80
2 / 2006 The corporate technical journal of the ABB Group www.abb.com/abbreview ABB Review a Embedded system technologies Canned solutions Trends in embedded systems page 9 Wireless sensor networks page 39 Making power lines sing page 50

Abb Embedded Sym

Embed Size (px)

Citation preview

Page 1: Abb Embedded Sym

2 / 2006

The corporate technical journal of the ABB Group

www.abb.com/abbreview

ABBReview

Pioneering spirits

A revolution in high dc current measurement

page 6

Team-mates: MultiMove functionality heralds a new era in robot applications

page 26

Best innovations 2004page 43

a

Embedded system

technologiesCanned solutions

Trends in embedded systemspage 9

Wireless sensor networkspage 39

Making power lines singpage 50

Page 2: Abb Embedded Sym

Tin cans have firmly established them-selves as the medium of choice for shipping and storing a broad range of products. An important part of this success story is their simplicity. If an opening-tool is needed at all, it is the modest and universally compatible can-opener.

Computer systems have long been the antithesis of this. Even simple tasks called for specialist skills. Not so for embedded systems! Here, the com-puter is usually contained within the device it controls and reacts directly to relevant events. In the extreme, it blends in so well that nobody knows it’s there – until they take a peek inside.

Page 3: Abb Embedded Sym

3ABB Review 2/2006

Editorial

“Embedded, Everywhere” is the title of a research agenda published in 2001 by the National Academy of Sciences in the USA. This agenda highlighted the importance the scien-tific community attached to research in the field of embed-ded computers. Embedded intelligent devices are today, five years later, pervasive and estimated to be more numer-ous than people on earth. By 2010, at least three embed-ded computers will exist for every living person. That equals 16 billion machines, growing to 40 billion by 2020. The European Union is currently formulating a major initia-tive covering research into the next generation of embed-ded device technologies, which will focus on the interac-tion between embedded networks. ARTEMIS1), as the pro-gram is called, rallies industries and academia to lead the development efforts, backed by funding from national gov-ernments and the EU. Asia is equally aware of the signifi-cance of embedded technologies for future economic growth and prosperity. Government-backed programs exist in Japan, Korea and China, each with its own flavor and emphasis. All these regional and national programs are driving towards the pervasive use of embedded devices in a multitude of applications across industries and large infrastructures, health and entertainment, fixed and mobile networks.

So what is this technology all about, and how do we, in ABB, use it? What challenges lie ahead? Where is the tech-nology going? This issue of ABB Review focuses on exactly these questions, and finds answers in almost all areas of ABB’s research and development.

Embedded computers have been around for some decades already. Their tasks have been restricted primarily to auton-omous applications and small networks involving limited interaction with humans. The term “embedded” refers to the nature of these tasks, which are mostly related to dedi-cated background functions that escape human awareness (as long as the tasks are performed correctly). For example, a modern car has well over 20 embedded computers taking care of systems for brakes, comfort, engine control etc. The next step foreseen for these embedded devices is their full membership of sensor-based networks as intelligent and communicative systems that are not only capable of ex-changing information among themselves, but also between different networks. The exponential increase in complexity compared to our current level of capability defines a tech-nology gap, which we now have to find ways of bridging,

hence the importance of initiatives like ARTEMIS in Europe and its counterparts in the USA and Asia.

Two guest-authors in this issue of ABB Review introduce the topic of embedded technologies to our readers. Dr. Kostas Glinos, of the European Commission, describes in the first article the importance the EU is attaching to this technology, and presents the ARTEMIS initiative in some detail. The second lead story is by Dr. Richard Zurawski, president of the ISA Group based in St Clara, California. He looks more carefully at the state of embedded hardware and software technologies and where the trends seem to be leading. A review of how ABB uses embedded systems in its own product portfolio completes the overview section.

Several applications of embedded systems are described next. This set of articles cover a wide range of products with enhanced customer benefits, which originate from the incorporation of embedded technologies already in the initial design-phase. The next section is devoted to a broad spectrum of com-munication capabilities of embedded devices. Wireless communication and industrial Ethernet, fieldbus and power-line carrier are just a few examples that illustrate how ABB can support information exchange on different levels and over different media. The last two sections of this issue of ABB Review describe software and hardware implementa-tions.

The breadth of technologies required to bring performance benefits to our customers is remarkable. Maintaining up-to-date knowledge in a number of rapidly changing fields as wide apart as low power applications to software generation and verification, from advanced signal processing to FPGA technologies is a great challenge for our global research and development teams. But then, tackling such challenges is why researchers love research in the first place. Enjoy your reading

Peter TerwieschChief Technology OfficerABB Ltd.

Embedded system technologies

Footnote1) ARTEMIS stands for Advanced Research & Technology for Embedded Intelligece

and Systems

Page 4: Abb Embedded Sym

4 ABB Review 2/2006

ABB Review 2/2006Embedded system technologies

Contents

6The challenge of embedded systemsManaging the revolution in embedded digital technolo-

gies, one of the fastest growing sectors in IT today.

9Trends in Embedded SystemsOpportunities and challenges for System-on-Chip

and Networked Embedded Systems technologies in

industrial automation.

14Embedded system technology in ABB Current and future challenges. Advances in performance

and functionality, with reductions in cost and size,

present developers with new challenges.

18Embedded power protectionEnhanced embedded applications in power system

automation handle protection aspects alongside many

additional dedicated applications.

23Drivers of changeWhy the DTC drive controller from ABB is fast becoming

the “torque” of the town.

26Roll and controlWhat do a lightweight train and a rolling-mill have in

common? The fast and flexible AC800 PEC plays a big

role in their control!

30Embedded systems extend automationSystem 800XA incorporates a multitude of embedded

devices.

35DriveMonitorNew lifecycle management software with its finger

on the system’s pulse.

39Wireless sensor networksIntroducing wireless sensor networks to the world of

industrial automation.

43High-performance EthernetABB broadens its range of Ethernet-compatible devices

to enhance communication.

46Fieldbuses for drivesAdvanced fieldbuses are improving drive connectivity.

48Motor medicalBoosting a motor’s productivity by watching

its health.

50Making power lines singPutting more power into communications –

ABB’s ETL600 transmits information across power lines.

Embedded system technologies

Page 5: Abb Embedded Sym

5ABB Review 2/2006

14

39

46

58

www.abb.com/abbreview

54Bright ideasProduct development processes in ABB’s distribution

automation business benefit from international

cooperation.

58Do-it-yourself roboticsThe FlexPendant software development kit brings

user-friendly robot programming to the desktop.

62Design patternsHow ABB created the AC800 PEC controller

66Wireless power in wireless productsFewer flexes, more flexibility. Bringing wireless power to

devices in hard-to-reach places cuts installation costs

and presents new opportunities for distributed electronic

devices.

70Coming of ageFPGAs bridge the gap between hard- and software.

75Signal processing in embedded systemsNew algorithms for device-level instrumentation enhance

performance and extend functionality.

Page 6: Abb Embedded Sym

6 ABB Review 2/2006

What do a mobile phone, an industrial robot, a cable modem, an MP3 player, and a car have in common? They are all examples of products that use em-bedded systems. In fact, embedded technologies are one of the fastest growing sectors in IT today. However, increasing pressure to bring innovative products to the market ever faster and at ever diminishing prices means that guaranteeing product quality, while reducing the cost, development time and system complexity, has become a tough challenge.

The challenge of embedded systemsKostas Glinos

ly in medicine, in particular in diag-nostic medical equipment, and in the increasing variety of intelligent devic-es that are implanted into the human body. Transportation is another area that has seen a rapid proliferation of embedded systems, be it cars, trucks, trains or airplanes.

The numbers are staggering: it is esti-mated that more than 90 percent of all computing devices are to be found in embedded rather than in desktop systems. In terms of market value, for example, the automotive sector alone accounts for about 5 percent of the

More remarkable, however, is the less visible revolution in embedded digital technology. Embedded digital tech-nology is found in all kinds of equip-ment and systems, and is used to increase functionality, as well as to improve operation at low cost. In-deed, embedded computers are now found in almost all technical devices, from simple everyday home applianc-es, to facilities and facility manage-ment such as heating, air condition-ing, elevators and escalators, and in production units from robotics to production automation and control systems. They are also used extensive-

In a space of less than four decades, digital information technology has to-

tally revolutionized the world in which we live. It has evolved from mainframe computers – mainly operated as hosts in computing centers – to the net-worked desktops and laptops we know today. Our everyday business and home life is deeply affected by this ex-tensive digital infrastructure: from keep-ing in contact with friends and relatives around the world, to staying in business in a globalized and highly competitive market. Computers have become every-day tools, deeply integrated into all kinds of social and business activities.

Europe is considered a world leader in embedded technologies for the aerospace, automotive, industrial, communication and consumer elec-tronics industries. However, this lead-ing position is threatened by global competition, fragmentation and lack of coordination across these industry sectors. Maintaining a leading position in embedded systems technology will require significant – and appropriately targeted – investment in research and development.

To address these issues, the Europe-an Commission has facilitated the development of an initiative called ARTEMIS. ARTEMIS is a broad alli-ance of industrial and research play-ers in the field of embedded system technologies. The ARTEMIS partner-ship draws upon many industrial sec-tors, including automotive, aerospace, consumer electronics, communica-tions, medical and manufacturing, where European industry remains strong.

Cou

rtes

y A

irbus

Page 7: Abb Embedded Sym

Embedded system technologies

7ABB Review 2/2006

The challenge of embedded systems

world semiconductor market (approxi-mately N 200 billion in 2005).

What is even more striking is how embedded systems are increasing the value of many products. For example, embedded systems now account for 20 percent of the total value of an average car and this is expected to increase to 36 percent in 2009. In the same year, embedded electronics and software will constitute 22 percent of the value of industrial automation systems, 41 percent of consumer electronics and 33 percent of medical equipment.

