Design and implementation of a Ultra wide-band low-noise amplifier 3.1-4.8 GHz

Examensarbete

LITH-ITN-ED-EX--06/017--SE

Design and implementation of a

Ultra wide-band low-noise

amplifier 3.1-4.8 GHz

Erik Ottosson

LITH-ITN-ED-EX--06/017--SE

Design and implementation of a

Ultra wide-band low-noise

amplifier 3.1-4.8 GHz

Examensarbete utfört i Elektronikdesign

vid Linköpings Tekniska Högskola, Campus

Norrköping

Erik Ottosson

Handledare Adriana Serban

Examinator Shaofang Gong

Rapporttyp Report category Examensarbete B-uppsats C-uppsats D-uppsats _ ________________ Språk Language Svenska/Swedish Engelska/English _ ________________ Titel Title Författare Author Sammanfattning Abstract ISBN _____________________________________________________ ISRN _________________________________________________________________ Serietitel och serienummer ISSN

Title of series, numbering ___________________________________

Datum

Date

URL för elektronisk version

Avdelning, Institution

Division, Department

Institutionen för teknik och naturvetenskap Department of Science and Technology

2006-04-21

x

x

LITH-ITN-ED-EX--06/017--SE

Design and implementation of a Ultra wide-band low-noise amplifier 3.1-4.8 GHz

Erik Ottosson

The main purpose of this master thesis work was to design and implement an ultra-wideband 3.1-4.8 GHz low-noise amplifier. Moreover, a deeper understanding of the design process when creating a RF application was expected.

The design starts on the schematic level where amplifier is matched to a 50 §Ù termination. First the match was done with lumped components and then with distributed components. The multi-section matching networks were made using two microstrip stubs to achieve a wider bandwidth. The design was then carried on to the layout level where finally the entire circuit was implemented in a PCB. The design and all the simulations were made using ADS from Agilent Technologies Inc.

The PCB was manufactured, assembled and finally measured with a network analyzer. This report shows a functional design with an approximate gain of 10 dB and a noise figure of 2.6 dB at the highest.

Upphovsrätt

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –

under en längre tid från publiceringsdatum under förutsättning att inga

extra-ordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,

skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för

ickekommersiell forskning och för undervisning. Överföring av upphovsrätten

vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av

dokumentet kräver upphovsmannens medgivande. För att garantera äktheten,

säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ

art.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i

den omfattning som god sed kräver vid användning av dokumentet på ovan

beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan

form eller i sådant sammanhang som är kränkande för upphovsmannens litterära

eller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press se

förlagets hemsida

Copyright

The publishers will keep this document online on the Internet - or its possible

replacement - for a considerable time from the date of publication barring

exceptional circumstances.

The online availability of the document implies a permanent permission for

anyone to read, to download, to print out single copies for your own use and to

use it unchanged for any non-commercial research and educational purpose.

Subsequent transfers of copyright cannot revoke this permission. All other uses

of the document are conditional on the consent of the copyright owner. The

publisher has taken technical and administrative measures to assure authenticity,

security and accessibility.

Design and implementation of an

Ultra-Wideband 3.1 - 4.8 GHz Low-Noise Amplifier

Erik Ottosson

Preface

This Master Thesis work is the last step towards the Master of Science examination in Electronics Design Engineering at Linköping University. The work has been carried out at the Department of Science and Technology at Linköping University, Campus Norrköping.

I would like to thank every one who helped me during this Master Thesis work. Special thanks to my supervisor Adriana Serban for her full assistance. I would also like to thank Magnus Karlsson, Allan Huynh and Pär Håkansson for all their help. Thanks to Johan Lönn and Jonas Olsson for their help with the assembly, my examiner Shaofang Gong for valuable input and comments on the design and finally my opponent Sang Yu for continuous support throughout the project.

Abstract

The main purpose of this master thesis work was to design and implement an ultra-wideband 3.1-4.8 GHz low-noise amplifier. Moreover, a deeper understanding of the design process when creating a RF application was expected. The design starts on the schematic level where amplifier is matched to a 50 Ω termination. First the match was done with lumped components and then with distributed components. The multi-section matching networks were made using two microstrip stubs to achieve a wider bandwidth. The design was then carried on to the layout level where finally the entire circuit was implemented in a PCB. The design and all the simulations were made using ADS from Agilent Technologies Inc.

The PCB was manufactured, assembled and finally measured with a network analyzer. This report shows a functional design with an approximate gain of 10 dB and a noise figure of 2.6 dB at the highest.

Sammanfattning

Huvudsyftet med detta examens arbete var att designa och implementera en UWB 3.1-4.8 GHz lågbrusförstärkare. Dessutom uppnå en djupare förståelse av hur design processen då man skapar en RF applikation.

Designen börjar på schemanivå där förstärkaren matchas till 50 Ohms terminering. Matchningen gjordes först med diskreta komponenter och sedan med distribuerade komponenter. Så kallade multi-stage matchande nätverk användes för att åstadkomma en större bandbredd. Designen fortsattes sedan på layoutnivå där tillslut hela kretsen sattes samman i en PCB. Design och simuleringar gjordes i ADS från Agilent Technologies Inc.

PCB tillverkades, monterades och mättes med en nätverksanalysator. Denna rapport visar en fungerande design med en förstärkning på ca 10 dB och en noise figure på 2.6 dB som mest.

