Parametric Study of Insertion Losses and Antenna Pattern ...Microstrip Patch Antenna parameters. In this paper we present a comparative study of return loss, insertion loss of the

International Journal of Electrical Electronics & Computer Science Engineering

Volume 5, Issue 3 (June, 2018) | E-ISSN : 2348-2273 | P-ISSN : 2454-1222

Available Online at www.ijeecse.com

26

Parametric Study of Insertion Losses and Antenna Pattern in Cascaded

Microstrip Coupled-Line Filter Sections for Bandpass Filter Responsein

Microwave Band G

Abdul M. Syed1, Kanti Prasad2 1Ph.D Scholar, 2Professor, Dept. of Electrical & Computer Engineering, University of Massachusetts Lowell, MA, USA

[email protected], [email protected]

Abstract: Bandpass Filters (BPF) designed using cascaded

(parallel-coupled) Microstrip Coupled-Line Filter (MCFIL)

sectionsand Microstrip Lines (MLIN) find applications in

the Frontend modules of MMIC communication chipsets.

Specifically BPF’s based on MCFIL design in Band G (4-6

GHz) of Microwave regime and current Long Term

Evolution (LTE) Band 46 emphasizes the selection of

substrates based on trade-off between MMIC transmission

medium loss metrics along with compromisations on

Microstrip Patch Antenna parameters. In this paper we

present a comparative study of return loss, insertion loss of

the comprehensive BPF,realized with most common

dielectric substrates at a operating frequency of 4.89 GHz.

The characteristic impedance of shielded MLIN achieved

using a computer program for FEM based Quasi-Static

analysis for ten different substrates is given. BPF design

incorporating Finite Element Method (FEM) based

Computational Electromagneticanalysis of shielded MLIN

and Chebyshev Approximation is discussed. Using

Keysight’s EM ADS simulator, the Full-Wave results of

return and insertion losses for BPF are presented. Our

theoreticalEM post-processed Antenna radiation pattern

results for the designed BPF are in good agreement with

numerical design models of Microstrip Patch Antennas

available in literature meeting narrowband, wide-beam, and

RF power steering requirements. Full-Wave analysis

presented here provides key parametric details in Wireless

Communications for the design of third order Chebyshev

Type-I distributed model BPF response using cascaded

MCFIL’s, MLIN for resolving Bandwidth (BW) limitations.

Keywords: Monolithic Microwave Integrated Circuits

(MMIC),Shielded MicrostripLine, Characteristic Impedance,

Bandpass Filter, Electromagnetics (EM),Type I Chebyshev

Filter, Radio Frequency (RF), Quasi-Static and Full-Wave

Analysis.

I. INTRODUCTION

BPF designs based on Microstrip technology is a very

popular topic in academia and is mature in industry as

well. Among those in literature search, dating back to a

couple of decades is tunable BPF in time domain [1].

The method in [1], demonstrates tuning the time

domain response of coupled–resonator filter’s return

loss such that the center frequency of each resonator is

precisely tuned using Discrete Inverse Fourier Transform (DIFT) of frequency response. However in

the last decade a tremendous amount of BPF in

Microwave bands are investigated by researchers, some

of which are listed in [2]-[12]. Miniaturization aspects

of BPF are thoroughly discussed and successfully

demonstrated for widely used Chebyshev response in

Microwave bands using the Microstrip technology such

as 4th order Ku-band BPF [2], and 4th order W-band

BPF [3], along with 5thand 7th order X-band hairpin

type BPF [4].Microstrip BPF with double octagonal

shaped resonators for S-band applications is presented

in [5], while fabricated Bell shaped resonators with

dual band BPF which are network matched using the

Butterfly radial stubs are experimentally verified in [6].

Novel Wi-Fi BPF with performance characteristics

improvements based on Modified Complementary Split

Ring Resonators (MCSRR’s) and using negative

permittivity behavior as in Metamaterials (MTM) is proven in [7]. Another such application of MTM in

wireless communications, we have recently discussed

in [20].Design of Microstrip-BPF of 3rd order for

WLAN applications is shown in [8] with parametric

data, insertion and return losses. Parallel-coupled lines

BPF using five soft substrates for 5th order Chebyshev

type filtering is presented in [9]. But implementations

with the substrates used in [9], have been known to

provide low APHC features highlighted in [14],

rendering them unsuitable for reliable communications

such as todays smartphones. Also designs in [10] with

optimization details for 5th order dual band BPF and [12] with 7th order BPF for band 28 are shown, but both of

them lack the comparative performance analysis with

other commonly used substrates in industry such as

Alumina for Microstrip technology or GaAs in MMIC

based designs. Moreover [11] emphasized lumped

element prototype model to give the 5th order hybrid

BPF response with Microstrip technology. Even though

[8]-[12] gave complete design and characterization

results for effective BPF responses, it did not take into

account the dispersion effects for characteristic

impedance(Z0m)which is critical at high frequencies for filter operation that causes shift in center frequency nor

do they address the antenna parameters that can help

one to better understand the channel utilization schemes

for shared antenna or the effective link power budget

analysis problems. We present here results of Z0m using

FEM based analysis recently given by us in [13] along

with design and loss characterization of a 3rd order

ChebyshevType-I distributed model BPF based on

MCFIL’s, MLIN’sfollowed by their antenna parameters using ADS [18].