The growth rate is currently exceeding 10 percent per annum in all applica-tion sectors, and by 2020, it is predict-ed that there will be over 40 billion embedded chips worldwide.

Thanks to significant advances in semiconductor technology – which is itself driven by customer demands for innovative products and services, with increasing functionality at ever dimin-ishing prices – embedded systems have evolved from the simple stand-alone, single-processor type comput-ers of the 1980s and early 1990s, to the sophisticated multi-processor systems in use today. The downside, however, is that systems are becoming ever more complex and harder to design, test and verify. As these sys-tems are themselves becoming more interconnected, they are also becom-ing more vulnerable. True interopera-bility is hampered by the lack of com-mon open standards and appropriate middleware. While many of the devel-

opments are still sector-specific, there are significant synergies between sec-tors that should be exploited. And engineers with the appropriate skills in, for example, system architecture, are in short supply. These problems must be overcome. For its part, Euro-pean industry is expected to invest more than N 22 billion per annum in embedded systems research and development by 20091). This is almost double what it invested in 2003.

Because of the above research and industrial challenges, and the impor-tance of embedded systems technolo-gy for key industrial sectors (from industrial automation and medical equipment to automotive and avion-ics), the European Commission has devoted a specific part of its Informa-tion Society Technologies (IST) pro-gram to embedded systems research. In the last three years alone, it has invested N 140 million in collaborative projects between industry, academia and research centers. These projects focus largely on systems design, safe-ty-critical systems, embedded comput-ing, middleware platforms, wireless sensor networks, and distributed and hybrid control systems. Embedded systems are also one of the six “pil-lars” of ICT research in the European Commission‘s proposals for the 7th Framework Programme, due to start in 2007.

In 2004, the Technology Platform ARTEMIS (Advanced Research and Technology for EMbedded Intelligence and Systems) was set up. ARTEMIS is an industry-led initiative to reinforce the EU’s position as a leading global player in the design, integration and supply of embedded systems2). Its manifesto, entitled “Building ARTE-MIS”, was signed by 20 executives from various EU companies, and is aimed at establishing and implement-ing a coherent and integrated Euro-pean strategy for embedded systems that covers aspects from research and development priorities, and the research infrastructures needed, to the

standardization policy as well as the educational curricula. This strategy has been recently published as the ARTEMIS “Strategic Research Agenda”.

The driving force behind ARTEMIS is the vision of a society where all systems, machines, and objects have become digital, communicating, self-managed resources. These transforma-tions are possible through advances in embedded systems technologies and their large-scale deployment, not only in industry and services, but in all areas of human activity. Such devel-opments have a range of important consequences for society and the economy: Life in our society – and its safety and security – will depend increas-ingly on embedded systems.

The competitiveness of European industries, in almost all sectors, will

Footnotes1) FAST Study on “Worldwide Trends and R&D

Programmes in Embedded Systems in view of

maximising the impact of a Technology Platform

in the area”2) http://www.cordis.lu/ist/artemis/index.html

ARTEMIS – European joint technology initiative for embedded systems

Music center – courtesy Nokia

A380 cockpit – courtesy Airbus

Page 8: Abb Embedded Sym

8 ABB Review 2/2006

The challenge of embedded systems

rely on innovation capabilities in the area of embedded systems.

Given the dramatically increasing importance of embedded systems to productivity growth, these tech-nologies will be critically important in redressing the present imbalance in productivity growth between Europe, the US and Asia.

Maintaining a leading position in embedded systems technology will require significant investment in research and development that is focused on specific joint priorities. While tackling the R&D challenges is necessary it is not, on its own suffi-cient. ARTEMIS will facilitate and stimulate European success in embed-ded systems by establishing an envi-ronment supportive of innovation, in which both co-operation and com-petition in technological development are enhanced. It will also proactively stimulate the emergence of a new

supply industry for components, tools and design methodologies, supporting embedded systems, and focus research and development to make more effec-tive use of resources to avoid frag-mentation and facilitate deployment.

Embedded digital tech-nology is found in all kinds of equipment and systems, and is used to increase functionality, as well as to improve operation at low cost.

While custom-designed embedded systems add high value for customers, and individual projects and products may be highly profitable, the markets themselves are highly fragmented. Traditionally, this has led to the frag-mentation of both the supply industry and R&D investment. The ARTEMIS strategy was conceived to overcome this fragmentation so as to increase the efficiency of technological devel-opment and, at the same time, facili-tate the establishment of a competitive market in the supply of technologies.

Embedded system technologies

The conception, design and deploy-ment of customized systems will add even greater value to most products and services in the future Information Society. Over the last few decades, Europe has been strong in this area, most notably with successes in mobile phones, bespoke systems for transport and aerospace, and industrial engi-neering. ARTEMIS aims to derive max-imum benefit from Europe’s strengths, while being cognizant of the strengths of global competitors. The ARTEMIS approach will remove barriers between application sectors, thus stimulating creativity and yielding multi-domain reusable results.

I strongly believe that by creating an environment that favors and supports innovation in embedded systems and by focusing our research and develop-ment resources to achieve common and ambitious objectives, we will not only maximize our impact in terms of industrial competitiveness, but we will also improve the quality of life, safety and security of people. Success in this endeavor can be achieved only if all parties – public or private, industrial or academic – work closely together and remain committed to their com-mon objectives. Rapid progress in this direction over the past year makes me confident that this will indeed be the case and that this collective effort will be successful.

Kostas Glinos

European Commission

The views expressed are those of the author and do

not necessarily represent the official view of the Euro-

pean Commission on the subject.

Kostas Glinos has been with the European Commission since 1992. He now leads the Embedded Systems unit of the IST Program. Before joining the Commission he worked with multinational com-panies and research institutes in the US, Greece and Belgium. He holds a Ph.D. in chemical engi-neering and a Masters’ in financial management.

Kostas Glinos

Concept car – courtesy DaimlerChrysler Ambient room – courtesy Philips Embedded components

Industrial robot – courtesy ABB

Page 9: Abb Embedded Sym

9ABB Review 2/2006

Trends in embedded systemsOpportunities and challenges for System-on-Chip and Networked Embedded Systems technologies in industrial automationGrant Martin, Richard Zurawski

Courtesy Philips

Advances in process technology and the availability of new design tools are broadening the scope of embedded systems; from being implemented as a set of chips on a board, to a set of modules in an integrated circuit. System-on-Chip (SoC) technology is now being deployed in industrial automation, enabling the creation of complex field-area intelligent devices. This trend is accompanied by the adoption of platform-based design, which facilitates the design and verifica-tion of complex SoC through the extensive re-use of hardware and software IP (intellectual property). A further important aspect of the evolution of embedded systems is the trend towards networking of embedded nodes using specialized network technologies, frequently referred to as Networked Embedded Systems (NES).

Page 10: Abb Embedded Sym

10 ABB Review 2/2006

Trends in embedded systems

System-on-Chip (SoC) represents a revolution in integrated circuit (IC)

design, enabled by advances in pro-cess technology, which allow the inte-gration of the main components and subsystems of an electronic product onto a single chip or integrated chip-set [1]. This development has been embraced by designers of complex chips because it permits the highest possible level of integration, resulting in increased performance, reduced power consumption, and advantages in terms of cost and size. These are very important factors in the design process, and the use of SoC is argu-ably one of the key decisions in devel-oping real-time embedded systems.

SoC can be defined as a complex inte-grated circuit, or integrated chipset, that combines the main functional ele-ments or subsystems of a complete end product in a single entity. Nowa-days, the most challenging SoC de-

signs include at least one programma-ble processor, and very often a combi-nation of at least one RISC (reduced instruction set computing) control pro-cessor and one digital signal processor (DSP). They also include on-chip com-munications structures – processor bus(es), peripheral bus(es) and some-times a high-speed system bus. A hier-archy of on-chip memory units, as well as links to off-chip memory, is especially important for SoC proces-sors. For most signal-processing appli-cations, some degree of hardware-based accelerating functional unit is provided, offering higher performance and lower energy consumption. For interfacing to the external world, SoC design includes a number of peripher-al processing blocks consisting of ana-logue components as well as digital interfaces (for example, to system bus-es at board or backplane level). Future SoC may incorporate MEMS-based (micro electro-mechanical system) sen-

sors and actuators, or chemical pro-cessing (lab-on-a-chip) 1 .

All interesting SoC designs comprise both hardware and software compo-nents. These include programmable processors, real-time operating sys-tems, and other elements of hardware-dependent software. Thus, the design and use of SoCs not only concerns hardware – it also involves system-level design and engineering, hard-ware–software tradeoffs and partition-ing, and software architecture, design and implementation.

System-on-a-Programmable-Chip Recently, the scope of SoC has broad-ened. From implementations using custom IC, application specific IC (ASIC) or application-specific standard part (ASSP), the approach now in-cludes the design and use of complex reconfigurable logic parts with em-bedded processors. In addition other application-oriented blocks of intellec-tual property, such as processors, memories, or special purpose func-tions bought from third parties are incorporated into unique designs.

These complex FPGAs (Field-Program-mable Gate Arrays) are offered by sev-eral vendors, including Xilinx (Virtex-II PRO Platform FPGA, Virtex-IV) and Altera (SOPC). The guiding principle behind this approach to SoC is to com-bine large amounts of reconfigurable logic with embedded RISC processors, in order to enable highly flexible and tailorable combinations of hardware and software processing to be applied to a design problem. Algorithms that contain significant amounts of control logic, plus large quantities of dataflow processing, can be partitioned into the control RISC processor with reconfigu-rable logic for hardware acceleration. Although the resulting combination does not offer the highest performance, lowest energy consumption, or lowest cost – in comparison with custom IC or ASIC/ASSP implementations of the same functionality – it does offer enor-mous flexibility in modifying the design in the field, and avoids expensive Non-Recurring Engineering (NRE) costs as-sociated with field changes. Thus, new applications, interfaces and improved algorithms can be downloaded to prod-ucts already working in the field.