Table of contents

1 Introduction ... 1

1.1 Background ... 1

1.2

Purpose... 1

1.3

Method ... 1

1.4

Outline... 2

2 Theory ... 3

2.1

General

RF theory... 3

2.1.1 Transmission line theory ... 3

2.1.2 Microstrip line ... 5

2.1.3 Scattering parameters... 7

2.1.4 Termination conditions... 8

2.2 Low-noise amplifier theory ... 9

2.2.1 Stability ... 10

2.2.2 Gain ... 10

2.2.3 Noise ... 12

2.2.4 Matching networks ... 12

2.2.4.1 Design approach ... 13

2.2.4.2 Matching networks using transmission lines... 15

2.2.4.3 Quality factor... 17

3 Designing the LNA ... 19

3.1 Design Specifications... 20

3.2

Matching

Networks... 21

3.3

Bias

Network ... 25

4 PCB Layout ... 28

5 Results ... 30

5.1 Simulation results... 30

5.1.1 LNA gain and noise figure... 31

5.1.1.1 Microstrip width 0.542 mm ... 31

5.1.1.2 Microstrip width ± 10%... 33

5.1.2 RF-choke ... 36

5.1.3 Matching networks ... 37

5.1.3.1 Input matching network... 37

5.1.3.2 Output matching network ... 38

5.2 Measurement results ... 39

5.2.1 LNA ... 39

5.2.1.1 LNA with added copper wire ... 40

5.2.1.2 LNA with shorter open stubs... 41

5.2.1.3 Microstrip width ± 10%... 42

5.2.2 RF-choke ... 44

5.2.3 Matching networks ... 45

5.2.3.1 Input matching network... 46

5.2.3.2 Output matching network ... 47

6 Discussion... 48

7 Conclusions ... 50

8 Further work ... 51

References ... 52

Appendix A: Possible microstrip matching designs ... 53

Appendix B: PCB design ... 54

List of figures

Fig 2.1: Segment of transmission line expressed with lumped elements ...3

Fig 2.2: Microstrip line ...5

Fig 2.3: Two-port network definition ...7

Fig 2.4: Typical front-end circuit ...9

Fig 2.5: Simplification of a single stage amplifier ...10

Fig 2.6: Eight possible combinations of matching networks ...13

Fig 2.7: Effect when adding a reactance. ...14

Fig 2.8: Single stub matching network ...15

Fig 2.9: Design example for single stub matching network. ...16

Fig 2.10: Constant Q contours. ...18 n Fig 2.11: Single stub (a) and double stub (b) match ...18

Fig 3.1: MAX2649 pin configuration and dimensions ...20

Fig 3.2: ADS Amplifier Design Guide interface. ...21

Fig 3.3: Input matching network with microstrip lines. ...22

Fig 3.4: Matching network with IMN as layout component ...23

Fig 3.5: Changes of gain curve characteristics (in dB). ...23

Fig 3.6: Matched LNA ...24

Fig 3.7: RF-choke...25

Fig 3.8: S21 (dB) simulation of radial stub ...26

Fig 3.9: RF-choke dimensions ...26

Fig 3.10: Shutdown design. ...27

Fig 4.1: LNA footprint ...28

Fig 4.2: SMD pad dimensions ...28

Fig 4.3: PCB structure ...29

Fig 5.1: Noise Figure simulation ...31

Fig 5.2: S21 simulation ...32

Fig 5.3: Noise Figure simulation strip width +10% ...33

Fig 5.4: S21 simulation strip width +10% ...33

Fig 5.5: Noise Figure simulation strip width -10% ...34

Fig 5.6: S21 simulation strip width -10% ...34

Fig 5.7: S21 simulation of the RF-choke ...36

Fig 5.8: S11 simulation of the RF-choke ...36

Fig 5.10: S22 simulation of IMN ...37

Fig 5.11: S11 simulation of OMN ...38

Fig 5.13: LNA device ...39 Fig 5.14: S21 measurement ...40 Fig 5.15: S21 simulation ...40 Fig 5.16: S21 measurement ...41 Fig 5.17: S21 measurement ...42 Fig 5.18: S21 measurement ...42

Fig 5.19: RF-choke test device ...44

Fig 5.20: S11 measurement of RF-choke ...44

Fig 5.21: S21 measurement of RF-choke ...44

Fig 5.22: IMN test device ...46

Fig 5.23: S11 measurement of IMN ...46

Fig 5.24: S22 measurement of IMN ...46

Fig 5.25: OMN test device ...47

Fig 5.26: S11 measurement of OMN ...47

Fig 5.27: S22 measurement of OMN ...47

Fig 6.1: DC path display ...48

Fig 6.2: Stub detachment ...49

Fig A.1: Design A ...53

Fig A.2: Design B ...53

Fig A.3: Design C ...53

Fig B.1: LNA PCB design ...54

Fig B.2: RF-choke PCB design ...54

Fig B.3: IMN PCB design ...55

Fig B.4: OMN PCB design ...55

List of tables

Table 3.1: Substrate properties ...20

Table 3.2: Amplifier properties ...20

Table 5.1: Simulated LNA gain change ...35

Table 5.2: Simulated LNA noise change ...35

List of abbreviations

ADS Advanced Design Systems

BPF Band Pass Filter

GND Ground

IMN Input Matching Network

LNA Low Noise Amplifier

OMN Output Matching Network

PCB Printed Circuit Board

RF Radio Frequency

SMA Sub Miniature version A

SMD Surface Mounted Device

SWR Standing Wave Ratio

UWB Ultra-wideband

1 Introduction

This chapter gives the reader an introduction to the report. The background and the purpose of the thesis work are presented followed by the method of the work and the outline of the report.

1.1

Background [1], [2]

Ultra-wideband radio (UWB) is a wireless communication technology based on multi-band OFDM and the spread spectrum technologies. It transmits data in short pulses which are spread out over a wide range of frequencies. Because of the wide-band used the data rate becomes very high. It can also share frequencies used by other applications because it transmits at low power levels it will not interfere with other signals. The technology was first used by the US military in the Second World War. Because the communication signals were divided over several different frequencies the enemy could only pick up on fragments of it. High data rates and low transmitted power make UWB systems very attractive for wireless communications. In a multi-band UWB transceiver the 7.5 GHz bandwidth is divided into 528 MHz sub-bands. The spectrum between 3.1 and 4.8 GHz is called Band Group 1. This report describes the work of designing a low- noise amplifier (LNA) for this band.

1.2 Purpose

The main purpose for this master thesis work was to design and implement an ultra-wideband 3.1-4.8 GHz low-noise amplifier, and to understand the design process when creating a RF module design. Another objective was to get to know the Advanced Design System (ADS) tool from Agilent Technologies Inc. better.

1.3

Method

The theoretical background needed for this thesis work was obtained from various books, articles and internet sites. The design and all the simulations were done using the ADS tool. The measurements were made using a network analyzer.

Section 1.4 Outline Chapter 1 Introduction

1.4 Outline

The report includes the following chapters:

• Chapter 2 contains the theoretical background related to this report. The RF theory and general LNA theory are presented here.

• Chapter 3 describes the entire design process from schematic representation to layout.

• Chapter 4 shows how the final PCB layout was designed.

• In Chapter 5 all the results including both simulations and measurements are given.

