pub.tik.ee.ethz.ch · Abstract The PermaSense project is a joint computer science and geoscience project, aiming to understand the in uence of climate change on permafrost. A reliable

Institut fürTechnische Informatik undKommunikationsnetze

Supervisors: Professor:

Dr. Jan Beutel Prof. Dr. Lothar Thiele

Dr. Stephan Gruber

Master Thesis

Acoustic Emission Sensing in

Wireless Sensor Networks

Josua Hunziker

5th May 2011

II

Abstract

The PermaSense project is a joint computer science and geoscience project,aiming to understand the inuence of climate change on permafrost. Areliable wireless sensor network (WSN) infrastructure has been developed,enabling scientists to constantly monitor the condition of certain rock wallson watch in near real time. However, as the network infrastructure is opti-mized for energy ecient operation and low data rates, only slow physicalprocesses can be monitored.

This thesis explores the possibilities of acquiring sensor data at higher sam-pling rates in the context of the existing PermaSense architecture, specif-ically focusing on the application of acoustic emission (AE) measurementson rock walls. To this end, the architectural concept of processing sensorsis introduced and a prototype AE sensing system is presented, integratingseamlessly into the existing PermaSense WSN infrastructure. Specically,the prototype's architecture and design are discussed, followed by a systemevaluation based on laboratory experiments.

The evaluation results show that the AE sensing system designed successfullyintegrates AE measurements into PermaSense. Furthermore, variations inmeasurement quality depending on the ambient temperature are character-ized qualitatively. Further steps include deploying the developed system inthe eld in order to gain experience in the targeted operating environment,as well as quantitatively analyzing and eliminating temperature dependentmeasurement errors.

III

IV

Acknowledgements

First and foremost, I would like to thank Dr. Jan Beutel for his excellentsupport, the constant feedback and all the insightful discussions throughoutthis project. You challenged me to dig deeper in every aspect. I also thankDr. Stephan Gruber for co-advising this thesis, providing the geoscienticfocus and being such a patient and grateful, yet demanding customer. Fur-thermore, I would like to thank Prof. Dr. Lothar Thiele for having givenme the opportunity to contribute to the research activity of the ComputerEngineering group in several student projects throughout the last few years.Many thanks also to the following people who supported me in various as-pects: Tonio Gsell, Roman Lim, Christoph Walser and Mustafa Yücel (Com-puter Engineering Group), Michael Lerjen (Communication Theory Group),Manuel Löhr and Joachim Sell (Physical Acoustics Corporation) as well asDr. Mauro Ciappa (Integrated Systems Laboratory).

I would also like to express my gratitude to my family and friends for alltheir constant support and care. My dear parents: You have invested somuch in my life no thank could ever measure it up. Romina, my ancée:You are a constant source of joy and strength in my life. Nobody else couldll what you are to me.

I thank God, my Father, my King. Your truth sets free!

V

VI

Contents

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Basic Concepts 5

2.1 Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . 5

2.1.1 The PermaSense WSN Architecture . . . . . . . . . . 6

2.2 Acoustic Emission Sensing . . . . . . . . . . . . . . . . . . . . 8

2.2.1 Sensor Technology . . . . . . . . . . . . . . . . . . . . 9

2.2.2 Data Processing and Signal Parameter Analysis . . . . 10

2.2.3 AE Sensing in Geosciences . . . . . . . . . . . . . . . . 11

3 System Specication 13

3.1 Preliminary Experiments . . . . . . . . . . . . . . . . . . . . . 13

3.1.1 Installation . . . . . . . . . . . . . . . . . . . . . . . . 13

3.1.2 Data Analysis from a Technical Perspective . . . . . . 14

3.2 Intented Experimental Setup . . . . . . . . . . . . . . . . . . 20

3.3 Network Bandwidth Considerations . . . . . . . . . . . . . . . 20

3.4 Functional System Specication . . . . . . . . . . . . . . . . . 23

4 High Data Rate Sensing in Low Bandwidth WSNs 25

4.1 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . 25

4.2 Raw Data Reduction Strategies . . . . . . . . . . . . . . . . . 26

4.2.1 Duty-Cycling . . . . . . . . . . . . . . . . . . . . . . . 26

4.2.2 Raw Data Classication . . . . . . . . . . . . . . . . . 27

4.3 Direct Data Aggregation . . . . . . . . . . . . . . . . . . . . . 27

4.3.1 Processing Sensors . . . . . . . . . . . . . . . . . . . . 29

5 AEBoard Hardware Design 31

5.1 Basic System Architecture . . . . . . . . . . . . . . . . . . . . 31

VII

5.2 AE Sensor Key Components . . . . . . . . . . . . . . . . . . . 32

5.2.1 Slave Processor . . . . . . . . . . . . . . . . . . . . . . 32

5.2.2 ADCs . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

5.3 Input Signal Conditioning Circuitry . . . . . . . . . . . . . . 35

5.4 Power Concept . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5.5 AEBoard Prototype . . . . . . . . . . . . . . . . . . . . . . . 39

5.5.1 Changes for Revision 1.2 . . . . . . . . . . . . . . . . . 41

6 PermAE Software Design 43

6.1 General Concept . . . . . . . . . . . . . . . . . . . . . . . . . 43

6.2 PermAE Messages . . . . . . . . . . . . . . . . . . . . . . . . 44

6.2.1 PermAE Output Messages . . . . . . . . . . . . . . . . 44

6.2.2 PermAE Dozer Commands . . . . . . . . . . . . . . . 45

6.3 Onboard Communication Protocol . . . . . . . . . . . . . . . 45

6.4 STM32 Implementation . . . . . . . . . . . . . . . . . . . . . 47

6.4.1 PermAE Initialization . . . . . . . . . . . . . . . . . . 50

6.4.2 Data Acquisition . . . . . . . . . . . . . . . . . . . . . 50

6.4.3 Data Analysis Controller Task . . . . . . . . . . . . . 52

6.4.4 Data Analysis Task . . . . . . . . . . . . . . . . . . . . 52

6.4.5 Communication Task . . . . . . . . . . . . . . . . . . . 54

6.4.6 Storage Manager Task . . . . . . . . . . . . . . . . . . 54

6.4.7 Event Statistics and Voltage Supervision Tasks . . . . 56

6.4.8 PermAE Initialization and Calibration . . . . . . . . . 56

6.5 TinyNode Implementation . . . . . . . . . . . . . . . . . . . . 57

6.5.1 AEBoard Drivers . . . . . . . . . . . . . . . . . . . . . 58

6.5.2 AEModule . . . . . . . . . . . . . . . . . . . . . . . . . 58

6.5.3 PermaDozer and DataControl . . . . . . . . . . . . . . 58

6.5.4 PermAE Top Level Application . . . . . . . . . . . . . 58

6.6 GSN Integration . . . . . . . . . . . . . . . . . . . . . . . . . 59

7 System Evaluation 61

7.1 Measurements at Room Temperature . . . . . . . . . . . . . . 61

7.1.1 Analog Frontend Characterization . . . . . . . . . . . 61

7.1.2 Power Consumption . . . . . . . . . . . . . . . . . . . 65

7.2 Temperature Cycle Tests . . . . . . . . . . . . . . . . . . . . . 66

7.2.1 Load Test . . . . . . . . . . . . . . . . . . . . . . . . . 66

7.2.2 Data Quality Test: AEBoard and USB AE Node . . . 68

7.2.3 Current Consumption . . . . . . . . . . . . . . . . . . 75

8 Conclusion and Outlook 77

VIII

A AEBoard and PermAE User Manual 79

A.1 Connectors and LEDs . . . . . . . . . . . . . . . . . . . . . . 79A.2 Deployment Preparation . . . . . . . . . . . . . . . . . . . . . 81A.3 STM32 Debugging . . . . . . . . . . . . . . . . . . . . . . . . 82A.4 Data Retrieval . . . . . . . . . . . . . . . . . . . . . . . . . . 83

A.4.1 Data Sent Over the WSN . . . . . . . . . . . . . . . . 83A.4.2 Raw Event Samples . . . . . . . . . . . . . . . . . . . 83

A.5 Sending Commands to the AEBoard . . . . . . . . . . . . . . 84

B PermAE Developer's Guide 87

B.1 STM32 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87B.1.1 PermAE . . . . . . . . . . . . . . . . . . . . . . . . . . 87B.1.2 PermAEInit . . . . . . . . . . . . . . . . . . . . . . . . 89B.1.3 EventGenerator . . . . . . . . . . . . . . . . . . . . . . 89B.1.4 Programming and Debugging the STM32 . . . . . . . 90

B.2 TinyNode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

C PermAE Output Messages 93

D Data Units and Conversion Functions 97

D.1 AE Data Units and Conversion Functions . . . . . . . . . . . 97D.1.1 Start Sample, Length and Risetime . . . . . . . . . . . 97D.1.2 Amplitude and Threshold . . . . . . . . . . . . . . . . 97D.1.3 Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . 98D.1.4 Threshold and Channel . . . . . . . . . . . . . . . . . 99

D.2 System Health Data Conversion . . . . . . . . . . . . . . . . . 99D.2.1 Onboard Voltages . . . . . . . . . . . . . . . . . . . . 99D.2.2 Onboard Currents . . . . . . . . . . . . . . . . . . . . 100D.2.3 Temperature and Humidity . . . . . . . . . . . . . . . 100

E AEBoard PCB Schematics and Layout 103

F Performance Measurements 111

F.1 Input Filter Linearity . . . . . . . . . . . . . . . . . . . . . . . 111F.2 Temperature Dependency of AE Parameters . . . . . . . . . . 114

G CD Contents 119

H Task Denition 121

IX

X

List of Tables

5.1 Comparison of three potential slave processor platforms. . . . 345.2 Comparison of two potential ADCs. . . . . . . . . . . . . . . 35

6.1 PermAE output messages . . . . . . . . . . . . . . . . . . . . 446.2 Dozer command messages . . . . . . . . . . . . . . . . . . . . 456.3 Onboard ACK and NACK messages . . . . . . . . . . . . . . 466.4 Onboard command messages . . . . . . . . . . . . . . . . . . 466.5 Onboard data message types . . . . . . . . . . . . . . . . . . 466.6 PermAE memory partition. . . . . . . . . . . . . . . . . . . . 506.7 The PermAE initialization block . . . . . . . . . . . . . . . . 516.8 Raw event data block on SD card . . . . . . . . . . . . . . . . 55

7.1 AEBoard power consumption. . . . . . . . . . . . . . . . . . . 657.2 Data quality test results for AEBoard channel one. . . . . . . 707.3 Parameter standard deviations for AEBoard channel one. . . 717.4 Data quality test results for AEBoard channel two. . . . . . . 727.5 Parameter standard deviations for AEBoard channel two. . . 727.6 Conversion factors between PermAE and AEWin. . . . . . . . 75

C.1 Message denition: aedata . . . . . . . . . . . . . . . . . . . . 94C.2 Message denition: aedaqhelathdata . . . . . . . . . . . . . . 95C.3 Message denition: aestatistics . . . . . . . . . . . . . . . . . 95C.4 Message denition: nodehealth . . . . . . . . . . . . . . . . . 96

D.1 Onboard voltage constants . . . . . . . . . . . . . . . . . . . . 100

XI

XII

List of Figures

2.1 Basic principle of a multihop WSN . . . . . . . . . . . . . . . 62.2 The PermaSense network architecture . . . . . . . . . . . . . 72.3 Piezoelectric AE sensor . . . . . . . . . . . . . . . . . . . . . 92.4 AE data acquisition chain . . . . . . . . . . . . . . . . . . . . 92.5 AE event parameters . . . . . . . . . . . . . . . . . . . . . . . 11

3.1 AE sensor positions for preliminary experiments at Jungfraujoch 143.2 Channel 4 event rate. . . . . . . . . . . . . . . . . . . . . . . . 153.3 The eect of lowering the ADC bit resolution . . . . . . . . . 183.4 The eect of downsampling . . . . . . . . . . . . . . . . . . . 193.5 Intended experimental setup of an AE measurement location 213.6 Increasing link bandwith requirements in multihop WSNs . . 22

4.1 Basic principle of duty-cycling . . . . . . . . . . . . . . . . . . 264.2 Basic principle of raw data classication . . . . . . . . . . . . 274.3 Basic principle of direct data aggregation . . . . . . . . . . . 284.4 The processing sensor architecture . . . . . . . . . . . . . . . 29

5.1 AEBoard architectural concept . . . . . . . . . . . . . . . . . 325.2 Slave processor data streams. . . . . . . . . . . . . . . . . . . 335.3 Simplied signal conditioning circuitry schematics . . . . . . . 375.4 Signal conditioning circuitry frequency response. . . . . . . . 375.5 AEBoard power concept . . . . . . . . . . . . . . . . . . . . . 405.6 The AEBoard prototype. . . . . . . . . . . . . . . . . . . . . . 41

6.1 Valid onboard message ows . . . . . . . . . . . . . . . . . . . 476.2 PermAE application structure on the STM32 . . . . . . . . . 496.3 AEBoard data acquisition wiring . . . . . . . . . . . . . . . . 516.4 Generated data acquisition clocks . . . . . . . . . . . . . . . . 526.5 PermAE application structure on the TinyNode . . . . . . . . 57

7.1 Input lter frequency response without preamplier . . . . . . 627.2 Input lter frequency response with preamplier . . . . . . . 63

XIII

7.3 Uncompensated AEBoard input lter error. . . . . . . . . . . 647.4 Compensated AEBoard input lter error. . . . . . . . . . . . 647.5 Compensated AEBoard and preamplier input lter error. . . 657.6 Load test setup. . . . . . . . . . . . . . . . . . . . . . . . . . . 677.7 System performance under varying temperature conditions. . 687.8 Data quality test setup. . . . . . . . . . . . . . . . . . . . . . 697.9 Test signal generated by the EventGenerator application. . . 697.10 Pro-cyclic versus anti-cyclic parameter behaviour. . . . . . . . 717.11 Temperature dependent mean parameter ranges.. . . . . . . . 737.12 The eect of constant peramplier temperature. . . . . . . . . 737.13 USB AE Node temperature dependent variations. . . . . . . . 747.14 AEBoard's power consumption at varying ambient temperature. 75

A.1 AEBoard connectors and LEDs. . . . . . . . . . . . . . . . . . 80

F.1 Compensated AEBoard input lter error (30kHz). . . . . . . . 111F.2 Compensated AEBoard input lter error (50kHz). . . . . . . . 112F.3 Compensated AEBoard input lter error (100kHz). . . . . . . 112F.4 Comp. AEBoard input lter error w. ext. preamp (30kHz). . 113F.5 Comp. AEBoard input lter error w. ext. preamp (50kHz). . 113F.6 Comp. AEBoard input lter error w. ext. preamp (100kHz). 114F.7 Data quality test parameter plots for AEBoard channel 1. . . 115F.8 Data quality test parameter plots for AEBoard channel 2. . . 116F.9 Data quality test parameter plots for the USB AE Node. . . . 117

XIV

1Introduction

1.1 Motivation

The PermaSense project [1] is a joint computer science and geoscience project,aiming to understand the inuence of climate change on permafrost. A spe-cial focus lies on the stability of rock walls in alpine environments, wherenatural hazards pose a constant danger to habitants and infrastructure.PermaSense makes use of state-of-the-art ultra low-power sensor nodes thatform a multihop wireless sensor network (WSN). Data samples measured aretransmitted over the WSN to a core station, where the data is relayed to abackend server using a WLAN link and the public internet infrastructure.

In the context of such an environmental monitoring system, low energy con-sumption is a crucial requirement for the wireless sensor nodes. The sensornode platform currently used in PermaSense can sense and transmit data for3-5 years from a single battery. In order to achieve this high energy eciency,platform architecture, operating modes and network protocol have been op-timized to allow for a very low average power consumption by utilizing thesensor node's sleep mode capabilities as extensively as possible. This setuphas proven to be very reliable and stable during the continuous operationof PermaSense measurement campaigns at Matterhorn and Jungfraujoch formore than three years at the time of writing.

In addition to the current measurements of rock temperature, rock humidityand crack movement, interest arose from the geoscientical side to capturephysical processes that expose faster variations. One phenomenon of par-

1

CHAPTER 1. INTRODUCTION

ticular interest are acoustic emissions (AE) during crack and ice formation.These acoustic emissions are generally very short in the order of a few mil-liseconds and exhibit frequencies of multiple kHz. As the architecture andoperating mode of the existing sensor nodes are optimized to capture muchslower processes and thus produce at most a few samples per minute, theyare not suitable for measuring and processing such fast processes. Therefore,new concepts are needed in order to allow the study of AE and other fastphenomena in the context of PermaSense.

In this thesis, a sensor platform architecture concept that is capable of per-forming high data rate measurements in a low-power WSN environment isproposed. Furthermore, an implementation of this concept that is tailored tomeasure, process and transmit AE data and fully integrates into the existingPermaSense WSN infrastructure is presented in detail.

1.2 Contributions

The contributions of this thesis are the following:

An architectural concept for performing high data rate measurementsin a low-power WSN environment.

A proof-of-concept implementation tailored for AE measurements inPermaSense, consisting of a hardware platform (AEBoard) and a dis-tributed software system running in this platform (PermAE).

An operational performance analysis and comparison to a commercialAE measurement system.

1.3 Related Work

The PermaSense project [1], its ndings and developments form the basiccontext for the work presented in this thesis. The PermaSense system archi-tecture has been described in [2, 3]. It has proved to deliver high-quality datareliably and over an extended period of time. However, the PermaSense ar-chitecture has so far primarily been used for measurements with a relativelylow data rate [4]. This thesis aims to extend the PermaSense ecosystem,in order to support applications that require higher sampling rates with-out compromising the standards of quality and reliability achieved so far.There are also eorts to integrate high-accuracy GPS measurements intoPermaSense, facing similar challenges but taking dierent approaches fordata reduction [5, 6, 7].

2

1.4. OUTLINE

High data rate WSN applications in general and processing of acoustic dataspecically have previously been studied in publications such as [8, 9]. Espe-cially the latter publication by Simon et al. inspired some concepts that canbe found in this thesis, specically the idea that a signicant part of the nec-essary data processing is done directly on the sensor node, thereby reducingthe amount of data to transmit over the network signicantly. Werner-Allenet al. presented another concept for data reduction in [10], transmitting rawsamples selectively based on a value metric for the measured data.

There are also projects that dealt with AE sensing in WSN specically.Grosse et al. presented a 4 channel AE device for WSN deployments in[11, 12]. Upon detection of AE activity, a number of samples is recorded andthen transmitted over the WSN, in the meanwhile stopping data acquisition.Bachmaier introduced another WSN AE device in [13], not sampling anyAE data at all but only estimating AE activity based on a xed threshold.Recently, Lédeczi et al. presented a WSN based AE monitoring systemfor bridges in [14], saving power by residing in sleep mode for most of thetime and only waking up and sampling upon increased AE activity that isindicated by a low-delity sensor. In all these projects, the AE signal is notcontinuously sampled in order to save energy and thus prolonging the systemlifetime. However, as the processes that produce AE activity in rockwalls arenot yet well understood, this thesis follows a dierent approach. In order toprovide as much insight as possible, we focus on characterizing as many AEevents as possible at a continuously high sampling rate. Considering theserequirements, power consumption has been optimized as a secondary goalonly.

1.4 Outline

Chapter 2 discusses the concepts of AE sensing and WSNs in greater de-tail. Chapter 3 describes the initial AE sensing experiments that initi-ated this thesis and derives the specications for an AE sensing systemin PermaSense from these preliminary results. In Chapter 4, our generalconcept for high data rate measurements in a low-power WSN environmentis presented. Chapters 5 and 6 document the AEBoard hardware platformdeveloped as well as PermAE, a distributed software system that runs onthe AEBoard and is able to measure and transmit AE data eciently. Theoverall system performance of AEBoard and PermAE is then evaluated inChapter 7. Finally, Chapter 8 summarizes the presented work and proposesfurther steps based on this thesis.

3

CHAPTER 1. INTRODUCTION

4

2Basic Concepts

This chapter aims to give a brief overview over the two basic concepts un-derneath this thesis: Wireless Sensor Networks (WSNs) as well as Acous-tic Emission (AE) sensing. It further provides references to comprehensivesources that cover these topics in-depth.

2.1 Wireless Sensor Networks

Wireless communication has become an ubiquitous technology these days.Not only do personal mobile devices like smartphones and tablet computersuse wireless communication interfaces as their primary channels, also in in-dustry and safety-critical applications, wireless technology is about to changetraditional patterns of human-machine interaction. Concepts like The In-ternet of Things [15] or Ubiquitous Computing [16] build on the existenceof small independent computing platforms that are connected to their envi-ronment and thus form the basic infrastructure for a "smart environment".

The term Wireless Sensor Network commonly refers to a distributed sys-tem, consisting of many light computing platforms that communicate overa wireless channel. Each of these so-called sensor nodes is sensing phe-nomena of its physical environment and forwards these measurements overthe network. In many common WSN architectures, all data harvested inthe network are collected at one or multiple distinct sink nodes, where thedata is either being forwarded over a dierent communication link, or canbe accessed directly.

5

CHAPTER 2. BASIC CONCEPTS

sink node

sensor node

to data relay

Figure 2.1: Basic principle of a multihop WSN

This approach of placing low-weight sensors unobtrusively in the environ-ment allows to measure and monitor phenomena very closely and in a natu-ral environment, as the nodes can be tailored and placed in such a way, thatthey have minimal impact on the process that is to be captured [17, 18].

There is a broad palette of applications for WSNs, ranging from militaryapplications over environmental monitoring, health applications, monitoringof infrastructure and buildings as well as a wide range of industrial prospects[19]. Their generally low cost, ease of deployment and low maintenancerequirements make WSNs an area of active research and high industrialinterest.

Nevertheless, developing and deploying reliable WSNs is still a challengingtask. One key constraint on autonomous wireless platforms is energy - theamount of energy available to the system ultimately determines the max-imum time of unattended operation and sensing. Although there exist avariety of concepts for energy harvesting on WSN nodes (e.g., [20]), stillmost of the platforms in practical use nowadays use batteries as their solepower supply. This requires a stringent power management on all nodes, of-ten resulting in putting the node into a low-power mode for most of the time,implying frequent disconnection and reconnection of nodes in the network.The resulting very dynamic network structure requires novel approaches fornetwork management and routing, as classical network protocols often as-sume rather static network topologies with infrequent changes. Thus, mul-tiple WSN-tailored network protocols have evolved over the past decade,e.g. Dozer [4], PEGASIS [21] or APTEEN [22], that aim to allow for anenergy-ecient WSN communication.

2.1.1 The PermaSense WSN Architecture

The PermaSense WSN infrastructure is tailored for maximum reliability evenunder harsh environmental conditions. Sudden temperature changes, snow,

6

2.1. WIRELESS SENSOR NETWORKS

ice, rockfall and other natural impacts drastically increase the probabilityof temporary or permanent disconnection of a node, compared to, e.g., anindoor deployment with rather controlled environmental conditions. Faulttolerance is therefore a key requirement on all scales.

The core WSN of PermaSense consists of sensor nodes based on the TinyNodeplatform by Shocksh SA [23, 24]. These TinyNodes are equipped witha custom-built sensor interface board (SIB), that connects to various sen-sors measuring rock temperatures, rock humidity, crack movement etc. Thesensor nodes run software based on the TinyOS operating system [25] andcommunicate over the ultra low-power Dozer network protocol [4], which iscongured to schedule communication slots every 30 seconds. All data isforwarded to a single sink node, which is part of the so-called CoreStation.The CoreStation is basically an embedded PC platform, featuring WLANand GPRS downlinks to forward all data to the data backend, which is basedon the GSN software [26] and features additional control, management andsupervision interfaces. Furthermore, sensors with higher data volume suchas high resolution photo cameras [27] and GPS receivers are also connectedto the backend over separate network links.

Figure 2.2: The PermaSense network architecture

In case of normal operation, samples arrive at the data backend on the or-der of a few minutes after measurement, assuming that no network links arecongested. Additionally, each sensor node in the WSN features a 1GB SD

7


storage, where measured sensor data can be buered in case of connectionloss (e.g., in case of the sensor being snowed in). Once the network connec-tion can be re-established, the accumulated data packets are ushed to thebackend.

The sensor nodes run on a single Li-SOCl2 battery which enabled them tooperate for more than three years. Larger components, like the core stationsas well as the high resolution cameras, are powered by a regenerative powersupply.

For more details about the current PermaSense architecture, please refer to[2, 3, 28].

2.2 Acoustic Emission Sensing

It is commonly known from daily experience, that materials often emitsounds under mechanical stress or when internal structures break. Gen-erally speaking, these (often inaudible) breaking sounds are referred to asAcoustic Emissions (AE):

Acoustic Emission (AE) refers to the generation of transientelastic waves produced by a sudden redistribution of stress in amaterial. When a structure is subjected to an external stimulus(change in pressure, load, or temperature), localized sources trig-ger the release of energy, in the form of stress waves, which prop-agate to the surface . . . Sources of AE vary from natural eventslike earthquakes and rockbursts to the initiation and growth ofcracks, slip and dislocation movements, melting, twinning, andphase transformations in metals. In composites, matrix crackingand ber breakage and debonding contribute to acoustic emis-sions. [29]

AE sensing technology began to evolve in the middle of the 20th century. Itis a so-called non-destructive testing (NDT) method, providing informationabout ongoing stress processes in the material sample under test. In contrastto other NDT methods, which are mostly applied before or after the materialis being stressed, AE sensing captures processes while they are taking place.This makes AE sensing particularly interesting for applications that requireconstant monitoring of certain structures, e.g. alarm systems for pipelineleaks. Often, this method is used to detect material failure at a very earlystage of damage and before the structure fails completely ([30], chapter 1).

8

2.2. ACOUSTIC EMISSION SENSING

2.2.1 Sensor Technology

AE signals are often being captured with piezoelectric sensors. These sen-sors are directly attached to the surface of the material sample under test,converting dynamic motions at the surface into an electrical signal.

Figure 2.3: Piezoelectric AE sensor (schematic representation). [29]

When using piezoelectric sensors, it must be considered that they are nor-mally operated at resonance and therefore do not allow broadband detectionof AE signals. Nevertheless, the frequency ranges of the expected signals areoften fairly well known, making it possible to choose the right sensor beforethe experiment. Thus, their ease of use as well as their high sensitivity makepiezoelectric sensors the instrument of choice for many AE applications, eventhough more sophisticated AE sensor concepts exist (e.g. sensors based onlaser systems or on ber optics).

Before A/D conversion, the electrical signal at the output of a piezoelectricsensor normally needs to be ltered and amplied. This process is alsoknown as signal conditioning. It is obvious, that these components ofan acquisition system are very sensitive and heavily inuence the overallmeasurement quality. The conditioned signal is then sampled and all furtherprocessing is then performed digitally.

AnalogFiltering

Amplification ADCProcessing

Figure 2.4: AE data acquisition chain

9


2.2.2 Data Processing and Signal Parameter Analysis

As most modern data analysis systems, today's AE systems almost exclu-sively perform the necessary signal detection, processing and characterizationtasks in the digital domain. AE activity normally occurs rapidly and ran-domly, resulting in distinct pulses of oscillation in the measured AE signal.Conventionally, a threshold is dened by the user in order to distinguish AEactivity from background noise. Once the AE signal crosses this threshold,an AE event is recognized. The event is considered to be terminated whenthe threshold has not been exceeded for a certain, user-dened amount oftime (i.e., hit-lockout time, dead time or posttrigger time). The AE sig-nal between the rst threshold crossing and the end of the posttrigger timeis also referred to as AE Hit or AE event. As especially in the early daysof AE testing technology, all of the signal processing had to be done in theanalog domain, some common signal parameters have evolved that still formthe basis for many of today's AE testing and analysis applications. Theseclassical AE signal parameters are ([30], chapter 4):

1. Event Count/Hit Count: A signal that exceeds the threshold andcauses the AE system to accumulate data. Hit counts are often usedto show the overall AE activity over a certain period of time.

2. Length/Duration: The amount of time between the rst thresholdcrossing of an event and the end of the posttrigger time.

3. Amplitude: The maximum signal amplitude within an event.

4. Rise Time: The time interval between the rst threshold crossing andthe time of the peak amplitude.

5. Pulse Count/Count: The number of times an AE event signal ex-ceeds the threshold value.

6. Energy: There is no single agreed denition of AE signal energy. Inthis thesis, we dene the energy of an AE event as the signal energycontained in the event, i.e., the sum of the squared sample amplitudes.Another popular energy denitions is the measured area under therectied signal envelope.

Figure 2.5 illustrates these aforementioned features of an AE event.

10

2.2. ACOUSTIC EMISSION SENSING

Threshold

Rise Time Amplitude

Length

Posttrigger

Pulse Count

Energy

Time

AE Signal

Figure 2.5: AE event parameters

From these basic parameters, further event parameters may be derived:

The Average Frequency is calculated by dividing Pulse Count byLength.

The RA Value is derived from Rise Time divided by Amplitude.This parameter can be used to classify cracks.

2.2.3 AE Sensing in Geosciences

AE sensing originally arose in the context of material sciences and is nowa-days particularly popular in the eld of civil engineering for structural healthmonitoring. Nevertheless, this method has a much broader scope of applica-tions, e.g., in geosciences for studying stability and structural displacementsin rock structures. In contrast to seismology, where large, long events atlow frequencies (i.e., earthquakes) are studied, AE measurements in rockfocuses on much smaller, shorter events that can give information about dis-placements in the order of a few micrometers ([30], chapter 11). In mining,scientists started to monitor rock pillars with AE measurements in the mid-dle of the 20th century. The development of these practical experiments was

11


accompanied by laboratory experiments, where rock samples were hydrauli-cally stressed and monitored for AE at the same time. Also nowadays, AE inrock structures under stress are still an area of active geoscientic research(see e.g., [31, 32, 33]).

The formation of ice within rock is an important driver of near-surface frostweathering [34] and rock damage at the depth of several meters [35], andin steep terrain, this process may be crucial for the slow preconditioning ofrock fall from warming permafrost areas [36]. However, the transfer of cor-responding theoretical insight and laboratory evidence to natural conditionscharacterized by strong spatial and temporal heterogeneity is nontrivial. Toprepare corresponding characterization of rock fracture in natural conditions,AE induced by natural thermal cycling and freeze/thaw in a high altituderock face are studied within the project PermaSense. This complements wellprevious investigations within PermaSense (e.g., [37, 38]). To the knowledgeof the author, no comparable study of AE in alpine rock walls exists so far.

12

3System Specication

This chapter describes the experiments and intentions, that led to the speci-cation of an AE sensing platform for PermaSense nally presented in Section3.4.

3.1 Preliminary Experiments

As no reference experiments of AE measurements in alpine rock walls exist,preliminary experiments had to be carried out in order to get a feeling for thetype of AE activity that has to be expected during more in-depth studies.For this purpose, AE experiments with conventional laboratory equipmenthave been carried out at the Jungfraujoch site in April 2010.

3.1.1 Installation

Six piezoelectric AE sensors have been installed at dierent positions inthe rock wall. The installation positions were selected in order to capturevarieties in rock structure and expected humidity, i.e., some sensors weremounted at dry and compact expositions, whereas others were placed atcracked positions featuring a constant ow of water. The sensors were con-nected to a laboratory computer running the AEWin software from PhysicalAcoustics Corporation [39] and AE data has been acquired for a period offour days. Both event parameters as well as the raw event waveforms havebeen captured.

13

CHAPTER 3. SYSTEM SPECIFICATION

The results of these measurements were very promising and reinforced in-terest in monitoring AE activity in rock walls over longer periods of time[40]. Thus, the development of a custom AE sensing platform that inte-grates into the existing PermaSense infrastructure was initiated, whereof arst prototype is presented in this thesis.