Embedded system technologies

1 A typical SoC device for consumer applications

External memory access

FlashRAMDMA

ICacheDCache

Microprocessor DSP

Peripheral bus

PLL

Test

PCI

USB

Bus bridge

Audio CODEC

RAMFlash

DCacheICacheSystem

bus

MPEG decode

Video I/F

Disk controller

100 base-T

Page 11: Abb Embedded Sym

11ABB Review 2/2006

Trends in embedded systems

Products in this area also include other processing and interface cores: these consist of multiply–accumulate (MAC) blocks aimed at DSP-type data-flow signal- and image processing applications, and high-speed serial interfaces for wired communications such as SERDES (serializer/de-serializ-er) blocks. In this sense, system-on-a-programmable-chip SoCs are not exactly application-specific, but not completely generic either.

The guiding principle behind this approach to SoC is to combine large amounts of reconfigurable logic with embedded RISC processors.

It remains to be seen whether system-on-a-programmable chip SoCs will be successful in delivering high-volume consumer applications, or whether they will be restricted to two main areas: Rapid prototyping of designs that will be re-targeted to ASIC or ASSP implementations; and high-end, relatively expensive elements of the communications infrastructure that require in-field flexibility, and which can afford the higher levels of cost and energy consumption in combina-tion with reduced performance.

Intermediate forms, such as the use of metal-programmable gate-array style logic fabrics, together with hard-core processor subsystems and other cores – as provided in the “Structured ASIC” offerings of LSI Logic (RapidChip) and NEC (Instant Silicon Solutions Plat-form) – represents an intermediate form of SoC between the full-mask approach and the field-programmable gate-array approach. Specific trade-offs here are much slower design cre-ation (a few weeks rather than a day or so); higher non-recurring engineer-ing than FPGA (but much lower than a full set of masks); and better cost, performance and energy consumption than FPGA (perhaps only 15–30 per-cent worse than an ASIC approach). Further interesting hybrid approaches, such as ASIC/ASSP with on-chip FPGA regions, are also emerging to give design teams more choices. A final

interesting variation is a combination of a configurable processor, which is implemented partly in fixed silicon, together with an FPGA region, which is used for instruction extensions and other hardware implementations in the field. Stretch Inc., a semiconductor company, for example, uses the Ten-silica configurable processor to imple-ment this type of platform SoC 2 .

Platforms and Programmable PlatformsRecent years have seen a more inte-grated approach to the design of com-plex SoC and the re-use of virtual com-ponents – this is called “platform-based design” [1, 2]. Platform-based design can be defined as a planned design methodology that reduces the time and effort required – as well as the risk involved – in designing and verifying a complex SoC. This is accomplished by extensive re-use of combinations of hardware [3] and software [4] IP. In contrast to IP re-use in a block-by-block manner, platform-based design assembles groups of components into a re-usable platform architecture. This re-usable architecture, together with libraries of pre-verified and pre-charac-terized, application-oriented hardware and software virtual components, con-stitutes a SoC integration platform.

There are several reasons for the growing popularity of the platform approach in industrial design. These include the increase in design produc-tivity, the reduction in risk, the ability to utilize pre-integrated virtual com-ponents from other design domains more easily, and the ability to re-use SoC architectures created by experts. Industrial platforms include full appli-cation platforms for specific product areas, such as Philips Nexperia and TI OMAP [5], reconfigurable SOPC platforms, and processor-centric plat-forms. Processor-centric platforms – such as those using multiple Tensilica-configured, extended processors, or ARM PrimeXsys – concentrate on the processor, its required bus architec-ture and basic sets of peripherals, along with RTOS (real-time operating systems) and basic software drivers.

Platform FPGAs and SOPC devices can be thought of as a “meta-platform”; that is, a platform for creating platforms. These devices contain a basic set of more generic capabilities and IP embed-ded processors, on-chip buses, special IP blocks such as MACs and SERDES, and a variety of other pre-qualified IP blocks. Design teams can obtain such devices from companies like Xilinx and

Embedded system technologies

2 Tensilica’s LX processor

Instruction fetch/decode

Designer-defined FLIX parallel execution pipelines – "N" wide

Base ISA execution pipeline

Processor controls

Trace/JTAG/OCD

Interrupts, breakpoints, timers

Local instruction memories

External bus interface

Local data memories

Xtensa local memory interface

Register file

Base ALU

Optional execution units

Vectra LX DSP engine

Data load/store unit

Load/store unit #2

Use

r-de

fined

exe

cutio

n un

its,

regi

ster

rile

s an

d in

terf

aces

User-defined execution unit

Processor interface (PIF) to system bus

User-defined

queues/ports up to

1M pins

Use

r-de

fined

exe

cutio

n un

its,

regi

ster

file

s an

d in

terf

aces

Base ISA feature Configurable functions Optional function Optional and configurable Designer-defined feature (TIE)

Page 12: Abb Embedded Sym

12 ABB Review 2/2006

Trends in embedded systems

Altera, and then customize the meta-platform to their own application space by adding application domain-specific IP libraries. This can then be delivered to derivative design teams.

Networked Embedded SystemsAnother important facet of the evolu-tion of embedded systems is the emer-gence of distributed embedded systems, frequently termed networked embed-ded systems, where the word “net-worked” signifies the importance of the networking infrastructure and commu-nication protocol. A networked embed-ded system is a collection of spatially and functionally distributed embedded nodes, interconnected by means of wireline and/or wireless communica-tion infrastructure and protocols, and interacting with the environment (via sensor/actuator elements) and each other. Within the system, a master node can also be included to coordinate computing and communication, in order to achieve specific objectives.

Controllers embedded in nodes or field devices, such as sensors and actuators typically provide on-chip signal conversion, data and signal processing, and communication func-tions. The ever-increasing functional-ity, processing and communication capabilities of controllers have been instrumental in the emergence of a widespread trend for the networking of field devices around specialized networks, frequently referred to as field area networks. (A field area net-work is normally a digital, two-way,

multi-drop communication link [6].) In general, the benefits of using special-ized (field area) networks are numer-ous and include: increased flexibility through combining embedded hard-ware and software; improved system performance; and ease of system in-stallation, upgrade, and maintenance.

Recent years have seen a more integrated approach to the design of complex SoC and the re-use of virtual components – this is called “platform-based design”.

Networked embedded systems are present in a variety of application do-mains: for example; automotive, train, aircraft, office building and industrial – primarily for monitoring and con-trol. Representative examples of net-work embedded systems include net-works connecting field devices such as sensors and actuators with field controllers; for instance, programma-ble logic controllers (PLCs) in indus-trial automation or electronic control units (ECUs) in automotive applica-tions. They are also used in man-ma-chine interfaces; for example, in dash-board displays in cars and SCADA (super visory control and data acquisi-tion) in industrial automation. The specialized network technologies employed are as diverse as the appli-cation areas. For instance: PROFIBUS,

PROFInet or EtherNet/IP (both sup-porting real-time communication) in industrial control and automation; LonWorks, BACnet, and EIB/KNX in building automation and control; CAN, TTP/C and FlexRay in automotive ap-plications; and Train Communication Network (TCN) in train automation. The diversity of requirements imposed by different application domains (soft/hard real-time, safety critical, network topology, and so forth) necessitated a variety of solutions, and the use of different protocols based on different operating principles. This has resulted in a plethora of networks developed for specific application domains [6] 3 .

Because of the nature of the commu-nication requirements imposed by applications, field area networks – unlike local area networks (LANs) – tend to have low data rates, small data packets, and typically require real-time capabilities, which may demand deter-ministic or time-bounded data transfer. However, data rates above 10 Mbit/s, typical of LANs, have become com-monplace in field area networks. For field area networks employed in in-dustrial automation (unlike in building automation and control) there is little need for routing functionality or end-to-end control. As a consequence, only layers 11) (physical layer), 2 (data link layer, including implicitly the medium access control layer), and 7 (application layer, which also covers user layer) of the ISO/OSI reference model [7] are used in these networks.

The need to guarantee a deterministic response requires the use of appropri-ate scheduling schemes, which are frequently implemented in applica-tion-domain specific real-time operat-ing systems or custom-designed, “bare-bone”, real-time executives.

The networked embedded systems used in safety-critical applications, such as x-by-wire, that adopt mecha-tronic solutions to replace mechanical or hydraulic solutions with electrical/electronic systems must be highly dependable to ensure a failsafe sys-tem. Examples of such embedded

Embedded system technologies

Footnote1) For a brief overview of the OSI model, see figure 1

on page 47.

3 Typical field area network architecture in industrial automation

Control network

Field areanetwork

(Fieldbus)

motor

switchgear drives instrument

controller

I/O modules

Page 13: Abb Embedded Sym

13ABB Review 2/2006

Trends in embedded systems

systems include fly-by-wire in aero-planes and steer-by-wire in automo-tive applications, where failure could endanger human life, property, or the environment. To avoid such risks, reliable, failsafe services must be delivered at the request of the system user. The dependability of x-by-wire systems is one of the main require-ments, as well as a constraint on the adoption of this kind of system.

Although the use of wireline-based field area networks is most common, wireless technology – including wire-line/wireless hybrid solutions – offers incentives in a number of application areas. In industrial automation, for in-stance, wireless device (sensor/actua-tor) networks can provide support for mobile operation in the case of mobile robots and the monitoring and control of equipment in hazardous and diffi-cult-to-access environments. A separate category is wireless sensor networks envisaged for monitoring purposes.

Opportunities and challenges in SoC and MPSoC There are many opportunities arising from the efficient and error-free design of SoC, and in particular Multi-Proces-sor System-on-Chip (MPSoC), which combines the advantages of parallel processing with the high integration capability of SoC. Other areas of inter-est include testing of embedded cores in SoC, power-aware computing, secu-rity in embedded systems, and devel-opment of safety-critical systems in the context of x-by-wire and various other applications [8].

