstract —In heparan ans of microsite wanton Is item wing a newer sition through trarattcietine models DH Keren ramon sch tron containing warm and ora lines awl eed and vray sored mrntip linn nl be teeted, The vali of the modsting ress verified experimentally ty ‘eng mme etiet t« cn of fu tes Sloine nthe frequency rage from 1 0 16 GH For fractal plo Sega cane ae pooner €5 mb na 1. Ivrropucrion ANY AUTHORS have dealt with the problem of developing a broad-band transition from microstrip 16 slotline {1}-[6]. Some intuitive approaches to this prob- Tem have been presented with a reported bandwidth be- tween about 1 octave [1] and 1 decade (3), the latter in the frequency range from 1 to 10 GHz. 'A problem quite similar to the subject considered here hhas been treated in (7}, where a nonplanar compensated balun has been presented for frequencies below 1 GHz, However, some basic results are similar to those given here. Due to the fact that a systematical description of this problem is still lacking, the present paper is intended to give an approach to this problem based on the use of transmission-line models. Inthe case of creular-area stubs, these lines will be synthesized as piecewise uniform lines ‘The validity of this rough approximation will be verified by experimental results ‘The transitions mentioned in this paper are sketched in Fig. 1. Concerning experimental verification, the question fof how to measure the frequency response of 2 single transition arises. A measurement of only the VSWR or reflection coefficient at the microstrip side when the slot line is matched (2, [4}-[6] ignores matching through radia- tion effects of the slot stub. Moreover, increasing line loss at higher frequencies also yields a low VSWR of the transition. Both effects suggest a bandwidth of the transi tion that may be too large. ‘When measuring the transmission and/or reflection coefficient of two transitions connected through a slotline ‘of length Z, (3, (6), impedance interactions occur between the transitions. As'a result, the measured bandwidth of such a configuration depends on the distance between the transitions, whichis somewhat arbitrary Mano recived October 30,187; revised Match 16,188 {The aihor ss wah the Technische Uniessst Bev Last fr Hochequensieck, D100 Bern 10, Wet Gera TERE Log Number #8276, I _) t 6) @ ry Microsite tnstions unde consideration; sd es: Fig. ‘icrostrp culty: dased lines sine cist a) Slers mire ‘inp shor and uniform 8/4 oie ©) Vaal Sort th imo AE pen scrosnp and worm 174 atin) Sldered mic ‘hort ta sine open creat troegh aor evi sotine (8) Great trough nonunion cea sine TL. SipLiriep Mopét For CincUrT ANALYSIS In order to clarify the effects, let us first consider simplified circuit models. It will be shown that the transi- tion of Fig. 1(b) offers ether an improved flatness of the frequency response or an improved bandwidth by means of a reactance compensation effect [7] compared to the transition of Fig, 1(a). This reactance compensation effect also takes place in the ease of nonuniform lines. However, if these Hines are of a circular shape (Fig. 1(@), only an improved flatness is observable, whereas the bandwidth is somewhat smaller than in the case ofa soldered transition Fig. 10). It should be pointed out that in the following approxi ‘mations any reactances due to the discontinuities are ne- lected Tm the presence of uniform microstrip and slotlines, equivalent cireuits for both of the transitions of Fig, (a) and (b) are shown in Fig. 2, where the soldered wire can be represented by a series inductance, which, however, will be neglected, Equivalent circuits as simple as those given in Fig. 2a) and (b) imply that perfect coupling between the ‘microstrip and slotline currents takes place If we choose the characteristic impedances of microst and slotine to be 2,7 2,=509 a) it is convenient to