Figure 3.1: AE sensor positions for preliminary experiments at Jungfraujoch

3.1.2 Data Analysis from a Technical Perspective

In order to dene the basic requirements and constraints for a system im-plementation, the data gathered in the aforementioned experiment has beenanalyzed from a technical perspective.

All sensor channels were congured with a xed threshold of 35 dB andposttrigger value of 1ms. The signals were sampled with a quantizationresolution of 24 bits at a sampling rate of 1MHz.

Event Rate

Depending on the sensor position, the number of events occurring dieredsignicantly. We could also observe that events often occur in bursts: longperiods of low AE activity follow on shorter periods with a much higheractivity level.

For the specication of an AE measurement system, the following event ratemeasures are therefore of particular interest:

The maximum event rate denes the processing an storage bandwidth

14

3.1. PRELIMINARY EXPERIMENTS

Figure 3.2: Observed event rate of channel 4 during the preliminary experiments.

(in events per second) an acquisition system must oer in order to beable to parametrize all occurring events, regardless of the current levelof AE activity.

The average event rate denes the required communication bandwidth(in events per second), given there's enough buer storage available tocompensate for event bursts.

For both of these values, the worst case that occurred during the preliminaryexperiments was chosen for the system specication in order to ensure normaloperation under heavy load. As for the number of occurred events, the sensorat position number four showed by far the most intense AE activity. Duringthe four days of experimentation, a peak burst of 300 events in one secondwas observed. On a minute scale, the peak event rate was at 480 eventsper minute (c.f. Figure 3.2). However, generally the activity peaks didnot exceed 100 events per minute. The average event rate over the wholeexperiment was in the order of only a few events per minute. Again, sensorposition four showed the highest overall activity level with an average of 3events per minute.

Generally, the event rate observed is highly dependent on data acquisitionparameters, i.e., threshold and posttrigger time: Choosing a higher thresh-old will decrease the number of events observed while also decreasing theoverall system sensitivity. Increasing the posttrigger time will concatenateevents, that would have been detected as distinct events otherwise. A suit-able compromise has to be found for every deployment, depending on theactivity characteristics expected.

15


Frequency Bandwidth

The frequency bandwidth of the observed signals depends heavily on thesensor devices used for measuring. As piezoelectric sensors generally showdistinct resonances in their frequency response, they already lter the me-chanical input signal signicantly. It is therefore important to know the typeof AE activity that is expected to be observed. As the speed of sound in(fractured) rock is relatively low (i.e., about 3600m/s) compared to, e.g.,steel due to its lower density, the expected AE frequencies are rather low,i.e., below 100 kHz ([30], chapter 11). In our preliminary experiments, piezo-electric sensors of the type R6α by Physical Acoustics Corporation were used[41]. The sensors' frequency response is measured in V/(m/s), i.e., outputvoltage as a function of the rock's speed of movement. The peak resonance ofthe R6α sensors lies at a frequency of 55 kHz, with a loss of less than 10 dB(V/(m/s)) in a range of 35 to 100 kHz. This also matches the observedaverage event frequencies on Jungfraujoch.

Event Length

Event length, together with the sampling rate, determines memory buersize requirements in order to store event samples. In the preliminary experi-ments, the observed event length was generally in the order of 1 up to 5ms.However, also event lengths of up to 180ms have been detected. Again,this parameter heavily depends on the set threshold and posttrigger time: Ahigher threshold or a longer posttrigger time would have resulted in more,but shorter events.

Amplitude Dynamic Range

The highest event amplitudes observed in the preliminary experiment were104 dB. As for parametrization, every signal or pulse not exceeding thethreshold of 35 dB is classied as noise, the dynamic range of the observedsignal is about 70 dB.

The event amplitudes' dynamic range is important in order to specify thenecessary ADC resolution. As a result of quantization, the signal to noiseratio of an ideal ADC with a resolution of N bits is given as

SNRdB[MAX](N) = 6.021 dB ·N − 1.763 dB [42]

Thus, a sampling resolution of at least 12 bits is needed to cover the givendynamic range, as SNRdB[MAX](12) = 70.5 dB.

In order to get a better impression of the eects of lowering the bit resolution,the captured waveforms have been resampled and reparametrized. Figure 3.3

16


compares the parametrization errors to the original parametrization whenlowering the sampling resolution from 24 bits to 20, 16 and 12 bits. It clearlyshows that a signal quantization of 12 bits introduces signicant errors inthe signal parametrization process, whereas a resolution of 20 bits performsalmost equally to the full resolution of 24 bits. A quantization resolutionof 16 bits oers a good trade-o between complexity and introduced error,as still for 85% of the expected events, all parameters show an error of lessthan 10%. The theoretical dynamic range of an ideal 16 bit ADC is given asSNRdB[MAX](16) = 94.6dB.

Sampling Rate

The second important parameter determining the error introduced by sam-pling and quantizing an analog signal is the sampling rate used. Highersampling rates improve the temporal resolution of the measurement and al-low to capture higher signal frequencies, but on the other hand thereby alsoincrease the required signal processing capabilities. As already discussedin Section 3.1.2, the expected AE signal frequencies are in the range of upto 100 kHz. According to the Nyquist-Shannon sampling theorem [43], thesampling frequency must be at least twice the highest frequency occurringin the measured signal in order to capture the signal unambiguously (i.e.,without aliasing eects). Sampling at the Nyquist rate would allow to recon-struct the measured signal perfectly, given that it is strictly bandlimited tothe specied bandwidth. Nevertheless, as signal reconstruction would intro-duce signicant processing overhead compared to parametrizing the sampledsignal, higher sampling rates are needed in order to minimize the error in-troduced. Figure 3.4 shows the eect of downsampling the measured events,originally sampled at 1MHz.

Obviously, sampling at the Nyquist rate of 200 kHz introduces signicant er-rors to the event parametrization. As expected, deeper analysis also showedthat this error is higher for events with average frequencies close to 100 kHzand lower for events with comparably low average frequencies. Sampling at900 kHz already introduces notable errors, while lowering the sampling ratefrom 900 to 500 kHz, the loss of parametrization precision is comparablysmall. Thus, a sampling rate of 500 kHz seems to perform reasonably well,again introducing parametrization errors of less than 10% for more than 85%of all occurring events.

17


Figure 3.3: The eect of lowering the ADC bit resolution on the captured AEevents. The reparametrization results are compared to the originalresult with a full resolution of 24 bits.

18


Figure 3.4: The eect of downsampling the captured AE events. Thereparametrization results are compared to the original result sampledat 1MHz.

19


3.2 Intented Experimental Setup

Having studied and analyzed the Jungfraujoch data [40], the geoscienticPermaSense partners dened an experimental setup for studying AE activity.Specically, the measured AE activity shall be correlated with the alreadyexisting temperature and moisture measurements, giving further insight intothe role of water in, e.g., crack formation.

As rock moisture and temperature are measured in multiple depths, it isalso desirable for the AE platform to sense not only general activity, butalso give a notion of how deep the event source was located. A planarlocation of the events is however of smaller interest, as signal attenuationin fractured rock is comparably high and thus AE signals are not expectedto travel over distances in the order of meters or more. Therefore, eventlocation requirements have been limited to linear depth location, requiringat least two AE sensors mounted in dierent depths. AE source depth canthen be estimated by measuring the arrival time dierence of an AE signalat the two sensors. Furthermore, the event rise time at both sensors alsogives information about the source location, i.e., the longer the rise time,the further was the AE source located.

Measuring the arrival time dierence between the two sensors requires exacttime synchronization in the order of microseconds: As the speed of soundin rock is about 3000m/s and the sensors will be mounted at a distance ofapproximately one meter, the arrival time dierences will be in the order ofless than one millisecond. As in the Dozer based PermaSense network, noexact time synchronization can be achieved between two sensor nodes, bothAE sensors belonging to one location need to be read and processed by asingle node in order to achieve the necessary temporal precision. Figure 3.5therefore sketches the intended experimental AE measurement setup.

3.3 Network Bandwidth Considerations

So far, the data streams measured in PermaSense have been transferred tothe backend completely, meaning that each and every measured sample hasnally been stored in the PermaSense database and could be retrieved fromthere to be analyzed. Naturally, this approach oers the highest degree ofreliability as well as exibility in data analysis, as all processing steps canbe done oine (i.e., not in real time) and non-destructively, i.e., withoutloosing any data. Currently, an average PermaSense node generates lessthan 100Bytes of payload data per minute. When considering the ndingsand requirements from Sections 3.1 and 3.2 however, it is easy to see that areasonable AE sensing platform will produce much more data:

20

3.3. NETWORK BANDWIDTH CONSIDERATIONS

rock

sensor rod(temperature / humidity)

sensor nodefor sensor rod

AE node

AE sensors

Figure 3.5: Intended experimental setup of an AE measurement location

Bandwidthall samples = N · fs · w= 2 · 500 · 106 s−1 · 16 bits

= 16 · 106 bits/s

= 2MB/s

= 120MB/min

where N denotes the number of channels, fs denotes the sampling frequencyand w denotes the bit width per sample. Thus, given that the AE sensingboard should integrate into the existing PermaSense WSN infrastructure,streaming all measured AE data samples to the backend is not feasible.

A rst step of reducing the amount of data to be transferred would be toonly forward raw samples when an event (i.e., a signal that exceeds a setthreshold) has been detected. Again referring to the ndings from Section3.1, we would expect an average event rate of about two events per minute,each event having a length of about 2ms. In this case, the required networkbandwidth for serving one AE sensor node would be

21


Bandwidthevent samples = dBandwidthall samples · fe · lee

= d16 · 106 bits/s · 2

60 s· 2 · 10−3 se

= 134Bytes/s

= 8040Bytes/min

with fe denoting the expected event frequency and le denoting the expectedevent length. Obviously, only forwarding event data samples would re-duce the required network bandwidth signicantly. When aggregating datastreams at a sinknode or over multiple WSN hops however, the requiredbandwidth for one point to point link is much higher than the amount ofdata produced by a single node as Figure 3.6 illustrates:

AE node

sink node

AE node

AE node

relay noderelay node

67 Bytes/s

67 Bytes/s

134 Bytes/s

67 Bytes/s

201 Bytes/s

required link bandwith

Figure 3.6: Increasing link bandwith requirements in multihop WSNs

Thus, the probability of congesting the PermaSense WSN with AE datastreams is still high, making the transmission of raw samples undesirablefrom a technical perspective.

However, when only considering the transmission of the event parametersdiscussed in Section 2.2.2, the required bandwidth can again be reduceddrastically:

Bandwidthevent parameters = fe · sp

=2

60 s· 23Bytes

= 46Bytes/min

where fe again denotes the expected event frequency and sp denotes themaximum payload size of one Dozer packet in PermaSense, assuming that allevent parameters t into one Dozer packet. Even with the worst case average

22

3.4. FUNCTIONAL SYSTEM SPECIFICATION

event rate of three events per minute, only 69Bytes of data per minute wouldbe produced. Thus, the required network bandwidth for an AE sensingplatform that only transmits the event parameters is of the same order asfor the existing sensors. This approach seems to be the most promising inorder to achieve a seamless and tight integration into the existing PermaSenseinfrastructure.

3.4 Functional System Specication

The ndings from Sections 3.1 to 3.3 formed the basis to dene the followingfunctional specication for a PermaSense AE sensor node:

The AE sensor node shall integrate seamlessly into the existing PermaSenseWSN infrastructure. Specically, it shall communicate its data over theDozer WSN.

Two piezoelectric sensors shall be sampled with a sampling rate of500 kHz and a quantization resolution of 16 bits. Input signals withfrequencies between 20 kHz and 100 kHz shall be captured.

The sensor data shall be processed directly on the sensor node. Theprocessing must thereby respect the following priorization:

1. Count the occurring events and transmit the cumulative numberover a period of two minutes. This counter value must be accurateand up to date under all circumstances, i.e., also under high AEactivity load.

2. Parametrize the occurring events and transmit the event param-eters. If AE activity load is too high to parametrize all events,parametrization may be done on a best eort basis.

3. Store the raw event samples on an internal storage device forlater oine inspection and analysis if resources are not busy withcounting and parametrizing events.

The processing must behave robust under high AE activity, i.e., if notall detected events can be parametrized and / or stored, the systemmust not run into an illegal state and return to normal operation whenAE activity decreases.

The two channels' measurements must be temporally correlatable withan accuracy of one sample.

The threshold and posttrigger values shall be adaptable in operation,i.e., with command messages sent to the AE sensor node.

23


The most important system voltages as well as on board temperatureand humidity shall be measured and transmitted to the backend everytwo minutes.

Energy consumption shall be minimized as far as possible. Multiplepower supply domains shall be switchable by the user in order to opti-mize the system's energy eciency. The system shall run at a supplyvoltage of 12V DC. If desired, a photovoltaic power supply may beused instead of batteries.

The AE sensor must feature an operating range from -40 to +65C,with a max. change rate of 5C per minute.

24

4High Data Rate Sensing in Low

Bandwidth WSNs

Before going into implementation details of the developed AE sensing plat-form, this chapter introduces some general design concepts. Specically, ageneric architectural concept for sensing data with high sampling rates in alow bandwidth WSN environment is presented, that may also be used forapplications other than AE sensing.

4.1 Problem Statement

Most WSN architectures and protocols are optimized to maximize unat-tended lifetime. As many sensor nodes are battery powered, energy is often akey factor limiting the independent operating timespan. However, especiallyin environmental monitoring, the physical processes captured often exposerelatively slow temporal variations, requiring only low sampling rates in theorder of a few samples per minute to measure them adequately. Thus, manyWSN operating approaches aim to save energy by putting the sensor nodesinto some low power mode for most of the time, regularly waking them upfor performing short data sampling and/or communication sequences. Thisinherently also decreases the available network bandwidth and node reactiv-ity.

Integrating measurements of processes that - due to their faster varying na-ture - require continuous sampling at higher rates thus produce data in the

25

CHAPTER 4. HIGH DATA RATE SENSING IN LOW BANDWIDTH WSNS

order of multiple MB/second thus requires novel architectural and opera-tional approaches. As the available bandwidth does not allow to streamall measured data through the WSN, some form of data reduction musttake place directly on the sensor node itself. Besides compressing the datato transmit over the network with generic statistical data compression al-gorithms such as Human coding [44], some more specic approaches arepossible that are discussed in the following.

4.2 Raw Data Reduction Strategies

The simplest possibility to reduce the required bandwidth for high datarate measurements is to reduce the amount of raw data samples that istransmitted over the WSN. To this end, two basic approaches are possiblewhich are presented in the following sections.

4.2.1 Duty-Cycling

Bandwidth requirements for high data rate measurements can be lowered byreducing the timespan, during which the phenomenon of interest is sampled,i.e., only sampling for short periods with intermediate longer periods whereno sampling takes place. E.g., consider an application where sensor data issampled with a sampling rate of fs = 1 kHz and a quantization resolutionof w = 8bits. Thus, during sampling, br = fs · w = 1 kBytes/second of rawdata are being produced. When reducing the sampling period to one minuteper hour, the average required bandwidth is only br

60 = 16Bytes/second; therequired bandwidth could be reduced by the same factor as the sampling timewas reduced. The advantage of this method lies in its exact predictability ofthe bandwidth requirements, comparable to measurements with much lowerconstant sampling rates.

In PermaSense, the current eorts to integrate GPS measurements are basedon a duty-cycling approach (c.f. [6, 7]).

time

sampling periods

Figure 4.1: Basic principle of sampling time reduction: Periodical short periods ofsampling reduce the average data rate.

26

4.3. DIRECT DATA AGGREGATION

4.2.2 Raw Data Classication

Although providing excellent predictability of bandwidth requirements, sam-pling time reduction is only suitable for special measurement cases. Often,it is unknown in advance when data should be sampled, as the processes ofinterest (e.g., AE activity) occur spontaneously and would not be capturedby periodic short measurement sequences. Thus, as already depicted in Sec-tion 3.3, the amount of transmitted raw data may also be reduced by somesort of data classication, basically deciding for each sample whether it isworthy to be sent over the network or not. E.g., in the case of AE measure-ments, this decision process is fairly simple, as all samples that belong toan event are forwarded and all others are discarded. Similar simple decisionprocesses might be dened for other phenomena, minimizing the requiredon-node processing power.

time

continuous sampling aperiodic interesting sequences

Figure 4.2: Basic principle of raw data classication: Data is sampled continuously,but only interesting sequences are forwarded.

The network bandwidth required with this approach can generally only beestimated, as it depends on the fraction of samples that need to be forwarded,which is unknown for spontaneous processes. With pf denoting the expectedfraction of samples to forward and br denoting the raw data rate being pro-duced when sampling, the required network bandwidth is equal to br · pf .For the specic case of AE sensing, a more detailed example is presented inSection 3.3.

4.3 Direct Data Aggregation

As already discussed in Section 3.3, for certain applications, reducing theamount of transmitted raw data is still not sucient in order to meet givennetwork bandwidth constraints of a low power WSNs. Either valuable datawould be discarded, or the WSN would quickly suer congestion. Thus,for such applications it is desirable to perform substantial parts of the dataprocessing directly on the sensor node itself, aiming to reduce the amountof transmitted data as much as possible without discarding information ofinterest. Basically, no raw samples are transmitted over the network at all,but merely meaningful aggregates thereof that can be used for further anal-ysis. In the case of AE sensing for example, the events may be detected and

27


parametrized directly on the sensor nodes, leaving only the AE parametersto be transmitted over the WSN.

Again, as the processes of interest are assumed to occur spontaneously, therequired network bandwidth can only be estimated based on expected ac-tivity rates. A bandwidth calculation example for the case of AE sensing ispresented in Section 3.3. Naturally, direct data aggregation especially makessense if local processing can reduce the data by multiple orders, as it is thecase for AE event detection and parametrization.

time

continuous sampling aperiodic interesting sequences

direct data aggregation

Figure 4.3: Basic principle of direct data aggregation: Data is sampled continu-ously, interesting sequences are rst aggregated and then forwarded.

In order to eectively employ such an approach, a precedent solid under-standing of the phenomenon studied is required: Opposed to the analysisof raw data that has arrived at the backend, data aggregation happens veryearly in the process. This prohibits the application of any other form ofanalysis on the samples measured than the predened aggregation functions.Nevertheless, for certain high data rate applications, this may be the onlyapproach to make a successful WSN integration feasible.

However, the low power optimization trend in WSN development did notonly concern the operating mode of the sensor nodes: Also the sensor nodehardware itself is optimized to operate very power eciently. This generallyalso implies that the processing capacity of most sensor nodes is very limited,tailored to sense and forward data and at most perform some very basicaggregation operations. In the context of direct data aggregation for highdata rate measurements, common sensor nodes simply lack the computationpower to perform the necessary signal processing tasks. On the other hand,building a heterogeneous WSN that integrates commodity sensor nodes aswell as higher performance platforms presumably implies a lot of hassle withcross-platform compatibility issues, complicating network management andmonitoring. As an approach to address this dilemma, we propose the conceptof "processing sensors".

28

4.3. DIRECT DATA AGGREGATION

4.3.1 Processing Sensors

The concept of processing sensors basically attempts to insert an additionallayer of abstraction between data acquisition and data transmission, signi-cantly reducing the amount of data to forward and thereby virtually reducingthe sensor's data rate. This is achieved by adding an additional processingelement between the physical sensor and the sensor node platform, servingas data acquisition controller, data processing instance and sensor data in-terface at the same time. Ideally, this data processing platform keeps its ownstate and is only loosely coupled to the sensor node, separating concerns asmuch as possible in order to allow virtually independent operation. Figure4.4 illustrates this basic principle:

sensor

common low data rate architecture

data

control interfacelow power

sensor node platform

processing sensor architecture

data

control interface

low power sensor node

platform

performantdata processing

platform

control interface

sensor

processing sensor

radio

radio

data

Figure 4.4: The processing sensor architecture. From a sensor node platform's per-spective, a processing sensor does not dier much from any other lowdata rate sensor.

The approach of processing sensors oers the following advantages over in-tegrating higher performance computing platforms directly into the WSN:

Manageability: Deploying and operating a WSN reliably is a diculttask. It becomes even harder, if the network consists of a heteroge-neous mix of dierent platforms. Relying on a single WSN optimizedplatform therefore eases network management signicantly.

Robustness and speed of development: A command based pointto point data interface is simpler to implement and less prone to cross-platform compatibility issues than a full WSN network stack.

Modularity: As long as the data interface between the sensor nodeplatform and the processing element stays unaltered, the two compo-

29


nents may be developed and improved independently. Changes in theWSN platform code or hardware do not aect the processing sensorplatform and vice versa.

Predictability: As each platform keeps and manages its own state,there's virtually no interference between data acquisition/processingand network communication. This makes it easier to predict data ac-quisition and processing performance as it is not inuenced by networklink quality and other factors related to communication.

Naturally, introducing a second processing element on a sensor node gener-ally increases its power consumption. It is therefore important to carefullytrade computation power (i.e., power consumption) versus the amount ofdata to transmit (i.e., network bandwidth requirements) for each individualapplication.

30

5AEBoard Hardware Design

This chapter presents an AE sensing hardware platform called the AEBoard,custom designed for performing AE measurements in PermaSense (c.f. Chap-ter 3).

5.1 Basic System Architecture

The AEBoard's system architecture basically follows the approach of a pro-cessing sensor (c.f. Section 4.3.1): An intelligent sensor reduces the amountof data to transmit drastically, such that existing WSN hard- and softwarecan be reused as if a low-data rate sensor was interfaced.

Specically, the AEBoard features a TinyNode WSN platform as systemmaster, controlling all sensors as well as taking care of the WSN communi-cation. The TinyNode receives AE parameter data over its serial interfacefrom a processing AE sensor, consisting of a slave microprocessor and the ac-tual AE sensing circuitry. Furthermore, two SD memory cards are included:One SD card is attached to the TinyNode to serve as network packet back-log memory in case of missing network connectivity or AE data bursts. Theother SD card may be used by the slave processor to store raw AE data of oc-curring events (c.f. Section 3.4). Figure 5.1 sketches this basic architecture.

Additionally, a combined sensor for temperature and humidity has been at-tached to the TinyNode for system health monitoring as well as measurement

31

CHAPTER 5. AEBOARD HARDWARE DESIGN

Sensor 1

Sensor 2

ADC

Am

pA

mp

ADC

SlaveProcessor MSP430

TinyNode

Rad

io

SDCard

SDCard

Figure 5.1: AEBoard architectural concept

calibration. The TinyNode also monitors several voltage levels and supplycurrents with its internal ADCs and is able to power cycle the slave processor(c.f. Section 5.4).

5.2 AE Sensor Key Components

As no similar measurements have been done in PermaSense so far, theAEBoard's processing sensor has been designed from scratch. The follow-ing sections describe the various considerations that were taken into accountwhen selecting the sensor's key components.

5.2.1 Slave Processor

The slave processor platform is the heart of a processing sensor: It controlsthe data acquisition (DAQ) process, performs signicant parts of the dataanalysis and interfaces the actual WSN platform to forward data. Further-more, the AE sensor slave processor has to store raw event data on its SDcard. Thus, besides being able to perform the necessary signal processing inreal time, the slave processor must be able to coordinate multiple internaland external data streams, as Figure 5.2 illustrates.

Therefore, a key requirement for the slave processor platform are powerfulcommunication interfaces and a memory architecture that oers enough sizeand bandwidth to serve all communication interfaces virtually in parallel.Ideally, the slave processor oers true parallelism for certain tasks (e.g.,parallel communication and signal processing).

Based on these requirements, candidate devices for the slave processor havebeen evaluated. Three platform variants have been considered:

Digital signal processor (DPS) platforms

32

5.2. AE SENSOR KEY COMPONENTS

Raw Samplesfrom Sensor Channel 2

Raw Samples to SD Storage

Parametersto TinyNode

Raw Samplesfrom Sensor Channel 1 Slave Processor

Input Buffer

Storage Buffer

Parameter Buffer

Figure 5.2: Slave processor data streams.

Programmable logic devices, i.e., several PLDs and FPGAs

General purpose microprocessor platforms (e.g., ARM Cortex basedplatforms)

Table 5.1 compares the three most promising devices (one of each platformvariant) in detail.

Out of the three components presented, the ARMCortex M3 based STM32F103RDhas been selected for the following reasons:

The STM32 oers comparably many communication interfaces, all ofthem oering direct memory access (DMA) capabilities. This allowsthe processing core to spend more capacity for signal processing insteadof managing communication tasks.

The included 64 kBytes SRAM oers enough space for buering thevarious data streams.

Native SDIO support oers high writing performance to the SD card.

An FPGA based solution would have oered more exibility, true hard-ware parallelism as well as assumably lower power consumption. How-ever, there is no FPGA know-how in PermaSense so far. Thus, as theAEBoard shall be productively integrated into PermaSense as soon aspossible, a microprocessor based approach seemed to be more compat-ible with further project developments.

The STM32 microprocessors are well available at comparably low prices.

33


ADSP2186M IGLOO AGL600 STM32F103RD

GeneralType DSP FPGA Microprocessor

Manufacturer Analog Devices Actel STMicroelectronics

CPUCore ADSP-2100 ARM Cortex M1 ARM Cortex M3fmax 75MHz 60MHz 72MHz

Bus width 16 bits 32 bits 32 bitsClocking external external / external external

MemoryRAM type SRAM Dual-Port SRAM SRAMRAM size 40 kBytes 13.5 kBytes 64 kBytesROM type n/a Flash FlashROM size n/a 1 kBit 384 kBytes

PowerSupply voltage 2.5V - 3.6V 1.2V - 1.5V 2.0V - 3.6V

Low power modes 1 2 4Wakeup source n/a ext Pin ext Pin

Est. Power at 60MHz 20mW 15mW 20mW

ConnectivitySPI ports 2 congurable 3USARTs 0 congurable 5

SDIO 0 congurable 1max GPIOs 8 235 51

MiscellaneousOperating temp. -40C to +85C -40C to +85C -40C to +85C

Price per unit 16CHF 75USD 12CHF

Table 5.1: Comparison of three potential slave processor platforms. All technicalvalues base on the corresponding datasheets [45, 46, 47].

Some preliminary tests with the STM32Discovery evaluation kit further ver-ied that the system specications can be met with the STM32 slave pro-cessor.

34

5.3. INPUT SIGNAL CONDITIONING CIRCUITRY

5.2.2 ADCs

When digitally capturing and processing analog data, the ADC naturallyheavily inuences the measurement quality achieved. Thus, choosing anappropriate ADC is crucial for good measurement quality. For the AEBoard,besides oering a quantization resolution of 16 bits at a sampling rate of500 kHz, the ADCs have to operate between -40C to +85C and oer lowpower consumption.

Many available low-power ADCs either only oer a resolution up to 12 bits ordon't achieve the sampling rates required. While for audio applications withsampling rates up to 96 kHz a broad variety of 16 bit ADCs exists, productsin the 16 bit / 500 kHz class are rather rare. Finally, two candidate ADCscould be identied, which are compared in Table 5.2.

ADS8327 AD7980

GeneralManufacturer Texas Instruments Analog Devices

ConversionMaximal sampling rate 500 kHz 1000 kHz

Resolution 16 bits 16 bits

MiscellaneousTyp. power dissipation at 500 kHz 10.6mW 3.5mW

Operating temp. -40C to +85C -40C to +125CInterface SPI SPI

Price per unit 30CHF 40CHF

Table 5.2: Comparison of two potential ADCs. All technical values base on thecorresponding datasheets [48, 49].

Because of its lower power consumption as well as the possibility to increasethe sampling rate above 500 kHz if necessary, the AD7980 ADC from AnalogDevices has been chosen for the AEBoard. Its output format is compatibleto the well-known SPI protocol, which is well supported by the STM32.

5.3 Input Signal Conditioning Circuitry

Before sampling the AE sensor's output voltage, several signal conditioningtasks have to be performed in order to ensure optimal sampling quality:

Filter the input signal with a bandpass lter in order to minimize

35


unwanted noise.

Adapt the signal's voltage range in order to utilize the ADC's fulldynamic range.

Drive the ADC with a low-impedant signal source for maximal quan-tization accuracy.

Piezoelectric sensors generally produce relatively weak output signals thatare not suitable to drive cables longer than a one or two meters. Therefore,the AE signals have to be amplied very close to the sensor. The PhysicalAcoustics Corporation oers both active sensors that already have the neces-sary amplier built into the housing (e.g., the PK6I sensor [50]) and passivesensors that need to be connected to an in-line amplier after at most twometers of cable (e.g., the R6α sensor [41] together with a IL-LP-6S preampli-er). In both cases, the preamplier is supplied with electric power over thesignal line by lifting it to a DC bias voltage of 4 to 7V. The sensor outputsignal then rides on this supply voltage, featuring a maximal amplitude thatis equal to the supply voltage.

On the other hand, the AD7980 ADC requires the constant reference voltagesVref and GND. It then samples input voltages between GND and Vref witha resolution of 16 bits. Input voltages smaller than GND or higher than Vrefare saturated at the respective output value [49].

Thus, the signal conditioning circuitry must bandpass the input signal to thespecied frequency range of 20 kHz to 100 kHz, and adapt its voltage rangesuch that the maximal ADC input voltage lies at Vref and its minimum liesat GND potential. For this purpose, an active lter has been designed, usingthe ADA4841 operational amplier by Analog Devices [51] which is recom-mended to drive the AD7980 ADC. The DC supply voltage is removed at thelter input by a serial capacitor, whereas a 50Ω serial resistor decouples thesignal from the supply voltage regulator. The signal conditioning circuitry'ssimplied schematics are shown in Figure 5.3.

Circuit simulation with LTSpice [52] was used to adequately dimension thepassive components. Figure 5.4 shows the simulated lter frequency re-sponse.

For this simulation, the input source was simply modelled as an AC sourcewith an oset voltage of 4.5V instead of exactly modelling the sensor /preamplier source. Thus, the 4.5V bias voltage aects the input signal,resulting in an overall -6 dB damping in the simulation results. The LTSpicesimulation les are also contained on the enclosed CD (c.f. Appendix G).

36

5.3. INPUT SIGNAL CONDITIONING CIRCUITRY

ADC

Am

p+

-

+4.5V

+4.5V

Input

100k

100k

5.1k

5.1k330R

2.2n

6.8n50R

4.5V

max 6.75V

min 2.25V

0V

max 2.25V

min -2.25V

2.25V

max 4.5V

min 0V

Figure 5.3: Simplied schematics of the signal conditioning circuitry for one AEchannel

Figure 5.4: Signal conditioning circuitry frequency response, simulated with LT-Spice. The solid line shows the lter amplication, which is relevantto dene the usable frequency range. The dotted line shows the lessimportant lter phase.

37


5.4 Power Concept

The AEBoard features various components requiring electrical power supply.In order to minimize the overall power consumption, the input voltage foreach component should be as low as low as possible in it's specied supplyrange. Thus, power consumers on the AEBoard have been divided into thefollowing supply voltage categories:

4.5V: The AE sensors need a supply voltage Vs of at least 4V, ad-ditional 0.5V have been added to provide enough headroom. Conse-quently, the operational ampliers also have to be powered with thesame voltage in order to support the required output range of 0V toVs.

2.8V: SD cards require a minimum supply voltage of 2.7V [53]. Adding0.1V of headroom, the STM32 as well as the TinyNode need to be pow-ered with at least 2.8V in order to meet the requirements for SD cardhosts.

2.5V: The AD7980 ADC requires a supply voltage of exactly 2.5V.Thus, it requires a separate power supply circuit. Furthermore, it alsorequires a reference voltage Vref that is equal to the maximum amplieroutput, i.e., 4.5V, as well as a digital input / output level voltage VIOof 2.8V.