Ever-increasing circuit densities and operating frequencies, as well as the use of system-on-chip designs, have resulted in enormous test-data vol-umes for today’s embedded core-based integrated circuits. Reducing data volume and time are two of the main challenges in testing these kinds of circuits. Other problems include: the growing disparity between perfor-mance of the design and the automatic test equipment, which makes at-speed testing – particularly of high-speed cir-cuits – a challenge, and results in in-creasing yield loss; high cost of manu-ally developed functional tests; and a growing cost of high-speed and high-pincount testers.

Growing power dissipation, resulting from the increase in density of inte-grated circuits and clock frequency, has a direct impact on the cost of packaging and cooling, as well as reli-ability and lifetime. These and other factors, such as battery-based power supply and device-restricted size (as in the case of hand-held devices), make designing for low power con-sumption a high priority for embed-ded systems. The design techniques and methodologies aimed at reducing both static and dynamic power dissi-pation tend to focus on the following areas: system/application level optimi-zation, which explores task implemen-tations exhibiting different power/en-ergy versus quality-of-service charac-teristics; energy-efficient processing subsystems like voltage and frequency scaling, dynamic resource scaling, and processor core selection; and energy-efficient memory subsystems, such as cache hierarchy tuning, novel hori-zontal and vertical cache partitioning schemes, as well as dynamic scaling of memory elements.

The relatively limited commercial band-width resources for computing, memo-ry, and communication of embedded device controllers (eg, field devices in industrial automation) poses consider-able challenges for the implementation of effective security policies, which, in general, are resource- demanding. This limits the applicability of the main-stream cryptographic protocols, even vendor-tailored versions. Operating s ystems running on small-footprint controllers tend to implement essential services only, and do not provide authentication or access control to pro-tect mission- and safety-critical field devices. A growing demand for remote access to process data at factory-floor level may expose automation systems to potential electronic security attacks, which may compromise the integrity of these systems and endanger plant safety. The system/plant availability requirement may render the updating of security software in running field devices impractical or too risky.

Grant Martin

Tensilica, USA

[email protected]

Richard Zurawski

ISA Group, USA

[email protected]

References

[1] H. Chang, L. Cooke, M. Hunt, G. Martin, A. Mc-

Nelly, L. Todd: Surviving the SoC Revolution: A

Guide to Platform-Based Design. Kluwer Academ-

ic Publishers, 1999.

[2] A. Sangiovanni-Vincentelli, G. Martin: Platform-

Based Design and Software Design Methodology

for Embedded Systems. IEEE Design and Test of

Computers 18 (2001) 6, 23–33.

[3] M. Keating, P. Bricaud: Reuse Methodology Manu-

al for System-on-a-Chip Designs. Kluwer Aca-

demic Publishers, 1998 (First Edition), 1999 (Sec-

ond Edition), 2002 (Third Edition).

[4] G. Martin, C. Lennard: Invited CICC paper.

Improving Embedded Software Design and

I ntegration for SOCs. Custom Integrated Circuits

Conference, May 2000, 101–108.

[5] G. Martin, H. Chang (Editors): Winning the SOC

Revolution: Experiences in Real Design. Kluwer

Academic Publishers, 2003.

[6] R. Zurawski (ed.): The Industrial Communication

Systems, Special Issue. Proceedings of the IEEE,

93 (2005) 6.

[7] Zimmermann H.: OSI Reference Model: The ISO

model of architecture for open system intercon-

nection. IEEE Transactions on Communications,

28(4): 425–432, 1980.

[8] R. Zurawski (ed.): Embedded Systems Handbook.

Taylor & Francis, 2005.

Grant Martin is Chief Scientist at Tensilica, Inc. He received his Bachelor’s and Master’s Degrees in Mathematics from the University of Waterloo, Canada. He worked at Burroughs in Scotland, BNR/Nortel in Canada, and Cadence Design Systems in San Jose, California, prior to joining Tensilica in 2004.

Grant Martin

Embedded system technologies

Richard Zurawski is President of the ISA Group, San Francisco. He held executive positions with San Francisco Bay area companies, Kawasaki Electric, Tokyo, and was a professor at the Institute of Industrial Sciences, University of Tokyo. He is Editor of a book s eries on Industrial Information Technology, CRC Press/Taylor & Francis. He holds an MSc in Elec-trical Engineering, and a PhD in Computer Science.

Richard Zurawski

Page 14: Abb Embedded Sym

ABB Review 2/2006

As the underlying technologies evolve, embedded systems are find-ing their way into an increasing range of ABB products and applications. Advances in this technology lead to higher performance and more func-tionality on the one hand, and re-duced cost and size on the other. While this benefits the end user, the ever increasing complexity of embed-ded systems imposes new challenges for developers. This article provides a brief introduction to the use and application of embedded system technology in ABB’s power and auto-mation products – and the challenges being faced now and in the future.

Embedded system technology in ABB Christoffer Apneseth

14

Page 15: Abb Embedded Sym

15ABB Review 2/2006

Embedded system technology in ABB

Embedded systems are special purpose comput-

er systems that are totally in-tegrated and enclosed by the devices that they serve or control – hence the term “embedded systems”. While this is a generally accepted definition of embedded sys-tems, it does not give many clues as to the special char-acteristics the systems pos-sess.

How is ABB applying embedded systems?To appreciate the purpose of embedded systems, it might be useful to answer some questions that will help to understand the underlying technology.

The first question is: How do embed-ded systems differ from general-pur-pose computer systems? The answer is, it depends. By definition, an embed-ded system is designed to perform a set of predefined tasks. This could range in complexity from simple super-vision of the operation of an electrical switch, to controlling the movements of a powerful and highly flexible industrial robot. The two solutions, accordingly, will look completely dif-ferent. The former would be optimized for very low cost, high volume pro-duction and the execution of a small set of pre-defined algorithms. The latter would be designed for comput-ing complex, programmable movement paths and transforming the signals that control the motors of the manipulator.

The second question to ask is: Why do we need embedded systems? The answer to this is that general-purpose computers, like PCs, would be far too costly for the majority of products that incorporate some form of embedded system technology. A general-purpose solution might also fail to meet a number of functional or performance requirements such as constraints in power-consumption, size-limitations, reliability or real-time performance.

Embedded systems – where are they found?ABB has been developing automation and power technologies for more than

a hundred years. The concepts that underlie some of these technologies have evolved slowly: modern power transformers, for instance, work ac-cording to the same principles as they did in the early days of electric power transmission. And despite huge prog-ress in switching technologies and material science, circuit breakers have been based on the same principles for the last fifty years. Now that small and powerful microcontrollers are avail-able at low cost, embedded system components are finding their way into these long-established products. Here, the embedded systems typically per-form a secondary function: they are used to supervise, protect or control the primary function of the product. The technology is a way of providing these attributes more cheaply, than

the alternatives, or with new value-added features.

Many other product families now offered by ABB could not have been conceived without embedded system technology. Examples are Distributed Control Systems (DCS) that can safely auto-mate and control large and complex industrial plants, such as oil refineries, power plants and paper mills. In the early days of industrial auto-mation, relay logic was used to perform simple control functions. With the advent of integrated circuits and the first commercial microcon-

trollers in the seventies and eighties, programmable industrial controllers were introduced to perform more complex control logic. Today, ABB’s Industrial IT Extended Automation System 800xA integrates widely dis-tributed and intelligent field devices with high-level system functions that optimize production assets, as well as the process itself.

The software part of a modern embedded system can consist of hundreds of thousand lines of code.

Challenges in industrial applications of embedded systemsThis issue of ABB Review discusses the wide range of opportunities and challenges associated with the integra-tion of embedded system technology into ABB’s portfolio of products and solutions. Many of the benefits and re-quirements are typical of embedded systems in general – such as low cost, small size, etc. – some challenges are more specifically associated with in-dustrial applications.

Industrial requirementsIndustrial requirements vary enor-mously from application to applica-tion, but special industrial require-ments typically include: Availability and reliability Safety

Modern transformer technology

Installation of advanced instrumentation in the field

Embedded system technologies

Page 16: Abb Embedded Sym

16 ABB Review 2/2006

Embedded system technology in ABB

Real-time, deterministic response

Power consumption Lifetime

Availability and reliabilityAutomation and power sys-tems must have very high availability and be extremely reliable in order to minimize the cost of operation (ie to minimize scheduled as well as unplanned maintenance time).

SafetyWhile customers demand high quality and reliability from most of their embed-ded systems, it is not neces-sarily critical if, say, a PDA (personal digital assistant) needs to be restarted after an application causes the system to fail. For industrial applications, however, the effect of a failure in the system could be devastating. A gas leakage at an oil platform, for exam-ple, must be detected and followed by a safe shutdown of the process. Otherwise, expensive assets – or even human lives – could be at risk. Simi-larly, instabilities in power transmis-sion and distribution networks should be detected before they are allowed to propagate and cause large black-outs. Economic security and personal safety depend on high-integrity sys-tems. ABB uses embedded systems in such mission-critical configurations. Special development processes and design methodologies are implement-ed to provide proven and certified high-integrity products.

Real-time properties‘Real-time’ is a term often associated with embedded systems. Because these systems are used to control or monitor real-time processes, they must be able to perform certain tasks reli-ably within a given time. But the defi-nition of ‘real-time’ varies with the application. A chemical reaction, for instance, may proceed slowly, and the temperature at a given point may need to be read no more than once per second. However, the schedule must be predictable. At the other end of the scale, protection devices for high-voltage equipment need to sam-ple currents and voltages thousands of

times per second in order to detect and, where necessary, act within a fraction of a power-cycle.

ABB utilizes its global reach to apply best prac-tices developed in one part of its organization to others to improve overall performance.

