Quai-Passive 5.8 GHz Front-End Design and Implementation for Vital Signs Detection

LiU-ITN-TEK-A--18/003--SE

Quai-Passive 5.8 GHz Front-End

Design and Implementation for

Vital Signs Detection

Henrik Kalvér

LiU-ITN-TEK-A--18/003--SE

Quai-Passive 5.8 GHz Front-End

Design and Implementation for

Vital Signs Detection

Examensarbete utfört i Elektroteknik

vid Tekniska högskolan vid

Linköpings universitet

Henrik Kalvér

Handledare Qin-Zhong Ye

Examinator Adriana Serban

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

http://www.ep.liu.se/

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.

According to intellectual property law the author has the right to be

mentioned when his/her work is accessed as described above and to be protected

against infringement.

For additional information about the Linköping University Electronic Press

and its procedures for publication and for assurance of document integrity,

please refer to its WWW home page:

http://www.ep.liu.se/

Abstract

This thesis investigates the possibility to measure vital signs, such as heartbeat and respiratory rate, by developing a RF-front end for wireless detection. The RF-front has been developed and manufactured as a continuous wave Doppler radar receiver, which utilizes quadrature demodulation by means of a multi-port correlator together with power detectors for down conversion. This thesis has been part of an ongoing research project at Link¨opings University, to develop a sensor platform for wireless vital signs detection. This sensor platform has been broken down into two major parts, a radar RF front-end system and a back-end digital signal processing system. The back-end system consist of data acquisition- and a processing-part.

It was shown that very low-frequency signals emulating vital signs can be de-tected, when direct frequency conversion and demodulation are performed with the multi-port detector. Due to the limitations of the instruments, 10-Hz signals were demonstrated.

Acknowledgements

I want to give thanks to Adriana Serban, Qin-Zhong Ye and Gustav Knutsson at the Department of Science and Technology at Link¨oping University for their guidance through interesting discussions and support in providing necessary components and equipment.

Special thanks to Tobias Pettersson at Link¨oping University for providing moti-vation when it was needed and a good sounding board.

Acronyms

Abbr. Description

ADC Analog to Digital Converter DSP Digital Signal Processor WPD Wilkinson Power Divider PCB Printed Circuit Board RF Radio Frequency LO Local Oscillator CW Continuous Wave EM Electromagnetic

Contents

1 Introduction 1 1.1 Background . . . 1 1.2 Goal . . . 1 1.3 Specification . . . 2 1.4 Method . . . 2 1.5 Delimitation . . . 2 2 Microwave front-end 3 2.1 Concept . . . 3 2.2 Multi-port correlator . . . 4

2.3 Wilkinson Power Divider . . . 8

2.4 Quadrature 90◦ _{Hybrid . . . . 9}

2.5 Power Detector . . . 12

2.5.1 Diode HSMS286 . . . 13

3 Simulation 15 3.1 Simulation platform for Vital-sign detection . . . 15

3.1.1 The Substrate . . . 16

3.2 Wilkinson Power Divider Design . . . 16

3.3 Quadrature 90◦ _{Hybrid Design . . . 17}

3.4 Multi-port Correlator Design . . . 19

3.5 Diode Power Detector Design . . . 20

3.5.1 Diode Models in ADS . . . 20

3.5.2 Matching network design . . . 22

3.5.3 The output filter . . . 22

3.5.4 Power detector simulation . . . 23

3.6 Multi-port correlator detector simulation on layout level . . . 26

4 Manufacture 29 4.1 Power Detector . . . 29

4.2 Multi-port correlator . . . 30

4.3 Multi-port correlator with detectors . . . 31

5 Result 33 5.1 Power detector . . . 33

5.2 Multi-port correlator . . . 36

6 Discussion 43

7 Conclusion 45

List of Figures

2.1 Transmitted and received signal.[6] . . . 4

2.2 Block diagram of a) the multi-port receiver and b) equivalent, classical receiver.[7] . . . 5

2.3 Block diagram of the multi-port receiver with parallel diode configuration.[7] 6 2.4 Block diagram of the multi-port receiver with anti-parallel diode configuration.[7] 6 2.5 Wilkinson Power Divider. [3] . . . 8

2.6 Wilkinson Power Divider. [3] . . . 8

2.7 Wilkinson Power Divider. [3] . . . 9

2.8 Model of the Quadrature 90◦ _{Hybrid. [3] . . . 10}

2.9 Quadrature 90◦ _{Hybrid in normalized form.[3] . . . 10}

2.10 Even- and odd-mode excitation.[3] . . . 11

2.11 Block diagram of a power detector circuit . . . 12

2.12 Input voltage vs. output power of HSMS286 diode.[8] . . . 13

3.1 Multi-port correlator system simulation testbench,[2]. . . 15

3.2 Design guide for passive components: Wilkinson power divider design at 5.8 GHz. . . 16

3.3 Wilkinson power divider – S-paramteres simulation results, schematic level. . . 16

3.4 Wilkinson power divider – layout-like symbol and layout simulation set-up. . . 17

3.5 Quadrature 90◦ _{Hybrid made with P assiveCircuit in ADS. . . 17}

3.6 Quadrature 90◦ _{Hybrid S-parameters simulation results. . . 18}

3.7 Quadrature 90◦ _{Hybrid EM component. . . 18}

3.8 Quadrature 90◦ _{Hybrid EM component S-parameters simulation} re-sult. . . 18

3.9 5.8 GHz multi-port correlator, the layout. . . 19

3.10 5.8 GHz multi-port correlator simulation results, layout level. . . 20

3.11 Quadrature 90◦ _{Hybrid EM component. . . 20}

3.12 Package model SOT-3230. . . . 21

3.13 Package model SOT-323C. . . 21

3.14 Matching for parallel diode HSMS286 circuit. . . 22

3.15 Matching for anti-parallel diode HSMS286 circuit. . . . 22

3.16 Lowpass-filter after the diode. . . . 23

3.18 Power detector with SOT323C. . . . 24

3.19 S-parameter simulation result of power detector with single diode. . . 24

3.20 S-parameter simulation result of power detector with anti-parallel HSMS286. . . . 24

3.21 Simulation of diode curve of the power detector with single diode. . 25

3.22 Simulation on Port 1 of the diode curve of the power detector with anti-parallel diodes. . . . 25

3.23 Simulation on Port 2 of the diode curve of the power detector with anti-parallel diodes. . . . 25

3.24 Mulit-port correlator with parallel diodes detectors. . . . 26

3.25 Mulit-port correlator with anti-parallel diodes detectors. . . . 26

3.26 I(t) and Q(t) signals for the parallel diode, multi-port detector. . . . 27

3.27 The multi-port correlator with parallel detectors circuit I- and Q-phase baseband signals. . . . 27

3.28 I- and Q-signal from the anti-parallel circuit. . . 28

3.29 The multi-port correlator with anti-parallel detectors circuit I- and Q-phase baseband signals. . . 28

4.1 Single diode power detector, module layout.. . . . 29

4.2 Anti-parallel diode power detector, module layout. . . . 29

4.3 Single diode power detector, manufactured module. . . 30

4.4 Anti-parallel diode power detector, manufactured module. . . . 30

4.5 Multi-port correlator, module layout. . . . 30

4.6 Multi-port correlator, manufactured module, top-view. . . . 31

4.7 Multi-port RF front-end with parallel diodes, module layout. . . . . 31

4.8 Multi-port RF front-end with anti-parallel diodes, module layout. . . 32

4.9 Multi-port RF front-end with parallel diodes, manufactured module layout, top-view. . . . 32

4.10 Multi-port RF front-end with anti-parallel diodes, manufactured mod-ule, top-view. . . 32

5.1 Input power sweep, 1-dB stepwise, single diode power detector. . . 34

5.2 Input power sweep, 1-dB stepwise, anti-parallel diode power detector. Port 1 active, Port 2 50 Ω terminated. . . 34

5.3 Input power sweep, 1-dB stepwise, anti-parallel diode power detector. Port 2 active, Port 1 50 Ω terminated. . . 35

5.4 Input power toggling between -10 dBm and -9.5 dBm, single diode power detector. . . 35

5.5 Input power toggling between -7 dBm and -6.5 dBm, anti-parallel diode power detector. . . 36

5.6 S-parameter measurement set-up for the multi-port correlator. . . 36

5.7 S-parameter measurement: S13, S14, S15 and S16 of the multi-port correlator. . . 37

5.8 S-parameter measurement: S12, isolation between port 1 and port 2. . 37

5.9 Measurement: spectrum of the demodulated 10 kHz I-signal. . . 38

5.11 Measurement, anti-parallel diodes: I and Q signals down-converted

at Low-IF of 3 kHz. Frequency modulation at 10 Hz, not visible in the photography. . . 40