Additionally, in order to gain ner control over the AEBoard's operation andsave power when necessary, the three following possible operating modes havebeen dened:

Standby: Only the TinyNode's power supply is turned on. The slaveprocessor as well as the DAQ circuitry are cut from power supply com-pletely.

Data transfer: In addition to the Standby mode, the slave processoris supplied with power. This mode allows for data exchange betweenthe TinyNode and the STM32, but no AE data can be sensed.

AE sensing: All components are powered and operational.

Consequently, as the actual sensing circuitry shall be completely hidden fromthe TinyNode, it only controls the STM32's power supply. The slave pro-cessor in turn controls all DAQ components' supply and reference voltages.

In order to ensure optimal measurement quality and minimize coupling be-tween the two channels, each input channel is biased by a separate voltage

38

5.5. AEBOARD PROTOTYPE

regulator. Furthermore, the ADC reference voltage as well as the ampliersupply voltage are generated by distinct sources.

Initial estimates assumed a constant operating current of about 40mA in theAE sensing mode. As even high capacity batteries would discharge in a fewdays under this load, the AEBoard is designed to be fed by a solar powersupply, providing an input voltage of 12V. For power eciency reasons,these 12V are converted to an intermediate voltage of 5.5V by a switchedcapacitor voltage regulator. This intermediate voltage is then converted tothe required voltage levels by linear regulators. Figure 5.5 sketches thispower concept.

Due to a voltage drop of approximately 0.5V at the power switch transistorsused, the TinyNode and its SD card are fed with a supply voltage of 3.3V toensure the required 2.8V for the STM32 after the power switch. For systemhealth supervision purposes, voltage and current are measured at variouspoints: Voltages and currents in the prescaling and digital master domainare sensed by the TinyNode, voltages in the sensing domain are supervisedby the slave processor.

5.5 AEBoard Prototype

The hardware concept described has been implemented in an AEBoard pro-totype. In addition to the features already mentioned, the following compo-nents and features have been added:

JTAG programming and debugging ports for both the TinyNodeand the STM32. The 20 pin port for the STM32 is thereby compatiblewith the ST-LINK programmer by STMicroelectronics, the TinyNode's14 pin port is compatible with various MSP430 JTAG programmerdevices.

Reset buttons for both the TinyNode and the STM32.

Serial console connector for the STM32 which could be used tooutput debug messages to a PC. It is pin compatible with the TTL-232R-3V3 USB TTL serial cable from FTDI chip [54].

Three debug LEDs for the STM32.

Serial communication debug LEDs that indicate data ow be-tween the TinyNode and the STM32.

Two shared GPIO connections between the TinyNode and theSTM32 in addition to the serial interface.

39


+12V

switch

ed

12

V to

5.5

Vlin

ea

r5

.5V

to 3

.3V

TinyN

ode

SD

ST

M32

SD

+5.5V

+3.3V

reference2.8V

to 2.5VA

DC

1

+2.5V

VD

D

+2.8V

(VIO

)AD

C2

+2.8V

(VIO

)Am

p1

Am

p2

Sensor1

Sensor2

reference5V

to 4.5V

reference5V

to 4.5V

reference5V

to 4.5V

+4.5V

+4.5V

+4.5V

Prescalin

g D

om

ainD

igital M

aster Do

main

Dig

ital Slave D

om

ainS

ensin

g/C

on

version

Do

main

reference5V

to 4.5V

+4.5V

(RE

F)

current sense

current sense

voltage sensevoltage sense

voltage sense

voltage sense

voltage sense

voltage sense

voltage sense

voltage sense

Po

wer M

od

es:

Stan

db

y: S

1, S2, S

3 open D

ata transfer: S

1 closed, S2 &

S3 open

AE

sensin

g: S

1, S2, S

3 closed

Po

wer M

od

es:

Stan

db

y: S

1, S2, S

3 open D

ata transfer: S

1 closed, S2 &

S3 open

AE

sensin

g: S

1, S2, S

3 closed

S1

S2

S3

current sense+

2.8V

+5V +

2.8V

Figure 5.5: AEBoard power concept

40

5.5. AEBOARD PROTOTYPE

Power Supply

Analog part ADCs Slave Processor STM32

Master Processor TinyNode

SD Cards

Signal Inputs

Figure 5.6: The AEBoard prototype.

Switchable boot mode for the STM32 to recover the processorfrom corrupted program memory by booting it from the factory boot-loader.

Switchable processor supply voltage to 2.8V or 3.3V. This featurehas been removed for further hardware revisions, as a supply voltage of3.3V is required for proper functionality (c.f. Sections 5.4 and 5.5.1).

ESD protection diodes at the signal inputs.

Test pins at various points to measure supply and reference voltagesmanually.

For detailed design schematics and PCB layout gures of the so far manu-factured hardware revision 1.1, please refer to Appendix E.

5.5.1 Changes for Revision 1.2

Although the rst AEBoard prototype basically worked as expected, a fewchanges have been made for further hardware revisions (the component des-

41


ignators refer to the respective schematics). These changes have alreadybeen included in the latest schematics and layout les that may be found onthe enclosed CD (c.f. Appendix G) or in the PermaSense SVN repositoryunder /trunk/pcb/projects/0011_acoustic_emission.

Pins 5 and 13 of the STM32's JTAG interface have been interchangedin order to allow JTAG access. This bug could be circumvented byeither patching the access ribbon cable or using the ARM specic singlewire debug (SWD) protocol instead of JTAG for programming anddebugging.

The current supervision ICs' supply voltage has been corrected from5.5V to 3.3V. As a consequence of the wrong supply voltage, thecurrent supervision circuits in revision 1.1 are not usable.

An LED (D10) has been added to indicate the the presence of supplypower.

A pull-down resistor (R77) has been added to the STM32 reset cir-cuitry, preventing spontaneous resets due to a oating reset controlline.

The TinyNode's supply voltage has been xed to 3.3V by removingjumper W2 as well as resistor R20, as an output voltage of 2.8V at thelinear voltage regulator U9 is not sucient for powering the STM32'sSD card after the voltage drop of 0.5V at the switching transistor (c.f.Section 5.4).

42

6PermAE Software Design

This chapter describes a distributed piece of software called PermAE, thatruns on the AEBoard and meets the system specications described in Chap-ter 3.

6.1 General Concept

Consequently following the principle of a processing sensor (c.f. Chapter4), PermAE is split into two virtually independent parts: One part is runningon the STM32 slave processor and is responsible to control and supervise thedata acquisition process, process the raw data, forward detected AE eventparameters as well as status messages to the TinyNode and store raw eventsamples on its SD card. The other part of PermAE runs on the TinyNodeand basically controls the slave processor, manages the WSN connection,forwards received AE parameter and status messages to the data sink andmonitors air temperature and humidity. The two devices communicate overa UART port by a set of well dened commands and messages as presentedin the following section. This serial port, together with the TinyNodes ca-pability of power cycling the STM32, forms the only interface between thesetwo components, allowing to develop and test many of the respective internalmatters independently (c.f. Figure 5.1).

43

CHAPTER 6. PERMAE SOFTWARE DESIGN

6.2 PermAE Messages

As producing data messages is the main purpose of a sensor node in PermaSense,the expected output messages are specied rst. Note that from an enduser's perspective, it does not matter whether a message was produced bythe STM32 or by the TinyNode. Rather, a user should not even notice thattwo independent processors are involved.

6.2.1 PermAE Output Messages

Table 6.1 briey outlines the messages generated by PermAE and nallystored in the backend database.

Message Description Contents

aedata Information about onemeasured AE event.

Occurrence time infor-mation, event parame-ters, posttrigger used,threshold used and mea-surement channel

aestatistics Event counter values,aggregated over twominutes.

Total events mea-sured, total eventsparametrized, totalraw events stored (c.f.Section 3.4)

aedaqhealthdata AEBoard health andsystem status data.

Monitored supply volt-ages and currents,monitored referencevoltages, number of freeraw event blocks on theSD card, power state ofthe STM32 (on / o).

nodehealth PermaDozer standardmessage representingthe TinyNode's status

Packet count, systemvoltages, air temper-ature and humidity,SD card status andTinyNode uptime.

Table 6.1: PermAE output messages

Additionally, further standard PermaDozer messages (eventlogger, rssi, state-counter) are generated. For detailed format specications, refer to AppendixC. The values' units are described in detail in Appendix D.

44

6.3. ONBOARD COMMUNICATION PROTOCOL

6.2.2 PermAE Dozer Commands

According to the functional system specication in Section 3.4, thresholdand posttrigger values must be adaptable from the backend. Furthermore,data acquisition (i.e., the whole processing sensor) may be power cycledmanually. To this end, PermAE uses the Dozer beacon mechanism to receivecommands. As the maximum number of Dozer commands is limited to 16per deployment, only one new command type (AEBOARD_CTL_CMD)has been dened. The command argument then encodes the message thatis sent to PermAE as specied in Table 6.2:

Command Description Argument to send

DAQ_OFF turn data acquisition o 256DAQ_ON turn data acquisition on 512SET_THR set the threshold value 1024 + threshold [dB]SET_PTRG set the posttrigger value 2048 + posttrigger [samples]

Table 6.2: Dozer command messages. To set a threshold of, e.g., 53 dB, the argu-ment must be set to 1024 + 53 = 1077. To set a posttrigger value of,e.g., 600 samples, the argument must be set to 2048 + 600 = 2648.

Note that the threshold value is limited to 7 bits and therefore to a maxi-mum value of 127 dB. Likewise, posttrigger values are limited to 11 bits or amaximum value of 2047 samples.

The Dozer commands can easily be sent to a node running PermAE byutilizing the dozer_command virtual sensor in the PermaSense webinterface(c.f. Section A.5).

6.3 Onboard Communication Protocol

As already outlined, the two parts of PermAE communicate over a serialUART interface. Communication is achieved by a simple command basedprotocol. Generally, every communication round is started by the TinyNodesending a command. The STM32 may then respond according to the fol-lowing protocol. In order to request an exchange of messages (e.g., to for-ward AE event parameters to the TinyNode), the STM32 raises one of theshared GPIO pins (called the ARM_COMM_FLAG) and waits for theTinyNode to start communicating. All predened commands and messageshave a length of 1Byte.

The onboard communication protocol uses a simple acknowledgement scheme.Table 6.3 denes the messages used. If any party sends a NACK message

45


at any point in a message exchange, the communication ow is aborted andthe protocol must be reset on both sides.

Hex ASCII Description

0x41 A acknowledge (ACK)0x4E N not acknlowledge (NACK)

Table 6.3: Onboard ACK and NACK messages

To initiate a communication round, the TinyNode sends one of the com-mands specied in Table 6.4.

Hex ASCII Description

0x53 S start data acquisition0x4F O stop data acquisition0x45 E set threshold value0x49 I set posttrigger value0x4D M send a message

Table 6.4: Onboard command messages

When requested to send a message, the STM32 responds with a messageindicating the data message type to follow. Table 6.5 species the possibletypes.

Hex ASCII Description Size Dozer equivalent

0x45 E AE event data 23Bytes aedata0x53 S AE statistics 6Bytes aestatistics0x48 H health data 14Bytes n/a

Table 6.5: Onboard data message types

AE event data messages and AE statistics messages are equivalent to therespective Dozer messages and may be directly forwarded by the TinyNode.The health data is rst extended by additional values measured by theTinyNode and only then forwarded in an aedaqhealthdata message (c.f.Appendix C).

The possible valid communication ows are nally specied in Figure 6.1.

46

6.4. STM32 IMPLEMENTATION

TinyNode STM32Start / stOp DAQ

start / stop DAQAck

TinyNode STM32set thrEshold

Ack

threshold [1 Byte]

Ack

TinyNode STM32 TinyNode STM32send Message

Ack

set posttrIgger

Ack

posttrigger lower byte

Ack

stop DAQset thresholdstart DAQ

posttrigger upper byte

Ackstop DAQset posttriggerstart DAQ

AE Event /AE Statistics /system Health

Ack

message

Figure 6.1: Valid onboard message ows

Both parties must send a NACK message if either an invalid command /message was received or if an expected response has not arrived within 50ms.The protocol state must then be reset on both sides and communicationrestarts.

6.4 STM32 Implementation

The STM32 slave processor is mainly responsible for acquiring and process-ing AE sensor data as well as forwarding detected event parameters. Fur-thermore, as many raw events as possible shall be stored on the SD cardattached and the DAQ system's health shall be supervised regularly. As allthese tasks must virtually run in parallel, the FreeRTOS real time operat-ing system kernel [55] has been utilized to implement this part of PermAEeciently.

FreeRTOS features a fully preemptive task model with execution timeslotsof a xed maximum length. Tasks preferably communicate over FreeRTOS'FIFO queues. Reading from an empty queue or writing to a full queue mayor may not block the respective task depending on the programmer's deci-sion. When either an execution timeslot has expired or after a read or write

47


operation on a queue, the scheduler gives control to the highest priority taskthat is not blocked, explicitly delayed or suspended. For more informationabout the FreeRTOS execution model, please refer to the FreeRTOS docu-mentation [56, 57] on the enclosed CD.

The STM32 application has been split into the following tasks:

The data analysis controller task enables and disables the data ac-quisition process. It however only does so when receiving an accordingcommand from the TinyNode via the communication task.

Two data analysis tasks analyze the incoming raw data samples,detect AE events and parametrize them. Each data analysis task isresponsible for analyzing one input channel's data.

The communication taskmanages the serial interface to the TinyNode.It therefore implements the communication protocol specied in Sec-tion 6.3.

The storage manager task manages the SD card resource and writesincoming raw event sample buers onto the SD card.

The event statistics task reads and resets the event counter valuesevery two minutes and generates the according aestatistics messages.Likewise, the voltage supervision task measures the supervised sys-tem voltages with the internal ADCs and generates the system healthmessages every two minutes.

Tasks are listed according to their assigned priorities, i.e., the data analysiscontroller tasks has the highest priority and suspends all other tasks, whereasthe event statistics and voltage supervision tasks have the lowest priorities.These task priorities directly reect the output data priorization specied inSection 3.4: Data acquisition and event count is done with highest priority,forwarding of parameter is done with medium priority and storage of eventson the internal SD card is done with low priority. System supervision tasksare nally carried out with lowest priority as they are not critical for the de-tection of AE activity. Figure 6.2 depicts the PermAE application structureon the STM32.

For eciency reasons, all necessary data buers are allocated statically dur-ing initialization. Thus, the tasks only pass pointers to these static buersvia FreeRTOS queues. Consequently, PermAE also requires feedback queues,containing pointers to empty / processed buers that may be lled with newdata. These feedback queues have been omitted in Figure 6.2 for simplicity.Table 6.6 displays PermAE's memory partition.

48


Dat

a A

naly

sis

Cha

nnel

2

Prio

4

QS

D C

ard

Q

Q

Inte

rrup

tS

ervi

ceR

outin

e raw

dat

a

Q

DA

Q v

ia D

MA in

terr

upt

Tin

yNod

e

AD

C1

AD

C2 SP

I

SP

I

task

s

Qm

essa

ge q

ueue

s

perip

hera

ls

Q

Dat

a A

naly

sis

Cha

nnel

1

Prio

4

raw

dat

a

Q

Sto

rage

Man

ager

Prio

2

Com

mun

icat

ion

Prio

3S

DIO

even

tra

w d

ata

even

tpa

ram

s

Eve

ntS

tatis

tics

Prio

1

Vol

tage

Sup

ervi

sion

Prio

1

heal

thm

essa

ges

stat

istic

sm

essa

ges

Dat

a A

naly

sis

Con

trol

ler

Prio

5

QS

TAR

T /

ST

OP

com

man

ds

enab

le /

dis

able

data

acq

uisi

tion

UA

RT

Inte

rrup

tS

ervi

ceR

outin

e

QUA

RT

mes

sage

s /

com

man

ds

Figure 6.2: PermAE application structure on the STM32

49


Size [kB] Description # of Buers

5 Task Stacks n/a36 Raw Data Input Buers 189 Parameter Buers 4004 Storage Buers 27 Queues n/a

Table 6.6: PermAE memory partition. Total memory size is 64 kB.

6.4.1 PermAE Initialization

Before data acquisition and analysis may start, the PermAE applicationmust be initialized. An initialization task (not included in Figure 6.2) takescare of allocating buer memory and initializing all required data structures.

In order to ensure correct initialization after power cycling the STM32,PermAE stores some information in an initialization block on the SD card.This block is initially generated by the PermAEInit application (c.f. Section6.4.8). Table 6.7 species its format and contents.

The PermAE initialization task starts by reading this initialization block andchecking the magic bytes as well as the device ID in order to ensure thatthis SD card has been initialized with the current AEBoard. It then usesthe rest of the initialization block's information to prepare data acquisition,analysis and storage. Finally, after all other tasks and data structures havebeen created successfully, the initialization task suspends itself forever.

For debugging purposes, the initialization task features verbose status mes-sages on the AEBoard's debug console connector, which can be checked byconnecting the AEBoard to a terminal emulation software such as, e.g., mini-com (c.f. Appendix A).

6.4.2 Data Acquisition

The AEBoard connects the AD7980 ADCs to two of the STM32's SPI in-terfaces. The ADCs' operating mode requires their controller to initiatesampling by pulling their CNV input to high, waiting for a xed amount oftime and then releasing CNV again. The sample may then be read by gen-erating 16 clock pulses on the SCK input and reading the output line SDOat each raising clock edge [49]. In PermAE, both AE channels are sampledsynchronously. The conversion clock, as well as the SPI clock, are generatedby two timer peripherals on the STM32. Consequently, the SPI peripheralsare congured to operate in slave mode. Figure 6.3 illustrates this wiring.

50


Oset Size [Bytes] Name Description

0x00 4 magic bytes The static string AEP0x04 12 device ID The STM32's unique de-

vice identier0x10 4 serial number The current AE event

counter value0x14 4 blocks stored The number of raw

events stored on the SDcard

0x18 2 stored event size The size of one raw eventstored on the SD card

0x1A 2 stored event samples The number of raw sam-ples per stored event

0x1C 2 channel 1 oset Oset calibration valuefor channel 1 (c.f. Sec-tion 6.4.8)

0x1E 2 channel 2 oset Oset calibration valuefor channel 2 (c.f. Sec-tion 6.4.8)

0x20 2 threshold The event detectionthreshold

0x22 2 posttrigger The event detection pos-sttrigger value

0x24 476 reserved Reserved for future use

Table 6.7: The PermAE initialization block

STM32

SPI1

ADC2 SPI2

MOSISDO

MOSISDO

SCK SCK

SCK SCK

SCK Timer

CNV Timer

ADC1

CNV

CNV

Figure 6.3: AEBoard data acquisition wiring

When data acquisition is enabled by the data analysis controller task, the

51


CNV timer generates a periodic acquisition pulse with a frequency of 500 kHz.The SCK timer basically generates a continuous clock signal, but it's outputsignal is masked by a windowing timer such that per sampling period, only16 well-timed pulses appear at the SCK output.

t

2us

CNV

SCK window

SCK internal

SCK output

Figure 6.4: Generated data acquisition clocks

The received samples are transferred to a raw data buer via DMA, onlygenerating an interrupt when the raw sample buers are full (i.e., after read-ing 1024 samples) and thereby minimizing unnecessary interrupt overhead.An interrupt service routine re-initializes the DMA controller with new in-put buers and passes the full buers to the data analysis tasks (c.f. Figure6.2). Unfortunately, this process takes more than two microseconds andconsequently, one sample per channel is lost at each buer switch. To min-imize the introduced error, the missing sample is estimated by calculatingthe mean value of its preceding and its subsequent sample.

6.4.3 Data Analysis Controller Task

Even though the data analysis controller task is the highest priority task inPermAE, it is generally the least active. Its sole purpose is to enable anddisable the data acquisition process and to thereby indirectly control thedata analysis tasks as outlined in Algorithm 1.

6.4.4 Data Analysis Task

Each AE channel's raw data is analyzed in a separate task. Once a sam-ple's absolute value exceeds the set threshold, an event is detected. Eventparametrization runs until the threshold has not been exceeded for the num-ber of samples that is dened by the posttrigger value. Assuming that therst threshold crossing occurred at sample i and the event ends at samplej (i.e., the last threshold crossing occurred at sample j − posttrigger), Al-

52


Algorithm 1 PermAE data analysis control task

1: initialize data analysis;2: while true do3: wait for DAQ_START;4: initialize data acquisition;5: start data acquisition;6: wait for DAQ_END;7: stop data acquisition;8: end while

Algorithm 2 PermAE event parametrization

1: length := j − i;2: amplitude := 0;3: risetime := 0;4: energy := 0;5: count := 0;6: for k = [i . . . j] do7: if abs( samples[k] ) > amplitude then8: amplitude := abs( samples[k] );9: risetime := k − i;

10: end if11: energy = energy + (samples[k])2

512 ;12: if samples[k − 1] < threshold and samples[k] >= threshold then13: count := count +1;14: end if15: end for

gorithm 2 is equivalent to PermAE's event parametrization algorithm (c.f.Section 2.2.2 and Figure 2.5):

There are two points noteworthy about this implementation:

In line 11, the squared samples are rst divided by a scaling factor of512 before being added to the energy parameter. This scaling preventsenergy parameter values from overowing the assigned range of 32 bits.

Some AE parametrization algorithms use extended criteria to denea threshold crossing in order to suppress high frequency oscillations.PermAE however counts every single exceeding of the positive thresh-old as depicted in Figure 2.5.

53


AE Event Metadata

In addition to the actual event parameters, metadata is added to the eventparameters: A global event counter is incremented every time an event wasparametrized. This counter value is attached to the event data as the event'sserial number. The serial number can only be reset by re-initializing the SDcard. Thus, it uniquely identies every event of a sensor's deployment. Theevent's serial number is also crucial for correlating raw event data from theSD card with parameters stored in the PermaSense database. Furthermore,in order to temporally correlate the AE events precisely (c.f. Section 3.4),a sample counter value indicating the start sample's counter value is addedto each parameter set. E.g., if event A's sample counter value is equal to468194 and event B's sample counter value is equal to 468276, for event A,the rst threshold exceedance occurred 82 samples (=164µs) before eventB's rst threshold exceedance.

As the sample counter value is only 32 bits wide, it overows and restartsfrom zero every 232 samples, which is equivalent to 143minutes at a samplingrate of 500 kHz. It is furthermore reset when data acquisition is stopped andrestarted.

Details about converting the parameter values into physical units are given inAppendix D. Once parameter detection has nished, the detected parametersare passed to the communication task, whereas the event's raw samples andsome metadata are passed to the storage manager task (c.f. Figure 6.2).

6.4.5 Communication Task

The communication task basically implements the communication protocolspecied in Section 6.3. As soon as PermAE on the STM32 is fully initialized,the ARM_COMM_FLAG line is raised to signal communication readiness.Data acquisition is only enabled or disabled upon the accordant TinyNodecommands.

If there are multiple data messages to forward, they are transmitted instrictly prioritized order, depending on their type as shown in Algorithm3.

6.4.6 Storage Manager Task

The storage manager task controls and manages access to the SD card stor-age. The SD card is accessed via the STM32's SDIO peripheral, allowingecient data input / output using bulk DMA transfers.

Raw event data buers are received from the data analysis tasks and written

54


Algorithm 3 PermAE data message transmission

1: if AE parameters waiting then2: transmit AE parameters;3: else if AE statistics messages waiting then4: transmit statistics message;5: else if Health data message waiting then6: transmit health message;7: end if

to the SD card unaltered. For eciency reasons, the raw data samples of allevents are stored in blocks of xed size, which is dened at device calibrationtime (c.f. Section 6.4.8). E.g., if this event buer size is set to 2048Bytes,one raw event block on the SD card would have the following format:

Oset Size [Bytes] Name Description

0x00 4 serial number The event's serial number (c.fSection 6.4.4)

0x04 4 start sample The event's start sample (c.fSection 6.4.4)

0x08 2 length The event's length parameter0x0A 2 threshold The detection threshold used0x0C 2 posttrigger The detection posttrigger value

used0x0E 2032 event data The raw event samples

Table 6.8: Raw event data block on SD card

Thus, maximally 1016 samples per event will be stored. Longer AE eventsare simply cropped to 1016 samples, shorter AE events do not utilize thewhole storage block. Nevertheless, xing a stored event's block size allowsfor very ecient write operations, as SD cards are always written in blocksof 512Bytes. This, however, also limits the possible storage block sizes tomultiples this basic block size.

As no PC readable le system is used to store the raw AE events, the datastored on a PermAE SD card must rst be preprocessed in order to be usedin third party software. For this purpose, a PHP script is provided on theenclosed CD that generates a simple-formatted ASCII le for each eventstored on the memory card. These les may then further be processed withthird party software tools, e.g., with MATLAB. For details, refer to SectionA.4.2.

The storage manager task also periodically updates the PermAE initializa-

55


tion block stored on the SD card at address 0 (c.f. Sections 6.4.1 and 6.4.8),ensuring initialization block actuality and validity even after power cycles.

6.4.7 Event Statistics and Voltage Supervision Tasks

This rather simple tasks generate information blocks in two minute intervals.The event statistics task reads and resets the event counter values (measuredevents, parametrized events, stored events) and generates a statistics messagethereof. Furthermore, the voltage supervision task samples the followingsystem voltages every two minutes (c.f. Figure 5.5):

Bias voltages for channel 1 and 2 (4.5V)

ADC supply voltage (2.5V)

ADC reference voltage (4.5V)

Operational amplier supply voltage (4.5V)

These voltages, together with the number of free event blocks on the SDcard, are forwarded to the communication task as a data acquisition healthmessage.

6.4.8 PermAE Initialization and Calibration

In order to use the STM32 PermAE implementation, some initial systemparameters must be set (c.f. Section 6.4.1):

The initial threshold value

The initial posttrigger value

The block size of one raw event data block (c.f. Section 6.4.6)

Furthermore, the AE measurement channels need to be calibrated: Foreach channel, the input signal theoretically features a static oset of ex-actly 2.25V, which would correspond to an ADC output value of 32768 (c.f.Figure 5.3 and Appendix D). However, due to component inaccuracies, thisoset voltage varies for every AEBoard input channel, making a preliminarycalibration necessary.

For calibration and initialization, a separate application called PermAEInitis provided. PermAEInit measures each AE channel's oset value by sam-pling 1024 samples and calculating the mean value. Therefore, it is important

56

6.5. TINYNODE IMPLEMENTATION

PermAE Application

PermaDozer /DataControl

AEModule

AEBoard Drivers

Radio

SD card

Dozer commandspackets

STM32

STM32 commands

TinyOS

SHT21

ADCs

packets

system voltages

temperature / humidity

UART

DatasourceAcoustic Emissions

DatasourceAEBoard Health

data messages

ADCssystem voltages

data packets

data packets

Figure 6.5: PermAE application structure on the TinyNode

to remove the AE sensors before calibrating to make sure that no unwantedsensor stimulation occurs. After measuring the channel osets, PermAEInitgenerates the initialization block and writes it onto the SD card (c.f. Section6.4.1).

The mentioned detection and storage parameters are compiled statically intothe application's binary, requiring recompilation of PermAEInit to changethem (c.f. Appendix B). For further information about initializing an AEBoard,please refer to Appendix A.

6.5 TinyNode Implementation

In contrast to the implementation on the STM32, the TinyNode part ofPermAE is built upon an already existing software stack consisting of theTinyOS operating system [25] and the PermaDozer network stack [4, 2].Building on this existing infrastructure, the PermAE TinyOS applicationonly has to implement the communication protocol specied in Section 6.3and provide data source modules as well as a hardware description layer forthe AEBoard. Figure 6.5 depicts the application's basic architecture.

All in all, much of the necessary code could be reused from other PermaSenseapplications, greatly facilitating a seamless integration into the existing PermaSenseinfrastructure.

57


6.5.1 AEBoard Drivers

The AEBoard driver layer abstracts the AEBoard hardware resources: SDcard and SPI bus, the UART interface to the STM32, the MSP430 ADCsfor measuring system voltages and currents as well as the SHT21 temper-ature and humidity sensor [58]. As most of the components attached tothe TinyNode have already been used on other PermaSense platforms (i.e.,the sensor interface board (SIB) and the PermaDozer base board), most ofthe hardware layer code could be reused from the already existing codebasein a slightly adapted form. Comparably lightweight datasource plugins forthe PermaDozer DataControl module implement the actual measurementroutines.

6.5.2 AEModule

The AEModule basically implements the onboard communication protocol(c.f. Section 6.3). On one hand, it provides a control interface to the PermAEtop level application to send command messages to the STM32 (start / stopdata acquisition, setting the threshold and posttrigger values). On the otherhand, it reads data messages from the STM32 and injects them into thePermaDozer message queue.

6.5.3 PermaDozer and DataControl

The PermaDozer and DataControl libraries already available from otherPermaSense applications manageWSN communication and SD storage, there-for utilizing the AEBoard driver hardware abstractions. The existing SDstorage engine was slightly adapted for PermAE: In other PermaSense ap-plications, data was written to the SD card only every two minutes. Asin these applications, the data rate is well known, queue overows couldeasily be avoided. In PermAE however, data packets arrive irregularly andin bursts, sooner or later overowing the storage module's xed-size inputqueue. Thus, the PermAE storage queue adds incoming data packets to theSD card storage buer immediately, instead of rejecting new packets whenthe input queue is full. Furthermore, the PermaDozer component generatesa few other standard messages (eventlogger, rssi, statecounter) for networkmanagement purposes.

6.5.4 PermAE Top Level Application

The PermAE module is the top level TinyOS application. After initializingthe hardware, it turns on the STM32 and waits for the ARM_COMM_-

58

6.6. GSN INTEGRATION

FLAG pin to be raised (c.f. Section 6.3). Then, it sends a START_DAQcommand to initiate sampling of AE data. Once up and running, the appli-cation behaves as follows:

When the ARM_COMM_FLAG pin is raised again, a message readoperation by the AEModule is initiated.

Upon reception of a WSN command (c.f. Section 6.2.2), the commandis parsed and an according command is sent to the STM32 throughthe AEModule.

Every two minutes, PermaDozer initiates periodic measurements. PermAEthen starts the system health measurement routines to generate anodehalth as well as an aedaqhealthdata packet. The latter packetbases on the last received health message from the STM32 and is ex-tended with some values measured by the TinyNode.

For further information about the generated packets, refer to Appendices Cand D.

6.6 GSN Integration

As the PermAE implementation on the TinyNode bases on the same mech-anisms as all other PermaSense WSN nodes, AE data integration into GSNis a fairly simple process. The TinyOS mig tool converts the message for-mat denitions in the PermAE header les to Java classes, which in turncan be integrated into the PermaSense GSN instances. As the mig Makelehas been adapted accordingly, integrating AE messages into another GSNinstance is just a matter of running the mig tool and copying the generatedclass les to the respective GSN instance path. For closer details about thisprocess, refer to the PermaSense documentation.

As usual in PermaSense, data calibration/conversion to physical units isdone in between the Private and the Public GSN instances. I.e., whenaccessing AE data on the Public GSN instance, the data displayed hasalready been converted to physical units (c.f. Appendices C and D).

59


60

7System Evaluation

To complete the presentation of the AEBoard and PermAE, the system'sperformance characteristics are evaluated in this chapter. A special focusthereby lies on the reliable system operation under harsh environmental con-ditions.

7.1 Measurements at Room Temperature

7.1.1 Analog Frontend Characterization

In order to ensure a good quality of the measured samples, the analog fron-tend has been characterized in terms of frequency response and linearity.For both measurements, the input signal has once been applied directly atthe AEBoard signal input, and once at the input of the IL-LP-6S externalpreamplier by Physical Acoustics.

Input Frequency Response

Figure 7.1 shows the AEBoard's analog frontend frequency response. Theinput signal has been applied at the AEBoard input test pins and no pream-plier has been attached at the input channel. The output was measureddirectly at the ADC input.