Power consumptionAt first glance, the power consump-tion of industrial electronics may appear insignificant because of the abundance of power that is available. However, this power is not always available, and the need to keep instal-lation costs low has created a demand for electrical protection devices that do not require a separate power sup-ply for the electronics; these devices are self-sufficient with respect to pow-er and meet their needs by extracting small amounts of energy from their surroundings. Wireless sensors for building-, factory- or process-automa-tion must offer years of battery life or a completely autonomous mode of operation. Self-sufficient power sup-plies can be designed to extract min-ute levels of energy from electromag-netic or solar power, temperature gradients or vibration in the environ-ment. This is frequently referred to as energy “harvesting.” Even when power is available, low-power design can be used to reduce the generation

of excessive heat that would otherwise necessitate expen-sive and error-prone cooling devices.

Lifecycle issuesYet another requirement that is frequently imposed on in-dustrial embedded systems is a long lifetime of the product itself and the lifecycle of the product family. While mod-ern consumer electronics may be expected to last for less than five years, most industri-al devices are expected to work in the field for 20 years or more. This imposes chal-lenges not only on the robust-ness of the electronics, but

also on how the product should be handled throughout its lifecycle: Hard-ware components, operating systems and development tools are constantly evolving and individual products eventually become obsolete.

Key issues in developing embedded systemsSome challenges involved in the design of embedded systems have not really changed in the last couple of decades. The drive for increased performance at reduced cost and size, for instance, will continue as long as developments in the underlying technologies will permit. Other chal-lenges involved in embedded system design are changing rapidly. Three areas should be given particular atten-tion: complexity, connectivity and usability.

ComplexityWhile the steadily increasing transistor density and speeds of integrated cir-cuits offer tremendous opportunities, these improvements also present de-velopers (individuals, teams, organiza-tions) with a huge challenge: how to handle the added complexity? A mod-ern embedded system can consist of hundreds of thousand lines of soft-ware code.

More and more products now include complex embedded systems and the development organizations must evolve with the products and their technologies. It is necessary to estab-lish suitable development processes,

Embedded system technologies

Wireless power transmission a power supply b primary coilc switches with secondary coils

a

b

c

Page 17: Abb Embedded Sym

Robot arm equipped with wireless proximity switch

17ABB Review 2/2006

Embedded system technology in ABB

methods and tools. ABB utilizes its global reach to apply best practices developed in one part of its organiza-tion to others to improve overall per-formance. Developing product plat-forms also ensures re-use of technolo-gy and increased efficiency.

The emergence of SoC has enabled extremely powerful systems to run on configurable platforms that contain all the build-ing blocks of an embed-ded system.

ConnectivityBefore the widespread deployment of digital communication, most embed-ded systems operated in a stand-alone mode. They may have had some capa-bilities for remote supervision and control, but, by and large, most func-tions were performed autonomously.

This is changing rapidly. Embedded systems are now often part of sophis-ticated distributed networks. Simple sensors with basic transmitter elec-tronics have been replaced by com-plex, intelligent field devices. As a consequence, individual products can no longer be designed in isolation; they must have common components. Communication has gone from being a small part of a system to being a significant function. Where serial peer-to-peer communication was once the only way to connect a device to a control system, field buses are now able to integrate large numbers of complex devices. The need to connect different applications within a system to information and services in field devices drives the introduction of standard ICT technologies like Ether-net and web-services.

UsabilityComplex field devices are often pro-grammable or configurable. Today’s pressure transmitters can contain several hundred parameters. The in-teraction with a device – either from a built-in panel or from a software application in the system – has be-come more complex. The task of hiding this complexity from the user through the creation of a user-friendly device has sometimes been underesti-mated. Most other requirements are easily quantifiable or absolute, but “usability” is somewhat harder to define. Yet an embedded system that is intuitive and simple to operate will reduce the cost of commissioning and maintenance. It will reduce errors and be a key factor in the overall custom-er satisfaction.

That is why usability is given a high priority in the design and develop-ment of ABB products, from the con-ceptual stage, right through to the final testing.

Embedded systems – latest trendsABB is shaping the future of power and automation through innovative products and solutions, and embed-ded systems technologies are increas-ingly important in what the company does. That is why, to stay ahead of the game, ABB must anticipate the emerging trends and opportunities.

One such trend is SoC – Systems on Chip. The emergence of SoC has enabled extremely powerful systems – including hardware and software – to run on configurable platforms that contain all the building blocks of an embedded system; microprocessors, DSPs, programmable hardware logic, memory, communication processors and display drivers, to give but a few examples.

Other trends are related to built-in wireless communication and self-con-figurable networked devices. These trends enable extended use of intelli-gent field devices in applications where wiring costs for such devices are prohibitive. ABB is at the forefront of developing technologies and appli-cations that benefit from the latest advances in research combined with technologies from other industries such as telecommunications and con-sumer electronics.

Simple sensors with basic transmitter electronics have been replaced by complex, intelligent field devices.

Exactly what power and automation systems will look like twenty years from now is impossible to predict. But whatever developments we witness, embedded systems will be key enablers and drivers for change.

Christoffer Apneseth

ABB Corporate Research, ABB AS

Billingstad, Norway

[email protected]

Communication module for radio transmission

Embedded system technologies

Page 18: Abb Embedded Sym

Thema

Embedded power protectionEmbedded applications in power system automationKornel Scherrer

Since the infancy of electrification, more than 130 years ago, protecting assets from power failures has been a main objective. New embedded information technologies incorpo-rated in power system automation are now handling protection as-pects plus many additional dedicat-ed applications. This evolution and its future trends are discussed in this overview article of power system automation, as applied to the generation, transmission and the distribution of electricity.

18 ABB Review 2/2006

Page 19: Abb Embedded Sym

19ABB Review 2/2006

Power system automation has its ori-gin in the protection of high- or

medium-voltage equipment from dam-age in case of a power system failure. This equipment includes power switch-ing devices, circuit breakers, and pow-er lines, and also motors and genera-tors. The first protective devices were developed more than 130 years ago at the time when the first electrification projects began. Protective devices at that time were built on electromechani-cal principles and their operation was mechanical. Even today, such electro-mechanical relays exist in large num-bers in many power systems world-wide. As electronics and semiconduc-tor technologies emerged, power sys-tem protection also took advantage of new possibilities and a second genera-tion of protective devices was designed with electronic components. These solid-state relays enabled new applica-tions, incorporating enhanced protec-tion functions in addition to power measurements, alarm triggering and basic trending. Eventually, when mi-croprocessors became commercially available in the early eighties, numeri-cal protection emerged. Microprocessor technology has enabled a wealth of new functionality. These embedded numerical devices now deliver key benefits in protection, control, moni-toring, and self-supervision, as well as in the field of data communication.

Power system automation business driversWhile in the past the sole purpose of a protective device was to protect high- and medium-voltage equipment, today’s power transmission and distri-bution business environment imposes new requirements that call for new solutions. Technical considerations are complemented by a great number of new challenges. Electricity market deregulation, utilities’ customer-focus, customer retention, power quality and reliability, value added service, finan-cial performance, reduced operation and maintenance cost, and asset man-agement are just a few of the chal-lenges that drive the implementation of modern automation solutions in the power delivery process. Real-time data communication is a key feature and ubiquitous access to process in-formation is key to reaping the bene-fits of advanced solutions.

Power system automation application areasPower system automation is a distinct variant of general industrial automa-tion. Due to the proximity of high- and medium-voltage equipment, pow-er system automation solutions have more stringent requirements. Com-pared to industrial automation, the key differences include higher voltage sig-naling, high current and voltage sens-ing, system time synchronization of 1ms accuracy for event time tagging, short typical response time (in the range of some milliseconds) and more stringent EMC (electromagnetic com-patibility) and EMI (electromagnetic interference) testing requirements. In the following section, some typical power system automation applications are introduced and characterized.

The number of embedded system components is increasing rapidly and these components with their varied tasks cover the entire chain of the electrical power delivery process from production to consumption. A key criterion for the characterization of an embedded system or system compo-nent is its ability to react to process events or conditions within a deter-ministic timeframe. Such real-time applications are typically executed cyclically. The cycle time determines the fastest response time and must therefore be designed specifically for the application. In general, applica-tions closer to the power process require shorter cycle times than appli-cations in remote locations such as network control centers. The title pic-ture depicts a typical power delivery

Embedded system technologies

Embedded power protection

Page 20: Abb Embedded Sym

20 ABB Review 2/2006

Embedded power protection

structure involving several automation applications with different characteris-tics. In general, the basic functionality of power system automation includes protection of power system equip-ment, control of power flow, monitor-ing of the power process and condi-tion monitoring of the equipment.

Power stationIndustrial control is the predominant automation technology at the power station. However, higher voltage de-vices such as power generators utilize power system automation devices. Typical functionality includes: Generator protection and control Functions to verify synchronous operation (Synchrocheck), ensuring proper timing when the generator is connected to the power transmis-sion network.

Circuit breaker protection and con-trol

Power system automation devices are usually integrated into the power plant automation system, enabling central control of the complete station.

Power transmission networkSubstations are usually located at each end of a power line. The most typical application in the transmission net-work is the power line protection function, which is embedded as a dedicated task in the automation sys-tem that is installed in the substation. Line differential protection is built on two electronic devices that measure voltage and current at both ends of the line. Specialized communication links transmit these measurements, which under normal operating condi-tions would show zero differences. A difference in measured quantities would indicate a fault on the power line and circuit breakers would be operated (tripped) in a matter of a few milliseconds, disconnecting the line from the transmission network. Such faults can be temporary, as in the case of a lightning strike, or per-manent, as in the case of a fallen tree. In the case of a temporary fault, auto-mation functions will reconnect the line automatically.

Another common application is the line distance protection that performs

a similar function but is based on the power line impedance rather than voltage or current differences. In the event of a line fault, the embedded device will not only disconnect the line, but it will provide some indica-tion of how far from the substation the failure is thought to have oc-curred. Automation devices in a sub-station are generally connected to a remote communication terminal or gateway, which exchanges informa-tion with the network control center.

Today’s power transmis-sion and distribution busi-ness environment impos-es new requirements that call for new solutions.