• Chapter 6 covers the discussions made regarding the results and the design process in this thesis.

• Chapter 7 lists the conclusions drawn based on this thesis work. • Chapter 8 gives suggestions for further work in this area.

2 Theory

The general theoretical background needed for this thesis work is given here. To get a deeper understanding one may see the books and articles listed in the references.

2.1

General RF Theory [3]

The Kirchoff’s voltage and current laws of the circuit theory do not apply at high frequencies. When the wavelength becomes smaller than the length of the transmission line reflections may occur because the terminations are not matched to the characteristic impedance of the wire.

2.1.1

Transmission Line Theory

The Kirchoff’s voltage and current laws can not be applied to high-frequency circuits. However, if one looks at a segment of the transmission line, much smaller than the wavelength, these laws can still be used.

Fig 2.1: Segment of transmission line expressed with lumped elements

By using Kirchoff’s voltage law we conclude from fig. 2.1:

) ( ) ( ) ( ) (R+ jωL I z ∆z+V z+∆z =V z (2.1)

which can be expressed as the differential equation:

) ( ) ( ) ( z I L j R dz z dV _{= + ω} − (2.2)

Section 2.1 General RF Theory Chapter 2 Theory

Applying the Kirchoff’s current law to the node a in fig. 2.1 yields:

) ( ) ( ) ( z V C j G dz z dI _ω + − = (2.3)

The standard second-order differential equation is expressed as: 0 ) ( ) ( 2 2 2 = −k V z dz z V d (2.4) 0 ) ( ) ( 2 2 2 = −k I z dz z I d (2.5) where k is the complex propagation constant described as:

) )( (R j L G j C jk k k = _r + _i = + ω + ω (2.6)

When solving equation (2.4) and (2.5) exponential functions for the voltage and the current are found:

kz kz e V e V z V( )= + − + − (2.7) kz kz e I e I z I( )= + − + − (2.8)

From the current and voltage expressions the characteristic line impedance can be derived: ) ( ) ( ) ( 0 C j G L j R k L j R Z ω ω ω + + = + = (2.9)

The characteristic line impedance is not an impedance in the normal circuit sense because both the voltage and the current are considered as positive and negative waves.

Section 2.1 General RF Theory Chapter 2 Theory

2.1.2 Microstrip

Line

The signal moving through the transmission line can be considered to be a wave

with a certain phase velocity, v_p, defined as:

r r p c f v µ ε λ = = (2.10)

where is the speed of light, c ε_r is the dielectric constant and µ_r is the relative

permeability.

Planar printed circuit boards (PCBs) are used in most electronic systems. The PCB makes it easy to access components on the board and is fairly cheap to manufacture. On the other hand, the problems with PCBs are the high loss and interference between the microstrip lines. A microstrip line is applied on a dielectric substrate with a certain thickness, as shown in fig. 2.2.

Fig 2.2: Microstrip line

If one neglects the thickness t of the conductor in comparison to the thickness d of

the substrate the characteristic impedance of the microstrip can be presented in

form of its physical dimension and the dielectric constant

0 Z

r

ε . For a thin line,

: 1 /d < w ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ₊ = h w w h Z Z eff f 4 8 ln 2 0 _{π ε} (2.10)

where Z_f is the wave impedance in free space and ε is the effective dielectric _eff

constant. Ω = = 376.8 0 0 ε µ f Z (2.11)

Section 2.1 General RF Theory Chapter 2 Theory ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − + ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + − + + = − 2 2 1 1 04 . 0 12 1 2 1 2 1 d w w d r r eff ε ε ε (2.12)

For a wide line, w/d >1,

⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + + + = 444 . 1 ln 3 2 393 . 1 0 d w d w Z Z eff f ε (2.13) with 2 1 12 1 2 1 2 1 − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + − + + = w d r r eff ε ε ε (2.14)

With the above statements the phase velocity of the microstrip line can be expressed as: eff p eff p f c f v c v ε λ ε ⇒ = = = (2.15)

Section 2.1 General RF Theory Chapter 2 Theory

2.1.3 Scattering

Parameters

The scattering parameters, or S-parameters, are used to describe input and output relations of a two-port network. As seen in fig. 2.3 below the incident normalized

power wave a_n and the reflected normalized power wave are defined as: b_n

(

)

(

)

where the index n is referring to the port number 1 or 2 and is the

characteristic impedance of the lines connected to the input and the output of the two-ports.

0 Z

Fig 2.3: Two-port network definition

The S-parameters are defined as follows:

⎭ ⎬ ⎫ ⎩ ⎨ ⎧ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ = ⎭ ⎬ ⎫ ⎩ ⎨ ⎧ 2 1 22 21 12 11 2 1 a a S S S S b b (2.17)

where describes the reflection at the input, is the output reflection

coefficient, is the forward voltage gain and is the reverse voltage gain.

11

S S₂₂

21

Section 2.1 General RF Theory Chapter 2 Theory

2.1.4 Termination

Conditions

At a distance d away from the load the input impedance is given by the expression:

) tan( ) tan( ) ( 0 0 0 d jZ Z d jZ Z Z d Z L L in β β + + = (2.18)

where β is the wave number defined as:

λ π π β = 2 = 2 p v f (2.19) If the transmission line is short circuited, i.e., the load impedance is equal to zero

the input impedance is described as: )

tan( )

(d jZ₀ d

Z_in = β (2.20)

If d reaches a quarter-wave length the input impedance will be:

L L L in Z Z jZ Z jZ Z Z d Z 2 0 0 0 0 2 2 tan 2 2 tan ) 4 / ( = ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + = = λ λ π λ λ π λ (2.21)

Because the load impedance is equal to zero the input impedance will be infinite, which represents an open-circuit condition. The open circuit transmission line input impedance is expressed:

) tan( 1 ) ( ₀ d jZ d Z_in β − = (2.22)

Section 2.2 LNA Theory Chapter 2 Theory

2.2

Low-noise amplifier theory [3], [4]

Since an LNA is the first gain stage in the receive path, its noise figure dominates that of the system. Therefore the main issue when designing an LNA is the balance between a high gain and a low noise figure.