5.12 . Measurement, anti-parallel diodes: I signal down-converted at

Low-IF of 3 kHz with the RC-filter discarded. . . 40

5.13 Measurement, parallel diodes: I and Q signals down-converted at

Low-IF of 3 kHz. Frequency modulation at 10 Hz, not visible in the photography. . . 41

5.14 Measurement, parallel diodes: I and Q signals down-converted at

Low-IF of 3 kHz with the RC-filter discarded. Frequency modulation at 10 Hz, not visible in the photography. . . 41

List of Tables

2.1 SPICE Parameters for HSMS286 . . . 14 3.1 Rogers4350B properties . . . 16 3.2 Diode (HSMS286) parameters . . . 21

Chapter 1

Introduction

1.1

Background

The Doppler vibration radar has been studied since the 1970s [1]. The Doppler effect can be used to detect in real-time, small periodic movements, including movements caused by heart and respiration beat, i.e., to detect vital signs of a human. Such radars use direct-conversion receiver architecture in either monostatic or bistatic radar configuration. The signal to be detected is extracted from the backscattered signal in form of a phase that differs when the received signal is “compared” to the transmitted signal.

In this project, a microwave passive technique is used to design, manufacture and measure a Doppler radar front-end prototype. The front-end relies on the multi-port technology, a technology mostly used for radio application at GHz frequency for high data rates communication. As compared to radio applications, the project challenge is to detect very weak and low frequencies signals corresponding to the vital signs.

The presented Master thesis project is a part of a research project at Link¨oping University, Campus Norrk¨oping founded by the Norrk¨oping Municipality. Two other master-level projects were conducted within this project frame, one dedicated to the design of the RF front-end and the other to the digital signal processing implemen-tation. The project presented in this report is intended to continue and push further the previous project so that first prototypes will be available.

1.2

Goal

The goal of this thesis has been to develop a RF-front end for the vital sign detector. This includes early simulations aswell as manufacturing and testing of the passive circuits. This include a system that receives, process and separates the signal into its quadrature components, I(t) and Q(t). Before going to the DSP, the signals passes through a detector and a lowpassfilter to filter the 5.8 GHz signal. The com-plete circuit including sub-circuits will be manufactured at Link¨opings Universitet, Campus Norrk¨oping, as a Printed Circuit Board (PCB).

1.3

Specification

The multi-port RF front-end will be designed at 5.8 GHz operation frequency. In-dividual modules will be manufactured and tested before the entire RF front-end for vital sign detection will realized. The substrate will be used is ROGERS4350B substrate.

1.4

Method

To complete the tasks and achieve the goals, the project will start with literature study. The know-how accumulated in previous research and Master thesis reports will be analyzed for further use and progression of the entire project. However, to assure a unitary project data base and consistent knowledge and design skills, all the components will be developed from zero. The design will be done in Advanced Design System (ADS) from Keysight Technologies (former Agilent). The microwave modules will be manufactured at Link¨oping University, Campus Norrk¨oping, making use of the PCB laboratory. For measurements, the instruments in the RF laboratory will be used.

1.5

Delimitation

The research project include a RF front-end system and a back-end signal processing system. This thesis will only handle the RF front-end system, with a close relation-ship to a student that is doing a thesis about the back-end system. The thesis will also not include the development of the 5.8 GHz antenna. Instead and antenna will be bought or used from previous work. In ADS a test-bench has already been developed so that will not be done in this thesis.[2]

Chapter 2

Microwave front-end

2.1

Concept

To be able to detect vital signs of a human, a continuous wave (CW) Doppler radar at 5.8 GHz is proposed. Figure 2.1 shows a generic description of how the CW Doppler radar works. The basic idea is to transmit a continuous wave signal, i.e., in this project a 5.8 GHz unmodulated signal. Reaching the target, the signal is back scattered and received by the radar device. The received signal is changed in its phase due to the propagation but also due to the periodic small movement of the tar-get, i.e., the received signal is now a modulated signal. The modulation appears as a change in the carrier frequency with a rate determined by the periodic movement of the heartbeat and respiration rate,[4]. To detect these low-frequency signals, the received RF signal must be demodulated while it is down-converted from the carrier frequency to either an intermediary frequency (IF) or to 0-frequency band. Due to the periodicity of the vital signals and due to the very sensitive phase variation with small movements, the low-frequency signal might be difficult to be detected as it passes through nulls located at every quarter wavelength, [5]. To avoid this problem, the quadrature demodulation is adopted in many Doppler radar topologies, i.e., the output signals are in-phase (I) and in quadrature-phase (Q) signals. The advantage of having the demodulated signal as two signals in quadrature, is that when one of either I(t) or Q(t) will be at a null point, the other will be at the optimal detection point.

In this project, the quadrature demodulation will be implemented by using the multi-port receiver front-end architecture. A multi-port receiver consists of one multi-port correlator and power detectors using zero-biased Schottky diodes. The two input signals are the backscattered signals (RF) and the local-oscillator signal (LO) and the outputs are the I(t) and Q(t) quadrature signals, usually in differential form. These output signals called raw data is the data to be further processed by the DSP block. The DSP will extract the phase information from I(t) and Q(t) signals and calculate the spectrum to identify the vital signs[6], signal processing unit (DSP) that performs operations upon raw data.

Figure 2.1: Transmitted and received signal.[6]

2.2

Multi-port correlator

The multi-port correlator is a passive circuit composed of several microwave pas-sive components such as Wilkinson Power Dividers and Quadrature 90◦ _hybrids.

The outputs from the multi-port correlator are linear (±) combination signals of controlled-phase shifted input signals. By using transistors or diodes connected to the output ports of the multi-port correlator, the multi-port can be used to perform a series of high frequency signal processing, e.g., direct-carrier complex quadrature modulation and quadrature down-conversion.

Figure 2.2a shows a block diagram over a multi-port correlator system. It can be assumed that the RF signal at P1 is modulated. This either purposely done or

by vibration or movement distance to a target, as in sensor- or radar applications. At port P2 the local oscillator (LO) is connected. The output of the multi-port

correlator are at ports 3-6 (P3− P6) and are connected to square-low (·)2 matched

power detectors. The output signals are linear, phase-shifted combinations of the RF and LO input signls, as can be deduced from Figure 2.2a. The output signals S3 to

S6 form ideally two pairs of differentials signals, i.e., S4,S3 and S6,S5 corresponding

to the in-phase (I+

and I−_{) and quadrature-phase signals (Q}+

and Q−_{), respectively.}

Differential amplifiers can be used, in order to obtain amplified, single-ended output quadrature signals I and Q.

The circuit can be easily changed to operate either in direct conversion or low intermediate frequency (IF) schemes. The equivalent, block diagram of a typical direct-conversion receiver is shown in Figure 2.2b.

Figure 2.2: Block diagram of a) the multi-port receiver and b) equivalent, classical

receiver.[7]

The inputs signals for a typical application are Equations 2.1 and 2.2:

SLO(t) = ALOcos(ω0t) (2.1)

SRF(t) = ARFcos(ω0t − ϕ(t)) (2.2)

For detection of vibrations or vital signs, the multi-port front-end is used in a Doppler radar configuration. Hence, the phase difference ϕ(t) in 2.2 can be directly expressed in the form:

ϕ(t) = kz = 2π

λ 2R(t) (2.3) R(t) = R0+ r(t) = R0+ a(t)cos(ωV St) (2.4)

Where a(t) is the amplitude of the respiration/heartbeat and ωV S is the angular

frequency of the vital sign.