While the transmission network oper-ates at alternating current (AC), high voltage direct current (HVDC) is usu-ally employed for very long distance power transmission. Power at both ends of the line needs to be converted from AC to DC and from DC to AC by thyristor controlled converters. These circuits require highly sophisticated and very powerful control and protec-tion equipment, executing at cycle times as short as 100 ns.

Transmission substationAt the substation, large oil-insulated power transformers convert voltage levels from the transmission voltage, which might be 240 kV, to the distri-bution voltage, which might be 110 kV. Specific arrangements of cir-cuit breakers enable reliable control of the power flow. Many embedded systems are installed for automation purposes. In general, one distinguish-es between object protection func-tions, such as line protection, trans-former protection, and breaker protec-tion, and system protection functions, such as busbar protection. Short cir-cuits in the substation can become as high as 100,000 Amps, so protective devices need to react in 10 to 20 ms by disconnecting the faulting part of the station.

For dependability reasons, separate embedded devices are used for pro-

tection and control. Thus, a substation will need many dozens of automation devices, and large stations can require several hundred. The automation de-vices are modular system components, which vary in their number of process inputs and outputs, as well as in their computing power.

Primary distribution substationThe primary distribution substation performs the same functions as a transmission substation but on lower voltage levels. Smaller power trans-formers convert voltage levels from, for example, 110 kV to 38 kV. At this level, protection and control are gen-erally integrated in a single device, executing all functions concurrently. The energy involved in a fault is less critical than in a transmission system and thus, real-time response require-ments are somewhat relaxed. Howev-er, operating times are still in the range of a few tens of milliseconds.

Secondary distribution substationThe secondary distribution substation is located closer to consumers and at lower voltage levels. It may or may not include a transformer and the complete arrangement is considerably less complex than in the primary sub-station.

Sophistication in automation is also very limited and most often reduced to simple protection functions. Devic-es are standardized and available at very low cost. Most often, no commu-nication is employed at this level of the distribution network.

Distributed power generator stationThe most common application of a distributed power generator is an emergency backup power supply for critical consumers, such as hospitals, industrial applications or mission-criti-cal infrastructures. A key application in such stations is the transfer switch from the standard power source to the backup supply. Appropriate embed-ded automation functions ensure cor-rect operation of all devices involved, including the ability to disconnect the power line, start the generator, and connect the generator to the critical consumer. In case the power supply needs to remain uninterrupted, as is commonly required in information

Embedded system technologies

Page 21: Abb Embedded Sym

21ABB Review 2/2006

Embedded power protection

server farms, large batteries or fly-wheel technology would be employed to overcome the generator startup de-lay. Complete fast transfer operations can be executed within a few millisec-onds, leaving critical computer equip-ment uninterrupted.

Feeder automationThe application of protection and control devices outside the substation and on the power distribution line is called feeder automation. Typical

functionality includes overcurrent pro-tection, fault location, breaker reclos-ing. Smart and quick restoration of faulted distribution feeders are other good examples of advanced embed-ded automation functions.

Industry networkLarge power consumers such as indus-trial parks, chemical plants and facto-ries operate and maintain their own on-site power distribution network in order to power motors and other large

equipment. A great number of devices are installed to perform protection, control and measurement functions.

These power system automation de-vices are typically integrated into the overall process control system.

Network control centerThe network control center is the cen-tral location for network operation. Large supervisory control and data acquisition (SCADA) systems collect information from all substations and perform complex computations. At this level, energy management appli-cations are executed, enabling proper and stable operation of the genera-tors, transmission network and con-sumers. Complex power flow calcula-tions are performed to monitor critical conditions and enable appropriate ac-tions to be taken by network control personnel.

Due to the proximity of high- and medium-voltage equipment, power system automation solutions have more stringent require-ments.

Embedded power system automation devices perform real-time-critical functions on all levels of the system and control hierarchy. The graph in 1 classifies the applications mentioned above according to their real-time response requirement.

Technology TrendsThe future of embedded components in power system automation will be determined by three distinct technolo-gy trends:

Electronics integrationAs integrated circuit technology ad-vances, more and more functionality will be incorporated into single auto-mation devices. Because of higher CPU clock speeds and increased memory, a single embedded device will be capable of executing new and additional functionality, which will need to be processed by multiple devices, or even off-line.

Embedded system technologies

Early implementation of numerical power system protection and control devices used specialized digital sig-nal processing (DSP) units. Today’s implementations are leveraging the vast computing power available in general purpose central processing units (CPU). As such, PowerPC mi-crocontrollers deliver high comput-ing power at low power consump-tion and, therefore, low power dissi-pation. Random Access Memory (RAM) is utilized for the program execution memory and erasable read only memory (EPROM) stores pro-gram and configuration information. A typical configuration can include a 400 MHz PowerPC, 64 Mbytes of EPROM and 64 Mbytes of RAM. The CPU can be complemented with

field programmable gate arrays (FPGA ) that integrate logic and sig-nal pre-processing functionality. An automation device usually includes a number of printed circuit board assemblies (PCBA), accommodating requirements for the diversity and number of different input and out-put circuitry. High-speed serial com-munication is built in for inter-mod-ule communication that enables the CPU to send and acquire data from the input and output modules. Application-specific circuits are de-signed to optimize overall technical and economical objectives. The pic-ture below shows a sample of a high-performance CPU module, con-nected to a binary input and Ether-net communication module.

Substation automation technology

a EPROMb Signal preprocessing FPGAc Device internal 100 bit/s serial communicationd Power supply

e Multi-port Ethernet switch with optical and electrical 100Mbits/s Ethernet media access

f 18-300V binary inputsg Binary input processing ASICh RAMi PowerPC micro-controller

a

b

c

d

e

f

h

g

i

Page 22: Abb Embedded Sym

22 ABB Review 2/2006

Embedded power protection

Moreover, modern system implemen-tations are based on more generic electronics and software platforms, allowing for the most economical configuration of specific applica-tions.

Switchgear integrationEmbedded systems will also be inte-grated into the switchgear apparatus itself. Automation devices are current-ly mounted in switchgear panels and connected to the apparatus by extensive wiring.

Thus, the apparatus and its automa-tion functionality represent a compre-hensive functional unit that is also referred to as the intelligent appara-tus. Hardware engineering activities, such as drafting and wiring will be substituted by software engineering and configuration.

Integrated electronics in low-voltage equipment is already well established and state-of-the-art. In medium voltage, the first intelligent circuit breakers have been launched and market accep-tance is growing. On high-voltage levels, research is ongoing and market accep-tance still needs to be estab-lished 2 . However, what is common to all application areas is the continuous drive towards more integration.

It is only the rate of progress that dif-fers.

Data communicationThe strongest trend however, is to-wards more and higher-speed commu-nication, which in general means Industrial Ethernet implementation. The new utility industry standard, IEC61850, is fostering inter-operability on all levels of power automation systems, boosting the benefits and the acceptance of base communication technology. Future devices will in-clude integrated multi-port network functionality, such as routing and switching capabilities, as well as high-ly accurate time synchronization. Additionally, most of the commonly used protocols, such as Modbus and DNP (Distributed Network Protocol) will be extended for Ethernet net-

working, enabling the utilization of a multitude of standard protocols in a single Ethernet network.

Today’s protection and control devices have the potential to become fully capable communication network nodes with automation functionality.

Industrial control is the predominant automation technology at the power station.

Future trends in embedded power protection technologyHighly sophisticated embedded sys-tems are employed in great numbers in the electrical power delivery pro-cess at all levels. The main function of these systems is to protect the power system components, control the power flow, and monitor the pro-cess, as well as the condition of its equipment.

Power system automation devices are integrated in communication networks for the exchange of information be-tween several such devices, as well as with supervisory systems.

Technology trends predict an even higher level of functional complexity per device and also deeper integration with medium- and high-voltage appa-ratus. The need to enhance automa-tion and communication will continue to grow. To meet this demand of the future automation devices must be equipped with sophisticated data communication and networking capa-bilities.

Kornel Scherrer

Distribution Automation

ABB Management Services Ltd.

Zürich, Switzerland

1 Real-time requirements for different embedded applications in a power system hierarchy

Transmission network

Network control centre

Primarysubstations

Distributedpower

generation

Secondarysubstations

Industrynetworks

LVnetwork

1 s

100 ms

10 ms

1 ms

1 ms

Typi

cal a

pplic

atio

n cy

cle

time

Time synchronization: analog data acquisition: 1..30 µs system events: 1 ms

HVDC, 100 ns

Embedded system technologies

2 Device integration related to voltage levels

Low voltage < 1 kV

Medium voltage 1..20 kV 10..52 kV

High voltage > 70 kV

Leve

l of i

nteg

ratio

n

Page 23: Abb Embedded Sym

23ABB Review 2/2006

Direct Torque Control (DTC) is a control method that gives elec-

tronic variable speed motor control-lers (AC drives) an excellent torque response time 1 . For AC induction machines, it delivers levels of perfor-mance and responsiveness reaching the machine’s theoretical limits in terms of torque and speed control.

DTC uses a control algorithm that is implemented on a microcontroller em-bedded in the drive. The technology

was first used commercially by ABB in 1995, and rapidly became the pre-ferred control scheme for AC drives, especially for demanding or critical applications, where the quality of the control system could not be compro-mised. To understand the interplay of control theory and progress in embed-ded control, the history of DTC should be considered.

The emergence of a new technologyThe main function of a variable speed

drive (VSD) is to control the flow of energy from the mains to a process via the shaft of a motor. Two physical quantities describe the state of the shaft: torque and speed. Controlling the flow of energy depends on con-trolling these quantities.

In practice, either of these can be con-trolled and the implementation is re-ferred to as “torque control” or “speed control”. When a VSD operates in torque control mode, the load deter-

When people step onto an escalator, they don’t expect it to slow down un-der the extra load: Rather, the power output should rise to maintain a con-stant speed. In industrial applications; belts, shafts and pumps are similarly expected to maintain preset speeds or torque values, regardless of changing conditions. Such requirements are not

fulfilled through inherent properties of the motors, but through the use of motor control systems (drives).