The system's input is attenuated by 6 dB, due to the bias voltage providing

61

CHAPTER 7. SYSTEM EVALUATION!"#$%&''$($#)%

*+,-$.

/0111 .0/23 3043. /10111 ./0233 3403/4 /110111 ./20335 3430/26

(/7

(/4

(/3

(/.

(/1

(7

(4

(3

(.

1

89-:;-'<=$>?@A

#BCDEFE<+GE&'$>HIA

Figure 7.1: Frequency response of the onboard signal conditioning circuitry.

power to the external preamplier (c.f. Section 5.3). If more input gain isrequired, the operational amplier feedback path's resistor value could beincreased (c.f. Figure 5.3 and Appendix E).

The frequency response has also been measured with the external pream-plier included in the signal path. The input signal has been applied atthe preamplier input, the output was measured again directly at the ADCinput. Figure 7.2 depicts this measurements' results.

For the whole signal conditioning chain, the point of maximal amplicationlies at a frequency of 35 kHz with an amplication of 22 dB. In the frequencyrange of 20 to 100 kHz, the loss compared to the maximal amplication doesnever exceed 1 dB. Thus, the input lter matches the system specicationsin Section 3.4, requiring an input frequency bandwidth of 20 to 100 kHz.

Filter Linearity

Besides the frequency response, the input lter's linearity is another factorthat heavily aects the sampled signal's quality. A linear lter only changesamplitude and phase of an input signal's frequency parts, without introduc-ing frequencies that have not been part of the original signal. Thus, if a linearlter is fed with a perfectly sinusoidal signal, its output is a phase-shifted,amplied copy of the input signal.

The signal conditioning circuitry's linearity has been analyzed at input fre-quencies of 30, 50 and 100 kHz. The sinusoidal input signal was attached atthe AEBoard's input connector and the lter output has been measured at

62

7.1. MEASUREMENTS AT ROOM TEMPERATURE

!"#$%&'('$)*

!+,#'-

./000 1/.2- -/3-1 .0/000 1./2-- -3/-.3 .00/000 1.2/--4 -3-/.25

(6

(3

(-

(1

0

1

-

3

6

.0

.1

.-

.3

.6

10

11

1-

7"#89#:;<'=>?@

$%&ABCB;+DBE:'=FG@

Figure 7.2: Frequency response of the signal conditioning circuitry including theexternal preamplier.

the ADC input. As the AEBoard input lter is inverting the input signal,the sum of both signals is equal to the signal change introduced by ltering.Figure 7.3 shows the input and output signal for the 50 kHz measurement,as well as their sum.

Obviously, the error signal is also sinusoidal, indicating a linear lter char-acter. Indeed, after shifting and amplifying the output signal adequately(i.e., compensating the linear ltering), the dierence between the signals issmaller than 5% as shown in Figure 7.4.

However, when adding the external preamplier to the signal conditioningchain, the linearity is lost. Figure 7.5 clearly shows, that even after com-pensating phase shift and amplication, signicant nonlinear distortions ofthe output signal remain. Because the IL-LP-6S preamplier also inverts itsinput signal, the output signal must now be subtracted from the input inorder to nd the error introduced.

Nevertheless, although this nonlinear behaviour certainly aects the rawevent data that is stored on the SD card, it should not notably aect theAE event parametrization results. However, further analysis based on thestored raw samples that includes frequency domain calculations should takethese nonlinear eects into account.

For all linearity measurement results, please refer to Section F.1.

63

CHAPTER 7. SYSTEM EVALUATION

Figure 7.3: Uncompensated error introduced by the AEBoard input lter.

Figure 7.4: Phase and amplitude compensated error, introduced by the AEBoardinput lter.

64

7.1. MEASUREMENTS AT ROOM TEMPERATURE

Figure 7.5: Phase and amplitude compensated error, introduced by the externalpreamplier and AEBoard input lter.

7.1.2 Power Consumption

The AEBoard's power consumption has been measured at room temperaturefor various isolated system parts as well as for the whole system in operation.Table 7.1 summarizes these measurements.

Power Domain InputVoltage

InputCurrent

InputPower

Digital Master Domain 3.3V 23mA 75mWDigital Slave Domain 2.8V 44mA 122mWSensing / Conversion Domain 5.5V 13.7mA 75mWSystem in operation 12V 48mA 576mW

Table 7.1: AEBoard power consumption (for domain names, refer to Figure 5.5)

The dierence between the total system power consumption and the accu-mulated subsystem consumption is due to losses in the power supply system.These losses clearly dominate the overall power consumption with a total of304mW, probably leaving much potential for further optimization.

From the beginning, the AEBoard was designed to support a photovoltaicpower supply. Nevertheless, the Li-SOCl2 batteries currently used in PermaSense(the LSH 20 model by Saft batteries [59]) feature a nominal capacity of

65


13Ah per cell. When combining three of these 3.6V batteries to a 10.8Vpower supply, an AEBoard could continuously acquire data for about 10days. This may be sucient for shorter AE sensing campaigns. For long-term measurements however, a photovoltaic power supply is inevitable.

7.2 Temperature Cycle Tests

In order to evaluate the system performance under realistic environmentalconditions, the following performance tests have been carried out under vary-ing temperature conditions. To this end, an automatic climate chamber hasbeen utilized to ensure controlled experimental conditions. The followingquestions have been studied specically:

Is the system operating stable under quickly varying temperature con-ditions?

Does temperature aect the system's performance in terms of maximalAE activity load that can be processed and stored?

Does temperature aect the system's power consumption?

How does the signal conditioning circuitry behave under temperaturecycles and how is the parametrization result aected?

To answer these questions, two experiments have been carried out in thetemperature chamber, as discussed in the following sections. As the experi-mental conditions shall be reproducible, no piezoelectric AE sensor has beenused for these experiments. Instead, the STM32Discovery evaluation kitfor the STM32 platform has been utilized to simulate the output of a piezo-electric AE sensor. Thus, the following experiments are strictly electricaland do not consider any mechanical or electrical characteristics of the AEsensor itself. The application used for event simulation is also included onthe enclosed CD or in the PermaSense SVN repository (c.f. Appendix B andG).

7.2.1 Load Test

The rst temperature test setup was designed to evaluate the overall systemstability under changing temperature conditions as well as measuring thesystem's event processing capabilities over an extended period of time and fordierent temperatures. The AEBoard, as well as the external preampliers,have been placed in the climate chamber, whereas the power supply and the

66

7.2. TEMPERATURE CYCLE TESTS

Climate Chamber

AEBoard

Preamp

Preamp

Channel 1

Channel 2

event generator

power supply

5kOhm

1kOhm

Figure 7.6: Load test setup.

event generator were kept at room temperature in order to ensure stableconditions. Figure 7.6 sketches this setup.

Every ve minutes, the event generator produced fteen AE events per sec-ond on both acquisition channels over a period of 30 seconds, resulting in900 generated AE events per stimulation period. This event rate is highenough to be above all expected event bursts occurring in a real measure-ment deployment (c.f. Section 3.1.2). All generated AE events were identical,featuring an amplitude of 80 dB (1V) at the ADC input and an event lengthof about 2700 samples (1.35ms). For each stimulation period, the accordingaestatistics messages have been analyzed in order to evaluate the system'sparametrization and storage performance. During the experiment durationof totally seven hours, the ambient temperature was varying between -35Cand +50C. Figure 7.7 shows the results of this load test experiment.

Obviously, the varying temperature conditions did not aect the system'sperformance in terms of operational stability as well as parametrization andstorage capability. Furthermore, according to the preliminary experimentalresults described in Section 3.1, the applied load is unrealistically high. Ad-ditional tests showed that by reducing the stimulation period to only 15 sec-onds, all occurring events during a stimulation period could be parametrized,and over 90% of the events could be stored on the SD card. The system'sperformance bottleneck in terms of parametrization capability is the serialport's communication speed to the TinyNode - as more events arrive thancan be forwarded, the acquisition system slowly runs out of input buers.SD storage performance however is not limited by the speed of the SDIOinterface. Rather, as the storage task features a lower priority than the com-munication task, it simply does not have enough processor time to store allevents. Raising the storage task's priority would solve this problem but also

67


!"#$%$&

'%()*+

,,-,, ,.-,, ,/-,, ,+-,, ,0-,, ,1-,, ,2-,, ,3-,,

2,,

3,,

4,,

5,,

.,,,

60,

6+,

6/,

6.,

,

.,

/,

+,

0,

1,

2,

7)%&89):

'%9%;)$9<=):

#$>9):

?);@)9%$89)

?<;)*ABC

D8;E)9*>F*"G)H$&

?);@)9%$89)*AIJC

Figure 7.7: System performance under varying temperature conditions.

violate the system specication (c.f. Sections 3.4 and 6.4).

7.2.2 Data Quality Test: AEBoard and USB AE Node

Even tough temperature seems not to inuence the system's overall relia-bility and speed, it certainly aects the data acquisition system's analogcomponents and thus also the parametrization results. In order to investi-gate on these eects, data acquisition quality has been tested and comparedto a commercially available system.

Again, the AEBoard has been placed in the climate chamber, together withthe external preamplier for channel one. The preamplier for channel twohowever was placed outside the chamber to allow for analysis of the pream-plier's impact on data quality. Furthermore, an USB AE Node by PhysicalAcoustics Corporation [60] together with a preamplier was also placed inthe climate chamber in order to get reference measurements of a state ofthe art AE measurement device. This USB device was connected to a PCrunning the AEWin software [39]. Figure 7.8 depicts the data quality testsetup.

All AE preampliers were fed with the same input signal generated by theaforementioned event generator based on the STM32Discovery evaluationkit and the EventGenerator software. The test signal, as shown in Figure7.9, simulates an AE event consisting of three frequency components at 35,50 and 60 kHz, with the 50 kHz oscillation being the dominant frequencycomponent. This characteristics match well the average waveform that

68


Climate Chamber

AEBoard

Preamp

Preamp

Channel 1

Channel 2

event generator

power supply

Preamp USB AE Node

Input

USBPC

5kOhm

1kOhm

Figure 7.8: Data quality test setup.

has been captured during the preliminary experiments. For detection, athreshold value of 52 dB was used in conjunction with a posttrigger time of400 samples (800µs).

Figure 7.9: Test signal generated by the EventGenerator application.

69


The test signal was generated exactly once every 60 s over the total exper-iment duration of 18 hours. Ambient temperature varied in the range of-30C up to +40C during that timespan. The measured parameters for allcaptured test signals have then been compared. In an ideal system, thesemeasured parameters would have remained constant over the whole experi-ment.

Generally, this data quality test showed that some AE parameters are heavilyinuenced by temperature variations, while others are not. Section F.2 showsthe result plots for all parameters and channels, while the most outstandingeects are discussed in the following.

Temperature Dependent Parameter Variations

Table 7.2 gives an overview of the results for AEBoard channel one, whichis the most relevant measurement for the estimation of the AEBoard's dataquality in a productive deployment. The mean values at room temperatureare compared to the mean values at -30C and +40C. The B column in-dicates, whether the parameter exposes pro-cyclic1 or anti-cyclic2 behaviourwith temperature.

Param. Mean +15C Mean -30C Mean +40C B

Length 1266 1321 (+4.3%) 1229 (-2.9%) aRisetime 218 218 (+0.0%) 218 (+0.0%) n/aAmplitude 20421 19376 (-5.2%) 20532 (-0.5%) pCount 73 77 (+5.5%) 70 (-4.1%) aEnergy 382e6 418e6 (+9.4%) 317e6 (-17.0%) a

Table 7.2: Data quality test results for AEBoard channel one. p stands for pro-cyclic, a for anti-cyclic behaviour. See Section F.2 for the accordingplots.

The table shows, that the risetime parameter was not inuenced by thetemperature variations at all and in fact, it almost stayed constant over thewhole experiment, thus featuring an absolute precision of less than 5µs. Theother parameters however were inuenced stronger by temperature changes.Especially the energy parameter exposes a bandwidth of almost 30% of itsaverage room temperature value.

Figure 7.10 depicts the pro-cyclic parameter behaviour of the amplitude pa-rameter versus anti-cyclic behaviour at the example of the energy parameter.

1Pro-cyclic: The parameter value rises when temperature rises.2Anti-cyclic: The parameter value falls when temperature rises.

70

7.2. TEMPERATURE CYCLE TESTS!"#$%

!&'()*

++,++ +-,++ *+,++ *-,++

+

-+++

*++++

*-+++

.++++

.-+++

/++++

0/+

0.+

0*+

+

*+

.+

/+

1+

-+

2+3456#%78(

9(45("&%7"():;<=

9#4():>=

3456#%78():'($("#?=

9(45("&%7"():;<=

++,++ +-,++ *+,++ *-,++

+@A+

*@AB

.@AB

/@AB

1@AB

-@AB

0/+

0.+

0*+

+

*+

.+

/+

1+

-+

2+@$("'C

9(45("&%7"():;<=

9#4():>=@$("'C):'($("#?=

9(45("&%7"():;<=

++,++ +-,++ *+,++ *-,++

+

*++

.++

/++

1++

-++

2++

0/+

0.+

0*+

+

*+

.+

/+

1+

-+

2+

D#E(%#4(

9(45("&%7"():;<=

9#4():>=

D#E(%#4():7E=

9(45("&%7"():;<=

++,++ +-,++ *+,++ *-,++

+

*++

.++

/++

1++

-++

2++

0/+

0.+

0*+

+

*+

.+

/+

1+

-+

2+

<F7$%

9(45("&%7"():;<=

9#4():>=

!76E(E

9(45("&%7"():;<=

Figure 7.10: Pro-cyclic versus anti-cyclic parameter behaviour.

Another data quality indicator is the parameter variation for constant tem-perature. As all stimulation events are identical, in an ideal system thisvariations would be zero for phases with reasonably constant ambient tem-perature. Table 7.3 lists the constant temperature standard deviations for allAE parameters measured at channel one in absolute units as well as relativeto the mean value at the respective temperature.

Param. Std Dev +15C Std Dev -30C Std Dev +40C

Length 14.4 (1.1%) 24.8 (2.0%) 10.6 (0.8%)Risetime 0.5 (0.2%) 0.4 (0.2%) 0.5 (0.2%)Amplitude 175.4 (0.9%) 125.2 (0.6%) 183.6 (0.9%)Count 1.1 (1.5%) 1.3 (1.8%) 0.8 (0.9%)Energy 106e4 (0.3%) 65e4 (0.2%) 212e4 (0.6%)

Table 7.3: Parameter standard deviations at constant temperatures for AEBoardchannel one. See Section F.2 for the according plots.

Obviously, although temperature variations inuence the measurements heav-ily, parameter variations for constant ambient temperatures expose a stan-dard deviation of less than 2% for all parameters.

Thus, for reasonably constant ambient conditions, the parametrization resultcan reliably be reproduced. For changing temperature conditions however,the parametrization results should be adapted appropriately. To gain reliableunderstanding of the temperature dependent parameter variations however,further tests under controlled conditions are necessary.

71


Inuence of the External Preamplier

As already described at the beginning of Section 7.2.2, the external pream-plier for AEBoard channel two was placed outside the temperature chamberin order to analyze its eect on the temperature dependent parameter vari-ations. Like in the previous section, Tables 7.4 and 7.5 list the temperaturedependent variations as well as the parameter standard deviation at constanttemperatures.

Param. Mean +15C Mean -30C Mean +40C B

Length 1260 1276 (+1.3%) 1238 (-1.7%) aRisetime 218 218 (+0.0%) 218 (+0.0%) n/aAmplitude 20853 20931 (+0.4%) 20914 (+0.3%) n/aCount 72 73 (+1.4%) 71 (-1.4%) aEnergy 396e6 389e6 (-1.5%) 373e6 (-5.6%) a

Table 7.4: Data quality test results for AEBoard channel two. p stands for pro-cyclic, a for anti-cyclic behaviour. See Section F.2 for the accordingplots.

Param. Std Dev +15C Std Dev -30C Std Dev +40C

Length 15.5 (1.2%) 17.1 (1.4%) 9.6 (0.8%)Risetime 0.5 (0.2%) 0.5 (0.2%) 0.5 (0.2%)Amplitude 174.4 (0.8%) 190.9 (0.9%) 185.6 (0.9%)Count 0.9 (1.3%) 0.9 (1.3%) 0.6 (0.8%)Energy 424e3 (0.1%) 396e3 (0.1%) 933e3 (0.2%)

Table 7.5: Parameter standard deviations at constant temperatures for AEBoardchannel two. See Section F.2 for the according plots.

When comparing these values with the values from Section 7.2.2, we ndthat keeping the external preamplier's ambient temperature constant

signicantly reduces the temperature dependent parameter variationsfor all parameters (except the risetime parameter, which does not showany temperature dependency at all).

slightly reduces the parameter variation for constant AEBoard ambienttemperatures.

Figure 7.11 illustrates the rst nding by depicting the mean ranges fromTables 7.2 and 7.4.

72


!"#$%

!&'()*

+,-.. *+-.. .*-.. ./-..

.

.0.+

.0.*

.0.1

.0.2

.0.3

.0.,

.0./

.0.4

.0.5

.0+

61.

6*.

6+.

.

+.

*.

1.

2.

3.

78""($%

9(:;("&%8"()<=7>

9#:(

78""($%)?@$A8:;%#@$)<B>

9(:;("&%8"()<=7>

C($'%D E#A(%#:( B:;F#%8G( 7@8$% H$("'I

4.

43

5.

53

+..

+.3

++.

7D&$$(F)+

7D&$$(F)*J

Figure 7.11: Temperature dependent mean parameter ranges for both AEBoardchannels. For channel 1, the external preamplier was placed insidethe climate chamber, whereas the preamplier for channel 2 was heldat constant temperature. The mean parameter value at 15C is set to100%.

Figure 7.12 shows the energy parameter plots for both channels, furtherdemonstrating the reduction in temperature dependent parameter variations.!"#$%

!&'()*

++,++ +-,++ *+,++ *-,++

+./+

*./0

1./0

2./0

3./0

-./0

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

67&$$(8)9$(

.$("':

;(<=("&%>"()?@6A

;#<()?7A

.$("':)?'($("#BA

;(<=("&%>"()?@6A

++,++ +-,++ *+,++ *-,++

+./+

*./0

1./0

2./0

3./0

-./0

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

67&$$(8);CD

.$("':

;(<=("&%>"()?@6A

;#<()?7A

.$("':)?'($("#BA

;(<=("&%>"()?@6A

++,++ +-,++ *+,++ *-,++

+

*++

1++

2++

3++

-++

5++

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

E#F(%#<(

;(<=("&%>"()?@6A

;#<()?7A

E#F(%#<()?>FA

;(<=("&%>"()?@6A

++,++ +-,++ *+,++ *-,++

+

*++

1++

2++

3++

-++

5++

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

6D>$%

;(<=("&%>"()?@6A

;#<()?7A

!>8F(F

;(<=("&%>"()?@6A

Figure 7.12: The eect of keeping the external preamplier's ambient temperatureconstant at the example of the energy parameter.

This ndings indicate that keeping the preamplier's ambient conditionsconstant only reduces the eects on the parametrization result, but does notchange the eects' characteristics. Hence, the AEBoard's signal conditioningcircuitry's behaviour under temperature variations is qualitatively as well as

73


quantitatively well comparable to state of the art industry products such asthe IL-LP-6S preamplier.

USB AE Node Test Results

As for the USB AE Node, no temperature specication is given by the man-ufacturer, it was also exposed to the full temperature cycle in order to testout its limits. Basically, the results show a behaviour that is comparableto the AEBoard's results. However, the USB AE Node only seems to oper-ate reliably in a temperature range between -10C and +30C. Figure 7.13illustrates this with the USB AE Node's risetime and count plots.

!"#$%

!&'()*

++,++ +-,++ *+,++ *-,++

+./+

*./0

1./0

2./0

3./0

-./0

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

67&$$(8)9$(

.$("':

;(<=("&%>"()?@6A

;#<()?7A

.$("':)?'($("#BA

;(<=("&%>"()?@6A

++,++ +-,++ *+,++ *-,++

+./+

*./0

1./0

2./0

3./0

-./0

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

67&$$(8);CD

.$("':

;(<=("&%>"()?@6A

;#<()?7A

.$("':)?'($("#BA

;(<=("&%>"()?@6A

++,++ +-,++ *+,++ *-,++

+

*++

1++

2++

3++

-++

5++

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

E#F(%#<(

;(<=("&%>"()?@6A

;#<()?7A

E#F(%#<()?>FA

;(<=("&%>"()?@6A

++,++ +-,++ *+,++ *-,++

+

*++

1++

2++

3++

-++

5++

42+

41+

4*+

+

*+

1+

2+

3+

-+

5+

6D>$%

;(<=("&%>"()?@6A

;#<()?7A

!>8F(F

;(<=("&%>"()?@6A

Figure 7.13: USB AE Node temperature dependent variations. At temperaturesbelow -10C and above +30C, some parameter values start to exposestrong variability. See also Section F.2 for other parameter plots.

Furthermore, the USB AE Node requires a PC in order to acquire data.Most of these devices are also not specied for extreme temperatures. Thus,utilizing the USB AE Node for reference AE tests is only recommended ifthe environmental conditions are controlled or well known in advance.

Comparing PermAE with AEWin Results

A further goal of the data quality test was to compare results of the PermAErunning on the AEBoard with results from a commercial reference system,i.e., the USB AE Node with AEWin. Our ndings are that, qualitatively,both devices expose comparable behaviour under the same environmentalconditions (c.f. Section F.2). However, a quantitative comparison of theresults is dicult, as both the AEBoard and the USB AE Node featuredierent internal analog signal amplication and ltering. Thus, the signal

74


measured at both ADCs is never the same, even tough the devices havereceived the same stimulating signal.

For the three parameters length, rise time and energy, the conversion con-stants could roughly be estimated based on the data quality test results.Of course, only results within the USB AE Node's reliable operating tem-perature range have been considered. Table 7.6 lists the respective factors.

Parameter PermAE unit AEWin unit c (vAEWin = c · vPermAE)

Length samples µs 2µsRisetime samples µs 2µsEnergy V2 V2 80e3

Table 7.6: Conversion factors between PermAE and AEWin based on the dataquality test results.

For the event amplitude and pulse count parameters, the results from PermAEand AEWin dier signicantly. Probably, this dierence is due to dierentsignal amplication factors in the device's signal conditioning circuitry. Theconversion functions between PermAE and AEWin results for these two pa-rameters are most probably not linear and their identication would requirefurther in-depth investigations.

7.2.3 Current Consumption

During the data quality test, also the AEBoard's current consumption at asupply voltage of 12V has been constantly monitored (c.f. Section 7.1.2).As Figure 7.14 shows, the AEBoard's current consumption does not exhibita distinct temperature dependency.

!"#$%

!&'()*

++,++ +-,++ .+,++ .-,++

+

+/+.

+/+*

+/+0

+/+1

+/+-

+/+2

+/+3

+/+4

+/+5

+/.

60+

6*+

6.+

+

.+

*+

0+

1+

-+

78""($%

9(:;("&%8"()<=7>

9#:()<?>

78""($%)@A$B8:;%#A$)<C>

9(:;("&%8"()<=7>

D($'%? E#B(%#:( C:;F#%8G( 7A8$% H$("'I

4+

4-

5+

5-

.++

.+-

..+

7?&$$(F).

7?&$$(F)*J

Figure 7.14: AEBoard's power consumption at varying ambient temperature.

75


76

8Conclusion and Outlook

In this thesis, the architectural approach of processing sensors has been pre-sented, allowing to sense data with high sampling rates in a low-power op-timized WSN environment (c.f. Chapter 4). With processing sensors, a lowdata rate sensor is emulated by performing parts of the data processing al-ready on the sensor device and thus reducing the data volume drastically.Although this thesis focuses on the application of AE sensing, processingsensors may also be used for other applications, e.g., on sensor processing ofcamera images.

Based on the concept of processing sensors, an AE sensing platform forPermaSense has been implemented (c.f. Chapters 5 and 6). The neces-sary data acquisition hard- and software systems have been designed fromscratch, whereas the TinyNode could be used as WSN platform, thereby en-suring maximal compatibility with the existing PermaSense infrastructure.An evaluation of various system characteristics showed that the developedsystem indeed meets its specications and is also competitive with commer-cially available AE sensor systems (c.f. Chapter 7). However, the system'sstability and reliability in an outdoor deployment still needs to be proven.According experiments are scheduled for summer 2011.

System tests with heavy load showed that PermAE currently is not alwaysable to parametrize and store all events occurring (c.f. Section 7.2.1). Nev-ertheless, the performance achieved is sucient for the specic application,as the applied load was much higher than the expected activity peaks ineld experiments. However, it is not clear whether buer sizes and task pri-

77

CHAPTER 8. CONCLUSION AND OUTLOOK

orities chosen in the current PermAE implementation are optimal in orderto achieve maximal parametrization and storage throughput. As PermAE'sbasic structure is essentially not very complex, analytical modelling couldhelp to further optimize PermAE's performance.

As parametrization results vary signicantly with changing ambient tem-perature (c.f. Section 7.2.2), care must be taken when analyzing the resultsproduced in order to not interpret temperature dependent changes as changesin AE characteristics. From the perspective of possible applications in earlywarning systems, it is even desirable to automate the according error com-pensation processing and integrate it either in the parametrization softwareitself or at a later stage in the data backend.

Further steps based on this work may include the following:

Deploying the developed AE sensors in the eld in order to validatetheir robust operation over longer periods of time.

Optimizing the AEBoard's power consumption.

Optimizing PermAE's parametrization performance under heavy loadpossibly using analytical modelling techniques.

Quantitatively analyzing the measurement errors introduced by tem-perature variations in order to eliminate these errors before furtherdata analysis.

Building other high data rate sensors based on the concept of process-ing sensors.

Of course, from a non-technical perspective, further steps also include application-focused eld deployments in order to gain insight into crack formation pro-cesses in rock walls.

We hope that the work presented will eventually contribute to the develop-ment of reliable rockfall warning systems, potentially saving many humanlives from these natural hazards. Technology must always be about peopleafter all.

Das Gefährlichste an der Technik ist, dass sie ablenkt von dem,was den Menschen wirklich ausmacht, von dem, was er wirklichbraucht. Elias Canetti

78

AAEBoard and PermAE User Manual

This chapter gives a short overview of the AEBoard's functionality and ex-plains how to prepare it for deployment. All instructions refer to hardwarerevision 1.1 and may need to be adapted for further revisions.

A.1 Connectors and LEDs

Figure A.1 schematically depicts the AEBoard's connectors and LEDs.

Power supply connector: Connect to a power source with an outputvoltage between 7 and 20V.

On / O switch: Use to switch the main power supply on or o.

Conversion frequency jumper: Apply a jumper to x voltage con-version switching frequency. Should be removed for power eciencyreasons.

Digital supply voltage jumper: A jumper must be applied at theright position (3.3V) to ensure correct system operation. This jumperhas been removed in rev. 1.2 and digital supply voltage was xed to3.3V.

TinyNode reset button: Press to reset the TinyNode.

79

APPENDIX A. AEBOARD AND PERMAE USER MANUAL

On / Off Switch

Power Supply Connector

Conversion Frequency Jumper

Digital Supply Voltage Jumper

TinyNode Reset ButtonTinyNode JTAG connector

TinyNode SD card

TinyNode

STM32

STM32 Reset Button

STM32 SD card

Tx / Rx LEDsSTM32 power jumper

Debug LEDs 1-3

Console Connector

STM32 JTAG Connector

STM32 bootmode jumper

Sensor Connectors

Figure A.1: AEBoard connectors and LEDs.

TinyNode JTAG connector: Use to program or debug the TinyNode'sMSP430 microprocessor.

TinyNode SD card: Insert an SD card to enable TinyNode backlog-ging (c.f. Section A.2).

TX / RX LEDs: UART trac LEDs to indicate communicationbetween the TinyNode and the STM32.

STM32 power jumper: Apply a jumper in order to turn on theSTM32 once the main power supply is enabled. This jumper must beremoved for productive operation and is only intended for deploymentpreparation (c.f. Section A.2).

Debug LEDs 1-3: Display PermAE's status on the STM32:

LED1 (red): Indicates that the STM32 is turned on.

LED2 (orange): Indicates a software error if turned on (c.f. Sec-tion A.3).

LED3 (green): Toggles upon detection of an AE event.

Console connector: Attach e.g. a TTL-232R-3V3 USB TTL serialcable [54] to display debug messages on a PC console (c.f. SectionA.3).

80

A.2. DEPLOYMENT PREPARATION

STM32 JTAG connector: Use to program or debug the STM32microprocessor (c.f. Section A.2 and Appendix B).

STM32 SD card: Insert an SD card for the STM32 (c.f. SectionA.2).

STM32 bootmode jumper: Apply a jumper at the left position toensure correct system operation. Applying a jumper to the right wouldlet the STM32 boot from its factory bootloader (c.f. [47]).

STM32 reset button: Press to reset the STM32.

Sensor connectors: Attach the external peramplier's output di-rectly to these SMA connectors.

A.2 Deployment Preparation

The following guide assumes the reader to be already familiar with Per-maDozer and the TinyOS toolchain. Furthermore, the following prerequi-sites must be met in order to prepare an AEBoard for deployment:

A windows PC with the STM32 ST-LINK Utility software installed1.

An ST-LINK JTAG programmer.

A working TinyOS 2.x toolchain with all PermaSense specic librariesin order to build and ash PermAE for the TinyNode.

An SIB to format the TinyNode's SD card.

Once these requirements are met, go through the following steps:

1. Make sure that the main power supply is switched o.

2. Remove the TinyNode, both SD cards and both AE sensor from theAEBoard.

3. Apply the STM32 boot mode jumper at the left position and the digitalsupply voltage jumper at the right position (c.f. Section A.1).

4. Apply a jumper to the STM32 power jumper connector.

5. Connect the AEBoard to a PC using the ST-LINK JTAG programmerconnected to the STM32 JTAG port.

1Downloadable at http://www.st.com/internet/evalboard/product/219866.jsp

81


6. Switch the AEBoard's main power supply on.

7. On your PC, open the STM32 ST-LINK Utility software.

8. For AEBoard revision 1.1., make sure to either use a patched ribboncable (interchanging pins 5 and 13) or enable SWD instead of JTAGcommunication under Target - Settings....

9. Program the STM32 with the PermAEInit.hex binary image that canbe found in the Binaries folder either on the enclosed CD (c.f. Ap-pendix G) or under /trunk/soft/AEBoard in the PermaSense SVNrepository. This initializes the signal processing with a threshold of50 dB, a posttrigger of 400 samples (800µs) and a storage block sizefor raw events of 2048Bytes (1016 samples). To change these initial-ization values, PermAEInit needs to be recompiled as described inAppendix B.

10. Insert the STM32's SD card, reset the STM32 and wait until the LED3(green) is turned on. The SD card is now initialized.

11. Program the STM32 with the PermAE.hex binary image that can befound in the Binaries folder either on the enclosed CD (c.f. AppendixG) or in the PermaSense SVN repository under /trunk/soft/AEBoard.

12. Switch the main power supply o and remove the STM32 power jumperconnector.

13. Format the TinyNode's SD card with an SIB and the SibFormatSDCardapplication.

14. Build the TinyOS PermAE application (located in the PermaSense SVNrepository at /trunk/soft/tinyos-2.1/apps) with the desired nodeID and ash it to the AEBoard's TinyNode.

15. Insert the TinyNode's SD card and attach the TinyNode on the AEBoard.

16. Connect the sensors to the sensor connectors and the TinyNode's an-tenna to the TinyNode antenna connector.

The AEBoard is now ready to be deployed.