Figure 2.3 and 2.4 shows the block-diagram of the multi-port correlator used in this thesis with different detector setups. Compared to each other, the multi-port correlator with anti-parallel diodes has the advantaged to give output signals that are single-ended and almost doubled in amplitude compared to the multi-port correlator with parallel diodes. The co-design of this thesis with the digital signal processing (DSP) thesis imposes single-ended signals as input for the analog-to-digital converter (ADC). Hence, the topology in Figure 2.4 is the suggested end result of the thesis.

Figure 2.3: Block diagram of the multi-port receiver with parallel diode

configuration.[7]

Figure 2.4: Block diagram of the multi-port receiver with anti-parallel diode

configuration.[7]

The following equations will show how the multi-port correlator with parallel detectors work. Equations 2.5 are for the multi-port correlator with parallel diode detectors. P3 : SRF(t)e−j180 ◦ + SLO(t)e−j360 ◦ (2.5a) P4 : SRF(t)e−j270 ◦ + SLO(t)e−j270 ◦ (2.5b) P5 : SRF(t)e−j180 ◦ + SLO(t)e−j270 ◦ (2.5c) P6 : SRF(t)e−j270 ◦ + SLO(t)e−j180 ◦ (2.5d) Equations 2.6 are for the multi-port correlator with anti-parallel detectors, the difference is the 90◦ _{path at Port 3 and Port 5.}

P3 : SRF(t)e−j270 ◦ + SLO(t)e−j90 ◦ (2.6a) P4 : SRF(t)e−j270 ◦ + SLO(t)e−j270 ◦ (2.6b) P5 : SRF(t)e−j270 ◦ + SLO(t)e−j360 ◦ (2.6c) P6 : SRF(t)e−j270 ◦ + SLO(t)e−j180 ◦ (2.6d) Using the notations

α = ω0t − φ(t) (2.7a)

β = ω0t (2.7b)

The RF and LO signals at the output ports for the multi-port correlator with parallel detectors are:

P3 : −ARFcosα + ALOcosβ (2.8a)

P4 : −ARFsinα − ALOsinβ (2.8b)

P5 : −ARFcosα − ALOsinβ (2.8c)

P6 : −ARFsinα − ALOcosβ (2.8d)

And RF and LO signals at the output ports for the multi-port correlator with anti-parallel detectors are:

P3 : −ARFsinα + ALOsinβ (2.9a)

P4 : −ARFsinα − ALOsinβ (2.9b)

P5 : −ARFsinα + ALOcosβ (2.9c)

P6 : −ARFsinα − ALOcosβ (2.9d)

After squaring, the relevant terms for the down-conversion for the multi-port correlator with parallel diodes are:

v2

3 = −2ARFALOcos(α)cos(β) (2.10a)

v2 4 = +2ARFALOsin(α)sin(β) (2.10b) v2 5 = +2ARFALOcos(α)sin(β) (2.10c) v2 6 = +2ARFALOsin(α)cos(β) (2.10d)

The relevant terms after squaring for the down-conversion for the multi-port correlator with anti-parallel diodes are:

v2

3 = −2ARFALOsin(α)sin(β) (2.11a)

v2 4 = +2ARFALOsin(α)sin(β) (2.11b) v2 5 = −2ARFALOsin(α)cos(β) (2.11c) v2 6 = +2ARFALOsin(α)cos(β) (2.11d)

After assuming ideal filtering and summing of the currents at the output nodes, Equations 2.10 results in:

i3(t) = −kARFALOcos(α − β) (2.12a)

i4(t) = +kARFALOcos(α − β) (2.12b)

i5(t) = −kARFALOsin(α − β) (2.12c)

And Equations 2.11 results in:

i3(t) = −kARFALOcos(α − β) (2.13a)

i4(t) = −kARFALOcos(α − β) (2.13b)

i5(t) = +kARFALOsin(α − β) (2.13c)

i6(t) = +kARFALOsin(α − β) (2.13d)

The I- and Q- baseband signals can then be written as:

SI(t) = −AIcos(φ(t)) (2.14a)

SQ(t) = +AQsin(φ(t)) (2.14b)

Where AI = AQ = 2kARFALO.

2.3

Wilkinson Power Divider

The Wilkinson Power Divider (WPD) is a three port network that will appear to be loss-less when the outputs are matched. The WPD can be made for both arbitrary power dividing of a signal as well as a 3 dB power split. Isolation between the output ports is achieved with a resistor 2Z0 that can be seen in Fig. 2.5. [3]

Figure 2.5: Wilkinson Power Divider. [3]

The WPD can have two different modes, even and odd mode, in this thesis only even mode WPD will be used. In Figure 2.6 even is given when Vg2 = Vg3 = 2V0

and odd mode is where Vg2 = −Vg3 = 2V0. For even mode, Vg2 = Vg3 = 2V0, give

that no current flows through r

2 resistors or the short circuitline at port 1.

With this the circuit can be simplified by bisecting the circuit in Figure 2.7.

Figure 2.7: Wilkinson Power Divider. [3] When looking into port 2, the impedance Ze

in = Z

2

2 will be found, because of the

transmission line looks like a quarter-wave transformer. If then Z =√2, port 2 will be matched. This give Ve

2 = V0 since Z_ine = 1, to find V1e transmission line equation

is needed[3]. If x = 0 at port 1 and x = −λ

4 at port 2, the equation will be Equation

2.15: V (x) = V+ (e−jβx_{+ Γe}jβx_{) (2.15)} This give V1 e and V 2 e as following: Ve 2 = V (− λ 4) = jV + (1 − Γ) = V0 (2.16) V1e = V (0) = V + (1 + Γ) = jV0 Γ + 1 Γ − 1 (2.17) The reflection coefficient is the one seen at port 1 looking at the resistors nor-malized as the value 2. With the reflection equation we get:

Γ = ZL− Z ZL+ Z = 2 − √ 2 2 +√2 (2.18) that gives Ve 1 = −jV0 √ 2.

2.4

Quadrature 90

◦

Hybrid

The Quadrature 90◦ _{hybrid is a four-port 3 dB directional coupler with a 90}◦ _phase

shift between the output ports. Figure 2.8 shows a model of the Quadrature hybrid, the power from input 1 is divided between port 2 and 3, while port 4 is isolated when all ports are matched. [3]

Figure 2.8: Model of the Quadrature 90◦ _{Hybrid. [3]}

The Quadrature 90◦ _{hybrid is also known as a branch-line hybrid. It is usually}

made with microstrip lines and is attractive for its simple design and planar struc-ture. Equation 2.24 shows the scattering matrix. Because of the symmetry of the Quadrature hybrid any port can be used as input, this give that the output ports will always be on the opposite side of the input port.

S = −1√ 2     0 j 1 0 j 0 0 1 1 0 0 j 0 1 j 0     (2.19) Figure 2.9 shows a Quadrature 90◦ _{Hybrid in normalize form, here each line}

represent a transmission line with indicated characteristic impedance normalized to Z0. All grounds are connected to a common ground and amplitude of the incoming

signal A1 = 1 is incident at port 1.

Figure 2.9: Quadrature 90◦ _{Hybrid in normalized form.[3]}

Similarly to the WPD, the Quadrature hybrid can be decomposed into the su-perpostion of an even-mode excitation and odd-mode excitation. If the circuit in Figure 2.9 is decomposed it will result in the circuits in Figure 2.10. These are even-mode excitation (a) and odd-mode excitation (b). These two circuits can then be decomposed into two-port circuits, as seen in Figure 2.10 to the right.

Figure 2.10: Even- and odd-mode excitation.[3] Because each incident wave have an amplitude of ±1

2, the amplitude can be

expressed as Equations 2.20-2.23 on each port. S(1, 1) = 1 2Γe+ 1 2Γo (2.20) S(2, 1) = 1 2Te+ 1 2To (2.21) S(3, 1) = 1 2Te− 1 2To (2.22) S(4, 1) = 1 2Γe− 1 2Γo (2.23) Where Te,o are the transmission coefficient and Γe,o are the reflection coefficient

for the even- and odd-mode. To confirm Equation 2.24 , a calculation of Te,o and

Γe,o can be done by multiplying ABCD matrices of each cascade component. The

calculation of the even-mode circuit will look like: A B C D  e = 1 0 j 1  Y = j  0 √j 2 j√2 0  λ 4 1 0 j 1  Y = J = √1 2 −1 j j −1  (2.24) Where the admittance of the shunt open-circuit λ

8 stubs is Y = jtanβl = j.

