TUTORIAL A GENERAL D ESIGN PROCEDURE FOR BANDPASS FILTERS D ERIVED FROM LOW PASS PROTOTYPE ELEMENTS: PART II P art II of this bandpass filter tutorial features several examples that illustrate the general design procedure for bandpass filters using resonators and coupling methods of different types. BANDPASS FILTER DESIGN The general bandpass filter design procedure utilizes the normalized, low pass elements of the prototype filter to determine the r e q u ir e d r e son at or …several examples that coupling and coupling to the external circuit, illustrate the general design that is, the source and procedure for bandpass filters load. In each case, the e xt e r n al sou r ce an d using resonators and coupling load are assumed to be methods of different types. 50 Ω, although the filter could be matched to other real impedances. The general design procedure is listed in Table 1. To illustrate the general bandpass filter design procedure, several examples are offered. It should be mentioned that the coupling parameters, that is, the capacitor matrix, or even and odd mode impedances, for several types of distributed resonators, may be determined with the aid of an electromagnetic (EM) simulator. This is an extremely valuable design tool because it eliminates the fabrication of models in order to determine the coupling of symmetrical, distributed resonators. The illustrated examples include a lumped element, Chebishev, bandpass filter using πtype resonators, a coaxial cavity bandpass filter using λ/8 resonators with capacitive tuning (comb-line), a coaxial cavity bandpass filter using λ/4 resonators (interdigital) and a half wavelength via-hole direct-coupled filter. Also described are a dielectric resonator, a bandpass filter using disk-shaped resonators, an inductively coupled filter in coplanar waveguide and a stripline or microstrip, Chebishev, bandpass filter using λ/2, side-coupled resonators. Lumped Element Bandpass Filter The lumped element, π-type resonator is used in a five-section, 0.01 dB ripple, Chebishev filter. The low pass prototype elements (g-values) may be calculated using the equa- K.V. P UGLIA M/A-COM Inc. Lowell, MA Reprinted with permission of MICROWAVE JOURNAL® from the January 2001 issue. © 2001 Horizon House Publications, Inc. TUTORIAL TABLE I BANDPASS FILTER DESIGN PROCEDURE TABLE II Determine the Filter Parameters: Center frequency, f0 Bandwidth, ∆f In-band ripple, r dB Number of resonators, n Rejection requirements, R(f') Insertion loss, L(f) Resonator type: lumped element, comb-line, interdigital, coaxial cavity, dielectric … Special conditions: MFTD, Bessel, Gaussian … Determine the Low Pass Prototy pe Elements: Calculate elements Acquire from tabular data Calculate Coupling Coefficients: k i,i +1 = ∆f fo 1 for i = 1 to n –1. g i g i +1 Determine Resonator Reactance or Susceptance Slope Parameter: π-TYPE RESONATOR BANDPASS FILTER DATA i gi Cπ,I,I +1 Cia Cib 0 1.0000 16.4 1 0.7563 5.1 37.7 45.6 2 3 1.3049 3.5 45.6 47.1 1.5773 3.5 47.1 47.1 4 1.3049 5.1 47.1 47.1 5 0.7563 16.4 45.6 37.7 6 1.0000 α or β. Determine Coupling Reactance or Susceptance: Ki,j+1 or Ji,i+1 for i = 1 to n–1; Ki,i+1 = αki,i+1 and Ji,i+1 = βki,i+1. and the susceptance slope parameter is calculated as β = 1(2πf0)C = 0.063662 mhos The resonator coupling elements are determined using Determine Input and Output Coupling Parameters. Determine Filter Elements (for lumped element resonators). Determine Filter Dimensions (for distributed resonators). Optional: 2πf0C π Computer simulation Estimate filter insertion loss Estimate filter group delay ▼ Fig. 1 = βk i, i + 1 for i = 1 to i = n – 1 Schematic of a π-type resonator. L C i, i + 1 C BANDPASS FILTER DESIGN PROCEDURE The bandpass filter design procedure is listed in Table 1. The values of the inductance and capacitors may be determined using 1 f0 = 2π and the input and output coupling are calculated as 2πf0C 0,1 = and 2 = 100 MHz LC 2πf0C n ,n + 1 = where tions within the appendix or may be determined from available tabular data. The filter parameters are f0 = 100 MHz ∆f = 5 MHz n=5 r dB = 0.01 dB g0 = g6 = 1.0000 g1 = 0.7563 g2 = 1.3049 g3 = 1.5773 g4 = 1.3049 g5 = 0.7563 The unloaded quality factor for this type of resonator at this frequency is Q u = 250. A schematic of the resonator is shown in Figure 1. L≈ 10 = 100.0 nH f0 and C= 2 (2πf0 )2 L = 50.66 pF The inductance formula provides a good estimate of a convenient value of the resonator self-inductance from which the resonator capacitance may be calculated. Next, the coupling values are calculated using k i, i + 1 = ∆f f0 1 g i , g i+1 for i = 1 to i = n – 1 ∆f β f0 g0 g1Z0 β ∆f f0 gn gn + 1Z0 The input and output coupling methods need not be the same as the resonator coupling. However, the input and output coupling must satisfy the relationship βZ0 g gf Qe = = 0 10 2 in ∆f J ( 0, 1 ) and Qe out = βZ0 (Jn ,n +1 ) 2 = gn gn + 1f0 ∆f Tab le 2 lists the π-type resonator, bandpass filter data. A schematic for the five-section, πt yp e r e son at or b an d p ass filt e r is TUTORIAL n ∑ gk = 1.98 dB k =1 The computer simulation of the transmission group delay is also presented in the results. The midband group delay may be estimated from the formula ( ) τd f0 = 1 2π∆f n ∑ gk = 181 ns k =1 It should be mentioned that the on ly filt e r e le m e n t s wh ich we r e tuned in order to achieve the desired passband were the inductors of the πresonators. COAXIAL CAVITY BANDPASS FILTER λ/ 8 RESONATORS The second bandpass filter design example is the coaxial cavity type filter using λ/8 resonators with capacitive tuning or comb-line filter, so called because the resonators are fixed to ground on a single surface in much the same configuration of the teeth of a comb. This is a very popular filter structure because a moderate value of unloaded resonator quality factor may be achieved (and therefore low loss), and because a wide rejection band is indigenous to the structure by virtue of the wide separation in resonant modes. The resonator structure employed for the example filter is that depicted in Part I. Unlike the traditional comb-line filter, which has no metallic obstacle between resonators, the filter example employs rectangular cavities with coupling slots as shown in Figure 4. This structure yields a somewhat higher unloaded quality factor because more of the field is enclosed. The coupling between resonators is controlled via the width of the slot, w. The coupling coefficient between resonators may be determined by measurements of symmetrical-cou- 5.1 100 3.5 45.6 47.1 45.6 100 3.5 47.1 47.1 100 47.1 45.6 100 5.1 16.4 45.6 37.7 INDUCTOR VALUES IN nH CAPACITOR VALUES IN pF ▲ Fig. 2 Schematic of a five-section π-type resonator bandpass filter. pled resonators as a function of the slot width or via analysis of the self and mutual capacitance using an EM simulator. Both methods were employed and resulted in acceptable agreement, that is, less than 5 percent over the slot widths used in the filter, between the experimentally determined coefficients and the measured values as shown in Figure 5. To illustrate the filter design procedure a PCS1900 transmit band filter (1930–1990 MH z) is required with minimum insertion loss (< 1.5 dB) and volume as the principal design specifications. A filter rejection of 60 dB, minimum, in the PCS1900 receive band (1850–1910 MH z) is also required. From the earlier data, an eight-resonator filter is required to achieve the rejection requirements. This may be verified from direct calculation using L (f a ) 10 –1 –1 10 cosh r dB 10 – 1 10 n= cosh –1 ∆fn ( ) where L(fa) = the attenuation required at the specific rejection frequency fa r db = the in-band ripple factor ∆fn = the normalized bandwidth at the rejection frequency, ▼ Fig. 4 Coaxial cavities with slot coupling. w that is ∆fn = 