Upload
phungxuyen
View
213
Download
0
Embed Size (px)
Citation preview
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
CHAPTER 2. BASIC CONCEPTS
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
CHAPTER 2. BASIC CONCEPTS
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
CHAPTER 2. BASIC CONCEPTS
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
CHAPTER 3. SYSTEM SPECIFICATION
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
3.1. PRELIMINARY EXPERIMENTS
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
CHAPTER 3. SYSTEM SPECIFICATION
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
3.1. PRELIMINARY EXPERIMENTS
Figure 3.4: The eect of downsampling the captured AE events. Thereparametrization results are compared to the original result sampledat 1MHz.
19
CHAPTER 3. SYSTEM SPECIFICATION
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
CHAPTER 3. SYSTEM SPECIFICATION
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
CHAPTER 3. SYSTEM SPECIFICATION
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
CHAPTER 4. HIGH DATA RATE SENSING IN LOW BANDWIDTH WSNS
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
CHAPTER 4. HIGH DATA RATE SENSING IN LOW BANDWIDTH WSNS
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
CHAPTER 5. AEBOARD HARDWARE DESIGN
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
CHAPTER 5. AEBOARD HARDWARE DESIGN
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
CHAPTER 5. AEBOARD HARDWARE DESIGN
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
CHAPTER 5. AEBOARD HARDWARE DESIGN
+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
CHAPTER 5. AEBOARD HARDWARE DESIGN
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
6.4. STM32 IMPLEMENTATION
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
6.4. STM32 IMPLEMENTATION
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
6.4. STM32 IMPLEMENTATION
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
6.4. STM32 IMPLEMENTATION
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
CHAPTER 6. PERMAE SOFTWARE DESIGN
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
CHAPTER 7. SYSTEM EVALUATION
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
CHAPTER 7. SYSTEM EVALUATION
!"#$%$&
'%()*+
,,-,, ,.-,, ,/-,, ,+-,, ,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
7.2. TEMPERATURE CYCLE TESTS
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
CHAPTER 7. SYSTEM EVALUATION
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
CHAPTER 7. SYSTEM EVALUATION
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
7.2. TEMPERATURE CYCLE TESTS
!"#$%
!&'()*
+,-.. *+-.. .*-.. ./-..
.
.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
CHAPTER 7. SYSTEM EVALUATION
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
7.2. TEMPERATURE CYCLE TESTS
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
CHAPTER 7. SYSTEM EVALUATION
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
APPENDIX A. AEBOARD AND PERMAE USER MANUAL
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
APPENDIX A. AEBOARD AND PERMAE USER MANUAL
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
APPENDIX A. AEBOARD AND PERMAE USER MANUAL
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
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
APPENDIX B. PERMAE DEVELOPER'S GUIDE
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
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
APPENDIX B. PERMAE DEVELOPER'S GUIDE
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
uint16 unsigned 16 bit integer number
uint24 unsigned 24 bit integer number
uint32 unsigned 32 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
Field Type Conv Unit Meaning
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
Field Type Conv Unit Meaning
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
Field Type Conv Unit Meaning
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
APPENDIX D. DATA UNITS AND CONVERSION FUNCTIONS