Values for all the individual matrices can be found in table 4.1[3], this can be used with Table 4.2[3] to convert the ABCD parameters to S parameters, which are equivalent with the transmission and reflection coefficients and will give,

Γe= A + B − C − D A + B + C + D = −1+j−j+1 √ 2 −1+j+j−1 √ 2 = 0 (2.25)

Te= 2 A + B + C + D = 2 −1+j+j−1_√ 2 = −1√ 2(1 + j). (2.26) With same use of Equation 2.24 but for odd-mode circuit will lead too Equation 2.27, A B C D  o = √1 2 1 j j 1  (2.27) which gives the transmission and reflection coefficients as,

Γo = 0 (2.28)

To =

1 √

2(1 − j). (2.29) Using the results from Equations 2.25, 2.26, 2.28 and 2.29 will give:

S(1, 1) = 0 (2.30) S(2, 1) = −√j 2 (2.31) S(3, 1) = −√1 2 (2.32) S(4, 1) = 0. (2.33) Which agrees with the scattering matrix in Equation 2.19.

2.5

Power Detector

Figure 2.11 shows a block diagram of a power detector circuit, which is split into three blocks. First of is the matching network that is a crucial part when handling higher frequencies, second is the diode circuit that is the controlling part, what operations the circuit is supposed to do, and last a filter is applied to take away unwanted frequencies.[3]

Figure 2.11: Block diagram of a power detector circuit

The classical pn−junction diode commonly used at low frequencies cannot be im-plemented in high frequency applications, due to large junction capacitance. How-ever for high frequencies, Schottky diodes are used instead because it relies on a

semiconductor-metal junction and that results in a much lower junction capaci-tance. A junction diode can be expressed as nonlinear resistor with a small-signal V − I relationship as described in Equation 2.34:

I(V ) = Is(eαV − 1) (2.34)

Where α = _nkTq , q is the charge of an electron, k is Boltzmann’s constant, T is temperature, n is the ideality factor and Is is the saturation current. Schottky

Diodes primary application is in frequency conversion of input signals. The usage for a Schottky diode in this thesis is for the frequency conversion operation detection (demodulation of an modulated signal).

In detector circuits the diode is used to demodulate and amplitude-modulated the RF carrier. The Equations 2.10 and 2.11 describes the ”square-law behavior” of a diode. This is a region of the diode curve that give voltage output square compared to the input power. This region only has a certain range and if the input power is too low the signal will disappear in noise and if the input power is too high the signal will either be linear or saturated.

2.5.1

Diode HSMS286

The HSMS286 diode from AvagoT echnologies is a DC biased detector Schottky diode that is optimized from 915 MHz to 5.8 GHz. The diode is designed for applications that requires small signal detection (Pin < −20dBm) and are used

without bias current for frequencies over 4 GHz. Figure 2.12 shows the output voltage vs. input power for the HSMS286 diode from -50 dBm to 0 dBm.[8]

Figure 2.12: Input voltage vs. output power of HSMS286 diode.[8]

To be able to realized the diode the manufacture (Avago Technologies) give out SPICE parameters, which can be seen in Table 2.1.

Table 2.1: SPICE Parameters for HSMS286

Parameter Units Value

B_{v V 7.0} CJ O pF 0.18 EG eV 0.69 IBV A 1E-5 IS A 5E-8 N _1.08 RS 6.0 P_{B(V J) V 0.65} PT(XT I) 2 M _0.5

HSMS286 are delivered in different packages. In this thesis package SOT-3230 for power detectors with single diode and SOT-323C for power detectors with anti-parallel diodes, was used.

Chapter 3

Simulation

3.1

Simulation platform for Vital-sign detection

First simulations of the new vital sign detector, i.e., the new six-port correlator and the new power detectors were performed by using a simulation template in ADS. This template was previously developed as a part of the research project and it is shown in Figure 3.1. The template includes the model of the transmitter. It mainly consists of LO signal source at 5.8 GHz. The LO signal is divided by a power splitter so that a part is sent towards the target via one transmit antenna, and the other part is used (coupled) as the LO signal at one input of the detector. To model the radar operation, the sent LO signal is modulated by two low-frequency signals corresponding to the heart and respiration frequency. The received signal is amplified by two low-noise amplifiers (LNAs) before coming into the multi-port correlator detector system,[2].

PhaseModNoise

Receiver

Transmitter

CW Doppler Radar Architecture for Non-Contact Vital Signs Detection (NCVSD) 5.8 GHz

Channel Model diode Rx1 Tx_Signal TxRx V4 V3 I_BB Rx_Signal I Rx Rx3 Tx Q V6 Q_BB Tx1 Rx2 Txx TxLO Signal_Generator Tx3 V5 vpa vpb Attenuator PM_ModTuned VtSine PhaseShiftSML Amplifier2 Amplifier2 OpAmp Amplifier2 PwrComb C Antenna_Model_Rx R VtSine Attenuator R Attenuator emModel V_Noise P_1Tone MatchingNetwork_ShortCircuitStub2 MatchingNetwork_ShortCircuitStub2 MatchingNetwork_ShortCircuitStub2 emModel MatchingNetwork_ShortCircuitStub2 PhaseNoiseMod PwrSplit2 emModel DiodeModel_HSMS286Y PwrSplitComp Antenna_Model_Tx DiodeModel_HSMS286Y RadialStub_Circuit2_GND_v1 Attenuator R C Term DiodeModel_HSMS286Y DiodeModel_HSMS286Y Vf_Square SampleHoldSML SampleHoldSML R R Term Term PwrComb PwrSplitComp OpAmp RadialStub_Circuit2_v1 RadialStub_Circuit2_v1 emModel emModel RadialStub_Circuit2_v1 RadialStub_Circuit2_v1 emModel SixPort_Circuit RadialStub_Circuit2_GND_v1 emModel emModel RadialStub_Circuit2_GND_v1 emModel emModel emModel emModel emModel RadialStub_Circuit2_GND_v1 C C ATTEN1 MOD1 Respiration-induced PS3 Power_Amplifier LNA1 AMP6 LNA2 I__58 C42 I__66 R18 Heartbeat-induced ISL_Tx-RX R9 ISL_Tx-LO I__140 SRC5 PORT1 I__33 I__128 I__132 MOD2 PWR3 I__133 I__134 I__57 I__7 I__135 I__136 ATTEN2 R17 C30 Term3 I__0 I__129 SRC4 SAMP1 SAMP2 R12 R19 Term1 Term2 I__30 I__56 AMP8 I__75 I__137 I__130 I__97 I__131 I__138 I__139 C32 C39 VSWR=1 Loss=Atten4 dB Rout=50 Ohm Fnom=RFfreq Sensitivity=180/pi Phase=0 Damping=0 Delay=0 nsec Freq=0.5 Hz Amplitude=V_Resp V Vdc=0 V ZRef=50. Ohm Phase=Dist_PhShift Psat=29 TOI=37 NF=5.5 dB S21=dbpolar(14.39,-21.68) GainComp=1.0 dB GainCompPower=16 Psat=17.5 TOI=28 NF=1.8 dB S21=dbpolar(29.41,-56.43) VCC=3 V VEE=-3 V Zero1= Pole1= BW=0.75 MHz VOS=0 V IOS=0 A SlewRate=1e+6 CCom=0 F RCom=1 MOhm CDiff=0 F RDiff=1 MOhm Rout=50 Ohm CMR= Gain=10 dB GainComp=1.0 dB GainCompPower=16 Psat=17.5 TOI=28 NF=1.8 dB S21=dbpolar(29.41,-56.43) C=0.016 uF R=1 MOhm Phase=0 Damping=0 Delay=0 nsec Freq=1 Hz Amplitude=V_Heart V Vdc=0 V VSWR=1 Loss=ISL_TxRx dB R=100 Ohm VSWR=1 Loss=ISL_TxLO dB V_Noise=1 uV Freq=RFfreq P=polar(dbmtow(Pin),0) Z=50 Ohm Num=1 QL=50 NF=3 dB Fcorner=1 kHz Rout=50 Ohm Fnom=RFfreq S31=0.707 S21=0.707 VSWR=1 Loss=Atten_Tx dB R=1 MOhm C=0.016 uF Z=50 Ohm Num=3 Harmonics=16 Weight=no Delay=8.2 usec Fall=0.1 nsec Rise=0.1 nsec Freq=10 Hz Vdc=0 V Vpeak=1 V R=1 MOhm R=1 MOhm Z=50 Ohm Num=5 Z=50 Ohm Num=2 VCC=3 V VEE=-3 V Zero1= Pole1= BW=0.75 MHz VOS=0 V IOS=0 A SlewRate=1e+6 CCom=0 F RCom=1 MOhm CDiff=0 F RDiff=1 MOhm Rout=50 Ohm CMR= Gain=10 dB C=0.016 uF C=0.016 uF