Fig 2.4: Typical front-end circuit

The interface between the antenna and the LNA entails an interesting issue. From the noise point of view, we may require a transformation network to precede the LNA so as to obtain minimum noise figure. From the signal power point of view, we may utilize conjugate matching between the antenna and the LNA. Both these methods have its advantages and its drawbacks but the last one is dominant in today’s systems because the LNA is often designed to have a 50 Ω resistive input impedance. This is due to the fact that the band pass filter following the antenna is used in various transceiver systems and must therefore operate with a standard termination impedance, i.e., 50 Ω. If the source and load impedances seen by the filter deviate from 50 Ω, then the passband and stopband characteristics may show considerable loss and ripples.

The quality of the input matching is expressed by the input return loss, defined

as20logΓ, where Γ is:

0 0 R Z R Z in in + − = Γ (2.23)

Section 2.2 LNA Theory Chapter 2 Theory

2.2.1 Stability

In addition to the above parameters, the stability of an LNA is also of concern. In the presence of feedback paths from the output to the input, the circuit may become unstable for certain combinations of source and load impedances. This could make the LNA oscillate at the extremes of manufacturing variations and perhaps at unexpected high or low frequencies.

The stability of the LNA is characterized by the Stern stability factor or Rolett factor, K, defined as:

12 21 2 22 2 11 2 2 1 S S S S K = + ∆ − − (2.24) where 21 12 22 11S S S S − = ∆ (2.25)

If K > 1 and ∆ < 1, the circuit is unconditionally stable, i.e., it does not oscillate with any combination of source and load impedances. One way to stabilize the amplifier is to add a series or shunt resistor at either the input or the output. If the resistor is placed at the input, the noise generated by it will be amplified. This is obviously not desirable, which is why the resistor should be placed at the output.

2.2.2 Gain

There are various power gain definitions that are important in order to understand how an RF amplifier functions.

Section 2.2 LNA Theory Chapter 2 Theory

With the simplification in fig. 2.5 the incident power , which is the power

launched toward the amplifier, can be expressed as:

inc P 2 2 2 1 1 2 1 2 ´ S in S inc b b P Γ Γ − = = (2.26)

In order to get the input power of the amplifier the reflected power waves have to be taken in to account:

(

)

When the amplifier is properly conjugate-matched the available power is

defined as: A P 2 2 1 2 1 * S S in A b P P S in = − Γ = _{Γ =Γ} (2.28)

where Γ_S is the source reflection coefficient.

To quantify the gain of the amplifier placed between source and load the

transducer power gain G_T has to be calculated as follows:

A L T P P G = = source the from power available load the to delivered power (2.29) where

(

)

This results in:

(

) (

)

(

)(

)

Section 2.2 LNA Theory Chapter 2 Theory

2.2.3 Noise

The most important parameter of the low noise amplifier is the noise figure. As mentioned earlier an amplifier with both a minimum noise figure and a maximum gain cannot be designed. Therefore a compromise has to be done between the noise, gain and stability. The noise figure F is defined as:

(

)

where is the minimum (or optimum) noise figure, is the equivalent noise

resistance and is the optimum reflection coefficient.

min

F R_n

opt Γ

2.2.4 Matching

networks

Matching networks are used to match the input and output impedances of the amplifier with the termination impedances. This is necessary to achieve maximum power transfer. The matching networks can also be used to minimize noise and linearize the frequency response. In short, the general purpose of the matching network is to change the impedance from one value to another more suitable for that specific design.

Section 2.2 LNA Theory Chapter 2 Theory

2.2.4.1 Design approach

The easiest way to create a matching network is to use discrete components. Although these networks can only be used in low frequency applications they are easy to analyze. The so called L-type or two-component networks are the simplest design approach. The design consists of one series and one shunt component, either a capacitance or an inductance. This leads to eight different possible combinations of networks.

Fig 2.6: Eight possible combinations of matching networks

When designing the networks the Smith Chart is a useful tool. The effect of adding a reactive component to a complex load is the following:

• Adding a reactance in series with the load impedance results in a motion along the constant resistance circle shown in fig. 2.7.

• Adding a shunt reactance results in a motion along the constant conductance circles shown in fig. 2.7.

Section 2.2 LNA Theory Chapter 2 Theory

The direction of the motion depends on whether it is a capacitance or an inductance. This motion makes it possible to go from one impedance to any desired impedance. The process is simple, starting with plotting out the normalized source and load impedances. An amplifier is usually matched to the standard impedance 50 Ω, i.e., in the centre of the Smith Chart. Draw the constant resistance and conductance circles that coincide with the source and the complex conjugate of the load impedance points. Find out where the circles intersect each other, the number of intersections gives the number of possible L-type networks. Now following the path between the load and source the values of the reactances and the susceptances can be found, which lead to the computation of the discrete components.

Section 2.2 LNA Theory Chapter 2 Theory

2.2.4.2 Matching networks using transmission lines

The discrete components cannot be used in high frequency circuits and therefore the design with distributed elements is a good alternative. The design uses open and short stubs instead of the lumped components. The impedance of the stub depends on its length and its characteristic impedance.

Fig 2.8: Single stub matching network

The concept is to select the length of the stub such that it produces a susceptance sufficient to move the load admittance to the standing wave ratio SWR circle that passes through the input impedance point. The standing wave ratio circles have its centre in the middle of the Smith Chart.

0 0 1 1 Γ − Γ + = SWR (2.33)

The design process is the following. Start with plotting out the SWR circle that

passes through the source impedance point and the constant conductance circle

that passes through the load impedance point . Find out where the circles

intersect each other and choose the design path from load to source. The length of

the stub is determined by the difference in susceptance when moving from to

the intersection point A. When creating an open stub one starts in the open point of the Smith Chart and move along the outer perimeter toward the generator to the difference in susceptance point, see fig. 2.9. The length of the stub is shown in terms of the wavelength, in the example below

S z L z L z λ 067 . 0 = S

l . Finally the length of

the series transmission line is found in outer perimeter of the chart when moving along the constant SWR circle to the load impedance point, in the example the length is determined to be 0.266 of the wavelength.

Section 2.2 LNA Theory Chapter 2 Theory

Fig 2.9: Design example for single stub matching network.

To achieve a wider bandwidth of the frequency response a double stub matching network can be designed, more on this topic is covered in Section 2.2.4.3.

Section 2.2 LNA Theory Chapter 2 Theory

2.2.4.3 Quality factor

There are always at least two different possible L-type matching networks to choose between. So the question is why one design should be chosen over another. The choice may depend on the frequency response or perhaps on the quality factor of the network.