II. CASCADED MICROSTRIP COUPLED-LINE

FILTER SECTIONS

A single shielded MLIN is shown in Fig. 1.(a). The

dimensions of shielded MLIN in Fig. 1. (a), are selected

http://www.ijeecse.com/

mailto:[email protected]

mailto:2kanti_prasad@uml.




27

as a = 5 mm, b = 3 mm, conductor strip width or line

width w = 1.666 mm (2X0.833 mm), substrate height h

= 25 mils, the thickness of conductor strip (copper foil)

t = 0.15 mm. These specifications meet the design

conditions of a wide strip line having w

h≥ 1 and

satisfies Wheeler’s incremental inductance rule (t > 4δ,

δ being the skin depth) [13]-[14]. Fig. 2.(a) shows a

SymmetricalMicrostrip Coupled Line (MCLIN) having

4 ports.An MCFIL section with port voltages and port

currents as shown in Fig. 3. (a), is formed by making

the ports 2 and 4 of Fig. 2. (a) open, due to ease of

fabricating an open circuit than creating a short circuit

in microstrip technology [16]. Currents I1 (I1 = i1 + i2)

and I3(I3 = i3 – i4) are the port currents with i1, i3 being their respective current sources driving the transmission

line in even mode. Open circuit currents at ports 2 and

4 are both 0 such that the current sources i2, i4 drive the

line in odd mode. V1 and V3 are the port voltages given in terms ofImpedance (image) parameters Z, in [16] as

V1 = Z11I1 + Z13I3 (1)

V3 = Z31I1 + Z33I3 (2)

(a) (b) (c)

Fig. 1. (a). Microstrip Line (MLIN) in an Shielded

Enclosure, (b). “Linecalc” Utility Depict of MLIN in ADS, and (c). MLIN Symbol View in ADS [13], [18]

(a) (b) (c)

Fig. 2. (a). Microstrip Coupled Lines (MCLIN) in an

Shielded Enclosure, (b). “LineCalc” Utility Depict of

MCLIN in ADS, and (c). MCLIN Symbol View in ADS [13], [18]

Admittance inverters constructed using Quarter-wave

transformers are used for transformation of shunt

connected elements into series element connection and

vice versa [16]. Fig. 3 (c) show the equivalent circuit of

MCFIL using Admittance Inverter. By considering

excitations in even and odd modes and applying the

superposition theorem, the open circuit Z matrix is derived in [16].

(a) (b)

(c)

Fig. 3. (a). Individual MicrostripCoupled-Line Filter

(MCFIL) Section Unit, (b). MCFIL Symbol View in

ADS [18], and (c). Equivalent Circuit of MCFIL using the Admittance Inverter [16]

Coupling factor C and the equations for calculating

MCFIL line impedances for even (Z0e), and odd (Z0o)

modes along with the BPF design equations are given

in [14], [16]-[17]. Based on our design specifications

for a BPF in Microwave band G (4-6 GHz) [15], we

have employed the 3rd order Chebyshev polynomial

approximation, and we achieved n–1 = 2 peaks within

the passband as shown in Fig. 12. With an order of

filter n = 3 and number of stages (sections) = n+1 = 4

so that in Fig. 4, the Impedance-Admittance Interval

terms for 2nd, 3rd stages Z0Ji as well as for 1st, 4th stages

Z0Jj+1 are dictated by Z0Ji = π ∆

2 √gi−1 gi , and Z0Jj+1 =

√π ∆

2gj gj+1 respectively.

where i = 2, 3 and j = 0, 3, Δ = 0.1 is the fractional

Bandwidth, Ripple factor = 0.1 dB and gn are the Low-

pass (LP) prototype values interpreted using gi-1, gi, gj,

gj+1, the normalized distributed elements which are

being obtained using Chebyshev table given in [21]-[22].Prime design equations are

C (dB) = (Z0e−Z0o

Z0e+Z0o) (3)

Z(0e)k+1 = Z0 .[1 + Z0Jk+1 + (Z0Jk+1)2] (4)

Z(0o)k+1 = Z0 .[1 – Z0Jk+1 + (Z0Jk+1)2] (5)

where Inverter constant k = 0 to n, Z0 = 50Ω is the

characteristic impedance and Admittance Inverter

constants Jk+1 are Filter coefficients (for k+1

Admittance Inverters), being determined from the LP

prototype values. For Π or T network, the prototype

values from [21] are Rs = g0 = 1, RL = g4 = 1, g1 = 1.433,

g2 = 1.594, g3 = 1.433. The MCFIL even and odd mode

characteristic impedances are calculated using equations (4), (5) respectively and are summarized in Table III.