2 fa – f0 ∆f Because the reject band is close to the passband, the passband was deliberately narrowed to 55 MH z so that coupling adjustment could be facilitated in order to increase the final filter bandwidth to 60 MHz. The filter parameters are f0 = 1960 MHz ∆f = 55 MHz n=8 r dB = 0.05 dB π θ0 = 4 Q u = 2250 1 = v0C g mhos Y0 = 60 ▼ Fig. 3 (dB) 4.343f0 ∆fQ u 37.7 100 21 ( ) IL f0 = 16.4 S11 (dB), shown in Figure 2. A computer simulation was conducted on the fiveresonator bandpass filter using only the inductors as variable or tuning elements. The results of the simulation are shown in Figure 3. Note that the bandwidth is 5.12 MHz as compared to the design bandwidth of 5.0 MHz, and that the midband insertion loss is approximately 2 dB. The midband insertion loss of a bandpass filter may be estimated from the equation Computer simulation results. 0 800 −10 700 −20 600 −30 500 −40 400 −50 300 −60 200 −70 100 −80 90 100 110 FREQUENCY (1 MHz/ div) 0 TUTORIAL The LP prototype g-values are g0 = 1.0000 g1 = 1.0437 g2 = 1.4514 g3 = 1.9899 g4 = 1.6502 g5 = 2.0457 g6 = 1.6053 g7 = 1.7992 g8 = 0.8419 The coupling values are calculated using K i, i + 1 = ∆f 1 • f0 g ig i+1 The slot width may be determined from the measured coupling data of symmetrical coupled resonators or from an EM simulator that provides the capacitance matrix data for the particular structure. F or this example, both techniques are utilized. F irst, the measured coupling data from a pair of symmetrical coupled resonators was subjected to polynomial regression of degree three. Next, the coefficients of the polynomial were used to generate an equation for the coupling coefficient as a function of the slot width. Finally, a plot of the coupling coefficient was used to determine the slot width for each coupled section. The measured coupling data is displayed in Table 3 for three values of slot width for square cavities of 0.875 inches and center conductor diameters of 0.375 inches. These dimensions produce a resonator characteristic impedance of approximately 60 Ω. Performing a polynomial regression produces an equation for the coupling coefficient Kc as a function of the slot width, w such that [ ( )] ( ) π 1 + K i, i + 1 2 Filter Simulation In order to conduct a simulation of the filter characteristics, an equivalent circuit is required. A suitable equivalent circuit may be constructed wit h t h e aid of t h e t wo-wir e lin e equivalent of the coupled line shown in Figure 7. Y0 cot θ0 + θ0 csc 2 θ0 2 and the mutual coupling capacitors β= β • k i, i + 1 C g = υ0 2 The filter configuration is shown in Figure 6. Note that all resonators are the same diameter and that coupling to the input and output is via direct “tap” or contact to the first and eighth resonator at a low impedance point on the resonators. Alternate coupling mechanisms are via capacitive probe at a high electric field point of the resonator or via transformer coupling. Generally, the tap is preferred because it eliminates the need for another machined part, provid e s a m or e com p act filt e r, an d makes the filter’s input and output a direct short to ground at DC. The design data for the comb-line filter is listed in Table 4. The correlation coefficient R for the polynomial regression was 1.00. Alternatively, an E M simulator may be utilized to determine the selfcapacitance C g and the mutual capacitance C m for various slot widths. Subsequent data may also be subject to polynomial regression analysis to determine the correct slot width for each coupled resonator. The susceptance slope parameter is calculated using Filter configuration. MEASURED 4.250 0 0.030 COUPLING K(w) C i, i + 1 = Kc(w) = –0.02717 + 0.10865w – 0.03150w2 ▼ Fig. 6 SIMULATED becomes 0.026 4 × 40 tap 0.25 Dp. mim 18 plcs 4.125 2 0.022 0.125 0.018 1.125 2.125 3.125 0 0.014 0.125 0.1875 0.125 R. typ 0.010 0.40 0.44 0.48 0.52 0.56 0.60 SLOT WIDTH w (" ) 0.125 R. typ 0 0.405 0.547 7 0.626 0.452 0.452 0.547 1.0625 1.125 1.1875 ▲ Fig. 5 Coupling coefficient for λ/8 lines vs. slot width. 