3.1.1

The Substrate

The Rogers4350B substrate was used for both the simulations and the manufacturing of the PCBs. Table 3.1 shows the properties of the substrate.

Table 3.1: Rogers4350B properties

Material Properties Dielectric Thickness 0.508 mm Dielectric Constant 3.66 Dissipation Factor 0.004 Metal Thickness 0.035 mm Metal Conductivity 5.8×107 S/m Surface Roughness 0.001 mm

3.2

Wilkinson Power Divider Design

The Wilkinson power divider (WPD) was designed in ADS for the operation fre-quency 5.8 GHz and for the substrate Rogers 4350B. The ADS design guide for microwave passive components was used, as illustrated in Figure 3.2.

Figure 3.2: Design guide for passive components: Wilkinson power divider design

at 5.8 GHz.

To determine if it was matched for 5.8 GHz and it had a -3 dB loss a S-parameter simulation was done. Figure 3.3 shows S-parameters simulation results on schematic level. Port 2 and Port 3 have a loss of -3.061 dB (S(2,1)=S(3,1)=-3.061 dB) and at Port 1 there is no reflection back (S(1,1)=-37.648 dB).

Figure 3.3: Wilkinson power divider – S-paramteres simulation results, schematic

Once the WPD was designed on schematic level by means of the Passive Circuit design guide, the corresponding layout was generated and simulated. Electromag-netic (EM) simulations were performed in Momentum and a layout-like symbol was generated for the WPD, as shown in Figure 3.4.

Figure 3.4: Wilkinson power divider – layout-like symbol and layout simulation

set-up.

3.3

Quadrature 90

◦

Hybrid Design

The Quadrature 90◦_{Hybrid was designed in ADS following the same methodology as}

that used for the WPD. The design guide was firstly used to generate the schematic level, shown in Figure 3.5.

Figure 3.5: Quadrature 90◦ _{Hybrid made with P assiveCircuit in ADS.}

S-parameters simulations were performed in ADS to see if the Quadrature 90◦

Hybrid was matched for 5.8 GHz and if it gave a -3 dB loss for port 2 and port 3, also so port 4 was isolated. Figure 3.6 shows the result from the S-parameter simulation. Port 2 has a loss of -2.993 dB (S(2,1)=-2.993 dB), Port 3 has a loss of -3.222 dB (S(3,1)=-3.222 dB), Port 4 is isolated from Port 1 (S(4,1)=-39.209 dB) and Port 1 has no reflection back (S(1,1)=-33.074 dB).

Figure 3.6: Quadrature 90◦ _{Hybrid S-parameters simulation results.}

Based on the schematic of the hybrid, the layout was generated and simulated in Momentum. The layout-like symbol of the 90◦ _{quadrature component is shown in}

Figure 3.7. It indicates that this component was simulated with the electromagnetic solver in Momentum, for the S-parameters.

Figure 3.7: Quadrature 90◦ _{Hybrid EM component.}

Simulation results are shown in Figure 3.8. Port 2 has a loss of -2.716 dB (S(2,1)=-2.716 dB), Port 3 has a loss of -3.542 dB (S(3,1)=-3.542 dB), Port 4 is isolated from Port 1 (S(4,1)=-33.931 dB) and Port 1 has no reflection back (S(1,1)=-33.931 dB).

3.4

Multi-port Correlator Design

As described in Section 2.2, the multi-port correlator consists of one Wilkinson power divider and three 90◦ _{hybrid couplers. It represents the core of the microwave}

processing that mainly controls the phase and amplitudes of the signals confined in the passive network. It is ideally assumed that the output signals I and Q have the same attenuation and a phase difference of 90◦_{. To realize this, the symmetry of the}

layout along the x- and y-axes are of great importance. In this project, the layout of the correlator was directly generated from the layout of the individual passive components. Figure 3.9 shows the final layout of the multi-port correlator. Some effort was invested in appropriate compactness and shape to fit future RF-module block.

Figure 3.9: 5.8 GHz multi-port correlator, the layout.

Simulation results of the multi-port correlator are shown in Figure 3.10. The port assignment is the same as that used in Chapter 2. As can be seen, Port 1 has no reflection back (S(1,1) = -27.722 dB), Port 2 and Port 1 are isolated from each other (S(2,1) = -30.101 dB), due to Port 1 and Port 4 in Quadrature 90◦ _Hybrid.

Port 3 and Port 6 has a -5.927 dB loss which is good compared to the theory were it should be -6 dB. Port 4 and Port 5 has a -6.768 dB loss which can be considered good. Port 7 is also isolated (S(7,1) = -29.915 dB) from Port 1 for the same reason as Port 1 and Port 2 are isolated.

Figure 3.11, the phase difference between Port 3 and Port 4, and between Port 5 and Port 6 are evaluated. It can be seen that the phase difference is very close to 180◦ _{phase difference, verifying that the multi-port correlator process the signals}

Figure 3.10: 5.8 GHz multi-port correlator simulation results, layout level.

Figure 3.11: Quadrature 90◦ _{Hybrid EM component.}

3.5

Diode Power Detector Design

As described in Section 2.5, the power detectors play an important role in detect-ing real-world, weak signals at very low frequency. Therefore, the design of these detectors is a key factor for the performance of the entire vital sign detector using multi-port technique. In this Section, the design of the power detectors is presented. Two power detectors topologies are designed, called “parallel diode” power detec-tor and anti-parallel diode power detecdetec-tor. The design of any of these detecdetec-tors includes the diode model generation in ADS, the design of the matching networks and the simulation of the power detectors using the layout level components and the accurate model of the diodes.

3.5.1

Diode Models in ADS

The diode HSMS286 that was used came in two different packages, depending on if it would be with the anti-parallel or with the parallel vital sign detection circuit. In Table 3.2 the parameters for the diode model can be seen. Because different packages was needed, SOT-3230 and SOT-323C were used.

Table 3.2: Diode (HSMS286) parameters

Parameter Units Values

B_{V V 7.0} CJ O pF 0.18 EG EV 0.69 IBV A 1E-5 IS A 5E-8 N _{None 1.08} RS Ω 6.0 P_{B(V J) V 0.65} PT(XT I) None 2 M _{None 0.5}

The diode package SOT-3230 was used for the multi-port correlator with parallel power detectors in ADS and can be seen in Figure 3.12.

Figure 3.12: Package model SOT-3230.

The diode package SOT-323C was used for the multi-port correlator with anti-parallel power detectors in ADS and can be seen in Figure 3.13.

3.5.2

Matching network design

As the amplitudes of the incoming signals are expected to be very small, matching networks at the input of the power detectors are used for both power detectors. In this way, loss of input power through reflections are avoided. Generation of the matching networks detectors was done at 5.8 GHz in ADS using the Smith chart tool. The matching networks are shown in Figure 3.14 and 3.15.

Figure 3.14: Matching for parallel diode HSMS286 circuit.