All L-type matching networks can be described by a loaded quality factor ,

called loaded because it describes the power loss associated with the network and the external load.

L Q

BW f

Q_L = 0 (2.34)

where is the resonance frequency and BW is the 3 dB bandwidth. Because it is

hard to find the bandwidth of a matching network a so-called nodal quality factor could be calculated. As the name implies it describes the quality factor in each

node of the network defined as either the ratio of the reactance to the

corresponding resistance or as the ratio of the susceptance to the

corresponding conductance . 0 f n Q S X S R B_P P G P P S S n G B R X Q = = (2.35)

The relation between the loaded and the nodal quality factor is defined;

2

n L

Q

Q = (2.36)

To simply the matching design the constant contours can be added in the

Smith Chart when designing the networks.

n Q

Section 2.2 LNA Theory Chapter 2 Theory

Fig 2.10: Constant Qn contours.

Equation (2.34) shows that to keep the bandwidth high the quality factor should be as low as possible. As seen in fig. 2.11 when matching with a single stub network a quality factor of 2 is achieved and when doing the same match with a double stub design the quality factor is lowered to 1.1. This means that to create a design for wide band application a double stub network is preferable.

3

Designing the LNA

When designing the LNA the ADS tool provided by Agilent Technologies was used. The process began on the schematic level where all components are considered ideal. The simulations on this level are fast because of its primitive nature. The emphasis in this stage of the process is to get a basic structure of the circuit. The circuit is trimmed to meet the specifications before the design process moves on to the next level.

The next level is the layout level where components and transmission lines are no longer considered ideal. ADS uses the Momentum tool to simulate the actual physical dimensions of the signal paths. This electromagnetic simulation leads to a result more similar to the truth.

When a satisfying result has been achieved on the schematic layer, a part of the schematic circuit is replaced with its corresponding layout circuit. This is done by creating a so called “layout component” and adding it to the schematic. This part will then be simulated using the Momentum electromagnetic simulator while the rest of the circuit is simulated with the simpler schematic simulator. This way it is easy to ensure that the desired characteristics of the circuit are preserved on the layout level.

Section 3.1 Design Specifications Chapter 3 Designing the LNA

3.1

Design specifications [5], [11]

The overall design principles applied on the LNA is:

• Design input matching network giving a close to minimum noise figure. • Design a frequency compensating output matching network to create a

high and flat gain over the entire specified bandwidth.

• Extend the band of the IMN and OMN by using multiple-section reactive networks and by optimizing the quality factor Q of the networks towards low and constant values.

Table 3.1 lists the properties of the Rogers substrate used in this design and table 3.2 displays the properties of the MAX2649 low-noise amplifier.

Table 3.1: Substrate properties

Dielectric thickness 0.254 mm Dielectric constant 3.48 ± 0.05 Dissipation factor 0.0037 Metal thickness 35 um Metal conductivity 5.8 × 107 _S/m Surface roughness 0.001 mm RO4350B

Table 3.2: Amplifier properties

Band of operation 4.9 - 5.9 GHz

Low noise figure 2.1 dB

High gain 17 dB

Single-supply op. 2.7 - 3.6 V

Chip size 1 × 1.5 mm

Section 3.2 Matching Networks Chapter 3 Designing the LNA

3.2 Matching

Networks

To obtain the maximum power the LNA must be matched to the standard 50 Ω termination impedance. In order to do this the S-parameters of the MAX2649 must be simulated. Maxim provides an S-parameter file and with the help of the ADS S2P block the component can be simulated. The ADS amplifier design guide (DesignGuide > Amplifier > S-parameter Simulations > S-params., Noise Fig.,

etc.) was used because it is a very good interface containing all the parameters

needed. Including gain-, noise- and stability-curves, source and load impedances for gain and noise matching and Smith chart etc.

Fig 3.2: ADS Amplifier Design Guide interface.

First of all the stability of the LNA needs to be analyzed. In order to stabilize it a shunt resistor was added at the output. As mentioned earlier in Section 2.2.1 this is the way to add as less noise as possible to the system. After sweeping the parameter the impedance was calculated to be 117 Ohm making the amplifier unconditionally stable. It is important to keep an eye on the Rolett factor during the design process because when you add the matching microstrip networks the

Section 3.2 Matching Networks Chapter 3 Designing the LNA

amplifier may become unstable and a new sweep is necessary to determine the new impedance.

By moving the slide bar in the Design Guide window to the desired matching frequency the source and load impedances can be found. There are two ways of matching the amplifier, matching for noise or for gain. Because this is a low noise amplifier the obvious choice is to match for minimum noise. Now the design of the matching networks can begin. The design is made with the ADS Smith Chart tool (Tools > Smith Chart...). With this tool it is easy to build the network with ideal distributed components. The matching network designed is a multi-section type network, i.e., a double stub network shown in fig. 3.3. This allows a low quality factor which leads to a wider IMN and OMN band as discussed in Section 2.2.4.3. After the design was completed the network was added in the schematic design and simulated to verify the match. The design of the output matching network was done in the same way.

Next the ideal transmission lines were replaced by microstrip lines. The physical width and length was calculated with the help of the LineCalc tool in ADS (Tools

> LineCalc > Start LineCalc) by submitting the characteristics of the substrate

and the metal along with the electrical length and the characteristic impedance of the transmission line.

MLSC TL19 L=3.221 mm MLOC TL17 L=5.01 mm MLIN TL14 L=2 mm Port P2 Port P1 MLIN TL12 L=1 mm W=0.2 mm MSTEP Step1 MTEE_ADS Tee2 MTEE_ADS Tee1

Fig 3.3: Input matching network with microstrip lines.

The microstrip design demands T-junctions and because of their dimensions the length of the stubs may have to be altered in order to keep the match. A parameter sweep was made to determine the optimal microstrip lengths. Because the dimensions of the MAX2649 solder pads are smaller than the width of the

Section 3.2 Matching Networks Chapter 3 Designing the LNA

was where the difficult part of the design started. When simulating the circuit it was clear that the length of the stubs had to be redesigned in order to keep the gain high and flat over the whole frequency spectrum. The open stub was ultimately determined to be 7.5 mm and the short stub to be 4.5 mm.