0.449 4 1.625 0.325 2.125 2.250 TABLE III 0.1875 Coupling Coefficient kc 0.400 0.01125 0.500 0.01928 0.600 0.02668 1.625 2.625 3.625 0.375 Dia OD 0.228 Dia ID 0.750 Long 8 plcs. typ MEASURED COUPLING DATA Slot Width (" ) w 0.625 2.0625 1.000 0.750 Ref 0 0.200 0 Dimensions in inches TUTORIAL If the coupled line equivalence is applied repeatedly, a complete filter equivalent circuit may be realized, as shown in Figure 8, where the tuning capacitance C t, resonator quality factor Q u and external circuit coupling has been included. The complete filter was subjected to simulation and optimization using only the tuning capacitance and external coupling taps as the variable. The simulation results are shown in Figure 9. Note that the actual bandwidth is 52.8 MH z versus the design bandwidth of 55.0 MHz, and that the return loss is consistent with the 0.05 dB Chebishev response. The simulated midband insertion loss and group delay are 0.85 dB and 37.1 ns, re- spectively. The midband insertion loss and group delay estimates are ( ) IL f0 = 4.343f0 ∆fQ u n ∑ gk = 0.85 dB k =1 and ( ) τd f0 = 1 2π∆f n ∑ gk = 35.9 ns k =1 INTERDIGITAL BANDPASS FILTER (λ/ 4-LINES) The interdigital bandpass filter is another popular type of microwave filter implementation. They typically have lower loss than comb-line structures and are easier to tune. They require resonators that are fixed to ground at opposite sid e s of t h e su p TABLE IV porting housing as COMB-LINE FILTER DESIGN DATA sh own in Fi g u r e 10. Therefore, the G Coupling Mutual Slot construction is not Values Coefficient Capacitance Width gi ki,I+1 Ci,I+1 Wi,I+1 as suitable for manu fact u r e as t h e 1.0437 comb-line filter. 0.02280 0.18402 0.547 I n m ost case s, 1.4514 the resonator length 0.01651 0.13327 0.465 is sligh t ly sh or t e r 1.9899 than λ/4 (typically, 0.01549 0.12498 0.452 0.9λ/4), wh ich al1.6502 lows the filter to be 0.01527 0.12326 0.449 tuned to the center 2.0457 frequency with the 0.01549 0.12498 0.452 tuning elements just breaking the cavity 1.6053 0.01651 0.13327 0.465 wall. This facilitates both ease of tuning 1.7992 and maximum un0.02280 0.18402 0.547 loaded quality fac0.8419 tor Q u of each reson at or. Th e cou - Index i 1 2 3 4 5 6 7 8 Yse = VoCm PORT PORT 2 1 θ0 Yse = VoCg PORT 1 θ0 PORT 2 θ0 ▲ Fig. 7 Coupled line equivalent circuit. θ0 pling to the input and output is accomplished via contact at the low impedance point of the resonator. The design procedure for interdigital bandpass filters is very similar to the comb-line filter design procedure. In this case, the example illustrates the design of a 500 MHz bandwidth filter centered at 10 GHz. The filter parameters are f0 = 10 GHz ∆f = 500 MHz n=5 r dB = 0.1 dB θ0 = π/4 Q u = 2500 Zc = 70 Ω g0 = 1.0000 g1 = 1.1468 g2 = 1.3712 g3 = 1.9750 g4 = 1.3712 g5 = 1.1468 The coupling values are calculated using k i, i + 1 = ∆f 1 • f0 g i , g i+1 the susceptance slope parameter becomes πY0 4 and the mutual coupling capacitors are determined using β= C i, i + 1 = πC gk i, i + 1 4 The calculation of the resonator coupling dimensions must be precede d b y se le ct ion of t h e r e son at or ground plane spacing. In order to properly define a distributed resonator and reduce evanescent modes, certain geometric and aspect ratios must be maintained. I t was found empirically that selection of ground plane spacing in accordance with Z0 πλ 0 138 10 h≤ 32 produced acceptable results. The formula results from a resonator aspect ratio (ratio of length to diameter) of 2 × 1. The approximate ground plane spacing for a resonator characteristic impedance of 60 Ω is listed in Table 5. Ω Ω 045 GROUND PLANE SPACING Z7 ,8 ,7 =2 =2 827 3Ω Z6 Z4 Z5 ,5 ,6 =3 =3 05 01 5Ω 3Ω 01 =3 ,4 Z3 TABLE V Filter Frequency f0 (GHz) Ground Plane Spacing h (" ) 1 3.000 2 1.500 4 0.750 8 0.375 10 0.300 12 0.250 15 0.215 18 0.175 = 6 7 5 Filter equivalent circuit. 