introduce normalized impedance factors for shorted slot stub and open microstrip stub, respec- tively @) 8) (0018-9480 /88 /0800-1272801.00 ©1988 IEEE. \ ® 1 pen ie shorted let tub Ore eee seca Fig. 3.097 dB bandwidth agaist impedance ator Wansdon "ov seldeted short ad wal shst th N ‘ial shot, w= 2618 (optima band » Fig. 2 Simplified equivalent ccs fa microsite anton (a) with sedeed wie (6) wih itl Sor For simplification let us assume slotline stub and micro- strip stub to have the same electrical length (By/— Bl, = Bl) Analyzing the transition of Fig. (a) (equivalent circuit Fig. 2G), the magnitude of the transmission coefficient becomes 1 Wal “ : alae eats fore sini gale cit of Fi 2 singe tan ang anf ln Hy ee hes ‘A ~0.97 dB bandwidth can be defined when choosing fang: nee te In the ial short (Fi 2(b)) the ete In the case of a virtual short (Figs. 1(b) and 2(6) ane eee ‘magnitude of the transmission coefficient becomes =r -aean Aonm r SI = Alcoa arctan» pee Pe > Go 4 which is plotted in Fig. 3. It should be noted that the Sareea ~0,97 dB bandwidth of the insertion loss corresponds to a VSWR = 26, ifthe network is lossless dominant at the band limits exe HANSACTIONS ON MICLOWAVE THORY AND TICHMOUES YOU 36,80. 8, AUGUST 1988 TABLE L Iurtuence oF Vanous Vaz. StoRE ConOVHONS ON THE BAXDBADTH OFA SINGLE THANSITON Tnpedinse ‘Bandi Factor Brame Sader shore © Bones Viruslshet wee Moyen Viewalshont #26 Bane BX ge irl shor 180 Ret 13 Rue ial short oe and it is evident from the above equation that choosing w yields a maximum flat frequency response in the passband, Tn this case where v=w, the ~097 dB bandwidth remains unchanged between soldered or virtual short, so (6), as well asthe calculations of Fig 3 is also valid when dealing with a virtual shor! “To clarify the results discussed above, numerical caleula- tions of the frequency response have been carried out for the equivalent circuits of Fig. (a) and (). Due to the symmetry properties of the frequency response itis suffi cient to ealeulae only the lower frequency range. It can be seen from Fig. 4 that the bandwidth is im- proved with increasing = w, as already predicted by Fig, 3 and {7}, The ~0.97 dB bandwidth isthe same in ease of fa soldered or a virtual short with = w. Furthermore, an improved flatness of the Trequency response in the pass- band is noticeable. ‘However, a more interesting case is that of an improved ‘bandwidth by means of a virtual short. Considering (7) an extreme value of [Sy] can be found through aisal 2 ) leading toa maximally fat frequency response if cot? woafne ) 6 and an optimum value of 1 3l- “Toot 1S Ge Feet Bi Considering the ~0.97 4B frequencies of a soldered short, which are given through (5), an optimal relation between both impedance factors can be deduced by insert ing (5) in (9), yielding (10) dw, a) In this case, the transmission coefficient [S| atthe ~0.97 AB frequencies ofa soldered short i inereaSed optimally at a value of |S,x|=~01 4B. If we consider the maximum fachievable ~0,97 dB bandwidth of the optimal ease de- scribed by (10), we get an optimal relation between both, {impedance factors: w= 26180 2) alt peo ot Sieted See Remarks ‘ora coord 1 fsqueny the pashnd ‘maximal ~0.97 sh bandwidt The soled stor Fig 5. Calculated sls for the simpli equals circuit of Fig 2 fora singe anion with ufo stabs, lower lequeney ngs: and a bandwidth of 1 arctan 555 T il-oman Ai\-osran Ate atin Tae as sketched in Fig. 3 (dotted line), yielding an optimally increased bandwidth compared tothe soldered short of (3) of approximately Bogs ™ (12-13) Bats (a4) ‘Summarizing this section, a comparison between sold- ‘ered and virtual short is given in Table T and depicted in Fig. 5 IIL, Traxstrions writ UNIFORM MICROSTRIP ‘AND SLOTLINES. ‘This section is intended to verify the results of Section TL experimentally. For practical applications, these transi tions offer a smaller bandwidth compared to transitions incorporating nonuniform stubs, which will be treated in Section IV. 