Figure 3.15: Matching for anti-parallel diode HSMS286 circuit.

3.5.3

The output filter

At the output of the power detectors, the diode current is transformed into a voltage signal, while the remains of AC frequencies and the possible leakage of the 5.8 GHz LO signal have to be low-pass filtered. Hence, a parallel RC circuit is added at the output of the diode. At the input, a 5.8 GHz notch filter is implemented as a

quarter-wave radial stub, as shown in Figure 3.16. The capacitor is 16 pF and the resistor is 1 MΩ.

Figure 3.16: Lowpass-filter after the diode.

3.5.4

Power detector simulation

The final power detectors schematics are shown in Figures 3.17 and 3.18. In Figure 3.17, one single diode in SOT3230 package with its matching network and the output filter is shown. The matching network is implemented with microstrip transmission lines with a shorted stub to permit the diode DC current path. In Figure 3.18 the anti-parallel diode power detector is shown, including the matching networks, the model of the two-diodes in a SOT323C package and the output filter.

Figure 3.18: Power detector with SOT323C.

Both configurations were at simulated for S-parameters to verify the 50 Ω match-ing at the input ports. From Figures 3.19 and 3.20 it can be seen that both the parallel diode and the anti-parallel diode power detectors are well matched at 5.8 GHz.

Figure 3.19: S-parameter simulation result of power detector with single diode.

Figure 3.20: S-parameter simulation result of power detector with anti-parallel

HSMS286.

To evaluate the sensitivity of the power detectors, the diode transfer character-istic was simulated. Simulation results are shown in Figures 3.21 and 3.22, for the parallel and anti-parallel diode power detector topologies, respectively. The input power was swept from -20 dBm to 35 dBm. The saturation point hits at around 32 dBm and about 2 V.

Figure 3.21: Simulation of diode curve of the power detector with single diode. Figure 3.22 displays the curve of the power detector with anti-parallel diodes when the sweep input is on Port 1 and Port 2 is shorted. It hits saturation point at 35 dBm at an amplitude of 1.2 V.

Figure 3.22: Simulation on Port 1 of the diode curve of the power detector with

anti-parallel diodes.

Figure 3.23 displays the curve of the power detector with anti-parallel diodes when the sweep input is on Port 2 and Port 1 is shorted. It hits saturation point at 35 dBm at an amplitude of -1.18 V.

Figure 3.23: Simulation on Port 2 of the diode curve of the power detector with

3.6

Multi-port correlator detector simulation on

layout level

The new multi-port vital sign detectors are shown in Figures 3.24 and 3.25. They include the multi-port correlator and the four power detectors, with parallel and with anti-parallel diode configurations, respectively. To evaluate the new designs, the simulation platform for vital sign detection was used as described in Section 3.1. Main signals to be detected are the heart beat at 1 Hz frequency and 0.5 Hz for the respiration rate. Through simulations, the length of the transmission lines through which the LO signal is supplied were optimized for coherent detection of the vital signs. Without this measure Equations 2.5-2.13 from Chapter 2 would not be satisfied.

Figure 3.24: Mulit-port correlator with parallel diodes detectors.

Figure 3.25: Mulit-port correlator with anti-parallel diodes detectors.

Figure 3.26 shows the result of the I- and Q-signals from the parallel circuit, were the I is the blue line and Q is the red line. To get both I−_{and I}+_{as an I-signal,}

an op-amplifier is used with 10 dB gain, same with the Q-signal. The I-signal has an peak to peak of 250 µV and the Q-signal has an peak to peak of 183 µV.

Figure 3.26: I(t) and Q(t) signals for the parallel diode, multi-port detector. In Figure 3.27, the spectrum of the I and Q signals is shown for the parallel-diode configuration. It can be seen that the two frequency components corresponding to the heartbeat and respiratory signals are visible in the spectrum. The respiratory signal has a power around -70 dBm for both the I and Q components and the heart signal has a power of -96.46 dBm for the I-signal is and -92.10 dBm for the Q-signal. These values indicate that the raw data must be amplified before the ADC processing, [6].

Figure 3.27: The multi-port correlator with parallel detectors circuit I- and

Q-phase baseband signals.

Figure 3.28 shows the results of the I- and Q-signals from the anti-parallel cir-cuit, were the I-signal is the red line and Q-signal is the blue line. This circuit does not need an op-amplifier to make the I−_{- and I}+

-signal to merge, because of the anti-parallel diodes. The I-signal has an peak to peak of 172 µV and the Q-signal has an peak to peak of 159 µV, which is almost as good as the parallel circuit with an op-amplifier. As expected, the peak-to-peak amplitude of the single-ended I and Q signals is higher as the differential signals are constructively added after inverting one of the components.

In Figure 3.29. the spectrum of the I and Q signals is shown for the parallel-diode configuration. It can be seen that the two frequency components corresponding to the heartbeat and respiratory signals are well visible in the spectrum above the noise floor. The respiratory signal has a power around -72.5 dBm for both the I and Q components and the heart signal has a power of -99.5 dBm for the I-signal is and -94.61 dBm for the Q-signal. These values indicate that the raw data must

be amplified before the ADC processing, [6]. To be noted that these values are the simulated values obtained without any amplification after the power detectors.

Figure 3.28: I- and Q-signal from the anti-parallel circuit.

Figure 3.29: The multi-port correlator with anti-parallel detectors circuit I- and

Chapter 4

Manufacture

In this Chapter, the manufactured microwave modules are shortly presented. After the design of each component in ADS, the layout information is transformed into Gerber files ready to be used in the printed circuit board (PCB) laboratory. The entire manufacturing process, including soldering of the components has taken place at PCB laboratory at Link¨oping University, Campus Norrk¨oping.

4.1

Power Detector

In Figure 4.1, the layout of the single diode power detector and in Figure 4.2, the layout of the anti-parallel diode power detector are shown. These microwave modules were designed for individual measurements, before the manufacturing of the entire multi-port module.

Figure 4.1: Single diode power detector, module layout..

In Figure 4.3 and Figure 4.4 the manufactured power detectors are shown in top-view. The back side of the modules is a common ground plane connected to the top ground plane by via holes. The radial stub, the matching networks and soldered components can be seen in Figure 4.3 and 4.4.

Figure 4.3: Single diode power detector, manufactured module.

Figure 4.4: Anti-parallel diode power detector, manufactured module.

4.2

Multi-port correlator

In Figure 4.5, the layout of the multi-port correlator is shown. Care was taken to the symmetry of all connections between the Wilkinson power divider and the 90 hybrid couplers. Also the distances between the output ports was determined to permit the soldering of the SMA contacts. The final multi-port correlator is shown in Figure 4.6 in top-view.

Figure 4.6: Multi-port correlator, manufactured module, top-view.

4.3

Multi-port correlator with detectors

Finally, two multi-port detectors for vital sign detection were manufactured, i.e., one with parallel diode power detectors and the other with anti-parallel diode power detectors. The final layouts of the two modules are shown in Figure 4.7 and Figure 4.8 respectively.

Figure 4.8: Multi-port RF front-end with anti-parallel diodes, module layout. The manufactured modules with all components soldered are shown in Figures 4.9 and 4.10, respectively.

Figure 4.9: Multi-port RF front-end with parallel diodes, manufactured module

layout, top-view.

Figure 4.10: Multi-port RF front-end with anti-parallel diodes, manufactured

module, top-view.

All the steps involved in this phase of the project are of importance for the final measurement results. As an example, poor soldered SMA contacts or compo-nents can give unexpected large losses or even malfunction that will effect the entire interpretation of the results.