0 150 −10 125 λ/ 4 100 30 75 −40 50 −50 25 NORMALIZED MUTUAL CAPACITANCE S11(dB) S21 (dB) Ω ° OUTPUT 5° ▲ Fig. 8 8 Z θ0 45 Ω ° ° 60 4 Zc INPUT 3 Z θ0 Zc 2 45 Ω ° =4 θ0 Zc = 60 θ0 = 45° 1 5° 0 =4 5° Z1 Z2 ,2 ,3 =2 =2 04 02 5Ω 7Ω TUTORIAL −60 0 1885 2035 FREQUENCY (15MHz/ div) ▲ Fig. 9 Simulation results. For the present example, a ground plane spacing of 0.275" has been selected, which requires a resonator diameter of 0.128" to obtain a resonator characteristic of 60 Ω. This value is calculated from Z0 = 4h 1 log = π d v εr 0C g 138 Cristal2 has given the closed-form expressions for the even and odd mode capacitance per unit length for this line configuration, from which the self and mutual capacitance may be calculated (see Appendix A). The graph of the mutual capacitance versus resonator center-to-center spacing is shown in Figure 11. The design data for the interdigital filter is listed in Table 6. Filter Simulation The equivalent circuit shown in Figure 12 is utilized to conduct a computer simulation. As in the case of the comb-line filter, a contact at the low impedance end of the res- 0.50 Interdigital filter construction. onator serves as the input and output coupling to the filter. The results of the computer simulation using only the end capacitance and tap points as variables are shown in Figure 13. Note that the actual bandwidth is 467 MHz versus the design bandwidth of 500 MHz, and that the return loss is consistent with the 0.1 dB Chebishev response. The sim- 0.388 0.30 0.20 0.1966 0.10 0.1498 0 0.25 0.30 0.35 0.40 0.45 0.50 CENTER-TO-CENTER SPACING (" ) ▲ Fig. 11 ▲ Fig. 10 0.364 0.40 Mutual capacitance vs. spacing. ulated midband insertion loss and group delay are 0.30 dB and 2.3 ns, respectively. The midband insertion loss and group delay estimates are ( ) IL f0 = 4.343f0 ∆fQ u n ∑ gk = 0.30 dB k =1 and ( ) τd f0 = 1 2π∆f n ∑ gk = 2.2 ns k =1 TABLE VI INTERDIGITAL FILTER DESIGN DATA Index i G Values gi 1 1.1468 2 3 4 5 Coupling Coefficient ki,I+1 Mutual Capacitance Ci,I+1 Center Spacing Si,I+1 0.03987 0.1966 0.364 0.03038 0.1498 0.388 0.03038 0.1498 0.388 0.03987 0.1966 0.364 1.3712 1.9750 1.3712 1.1468 TUTORIAL Z1,2 θ Ω ° Z2,3 θ ° Ω Ω Z The via-hole coupling reactance is calculated using 1910 Ω = 85° 85° Xi, i + 1 = 1 Zc = 60 Ω θ0 = ° Z = Ω ° 2 3 = 60 θ0 85° 4 Z = θ Ω ° = 5 Ω ° K i, i + 1 K 1 – i, i + 1 Z0 2 for i = 1 to i = n – 1 INPUT T OUTPUT 5° Interdigital filter equivalent circuit. 0 12 −10 10 −20 8 −30 6 −40 4 −50 2 −60 9 10 11 FREQUENCY (0.2 GHz/ div) 0 GROUP DELAY (ns) S11 (dB), S21 (dB) ▲ Fig. 12 X0,1 = 5° The filter is constructed on 25 mil alumina (Al2O 3) substrate material (εr = 9.9). The available resonator quality factor for this type of filter is 750 if pure alumina with high quality metalization is employed. A schematic of the filter is shown in Figure 15. The resonator coupling values are calculated using k i, i + 1 = Interdigital filter simulation results. ∆f f0 Xn ,n + 1 = n OUTPUT INPUT K0,1 ▲ Fig. 14 K1,2 Kn−1,n Kn,n+1 1 g i , g i+1 HALF WAVELENGTH VIA-HOLE COUPLED FILTER This bandpass filter design example illustrates the utilization of seriestype resonators coupled by impedance inverters. The general schematic of a bandpass filter, which employs series-type resonators coupled by impedance inverters, is shown in Figure 14. The design procedure is similar to those previously explored. The filter parameters are πZo 2 The impedance inverter values become f0 = 10 GHz ∆f = 700 MHz n=5 r dB = 0.1 dB g0 = g6 = 1.0000 g1 = 1.1468 g2 = 1.3712 g3 = 1.9750 g4 = 1.3712 g5 = 0.1468 θn = π – ∆f 1 f0 g0 g1 ∆f 1 f0 gn gn + 1 2 ( 1 φ j + φ j+1 2 3 4 5 IN OUT λ/ 2 λ/ 2 λ/ 2 λ/ 2 λ/ 2 1 2 3 4 5 IN ▲ Fig. 15 ) The actual via-hole diameter may be determined from the via-hole model within Series IV.