(b)





28

Fig. 4. Structure of Cascaded MCFIL Sections for BPF

Response

III. MATERIALS AND METHODS

(SIMULATIONS IN ADS)

A circuit schematic view of BPF (Fig. 4) is created as

shown in Fig. 5 along with an generated layout (Fig. 6),

which is then simulated using EM/Momentum

(“MomUW”) simulations in microwave mode (Full-

Wave analysis), with perfect boundary approximations

in ADS [18]. For simulation purposes, each of the

dielectric substrates from Table I are selected with their respective material properties. The parameter entry

mode (“String and Reference”)[18] for TLines-

Microstripsubcircuit and Simulation-S_Parameters

instance in ADS palette is opened and the values as

depicted in Fig. 5 for the case of Alumina substrate are

entered such that H, Hu (cover height), and T

corresponds to h, b, and t respectively as given in Fig.

1. (a)for MLIN’s while Hu is set at 3.9e+034 mils for

MCFIL’s. Also for Alumina substrate, the Line length

l, line width w values from Table IV and Kirschninget

al. dispersion model in [14] are assigned to Quarter-wavelength MLIN’s – TL1 and TL2 in BPF circuit of

Fig. 5. Frequency dependent Svensson/Djordjevic

dielectric loss model with linear frequency sweep for

2001 points is selected in ADS. A layout as shown in

Fig. 6 with appropriate pins for ports is generated by

synchronizaton of entire BPF design of Fig. 5. Using

ADS EM simulator feature in layout, emModel setup is

edited to initialize Momentum (“MomUW”)

simulations in microwave mode [18]. A FEM

symmetry plane is added so as to incorporate Z0mvalues

from Table IIof shielded MLIN with loaded dielectrics achieved by FEManalysis given in [13].

Fig. 5. Schematic View of BPF of Fig. 4 with Alumina

Substrate, MCFIL’s, MLIN’s, and Input-Output Port Terminations in ADS [18]

Global EM simulation option with adaptive frequency

plan having 2400 Node points (max), linear

extrapolation mode is choosen. Edge mesh having a

mesh density of 20 cells per wavelength with “Iterative

Dense” [18] matrix sover method, port-solvers are

enabled in the simulator. S-Parameter “TML” [18] port

calibration with 50 Ω reference impedance for P1 and P2 in Fig. 5 is mapped using the port editor and

fulfilling the partitioning rule by descending into layout

indeed by engulfing CLin’s encounteredin process, as casted by Fig. 6.

Fig. 6. ADS Layout, Generated from Schematic View of BPF Shown in Fig. 5

Fig. 7. “MSUB” module of Fig. 5 for Alumina

Substrate with Mapped Conductor, Cover and Layers in ADS [18]

Alumina substrate is then created as shown in Fig. 7 by

selecting the technology drop-down menu and editing

the desired material properties using material

definitions submenu. Cover is defined with perfect

conductor material, while the interface is a strip plane

with mapped layers. Conductor layer is sheet of copper with orientation angle 90° above interface and substrate

layer is defined such that bounding area layers are

inherited from the designed substrate. Copper foil of

thickness 0.15 mm (4.28 ounces) having electrical

conductivity of 5.96X107 S/m is taken as the strip

conductor. A small amount of loss tangent (tan δ) is

included in obtaining practical simulations to take into

account dissipation of each of the MLIN’s and

MCFIL’s distributed elements, as per an equivalent

microstrip Quarter-wavelength input-output matching

transformer. Port coefficient is k = 7.23, for length L

and width w of MLIN Quarter-wave matching sections. These act as a planar microstrip resonators having

source, load impedances of 50 Ω each for input- output

sections, with loaded stripline (Perfect Electric

Conductor-PEC, having metallization thickness t = 0.15 mm) needed for excitations.





29

Fig. 8. Momentum “MomUW” 3D Isometric view with

Conductor, and Substrate (Fig. 7) for (a).MLIN (Fig.

1), and (b). BPF (Fig. 5) using emModel’s Wireframe Mode in ADS [18].

BPF parametric details combined with defined

conductor, substrate and port characteristics in

emModel setup is viewed using ADS-3D view as

shown in Fig. 8. A tuning scheme shown by t in Fig.

5, for the Line length L is employed along with design

verification and design for testability including the

verify DRC, netlists LVS matching condition before we started ADS simulations.

IV. PARAMETRIC DATA

Table I. Dielectric Substrate Properties

Substrate Material εr (tan δ) Dielectric Loss

Tangent at 10 GHz

GaAs 12.3 0.0016

Si 11.7 0.0050

RO 3010 11.2 0.0022

RT Duroid 6010 10.5 0.0015

3M EPSILAM 10/

Keene DI-Clad

810

10.2 0.0015

Alumina 9.7 0.0002

FR-4/TMM4 4.3 0.0180

RO 4003C 3.4 0.0020

RT Duroid 5870 2.35 0.0012

RT Duroid 5880 2.20 0.0009

Table II. Quasi-Static Values using FEM based

Programand Computational Electromagnetics Method Shown in [13]

εr εre (0) εre (f) Line Length

L (mm) Z0m (Ω)

Z0m (f) (Ω)