A.3 STM32 Debugging

In order to more closely monitor an AEBoard's status, a TTL-232R-3V3 USBTTL serial cable [54] may used to connect the console connector to a PC'sUSB port. If PermAE and PermAEInit for the STM32 were compiled with

82

A.4. DATA RETRIEVAL

the CONSOLE_ON option (c.f. Appendix B), the STM32 then outputsdebug information over this console port. A terminal emulator (e.g., minicomunder Linux) may be used to display the AEBoard's data stream which isoutput at a baudrate of 2000000 bits/s per default. Console communicationis unidirectional and the STM32 does not process any data received over theconsole port.

If the STM32's LED2 (orange) is illuminated, a software failure occurred.This can be one of the following:

An initialization error. Use the console debugging port to get moreinformation about the initialization error that has occurred.

The data acquisition has run out of input buers, probably because thedata analysis tasks were too slow in processing the input data. Eventhough data is lost, the system continues normal operation as soon asempty input buers are available.

For PermAE code debugging, refer to Appendix B.

A.4 Data Retrieval

A.4.1 Data Sent Over the WSN

The data packets produced by PermAE may simply be retrieved over thePermaSense data webinterface, i.e., http://data.permasense.ch. The rel-evant virtual sensors are called according to the message types specied inAppendix C. On the Public GSN instance, data conversion to physicalunits (c.f. Appendices C and D) has already been performed by GSN.

For a detailed description of the messages produced by PermAE, refer toSection 6.2 as well as Appendix C. For a description of the exact eventparametrization algorithm as well as the timing correlation between events,refer to Section 6.4.4.

A.4.2 Raw Event Samples

The raw event data storage on the STM32's SD card does not use a PCreadable lesystem. Thus, in order to use the stored data for further analysis,it must be converted into les readable by third party software.

Raw data can be retrieved with the help of a PHP script, reading the rawSD card data stream and writing the raw events to ASCII les. The script isnamed convertStoredEvents.php and may be found in the AEBoard/Tools

83

http://data.permasense.ch


folder on the enclosed CD or in the PermaSense SVN repository. On a UnixPC, the dd tool may be used to forward the raw data to the PHP script.E.g., if the SD card can be accessed at /dev/sdb and the generated ASCIIles shall be written to /home/user/RawData, enter the following commandin a terminal:

sudo dd if=/dev/sdb | convertStoredEvents.php /home/user/RawData

When the command has nished, /home/user/RawData contains les namedevent_XX.txt, where XX denotes the event's serial number. The output leshave a very simple format, e.g.,

serial#: 0

startSample: 4166340

length: 1016

threshold: 580

posttrigger: 400

channel: 1

-745

...

-1056

The rst six lines refer to the relevant information contained in the initial-ization block (c.f. Section 6.4.6), the rest of the lines contain the event'ssample values in PermAE's internal data format (for converting these valuesto V or dB, refer to D.1.2).

A.5 Sending Commands to the AEBoard

As already described in Section 6.2.2, commands may be sent to the AEBoardin order to enable and disable data acquisition as well as controlling thethreshold and posttrigger values used for event detection.

The simplest way to send a command to a deployed AEBoard is via thedozer_command virtual sensor on the PermaSense data webinterface. Onthe GSN homepage, select the dozer_command virtual sensor according toyour deployment (e.g., jungfraujoch_dozer_command, open the Uploadtab and ll out the form as follows:

destination: The AEBoards TinyNode node ID.

command: Select AEBOARD_CTL_CMD.

arg: Fill in the encoded message as described in Section 6.2.2.

84

A.5. SENDING COMMANDS TO THE AEBOARD

repetitioncnt: You may enter a number of 3 repetitions.

To send the Dozer command, nally press the upload button. You maycheck your command's eect in incoming AE messages (the aedata virtualsensor) for threshold and posttrigger settings, or in the following aedaqhealth-data messages for the STM32's power state (c.f. Appendix C).

85


86

BPermAE Developer's Guide

B.1 STM32

For developing the PermAE software on the STM32 processor, the TrueSTU-DIO/STM32 IDE from Atollic has been used. A Lite Version of this IDEcan be downloaded on the Atollic website1. TrueSTUDIO is based on theEclipse IDE framework and oers an integrated workow for programmingthe STM32 microprocessors as well as in-circuit debugging.

For simplicity, the AEBoard directory, either on the enclosed CD or in thePermaSense SVN repository, directly contains a complete Atollic TrueSTU-DIO/STM32 workspace. For adapting the software, it is therefore sucientto download the IDE and open the AEBoard folder when prompted for thedesired workspace location.

The three projects contained in the STM32 workspace are described ingreater detail in the following sections.

B.1.1 PermAE

PermAE is the main application, running on the STM32 when the AEBoardis in operation. The /src directory contains all application source les, i.e.,C source and header les.

1http://www.atollic.com/index.php/targets/stm32

87

http://www.atollic.com/index.php/targets/stm32

APPENDIX B. PERMAE DEVELOPER'S GUIDE

PermAEParams.h: Parameters that aect the application's behaviour,e.g., the static task priorities and buer sizes. Furthermore, the dataoutput mode can be changed, i.e., if parameters are forwarded to theTinyNode or if they are output on the console port.

PermAE.h: Header le dening all application data structures, i.e.,tasks, queues etc.

CommunicationProtocol.h: Denitions for the onboard communica-tion protocol (c.f. Section 6.3).

FreeRTOSConfig.h: FreeRTOS conguration le. For details aboutconguring FreeRTOS, please refer to the FreeRTOS documentation[56, 57] on the enclosed CD.

main.c: The main program le. Basically, only the AEBoard is ini-tialized and then FreeRTOS is started immediately, starting with theinitialization task.

task*.c: The FreeRTOS tasks as described in Section 6.4.

ISR_*.c: Interrupt service routines for data acquisition and consoleport communication.

The STM32 workspace is congured to enable console text output when com-piling PermAE. If console output is not desired for any reason, the symbolCONSOLE_ON must be removed in Project - Properties - C/C++ General- Paths and Symbols - Symbols.

The PermAE application links to a Lib directory, containing rmware anddrivers for the STM32 in the STM32 directory, AEBoard specic drivers in theAEBoard directory as well as the FreeRTOS kernel in the FreeRTOS directory.Note that the rmware source les in STM32 are slightly adapted versions ofthe original vendor les. Specically, the following changes to the originalles have been made:

startup_stm32f10x_hd.s has been adapted in order to point to theFreeRTOS system hooks.

system_stm32f10x.c has been adapted to expect a 6MHz oscillatorinput instead of 8MHz.

Furthermore, note that FreeRTOS oers three possible memory allocationimplementations, dened in Lib/FreeRTOS/MemMang. As in PermAE allmemory is allocated statically, the simplest implementation (heap_1.c) hasbeen used. Make sure that always only one of the three implementation les

88

B.1. STM32

is included in compilation. You may edit the according source lter in Project- Properties - C/C++ General - Paths and Symbols - Source Location. Formore information about memory allocation in FreeRTOS, refer to [56].

B.1.2 PermAEInit

The PermAEInit application must be used to initialize an SD card to be usedon an AEBoard (c.f. Section 6.4.8). As some initialization parameters aredirectly compiled into the application, you may need to adapt PermAEInitaccordingly before initializing the AEBoards. The /src directory containsall application source les, i.e., C source and header les.

AEDetectionParameters.h: Change this le in order to adapt theinitial threshold, posttrigger and storage buer size parameters (c.f.Section 6.4.8)).

AEInitApplication.h: Some application specic data structures.

FreeRTOSConfig.h: FreeRTOS conguration le. For details aboutconguring FreeRTOS, please refer to the FreeRTOS documentation[56, 57] on the enclosed CD.

main.c: The main program le. Basically, only the AEBoard is ini-tialized and then FreeRTOS is started immediately, starting with theinitialization task.

taskInit.c: The only FreeRTOS task in this application, calibratingand initializing the AEBoard and writing the calibration data onto theSD card.

ISR_DAQ.c: Interrupt service routine for the calibration data acquisi-tion.

PermAEInit is linked to the same Lib directory as PermAE. Remember thatevery change in Lib aects both applications (c.f. Section B.1.1).

For further information about PermAE on the STM32, refer to Section 6.4as well as the comments in the sourcecode.

B.1.3 EventGenerator

The EventGenerator application was used to generate the stimulation eventsduring the temperature chamber tests. It basically outputs a well-denedanalog signal pulse at pin PA4 in congurable intervals. The EventGen-erator's directory structure is directly based on TrueSTUDIO's template

89


projects and runs on the STM32Discovery evaluation kit. The whole ap-plication code can be found in the main.c le. At the top of main.c, theparameter denitions may be changed in order to generate dierent stressperiods as described in Section 7.2.

The samples to be output at the STM32's DAC are stored in the eventBuffer.hle. This le has been generated using the genEventBuffer.php script. Forfurther documentation of the EventGenerator, please refer to the sourcecodecomments.

B.1.4 Programming and Debugging the STM32

For programming or debugging PermAE or any other application on theSTM32, the default debugging congurations shipping with TrueSTUDIOmay be used together with the ST-LINK programmer device. For the AEBoardrev. 1.1, make sure to either use a patched ribbon cable or enable SWDinstead of JTAG access in Debug Congurations - PROJECTNAME.elf -Debugger (c.f. Section 5.5.1).

For converting the nished software to a .hex le that can be ashed tothe STM32 using the ST-LINK Utility Software (c.f. Section A.2), theobjcopy.exe tool from the Codesourcery toolchain2 may be used. It canalso be found on the enclosed CD or in the PermaSense SVN repositoryat AEBoard/Tools. E.g., the following command makes a newly compiledversion of PermAE ready for the ST-LINK Utility Software:

objcopy.exe -O ihex PermAE.elf PermAE.hex

In order to use the EventGenerator software, simply attach your STM32Discoverydevice to the host PC and use the standard TrueSTUDIO debug congura-tion.

B.2 TinyNode

The TinyNode part of PermAE is heavily dependent on the PermaSense spe-cic TinyOS libraries. Therefore, its sourcecode is not included on the en-closed CD. Rather, the whole PermaSense specic TinyOS toolchain shouldbe installed as described in the PermaSense wiki3.

The PermAE application is located in apps/PermAE. This directory containsthe following les:

2http://www.codesourcery.com/sgpp/lite/arm3http://people.ee.ethz.ch/~nccr/permasense/wiki/TinyOS

90

http://www.codesourcery.com/sgpp/lite/arm

http://people.ee.ethz.ch/~nccr/permasense/wiki/TinyOS

B.2. TINYNODE

AcousticEmissions.h: The main header le, containing all PermAEspecic denitions and datatypes.

PermAEC.nc / PermAEP.nc: The top level application les.

AEModuleC.nc / AEModuleP.nC: The AEModule, handling the commu-nication with the STM32.

AEStorageQueue.h / AEStorageQueueC.nc / AEStorageQueueP.nc: Aslightly adapted version of the standard PermaDozer storage queue (c.f.Section 6.5.3).

AEParamControl.nc / ResetSplitControl.nc: Some interface deni-tions used in PermAE.

Makefile: The application's makele.

Furthermore, the PermAE application is dependent on some AEBoard spe-cic driver les. These are located in tos/sensorboards/AEBoard.

The AEBoard specic PermaDozer datasources can be found in

tos/lib/PermasenseData/datasources:

DataSourceAcousticEmissionsC.nc / DataSourceAcousticEmissionsP.ncas well as DataSourceAEBoardHealthC.nc / DataSourceAEBoardHealthP.nc.

For further details about the TinyNode part of PermAE, refer to Section6.5. For details about compiling and installing TinyOS applications on aTinyNode, please refer to the TinyOS and PermaSense documentation.

91


92

CPermAE Output Messages

In addition to some standard PermaDozer messages (eventlogger, rssi, state-counter), PermAE basically produces four dierent messages containing AEdata as well as system status information (c.f. Table 6.1). These messagesare specied in detail in Tables C.1 to C.4. Thereby, the following data typeconventions are used:

uint8 unsigned 8 bit integer number




In the Conv column, the conversion function to physical units is given withx denoting the eld's value. The Unit column then gives the according unitafter conversion. For a derivation of units, refer to Appendix D. When re-trieving data on the Public GSN instance (i.e., http://data.permasense.ch), data values have already been converted and are displayed in physicalunits.

93



APPENDIX C. PERMAE OUTPUT MESSAGES

aedata

Message ID: 0xCO

Field Type Conv Unit Meaning

serial uint32 x The event's serialnumber.

startSample uint32 2 · x µs Time of the event'srst threshold ex-ceedance.

length uint16 2 · x µs The event's lengthparameter.

risetime uint16 2 · x µs The event's rise timeparameter.

amplitude uint16 x·4.565536 V The event's ampli-

tude parameter.count uint16 x The event's pulse

count parameter.

energy uint32 x·4.52223

V2 The event's energyparameter.

posttrigger uint16 x samples The posttrigger valueused for parametriza-tion.

thresholdChannel uint8 c.f. D.1.4 Encoding of thethreshold value usedfor parametrizationand the AEBoardchannel the event wasmeasured on.

Table C.1: Message denition: aedata

94

aedaqhelathdata

Message ID: 0xC2


VccCh1 uint16 2 · x mV Bias voltage for AE channel1

VccCh2 uint16 2 · x mV Bias voltage for AE channel2

VccAdc uint16 1 · x mV ADC supply voltageRefAdc uint16 2 · x mV ADC reference voltageVccOpa uint16 2 · x mV Operation ampliers' supply

voltageVcc55V uint16 2.4 · x V 5.5V intermediate voltageI12V uint16 n/a n/a Not valid on AEBoard rev.

1.1 (c.f. Section 5.5.1)I55VAnalog uint16 n/a n/a Not valid on AEBoard rev.

1.1 (c.f. Section 5.5.1)I55VDigital uint16 n/a n/a Not valid on AEBoard rev.

1.1 (c.f. Section 5.5.1)SDEventsFree uint32 x Number of free event blocks

on the STM32's SD card

DAQPowerState uint8 x1: STM32 powered ON0: STM32 powered OFF

Table C.2: Message denition: aedaqhelathdata

aestatistics

Message ID: 0xC4


measured uint32 x The number of events detectedduring the last two minutes.

parametrized uint32 x The number of eventsparametrized and forwarded tothe TinyNode during the lasttwo minutes.

stored uint16 x The number of events stored onthe SD card during the last twominutes.

Table C.3: Message denition: aestatistics

95

APPENDIX C. PERMAE OUTPUT MESSAGES

nodehealth

Message ID: 0x80


sample uint16 x CurrentPermaSensesample number

uptime uint24 x s TinyNode uptimein seconds

sysvoltage uint16 1 · x mV TinyNode supplyvoltage

sdivoltage uint16 6 · x mV 12V externalsupply voltage

temperature uint16 x·175.7216384 − 46.85 C Onboard temper-

aturehumidity uint16 x·125

4096 − 6 % Onboard humid-ity

sibcurrent uint16 n/a n/a n/a

msptemperature uint16x· 1.5

4095−0.986

0.00355C MSP430 internal

temperatureashStatus uint16 x Bytes SD card storage

usedququeSize uint8 x Bytes Queue buer usedparentId uint16 x The parent

node's IDhopCount /childCount

uint8 x Dozer hop andchild count

Table C.4: Message denition: nodehealth

96

DData Units and Conversion Functions

The units and conversion functions used in this thesis are specied and de-rived in the following.

D.1 AE Data Units and Conversion Functions

D.1.1 Start Sample, Length and Risetime

As the AE signal sampling rate is xed to 500 kHz, the smallest time unitin PermAE is 2µs. Thus, the start sample, event length and event risetimeparameters are all measured in multiples of 2µs (c.f. Table C.1). Thus,with m denoting the time in samples and t denoting the time in seconds, thefollowing conversion functions apply:

t = 2 · 10−6 ·m [seconds]m = 5 · 105 · t [samples]

D.1.2 Amplitude and Threshold

Both amplitude and threshold refer to the signal level. This level can equiv-alently be measured in V or dB. The input signals at the AEBoard ADCsrange from 0 to 4.5V. This range is sampled with a resolution of 16 bit,

97

APPENDIX D. DATA UNITS AND CONVERSION FUNCTIONS

resulting in a base voltage unit of

4.5V

216= 68.66µV

and thus, the conversion functions from

All values in dB refer to a reference voltage of 0.1mV, i.e.,

xdB = 20 · log10

( xV0.0001V

)with xdB denoting the signal level in dB and xV denoting the signal level inV.

Thus, with xs additionally denoting the signal level in generic PermAE units,the following conversion functions apply:

xV = 68.66 · 10−6 · xs [V]xs = xV

68.66·10−6 [generic]

xdB = 20 · log10

(xs·68.66·10−6 V

0.0001V

)[dB]

xs = 10xdB20 · 0.0001V

68.66·10−6 V[generic]

xV = 10xdB20 · 0.0001 [V]

xdB = 20 · log10

(xV

0.0001V

)[dB]

D.1.3 Energy

During parametrization, the energy of a sample is calculated as

ε =∑ x2

s

29

with ε denoting the energy and xs denoting the sample value (c.f. Section6.4.4). Thus, the energy parameter's physical unit is V2. Given that thebasic sample value unit is equal to 4.5V

216, the energy parameter's conversion

functions are

e = ε·4.52223

[V 2]

ε = e·2234.52

[generic]

with e denoting the energy in V 2.

98

D.2. SYSTEM HEALTH DATA CONVERSION

D.1.4 Threshold and Channel

In order to t all AE event information into one PermaDozer packet, thethreshold used for event detection as well as the channel the event wasmeasured on were encoded into one 8 Byte value (c.f. Table C.1). ThethresholdChannel eld is built as follows:

TC = 2 · tdB + c− 1 c ∈ [1, 2]

where TC denotes the thresholdChannel eld value, tdB the threshold valuein dB and c the AE channel number the AE event was measured.

Thus, the thresholdChannel eld may be decoded as follows:

tdB = bTC2 c [dB]c = TC mod2− 1

D.2 System Health Data Conversion

D.2.1 Onboard Voltages

All onboard voltages are sampled with 12 bit ADCs, either by the STM32 orby the TinyNode (c.f. Figure 5.5). The reference voltages for these ADCs isthe corresponding device supply voltage, i.e., 2.8V for the STM32 and 3.3Vfor the TinyNode. In order to scale the measured voltages to this range,voltage dividers have been applied (c.f. Appendix E). Thus, with R1 and R2

denoting the divider resistors, vADC denoting the voltage at the ADC andvM the voltage to measure, it holds that

vADC = vM ·R2

R1 +R2

With x denoting the ADC output value received at the data backend andvREF denoting the ADC reference voltage, the following function may beused to convert it back to the original value:

vM = x ·vREF·

R2R1+R2

212[V]

Table D.1 lists the voltage divider values and conversion constants c =vREF·

R2R1+R2

212.

99


Field Message vREF [V] R1 [Ω] R2 [Ω] c [mV]VccCh1 aedaqhealthdata 2.8 100k 50k 2VccCh2 aedaqhealthdata 2.8 100k 50k 2VccAdc aedaqhealthdata 2.8 100k 200k 1RefAdc aedaqhealthdata 2.8 100k 50k 2VccOpa aedaqhealthdata 2.8 100k 50k 2Vcc55V aedaqhealthdata 3.3 100k 50k 2.4sysvoltage nodehealth 3.3 100k 200k 1sdivoltage nodehealth 3.3 100k 15k 6

Table D.1: Onboard voltage constants

D.2.2 Onboard Currents

For current measurements, the MAX9923H current sense ampliers by Maximare utilized on the AEBoard [61]. Their output voltage is then sampled withthe TinyNode's 12 bit ADC. As the MAX9923H features a constant gain of100V/V and input voltages are measured over 0.033Ω resistors, the relationbetween current I and output voltage vi is

I =100 · vi0.033Ω

[A]

Thus, as the output voltages are not fed through a voltage divider, the follow-ing conversion function from a current value xi (i.e., the I12V, I55VAnalogand I55VDigital elds in the aedaqhealthdata message) to the actual currentI applies:

I =100·xi· 3.3V

212

0.033Ω [A]

Note that due to a design error, current measurements on the AEBoard rev.1.1 are invalid (c.f. Section 5.5.1)

D.2.3 Temperature and Humidity

SHT21 output values

According to the SHT21 datasheet [58], temperature T and relative humidityH may be derived from the raw values xt and xh as follows:

T = xt·175.72214

− 46.85 [C]

H = xh·125212

− 6 [%]

100

D.2. SYSTEM HEALTH DATA CONVERSION

MSP430 Internal Temperature

The MSP430's internal temperature sensor value may be converted to C asfollows [62]:

TMSP430 =xmsp430· 1.5

4095−0.986

0.00355 [C]

101


102

EAEBoard PCB Schematics and Layout

The AEBoard's schematics and PCB layer prints are shown on the followingpages. Note that these schematics refer to AEBoard revision 1.1. Somechanges have been made for revision 1.2, c.f. Section 5.5.1. The accordingschematics for revision 1.2 may be found on the enclosed CD (c.f. AppendixG) or in the PermaSense SVN repository under

/trunk/pcb/projects/0011_acoustic_emission

103

Zeichnungs Titel:

Acoustic Emission Sensing - Analog Frontend

File:

Datum:

Zeichnungsnummer: 0001

2011-05-03

D:\Altium Projects\trunk\projects\0011_acoustic_emission\AnalogFrontend.SchDoc

16:20:16

Format:

A3

1.1Rev:

Ersteller:

Institut: ETH Zurich

Seite von1 5Josua Hunziker

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

A

B

C

D D

C

B

A

REF

1

VD

D2

IN+3

IN-4

GN

D5

CNV 6SDO 7SCK 8SDI 9VIO 10

ADC1AD7980

DAQ Channel 1

VOUT 1

-VS

2

IN+3

IN-4

PDN

5+V

S6

U1ADA4841

VCC_4.5_OPA

GND

GND

VCC_2.5

GND

VCC_2.8_SW_ADC

GND

GND

VCC_4.5_CH1_DECOUPLED

Signal Conditioning CircuitryGain=1, Voltage Shift=+2.25V-3dB Bandwidth: 4.6kHz to 220kHz

GND

GND

REF

1

VD

D2

IN+3

IN-4

GN

D5

CNV 6SDO 7SCK 8SDI 9VIO 10

ADC2AD7980

DAQ Channel 2

VOUT 1

-VS

2

IN+3

IN-4

PDN

5+V

S6

U2ADA4841

VCC_4.5_OPA

GND

GND

VCC_2.5

GND

VCC_2.8_SW_ADC

GND

GND

VCC_4.5_CH2_DECOUPLED

Signal Conditioning CircuitryGain=1, Voltage Shift=+2.25V-3dB Bandwidth: 4.6kHz to 220kHz

GND

GND

AD

C_R

EF_4

.5A

DC

_REF

_4.5

ADC_SCK

ADC_CNV

ADC_SCK

ADC_CNV

GND

VCC_4.5_OPA

GND

VCC_4.5_OPA

ADC Reference Voltage

VCC_5.5_SW

10uFTantalum

C15

GND

V+ 2SLE3

GND4

OUT 6

TP 5

NC8

TP 1

NC7

U3

REF194ES

GND

CH1SNS

12

P2

GNDGND

ESD Protection

CH1ADC

12

P3

GND

CH2ADC

12

P6

GND

CH2SNS

12

P5

GNDGND

ESD Protection

D2SM6T10A

D1SM6T10A

AD

C_R

EF_4

.5

ADC1_SDO

ADC2_SDO

P1SMA

P4SMA

GND

GND

6.8nF

C5

5.1k

R2

5.1kR1

100k

R4

100kR5

330R

R3

100nF

C2

2.2nFC6

10uFCeramic X5R

C1

10uFCeramic X5R

C7

100nF

C3

100nF

C4

6.8nF

C11

5.1k

R7

5.1kR6

100k

R9

100kR10

330R

R8

100nF

C8

2.2nFC12

100nF

C9

100nF

C10

100nFC14

100nFC13

Input Signal Properties:AC Waveform, 4.5V Offset, 2.25V AmplitudeFrequency Range: ca. 20-100kHz

Input Signal Properties:AC Waveform, 4.5V Offset, 2.25V AmplitudeFrequency Range: ca. 20-100kHz

PIADC101 PIADC102PIADC103

PIADC104

PIADC105PIADC106

PIADC107

PIADC108

PIADC109

PIADC1010

COADC1

PIADC201 PIADC202PIADC203

PIADC204

PIADC205PIADC206

PIADC207

PIADC208

PIADC209

PIADC2010

COADC2

PIC101PIC102

COC1

PIC201 PIC202

COC2

PIC301 PIC302

COC3

PIC401 PIC402

COC4

PIC501 PIC502

COC5

PIC601

PIC602COC6

PIC701PIC702

COC7

PIC801 PIC802

COC8

PIC901 PIC902

COC9

PIC1001 PIC1002

COC10

PIC1101 PIC1102

COC11

PIC1201

PIC1202COC12

PIC1301

PIC1302COC13

PIC1401

PIC1402COC14

PIC1501

PIC1502COC15

PID101PID102

COD1

PID201PID202

COD2

PIP101

PIP102

COP1

PIP201

PIP202

COP2

PIP301

PIP302

COP3

PIP401

PIP402

COP4

PIP501

PIP502

COP5

PIP601

PIP602

COP6

PIR101

PIR102COR1

PIR201 PIR202

COR2

PIR301 PIR302

COR3

PIR401 PIR402

COR4

PIR501

PIR502COR5

PIR601

PIR602COR6

PIR701 PIR702

COR7

PIR801 PIR802

COR8

PIR901 PIR902

COR9

PIR1001

PIR1002COR10

PIU101

PIU102PIU103

PIU104

PIU105PIU106COU1

PIU201

PIU202PIU203

PIU204

PIU205PIU206COU2

PIU301

PIU302PIU303

PIU304 PIU305

PIU306PIU307

PIU308

COU3

PIADC107NLADC10SDO

PIADC207NLADC20SDO

PIADC106

PIADC206NLADC0CNV

PIADC101

PIADC201

PIC101

PIC701

PIC1401 PIC1501PIU306

NLADC0REF0405

PIADC108

PIADC208NLADC0SCK

PIADC104

PIADC105

PIADC204

PIADC205

PIC102PIC202

PIC302 PIC402

PIC602

PIC702PIC802

PIC902 PIC1002

PIC1202

PIC1302

PIC1402 PIC1502

PID101

PID201

PIP102

PIP202

PIP302

PIP402

PIP502

PIP602

PIR501

PIR1001

PIU102

PIU202

PIU304

PIADC103

PIC601PIP301

PIR302

PIADC203

PIC1201PIP601

PIR802

PIC502 PIR201

PIC1102 PIR701

PIR101

PIR202 PIU104

PIR102

PIR301PIU101

PIR402

PIR502

PIU103

PIR601PIR702 PIU204

PIR602

PIR801PIU201

PIR902

PIR1002PIU203

PIU301

PIU305

PIU307

PIU308

PIADC102

PIADC202

PIC301

PIC901

PIADC109

PIADC1010

PIADC209

PIADC2010

PIC401

PIC1001

PIC501

PID102PIP101

PIP201

PIC1101

PID202PIP401

PIP501

PIC201

PIC801

PIR401

PIR901

PIU105PIU106

PIU205PIU206

PIC1301PIU302PIU303

104

Zeichnungs Titel:

Acoustic Emission Sensing - STM32 (ARM Cortex M3)

File:

Datum:


2011-05-03

D:\Altium Projects\trunk\projects\0011_acoustic_emission\STM32.SchDoc

16:20:16

Format:

A3

1.1Rev:

Ersteller:



1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

A

B

C

D D

C

B

A

BOOT060

NRST7

OSC_IN/PD05

OSC_OUT/PD16

PA0-WKUP14

PA115

PA216

PA317

PA420

PA521

PA622

PA723

PA841

PA942

PA1043

PA1144

PA1245

PA13/JTMS/SWDIO46

PA14/JTCK/SWCLK49

PA15/JTDI50

PB0 26

PB1 27

PB2/BOOT1 28

PB3/JTDO 55

PB4/JNTRST 56

PB5 57

PB6 58

PB7 59

PB8 61

PB9 62

PB10 29

PB11 30

PB12 33

PB13 34

PB14 35

PB15 36

PC0 8

PC1 9

PC2 10

PC3 11

PC4 24

PC5 25

PC6 37

PC7 38

PC8 39

PC9 40

PC10 51

PC11 52

PC12 53

PC13-TAMPER-RTC 2

PC14-OSC32_IN 3

PC15-OSC32_OUT 4

PD254

VBAT1

VDD_132

VDD_248

VDD_364

VDD_419

VDDA13

VSS_1 31

VSS_2 47

VSS_3 63

VSS_4 18

VSSA 12

U10

STM32F103RDT6

GND

VCC_2.8_SW_ARMGND

ARM RSTS2

Q9FDV301N

GND

AR

M_R

ST_C

TL

ARM_RST

ARM_RST

ARM Reset Switch

ARM_JTCKARM_JTMS

ARM Debug & Programming Port (ST-LINK)

GND

GND

ARM_OSC_INARM_OSC_OUT

ARM_OSC_IN

1k5

R36 ARM_OSC_OUT

ARM HF Oscillator

ARM Power Decoupling Capacitors

VCC_2.8_SW_ARM

GND

123

W6

VCC_2.8_SW_ARM

GND

P1connected: Flash MemoryP3 connected: System Memory

ARM_BOOT0

ARM Bootmode Selection

ARM_BOOT0

GND

GND

ARM_OSC32_IN

ARM_OSC32_OUT

ARM LF Oscillator

0R

R37

ARM_OSC32_INARM_OSC32_OUT

ARM_WKUP

ADC_SCK

ADC1_SDO

ADC2_SDO

ADC_SCK

ARM_SDIO_CMD

ARM_SDIO_CKARM_SDIO_D3ARM_SDIO_D2

ARM_SDIO_D0ARM_SDIO_D1

ADC_CNVADC_SCK

ARM SD Card



ARM_SDIO_CK

ARM_SDIO_CMD

GND

VCC_2.8_SW_ARM

51kR35

51kR34

51kR33

51kR32

51kR31

GND

ARM_TN_RXARM_TN_TX

TN_COMM_FLAGARM_COMM_FLAG

VCC_CH1_SNSVCC_CH2_SNS

VCC_ADC_SNSREF_ADC_SNSVCC_OPA_SNS

100nFC46

100nFC50

100nFC48

100nFC47

100nFC49

10nFC44

27pF

C41

27pF

C42

15pF

C43

15pF

C45

100nF

C40GND

32.768kHz12.5pF load, <50k ESR

Y2FC-135

6MHz

Y1PBRC6.00GR50X000

fmax=72MHz384K Flash Memory64K SRAM

12345

CONSOLE

6

P9

GND

CONSOLE_CTS

CONSOLE_TXDCONSOLE_RXDCONSOLE_RTS

ARM Console Connector (ready-to-use with TTL-232R-3V3)