™ This model was utilized to obtain data that was further refined using polynomial regression. The results are shown in Figure 16. The design procedure is very similar to the direct-coupled waveguide filter using inductive iris coupling found in Matthaei, Young and Jones.1 and 1 2 and and the input and output inverter values are K n ,n + 1 = R b α K 1 – n ,n + 1 Z0 for j = 0 to j = n Ki,i+j = αki,i+1 for i = 1 to i = n–1 K 0, 1 = R a α K n ,n + 1 2Xj, j + 1 φ j = tan –1 Z0 α= Bandpass filter using series resonators. 2 The coupling reactance is the result of the via-hole self-inductance, which is controlled by the via-hole diameter and/or number. The length of the resonators θn must be reduced by an amount commensurate with the magnitude of the coupling reactance using the equations for i = 1 to i = n – 1 and the reactance slope parameter is 2 K 1 – 0, 1 Z0 and ▲ Fig. 13 1 K 0, 1 OUT Half-wavelength via-hole coupled filter. TUTORIAL References TABLE VII VIA-HOLE COUPLED FILTER DESIGN DATA G Values Index i Coupling (k) Impedance Inversion (K) Via Reactance (Ω) Via Diameter (mils) Resonator Length (°) 0 1.0000 – 15.487 17.130 3.0 – 1 1.1468 0.05585 4.387 4.421 10.0 157.78 2 1.3712 0.04256 3.343 3.358 14.0 171.16 3 1.9750 0.04256 3.343 3.358 14.0 172.35 4 1.3712 0.05585 4.387 4.421 10.0 171.16 5 1.1468 – 15.487 17.130 3.0 157.78 6 1.0000 – – – – – The filter data are listed in Table 7. The via-hole, half-wavelength, direct-coupled filter was simulated using the previously displayed equiv- alent circuit. The results are shown in Figure 17. This via-hole, direct-coupled filter is particularly suited to wider bandwidth filters because the dimensions of the filter are more realizable than the side-coupled filters. There are also several implementations in addition to the microstrip medium, including stripline, finline, coplanar waveguide and slotline. The implementation of the inductive reactance, direct-coupled filter in coplanar waveguide is shown in Figure 18. This implementation results in low insertion loss if an air-dielectric is utilized. 16 12 8 4 0 2 4 S11 (dB), S21 (dB) ▲ Fig. 16 6 8 10 12 14 16 18 20 VIA-HOLE DIAMETER (mils) Via-hole reactance vs. diameter. 0 6 −10 5 −20 4 −30 3 −40 2 −50 1 −60 9 ACKNOWLEDGMENT The principal reference for the content of this article is the work of Matthaei, Young and Jones.1 Many of the concepts within this reference have been investigated and interpreted in order to provide a greater intuitive understanding of the bandpass filter design process. This text is strongly recommended for those having little familiarity with this work. ■ GROUP DELAY (ns) REACTANCE (Ω) 20 0 10 11 FREQUENCY (0.2 GHz/ div) ▲ Fig. 17 Via-hole coupled filter simulation data. w0,1 w1,2 w2,3 w3,4 w4,5 w5,6 IN ▲ Fig. 18 OUT 1 2 3 4 5 θ1 θ2 θ3 θ4 θ5 Inductive-coupled filter in coplanar waveguide. 1. G.L. Matthaei, L. Young and E.M.T. Jones, Microwave Filters, Impedance-matching Networks and Coupling Structures, McGraw-Hill, New York, 1964. 2. E.G. Cristal, “Coupled Circular Cylindrical Rods Between Parallel Ground Planes,” IEEE Transactions on Microwave Theory and Techniques, Vol. MTT-12, July 1964, pp. 428–439. 3. W.J. Getsinger, “Coupled Rectangular Bars Between Parallel Plates,” IRE Transactions on Microwave Theory and Techniques, Vol. MTT-10, January 1962, pp. 65–72. 4. A.I. Zverev, Handbook of Filter Synthesis, John Wiley and Sons, New York, 1967. 5. G.L. Matthaei, “Interdigital Band Pass Filters,” IEEE Transactions on Microwave Theory and Techniques, Vol. MTT-10, No. 6, November 1962. 6. G.L. Matthaei, “Comb-line Band Pass Filters of Narrow or Moderate Bandwidth,” Microwave Journal, August 1963. 7. S.B. Cohn, “Parallel-coupled Transmissionline-Resonator Filters,” IEEE Transactions on Microwave Theory and Techniques, Volume MTT-6, No. 2, April 1958. 8. R.M. Kurzrok, “Design of Comb-line Band Pass F ilters,” IEEE Transactions on Microw ave Theory and Techniques, Vol. MTT-14, July 1966, pp. 351–353. 