12.3 8.820 9.227 5.16 47.68 49.04

11.7 8.396 8.769 5.29 48.79 50.14

11.2 8.049 8.393 5.40 49.79 51.13

10.5 7.570 7.876 5.57 51.29 52.62

10.2 7.359 7.649 5.65 51.97 53.30

9.7 7.013 7.277 5.79 53.18 54.49

4.3 3.281 3.332 8.46 75.73 76.82

3.4 2.658 2.688 9.40 83.17 84.20

2.35 1.932 1.943 11.03 95.37 96.22

2.20 1.824 1.833 11.35 97.59 98.41

For effective transmission (S21) and minimal reflection (S11) characteristics, the optimized Line lengths

achieved for practical design tolerant thicknesses are

given in Tables III and IV for the case of BPF designed

with various substrates.In Tables I and II εr, εre are

relative permittivity and the effective relative

permittivity respectively.The complete design

equations for Z0m, εre, effective strip width effw and

Dispersion effects governed by εre (f), Z0m (f) along

with theoretically calculated Quality factor Q of the

Microstrip Lines given by equation (6) that dictates loss

metrics and Figure of Merit (FOM), were givenrecently by us in [13].

1

Q=

1

Qc+

1

Qd+

1

Qr=

1

Q0+

1

Qr=

1

QT (6)

where Q𝑐 , Qd, Qr account for the conductor, dielectric,

and radiation losses respectively due to discontinuities

on MLIN, by surface-wave propagation. In equation (6),

Q0 is the circuit Q (Loaded Q) while Q or QT is total Q

(Unloaded Q). Fig. 9 compares εre (f) and Zom (f) values

using FEM based Quasi-Static Analysis [13] at f = 4.89

GHz for the MLIN’s designed for various εr as given in Table II. Parametric data for length, width and spacing

in Tables III and IV are populated using equations (3)-

(5) along with the aid of ADS “LineCalc” utility [18], parameters being highlighted in Fig. 1, Fig. 2 and Fig.3.

Table III. Dimensions for 1, 4 and 2, 3 Symmetrical

MCFIL Sections of Fig. 4 by Applying Chebyshev

Approximations, and using Distributed BPF Model Along with ADS “LineCalc” Utility [18]

εr

Z0e1, Z0e4 = 72.03486 Ω,

and

Z0o1, Z0o4 = 38.92666 Ω

Z0e2, Z0e3 = 55.73669 Ω,

and

Z0o2, Z0o3 = 45.34349 Ω

w1, w4 (mm)

l1, l4 (mm) w2, w3 (mm)

l2, l3 (mm)

12.3 0.218937 5.368460 0.324378 5.096810

11.7 0.238311 5.474140 0.348391 5.198410

11.2 0.257015 5.574370 0.371475 5.294760

10.5 0.285101 5.722120 0.406001 5.436770

10.2 0.298075 5.789420 0.421909 5.501450

9.7 0.321475 5.909470 0.450552 5.616860

4.3 0.793637 8.143610 1.030850 7.792080

3.4 0.973768 8.940350 1.256300 8.590060

2.35 1.285670 10.26470 1.651470 9.951170

2.20 1.346530 10.51540 1.729160 10.21400

Table IV. Coupled Lines Spacing for Symmetrical (1,4

and 2,3) MCFIL Sections of Fig. 4 and Single MLIN Dimensions of Fig. 1, using ADS “LineCalc” Utility

[18], and Table III

εr

MCFIL

MLIN

Spacing S1, S4

(mm), with Spacing S2,

S3 (mm), with





30

Coupling C (dB) = -10.504737

Coupling C (dB) = -

19.758334 W (mm)

L (mm)

12.3 0.430720 1.194190 0.392782 5.715910

11.7 0.424409 1.184800 0.417479 5.825860

11.2 0.418549 1.176290 0.441151 5.930140

10.5 0.410138 1.164430 0.476454 6.083860

10.2 0.406398 1.159280 0.492683 6.153880

9.7 0.399865 1.150470 0.521857 6.278810

4.3 0.307361 1.043690 1.107470 8.635010

3.4 0.285318 1.024670 1.334230 9.502570

2.35 0.258244 1.011790 1.731770 10.98890

2.20 0.254259 1.011720 1.809930 11.27700

Fig. 9. εre (f), Zom (f) using Quasi-Static Analysis of MLIN

V. MICROSTRIP – BPF: RESULTS AND LOSS

CHARACTERIZATION

Chebyshev filter is a high Q filter where ripples in

passband with non-flat passband response are being

allowed [21]. We employed 3rd order Chebyshev Type-

I distributed model in Microwave band G [15] BPF,

whose design specifications fit applications in “U-NII-1” for LTEBand 46-with Full Duplex links, having a