Chapter 5

Result

To evaluate the RF front-end part of the multi-port vital sign detector, several measurements were performed. In this Chapter, the measurement set-up and the measurement results are presented and discussed. At first, the power detectors and the multi-port correlator were individually tested. Then, the entire RF front-end detector was measured. Measurement scenarios were defined to investigate the correct operation and to determine some detector parameters. These experiments were done within the limits permitted by the existing instruments at the radio frequency (RF) laboratory. The used instruments are:

• Agilent E8267D Vector Signal Generator [9]

• Rhode&Schwarz Signal Generator SMIQ 06B [10] [11] • Rhode&Schwarz Vector Network Analyzer ZVM [12] • LeCroy 9310AM Oscilloscope [13]

• Agilent E4407B ESA-E Series Spectrum Analyzer [14]

5.1

Power detector

The manufactured power detectors shown in Figure 4.3 and 4.4 were tested. These were 1) single-diode power detector supposed to be used as parallel diode power detectors and 2) the anti-parallel power detectors. Rhode&Schwarz Signal Generator SMIQ 06B [9] was used as a source of variable input power and the response of the diode was measured with LeCroy oscilloscope. For all power detectors, the input power was swept from -15 dBm to 10 dBm with a power-step of 1 dB and a time step of 15 ms.

In Figure 5.1, measurement results of single-diode power detector are shown. The curve shows the output voltage in time, when the input power is increased every 15 ms with 1 dB. The oscilloscope has 50 ms per square on the horizontal and 40 mV per square on the vertical axis.

Figure 5.1: Input power sweep, 1-dB stepwise, single diode power detector. When doing measurements on the detector with anti-parallel diodes one input was terminated with 50 Ω impedance while the other had the signals coming into it. In Figure 5.2 the measurement result can be seen from having the signal generator on Port 1, the saturation point is now lower because there was power going through the other diode back to Port 2. The saturation point is at 93.1 mV and the curve is recognized as a diode curve. Comparing this to the simulation, it gives lower result but have a similar curve.

Figure 5.2: Input power sweep, 1-dB stepwise, anti-parallel diode power detector.

Port 1 active, Port 2 50 Ω terminated.

The same measurement was done with Port 2 and now Port 1 was terminated with 50 Ω impedance. Figure 5.3 shows the measurement result from Port 2, the curve is negative and this is because these diodes reaction on negative pulses. The saturation points is at -116.2 mV and also have a recognizable diode curve. This also gives a lower result then the simulations, but the curve can still be recognized when compared.

Figure 5.3: Input power sweep, 1-dB stepwise, anti-parallel diode power detector.

Port 2 active, Port 1 50 Ω terminated.

A second measurement scenario for the power detectors was aimed to emulate the very weak vital signs signals as periodic signals at the input of the power detec-tors. Through measurements, the smallest input power level and input power level difference that still can be detected were determined. It was found that the lowest input power difference that can be detected was between -10 dBm and -9.5 dBm input power, that corresponds to a 6 mV amplitude difference at the input. At the output, the measured amplitude of the detected signal is 1.9 mV, as shown in 5.4.

Figure 5.4: Input power toggling between -10 dBm and -9.5 dBm, single diode

power detector.

For the anti-parallel power detectors, the same test was performed, but for one diode of two, at a time. Measurements shown in Figure 5.5 indicate an output signal with an amplitude of 1.2 mV, while the input signal switches between -7 dBm and -6.5 dBm. However, due to the way of operation of the anti-parallel diodes, the effective amplitude at the output of the detector will be twice the 1.2 mV, hence around 2.4 mV. To conclude, all deigned power detectors work properly and confirm simulation results presented in Chapter 3.

Figure 5.5: Input power toggling between -7 dBm and -6.5 dBm, anti-parallel diode

power detector.

5.2

Multi-port correlator

As described in Chapter 2 and Chapter 3, the multi-port correlator is a passive mi-crowave component designed for 5.8 GHz frequency operation and implemented with microstrip transmission lines on ROGERS4350B substrate. To evaluate the quality of the correlator design, S-parameters measurements have to be performed. For this, Rhode&Schwarz Vector Network Analyzer ZVM [12] was used. The photography illustrating the measurement set-up is shown in Figure 5.6.

Figure 5.6: S-parameter measurement set-up for the multi-port correlator. The measurements of the transmission coefficients between port 1 and the ports 3 to 6 are shown in Figure 5.7. It can be seen that losses at the center frequency of 5.8 GHz are between -6.5 dB to -6 dB for all S1j, j = 3 to 6, values that are close to

Figure 5.7: S-parameter measurement: S13, S14, S15 and S16 of the multi-port

correlator.

The measurement result for the isolation between port 1 and 2 is -19 dB at 5.8 GHz, as shown in Figure 5.8. The measured isolation does not reach the good value predicted by simulation results shown in Figure 3.10, but still it is good enough to consider the ports isolated.

Figure 5.8: S-parameter measurement: S12, isolation between port 1 and port 2.

5.3

Multi-port correlator with detectors

First measurements of the entire multi-port detector were performed in the RF laboratory at Link¨oping University, Campus Norrk¨oping. The tests were designed within the possibilities and limits of the measurement equipment to the date of the experiments. Main test scenario is summarized as follows:

• Generation of a frequency modulated (FM) signal with the carrier at 5.8 GHz and with the modulating signal at the lowest frequency value that can be generated and measured in the laboratory. This is the RF modulated signal, emulating the Doppler radar backscattered signal.

• Generation of the local oscillator signal, the LO signal. • Test with spectrum analyzer of the RF and LO signals. • Measurement of the parallel diode multi-port detector. • Measurement of the anti-parallel diode multi-port detector.

Expected results are the directly down-converted, low-frequency I and Q signals at ports 3, 4 and 5, 6. As shown in Chapter 2, the I and Q spectrum content corresponds to the vital signs, i.e., to signals of very low frequency. These results are expected under the assumption that the LO and RF signals have the same frequency, i.e., 5.8 GHz, as shown in Chapter 2 and indicated by Equations 2.1 and 2.2.

The RF modulated signal was generated using the Rhode&Schwarz Signal Gen-erator SMIQ 06B [10], [11]. Frequency-domain, i.e., spectrum measurements were performed at the outputs of the multi-port detector using Agilent E4407B ESA-E Series Spectrum Analyzer. Time-domain measurements were performed using LeCroy 9310AM oscilloscope.

Frequency-Domain Measurements

In Figure 5.9, the spectrum of the demodulated signal at the I output port is shown. As the minimum frequency that could be detected with the spectrum analyzer was 9 kHz, the 5.8 GHz carrier was frequency modulated with 10 kHz, frequency com-ponent that can be seen in Figure 5.9. The LO signal input power was 0 dBm and the RF-signal input power was 0 dBm and the detected signal has a power of -80 dBm.

Figure 5.9: Measurement: spectrum of the demodulated 10 kHz I-signal. In Figure 5.10, the same measurement was performed, this time with the mod-ulating signal of 100 kHz frequency.

Figure 5.10: Measurement: spectrum of the demodulated 100 kHz I-signal. Both experiments to determine the spectrum at the I and Q outputs, proof that the multi-port detector performs as expected, i.e., it down-converts the RF signal from 5.8 GHz to the very low frequency of 10 kHz. The detected spectrum is at much higher frequency than those of the heart and respiration beat, but this due to the limitation of measurement equipment.

Time-Domain Measurements

To verify the detector for lower frequency than those permitted by the spectrum analyzer, time-domain measurements using an oscilloscope were performed. The modulating frequency is 10 Hz, which is within the expected range of frequencies for vital signs.

A first conclusion is that the measurements results have shown that the 10 Hz signal can be detected and that the multi-port detector operates correctly. However, this conclusion cannot be proofed in this report by publishing a “static” photography of the oscilloscope, situation that is explained in the following text.

The main problem was that the RF carrier frequency was not equal with the LO frequency. This can be easily explained by the fact that the two signals were generated by two different instruments.

After iterations, the minimum frequency difference between the carrier frequency of the RF signal and the LO signal that could be achieved was around 3 kHz. Theoretically, this can be interpreted as signal of frequency of 3 kHz modulated in frequency by a 10 Hz signal. The signal present at the outputs of the multi-port detector is of the form “low-IF with FM”. This signal could be clearly observed on the display of the oscilloscope, as an in-time varying signal with its period modulated by the 10 Hz signal.

A second problem was that, even if for a moment the RF and LO signals could have the same frequency, the frequencies drifted soon apart, the warmer the equip-ment got. At the end, it was impossible to manually get two perfect equal in frequency signals.