'A transition with uniform lines has been treated in (6) using an equivalent circuit which is quite similar to that given in Fig. 2b), yielding 2 bandwidth of 1.45 octaves in the frequency range from 26 to 7.1 Gl, ifa VSWR <15 is accepted for two transitions in cascade. slot line in microstrip line Fig 6 Mirostp/slotine wanton wth uniform stabs Fig Cla onion cc ol sg mit ‘ral mirostp thos. Zn, $0 Gy vo" vital micron hor, aan A. Single Transitions ‘The transitions considered here are sketched in Fig, 2a) and (b), respectively. A more detailed view of @ transition is given in Fig. 6, The dimensions are chosen for a center frequency of about 5.0 GHz and an alumina substrate with €,=9.7 and a thickness of h= 0.635 mm. Using the for mulas given in (8), the following microstrip dimensions were used: microstrip width w, = 0.57 mm and microstrip stub length (in case of a vietal short) [,=5.7 mm, The slotline characteristics have been calculated with a com: puter program referring to [9}, The dimensions are slotline width w,= 0.05 mm (yiekling a characteristic impedance which slightly exceeds the value of $0 9) and a slot stub length of /, = 7.3 mm, Tn the case of a soldered short, a 0.57-mm-wide copper sheet is fed through the substrate and soldered at both sides. ‘The caleulated results for both virtual and soldered iicrostrip shorts are given in Fig. 7. The ~097 dB ‘bandwidth is approximately the same in both cases, con: sistent with the results of Section I, if both impedance factors are equal (w= 0). The small differences between the upper and lower ~0.97 dB frequency limits in the case of soldered and virtual shorts are due to slotline disper: transition] | sot tne eansition Fig. Cascade of two wanton separated by alone sion. The achievable -097 4B bandwidth is about 1.7 This bandwidth can easly be increased if we remember that an impedance relation of w = 2618P (eg (12)) offers an optimum ~097 dB bandwidth, For this, a characters- tic impedance of about 19 2 is necessary for the open- ended microstrip stub, which corresponds to 2 microstrip stub width of 30 mm. Due to the changed dispersion characteristics, the microstrip stub length is chosen to be J, 5.2 mm, The results are also given in Fig. 7 and it can be seen that the ~0.97 dB bandwidth is increased to @ value of about 1.9 octaves by simply broadening the mi- p, the results of Fig. 13 seem to be ‘contrary to this prediction, To explain this fact, let us ‘consider the relation w/o for different slot stub and mi- crostrip stub diameters, where a OE as) soa S (18) Zn, ital mrostp shor wih iferent mares stb meer ‘The results are plotted in Fig, 14, where the characteris tic impedances in the middle of the stubs have been chosen for comparison. It can be seen that the relation (W/0) nas ‘exceeds unity only if both stubs are strongly detuned. This isthe case if the electrical lengths of both stubs are too ferent for & reactance compensating effect. Due to the fact that w/o <1 in the range of interest (f,!,~ Buy) the results of Fig. 13 are absolutely consistent with the basic result of Section I ‘The restriction w//e <1 means in practice that the char- acteristic impedances of a circular microstrip stub cannot bbe made as low as necessary for a successful bandwidth improvement. For design purposes itis useful to know optimal combi nations of slot stub and microstrip stub diameters. The ‘modeling results for various transitions consisting of dif- ferent slot stub and microstrip stub diameters are sum- ‘marized in Fig. 15. ‘Compared to the 0.97 dB band limits of a soldered transition, the upper and lower band limits are shifted towards lower frequencies if an optimal combination of slot stub and microstrip stub diameter is