Fig 3.4: Matching network with IMN as layout component

From start it looked like the output matching network did not had to be redesigned but when the solder pads to the stabilizing resistor was added the characteristics of the simulation result changed drastically. The noise figure became much higher and the gain curve peaked at 4.8 GHz. Without a flat gain curve the signal coming through the amplifier may be distorted. Furthermore the noise figure must be kept below 3 dB. In fig. 3.5 below the difference in simulation result throughout the design process is shown.

Section 3.2 Matching Networks Chapter 3 Designing the LNA

By radically changing the length of the open and short stub the curve regained the characteristics desired. The noise however was still an issue. Because the width of the solder pad is 1.25 mm a microstrip with the same length was needed to attach it to. The added strip and the solder pad were the parts contributing with the most noise. The solder pad has a standard size which can not be changed so this meant that the microstrip had to be shortened as much as possible. By making the strip half as short the noise figure was reduced to 2.6 dB.

Section 3.3 Bias Network Chapter 3 Designing the LNA

3.3 Bias

Network

The LNA is an active component which means that a power supply is needed in order for it to function. The bias network of this design includes an RF-choke to block the RF signal to the power supply path and a shutdown function which makes the amplifier able to save power when it is not being used.

3.3.1 RF-choke

The output of the LNA is connected to a DC supply. To prevent the signal from flowing to the DC supply path a RF-choke is needed. It can be designed using lumped elements, i.e. inductors, or with distributed components. In the figure below the RF-choke design using distributed elements is shown.

Fig 3.7: RF-choke

The radial stub creates a virtual RF-ground in the junction. By choosing the transmission line length equal to one quarter of the wavelength, the RF-signal will experience high impedance at the T-junction (see Section 2.1.4) preventing it from reaching the DC supply and instead continuing on to the RF-output.

Section 3.3 Bias Network Chapter 3 Designing the LNA

The easiest way to design this component is to use a single radial stub. The reason for using a choke with a so-called butterfly radial stub in this design is because of the fact that this is a wideband LNA. This means that a wide variety of frequencies must be prevented from reaching the DC supply. The butterfly choke offer a some what wider virtual ground bandwidth compared to the single radial stub (see fig. 3.8) and that is why the design of a butterfly radial stub was chosen.

Fig 3.8: S21 (dB) simulation of radial stub

The bandwidth of the LNA reaches from 3.1-4.8 GHz and therefore the frequency chosen to the centre of the virtual ground was 4 GHz. By using the tool LineCalc in ADS and adding the characteristics of the Rogers substrate, the quarter-wavelength transmission line was computed to be 11.47 mm. A parameter sweep of the angle and length of the radial stub was necessary to obtain the desired simulation result. The physical dimensions of the RF-choke can be seen in fig. 3.9 below. MBSTUB Radial_stub D=0.542011 mm Angle=60 W =0.542011 mm MLIN TL1 L=11.47 mm W =0.542011 mm Port VCC

Section 3.3 Bias Network Chapter 3 Designing the LNA

3.3.2

Shutdown [5]

The shutdown network was designed exactly according to the MAX2649 datasheet. It contains a 100 Ohm resistor in series with a 20 pF shunt capacitor. To make the LNA active the shutdown logic signal should be set high and when it is set low the shutdown mode is in effect.

R R1 R=100 Ohm Port P2 Port P1 C C1 C=20 pF

4 PCB

Layout

The different parts of the design are put together into a PCB layout. A footprint of the LNA has to be created to attach the chip. The footprint can be seen in fig. 4.1 below and was designed according to the package dimensions in Section 3.1. The IMN and OMN were attached to the correspondent RF-in and RF-out pin stated by the configuration also shown in Section 3.1.

Fig 4.1: LNA footprint

The power supply path needs decoupling capacitors together with the RF-choke. The choke blocks the frequencies in the specified spectrum but the decoupling is needed to block low frequencies in the MHz region. The input and output port also need DC blocking capacitors. To be able to mount these components on the board SMD pads were designed, see fig. 4.2. The component sizes used were 0603 for the capacitors and 0805 for the resistors. Because the width of the traces is 0.542 mm in this design the DC blocking pads was designed to match that width. Otherwise unwanted parasitic effects could emerge when measuring the device. The discrete components used in the design are displayed in Appendix C.

Chapter 4 PCB Layout

To be able to apply a logic signal and supply voltage to the device, vias for pin connectors were created with a diameter of 1 mm. When the designed parts were integrated the ground plane was added with a clearance of twice the size of the trace width, i.e., 1.1 mm. The ground plane on the primary side (i.e., circuit side) has to be connected with the secondary side ground plane, in this design called “Metal 4”. This was done by adding several small through-hole vias. Without the vias in the ground plane small currents may occur, causing effect leakage and interference problems. The dielectric material used in this design was Rogers with a thickness of only 0.254 mm. To make the circuit board more robust a FR-4 substrate and pre-preg was added with a metal between each layer. The complete structure of the PCB is shown in fig. 4.3.

5 Results

The results in this chapter are divided into two sub groups; layout simulation and measurement results.

5.1 Simulation

Results

The simulations of the matching networks, the broadband RF-choke and the entire LNA are displayed in this section. All simulations were made using the ADS design tool from Agilent Technologies.

Section 5.1 Simulation Results Chapter 5 Results

5.1.1

LNA gain and noise figure

Designs with three different microstrip widths were created for comparison reasons. One with the original 50 Ω transmission line width of 0.542 mm (at 4.5 GHz) and the other two plus and minus 10 % in width.

5.1.1.1 Microstrip width 0.542 mm

The noise figure of the LNA is shown in fig. 5.1 (the blue curve), at most 2.626 dB at 4.8 GHz. The red curve characterizes the minimum noise figure of the circuit.

Section 5.1 Simulation Results Chapter 5 Results

In fig. 5.2 the gain of the LNA is displayed. The result shows a flat gain of 14.295 dB at 3.1 GHz and 14.599 dB at 4.8 GHz.

Section 5.1 Simulation Results Chapter 5 Results

5.1.1.2 Microstrip width ± 10%

The noise figure simulation of the LNA with wider transmission lines shows a slightly higher result, 2.799 dB at 4.8 GHz (see fig 5.3).

Fig 5.3: Noise Figure simulation strip width +10%

In fig. 5.4 the gain of the LNA with microstrip width +10% is displayed. The result shows a somewhat flat gain of 13.666 dB at 3.1 GHz and 14.819 dB at 4.8 GHz.