An important criterion for such a con-trol system is its responsiveness. How long does it take to respond to and compensate for a parameter change? Progress in microprocessor technolo-

gy is not only permitting faster data throughput in such systems, but is also allowing increasingly sophisticat-ed mathematical functions to be im-plemented. ABB’s Direct Torque Con-trol (DTC) relies on powerful digital signal processors (DSP) that deliver a very fast response time and an accu-rate and responsive control system.

Drivers of changeEmbedded DSP-based motor controlIlpo Ruohonen

Block diagram of DTC

PID

Torque reference

Speed reference

Flux optimising On/Off

Flux braking On/Off

Speed controller + acceleration compensator

Actual speed

Torque reference controller

Flux reference controller

Internal torque reference

Internal flux reference

Torque comparator

Flux comparator

Actual torque

Actual Flux

Adaptive motor model

Torque status

Control signals

Flux status

Switch positions

Motor current

Optimum pulse

selector

ASIC

DC bus voltage

Mains

Rectifier

DC bus

Inverter

M 3 ~

Switch position commands

+

-

DTC Core

U

U

f

f

Page 24: Abb Embedded Sym

24 ABB Review 2/2006

Drivers of change

mines the speed. Likewise, when op-erated in speed control mode, the load determines the torque. In either case, there is a relationship between the torque, the actual current and the actual flux in the machine. The idea of DTC is that motor flux and torque are used as primary control variables. This is contrary to the way in which tradi-tional AC drives control input frequen-cy and voltage, but is similar in princi-ple to what is done with a DC drive.

Also, with traditional PWM (pulse width modulation) and flux vector drives, the voltage applied to the motor requires a modulator stage. This stage adds to the signal processing time and therefore limits the respon-siveness of the control system, and hence the torque and speed response time.

The fact that DTC does not require a modulator is one of the reasons why this control method offers such fast response times – ten times faster than can be achieved through conventional flux vector control. Moreover, DTC achieves this fast field-oriented control without the need for speed feedback: It uses advanced motor theory to calcu-late the motor torque and stator flux.

The reaction time of DTC is so fast that it opens new possibilities for variable speed control. For example, a DTC drive is ideal for protecting the mechanics from overload and load shocks.

DSP enables breakthrough Although the advantage of DTC was understood in theory, it could not be implemented until progress in embed-ded control made it possible to exe-cute the primary control cycles at a sufficiently high frequency. Conven-tional microprocessors, as used in personal computers, do not achieve a sufficiently high data throughput. It was the introduction of digital signal processors (DSPs) that made the im-plementation of DTC possible. DSPs were first developed for the telecom-

munications industry, but have now found widespread use in drive con-trol. A modern DTC drive calculates the actual torque on the motor shaft at least 40,000 times per second (ev-ery 25 µs). This provides an extremely fast reaction to load changes on the motor shaft, as well as to changes in the speed or torque reference made by the user.

Today’s drives are smaller, faster, more efficient, more reliable, and easier to use than the previous genera-tion – all thanks to prog-ress in embedded control.

The reaction time of DTC is so fast that it opens new possibilities for vari-able speed control. For example, a DTC drive is ideal for protecting the mechanics from overload and load shocks. Also, fast torque control means that sophisticated algorithms can be readily implemented for the damping of mechanical vibrations in applications where mechanical reso-nances are inherently present. Similar-ly, a DTC drive can rapidly detect a loss of load torque caused by a me-chanical failure – for example a bro-ken conveyor belt – and act to pre-vent further damage. Because of its fast time response, there are many other examples of DTC being integrat-ed into protective functions for both machine and motor.

Whereas simpler control methods – such as sensorless vector control – are typically used in low-power drives for less demanding applications, DTC is preferred in the more demanding applications that require a very fast torque response time for optimum performance. Because high power drives are significant investments, DTC is also used in all such drives from ABB, regardless of the application.

Spreading to other applicationsWith the advent of DTC there is little to improve in the control method of variable speed drives: It is no longer the frequency converter that limits the performance of a variable speed drive, but the motor itself. Research

An Ideaplast plant in Italy with a single-lineextruder (detail of the extruder‘s head and winding film rolls)

Reliable conveyor operation is essential in bakery automation (Fazer Bakery, Finland)

A pressure-boosting station (Pietersaari Finland). The drives are equipped with DTCand intelligent pump control (IPC)

Embedded system technologies

Page 25: Abb Embedded Sym

25ABB Review 2/2006

Drivers of change

has now shifted towards the applica-tion of DTC in other settings. Some exciting new developments of embed-ded drive control have been opened.

One of these is the application of DTC to permanent magnet motors. Although the principal of permanent magnet motors has been known for some time, their commercial break-through had to wait until magnetic materials were sufficiently developed.

NdFeB (neodymium iron boron) mag-nets have been available since 1987, but there have been several further improvements in the material compo-sition before the mechanical and magnetic properties of these magnets allowed them to be used in the pro-duction of motors. Since then, produc-tion techniques have steadily im-proved – today powerful permanent magnet motors are commercially viable.

The result is a line current that is practically sinusoidal and free of disturbances.

The permanent magnet motor is a synchronous motor – operating on somewhat different principles than an asynchronous motor. ABB has created

a modified version of DTC specifically for permanent magnet motors. This combination of DTC and permanent magnet motors (PM-DTC) offers sever-al benefits. Although compatible with traditional drives, the motors come in standard IEC frames and mechanical dimensions, the PM-DTC combination offers more accurate control, without the need for encoders, and high torque at low speeds. As a result it has been possible to displace gear-boxes from paper machines. PM-DTC drives can lead to substantial cost savings. Compared with traditional solutions these drives use fewer com-ponents (no gears, no couplings, no encoders), require less engineering, save space, reduce maintenance costs, have a lower noise levels, higher availability and higher energy efficien-cy. Many of these benefits can be traced back to the development of DTC and advancements in embedded control. Although paper machines where among the first applications in which PM-DTC technology was applied, other applications can be found in ship propulsion and wind turbines.

Another new application of DTC is on the front-end of the drive. With modifications, ABB has applied DTC to the supply unit that is connected to the mains and provides the invert-er unit with power. With the help of DTC it has been possible to create a drive that produces only very low harmonic distortion.

Traditional drives are supplied with mains power rectified through a pas-sive diode bridge. The problem with this method is that the diode bridge distorts the voltage in the mains. This voltage distortion can affect other equipment connected to the same grid. A very effective way to mitigate this is to use a drive with a so-called “active front-end” that uses DTC for its control. The DTC supply unit con-trols the line current and removes low harmonic distortions. High harmonic distortions are removed using a small filter. The result is a line current that is practically sinusoidal and free of disturbances.

Traditional solutions are based on in-creasing the number of pulses in the

supply unit, 12-pulse or 24-pulse in-verters, and the use of a bulky phase-shift transformer. The active front-end with DTC does not need such a trans-former and the whole package is con-siderably smaller. These examples illustrate an important trend: Progress in electronics has led to increased embedded processing power and memory in the drive. This in turn has led to the successful implementation of a superior control method: DTC. The advantages of DTC have them-selves led to new applications and new functionality. Today’s drives are smaller, faster, more efficient, more reliable, and easier to use than the previous generation – all thanks to progress in embedded control.

Ilpo Ruohonen

ABB Oy

Helsinki, Finland

[email protected]

Control systems come in two basic varieties. Closed-loop control sys-tems have encoders in the motor to report its status. This is used as feedback information for the con-trol algorithm. Open loop systems are simpler because these encoders are omitted – but at the price of a lower control accuracy. Can the ac-curacy of a closed-loop be achieved without encoders? ABB’s DTC does exactly this – it uses mathematical functions to predict the motor sta-tus. The accuracy and repeatability delivered is comparable to closed-loop systems, but with the added bonus of a higher responsiveness (up to ten times as fast).

Motor control

Modern offices are full of sensitive equipmentthat requires harmonics in the network be keptas low as possible. Low harmonic drives withDTC are ideal for this type of environment.

Embedded system technologies

Page 26: Abb Embedded Sym

26 ABB Review 2/2006

Fast control in a traction applicationTo lower operating costs while raising attractivity, modern trains are becoming increasingly light and agile. The on-board power converters must follow suite by delivering greater speed, responsiveness and reliability while fitting into a smaller footprint. Enter ABB’s CC750® power converter!

The CC750® low-voltage IGBT con-verter is at the heart of the power

circuit of the FLIRT 1 type2) [1] of Swiss Federal Railways (SBB) as well as of GTW-type vehicles for the oper-ators THURBO (Thurgau-Bodensee Bahn) and RM (Regionalverkehr Mit-

telland). All these trains are manufac-tured by Stadler Rail AG. Since their first commercial operation in Decem-ber 2003, a total of about 250 vehicles have been commissioned. Their pow-

The great reliability, speed and precision required of power converters and drives call for high performance controllers. ABB’s AC 800PEC controller is integrated into the company’s successful and widely adopted 800xA control system. The AC 800PEC is suitable for a wide range of applications – not limited strictly to power electronics control, but including fields such as rolling plants in the metals industry where it is has a part in controlling the complete process. The following two examples illustrate the successful integra-tion of the AC 800PEC1) in locomotive drives and rolling plants.

Roll and controlThe AC 800PEC control platform in a broad range of applications Armin Eichmann, Andreas Vollmer

er converters are all controlled by AC 800PEC units from ABB.

System ConfigurationThe CC750® was developed as a trac-

1 ABB’s CC750® power converters are an integral part of the FLIRT-type modern lightweight trains.

Page 27: Abb Embedded Sym

27ABB Review 2/2006

Roll and control

tion converter for use in regional and suburban electrical multiple unit trains. The CC750® has an integrated auxiliary supply and is suitable for several catenary voltage supplies in-cluding 15 kV / 16.7 Hz and 25 kV / 50 Hz. It uses IGBT (Insulated Gate Bipolar Transistor) modules with 1200 V blocking voltage in both its traction supply circuit and in the auxiliary converter.