Frequency Division Link (FDL) for uplink (FDL Low)

of 5.125 GHz, and downlink (FDL High) of 5.925 GHz

[23]. The operating or center frequency fc = 4.89 GHz

for θ = π/2 in Fig. 3 (c), BW (3 dB) = 0.5 GHz, and the

BW (30 dB) = 1.5 GHz with higher and lower band

edges as fH = 5.15 GHz, fL = 4.65 GHz selected for

steeper initial descent into passband followed by

rollback in the passband with non-flatband

characteristics. Table V gives individual ADS

simulation results for an BPF with optimized MLIN Line length l (Table IV) and tuned MCFIL line lengths

(Table III), while Fig. 11and Fig. 12 shows the

combined ADS S-parameter results for all substrates of

Table I. Apart from losses Q𝑐 , Qd , Qr , the Average

Power Handling Capacity (APHC) of MLIN, MCFIL is

an important design factor dictated by thermal

conductivity of substrate and temperature dependency

of strip conductor [14]. From the results given in

Tables III, IV, and V, we can infer that a trade-off

exists between S11, Line length and Insertion Loss (IL)

values. However in case of Alumina, the Line lengths

given in Tables III and IV, achieved using “LineCalc”

utility of ADS [18] providedoptimized values for S11, and IL as given in Table V, while conceiving better

APHC feature to handle about 5 kW of Continuous Wave (CW) power at 10 GHz [14].

Table V. Cascaded MCFIL’s (Fig. 5) Results, using

EM-Momentum ADS Simulations (Full-wave analysis) for BPF response

Substrate Material εr S11

(dB)

IL

(dB)

GaAs 12.3 -9.284 -4.589

Si 11.7 -9.359 -3.087

RO 3010 11.2 -7.765 -5.137

RT Duroid 6010 10.5 -6.804 -5.066

3M EPSILAM 10/

Keene DI-Clad 810 10.2 -6.391 -5.229

Alumina 9.7 -5.333 -5.099

FR-4/TMM4 4.3 -3.061 -15.594

RO 4003C 3.4 -1.085 -15.433

RT Duroid 5870 2.35 -0.937 -16.188

RT Duroid 5880 2.20 -0.839 -16.874

Fig. 10.Group Delay ‘g’ (pico sec) in S-domain using

Full-Wave Simulations of Cascaded MCFIL’s, MLIN’s Schematic, as Shown in Fig. 5 for Alumina

Fig. 11. Return Loss (S11) using Full-Wave Simulations

of Cascaded MCFIL’s with MLIN’s Schematic as shown in Fig. 5

0

20

40

60

80

100

0

5

10

15

20

25

30

12.3 11.7 11.2 10.5 10.2 9.7 4.3 3.4 2.35 2.2

Z0

m (

f) Ω

ε re

(f)

εr

Epsilon (εr) vs. εre (f), Z0m (f) (Ω)

εre (f)Z0m (f) …

f = 4.89 GHz





31

Fig. 12. Insertion Loss (S21) using Full-Wave

Simulations of Cascaded MCFIL’swith MLIN’s Schematic as shown in Fig. 5.

Group delay ‘g’ or delay (1,2)giving a measure of

relative values of delay (time delay of amplitudes),

from the input to output ports at various frequencies in

the momentum simulation quantifies the “Full-Wave” –

Frequency domain loss characterization in ADS [14],

[17]. The group delay ‘g’ of BPF is approximately

proportional to filter order and inversely proportional to

BWwhile ILat filters band edges (fH,fL) is equal to IL at band center (fc) times the ratio of group delays of band

edges to the band center [17], [24]. Our achieved group

delay is about 38 ps as shown with marker ‘g’ in Fig.

10, with 0.7 ns rise time of a pulsed signal which is in

agreement with 40 ps value of ‘g’, seen commonly in

most of the Microstrip lines [14]. Parametric results

encapsulated into momentum visualizations for the case

study of Alumina substrate based BPF, are depicted by

Fig. 13, and Fig. 14. Also the Effective Isotropic

Radiated Power (EIRP) which is signal output power

due to its concentration into a confined area by antenna parameters is given in [14].

EIRP = [PT – Lc + Ga] (7)

where PT = Output Transmitter power in dBm,

Lc = Cable loss in dB and Ga = Antenna Gain in dBi.

Neglecting Lc for Microstrip Lines case, the antenna

parameters including EIRP from the post-processed

results of ADS emModel for various substrates are shown in Table VI.

Table VI. Antenna Parameters Measured using ADS

EM-Momentum Setup of Isotropic ‘i’ (point source) Input 0.9-1.6 mw

εr Gain

(dBi)

Directivity

(dBi)

Radiation

Efficiency (%)

EIRP

(μw)

12.3 -1.64 7.231 12.97 115.318

11.7 -2.53 6.698 11.94 105.220

11.2 -0.99 6.586 17.44 184.289

10.5 -0.15 6.504 21.60 251.246

10.2 0.036 6.471 22.72 270.146

9.7 0.34 6.457 24.44 302.761

4.3 -1.02 6.862 16.27 152.159

3.4 4.02 6.928 51.13 200.770

2.35 6.05 7.917 65.00 2740.31

2.20 6.41 8.023 69.03 2615.17

Fig. 13. Post-Processed E Field - Momentum

Visualizations of MLIN (Port 1 to Port 2 Vision:Upper

Image) and BPF based on MCFIL’s, MLIN’s(Port 1 Vision: Lower Image) Realised with Alumina Substrate