Section 5.1 Simulation Results Chapter 5 Results

The noise figure (see figure 5.5) in this design is significantly lower compared to the original design, only 2.323 dB at 4.8 GHz.

Fig 5.5: Noise Figure simulation strip width -10%

The gain of the LNA with thinner transmission line width also shows a flat gain of approximately 14 ± 0.4 dB. The S21 simulation is displayed in figure 5.6.

Section 5.1 Simulation Results Chapter 5 Results

Table 5.1 lists the gain at three different frequencies for the designs of different microstrip widths. The simulated result shows that the change in trace width has little effect on the amplifiers gain. The highest deviations can be found at the low frequency while almost no change appears at the high frequency.

Table 5.1: Simulated LNA gain change

LNA gain change

Width variation 3.1 GHz 4 GHz 4.8 GHz minus 10% (dB) 13.6 14.3 14.4 nominal (dB) 14.3 15.7 14.6 plus 10% (dB) 13.7 14.2 14.8 Tolerance (%) 3.8 2.8 0.1

The noise however has a wider variation as seen in table 5.2. At 3.1 GHz the noise figure only deviate 5.2% from the nominal value but at 4.8 GHz the result is over 18%.

Table 5.2: Simulated LNA noise change

LNA noise figure change Width variation 3.1 GHz 4.8 GHz minus 10% (dB) 1.492 2.323 nominal (dB) 1.355 2.626 plus 10% (dB) 1.421 2.799 Tolerance (%) 5.2 18.1

Section 5.1 Simulation Results Chapter 5 Results

5.1.2 RF-choke

The RF-choke simulated shows a loss of 0.366 dB at the low frequency and 0.296 dB loss at the high frequency which is quite acceptable, see fig. 5.7.

Fig 5.7: S21 simulation of the RF-choke

Fig. 5.8 shows the bandwidth of the RF-blocking. -12.392 dB attenuation at the low frequency and -14.690 dB attenuation at the high frequency are good enough.

Section 5.1 Simulation Results Chapter 5 Results

5.1.3 Matching

Networks

5.1.3.1 Input matching network

Fig. 5.9 shows the reflection at the input of the IMN and fig. 5.10 the reflection at the output.

Fig 5.9: S11 simulation of IMN

Section 5.1 Simulation Results Chapter 5 Results

5.1.3.2 Output matching network

Fig. 5.11 shows the reflection at the input of the OMN and fig. 5.12 the reflection at the output.

Fig 5.11: S11 simulation of OMN

Section 5.2 Measurement Results Chapter 5 Results

5.2 Measurement

Results

The measurements of the matching networks, the RF-choke and the entire LNA are presented in this section. All the measurements were made using a Rohde & Schwarz ZVM vector network analyzer.

5.2.1 LNA

The short-circuited stubs had to be detached from the ground plane, creating open stubs. This was made in two different attempts. First a copper wire of quarter-wave length was soldered on to the stubs and second a matching with shorter open stubs was made. More on how these measurements were set up is covered in Section 6.

Section 5.2 Measurement Results Chapter 5 Results

5.2.1.1 LNA with added copper wire

The result presented in fig. 5.14 shows the forward gain (S21) of the LNA. The bandwidth is wide enough but a slight displacement results in a 5 dB gain at 4.8 GHz compared to approximately 10 dB at the centre frequency. The sources causing this error are many, for example, the copper wire must lie firmly on the ground plane otherwise the electrical length of the stub will be changed. However the measurement clearly indicates that the design works.

Section 5.2 Measurement Results Chapter 5 Results

5.2.1.2 LNA with shorter open stubs

In the measurement in fig. 5.16 the stubs were shortened to the length of Design B displayed in Appendix A. However, as discussed in Section 3.2 the matching lengths of the stubs on schematic level is seldom the same as the lengths on the layout level. The result shows a gain of approximately 6 dB, not as high as the previous measurement but the bandwidth is much wider here.

Section 5.2 Measurement Results Chapter 5 Results

5.2.1.3 Microstrip width ± 10%

Changing the microstrip width -10% from 0.542 mm to 0.488 mm shows a gain of approximately 9.5 dB as seen in fig. 5.17. The measurement was made by simply detach the short-circuited stubs from the ground plane.

Fig 5.17: S21 measurement

By adding 10% in microstrip width the measurement in figure 5.18 was achieved. The result shows a gain of approximately 8 dB. It is obvious that the trace width does affect the result in a way, a 1.5 dB loss in the S21 measurement.

Section 5.2 Measurement Results Chapter 5 Results

Compared to table 5.1 the measured result shows much more significant gain change when the microstrip width is altered.

Table 5.3: Measured gain changes

LNA gain change

Width variation 3.1 GHz 4 GHz 4.8 GHz minus 10% (dB) 9.6 9.5 8.6 nominal (dB) 9.1 8.5 7.7 plus 10% (dB) 8.1 8.0 6.8 Tolerance (%) 17.4 20.0 25.7

Section 5.2 Measurement Results Chapter 5 Results

5.2.2 RF-choke

Fig 5.19: RF-choke test device

The voltage supply path of the RF-choke test device in fig. 5.19 is connected to ground reference. As seen in fig. 5.20 the RF-choke has a wide blocking bandwidth and at the centre frequency it reaches -17.46 dB which is quite sufficient. The difference in curve characteristics compared to the simulated result in fig. 5.8 is due to the SMA contacts.

Fig 5.20: S11 measurement of RF-choke

In fig. 5.21 the signal loss when passing the choke is shown. The result shows a -3 dB bandwidth of 4.3 GHz and covers the frequencies from 1.8 GHz to 6.1 GHz.

Section 5.2 Measurement Results Chapter 5 Results

5.2.3 Matching

Networks

All measurements done on the matching networks differ from the simulated result. The main reason for this is believed to be the SMA-contacts added. They are however needed in order to measure the device using the network analyzer. The solder mask added when manufactured can also affect the measurements. Although the impedances measured are not the same as the simulated ones, the shapes of the curves are similar. The measurements also underline the mismatch made in order to reach the goal of a flat gain curve, see Section 3.2. When measuring the S11 parameter port 2 is connected to a 50 Ω termination and when measuring the S22 parameter port 1 is terminated.