9. M. D ishal, “A Simple D esign Procedure for Small Percentage Round Rod Interdigital F ilters,” IEEE Transactions on Microw ave Theory and Techniques, Vol. MTT-13, September 1965, pp. 696–698. 10. S.B. Cohn, “Dissipation Loss in Multiplecoupled-resonator Filters,” Proceedings of t h e I R E , Vol. 47, Au gu st 1959, pp. 1342–1348. Kenneth V. Puglia holds the title of Distinguished Fellow of Technology at the M/A-COM division of Tyco Electronics. He received the degrees of BSEE (1965) and MSEE (1971) from the University of Massachusetts and Northeastern University, respectively. He has worked in the field of microwave and millimeter-wave technology for 35 years, and has authored or co-authored over 30 technical papers in the field of microwave and millimeter-wave subsystems. Since joining M/A-COM in 1971, he has designed several microwave components and subsystems for a variety of signal generation and processing applications in the field of radar and communications systems. As part of a European assignment, he developed a high resolution radar sensor for a number of industrial and commercial applications. This sensor features the ability to determine object range, bearing and normal velocity in a multi-object, multi-sensor environment using very low transmit power. Puglia has been a member of the IEEE, Professional Group on Microwave Theory and Techniques since 1965. TUTORIAL APPENDIX A NORMALIZED SELF AND MUTUAL CAPACITANCE OF ROUND RODS BETWEEN PARALLEL GROUND PLANES OF ROUND RODS BETWEEN PARALLEL GROUND PLANES Upon completion of specific elements of the design procedure and the calculation of the normalized self and mutual capacitance per unit TABLE AI length, rod diameter and center-to-center spacing may be calculated. APPROXIMATE VALUES First, the filter ground plane spacing h must be selected. The selection FOR GROUND PLANE SPACING of a suitable ground plane spacing is bounded by the out-of-band spurious response, passband frequency and insertion loss. Certainly, the Filter Frequency Ground Plane ground plane dimension should be less than one quarter wavelength f0 (GHz) Spacing h (" ) and no smaller than that required to achieve an unloaded Qu consistent with the insertion loss requirements. The maximum ground plane 1 3.000 spacing may be estimated in accordance with 2 1.500 Z0 πλ h ≤ 0 10 138 4 0.750 32 8 0.375 This formula provides the approximate values for ground plane spacing listed in Table A1: 10 0.300 Cristal2 has given the closed-form expressions for the even and odd 12 0.250 mode capacitance per unit length from which the self and mutual capacitance may be calculated from 15 0.215 C 0 (c ) = 18 0.175 4 d 1 ∞ – ln 1 – b + 2 ∑ –1 2 2 c m =1 b ( ) m m πc ln tanh 2 b and C e (c ) = NORMALIZED MUTUAL CAPACITANCE ε πd 1 4 b ln 2π 4 1– d 2b ε 4 d 1 ∞ m πc + ln 1 – b + 2 ∑ ln tanh 2 2 c 2 b m =1 b For reasonable accuracy, the parameter m may be limited to 100. The self and mutual capacitance is calculated using1 C g(c) = C e(c) and C m(c) = 0.25[C 0(c)–C e(c)] The equations have been programmed using Mathcad for a ground plane spacing of 0.275 inches and a rod diameter of 0.128 inches, that is, Zo = 60 Ω. The data are shown in Figures A1 and A2. ▲ Fig. A1 Mutual capacitanve of coupled lines for h = 0.275 and d = 0.128". NORMALIZED CAPACITANCE TO GROUND πd 1 4 b ln 2π 4 1– d 2b 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 250 300 350 400 450 500 CENTER-TO-CENTER SPACING (mil) 7.0 6.5 6.0 5.5 5.0 4.5 4.0 250 300 350 400 450 500 CENTER-TO-CENTER SPACING (mil) ▲ Fig. A2 Normalized self-capacitance to ground for h = 0.275” and d = 0.128”. TUTORIAL APPENDIX B ALTERNATE FILTER IMPLEMENTATIONS In addition to the examples within Part II of this tutorial, there are several alternative filter implementations that lend themselves to the general design procedure and should be cited. These alternative filter implementations