Fig. 14. Post-Processed E Field - Momentum

Visualizations of MLIN (Port 2 to Port 1 Vision: Upper

Image) and BPF based on MCFIL’s, MLIN’s (Port 2 Vision: Lower Image) Realised with Alumina Substrate

VI. QUANTIFYING ENGG. DESIGN AND

DISCUSSION OF RESULTS

To meet MMIC Yield and Reliability requirements of

the designed BPF for fabrication purposes, we adopted

an statistical approach where Poisson’s (Delta), Triangular and Rectangular distribution functions are

choosen for Yield Analysis while Log-Normal

distribution and Cumulative Distribution Functions

(CDF) are used for Failure Analysis.For a single wafer

from a virtual batch production, we set out the selection

criteria to contain 400 chip sites with 200 good chips,

whose examination shall show us that no good chips

can be found over approximately 25% of wafer area

corresponding to a Y0 of atleast 75%, using notation

that the fraction of chip sites that yield bad chips is (1-

Y0) and Yield (Y) is probability that a chip has no defects atall. From yield of good chips the estimated





32

D0A (Average number of defects per chip having an

area A), using the Matlab’s Symbolic Math Toolbox is

found to be approximately 0.405, 0.420, 0.437 for

Poisson, Triangular and Rectangular distribution

functions respectively. We continued the statistical analysis by using a data-fitting technique for 10 MMIC

devices comprising of MLIN, MCLIN’s in cascade that

provides BPF response of Fig. 5. Then using the

Matlab’s Statistics and Machine Learning tool box

(SML) [19], with a model test program we performed

accelerated aging at elevated temperatures such that a

different device fails at each of the 6 test times namely

300, 700, 1600, 2600, 3500, and 4200 hours. Device

Under Test (DUT) is tested for electrical characteristics

every 100 hours with SML. After 5000 hours, 4 out of

10 devices were still operating, when the DUT is removed from testing inside SML’s created

functionality test program. A scale parameter value (σ)

or standard deviation of Log-Normal distribution of

1.88 was achieved. Also with a median life of 8.8X105

hours (≈ 100 years), the expected fraction of failed

systems in 10 years period, with a CDF or Failure

function of 10 FIT (1 FIT = 1 Failure/109 Device-hour)

was calculated to be 0.2367, having a Reliability of

76.33 %. To improve the device Reliability, we then

placed a group of 20 devices and tested after every 10

hours of aging. The devices failed at times 30, 40, 60, 85 (2 failures), 100, 150 (2 failures) and 290 hours,

where the testing of DUT was discontinued at this time.

In this instance σ = 1.6 and the expected fraction of

failed systems in 10 years period is calculated to be

0.1780. A Reliability of 82% is thus projected for our

MLIN, MCFIL based MMIC BPF devices which fits ±

3σ, the Design for Manufacturing (DFM) criterion in

microelectronics fabrication.To discuss and interpret

the Full-Wave results presented, we start by comparing

the popular Microstrip Line filters such as Hairpin,

Parallel-coupled, Stepped and Stub impedances, along

with BPF implementations using Capacitively Coupled Series or Shunt Resonators. Filters using multisection

Coupled resonators concepts such as those presented

here with cascaded MCFIL’s, MLIN’s are dominant for

narrowband response for BW less than 20 % due to

developments in microelectronics fabrication processes

[16]. Further we can see that Far-field(regions ≥ 2d/λ

farther from antenna, Fig. 18) where ‘d’ is the diameter,

has radius ‘r’that is not more than 3 times the line width

used (Fig. 17), otherwise the transmission line becomes

almost imperceptible in its Z characteristics w.r.t a

straight line conductor section [17], [24]. On top of placing the shielded enclosure shown in Fig. 1 (a), the

metal losses in Alumina substrate based microstrips can

be reduced in half by doubling its substrate thickness

since metal loss is proportional to square root of

frequency while dielectric loss is proportional to frequency only [14]-[15], [24].

Fig. 15. Antenna Gain, Directivity Pattern of BPF from

the Simulation Post-Results of Alumina Substrate using ADS emModel Setup in dBi Scale

Important antenna parameters such as Gain (G),

Directivity (D), Effective Area, Radiation Intensity,

Percentage Radiation Efficiency (ER), Far Field pattern

and EIRP were obtained from the ADS post-processed

EM results, which are shown in Table VI and Fig. 15 to

Fig. 18. We considered a lossless antenna which by hypothesis has an efficiency of 100 % (0 dB) named as

isotropic ‘i’ antenna, having equally uniform radiations

in all of its beam (Half Power Beamwidth-HPBW) and

sidelobes. For antenna patterns an infinitesimal area

‘dA’ is selected on a sphere of radius ‘r’ (having solid

angle with 4π Steradians subtended) which is given in [24] by

dA = r2.Sin θ dθdφ (8)

where θ (Theta) = Plane angle in radians = Arc length l

Circle radius r

φ (Phi) = Azimuth angle, and Elevation = (90-θ)°

Radiation pattern is given in spherical coordinates by E

(r,θ,φ) at each of the locus points on sphere’s surface,

which is depicted by Fig. 16. As radial E-field

component Er is close to zero for antennas in Far Field

region of Fig. 18, we have the amplitude pattern given by only the E (θ,φ) function in [24] as

|E|2 = |Er|2 + |Eθ|2 + |Eφ|2 ≈ |Eθ|2 + |Eφ|2 (9)

For an angle θmax.= 45°, magnitude of E_max. (θ) =

0.2723 V and the magnitude of E_max. (φ) = 0.01043

V for φmax. = 357°, so that the magnitude of E-Field is found to be |E| = 0.2725 V.