Section 5.2 Measurement Results Chapter 5 Results

5.2.3.1 Input Matching Network

Fig 5.22: IMN test device

The S11 measurement of the IMN shows an impedance of 37.71-j35.13 Ω at 3.1 GHz and 22.11+j108.7 Ω at 4.8 GHz which is a bit different compared to the simulated result in fig. 5.9.

Fig 5.23: S11 measurement of IMN

The S22 measurement results in 77.11+j96.95 Ω at the centre-frequency as shown in fig. 5.24 comparable with the simulated result in fig. 5.10.

Section 5.2 Measurement Results Chapter 5 Results

5.2.3.2 Output Matching Network

Fig 5.25: OMN test device

When measuring the S11-parameter of the output matching network device, see fig. 5.25, an impedance of 97.22-j0.715 Ω was displayed at 3.1 GHz and 104.7-j21.08 Ω at 4.8 GHz. The measured result is displayed in figure 5.26 and the corresponding simulated result in fig. 5.11.

Fig 5.26: S11 measurement of OMN

In figure 5.27 the S22 measurement of the device is presented. The impedance at 3.1 GHz was determined to 48.76-j32.51 Ω and at 4.8 GHz to 42.75+j37.14 Ω, the corresponding simulated result is shown in fig. 5.12.

6 Discussion

To avoid the trouble with the peak of the gain curve after adding the stabilizing resistor pad, one can first attach the pad and do a new S-parameter simulation. This means that new source and load impedances can be found to match the amplifier. This way no struggle with mismatching is needed because the amplifier is matched to the right impedances.

When matching an amplifier with external voltage biasing use open stubs to avoid a DC short-circuit. As seen in figure 6.1 the DC path, displayed in yellow, is directly connected to the ground plane through the short stub in the output matching network. This was not detected during the simulations because no voltage supply was applied at that stage. The node was simply connected to a ground reference because in RF sense, that is what a VCC is. The problem was corrected by adding a quarter-wave length copper wire and thus creating an open stub with the same characteristics as the short stub.

Chapter 6 Discussion

Measurements were also made by simply detaching the short-circuited stub from the ground plane as seen in figure 6.2. After that it was possible to evaluate the affect of different microstrip widths. This problem with the short-circuited stubs could have been avoided by using the DC Rule Check in ADS.

7 Conclusions

The main goal of this master thesis work was to design and implement a broadband low noise amplifier. A functional design has been designed and implemented so this means that the requirements are met. The work has also resulted in a deeper understanding of the RF application design process.

The following conclusions have been made based on this work:

• It is possible to create a broadband low-noise amplifier using distributed elements.

• Avoid using short-circuited stubs when matching the amplifier. The match should be done with open stubs when external biasing is needed for the output stage.

• Changing the width of the microstrips does not affect the result in a large sense. That is one of the advantages of using distributed components instead of lumped, i.e., it is more tolerable.

8 Further

work

The further work to be done is obviously to redesign the matching networks with only open stubs. The biasing and the overall function of the circuit have been proven to be working.

References

[1] http://www.ultrawidebandplanet.com/faq (2006-01-19)

[2] A. Serban, S. Gong, Low-Noise Amplifiers Designs at 5 GHz, IMAPS Nordic Conference, 12-14 September 2005, Tonsberg, Norway.

[3] Reinhold Ludwig & Pavel Bretchko, RF Circuit Design – Theory and

Application, Prentice Hall 2000, ISBN 0-13-122475-1

[4] Behzad Razavi, RF Microelectronics, Prentice Hall PTR 1998, ISBN 0-13-887571-5

[5] http://www.maxim-ic.com (2006-03-15)

[6] First report order, revision of part 15 of commissions rules regarding

ultra-wideband transmission systems, FCC., Washington, 2002.

[7] Heydari, P., A study of low-power ultra wideband radio transceiver

architectures, Wireless Communications and Network Conference, 2005, 13-17

March 2005, Pages: 758-763 Vol. 2.

[8] Win M. Z. and R. A. Scholtz, Ultra-Wide Bandwidth Time-Hopping

Spread-Spectrum Impulse Radio for Wireless Multiple-Access Communications, IEEE

Trans. Communications, vol. 48, pp 679-691, April 2000.

[9] Gonzales, G., Microwave Transistor Amplifier Analysis and Design, Prentice-Hall, Upper Saddle River, New Jersey, 1997.

[10] Pozar, D. M., Microwave Engineering, Wiley, New York, 2005.

Appendix A

Possible microstrip matching designs

Three different possible designs can be made when matching the amplifier. Design A was used in this work. Design B is made using only open stubs while Design C has an open stub IMN and a mixed OMN.

MLOC TL17 L=5.01 mm W=0.542011 mm MLSC TL19 L=3.221 mm W=0.542011 mm Term Term2 Z=Z0 MLOC TL18 L=6.3 mm W=0.542011 mm MLSC TL20 L=3.3 mm W=0.542011 mm Term Term1 Z=Z0 MLIN TL11 L=2.3 mm W=0.542011 mm MLIN TL8 L=1.6 mm W=0.542011 mm MLIN TL14 L=2 mm W=0.542011 mm S2P SNP1 2 1 Re f MTEE_ADS Tee2 MTEE_ADS Tee1 MTEE_ADS Tee4 MTEE_ADS Tee3 R R1 R=Rstab MGAP Gap1 S=1.0 mm W=1.25 mm MSTEP Step2 W2=0.2 mm W1=width MLIN TL16 L=0.6 mm W=0.2 mm MLIN TL12 L=1 mm W=0.2 mm MSTEP Step1 W2=0.2 mm W1=width

Fig A.1: Design A

MLIN TL10 MLIN TL11 MLIN TL9 MLIN TL8 R R1 MLIN TL6 MLIN TL7 MLIN TL2 MLIN TL1 Term Term 2 Z=Z0 S2P SNP1 2 1 Ref Term Term 1 Z=Z0

Fig A.2: Design B

MLIN TL10 MLIN TL11 MLIN TL9 MLIN TL8 R R1 MLIN TL6 MLIN TL7 MLIN TL2 MLIN TL1 S2P SNP1 2 1 Ref Term Term 1 Z=Z0 Term Term 2 Z=Z0

Appendix B

PCB Design

Appendix B PCB Design

Fig B.3: IMN PCB design

Appendix C