utilize resonators not fully exploited in recent literature and for which recent CAD tools have permitted extensive analysis and understanding. Two specific types of resonators are suggested for special applications: cylindrical dielectric resonators and helical resonators. DIELECTRIC RESONATOR DIELECTRIC RESONATOR FILTER Figure B1 shows the electrical and mechanical schematic of a three-section bandpass filter using cylindrical dielectric resonators. Resonators of this type are not typical of conventional lumped or distributed element structures due to the existence of multiple resonant frequencies or modes. This resonator type permits the design of narrowband, low loss filters due to the high unloaded quality factor (Q u > 10,000) in the principal or dominant resonant mode. DIELECTRIC RESONATOR COUPLING SLOT OUTPUT COUPLING LOOP TUNER COUPLING SLOT TUNER TUNER DR2 DR3 DR1 Dielectric resonator filter. TUNER TUNER TUNER TUNER SLOT Helical resonator filter. TUNER SLOT SLOT ▲ Fig. B2 OUTPUT COUPLING LOOP LOW DIELECTRIC SUPPORT H-FIELD LINES OF TE01δMODE ▲ Fig. B1 DIELECTRIC RESONATOR SLOT A note of caution is in order with respect to the multiple resonator modes and the environment necessary to suppress excitation of these modes and thereby eliminate loss spikes and spurious responses in the filter transfer function. There are two techniques suitable for such a determination. The first technique requires experimental characterization of the resonator modes and mode coupling via the methods described in Part I. The second, and until recently more difficult analysis, involves analytical determination of the modes and mode coupling through the utilization of a three-dimensional electromagnetic simulation tool. Some sophisticated filter designers have made use of auxiliary modes to design socalled dual-mode or multi-mode filters that utilize two or more of the resonant modes to obtain very compact, low loss filters with excellent rejection band characteristics. The general design procedure is not suitable for this task. The input, output and inter-resonator coupling is accomplished via the magnetic field which is orthogonal to the circular plane of the cylindrical resonator for the dominant TE 01δ mode. The equivalent circuit of the dielectric resonator filter represents the electrical behavior in the principal mode only. It has acceptable accuracy if the resonator environment and mounting arrangement has demonstrated elimination of the auxiliary modes. To obtain the highest unloaded quality factor available from the cylindrical dielectric resonator, a low loss dielectric support structure is required to separate the resonator from the electrical ground boundary. HELICAL RESONATOR FILTER The helical resonator filter is another filter type for which the general design procedure is applicable. Helical resonator filters are particularly useful within the VHF and UHF bands for low loss, narrowband applications, and where the greater volume of a distributed resonator filter is prohibitive. The helical resonator is at least partially distributed due to the combination of self-inductance and the electrical length of the coil structure. The helical resonator is tuned via the distributed capacitance of the tuning screw at the high electric field point of the resonator. A direct tap at the low electric field point of the resonator accomplishes input and output coupling. The resonator spacing and coupling slot at the high magnetic field point of the helical structure facilitates inter-resonator coupling. Figur e B2 shows the mechanical and electrical schematic of a typical five-section helical resonator filter. Once again, resonator parameters and coupling coefficients may be determined experimentally or analytically by the use of EM simulation tools. Usually, a low loss dielectric form that facilitates uniformity, placement and structure supports the helical resonator.