33

Fig. 16. Radiation Intensity Pattern of BPF from the

Simulation Post Results of Alumina Substrate using ADS emModel Setup (Magnitude in W/Steradians)

Fig. 17. Antenna Effective Area (m2) for BPF from the

Simulation Post-Results of Alumina Substrate using ADS emModel Setup

Antenna Radiation Efficiency (ER), Total Efficiency of

Antenna (ET), and the relationship between Antenna

Gain (G), Directivity are given in [24] by equations (10) to (12).

ER = Power Radiated

Input Power =

PR

PI (10)

ET = ER. ML (11)

G = ER. D (12)

where ML = Mismatch losses due to impedance

mismatching when an antenna is connected to transmission line or receiver.

Radiation pattern (or Antenna Gain pattern) that

describes antenna gain in dB as a function of angle is

shown in Fig. 16. Smartphones with built-in Wi-Fi

antennas have efficiencies ranging between 20 % (-7 dB) to 70 % (-1.5 dB) [24] as evident in Table

VI.However smartphones that communicate with point-

to-point networks using the built-in cellular antennas

irrespective of their orientation and relative positions

prefer lower gain G [24]. From Table VI, we can see

that Alumina substrate still remains the viable choice

for Microstrip line designs as it provides the EIRP of ~

303μw which is within the FCC [23]regulated value as

well as a better Directivity of 6.4 dBi and Gain of 0.34

dBi, suitable for BPF for Frontend modules in personal

wireless applications like smartphones and in advanced RF switches, Wi-Fi devices.

Fig. 18. Far Field Computation of BPF from the

Simulation Post-Results of Alumina Substrate using ADS emModel Setup

VII. CONCLUSIONS

We presented a detailed parametric study of BPF

transmission (S21) and reflection (S11) characteristicsin

ADS by applying the concepts of Advanced RF,

Microwaves, and Applied Electromagnetics

applications. Design methodology and numerical data

presented will provide efficient RF/Microwave device

BPF response visualizations, along with planning MMIC Multi-chip module (MCM) transceivers

designbased on MLIN, MCFIL. The comparative

antenna pattern given will help in selection of substrate

for the case of Microstrip Patch Antennas that are

commonly integrated in Frontend modules of consumer

wireless applications such as smartphones or for

System-on package (SOP) implementations. Bandwidth

enhancements together with compact shared antennas

that features combining multiple carriers a.k.a “Carrier

Aggregation” in Wireless world can be addressed with

our integrated BPF structure and incorporating custom

Microstrip Patch Antennas using the same ADS design platform. EIRP values achieved from the post-

processed simulation results are within the maximum

allowed FCC regulated valueof 4 watts w.r.t an

integrated antenna having fixed EIRP for point-to-

multipoint communications in “Unlicensed National

Information Infrastructure” (U-NII-1) band [23], having

frequency range of 5.15-5.25 GHz. Finally using the

MLIN, MCFIL’s realized with Alumina substrate at a

operating frequency of 4.89 GHz, we achieved a small

improvement in antenna radiation efficiency of ~ 24 %

(compared with other dielectrics in our study excluding high impedance lines (8), (9), and (10) of Table I with low APHC).

VIII. REFERENCES

[1] J. Dunsmore, "Tuning band pass filters in the time

domain," 1999 IEEE MTT-S International

Microwave Symposium Digest (Cat.

No.99CH36282), Anaheim, CA, USA, 1999, pp. 1351-1354 vol.3.

[2] M. F. Islam, M. A. M. Ali and B. Y. Majlis,

"Miniaturized Bandpass Filter for Ku-band

Applications," 2007 5th Student Conference on

Research and Development, Selangor, Malaysia, 2007, pp. 1-4.

[3] L. B. Lok, C. J. Hwang, I. G. Thayne and K.

Elgaid, "Demonstration of a W-band microstrip

parallel coupled-line bandpass filter in GaAs

technology," IET Seminar on Passive RF and

Microwave Components, Birmingham, 2010, pp. 37-51.





34

[4] H. Xu and W. Sheng, "The X-band microstrip

filter design," 2017 7th IEEE International

Symposium on Microwave, Antenna, Propagation,

and EMC Technologies (MAPE), Xi'an, 2017, pp. 351-355.

[5] T. Praludi, Y. Sulaeman, Y. Taryana and B. E.