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
APPENDIX D. DATA UNITS AND CONVERSION FUNCTIONS
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:
Zeichnungsnummer: 0001
2011-05-03
D:\Altium Projects\trunk\projects\0011_acoustic_emission\STM32.SchDoc
16:20:16
Format:
A3
1.1Rev:
Ersteller:
Institut: ETH Zurich
Seite von2 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
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_D0ARM_SDIO_D1
ARM_SDIO_D2ARM_SDIO_D3
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
PISD10GNDPISD10SW PISD10SWD
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:
Zeichnungsnummer: 0001
2011-05-03
D:\Altium Projects\trunk\projects\0011_acoustic_emission\TinyNode.SchDoc
16:20:17
Format:
A3
1.1Rev:
Ersteller:
Institut: ETH Zurich
Seite von3 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
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
PISD20GNDPISD20SW PISD20SWD
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
PISD20GNDPISD20SW PISD20SWD
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
PIJ2021NLVCC05050SNS
PIJ2019NLVCC012000SNS
106
Zeichnungs Titel:
Acoustic Emission Sensing - Power Supply
File:
Datum:
Zeichnungsnummer: 0001
2011-05-03
D:\Altium Projects\trunk\projects\0011_acoustic_emission\Power.SchDoc
16:20:17
Format:
A3
1.1Rev:
Ersteller:
Institut: ETH Zurich
Seite von4 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
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
PIR1201 PIR1202COR12
PIR1301 PIR1302COR13
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
PIQ601PIR2301 PIR2401
PIW401
PIQ701PIR2501 PIR2601
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:
Zeichnungsnummer: 0001
2011-05-03
D:\Altium Projects\trunk\projects\0011_acoustic_emission\HealthSens.SchDoc
16:20:17
Format:
A3
1.1Rev:
Ersteller:
Institut: ETH Zurich
Seite von5 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
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
PAADC105PAADC104PAADC103PAADC102PAADC101
COADC1
PAADC206PAADC207PAADC208PAADC209PAADC2010
PAADC205PAADC204PAADC203PAADC202PAADC201
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
PAC4601 PAC4602COC46
PAC4701 PAC4702 COC47
PAC4801PAC4802
COC48
PAC4901PAC4902COC49
PAC5001 PAC5002COC50
PAC5101PAC5102COC51
PAC5201PAC5202
COC52
PAC5301 PAC5302
COC53
PAC5401PAC5402
COC54
PAC5501PAC5502COC55
PAC5601PAC5602
COC56
PAC5701PAC5702 COC57
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
PAJ202PAJ204PAJ206PAJ208PAJ2010PAJ2012PAJ2014PAJ2016PAJ2018PAJ2020PAJ2022PAJ2024PAJ2026PAJ2028PAJ2030
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
PAP1402 PAP1401COP14
PAP1502 PAP1501
COP15
PAP1602 PAP1601COP16PAP1702 PAP1701COP17
PAP1802
PAP1801 COP18
PAP1902
PAP1901 COP19
PAQ103PAQ102
PAQ101
COQ1
PAQ202 PAQ203PAQ201COQ2
PAQ302 PAQ303PAQ301COQ3
PAQ402 PAQ403PAQ401COQ4
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
PAR2301PAR2302COR23 PAR2401
PAR2402 COR24
PAR2501PAR2502COR25 PAR2601
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
PAR5201 PAR5202COR52
PAR5301PAR5302 COR53
PAR5401PAR5402 COR54
PAR5501 PAR5502
COR55
PAR5601
PAR5602 COR56
PAR5701PAR5702COR57
PAR5801PAR5802
COR58
PAR5901 PAR5902
COR59
PAR6001PAR6002
COR60
PAR6101 PAR6102
COR61
PAR6202 PAR6201 COR62
PAR6301PAR6302COR63
PAR6401PAR6402COR64
PAR6501PAR6502 COR65
PAR6601PAR6602COR66
PAR6702
PAR6701
COR67 PAR6801
PAR6802 COR68
PAR6901 PAR6902COR69
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
PASD20SWDPASD20SW PASD20WP
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
PAU10056PAU10055PAU10054PAU10053PAU10052PAU10051
PAU10050PAU10049
PAU10048PAU10047PAU10046PAU10045PAU10044PAU10043PAU10042PAU10041PAU10040PAU10039PAU10038PAU10037PAU10036PAU10035PAU10034PAU10033PAU10032PAU10031PAU10030PAU10029PAU10028PAU10027PAU10026PAU10025PAU10024PAU10023PAU10022PAU10021PAU10020PAU10019PAU10018PAU10017
PAU10016PAU10015PAU10014PAU10013PAU10012PAU10011PAU10010PAU1009PAU1008PAU1007PAU1006PAU1005PAU1004PAU1003PAU1002PAU1001COU10
PAU1101 PAU1102 PAU110NC
PAU1103PAU1104
COU11
PAU1205PAU1206