VCC_2.8_SW_ARM

51kR41

51kR40

51kR39

51kR38

CONSOLE_RXDCONSOLE_TXD

CONSOLE_CTSCONSOLE_RTS

330R

R42

LED1RED

D5

ARM_LED1ARM_LED2ARM_LED3

330R

R43

LED2ORANGE

D6

330R

R44

LED3GREEN

D7

GND

GND

GND

ARM_LED1

ARM_LED2

ARM_LED3

ARM Status LEDs

ARM_PA4

ARM_PA6

ARM_PA8

ARM_JTDI

ARM_PA8ARM_PA6ARM_PA4

ARM_JTDO

ARM_PB1

ARM_JTRSTARM_PB5

ARM_PB5

ARM_PB12ARM_PB9ARM_PB8

ARM_PC13ARM_PC4ARM_PC3

ARM Breakout Pins for unused GPIOs

ARM_PB8ARM_PB9

ARM_PB12

DAQ_PW_CTL

ARM_PC13

ARM_PC3ARM_PC4

ARM JTAG

1 23 45 67 89 1011 1213 1415 1617 1819 20

P8VCC_2.8_SW_ARMVCC_2.8_SW_ARM

GND

ARM_JTDO

GND

ARM_JTMSARM_JTCK

ARM_JTDIARM_JTRST

ARM_PB1

123

PC3 PC4 PC13

P11

12345

PB1 PB5 PB8 PB9 PB12

P12

123

PA4 PA6 PA8

P10

DAT2CS/CD/DAT3MOSI/CMDGNDVDDSCLKGNDMISO/DAT0DAT1 G

ND

SWD

SW

WP SD1

PIC4001 PIC4002

COC40

PIC4101PIC4102

COC41

PIC4201PIC4202

COC42

PIC4301PIC4302

COC43PIC4401

PIC4402COC44

PIC4501PIC4502

COC45

PIC4601

PIC4602COC46

PIC4701

PIC4702COC47

PIC4801

PIC4802COC48

PIC4901

PIC4902COC49

PIC5001

PIC5002COC50

PID501 PID502

COD5

PID601 PID602

COD6

PID701 PID702

COD7

PIP801 PIP802

PIP803 PIP804

PIP805 PIP806

PIP807 PIP808

PIP809 PIP8010

PIP8011 PIP8012

PIP8013 PIP8014

PIP8015 PIP8016

PIP8017 PIP8018

PIP8019 PIP8020

COP8

PIP901

PIP902

PIP903

PIP904

PIP905

PIP906

COP9

PIP1001

PIP1002

PIP1003

COP10

PIP1101

PIP1102

PIP1103

COP11

PIP1201

PIP1202

PIP1203

PIP1204

PIP1205

COP12

PIQ901PIQ902

PIQ903

COQ9

PIR3101

PIR3102

COR31PIR3201

PIR3202

COR32PIR3301

PIR3302

COR33PIR3401

PIR3402

COR34PIR3501

PIR3502

COR35

PIR3601PIR3602

COR36

PIR3701PIR3702

COR37

PIR3801

PIR3802

COR38PIR3901

PIR3902

COR39PIR4001

PIR4002

COR40PIR4101

PIR4102

COR41

PIR4201 PIR4202COR42

PIR4301 PIR4302

COR43

PIR4401 PIR4402

COR44

PIS201

PIS202COS2

PISD101

PISD102

PISD103

PISD104

PISD105

PISD106

PISD107

PISD108

PISD109

PISD10GNDPISD10SW PISD10SWD

PISD10WPCOSD1

PIU1001

PIU1002

PIU1003

PIU1004

PIU1005

PIU1006

PIU1007

PIU1008

PIU1009

PIU10010

PIU10011

PIU10012PIU10013

PIU10014

PIU10015

PIU10016

PIU10017

PIU10018PIU10019

PIU10020

PIU10021

PIU10022

PIU10023

PIU10024

PIU10025

PIU10026

PIU10027

PIU10028

PIU10029

PIU10030

PIU10031PIU10032

PIU10033

PIU10034

PIU10035

PIU10036

PIU10037

PIU10038

PIU10039

PIU10040

PIU10041

PIU10042

PIU10043

PIU10044

PIU10045

PIU10046

PIU10047PIU10048

PIU10049

PIU10050

PIU10051

PIU10052

PIU10053

PIU10054

PIU10055

PIU10056

PIU10057

PIU10058

PIU10059

PIU10060

PIU10061

PIU10062

PIU10063PIU10064

COU10

PIW601

PIW602

PIW603

COW6

PIY101PIY102 COY1

PIY201PIY202

COY2

PIC4401PIQ901PIS202 PIU1007NLA\R\M\0\R\S\T\

PIU10023NLADC10SDO

PIU10036NLADC20SDO

PIU10015NLADC0CNV

PIU10021

PIU10026

PIU10034NLADC0SCK

PIU10060

PIW602

NLARM0BOOT0

PIU10059NLARM0COMM0FLAG

PIP809

PIU10049NLARM0JTCK

PIP8013

PIU10050NLARM0JTDI

PIP805

PIU10055NLARM0JTDO

PIP807

PIU10046NLARM0JTMS

PIP8015

PIU10056NLARM0JTRST

PIR4201

PIU10025NLARM0LED1

PIR4301

PIU10037NLARM0LED2

PIR4401

PIU10038NLARM0LED3PIC4301

PIU1003

PIY201

NLARM0OSC320INPIR3701

PIU1004NLARM0OSC320OUT

PIC4101

PIU1005

PIY102

NLARM0OSC0IN

PIR3601

PIU1006NLARM0OSC0OUT

PIP1001

PIU10020NLARM0PA4

PIP1002

PIU10022NLARM0PA6

PIP1003

PIU10041NLARM0PA8

PIP1201

PIU10027NLARM0PB1

PIP1202

PIU10057NLARM0PB5

PIP1203

PIU10061NLARM0PB8

PIP1204

PIU10062NLARM0PB9

PIP1205

PIU10033NLARM0PB12

PIP1101

PIU10011NLARM0PC3

PIP1102

PIU10024NLARM0PC4

PIP1103

PIU1002NLARM0PC13

PIQ902

NLARM0RST0CTL

PISD105

PIU10053

NLARM0SDIO0CK

PIR3302

PISD102

PIU10054

NLARM0SDIO0CMD

PIR3402

PISD107

PIU10039

NLARM0SDIO0D0

PIR3502

PISD108

PIU10040

NLARM0SDIO0D1

PIR3102PISD109

PIU10051

NLARM0SDIO0D2PIR3202

PISD101

PIU10052

NLARM0SDIO0D3

PIU10029NLARM0TN0RX

PIU10030NLARM0TN0TX

PIU10014NLARM0WKUP

PIP902

PIR3802

PIU10045

NLCONSOLE0CTS

PIP906

PIR4102

PIU10044

NLCONSOLE0RTS PIP905

PIR4002

PIU10042

NLCONSOLE0RXD PIP904

PIR3902

PIU10043

NLCONSOLE0TXD

PIU10035NLDAQ0PW0CTL

PIC4002

PIC4102

PIC4202

PIC4302

PIC4402

PIC4502

PIC4602 PIC4702 PIC4802 PIC4902 PIC5002

PID502

PID602

PID702

PIP803 PIP804

PIP806

PIP808

PIP8010

PIP8012

PIP8014

PIP8016

PIP8018

PIP8020

PIP901

PIQ903PIS201

PISD103

PISD106


PISD10WP

PIU10012

PIU10018

PIU10028

PIU10031

PIU10047

PIU10063

PIW601

PIC4201 PIR3602

PIY101

PIC4501 PIR3702

PIY202

PID501PIR4202

PID601PIR4302

PID701PIR4402

PIP8011

PIP8017

PIP8019

PIP903

PIU1009NLREF0ADC0SNS

PIU10058NLTN0COMM0FLAG

PIC4001

PIC4601 PIC4701 PIC4801 PIC4901 PIC5001

PIP801 PIP802 PIR3101 PIR3201 PIR3301 PIR3401 PIR3501

PIR3801 PIR3901 PIR4001 PIR4101

PISD104

PIU1001

PIU10013

PIU10019

PIU10032

PIU10048

PIU10064

PIW603

NLVCC02080SW0ARM

PIU1008NLVCC0ADC0SNS

PIU10016NLVCC0CH10SNS

PIU10017NLVCC0CH20SNS

PIU10010NLVCC0OPA0SNS

105

Zeichnungs Titel:

Acoustic Emission Sensing - TinyNode Connector

File:

Datum:


2011-05-03

D:\Altium Projects\trunk\projects\0011_acoustic_emission\TinyNode.SchDoc

16:20:17

Format:

A3

1.1Rev:

Ersteller:



1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

A

B

C

D D

C

B

A

TDO1 VCC 2

TDI3 VReg 4

TMS5 GND 6

TCK7 nRST 8

URXD19 nREGE 10

UTXD11 P5.0/STE1 12

P1.2/TA113 P5.1/SIMO1 14

P1.3/TA215 P5.2/SOMI1 16

E_EVREF17 P5.3/UCLK1 18

P6.7/ADC7/DAC119 P1.6/TA1 20

P6.6/ADC6/DAC021 P2.3/CA0/TA1 22

P6.5/ADC523 P2.4/CA1/TA2 24

P6.4/ADC425 P4.0/TB0 26

P6.3/ADC327 T4.1/TB1 28

P6.2/ADC229 E_IVREF 30

J2 53364-3071

TDO1 VCC_IN 2

TDI3 VCC_OUT 4

TMS5 NC 6

TCK7 TEST 8

GND9 NC 10

RST NC 12

NC13 NC 14

J1

VCC_2.8

TinyNode JTAG

GND

TN_RST

TN_JTCK

TN_JTMS

TN_JTDI

TN_JTDO

TN_JTCK

TN_JTMS

TN_JTDI

TN_JTDO

TNS3

GND

TN_RST

TinyNode Reset

TN_RST

GND

VCC_2.8

ARM_TN_RX

ARM_TN_TX

TN_COMM_FLAG

ARM_COMM_FLAG

GND

SHT21_SDA

SHT21_SCL

10KR51

100nFC53

Temperature/Humidity Sensor

VCC_2.8

VSS 2

SDA1

SCL4

VDD 3U11

SHT21

10KR50

SHT21_SCL

SHT21_SDA

TN_SD_CS

TN_SD_MOSI

TN_SD_MISO

TN_SD_SCK

TinyNode SD Card

TN_SD_MISO

TN_SD_CS

TN_SD_SCK

TN_SD_MOSI

VCC_2.8

51kR49

51kR48

51kR47

51kR46

51kR45

ARM_TN_TX

ARM_TN_RX

VCC_2.8_SW_ARM

GND

330R

R54

GND

1MR52

100nFC55

100nFC56

TN_TXORANGE

D8

VCC_2.8_SW_ARM

GND

330R

R57

GND

1MR55

100nFC58

100nFC59

TN_RXORANGE

D9

UART Traffic LEDs

GND

GND

TRIG2

OUT 3

RST4

CVOLT 5

THR6

DIS 7

V+8 GND 1

U12

LMC555CMM

TRIG2

OUT 3

RST4

CVOLT 5

THR6

DIS 7

V+8 GND 1

U13

LMC555CMM

VCC_12.0_SNS

I_VCC_12.0_SNS

VCC_5.5_SNS

I_VCC_5.0_A_SNS

VCC_2.8_SNS

I_VCC_5.0_D_SNS

ARM_WKUP

100KR53

100KR56

ARM_RST_CTL

100nF

C51GND

100nFC52

100nF

C54

100nF

C57

ARM_PW_CTL

DAT2CS/CD/DAT3MOSI/CMDGNDVDDSCLKGNDMISO/DAT0DAT1 G

ND

SWD

SW

WP SD2

GND

PIC5101 PIC5102

COC51

PIC5201

PIC5202COC52

PIC5301

PIC5302COC53

PIC5401 PIC5402

COC54

PIC5501

PIC5502COC55

PIC5601

PIC5602COC56

PIC5701 PIC5702

COC57

PIC5801

PIC5802COC58

PIC5901

PIC5902COC59

PID801 PID802

COD8

PID901 PID902

COD9

PIJ101 PIJ102

PIJ103 PIJ104

PIJ105 PIJ106

PIJ107 PIJ108

PIJ109 PIJ1010

PIJ1011 PIJ1012

PIJ1013 PIJ1014

COJ1

PIJ201 PIJ202

PIJ203 PIJ204

PIJ205 PIJ206

PIJ207 PIJ208

PIJ209 PIJ2010

PIJ2011 PIJ2012

PIJ2013 PIJ2014

PIJ2015 PIJ2016

PIJ2017 PIJ2018

PIJ2019 PIJ2020

PIJ2021 PIJ2022

PIJ2023 PIJ2024

PIJ2025 PIJ2026

PIJ2027 PIJ2028

PIJ2029 PIJ2030

COJ2

PIR4501

PIR4502

COR45PIR4601

PIR4602

COR46PIR4701

PIR4702

COR47PIR4801

PIR4802

COR48PIR4901

PIR4902

COR49

PIR5001

PIR5002

COR50PIR5101

PIR5102

COR51

PIR5201

PIR5202COR52

PIR5301

PIR5302COR53

PIR5401 PIR5402

COR54

PIR5501

PIR5502COR55

PIR5601

PIR5602COR56

PIR5701 PIR5702

COR57

PIS301

PIS302COS3

PISD201

PISD202

PISD203

PISD204

PISD205

PISD206

PISD207

PISD208

PISD209


PISD20WPCOSD2

PIU1101

PIU1102

PIU1103

PIU1104

COU11

PIU1201

PIU1202

PIU1203

PIU1204

PIU1205

PIU1206

PIU1207

PIU1208

COU12

PIU1301

PIU1302

PIU1303

PIU1304

PIU1305

PIU1306

PIU1307

PIU1308

COU13

PIJ2020NLARM0COMM0FLAG

PIJ2015NLARM0PW0CTL

PIJ2026NLARM0RST0CTL

PIC5701

PIJ209NLARM0TN0RX

PIC5401

PIJ2011NLARM0TN0TX

PIJ2024NLARM0WKUP

PIC5102

PIC5202

PIC5302

PIC5502PIC5602

PIC5802PIC5902

PID802

PID902

PIJ109

PIJ206

PIS301

PISD203

PISD206


PISD20WP

PIU1102

PIU1201

PIU1301

PIJ2029NLI0VCC05000A0SNS

PIJ2027NLI0VCC05000D0SNS

PIJ2025NLI0VCC012000SNS

PIC5402

PIR5301PIU1202

PIC5601

PIR5201PIU1206

PIU1207

PIC5702

PIR5601PIU1302

PIC5901

PIR5501PIU1306

PIU1307

PID801PIR5402

PID901PIR5702

PIJ102

PIJ106

PIJ108

PIJ1010

PIJ1012

PIJ1013 PIJ1014

PIJ204

PIJ2010

PIJ2017

PIJ2030

PIR4502

PISD208

PIR4902PISD209

PIR5401PIU1203

PIR5701PIU1303

PIU1205

PIU1305

PIJ2028

PIR5002

PIU1104NLSHT210SCL

PIJ2013

PIR5102PIU1101

NLSHT210SDA

PIJ1011

PIJ208

PIS302

NLT\N\0\R\S\T\

PIJ2012

PIR4802

PISD201NLT\N\0\S\D\0\C\S\

PIJ2022NLTN0COMM0FLAG

PIJ107

PIJ207NLTN0JTCK

PIJ103

PIJ203NLTN0JTDI

PIJ101

PIJ201NLTN0JTDO

PIJ105

PIJ205NLTN0JTMS

PIJ2016

PIR4602

PISD207NLTN0SD0MISOPIJ2014

PIR4702

PISD202NLTN0SD0MOSI

PIJ2018

PISD205NLTN0SD0SCK

PIC5101PIC5201

PIC5301

PIJ104

PIJ202

PIR4501 PIR4601 PIR4701 PIR4801 PIR4901

PIR5001 PIR5101

PISD204

PIU1103

PIJ2023NLVCC02080SNS

PIC5501

PIC5801

PIR5202PIR5302

PIR5502PIR5602

PIU1204

PIU1208

PIU1304

PIU1308



106

Zeichnungs Titel:

Acoustic Emission Sensing - Power Supply

File:

Datum:


2011-05-03

D:\Altium Projects\trunk\projects\0011_acoustic_emission\Power.SchDoc

16:20:17

Format:

A3

1.1Rev:

Ersteller:



1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

A

B

C

D D

C

B

A

Sensor Power Channel 2

VCC_5.5_SW VCC_4.5_CH2_DECOUPLED

10uFTantalum

C33

GNDSignal DecouplingLowpass, fc=306Hz

V+ 2SLE3

GND4

OUT 6

TP 5

NC8

TP 1

NC7

U7

REF194ES

GND

Sensor Power Channel 1

VCC_5.5_SW VCC_4.5_CH1_DECOUPLED

10uFTantalum

C32

GND

Signal DecouplingLowpass, fc=306Hz

V+ 2SLE3

GND4

OUT 6

TP 5

NC8

TP 1

NC7

U8

REF194ES

GND

Op Amp Power

VCC_5.5_SW VCC_4.5_OPA

10uFTantalum

C21

GND

V+ 2SLE3

GND4

OUT 6

TP 5

NC8

TP 1

NC7

U4

REF194ES

GND

ADC Power

VCC_2.8_SW_ADC VCC_2.5

10uFTantalum

C20

GND

GND

V+ 2SLE3

GND4

OUT 6

TP 5

NC8

TP 1

NC7

U5

REF192ES

Q3

FDV302P

ARM Power Switch

VCC_2.8_SW_ARM

100RR23

470KR24

VCC_2.8

ARM_PW_CTL

VCC_12.0

MODE2

GND1

RSET3

VC4

VFB5 SS 6

SENSE 7

VCC 8

GATE 9

CAP 10

11

U6 LTC3824

GND

W1: placed: fixed freq, unplaced: burst mode

392KR12

10KR13

3.3nF

C26GND

GND

100nF

C27

GND475K1%

R1851RR17

100pFC36

80K61%

R16

GND

Q1Si7465DP

0R025R11

100nF>= 16V

C25

100nF>= 63VC28

100pFC24

GND

33uFC16

GND

GND

2.2uFC19

D3SS3H9

GND

22uHL1

100uFC29

GND

VCC_5.5_nSENSED_D

1 2

W1

Switching Voltage Regulator (Output: 5V, Current Limiter set to 2A)

VCC_2.8 VCC_5.5_D

GND

GND

10nF

C3910uFTantalum<3R ESR

C37

GND

4.7uFTantalum

C38

Linear Voltage Regulator (Input: 5.5V, Output: 2.84V/3.33V)

OUT1

SENSE/ADJ2

BYP3

GND4 SHDN 5

NC 6

NC 7

IN 8U9

LT1962

324kR20

249kR27

GND

123

W2

422kR19

P1connected: 2.84VP3 connected: 3.33V

GND

Q6FDV301N

GND

3M3R29

GND

Q4

FDV302P

ADC Power Switch

VCC_2.8_SW_ADC

100RR25

470KR26

VCC_2.8

DAQ_PW_CTL

GND

Q7FDV301N

GND

3M3R30

GND

Q2

FDV302P

DAQ Power Switch

VCC_5.5_SW

100RR21

470KR22

VCC_5.5_A

DAQ_PW_CTL

GND

Q5FDV301N

GND

3M3R28

GND

VCC_4.5_CH1 VCC_4.5_CH2

VCC_5.5_nSENSED_A

VCC_5.5

GND

V12DC_IN

1234

P72

3

1

S1

MFP106D

Q8FDS6681Z

Power Input (reverse polarity protected)

D4SM6T18CA

VCC_12.0_nSENSED

100nFC22

100nFC35

100nFC34

100nFC23

100nFC17

100nFC31

100nFC18

100nFC30

52RR14

52RR15

12

ARM_ONW4

12

ADC_ONW5

12

DAQ_ONW3

PIC1601 PIC1602

COC16

PIC1701

PIC1702COC17

PIC1801

PIC1802COC18

PIC1901 PIC1902

COC19 PIC2001

PIC2002

COC20PIC2101

PIC2102COC21PIC2201

PIC2202COC22

PIC2301

PIC2302COC23

PIC2401

PIC2402COC24

PIC2501PIC2502

COC25

PIC2601PIC2602

COC26

PIC2701PIC2702

COC27 PIC2801PIC2802

COC28 PIC2901PIC2902

COC29

PIC3001

PIC3002COC30

PIC3101

PIC3102COC31

PIC3201

PIC3202COC32 PIC3301

PIC3302COC33PIC3401

PIC3402COC34

PIC3501

PIC3502COC35PIC3601

PIC3602COC36

PIC3701

PIC3702

COC37 PIC3801

PIC3802COC38

PIC3901 PIC3902

COC39

PID301PID302

COD3

PID401

PID402COD4

PIL101PIL102

COL1

PIP701

PIP702

PIP703

PIP704

COP7

PIQ101PIQ102

PIQ103COQ1

PIQ201

PIQ202

PIQ203COQ2

PIQ301

PIQ302

PIQ303COQ3

PIQ401

PIQ402

PIQ403COQ4

PIQ501PIQ502

PIQ503COQ5

PIQ601PIQ602

PIQ603COQ6

PIQ701PIQ702

PIQ703COQ7

PIQ801

PIQ802

PIQ803

COQ8

PIR1101

PIR1102COR11



PIR1401

PIR1402COR14

PIR1501

PIR1502COR15

PIR1601

PIR1602COR16

PIR1701

PIR1702COR17

PIR1801

PIR1802COR18

PIR1901

PIR1902COR19

PIR2001

PIR2002COR20

PIR2101

PIR2102COR21

PIR2201

PIR2202COR22

PIR2301

PIR2302COR23

PIR2401

PIR2402COR24

PIR2501

PIR2502COR25

PIR2601

PIR2602COR26

PIR2701

PIR2702COR27

PIR2801

PIR2802COR28

PIR2901

PIR2902COR29

PIR3001

PIR3002COR30

PIS101

PIS102

PIS103COS1

PIU401

PIU402PIU403

PIU404 PIU405

PIU406PIU407

PIU408

COU4

PIU501

PIU502PIU503

PIU504 PIU505

PIU506PIU507

PIU508

COU5

PIU601

PIU602

PIU603

PIU604

PIU605 PIU606

PIU607

PIU608

PIU609

PIU6010

PIU6011

COU6

PIU701

PIU702PIU703

PIU704 PIU705

PIU706PIU707

PIU708

COU7

PIU801

PIU802PIU803

PIU804 PIU805

PIU806PIU807

PIU808

COU8

PIU901

PIU902

PIU903

PIU904 PIU905

PIU906

PIU907

PIU908

COU9

PIW101 PIW102

COW1

PIW201

PIW202

PIW203

COW2PIW301

PIW302

COW3

PIW401

PIW402

COW4

PIW501

PIW502

COW5

PIQ602PIR2902NLARM0PW0CTL PIQ502PIQ702 PIR2802PIR3002

NLDAQ0PW0CTL

PIC1602

PIC1702 PIC1802

PIC1902

PIC2002PIC2102PIC2202 PIC2302

PIC2602

PIC2702 PIC2802 PIC2902

PIC3002PIC3102

PIC3202 PIC3302PIC3402PIC3502

PIC3702 PIC3802

PID302

PID401PIP701

PIP702

PIQ503PIQ603 PIQ703

PIQ802

PIR1201

PIR1601

PIR2701

PIR2801PIR2901 PIR3001

PIU404 PIU504

PIU601

PIU6011

PIU704PIU804

PIU904

PIW101

PIW302PIW402 PIW502

PIC2402

PIQ101

PIR1101

PIU607

PIC2502PIU6010

PIC2601 PIR1301

PIC2701PIU606

PIC3601

PIR1701

PIC3902PIU903

PID301

PIL102

PIQ103

PID402PIP703

PIP704

PIQ801

PIQ102PIU609

PIQ202PIR2102

PIQ302PIR2302

PIQ402PIR2502

PIQ501PIR2101 PIR2201

PIW301


PIW401


PIW501

PIQ803 PIS102

PIR1202 PIU603

PIR1302 PIU604

PIR1602 PIR1702

PIR1802

PIU605

PIR1901PIW203

PIR2001

PIW201PIR2702

PIU902

PIW202

PIS103

PIU401

PIU405

PIU407

PIU408

PIU501

PIU505

PIU507

PIU508

PIU602

PIW102

PIU701

PIU705

PIU707

PIU708

PIU801

PIU805

PIU807

PIU808

PIU906

PIU907

PIC2001PIC2301PIU506

PIC3701

PIC3901

PIQ303 PIQ403

PIR1902 PIR2002PIR2402 PIR2602

PIU901

PIC1801

PIQ401

PIU502PIU503

PIQ301

PIC3201PIC3501

PIR1401PIU806

NLVCC04050CH1

PIR1402

PIC3301PIC3401

PIR1501PIU706

NLVCC04050CH2

PIR1502

PIC2101PIC2201PIU406

PIC2901

PIC3602

PIL101

PIR1801

NLVCC0505

PIQ203

PIR2202PIC3801

PIU905

PIU908

PIC1701

PIC3001PIC3101

PIQ201

PIU402PIU403

PIU702PIU703PIU802PIU803

PIC1601

PIC1901

PIC2401

PIC2501

PIC2801

PIR1102

PIU608

PIS101NLVCC012000nSENSED

107

Zeichnungs Titel:

Acoustic Emission Sensing - Board Health Sensing

File:

Datum:


2011-05-03

D:\Altium Projects\trunk\projects\0011_acoustic_emission\HealthSens.SchDoc

16:20:17

Format:

A3

1.1Rev:

Ersteller:



1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

A

B

C

D D

C

B

A

VCC Voltage Dividers

VCC_12.0

GND

100kR58

15kR60

VCC_12.0_SNS

0-19.3V

VCC_5.5

GND

100kR63

50kR65

VCC_5.5_SNS

0-8.4V

VCC_2.8

GND

100kR68

200kR70

VCC_2.8_SNS

0-4.2V

VCC_2.5

GND

100kR72

200kR75

VCC_ADC_SNS

0-4.2V

ADC_REF_4.5

GND

100kR59

50kR61

REF_ADC_SNS

0-8.4V

VCC_4.5_OPA

GND

100kR64

50kR66

VCC_OPA_SNS

0-8.4V

VCC_4.5_CH1

GND

100kR69

50kR71

VCC_CH1_SNS

0-8.4V

VCC_4.5_CH2

GND

100kR73

50kR76

VCC_CH2_SNS

0-8.4V

VCC_5.0

GND

VCC_2.8

GND

VCC_2.5

GND

VCC_4.5_OPA

GND

VCC_5.0

12

P18

VCC_2.8

12

P19

VCC_2.5

12

P15

VCC_4.5_OPA

12

P16

Voltage Breakout for Probes

VCC_2.8_SW_ARM

GND

VCC_2.8_SW

12

P14

VCC_4.5_CH1

GND

VCC_4.5_CH1

12

P13

VCC_4.5_CH2

GND

VCC_4.5_CH2

12

P17

VCC_12.0_nSENSEDVCC_12.0

RSB1

RS+2

RS-3

N.C.4

GND5 SHDN 6

REF 7

FB 8

OUT 9

V_DD 10U14

MAX9923H

GND

VCC_5.5_D

I_VCC_12.0_SNS

GND

100nFC61

VCC_5.5_nSENSED_DVCC_5.5_D

RSB1

RS+2

RS-3

N.C.4

GND5 SHDN 6

REF 7

FB 8

OUT 9

V_DD 10U15

MAX9923H

GND

VCC_5.5_D

I_VCC_5.0_D_SNS

GND

100nFC63

VCC_5.5_nSENSED_AVCC_5.5_A

RSB1

RS+2

RS-3

N.C.4

GND5 SHDN 6

REF 7

FB 8

OUT 9

V_DD 10U16

MAX9923H

GND

VCC_5.5_D

I_VCC_5.0_A_SNS

GND

100nFC65

Current Sense Amplifiers (0-850mA each)