chosen. The shift of the lower band limiting frequency, which is typically below 100 MH, is depicted in Fig. 16, whereas the shit of the upper band limiting frequency is typically of the order of a few GHz (Fig. 17) As an example let us compare a transition with soldered and virtual short, respectively. For instance, let us consider senpexr: sacuostur/SLomLMe RANSTIONS 0 2 & 6 8B d.mm ig 15. Optinal misostip stab dimer sais sot sth dante nodling ets 7 d,lmm Loe Fie 16. Calelted fouecy si ofthe her —097 dB band tit “fr dierent aicoip sto ancy lated Yo the loner bad bts ofa sokced mos sbr anton 4 transition with a circular slot stub with a diameter of 4,=40 mm ‘A transition with a soldered microstip short has ~0.97 4B band limits of 1.65 GHz and 1491 GHz (Fig, 12), yielding a bandwidth of 3.18 octaves (Fig. 13). In the case fof a virtual short with an optimal microstrip stub diameter of Jy, = 4.0 mm (see Fig. 15), the lower band limit is shifted 45 Mllz toward lower frequencies (see Fig. 16), yielding a lower band limiting frequency of about 1.6 GHz The ‘upper band limit is shifted 1.7 GHz toward lower frequer cies (see Fig. 17), yielding an upper band limiting frequency of about 132 GHz and a total 097 dB bandwidth of 3.04 octaves (Fig, 13) B. Experimental Verification AAs in the case of uniform slotlines and/or microstrip lines (Section IID), experimental verification has been car ried out by considering a cascade of two transitions sep- etoned optimum" FL, eturee 10 = 4 Fig 17. Calculated freuen sift ofthe oper ~097 4B band iis Tor erent nitsp sab damete. elated to he Upper band ofa soldered microst-shoettasion arated by a slotline (Fig, 8). Due to the fact that we now deal with transitions which offer a large bandwidth, prob- Jems arise from our measurement setup. Even though an automatic network analyzer system was used, itis difficult to achieve reproducible results between microstrip calibra- tion and measurement procedure, especially at frequencies above 12 GHz. Reproducibility problems are mainly due to nonsoldered coax/microstrip transitions. Thus, seatter- ing of measured data points increases with increasing frequency. Moreover, two-port error models forthe transi tions between coaxial and microstrip line may be 100 simple to achieve an error-compensated measurement. ‘The reproducibility of the transmission coefficient ifthe substrate is removed ftom the mount and fixed again has bbeen found by measurement to be within 0.8 dB in the frequency range from 12 GHz to 16 GH, whereas the error is negligible at Frequencies below 10 GHz. ‘The experiments reported here are limited to three dif- ferent slot stub diameters in conjunction with a soldered ‘microstrip short and one virtual microstrip short, respec- tively (see Table II), The dimensions (slot and microstrip stub diameters) are chosen in order to realize fixed lower frequency limits of 1, 2, and 4 GH, respectively. It must be pointed out that these dimensions do not agree with ‘optimal combinations of d,, and d, referring to Fig. 15 and hence do not offer the maximally achievable band- wid, ‘The experimental results together with the modeling results are given in Figs. 18-20. The theoretical and ex- perimental results agree satisfactorily, especially at lower 180 IEEE TRANSACTIONS ON MICHOWAYE THEORY AND TECHNZQUE, VOL. 36,808, AGUS 1988 TABLE I CCatcutaren BaNowion oF SNE MichostRF/S.OTLINE TRANSITIONS * Slot Stub Diameter “097 dB band aad igh ta 25 29.205 Gi 50 16. Beane &o O98 13 Ge inal Spore “ =097 dB band ts tion) dy fmm sige estos) 20 403-05 Gkie 30 205... 128 GH Ed Ton 105 Ge 3 1p FREQUENCY «KE TRANSMISSION COEFFICIENT . dB / Fig 18, Fig 18 Teogth 1,18 mm Theocucl (>) apd expenmenta ® “Transmision coefcient of 3 cascade of two mitostp sine ranstons, spate by 4 sotine of (re (a) crop shot soldered sone open cla,

    You might also like