PAU1207PAU1208
PAU1204PAU1203
PAU1202PAU1201
COU12
PAU1305PAU1306
PAU1307PAU1308
PAU1304PAU1303
PAU1302PAU1301
COU13PAU1406PAU1407
PAU1408PAU1409PAU14010
PAU1405PAU1404
PAU1403PAU1402PAU1401
COU14
PAU1506PAU1507PAU1508PAU1509PAU15010
PAU1505PAU1504PAU1503PAU1502PAU1501
COU15
PAU1606PAU1607PAU1608PAU1609PAU16010
PAU1605PAU1604PAU1603PAU1602PAU1601
COU16
PAW101
PAW102COW1
PAW202PAW201 PAW203COW2
PAW301
PAW302
COW3
PAW401
PAW402 COW4
PAW501
PAW502
COW5
PAW602PAW601 PAW603COW6
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
PAADC106PAADC107PAADC108PAADC109PAADC1010
PAADC105PAADC104PAADC103PAADC102PAADC101
COADC1
PAADC206PAADC207PAADC208PAADC209PAADC2010
PAADC205PAADC204PAADC203PAADC202PAADC201
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
PAC4601 PAC4602COC46
PAC4701 PAC4702 COC47
PAC4801PAC4802
COC48
PAC4901PAC4902COC49
PAC5001 PAC5002COC50
PAC5101PAC5102COC51
PAC5201PAC5202
COC52
PAC5301 PAC5302
COC53
PAC5401PAC5402
COC54
PAC5501PAC5502COC55
PAC5601PAC5602
COC56
PAC5701PAC5702 COC57
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
PAJ202PAJ204PAJ206PAJ208PAJ2010PAJ2012PAJ2014PAJ2016PAJ2018PAJ2020PAJ2022PAJ2024PAJ2026PAJ2028PAJ2030
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
PAP1402 PAP1401COP14
PAP1502 PAP1501
COP15
PAP1602 PAP1601COP16PAP1702 PAP1701COP17
PAP1802
PAP1801 COP18
PAP1902
PAP1901 COP19
PAQ103PAQ102
PAQ101
COQ1
PAQ202 PAQ203PAQ201COQ2
PAQ302 PAQ303PAQ301COQ3
PAQ402 PAQ403PAQ401COQ4
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
PAR2301PAR2302COR23 PAR2401
PAR2402 COR24
PAR2501PAR2502COR25 PAR2601
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
PAR5201 PAR5202COR52
PAR5301PAR5302 COR53
PAR5401PAR5402 COR54
PAR5501 PAR5502
COR55
PAR5601
PAR5602 COR56
PAR5701PAR5702COR57
PAR5801PAR5802
COR58
PAR5901 PAR5902
COR59
PAR6001PAR6002
COR60
PAR6101 PAR6102
COR61
PAR6202 PAR6201 COR62
PAR6301PAR6302COR63
PAR6401PAR6402COR64
PAR6501PAR6502 COR65
PAR6601PAR6602COR66
PAR6702
PAR6701
COR67 PAR6801
PAR6802 COR68
PAR6901 PAR6902COR69
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
PASD20SWDPASD20SW PASD20WP
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
PAU10056PAU10055PAU10054PAU10053PAU10052PAU10051
PAU10050PAU10049
PAU10048PAU10047PAU10046PAU10045PAU10044PAU10043PAU10042PAU10041PAU10040PAU10039PAU10038PAU10037PAU10036PAU10035PAU10034PAU10033PAU10032PAU10031PAU10030PAU10029PAU10028PAU10027PAU10026PAU10025PAU10024PAU10023PAU10022PAU10021PAU10020PAU10019PAU10018PAU10017
PAU10016PAU10015PAU10014PAU10013PAU10012PAU10011PAU10010PAU1009PAU1008PAU1007PAU1006PAU1005PAU1004PAU1003PAU1002PAU1001COU10
PAU1101 PAU1102 PAU110NC
PAU1103PAU1104
COU11
PAU1205PAU1206
PAU1207PAU1208
PAU1204PAU1203
PAU1202PAU1201
COU12
PAU1305PAU1306
PAU1307PAU1308
PAU1304PAU1303
PAU1302PAU1301
COU13PAU1406PAU1407
PAU1408PAU1409PAU14010
PAU1405PAU1404
PAU1403PAU1402PAU1401
COU14
PAU1506PAU1507PAU1508PAU1509PAU15010
PAU1505PAU1504PAU1503PAU1502PAU1501
COU15
PAU1606PAU1607PAU1608PAU1609PAU16010
PAU1605PAU1604PAU1603PAU1602PAU1601
COU16
PAW101
PAW102COW1
PAW202PAW201 PAW203COW2
PAW301
PAW302
COW3
PAW401
PAW402 COW4
PAW501
PAW502
COW5
PAW602PAW601 PAW603COW6
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.
Figure F.5: Phase and amplitude compensated error, with external preamplier,50 kHz input frequency.
113
APPENDIX F. PERFORMANCE MEASUREMENTS
Figure F.6: Phase and amplitude compensated error, with external preamplier,100 kHz input frequency.
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
APPENDIX F. PERFORMANCE MEASUREMENTS
!"#$%
!&'()*
++,++ +-,++ *+,++ *-,++
+
.++
/++
0++
1++
*+++
*.++
*/++
*0++
*1++
.+++
23+
2.+
2*+
+
*+
.+
3+
/+
-+
0+4($'%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
APPENDIX F. PERFORMANCE MEASUREMENTS
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
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
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
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
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