1nF

C64

1nF

C62

1nF

C600R033R62

0R033R67

0R033R74

PIC6001 PIC6002

COC60 PIC6101

PIC6102COC61

PIC6201 PIC6202

COC62 PIC6301

PIC6302COC63

PIC6401 PIC6402

COC64 PIC6501

PIC6502COC65

PIP1301

PIP1302

COP13

PIP1401

PIP1402

COP14

PIP1501

PIP1502

COP15

PIP1601

PIP1602

COP16

PIP1701

PIP1702

COP17

PIP1801

PIP1802

COP18

PIP1901

PIP1902

COP19

PIR5801

PIR5802COR58

PIR5901

PIR5902COR59

PIR6001

PIR6002COR60

PIR6101

PIR6102COR61

PIR6201

PIR6202COR62

PIR6301

PIR6302COR63

PIR6401

PIR6402COR64

PIR6501

PIR6502COR65

PIR6601

PIR6602COR66

PIR6701

PIR6702COR67

PIR6801

PIR6802COR68

PIR6901

PIR6902COR69

PIR7001

PIR7002COR70

PIR7101

PIR7102COR71

PIR7201

PIR7202COR72

PIR7301

PIR7302COR73

PIR7401

PIR7402COR74

PIR7501

PIR7502COR75

PIR7601

PIR7602COR76

PIU1401

PIU1402

PIU1403

PIU1404

PIU1405 PIU1406

PIU1407

PIU1408

PIU1409

PIU14010

COU14

PIU1501

PIU1502

PIU1503

PIU1504

PIU1505 PIU1506

PIU1507

PIU1508

PIU1509

PIU15010

COU15

PIU1601

PIU1602

PIU1603

PIU1604

PIU1605 PIU1606

PIU1607

PIU1608

PIU1609

PIU16010

COU16

PIR5902

PIC6102

PIC6302

PIC6502

PIP1302PIP1402 PIP1502 PIP1602 PIP1702PIP1802 PIP1902

PIR6001 PIR6101

PIR6501 PIR6601

PIR7001 PIR7101

PIR7501 PIR7601

PIU1405

PIU1407

PIU1505

PIU1507

PIU1605

PIU1607

PIC6402

PIU1609NLI0VCC05000A0SNS

PIC6202

PIU1509NLI0VCC05000D0SNS

PIC6002

PIU1409NLI0VCC012000SNS

PIC6001PIU1408

PIC6201PIU1508

PIC6401PIU1608

PIU1404

PIU1504

PIU1604

PIR5901PIR6102

NLREF0ADC0SNS

PIP1501

PIR7202

PIP1901

PIR6802

PIR6801PIR7002

NLVCC02080SNS

PIP1401 PIP1301

PIR6902

PIP1701

PIR7302

PIP1601

PIR6402

PIP1801

PIR6302

PIR7401PIU1603

PIC6101

PIC6301

PIC6501

PIR6701

PIU1406

PIU14010

PIU1503

PIU1506

PIU15010

PIU1606

PIU16010PIR7402 PIU1601

PIU1602

PIR6702 PIU1501

PIU1502

PIR6301PIR6502

NLVCC05050SNS

PIR5802

PIR6201PIU1403

PIR6202 PIU1401

PIU1402

PIR5801PIR6002

NLVCC012000SNS

PIR7201PIR7502

NLVCC0ADC0SNS

PIR6901PIR7102

NLVCC0CH10SNS

PIR7301PIR7602

NLVCC0CH20SNS

PIR6401PIR6602

NLVCC0OPA0SNS

108

CO

PAADC106PAADC107PAADC108PAADC109PAADC1010


COADC1



COADC2

PAC101PAC102

COC1

PAC201PAC202

COC2PAC301PAC302

COC3

PAC401PAC402

COC4

PAC501 PAC502COC5

PAC601PAC602

COC6

PAC701PAC702

COC7

PAC801PAC802

COC8

PAC901PAC902

COC9

PAC1001PAC1002

COC10

PAC1101 PAC1102

COC11

PAC1201PAC1202

COC12

PAC1301PAC1302COC13 PAC1401 PAC1402

COC14

PAC1501 PAC1502COC15

PAC1602

PAC1601

COC16

PAC1701

PAC1702

COC17

PAC1801PAC1802

COC18

PAC1901

PAC1902

COC19

PAC2001

PAC2002 COC20

PAC2101

PAC2102COC21PAC2201

PAC2202

COC22

PAC2301PAC2302

COC23

PAC2401PAC2402 COC24

PAC2501PAC2502COC25

PAC2601PAC2602COC26

PAC2701PAC2702COC27 PAC2802

PAC2801COC28

PAC2901

PAC2902COC29

PAC3001PAC3002

COC30

PAC3101PAC3102

COC31 PAC3201

PAC3202COC32

PAC3301

PAC3302

COC33

PAC3401PAC3402

COC34

PAC3501PAC3502

COC35

PAC3601 PAC3602

COC36 PAC3701

PAC3702 COC37PAC3801

PAC3802COC38

PAC3901PAC3902COC39

PAC4001PAC4002COC40

PAC4101PAC4102 COC41PAC4201 PAC4202COC42

PAC4301 PAC4302COC43PAC4401PAC4402COC44 PAC4501 PAC4502

COC45


PAC4701 PAC4702 COC47

PAC4801PAC4802

COC48

PAC4901PAC4902COC49


PAC5101PAC5102COC51

PAC5201PAC5202

COC52

PAC5301 PAC5302

COC53

PAC5401PAC5402

COC54

PAC5501PAC5502COC55

PAC5601PAC5602

COC56


PAC5801PAC5802COC58

PAC5901PAC5902

COC59

PAC6001PAC6002

COC60 PAC6101 PAC6102

COC61

PAC6201PAC6202COC62

PAC6301PAC6302COC63

PAC6401PAC6402COC64

PAC6501PAC6502

COC65

PAD101 PAD102COD1

PAD201 PAD202COD2

PAD301 PAD302COD3

PAD402PAD401COD4

PAD502PAD501COD5

PAD602PAD601

COD6

PAD702PAD701COD7PAD802

PAD801

COD8PAD902

PAD901

COD9

PAJ101

PAJ102

PAJ103

PAJ104

PAJ105

PAJ106

PAJ107

PAJ108

PAJ109

PAJ1010

PAJ1011

PAJ1012

PAJ1013

PAJ1014 COJ1

PAJ20

PAJ201PAJ203PAJ205PAJ207PAJ209PAJ2011PAJ2013PAJ2015PAJ2017PAJ2019PAJ2021PAJ2023PAJ2025PAJ2027PAJ2029


COJ2

PAL102 PAL101

COL1

PAP101PAP102 COP1

PAP202

PAP201

COP2

PAP302

PAP301

COP3

PAP401PAP402 COP4

PAP502

PAP501

COP5

PAP602

PAP601 COP6

PAP702PAP701 PAP703 PAP704 COP7

PAP801 PAP802

PAP803 PAP804

PAP805 PAP806

PAP807 PAP808

PAP809 PAP8010

PAP8011 PAP8012

PAP8013 PAP8014

PAP8015 PAP8016

PAP8017 PAP8018

PAP8019 PAP8020

COP8PAP901

PAP902

PAP903

PAP904

PAP905

PAP906COP9

PAP1001PAP1002PAP1003 COP10PAP1101PAP1102PAP1103 COP11PAP1201PAP1202PAP1203PAP1204PAP1205COP12

PAP1302 PAP1301COP13


PAP1502 PAP1501

COP15

PAP1602 PAP1601COP16PAP1702 PAP1701COP17

PAP1802

PAP1801 COP18

PAP1902

PAP1901 COP19

PAQ103PAQ102

PAQ101

COQ1

PAQ202 PAQ203PAQ201COQ2



PAQ502 PAQ503

PAQ501COQ5

PAQ602 PAQ603

PAQ601COQ6

PAQ702 PAQ703

PAQ701COQ7

PAQ803

PAQ802

PAQ801

COQ8

PAQ902

PAQ903PAQ901COQ9

PAR101 PAR102COR1

PAR201 PAR202

COR2

PAR301 PAR302

COR3PAR401 PAR402

COR4PAR501PAR502

COR5

PAR601 PAR602

COR6

PAR701 PAR702

COR7

PAR801 PAR802COR8

PAR901 PAR902

COR9

PAR1001 PAR1002COR10

PAR1102

PAR1101

COR11

PAR1201PAR1202

COR12 PAR1301

PAR1302COR13

PAR1401

PAR1402 COR14

PAR1501

PAR1502

COR15

PAR1601PAR1602COR16

PAR1701PAR1702COR17

PAR1801PAR1802

COR18

PAR1901 PAR1902 COR19PAR2001PAR2002COR20

PAR2101PAR2102COR21 PAR2201

PAR2202 COR22


PAR2402 COR24


PAR2602 COR26

PAR2701PAR2702COR27

PAR2801PAR2802

COR28

PAR2901PAR2902COR29

PAR3001PAR3002

COR30

PAR3101

PAR3102

COR31PAR3201

PAR3202

COR32

PAR3301PAR3302COR33

PAR3401PAR3402

COR34PAR3501PAR3502

COR35PAR3601PAR3602COR36 PAR3701

PAR3702

COR37

PAR3801PAR3802

COR38PAR3901PAR3902

COR39PAR4001PAR4002

COR40PAR4101PAR4102

COR41

PAR4201 PAR4202

COR42

PAR4301 PAR4302

COR43

PAR4401 PAR4402

COR44

PAR4501

PAR4502COR45PAR4601

PAR4602COR46

PAR4701PAR4702COR47

PAR4801PAR4802

COR48PAR4901PAR4902

COR49

PAR5001PAR5002

COR50

PAR5101PAR5102

COR51


PAR5301PAR5302 COR53


PAR5501 PAR5502

COR55

PAR5601

PAR5602 COR56

PAR5701PAR5702COR57

PAR5801PAR5802

COR58

PAR5901 PAR5902

COR59

PAR6001PAR6002

COR60

PAR6101 PAR6102

COR61

PAR6202 PAR6201 COR62

PAR6301PAR6302COR63

PAR6401PAR6402COR64


PAR6601PAR6602COR66

PAR6702

PAR6701

COR67 PAR6801

PAR6802 COR68


PAR7001

PAR7002 COR70

PAR7101 PAR7102

COR71PAR7201PAR7202

COR72

PAR7301 PAR7302

COR73

PAR7402

PAR7401

COR74

PAR7501PAR7502COR75

PAR7601 PAR7602

COR76

PAS103

PAS102

PAS101

COS1

PAS20NCPAS201

PAS202

COS2

PAS30NC PAS301

PAS302

COS3

PASD100PASD10GND

PASD108PASD107

PASD106

PASD105

PASD104

PASD103PASD102

PASD101

PASD109

PASD10SWDPASD10SW PASD10WP

COSD1

PASD200PASD20GND

PASD208PASD207

PASD206

PASD205

PASD204

PASD203PASD202

PASD201

PASD209


COSD2

PAU105

PAU106

PAU107

PAU108

PAU104

PAU103

PAU102

PAU101

COU1

PAU205

PAU206

PAU207

PAU208

PAU204

PAU203

PAU202

PAU201

COU2

PAU301

PAU305

PAU306

PAU307

PAU308

PAU304

PAU303

PAU302

COU3

PAU401

PAU405

PAU406

PAU407

PAU408

PAU404

PAU403

PAU402

COU4

PAU501

PAU505

PAU506

PAU507

PAU508

PAU504

PAU503

PAU502

COU5

PAU601PAU602

PAU603PAU604PAU605

PAU6010PAU609

PAU608PAU607PAU606

PAU6011COU6

PAU701

PAU705

PAU706

PAU707

PAU708

PAU704

PAU703

PAU702

COU7

PAU801

PAU805

PAU806

PAU807

PAU808

PAU804

PAU803

PAU802

COU8

PAU901

PAU902PAU903

PAU904

PAU908

PAU907PAU906

PAU905

COU9

PAU10064PAU10063

PAU10062PAU10061PAU10060PAU10059PAU10058PAU10057


PAU10050PAU10049

PAU10048PAU10047PAU10046PAU10045PAU10044PAU10043PAU10042PAU10041PAU10040PAU10039PAU10038PAU10037PAU10036PAU10035PAU10034PAU10033PAU10032PAU10031PAU10030PAU10029PAU10028PAU10027PAU10026PAU10025PAU10024PAU10023PAU10022PAU10021PAU10020PAU10019PAU10018PAU10017

PAU10016PAU10015PAU10014PAU10013PAU10012PAU10011PAU10010PAU1009PAU1008PAU1007PAU1006PAU1005PAU1004PAU1003PAU1002PAU1001COU10

PAU1101 PAU1102 PAU110NC

PAU1103PAU1104

COU11

PAU1205PAU1206

PAU1207PAU1208

PAU1204PAU1203

PAU1202PAU1201

COU12

PAU1305PAU1306

PAU1307PAU1308

PAU1304PAU1303

PAU1302PAU1301

COU13PAU1406PAU1407

PAU1408PAU1409PAU14010

PAU1405PAU1404


COU14

PAU1506PAU1507PAU1508PAU1509PAU15010


COU15



COU16

PAW101

PAW102COW1

PAW202PAW201 PAW203COW2

PAW301

PAW302

COW3

PAW401

PAW402 COW4

PAW501

PAW502

COW5


PAY102PAY101COY1 PAY202

PAY201

COY2PAC4401

PAQ901

PAS202

PAU1007

PAADC107

PAU10023

PAADC207

PAU10036

PAADC106

PAADC206

PAU10015

PAADC101

PAADC201

PAC101

PAC701

PAC1401

PAC1501PAR5902

PAU306

PAADC108

PAADC208

PAU10021

PAU10026

PAU10034

PAU10060

PAW602

PAJ2020

PAU10059

PAP809

PAU10049

PAP8013

PAU10050

PAP805

PAU10055

PAP807

PAU10046

PAP8015

PAU10056

PAR4201

PAU10025

PAR4301

PAU10037

PAR4401

PAU10038

PAC4301

PAU1003

PAY201PAR3701

PAU1004

PAC4101

PAU1005

PAY102PAR3601

PAU1006

PAP1001

PAU10020

PAP1002

PAU10022

PAP1003

PAU10041

PAP1201

PAU10027

PAP1202

PAU10057

PAP1203

PAU10061

PAP1204

PAU10062

PAP1205

PAU10033

PAP1101

PAU10011

PAP1102

PAU10024

PAP1103

PAU1002

PAJ2015

PAQ602PAR2902

PAJ2026

PAQ902PASD105

PAU10053

PAR3302

PASD102

PAU10054

PAR3402

PASD107

PAU10039

PAR3502PASD108

PAU10040

PAR3102PASD109

PAU10051PAR3202

PASD101

PAU10052

PAC5701

PAJ209

PAU10029

PAC5401PAJ2011

PAU10030

PAJ2024

PAU10014

PAP902

PAR3802

PAU10045

PAP906

PAR4102

PAU10044

PAP905

PAR4002

PAU10042

PAP904

PAR3902

PAU10043

PAQ502 PAQ702

PAR2802 PAR3002

PAU10035

PAADC104PAADC105

PAADC204PAADC205

PAC102PAC202 PAC302PAC402

PAC602

PAC702PAC802 PAC902 PAC1002

PAC1202

PAC1302PAC1402

PAC1502

PAC1602

PAC1702

PAC1802

PAC1902

PAC2002

PAC2102PAC2202

PAC2302

PAC2602

PAC2702 PAC2802

PAC2902

PAC3002

PAC3102

PAC3202

PAC3302PAC3402

PAC3502

PAC3702PAC3802

PAC4002

PAC4102PAC4202

PAC4302

PAC4402PAC4502

PAC4602

PAC4702

PAC4802

PAC4902

PAC5002

PAC5102

PAC5202

PAC5302

PAC5502

PAC5602

PAC5802

PAC5902

PAC6102

PAC6302

PAC6502

PAD101

PAD201

PAD302

PAD401

PAD502

PAD602

PAD702

PAD802PAD902

PAJ109

PAJ206

PAP102

PAP202

PAP302

PAP402

PAP502

PAP602

PAP701 PAP702

PAP803 PAP804

PAP806

PAP808

PAP8010

PAP8012

PAP8014

PAP8016

PAP8018

PAP8020

PAP901

PAP1302

PAP1402

PAP1502

PAP1602

PAP1702

PAP1802

PAP1902

PAQ503

PAQ603

PAQ703

PAQ802

PAQ903

PAR501

PAR1001

PAR1201 PAR1601

PAR2701

PAR2801

PAR2901

PAR3001

PAR6001

PAR6101

PAR6501

PAR6601

PAR7001

PAR7101 PAR7501

PAR7601

PAS201

PAS301

PASD103

PASD106

PASD10GND

PASD10SWPASD10SWD

PASD10WP

PASD203

PASD206

PASD20GND

PASD20SWPASD20SWD

PASD20WP

PAU104

PAU204

PAU304

PAU404

PAU504

PAU601 PAU6011

PAU704

PAU804

PAU904

PAU10012PAU10018

PAU10028

PAU10031

PAU10047

PAU10063

PAU1102

PAU1201

PAU1301

PAU1405PAU1407

PAU1505PAU1507

PAU1605PAU1607

PAW101

PAW302

PAW402

PAW502

PAW601

PAC6402

PAJ2029

PAU1609

PAC6202

PAJ2027

PAU1509

PAC6002

PAJ2025

PAU1409

PAADC103PAC601 PAP301PAR302

PAADC203

PAC1201PAP601

PAR802

PAC502 PAR201

PAC1102 PAR701

PAC2402

PAQ101

PAR1101

PAU607

PAC2502PAU6010

PAC2601

PAR1301

PAC2701

PAU606

PAC3601PAR1701

PAC3902

PAU903

PAC4201

PAR3602

PAY101 PAC4501

PAR3702

PAY202

PAC5402

PAR5301PAU1202

PAC5601

PAR5201

PAU1206

PAU1207

PAC5702

PAR5601PAU1302

PAC5901

PAR5501

PAU1306

PAU1307

PAC6001 PAU1408

PAC6201PAU1508

PAC6401PAU1608PAD301

PAL102PAQ103

PAD402

PAP703 PAP704

PAQ801

PAD501PAR4202

PAD601PAR4302

PAD701PAR4402PAD801PAR5402

PAD901PAR5702

PAQ102

PAU609

PAQ202

PAR2102

PAQ302

PAR2302

PAQ402

PAR2502

PAQ501

PAR2101 PAR2201

PAW301

PAQ601

PAR2301 PAR2401

PAW401

PAQ701

PAR2501 PAR2601

PAW501

PAQ803

PAS102

PAR101

PAR202 PAU102

PAR102

PAR301PAU106

PAR402 PAR502

PAU103

PAR601

PAR702 PAU202

PAR602

PAR801PAU206

PAR902

PAR1002PAU203

PAR1202

PAU603

PAR1302

PAU604

PAR1602PAR1702

PAR1802

PAU605

PAR1901PAW203PAR2001 PAW201PAR2702

PAU902

PAW202

PAR4502 PASD208

PAR4902

PASD209

PAR5401

PAU1203

PAR5701

PAU1303

PAU602

PAW102

PAR5901PAR6102

PAU1009

PAJ2028

PAR5002

PAU1104

PAJ2013

PAR5102

PAU1101

PAJ1011

PAJ208

PAS302

PAJ2012

PAR4802

PASD201

PAJ2022

PAU10058

PAJ107

PAJ207

PAJ103

PAJ203

PAJ101

PAJ201

PAJ105

PAJ205

PAJ2016

PAR4602PASD207

PAJ2014

PAR4702

PASD202

PAJ2018 PASD205

PAADC102

PAADC202

PAC301

PAC901

PAC2001PAC2301

PAP1501

PAR7202

PAU506

PAC3701

PAC3901

PAC5101

PAC5201

PAC5301

PAJ104

PAJ202

PAP1901

PAQ303

PAQ403

PAR1902PAR2002

PAR2402

PAR2602

PAR4501PAR4601

PAR4701

PAR4801 PAR4901

PAR5001

PAR5101

PAR6802

PASD204

PAU901

PAU1103

PAJ2023PAR6801

PAR7002

PAADC109PAADC1010

PAADC209PAADC2010

PAC401

PAC1001

PAC1801

PAQ401

PAU502

PAU503

PAC4001PAC4601

PAC4701

PAC4801

PAC4901

PAC5001

PAC5501

PAC5801

PAP801 PAP802PAP1401 PAQ301

PAR3101PAR3201

PAR3301

PAR3401 PAR3501

PAR3801 PAR3901 PAR4001 PAR4101

PAR5202

PAR5302

PAR5502

PAR5602

PASD104

PAU1001PAU10013PAU10019

PAU10032

PAU10048

PAU10064

PAU1204

PAU1208

PAU1304

PAU1308

PAW603

PAC3201PAC3501

PAP1301

PAR1401

PAR6902

PAU806

PAC501

PAD102

PAP101 PAP201

PAR1402

PAC3301PAC3401

PAP1701

PAR1501

PAR7302

PAU706

PAC1101

PAD202

PAP401 PAP501

PAR1502

PAC201

PAC801

PAC2101PAC2201

PAP1601

PAR401

PAR901

PAR6402

PAU107

PAU108

PAU207

PAU208

PAU406

PAC2901

PAC3602

PAL101

PAP1801

PAR1801

PAR6302

PAR6702

PAR7402 PAU1501PAU1502

PAU1601PAU1602

PAQ203

PAR2202

PAR7401

PAU1603

PAC3801

PAC6101

PAC6301

PAC6501

PAR6701 PAU905

PAU908

PAU1406

PAU14010

PAU1503

PAU1506

PAU15010

PAU1606

PAU16010

PAJ2021

PAR6301 PAR6502

PAC1301

PAC1701

PAC3001

PAC3101

PAQ201

PAU302

PAU303

PAU402

PAU403

PAU702

PAU703

PAU802

PAU803

PAC1601PAC1901

PAC2401

PAC2501

PAC2801

PAR1102

PAR5802

PAR6201

PAU608

PAU1403

PAR6202PAS101

PAU1401PAU1402

PAJ2019

PAR5801 PAR6002

PAR7201 PAR7502

PAU1008

PAR6901PAR7102

PAU10016

PAR7301PAR7602

PAU10017

PAR6401 PAR6602

PAU10010

109

CO



COADC1



COADC2

PAC101PAC102

COC1

PAC201PAC202

COC2PAC301PAC302

COC3

PAC401PAC402

COC4

PAC501 PAC502COC5

PAC601PAC602

COC6

PAC701PAC702

COC7

PAC801PAC802

COC8

PAC901PAC902

COC9

PAC1001PAC1002

COC10

PAC1101 PAC1102

COC11

PAC1201PAC1202

COC12

PAC1301PAC1302COC13 PAC1401 PAC1402

COC14


PAC1602

PAC1601

COC16

PAC1701

PAC1702

COC17

PAC1801PAC1802

COC18

PAC1901

PAC1902

COC19

PAC2001

PAC2002 COC20

PAC2101

PAC2102COC21PAC2201

PAC2202

COC22

PAC2301PAC2302

COC23


PAC2501PAC2502COC25

PAC2601PAC2602COC26

PAC2701PAC2702COC27 PAC2802

PAC2801COC28

PAC2901

PAC2902COC29

PAC3001PAC3002

COC30

PAC3101PAC3102

COC31 PAC3201

PAC3202COC32

PAC3301

PAC3302

COC33

PAC3401PAC3402

COC34

PAC3501PAC3502

COC35

PAC3601 PAC3602

COC36 PAC3701

PAC3702 COC37PAC3801

PAC3802COC38

PAC3901PAC3902COC39

PAC4001PAC4002COC40

PAC4101PAC4102 COC41PAC4201 PAC4202COC42

PAC4301 PAC4302COC43PAC4401PAC4402COC44 PAC4501 PAC4502

COC45


PAC4701 PAC4702 COC47

PAC4801PAC4802

COC48

PAC4901PAC4902COC49


PAC5101PAC5102COC51

PAC5201PAC5202

COC52

PAC5301 PAC5302

COC53

PAC5401PAC5402

COC54

PAC5501PAC5502COC55

PAC5601PAC5602

COC56


PAC5801PAC5802COC58

PAC5901PAC5902

COC59

PAC6001PAC6002

COC60 PAC6101 PAC6102

COC61

PAC6201PAC6202COC62

PAC6301PAC6302COC63

PAC6401PAC6402COC64

PAC6501PAC6502

COC65

PAD101 PAD102COD1

PAD201 PAD202COD2

PAD301 PAD302COD3

PAD402PAD401COD4

PAD502PAD501COD5

PAD602PAD601

COD6

PAD702PAD701COD7PAD802

PAD801

COD8PAD902

PAD901

COD9

PAJ101

PAJ102

PAJ103

PAJ104

PAJ105

PAJ106

PAJ107

PAJ108

PAJ109

PAJ1010

PAJ1011

PAJ1012

PAJ1013

PAJ1014 COJ1

PAJ20



COJ2

PAL102 PAL101

COL1

PAP101PAP102 COP1

PAP202

PAP201

COP2

PAP302

PAP301

COP3

PAP401PAP402 COP4

PAP502

PAP501

COP5

PAP602

PAP601 COP6

PAP702PAP701 PAP703 PAP704 COP7

PAP801 PAP802

PAP803 PAP804

PAP805 PAP806

PAP807 PAP808

PAP809 PAP8010

PAP8011 PAP8012

PAP8013 PAP8014

PAP8015 PAP8016

PAP8017 PAP8018

PAP8019 PAP8020

COP8PAP901

PAP902

PAP903

PAP904

PAP905

PAP906COP9

PAP1001PAP1002PAP1003 COP10PAP1101PAP1102PAP1103 COP11PAP1201PAP1202PAP1203PAP1204PAP1205COP12



PAP1502 PAP1501

COP15

PAP1602 PAP1601COP16PAP1702 PAP1701COP17

PAP1802

PAP1801 COP18

PAP1902

PAP1901 COP19

PAQ103PAQ102

PAQ101

COQ1




PAQ502 PAQ503

PAQ501COQ5

PAQ602 PAQ603

PAQ601COQ6

PAQ702 PAQ703

PAQ701COQ7

PAQ803

PAQ802

PAQ801

COQ8

PAQ902

PAQ903PAQ901COQ9

PAR101 PAR102COR1

PAR201 PAR202

COR2

PAR301 PAR302

COR3PAR401 PAR402

COR4PAR501PAR502

COR5

PAR601 PAR602

COR6

PAR701 PAR702

COR7

PAR801 PAR802COR8

PAR901 PAR902

COR9


PAR1102

PAR1101

COR11

PAR1201PAR1202

COR12 PAR1301

PAR1302COR13

PAR1401

PAR1402 COR14

PAR1501

PAR1502

COR15

PAR1601PAR1602COR16

PAR1701PAR1702COR17

PAR1801PAR1802

COR18

PAR1901 PAR1902 COR19PAR2001PAR2002COR20


PAR2202 COR22


PAR2402 COR24


PAR2602 COR26

PAR2701PAR2702COR27

PAR2801PAR2802

COR28

PAR2901PAR2902COR29

PAR3001PAR3002

COR30

PAR3101

PAR3102

COR31PAR3201

PAR3202

COR32

PAR3301PAR3302COR33

PAR3401PAR3402

COR34PAR3501PAR3502

COR35PAR3601PAR3602COR36 PAR3701

PAR3702

COR37

PAR3801PAR3802

COR38PAR3901PAR3902

COR39PAR4001PAR4002

COR40PAR4101PAR4102

COR41

PAR4201 PAR4202

COR42

PAR4301 PAR4302

COR43

PAR4401 PAR4402

COR44

PAR4501

PAR4502COR45PAR4601

PAR4602COR46

PAR4701PAR4702COR47

PAR4801PAR4802

COR48PAR4901PAR4902

COR49

PAR5001PAR5002

COR50

PAR5101PAR5102

COR51




PAR5501 PAR5502

COR55

PAR5601

PAR5602 COR56

PAR5701PAR5702COR57

PAR5801PAR5802

COR58

PAR5901 PAR5902

COR59

PAR6001PAR6002

COR60

PAR6101 PAR6102

COR61

PAR6202 PAR6201 COR62

PAR6301PAR6302COR63

PAR6401PAR6402COR64


PAR6601PAR6602COR66

PAR6702

PAR6701

COR67 PAR6801

PAR6802 COR68


PAR7001

PAR7002 COR70

PAR7101 PAR7102

COR71PAR7201PAR7202

COR72

PAR7301 PAR7302

COR73

PAR7402

PAR7401

COR74

PAR7501PAR7502COR75

PAR7601 PAR7602

COR76

PAS103

PAS102

PAS101

COS1

PAS20NCPAS201

PAS202

COS2

PAS30NC PAS301

PAS302

COS3

PASD100PASD10GND

PASD108PASD107

PASD106

PASD105

PASD104

PASD103PASD102

PASD101

PASD109


COSD1

PASD200PASD20GND

PASD208PASD207

PASD206

PASD205

PASD204

PASD203PASD202

PASD201

PASD209


COSD2

PAU105

PAU106

PAU107

PAU108

PAU104

PAU103

PAU102

PAU101

COU1

PAU205

PAU206

PAU207

PAU208

PAU204

PAU203

PAU202

PAU201

COU2

PAU301

PAU305

PAU306

PAU307

PAU308

PAU304

PAU303

PAU302

COU3

PAU401

PAU405

PAU406

PAU407

PAU408

PAU404

PAU403

PAU402

COU4

PAU501

PAU505

PAU506

PAU507

PAU508

PAU504

PAU503

PAU502

COU5

PAU601PAU602

PAU603PAU604PAU605

PAU6010PAU609

PAU608PAU607PAU606

PAU6011COU6

PAU701

PAU705

PAU706

PAU707

PAU708

PAU704

PAU703

PAU702

COU7

PAU801

PAU805

PAU806

PAU807

PAU808

PAU804

PAU803

PAU802

COU8

PAU901

PAU902PAU903

PAU904

PAU908

PAU907PAU906

PAU905

COU9

PAU10064PAU10063



PAU10050PAU10049

PAU10048PAU10047PAU10046PAU10045PAU10044PAU10043PAU10042PAU10041PAU10040PAU10039PAU10038PAU10037PAU10036PAU10035PAU10034PAU10033PAU10032PAU10031PAU10030PAU10029PAU10028PAU10027PAU10026PAU10025PAU10024PAU10023PAU10022PAU10021PAU10020PAU10019PAU10018PAU10017

PAU10016PAU10015PAU10014PAU10013PAU10012PAU10011PAU10010PAU1009PAU1008PAU1007PAU1006PAU1005PAU1004PAU1003PAU1002PAU1001COU10

PAU1101 PAU1102 PAU110NC

PAU1103PAU1104

COU11

PAU1205PAU1206

PAU1207PAU1208

PAU1204PAU1203

PAU1202PAU1201

COU12

PAU1305PAU1306

PAU1307PAU1308

PAU1304PAU1303

PAU1302PAU1301

COU13PAU1406PAU1407


PAU1405PAU1404


COU14



COU15



COU16

PAW101

PAW102COW1


PAW301

PAW302

COW3

PAW401

PAW402 COW4

PAW501

PAW502

COW5


PAY102PAY101COY1 PAY202

PAY201

COY2PAC4401

PAQ901

PAS202

PAU1007

PAADC107

PAU10023

PAADC207

PAU10036

PAADC106

PAADC206

PAU10015

PAADC101

PAADC201

PAC101

PAC701

PAC1401

PAC1501PAR5902

PAU306

PAADC108

PAADC208

PAU10021

PAU10026

PAU10034

PAU10060

PAW602

PAJ2020

PAU10059

PAP809

PAU10049

PAP8013

PAU10050

PAP805

PAU10055

PAP807

PAU10046

PAP8015

PAU10056

PAR4201

PAU10025

PAR4301

PAU10037

PAR4401

PAU10038

PAC4301

PAU1003

PAY201PAR3701

PAU1004

PAC4101

PAU1005

PAY102PAR3601

PAU1006

PAP1001

PAU10020

PAP1002

PAU10022

PAP1003

PAU10041

PAP1201

PAU10027

PAP1202

PAU10057

PAP1203

PAU10061

PAP1204

PAU10062

PAP1205

PAU10033

PAP1101

PAU10011

PAP1102

PAU10024

PAP1103

PAU1002

PAJ2015

PAQ602PAR2902

PAJ2026

PAQ902PASD105

PAU10053

PAR3302

PASD102

PAU10054

PAR3402

PASD107

PAU10039

PAR3502PASD108

PAU10040

PAR3102PASD109

PAU10051PAR3202

PASD101

PAU10052

PAC5701

PAJ209

PAU10029

PAC5401PAJ2011

PAU10030

PAJ2024

PAU10014

PAP902

PAR3802

PAU10045

PAP906

PAR4102

PAU10044

PAP905

PAR4002

PAU10042

PAP904

PAR3902

PAU10043

PAQ502 PAQ702

PAR2802 PAR3002

PAU10035

PAADC104PAADC105

PAADC204PAADC205

PAC102PAC202 PAC302PAC402

PAC602

PAC702PAC802 PAC902 PAC1002

PAC1202

PAC1302PAC1402

PAC1502

PAC1602

PAC1702

PAC1802

PAC1902

PAC2002

PAC2102PAC2202

PAC2302

PAC2602

PAC2702 PAC2802

PAC2902

PAC3002

PAC3102

PAC3202

PAC3302PAC3402

PAC3502

PAC3702PAC3802

PAC4002

PAC4102PAC4202

PAC4302

PAC4402PAC4502

PAC4602

PAC4702

PAC4802

PAC4902

PAC5002

PAC5102

PAC5202

PAC5302

PAC5502

PAC5602

PAC5802

PAC5902

PAC6102

PAC6302

PAC6502

PAD101

PAD201

PAD302

PAD401

PAD502

PAD602

PAD702

PAD802PAD902

PAJ109

PAJ206

PAP102

PAP202

PAP302

PAP402

PAP502

PAP602

PAP701 PAP702

PAP803 PAP804

PAP806

PAP808

PAP8010

PAP8012

PAP8014

PAP8016

PAP8018

PAP8020

PAP901

PAP1302

PAP1402

PAP1502

PAP1602

PAP1702

PAP1802

PAP1902

PAQ503

PAQ603

PAQ703

PAQ802

PAQ903

PAR501

PAR1001

PAR1201 PAR1601

PAR2701

PAR2801

PAR2901

PAR3001

PAR6001

PAR6101

PAR6501

PAR6601

PAR7001

PAR7101 PAR7501

PAR7601

PAS201

PAS301

PASD103

PASD106

PASD10GND

PASD10SWPASD10SWD

PASD10WP

PASD203

PASD206

PASD20GND

PASD20SWPASD20SWD

PASD20WP

PAU104

PAU204

PAU304

PAU404

PAU504

PAU601 PAU6011

PAU704

PAU804

PAU904

PAU10012PAU10018

PAU10028

PAU10031

PAU10047

PAU10063

PAU1102

PAU1201

PAU1301

PAU1405PAU1407

PAU1505PAU1507

PAU1605PAU1607

PAW101

PAW302

PAW402

PAW502

PAW601

PAC6402

PAJ2029

PAU1609

PAC6202

PAJ2027

PAU1509

PAC6002

PAJ2025

PAU1409

PAADC103PAC601 PAP301PAR302

PAADC203

PAC1201PAP601

PAR802

PAC502 PAR201

PAC1102 PAR701

PAC2402

PAQ101

PAR1101

PAU607

PAC2502PAU6010

PAC2601

PAR1301

PAC2701

PAU606

PAC3601PAR1701

PAC3902

PAU903

PAC4201

PAR3602

PAY101 PAC4501

PAR3702

PAY202

PAC5402

PAR5301PAU1202

PAC5601

PAR5201

PAU1206

PAU1207

PAC5702

PAR5601PAU1302

PAC5901

PAR5501

PAU1306

PAU1307

PAC6001 PAU1408

PAC6201PAU1508

PAC6401PAU1608PAD301

PAL102PAQ103

PAD402

PAP703 PAP704

PAQ801

PAD501PAR4202

PAD601PAR4302

PAD701PAR4402PAD801PAR5402

PAD901PAR5702

PAQ102

PAU609

PAQ202

PAR2102

PAQ302

PAR2302

PAQ402

PAR2502

PAQ501

PAR2101 PAR2201

PAW301

PAQ601

PAR2301 PAR2401

PAW401

PAQ701

PAR2501 PAR2601

PAW501

PAQ803

PAS102

PAR101

PAR202 PAU102

PAR102

PAR301PAU106

PAR402 PAR502

PAU103

PAR601

PAR702 PAU202

PAR602

PAR801PAU206

PAR902

PAR1002PAU203

PAR1202

PAU603

PAR1302

PAU604

PAR1602PAR1702

PAR1802

PAU605

PAR1901PAW203PAR2001 PAW201PAR2702

PAU902

PAW202

PAR4502 PASD208

PAR4902

PASD209

PAR5401

PAU1203

PAR5701

PAU1303

PAU602

PAW102

PAR5901PAR6102

PAU1009

PAJ2028

PAR5002

PAU1104

PAJ2013

PAR5102

PAU1101

PAJ1011

PAJ208

PAS302

PAJ2012

PAR4802

PASD201

PAJ2022

PAU10058

PAJ107

PAJ207

PAJ103

PAJ203

PAJ101

PAJ201

PAJ105

PAJ205

PAJ2016

PAR4602PASD207

PAJ2014

PAR4702

PASD202

PAJ2018 PASD205

PAADC102

PAADC202

PAC301

PAC901

PAC2001PAC2301

PAP1501

PAR7202

PAU506

PAC3701

PAC3901

PAC5101

PAC5201

PAC5301

PAJ104

PAJ202

PAP1901

PAQ303

PAQ403

PAR1902PAR2002

PAR2402

PAR2602

PAR4501PAR4601

PAR4701

PAR4801 PAR4901

PAR5001

PAR5101

PAR6802

PASD204

PAU901

PAU1103

PAJ2023PAR6801

PAR7002

PAADC109PAADC1010

PAADC209PAADC2010

PAC401

PAC1001

PAC1801

PAQ401

PAU502

PAU503

PAC4001PAC4601

PAC4701

PAC4801

PAC4901

PAC5001

PAC5501

PAC5801

PAP801 PAP802PAP1401 PAQ301

PAR3101PAR3201

PAR3301

PAR3401 PAR3501

PAR3801 PAR3901 PAR4001 PAR4101

PAR5202

PAR5302

PAR5502

PAR5602

PASD104

PAU1001PAU10013PAU10019

PAU10032

PAU10048

PAU10064

PAU1204

PAU1208

PAU1304

PAU1308

PAW603

PAC3201PAC3501

PAP1301

PAR1401

PAR6902

PAU806

PAC501

PAD102

PAP101 PAP201

PAR1402

PAC3301PAC3401

PAP1701

PAR1501

PAR7302

PAU706

PAC1101

PAD202

PAP401 PAP501

PAR1502

PAC201

PAC801

PAC2101PAC2201

PAP1601

PAR401

PAR901

PAR6402

PAU107

PAU108

PAU207

PAU208

PAU406

PAC2901

PAC3602

PAL101

PAP1801

PAR1801

PAR6302

PAR6702

PAR7402 PAU1501PAU1502

PAU1601PAU1602

PAQ203

PAR2202

PAR7401

PAU1603

PAC3801

PAC6101

PAC6301

PAC6501

PAR6701 PAU905

PAU908

PAU1406

PAU14010

PAU1503

PAU1506

PAU15010

PAU1606

PAU16010

PAJ2021

PAR6301 PAR6502

PAC1301

PAC1701

PAC3001

PAC3101

PAQ201

PAU302

PAU303

PAU402

PAU403

PAU702

PAU703

PAU802

PAU803

PAC1601PAC1901

PAC2401

PAC2501

PAC2801

PAR1102

PAR5802

PAR6201

PAU608

PAU1403

PAR6202PAS101

PAU1401PAU1402

PAJ2019

PAR5801 PAR6002

PAR7201 PAR7502

PAU1008

PAR6901PAR7102

PAU10016

PAR7301PAR7602

PAU10017

PAR6401 PAR6602

PAU10010

110

FPerformance Measurements

F.1 Input Filter Linearity

The following plots show the phase and amplitude compensated error plotsas presented in Section 7.1.1.