Sukoco, "Bandpass filter microstrip using

octagonal shape for S-band radar," 2017

International Conference on Radar, Antenna,

Microwave, Electronics, and Telecommunications (ICRAMET), Jakarta, 2017, pp. 145-148.

[6] A. Ahmad and A. R. Othman, "Narrow dual

bandpass filter using microstrip coupled line with

bell shaped resonator," 2016 International

Conference on Advances in Electrical, Electronic

and Systems Engineering (ICAEES), Putrajaya, 2016, pp. 550-554.

[7] J. Luo, Y. Yang and H. Zhang, "Design of Wi-Fi

Band-Pass Filter Based on HMSIW,"9thInt’l.

Conference on Intelligent Human-Machine

Systems and Cybernetics(IHMSC), Hangzhou, 2017, pp. 327-330.

[8] A. Gunjal, S. R. Gagare and R. P. Labade,

"Bandwidth enhancement in band pass filter (BPF)

using microstrip couple lines for WLAN (2.4GHZ)

applications," 2016 International Conference on

Electrical, Electronics, and Optimization Techniques (ICEEOT), Chennai, 2016, pp. 3629-3632.

[9] N. N. Al-Areqi, N. Seman and T. A. Rahman,

"Parallel-coupled line bandpass filter design using

different substrates for fifth generation wireless

communication applications," 2015

InternationalSymposium on Antennas and Propagation (ISAP), Hobart, TAS, 2015, pp. 1-4.

[10] M. B. Priya, K. Ramprakash, P. Pavithra and D. A.

Joe, "A cost effective dual band chebyshev

parallel coupled microstrip line bandpass filter for

ISM band applications," 2017 International Conference on Wireless Communications, Signal

Processing and Networking (WiSPNET), Chennai, 2017, pp. 623-626.

[11] M. RincyJuliet, R. Vallikannu and B. C. Kavitha,

"A novel design of hybrid microwave bandpass

filter using lumped element and microstrip line

technology," 2017 International Conference on

Intelligent Computing and Control (I2C2), Coimbatore, India, 2017, pp. 1-4.

[12] D. R. Basavaraju, H. V. Kumaraswamy and M.

Kothari, "Design and simulation of microstrip narrow band pass filter for Asian Pacific

Telecommunication band 28," 2017 2ndIEEE

International Conference on Recent Trends in

Electronics, Information & Communication

Technology (RTEICT), Bangalore, 2017, pp. 1313-1316.

[13] Abdul M. Syed and Kanti Prasad, “Full-Wave

Analysis of Shielded Microstrip Lines on

Dielectric Substrates using the Characteristic Impedance Computations with Loaded Dielectrics

by Finite Element Method,” International Journal

of Current Trends in Engineering & Technology, vol. 4, issue 2, pp. 143-151(March-April, 2018).

[14] K. C. Gupta, Ramesh Garg, Inder Bahl, and

Prakash Bhartia, “Microstrip Lines and Slotlines,” 2nd ed. Artech House, Norwood-MA, 1996.

[15] Sze, Simon M., “Physics of Semiconductor

Devices,” 2nd edition, John Wiley & Sons Ltd, 1981.

[16] David M. Pozar, “Microwave Engineering, Fourth

Edition,” John Wiley and Sons, Inc. 2012.

[17] Keysight Technologies, “Microwave Discrete and

Microstrip Filter Design,” Manuals for Advanced

Design System, ADS2016.07 Documentation Set,

EEs of EDA Division, June 2016, Santa Rosa, California, www.keysight.com.

[18] Keysight’s Premier High-Frequency and High

Speed Design Platform, “RF and Microwave

Design Suite,” Advanced Design System (ADS)

version 2016.01, Keysight Inc, Englewood, CO, 1985-2016.

[19] The Mathworks, “Matlab,” UMass. Lowell

student edition,licensed version no. 7.10.0.499 64-

bit (R2010a), Toolboxes/RF Tools, Natick, MA, 1984-2010.

[20] Abdul Syed and Kanti Prasad, “Intrinsic

Properties of Metamaterial Slab in Wireless

Power Transfer for Efficiency Improvement of

Wireless Communication System,” International

Journal of Current Engineering and Technology, vol. 8, no. 1, pp. 109-112(Jan/Feb 2018).

[21] Christopher Bowick, John Blyler, and Cheryl

Ajluni, “RF Circuit Design, Second Edition,” Elsevier Newnes, Inc. 2008.

[22] G. L. Matthaei, L. Young, and E. M. T. Jones,

“Microwave Filters, Impedance-Matching

Networks, and Coupling Structures,” Artech House, Dedham, Massachusetts, 1980.

[23] The United States of America – Federal

Communications Commission (FCC), www.fcc.gov.

[24] The World’s Microwave information resources at

Microwaves101.comAntennas Basics and Fundamentals at http://www.antenna-theory.com.


Documents

Parametric Study of Insertion Losses and Antenna Pattern ...Microstrip Patch Antenna parameters. In this paper we present a comparative study of return loss, insertion loss of the