In Figure 5.11, the I and Q output signals measured with the oscilloscope are shown for the anti-parallel diode multi-port detector. The signals have the residual 3 kHz frequency, i.e., the low-IF signal but during the measurements the 10 Hz low frequency modulation could be observed on the oscilloscope display. From Figure

5.11 it can be seen that the output signals of the multi-port detector with anti-parallel diodes are quadrature signals, balanced in amplitude.

Figure 5.11: Measurement, anti-parallel diodes: I and Q signals down-converted at

Low-IF of 3 kHz. Frequency modulation at 10 Hz, not visible in the photography.

In Figure 5.12, the same anti-parallel diode configuration is tested with the RC-filter disconnected, hence with only the radial stub, notch RC-filter at 5.8 GHz remained. Only the I signal is shown. It can be observed that the 5.8 GHz radial stub and the anti-parallel diode configuration effectively remove high frequency leakage and minimize the dc-offset

Figure 5.12: . Measurement, anti-parallel diodes: I signal down-converted at

Low-IF of 3 kHz with the RC-filter discarded.

For the multi-port detector with parallel-diode configuration, it can be seen in Figure 5.13, that the differential signals I+ and I- have different dc-offsets. Also it can be observed that the lower driving capability of each diode results in distorted output signals. The same behavior is shown for the case when the RC-filter is disconnected, see Figure 5.14, the amplitude of the signals is increased a alot (from 5 mV to around 50 mV) but the problem of dc-offsets remains.

Figure 5.13: Measurement, parallel diodes: I and Q signals down-converted at

Low-IF of 3 kHz. Frequency modulation at 10 Hz, not visible in the photography.

Figure 5.14: Measurement, parallel diodes: I and Q signals down-converted at

Low-IF of 3 kHz with the RC-filter discarded. Frequency modulation at 10 Hz, not visible in the photography.

Chapter 6

Discussion

In this project, several radio frequency (RF) modules were designed, manufactured and measured using the Link¨oping University, Campus Norrk¨oping facilities, e.g., the printed circuit board (PCB) laboratory and the RF laboratory.

Going through a complete design-to-prototype cycle has a multitude of benefits, e.g., that of in-depth understanding the problems that are usually present when designing RF circuits and systems; Or, understanding the way in which real RF circuit performance can be affected by the manufacturing process.

A main way to analyze the first prototype performances is to compare the ex-pected circuit performance as resulting from simulations to the measured ones. Usu-ally, accurate simulations and accurate component models predict well how the cir-cuit will work. But still there are differences that can be attributed either to the simulation accuracy and limitations or to the limitations in conducting desired ex-periments.

From this perspective, it can be concluded that the first multi-port detector prototype developed in this Master-degree project work has demonstrated that low-frequency signals can be detected using a direct conversion multi-port RF front-end architecture. For the first time in this project, the multi-port front-end was used for an applications that implies a 5.8 GHz signal, modulated in frequency by a 10-Hz signal. The proof was done through experiments on own developed RF front-end with signals generated by instruments. Problems that were met during the experiments are listed:

• The minimum frequency that could be detected in the spectrum was 9 kHz, much higher than the vital sign frequency specification, i.e., under 1 Hz. • The test with the RF and LO signals as generated by two different signal

gen-erators made impossible the detection only of the modulating signal. The error between the two GHz frequencies resulted in low-IF signal down conversion at the output ports of the detector. However, this problem is avoided in a real radar use of the multi-port detector, where the sent and received signal differ only through the Doppler effect in the phase of the received signal.

Chapter 7

Conclusion

The goal of this master thesis was the design, manufacturing and measurement of a multi-port RF-front end for a vital signs detection system. The multi-port front-end was thought to be followed by a digital signal processing (DSP) unit that should perform final digital processing on the raw data.

The thesis work makes partially use of the results of a previous project [2] that was more theoretically and RF design oriented. All the necessary microwave modules presented in this thesis are own designs of the author. After the design work using the dedicated tool ADS from Keysight Technologies, the microwave components, circuits and the entire RF detector front-end were manufactured and tested. Hence, through this work, an entire design-to-prototype cycle was completed.

At first, the multi-port correlator and two diode power detector topologies were separately designed and successfully tested. After this first step, the entire multi-port detector was designed and manufactured. It includes the six-multi-port correlator and the power detectors. Depending on which type of power detector are used, the parallel-diode and the anti-parallel diode multi-port detectors were available for measurements and evaluation of the functionality.

As shown in Chapter 5, the main conclusion of this thesis is that through exper-iments, the correct operation of the multi-port detector was demonstrated. It was shown that very low-frequency signals emulating vital signs can be detected, when direct frequency conversion and demodulation are performed with the multi-port detector. The task was to demonstrate weak signals around 1 Hz at the outputs of the detector. Due to the limitations of the instruments, 10-Hz signals were demon-strated, even without a last amplifying stage. Other conclusions refer to the type of power detectors to be used. It was shown that the anti-parallel diode multi-port detector can minimize problems that usually appear as a result of impairments be-tween I and Q paths microwave-signal processing in the six-port correlator and due to the isolation problem between port 1 and 2, e.g., dc-offsets.

Limitations of the project results were due to the required time to finish different parts of the project and due to operational range limits of the used measurement equipment. Further work mainly refers to the design of the final amplifier stage interfacing the RF multi-port module to the DSP module. Then the entire RF-DSP system should be verified, firstly with signals generated by instruments and finally, by real vital signals.

Bibliography

[1] J. C. Lin. Noninvasive microwave measurement of respiration vol. 63 1975-10. [2] Morales, O. Multi-Port Receiver/Sensor System Modelling for Vital Sign

Detec-tion ApplicaDetec-tions Link¨opings universitet, Tekniska h¨ogskolan.

[3] M Pozar, D. Microwave Engineering, Fourth Edition. John Wiley Sons, United States of America, 978-0470631553, 2012.

[4] Nieh, C-M. Wei, C. Lin, J. Concurrent Detection of Vibration and Distance Using

Unmodulated CW Doppler Vibration Radar With An Adaptive Beam-Steering Antenna, Vol. 63, No.6 IEEE Trans. Microw. Thery Techn. 2015-06.

[5] Y. Xiao, J. Lin, O. Boric-Lubecke, and V. M. Lubecke. Frequency tuning

tech-nique for remote detection of heartbeat and respiration using low-power double-sideband transmission in the Ka-band, vol. 54, no. 5 IEEE Trans. Microw.

The-ory Techn. 2006-05.

[6] Petterson, T. Implementation of vital sign detection algorithms on a

high-performance digital signal processor Link¨opings universitet, Tekniska h¨ogskolan.

2017.

[7] Adriana Serban. Norrk opings Fond f or Forskning och Utbildning (NFFU),

Radar/Radio Project Documentation. Oct. 9, 2017.

[8] AVAGO Technologies: HSMS-286x Series, 2006-08-22,

http://www.efo.ru/components/avago/catalog/files/pdf/5989-4023EN.pdf [9] Keysight: E8267D PSG Vector Signal Generator, 2016-02-10,

https://literature.cdn.keysight.com/litweb/pdf/5989-0697EN.pdf?id=473817 [10] RHODESCHWARZ: Operating Manual, Volume 1, 2015-08-03,

https://cdn.rohde-schwarz.com/pws/dldownloads/dlcommonlibrary/

dlmanuals/gb1/s/smiq1/SmiqbOperatingM anualV1en11.pdf

[11] RHODESCHWARZ: Operating Manual, Volume 2, 2015-08-03,

https://cdn.rohde-schwarz.com/pws/dldownloads/dlcommonlibrary/

dlmanuals/gb1/s/smiq1/SmiqbOperatingM anualV2en11.pdf

[12] RHODESCHWARZ: Vector Network Analyzers ZVM, ZVK, https://cdn.testequity.com/documents/pdf/zvm.pdf

[13] LECROY: 9310A Family Digital Oscilloscopes 400 MHz Bandwidth, 100 MS/s, https://www.testequipmentconnection.com/specs/LeCroy9310AM.P DF

[14] KEYSIGHT Technologies: ESA-E Series Spectrum Analyzer, Data Sheet, 2017-12-1