Figure F.1: Phase and amplitude compensated error, no preamplier, 30 kHz inputfrequency.

111

APPENDIX F. PERFORMANCE MEASUREMENTS

Figure F.2: Phase and amplitude compensated error, no preamplier, 50 kHz inputfrequency.

Figure F.3: Phase and amplitude compensated error, no preamplier, 100 kHz in-put frequency.

112

F.1. INPUT FILTER LINEARITY

Figure F.4: Phase and amplitude compensated error, with external preamplier,30 kHz input frequency.


113



F.2 Temperature Dependency of AE Parameters

The following gures show the data quality test results for both AEBoardchannels as well as for the USB AE Node (c.f. Section 7.2.2).

114

F.2. TEMPERATURE DEPENDENCY OF AE PARAMETERS!"#$%

!&'()*

++,++ +-,++ *+,++ *-,++

+

.++

/++

0++

1++

*+++

*.++

*/++

*0++

*1++

.+++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+4($'%5

6(78("&%9"():;<=

6#7():5=

4($'%5):>&78?(>=

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+

-+

*++

*-+

.++

.-+

3++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+@#>(%#7(

6(78("&%9"():;<=

6#7():5=@#>(%#7():>&78?(>=

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+

-+++

*++++

*-+++

.++++

.-+++

3++++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+

A78?#%9B(

6(78("&%9"():;<=

6#7():5=

A78?#%9B():'($("#C=

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+

*+

.+

3+

/+

-+

0+

D+

1+

E+

*++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+

<F9$%

6(78("&%9"():;<=

6#7():5=

!9?>(>

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+GH+

*GH1

.GH1

3GH1

/GH1

-GH1

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+G$("'I

6(78("&%9"():;<=

6#7():5=

G$("'I):'($("#C=

6(78("&%9"():;<=

Figure F.7: Data quality test parameter plots for AEBoard channel 1.

115


!"#$%

!&'()*

++,++ +-,++ *+,++ *-,++

+

.++

/++

0++

1++

*+++

*.++

*/++

*0++

*1++

.+++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+4($'%5

6(78("&%9"():;<=

6#7():5=

4($'%5):>&78?(>=

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+

-+

*++

*-+

.++

.-+

3++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+@#>(%#7(

6(78("&%9"():;<=

6#7():5=

@#>(%#7():>&78?(>=

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+

-+++

*++++

*-+++

.++++

.-+++

3++++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+

A78?#%9B(

6(78("&%9"():;<=

6#7():5=

A78?#%9B():'($("#C=

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+

*+

.+

3+

/+

-+

0+

D+

1+

E+

*++

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+

<F9$%

6(78("&%9"():;<=

6#7():5=

!9?>(>

6(78("&%9"():;<=

++,++ +-,++ *+,++ *-,++

+GH+

*GH1

.GH1

3GH1

/GH1

-GH1

23+

2.+

2*+

+

*+

.+

3+

/+

-+

0+G$("'I

6(78("&%9"():;<=

6#7():5=

G$("'I):'($("#C=

6(78("&%9"():;<=

Figure F.8: Data quality test parameter plots for AEBoard channel 2.

116

F.2. TEMPERATURE DEPENDENCY OF AE PARAMETERS!"#$%

!&'()*

++,++ +-,++ *+,++ *-,++

+

*+++

.+++

/+++

0+++

-+++

1+++

2/+

2.+

2*+

+

*+

.+

/+

0+

-+

1+3($'%4

5(67("&%8"()9:;<

5#6()94<

3($'%4)98=<

5(67("&%8"()9:;<

++,++ +-,++ *+,++ *-,++

+

*++

.++

/++

0++

-++

1++

2/+

2.+

2*+

+

*+

.+

/+

0+

-+

1+>#=(%#6(

5(67("&%8"()9:;<

5#6()94<>#=(%#6()98=<

5(67("&%8"()9:;<

++,++ +-,++ *+,++ *-,++

+

.+

0+

1+

?+

*++

*.+

*0+

2/+

2.+

2*+

+

*+

.+

/+

0+

-+

1+

@67A#%8B(

5(67("&%8"()9:;<

5#6()94<

@67A#%8B()9BC<

5(67("&%8"()9:;<

++,++ +-,++ *+,++ *-,++

+

-+

*++

*-+

.++

.-+

2/+

2.+

2*+

+

*+

.+

/+

0+

-+

1+

;D8$%

5(67("&%8"()9:;<

5#6()94<

!8A=(=

5(67("&%8"()9:;<

++,++ +-,++ *+,++ *-,++

+

*+++

.+++

/+++

0+++

-+++

1+++

E+++

2/+

2.+

2*+

+

*+

.+

/+

0+

-+

1+F$("'G

5(67("&%8"()9:;<

5#6()94<

F$("'G)9'($("#H<

5(67("&%8"()9:;<

Figure F.9: Data quality test parameter plots for the USB AE Node.

117


118

GCD Contents

The following directory structure is contained on the enclosed CD:

AEBoard

Binaries: PermAE and PermAEInit binaries (c.f. Section A.2).

PCB Design: Design les including a detailed bill of material, aswell as schematic and PCB prints of AEBoard rev. 1.1 and 1.2.

STM32 Workspace: Atollic IDE workspace (c.f. Section B.1.1).

Tools: convertStoredEvent.php (c.f. Section A.4.2) and objcopy.exe(c.f. Section B.1.4)

Datasheets and Manuals: Datasheets of PCB parts and Physical Acous-tics Equipment as well as the FreeRTOS documentation and manualsconcerning the STM32 processor.

Image Originals: The source les of some gures contained in thisthesis. All gures are in the OpenOce drawing format.

LTSpice: The LTSpice simulation les for the input circuitry simula-tion (c.f. Section 5.3).

MATLAB Scripts: MATLAB Scripts for data analysis of AEWin ex-ported data. Refer to the contained README.txt for details.

Presentations: Slides of the initial and of the nal presentation.

Thesis Sources: The LaTeX sources of this thesis.

119

APPENDIX G. CD CONTENTS

120

HTask Denition

The task denition for this master thesis is included in the following twopages.

121

Seite 1/2

Computer Engineering and Networks Lab

ETZ G 75 CH-8092 Zurich

Dr. Jan Beutel Gloriastr. 35 +41-44-6327032 +41-44-6321035 [email protected] www.tik.ee.ethz.ch/~beutel

MASTERARBEIT

Für Josua Hunziker

Betreuer: Jan Beutel

Stellvertreter: Stephan Gruber

Ausgabe: 22. September 2010 Abgabe: 06. Mai 2011

Acoustic Emission Sensing for Wireless Sensor Networks

Das PermaSense Projekt befasst sich mit der Beobachtung physikalischer Parameter des Permafrostes in den Schweizer Alpen. Zu diesem Zweck entwickelt und betreibt das Projekt verschiedene Messsysteme. Zu den existierenden Messungen von Temperaturen, Spaltausdehnung und Leitfähigkeitsprofilen sollen nun neue Sensoren entwickelt werden, die im Gegensatz zu den bisher eingesetzten Systemen schneller ablaufende Prozesse erfassen können. Die akustischen Emissionen während der Riss- und Eisbildung im Fels eignen sich hierfür in mehrfacher Hinsicht. Aus geowissenschaftlicher Sicht handelt es sich um relevante, jedoch wenig verstandene Prozesse im Permafrost, aus technischer Sicht stellt sich eine komplexe jedoch überschaubare Aufgabe in der Signalverarbeitung. In dieser Arbeit soll auf Basis der Feldversuche mit Akustik Sensoren am Jungfraujoch im April 2009 ein dediziertes System entwickelt werden. Das Auftreten relevanter Signaturen in Akustiksignalen ist sporadisch und bedingt durch die stark gedämpfte Ausbreitung im Fels als lokal zu betrachten. Dadurch ergibt sich ein Ansatz mit einer lokalen Vorverabeitung der Signale nahe am Sensor, einer Reduzierung der Datenmenge auf wesentliche Merkmale der zu detektierenden Signaturen und eine anschließende Übertragung mit einem drahtlosen Netzwerk. Der Sensor soll sich, falls möglich, in das stromsparende PermaDozer Netzwerk einbinden lassen und kann ggf. auch eine weitere Netzwerkressource zur Datenübertragung (WLAN) benutzen. Der Sensor soll für den Betrieb im Hochgebirge ausgelegt sein, keine manuellen Interventionen benötigen, und kann bei Bedarf von einer Solarzelle gespiesen werden. Aufgabenstellung

• Erstellen Sie einen Projektplan und legen Sie Meilensteine sowohl zeitlich wie auch thematisch fest. Erarbeiten Sie in Absprache mit dem Betreuer ein Pflichtenheft.

• Machen Sie sich mit den relevanten Arbeiten im Bereich Sensornetze, piezoakustischen Sensoren, Digitaler Signalverarbeitung vertraut. Führen Sie eine Literaturrecherche durch. Suchen Sie gezielt nach relevanten Publikationen. Prüfen Sie welche Ideen/Konzepte Sie aus diesen Lösungen verwenden können.

• Erstellen Sie eine Übersicht der Anforderungen an einen Akustik Sensor für das PermaSense Projekt. Beziehen Sie hierzu auch die bestehenden Erfahrungen mit ein. In Bezug auf Know-how soll besonders auf das Wissen der Gruppe von D. Amitrano (U.

122

Masterarbeit: Acoustic Emission Sensing in Wireless Sensor Networks

Seite 2/2

Grenoble) zurückgegriffen werden. Gegebenenfalls könnte auch ein Besuch im Labor in Grenoble zielführend sein.

• Erstellen Sie eine Funktionsspezifikation auf Basis einer eingehenden Analyse der vorhandenen Rohdaten der Feldversuche.

• Entwerfen Sie auf Basis der Funktionsspezifikation einen lauffähigen Prototyp. Implementieren Sie diesen.

• Validieren Sie den Prototyp mittels eines Referenzsystems. • Achten Sie darauf, dass die angestrebte Lösung auch in der Praxis anwendbar ist. Dies gilt

insbesondere für die Konzepte und Implementierung der Software für die Steuerung. • Dokumentieren Sie Ihre Arbeit sorgfältig mit einem Vortrag, einer kleinen Demonstration,

sowie mit einem Schlussbericht. Durchführung der Masterarbeit Allgemeines

• Der Verlauf des Projektes soll laufend anhand des Projektplanes und der Meilensteine evaluiert werden. Unvorhergesehene Probleme beim eingeschlagenen Lösungsweg können Änderungen am Projektplan erforderlich machen. Diese sollen dokumentiert werden.

• Sie verfügen über PCs mit Linux/Windows für Softwareentwicklung und Test. Für die Einhaltung der geltenden Sicherheitsrichtlinien der ETH Zürich sind Sie selbst verantwortlich. Falls damit Probleme auftauchen wenden Sie sich an Ihren Betreuer.

• Stellen Sie Ihr Projekt zu Beginn der Semesterarbeit in einem Kurzvortrag vor und präsentieren Sie die erarbeiteten Resultate am Schluss im Rahmen des Institutskolloquiums Ende Semester.

• Besprechen Sie Ihr Vorgehen regelmäßig mit Ihren Betreuern. Verfassen Sie dazu auch einen kurzen wöchentlichen Statusbericht (email).

Abgabe

• Geben Sie zwei Exemplare des Berichtes spätestens am 06. Mai 2011 dem betreuenden Assistenten oder seinem Stellvertreter ab. Diese Aufgabenstellung soll im Bericht eingefügt werden genauso wie das unterschriebene Unterschriftenblatt „Plagiat“ des Rektorats. Die Entsprechenden Richtlinien des Rektorats sind einzuhalten.

• Räumen Sie Ihre Rechnerkonten soweit auf, dass nur noch die relevanten Quellcode- und Binärdateien, Konfigurationsdateien, benötigte Verzeichnisstrukturen usw. bestehen bleiben. Der Programmcode sowie die Dateistruktur sollen ausreichend dokumentiert sein. Eine spätere Anschlussarbeit soll auf dem hinterlassenen Stand aufbauen können.

123

Bibliography

[1] The PermaSense Project, April 2011. [Online]. Available:http://www.permasense.ch 1, 2

[2] J. Beutel, S. Gruber, A. Hasler, R. Lim, A. Meier, C. Plessl, I. Talzi,L. Thiele, C. Tschudin, M. Woehrle et al., PermaDAQ: A scienticinstrument for precision sensing and data recovery in environmentalextremes, in Proceedings of the 2009 International Conference onInformation Processing in Sensor Networks. IEEE Computer Society,2009, pp. 265276. 2, 8, 57

[3] J. Beutel, S. Gruber, S. Gubler, A. Hasler, M. Keller, R. Lim, I. Talzi,L. Thiele, C. Tschudin, and M. Yücel, The PermaSense RemoteMonitoring Infrastructure, in International Snow Science WorkshopDavos, 2009. 2, 8

[4] N. Burri, P. Von Rickenbach, and R. Wattenhofer, Dozer: ultra-lowpower data gathering in sensor networks, in Proceedings of the 6thinternational conference on Information processing in sensor networks.ACM, 2007, pp. 450459. 2, 6, 7, 57

[5] J. Hunziker, J. Beutel, and M. Keller, Online GPS für PermafrostMonitoring, Semester Thesis, Computer Engineering and NetworksLab, ETH Zürich, 2009. 2

[6] P. Guillet, B. Buchli, and J. Beutel, Online GPS for PermaSenseWSN, Semester Thesis, Computer Engineering and Networks Lab,ETH Zürich, 2010. 2, 26

[7] F. Sutton, B. Buchli, and J. Beutel, GPS-equipped wireless sensornode for permasense, Semester Thesis, Computer Engineering andNetworks Lab, ETH Zürich, 2010. 2, 26

[8] M. Allen, L. Girod, R. Newton, S. Madden, D. Blumstein, andD. Estrin, Voxnet: An interactive, rapidly-deployable acoustic moni-toring platform, in Information Processing in Sensor Networks, 2008.IPSN'08. International Conference on. IEEE, 2008, pp. 371382. 3

124

http://www.permasense.ch

BIBLIOGRAPHY

[9] G. Simon, M. Maróti, Á. Lédeczi, G. Balogh, B. Kusy, A. Nádas,G. Pap, J. Sallai, and K. Frampton, Sensor network-based counter-sniper system, in Proceedings of the 2nd international conference onEmbedded networked sensor systems. ACM, 2004, pp. 112. 3

[10] G. Werner-Allen, S. Dawson-Haggerty, and M. Welsh, Lance: opti-mizing high-resolution signal collection in wireless sensor networks, inProceedings of the 6th ACM conference on Embedded network sensorsystems. ACM, 2008, pp. 169182. 3

[11] C. Grosse, M. Krüger, and S. D. Glaser, Wireless acoustic emissionsensor networks for structural health monitoring in civil engineering,in European NDT Conference (ECNDT), 2006. 3

[12] C. Grosse, G. McLaskey, S. Bachmaier, S. D. Glaser, and M. Krüger, Ahybrid wireless sensor network for acoustic emission testing in shm, inSensors and Smart Structures Technologies for Civil, Mechanical andAerospace Systems 2008, M. Tomizuka, Ed., vol. 6932, 2008. 3

[13] S. A. Bachmaier, Event-based acoustic emission technique for struc-tural health monitoring using wireless sensor networks, in 7th bInternational PhD Symposium in Civil Engineering, 2008. 3

[14] Ákos Lédeczi, T. Hay, P. Völgyesi, D. R. Hay, A. Nádas, and S. Ja-yaraman, Wireless acoustic emission sensor network for structuralmonitoring, IEEE Sensors Journal, vol. 9, no. 11, pp. 13701377,November 2009. 3

[15] N. Gershenfeld, R. Krikorian, and D. Cohen, The internet of things.Scientic American, vol. 291, no. 4, pp. 7681, 2004. 5

[16] M. Weiser, Ubiquitous computing, COMPUTER,, pp. 7172, 1993. 5

[17] I. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, Wirelesssensor networks: a survey, Computer networks, vol. 38, no. 4, pp.393422, 2002. 6

[18] J. Beutel, O. Kasten, F. Mattern, K. Römer, F. Siegemund, andL. Thiele, Prototyping wireless sensor network applications withBTnodes, Wireless Sensor Networks, pp. 323338, 2004. 6

[19] I. Akyildiz and M. Vuran, Wireless Sensor Networks. LibreDigital,2010. 6

[20] G. Park, T. Rosing, M. Todd, C. Farrar, and W. Hodgkiss, Energyharvesting for structural health monitoring sensor networks, Journalof Infrastructure Systems, vol. 14, p. 64, 2008. 6

125

BIBLIOGRAPHY

[21] S. Lindsey and C. Raghavendra, PEGASIS: Power-ecient gatheringin sensor information systems, in Aerospace Conference Proceedings,2002. IEEE, vol. 3. IEEE, 2002, pp. 31125. 6

[22] A. Manjeshwar and D. Agrawal, APTEEN: A hybrid protocol forecient routing and comprehensive information retrieval in wirelesssensor networks, in ipdps. Published by the IEEE Computer Society,2002, p. 0195b. 6

[23] H. Dubois-Ferrière, L. Fabre, R. Meier, and P. Metrailler, TinyNode:a comprehensive platform for wireless sensor network applications,in Proceedings of the 5th international conference on Information pro-cessing in sensor networks. ACM, 2006, pp. 358365. 7

[24] TinyNode. [Online]. Available: http://www.tinynode.com 7

[25] P. Levis, S. Madden, J. Polastre, R. Szewczyk, K. Whitehouse, A. Woo,D. Gay, J. Hill, M. Welsh, E. Brewer et al., Tinyos: An operatingsystem for sensor networks, Ambient Intelligence, vol. 35, 2005. 7, 57

[26] K. Aberer, M. Hauswirth, and A. Salehi, A middleware for fastand exible sensor network deployment, in Proceedings of the 32ndinternational conference on Very large data bases. VLDB Endowment,2006, pp. 11991202. 7

[27] M. Keller, M. Yücel, and J. Beutel, High-Resolution Imaging forEnvironmental Monitoring Applications, Relation, vol. 10, no. 1.149,p. 3558, 2009. 7

[28] J. Beutel, S. Gruber, A. Hasler, R. Lim, A. Meier, C. Plessl, I. Talzi,L. Thiele, C. Tschudin, M. Woehrle et al., Operating a sensornetwork at 3500 m above sea level, in Proc. of the 8th Int. Conf. onInformation Processing in Sensor Networks (IPSN), 2009. 8

[29] NDT Resource Center, Introduction to Acoustic EmissionTesting, April 2011. [Online]. Available: http://www.ndt-ed.org/EducationResources/CommunityCollege/Other%20Methods/AE/AE_Index.htm 8, 9

[30] C. Grosse and M. Ohtsu, Eds., Acoustic Emission Testing: Basicsfor Research, Applications in Civil Engineering. Springer BerlinHeidelberg, 2008. 8, 10, 11, 16

[31] D. Grgic and D. Amitrano, Creep of a porous rock and associatedacoustic emission under dierent hydrous conditions, J. geophys. Res,vol. 114, 2009. 12

126

http://www.tinynode.com

http://www.ndt-ed.org/EducationResources/CommunityCollege/Other%20Methods/AE/AE_Index.htm



BIBLIOGRAPHY

[32] T. Spies, J. Hesser, J. Eisenbl"atter, and G. Eilers, Measurements of acoustic emission during back-lling of large excavations, in Proceedings of the Sixth InternationalConference on Rock Bursts and Seismicity in Mines (Perth), 2005, pp.379383. 12

[33] G. Manthei, Characterization of acoustic emission sources in a rocksalt specimen under triaxial compression, Bulletin of the SeismologicalSociety of America, vol. 95, no. 5, p. 1674, 2005. 12

[34] B. Hallet, J. Walder, and C. Stubbs, Weathering by segregation icegrowth in microcracks at sustained subzero temperatures: vericationfrom an experimental study using acoustic emissions, Permafrost andPeriglacial Processes, vol. 2, no. 4, pp. 283300, 1991. 12

[35] J. Murton, R. Peterson, and J. Ozouf, Bedrock fracture by icesegregation in cold regions, Science, vol. 314, no. 5802, p. 1127, 2006.12

[36] S. Gruber and W. Haeberli, Permafrost in steep bedrock slopes and itstemperature-related destabilization following climate change, Journalof Geophysical Research, vol. 112, no. F2, p. F02S18, 2007. 12

[37] A. Hasler, S. Gruber, and W. Haeberli, Temperature variability andthermal oset in steep alpine rock and ice faces, The CryosphereDiscussions, vol. 5, pp. 721753, 2011. 12

[38] A. Halser, Thermal conditions and kinematics of steep bedrockpermafrost, Ph.D. dissertation, University of Zurich, 2011, draft.[Online]. Available: http://www.geo.uzh.ch/~ahasy/diss_hasler.pdf 12

[39] Physical Acoustics Corporation, AEWin Software, April 2011.[Online]. Available: http://www.acousticemission.com/index.aspx?go=products&focus=/Software/aewin.htm 13, 68

[40] S. Gruber, D. Amitrano, A. Hasler, J. Hunziker, and J. Beutel,Taking the lab to the mountains: rst experiences with acousticemission sensing at Jungfraujoch, 2010. [Online]. Available:http://www.mics.org/getDoc.php?docid=3426&docnum=1 14, 20

[41] Physical Acoustics Corporation, R6α Sensor Datasheet, 2010.[Online]. Available: http://www.pacndt.com/downloads/Sensors/Alpha/R6-Alpha.pdf 16, 36

[42] Maxim Integrated Products, ADC and DAC Glossary, April 2011.[Online]. Available: http://www.maxim-ic.com/app-notes/index.mvp/id/641 16

127

http://www.geo.uzh.ch/~ahasy/diss_hasler.pdf

http://www.acousticemission.com/index.aspx?go=products&focus=/Software/aewin.htm

http://www.acousticemission.com/index.aspx?go=products&focus=/Software/aewin.htm

http://www.mics.org/getDoc.php?docid=3426&docnum=1

http://www.pacndt.com/downloads/Sensors/Alpha/R6-Alpha.pdf

http://www.pacndt.com/downloads/Sensors/Alpha/R6-Alpha.pdf

http://www.maxim-ic.com/app-notes/index.mvp/id/641

http://www.maxim-ic.com/app-notes/index.mvp/id/641

BIBLIOGRAPHY

[43] Wikipedia, the free encyclopedia, Nyquist-Shannon samplingtheorem, April 2011. [Online]. Available: http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem 17

[44] D. Human, A method for the construction of minimum-redundancycodes, Proceedings of the IRE, vol. 40, no. 9, pp. 10981101, 1952. 26

[45] Analog Devices Inc., ADSP-2186M DSP MicrocomputerDatasheet, 2000. [Online]. Available: http://www.analog.com/static/imported-les/data_sheets/ADSP-2186M.pdf 34

[46] Actel Corporation, IGLOO Low-Power Flash FPGAs Datasheet,2010. [Online]. Available: http://www.actel.com/documents/IGLOO_DS.pdf 34

[47] STMicroelectronics, STM32Fxx Datasheet, September2009. [Online]. Available: http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00191185.pdf 34, 81

[48] Texas Instruments Inc., ADS8327 ADC Datasheet, January 2011.[Online]. Available: http://www.ti.com/lit/gpn/ads8327 35

[49] Analog Devices Inc., AD7908 ADC Datasheet, June 2009. [Online].Available: http://www.analog.com/static/imported-les/data_sheets/AD7980.pdf 35, 36, 50

[50] Physical Acoustics Corporation, PK6I Sensor Datasheet, 2010.[Online]. Available: http://www.pacndt.com/downloads/Sensors/PK/PK6I%20sensor%2064-08.pdf 36

[51] Analog Devices Inc, ADA4841 Operational Amplier Datasheet,December 2010. [Online]. Available: http://www.analog.com/static/imported-les/data_sheets/ADA4841-1_4841-2.pdf 36

[52] Linear Technology, LTSPice SPICE simulator, April 2011. [Online].Available: http://www.linear.com/designtools/software/ 36

[53] SD Group, SD Specications, Part 1, Physical Layer, SimpliedSpecication, September 2006. [Online]. Available: http://www.sdcard.org/developers/tech/sdcard/pls/Simplied_Physical_Layer_Spec.pdf 38

[54] FTDI Chip, TTL to USB Serial Converter Range of Cables Datasheet,September 2010. [Online]. Available: http://www.ftdichip.com/Support/Documents/DataSheets/Cables/DS_TTL-232R_CABLES.pdf 39, 80, 82

128

http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem

http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem

http://www.analog.com/static/imported-files/data_sheets/ADSP-2186M.pdf

http://www.analog.com/static/imported-files/data_sheets/ADSP-2186M.pdf

http://www.actel.com/documents/IGLOO_DS.pdf

http://www.actel.com/documents/IGLOO_DS.pdf

http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00191185.pdf



http://www.ti.com/lit/gpn/ads8327

http://www.analog.com/static/imported-files/data_sheets/AD7980.pdf

http://www.analog.com/static/imported-files/data_sheets/AD7980.pdf

http://www.pacndt.com/downloads/Sensors/PK/PK6I%20sensor%2064-08.pdf

http://www.pacndt.com/downloads/Sensors/PK/PK6I%20sensor%2064-08.pdf

http://www.analog.com/static/imported-files/data_sheets/ADA4841-1_4841-2.pdf

http://www.analog.com/static/imported-files/data_sheets/ADA4841-1_4841-2.pdf

http://www.linear.com/designtools/software/

http://www.sdcard.org/developers/tech/sdcard/pls/Simplified_Physical_Layer_Spec.pdf



http://www.ftdichip.com/Support/Documents/DataSheets/Cables/DS_TTL-232R_CABLES.pdf



BIBLIOGRAPHY

[55] The FreeRTOS Project, April 2011. [Online]. Available:http://www.freertos.org 47

[56] R. Barry, Using the FreeRTOS Real Time Kernel - ARM Cortex-M3Edition, Real Time Engineers Ltd., 2010. 48, 88, 89

[57] R. Barry, The FreeRTOS Reference Manual, Real Time EngineersLtd., 2011. 48, 88, 89

[58] Sensirion AG, Datasheet SHT21P, May 2010. [Online]. Available:http://www.sensirion.com/en/pdf/product_information/Datasheet_SHT21P_PWM.pdf 58, 100

[59] Saft S.A., LSH cell range, April 2011. [Online]. Available:http://www.saftbatteries.com/Produit_LSH_cell_range_303_8/Language/en-US/Default.aspx 65

[60] Physical Acoustics Corporation, USB AE Node Product Bulletin,April 2011. [Online]. Available: http://www.pacndt.com/downloads/LitDirectory/AE/USB_AE_Node.pdf 68

[61] Maxim Integrated Products, MAX9923H Datasheet, 2010.[Online]. Available: http://datasheets.maxim-ic.com/en/ds/MAX9922-MAX9923.pdf 100

[62] Texas Instruments Inc., MSP430x1xx Family, 2006. [Online].Available: http://focus.ti.com/lit/ug/slau049f/slau049f.pdf 101

129

http://www.freertos.org

http://www.sensirion.com/en/pdf/product_information/Datasheet_SHT21P_PWM.pdf

http://www.sensirion.com/en/pdf/product_information/Datasheet_SHT21P_PWM.pdf

http://www.saftbatteries.com/Produit_LSH_cell_range_303_8/Language/en-US/Default.aspx

http://www.saftbatteries.com/Produit_LSH_cell_range_303_8/Language/en-US/Default.aspx

http://www.pacndt.com/downloads/LitDirectory/AE/USB_AE_Node.pdf

http://www.pacndt.com/downloads/LitDirectory/AE/USB_AE_Node.pdf

http://datasheets.maxim-ic.com/en/ds/MAX9922-MAX9923.pdf

http://datasheets.maxim-ic.com/en/ds/MAX9922-MAX9923.pdf

http://focus.ti.com/lit/ug/slau049f/slau049f.pdf