The two converter systems are fully redundant – the vehicle can continue to operate at reduced power should one of them fail.

The main system configuration is shown in 2 . Two identical CC750® converter systems ( 2d and 2e ) are con-nected to the catenary 2a via a com-mon oil-cooled high voltage trans-former 2c . The two converter systems are fully redundant – the vehicle can continue to operate at reduced power should one of them fail.

Embedded Control SystemA decentralized concept was chosen for the control hardware 3 consisting of the following units: AC 800PEC Controller 3e , ABB’s high-end process control system. This can be programmed using MATLAB®/Simulink® and Real-Time Workshop®.

The PEBB 3b (Power Electronics Building Block) interface board, used as a universal remote I/O de-vice. This board controls and pro-tects the IGBT converters. The links to the IGBT drivers are bidirec-tional.

Combi I/O board 3c , a universal remote I/O device for high-speed traction applications.

Auxiliary modules 3a 3d , comprising power supplies and intermediate current and voltage transducers and the control of the switch and dis-connector devices.

Furthermore, the hardware arrange-ment includes AC current and DC voltage measurement (synchronous sampling), overcurrent protection and modulation and firing interlocking.

In order to ensure high tolerance to electromagnetic interference, commu-nication between the AC 800PEC Con-troller, the PEBB interface board and the Combi I/O board is assured by optical fibers. An additional optical link connects the converter control system to the higher level vehicle con-trol system via a CANopen bus. The connection to a host computer for programming and monitoring purpos-es is provided by an Ethernet link.

Since December 2003, a total of about 250 vehicles have been commissioned. Their power converters are all controlled by AC 800PEC units from ABB. Control Software of the AC 800PECHigh-speed digital control systems rep-resent the state of the art in power

Embedded system technologies

2 Traction converter arrangement on THURBO GTW with two CC750® units delivering a total of 1.1 MW of traction power.

a

b

c

d e

fg

h

i

j

l

m

n

k

a pantograph (15kV, 16 2/3 Hz cantenary)b main circuit breakerc transformerd and e CC750® power converter unitsf auxiliary transformer winding for train

heating supplyg grid inverter (390V input)h DC-link (750V)

i traction inverter (480V / 0 – 170 Hz, 750 kVA traction power)j and k asynchronous traction motorl three-phase auxiliary supply

(50 kVA / 3 x 400 Vac)m battery charger (12 kW / 36 Vdc)n brake chopper

3 Control hardware panel of CC750®

a

b c

d

e

a auxiliary moduleb PEBB interface boardc Combi I/O boardd auxiliary modulee AC 800PEC

Page 28: Abb Embedded Sym

Table 2 An impression of key data of a cold rolling mill

Maximum roll force = 30 MN Maximum mass of rolls in a stand =

40 tons Maximum mill acceleration = 2 m/s2

Maximum mill speed = 150 km/h Minimum strip thickness = 6 µm Thickness tolerance = 0.5 ... 1.0 %

28 ABB Review 2/2006

Roll and control

electronics. Typically, FPGAs (Field-Programmable Gate Arrays) using advanced VHDL (VHSIC Hardware Description Language) pro-gramming tools are used for highly time-critical func-tions in the microsecond-range and below. In the intermediate speed range (100 µs to millisecond-range) AC 800PEC provides a soft-ware layer based on MAT-LAB®/Simulink® with Real-Time Workshop® [2]. This environment allows high-level graphical programming on the conceptual abstrac-

tion level favored by control and system engineers. All cod-ing, downloading and monitor-ing functions are integrated into the platform. The engineer is spared time-consuming and error-prone low level coding.

Control systems typically consist of components with different time constants. The software therefore has sub-tasks that are executed at different intervals. In the control software for the CC750®, three software cycles have been implemented run-ning at cycle times of 1ms, 250 ms and 50 ms Table 1 .

Rolling mill in the metal industry

In the metals industry, demands on product quality and on plant produc-tivity and flexibility are steadily in-creasing. ABB’s new generation of rolling mill automation system in-cludes an integrated and advanced solution suite that meets the custom-ers’ needs for product quality and throughput. The use of ABB’s 800xA Automation Platform with the power-ful AC 800PEC controller permits mill-wide uniform automation, seamlessly integrating advanced solutions into the process control system.

In hot and cold rolling mills 4 , the demands on mill profitability, pro-

ductivity and product quality are on the rise. At the same time, mill flexi-bility has to match the growing variety of products. Strip quality and mill throughput are influenced by various factors such as mechanical design, electrical equipment, auxiliary sup-plies and control strategy, and the very many associated variables have to be tightly controlled to meet prod-uct quality targets.

An impression of key data of a cold rolling mill is given by Table 2 . To be able to control such a large and com-plex plant and meet the high demands on process speed and product quality, a powerful controller is needed to handle all required functions from low

level binary control up to advanced and sophisticated control solutions. The AC 800PEC is outstandingly well suited to meet these requirements. Beside the full integration into the 800xA Automation platform with com-munication to I/Os, Drives, various fieldbus systems and the Human Machine Interface, its strengths lie in its powerful programming capability (based on IEC 61131-3) and the CPU performance provided 5 .

The customer’s benefit is an improvement of the thickness deviation of up to 50 percent (product dependent).

The most demanding function in a rolling mill is the thickness control. Keeping the strip thickness within a narrow tolerance band is one of the

Embedded system technologies

tasks (examples) cycle time

Vehicle Control speed and torque 50 msvia CANopen instructionsAC 800PEC, Task C state machine, 1 msMATLAB®/ slow protectionSimulink® flux controllerwith Real-Time Workshop® Task B current controllers, 250 μs pantograph bounce detection Task A very fast current 50 μs controllers FPGA, VHDL modulators, ns-range very fast protection

Table 1 Software tasks and their cycle times

4 Everything under control at the rolling mill

Page 29: Abb Embedded Sym

29ABB Review 2/2006

Roll and control

of the rolling process, mechanical, electrical and hydraulic systems, in-strumentation, as well as the lubrica-tion and the control strategy must all fit together seamlessly 6 .

State-of-the-art thickness control algo-rithms are composed of single feed-forward control loops. These algo-rithms are limited in their achievable thickness performance because they

do not fully take into account the con-nection between thickness, roll posi-tion and tension [3].

By using the powerful AC 800PEC controller and its possibility of imple-menting C-Code beside the standard IEC 61131-3 program level, a new thickness control solution for cold rolling mills has been developed based on a MIMO (Multi-Input Multi-Output) control concept. The custom-er’s benefit is an improvement of the thickness deviation of up to 50 per-cent (product dependent).

Mill flexibility has to match the growing variety of products.

A powerful all-rounderThanks to the different programming levels of the AC 800PEC, this control-ler is well suited for a broad range of applications – from fast control algorithms in power electronics to process-wide control applications.

Armin Eichmann

ABB Switzerland Ltd.

Turgi, Switzerland

[email protected]

Andreas Vollmer

ABB Automation GmbH

Mannheim, Germany

[email protected]

References

[1] Peter Bruderer Stadler Rail Bussnang, Description

of FLIRT train, Railvolution 4/04 pages 58–72

[2] The Mathworks, User Manual Release 12.1, In

particular Matlab, Simulink, Real Time Workshop,

Stateflow, Stateflow Coder

[3] ABB in metals, http://www.abb.com

Footnotes1) For more background on the AC 800PEC, see also

“Design patterns” on page 62.2) FLIRT: “Flinker Leichter Innovativer Regional Trieb-

zug” or “Fast, Lightweight Innovative Regional

Train”3) Drawing is the process of forming metal sheet into

cylindrical or box-shaped parts using a punch. In

deep drawing the depth of the part is greater than

its diameter.

6 MIMO control concept with dynamic online parameter adaptation

Pass schedule, set-up and adaptation

Set pointControl

objectives

Controller design/adaption

On-line plant model

Controller parameter

Parameters

On-line estimation

MIMO controller

Dynamic disturbance FF

PIDsDynamic

decoupler

Mill

Embedded system technologies

mill’s most crucial requirements. The benchmark is set by the deep draw-ing3) of aluminum and steel sheets for cans or car body parts. The more the variation of thickness can be reduced, the smaller the minimum permissible thickness the mill can be operated at. This permits improve-ments through lower material usage, weight saving and overall cost-effi-ciency. To achieve effective control

5 Typical system configuration of the rolling mill application

OperateIT & MES Server Computer room Main control room

Thickness

MMS, TCP/IP

DriveBus

• Entry section and coil

preparation

• Exit section and

spool handling,

coil transportation

• Mill stand section and

roll change

• Hydraulic, lubrication

& roll oil

• Presetting &

recording

• Master refer-

ence control

• Coiler, mill

drive and

Deflector roll

control

• Roll force

control

• Position

control

• Thickness

control

• Tilt control

• Roll bending

and shifting

• Flatness

control

...

Flatness

Page 30: Abb Embedded Sym

Users expect — and demand — more functionality from automation systems than ever before. Embedded system components that reside within a con-trol system make much of this func-tionality possible. Advanced automa-tion solutions, such as ABB’s Extend-ed Automation System 800xA, require the integration of numerous embedded technologies to perform the wide vari-ety of productivity enhancing functions required by customers across the pro-cess industries. With plants that might be controlled remotely, and the very real need to keep production up and running around the clock for several years, process industry customers must have easy maintenance and reconfiguration options that do not interrupt production.

30 ABB Review 2/2006

Embedded systems extend automationSystem 800xA incorporates numerous embedded applications Kai Hansen, Tomas Lindström, Lars Mårtensson, Hans Thilderkvist

Page 31: Abb Embedded Sym

31ABB Review 2/2006

Embedded systems extend automation

Embedded systems are micropro-cessor-controlled computer sys-

tems that form an integral part of a larger system or piece of equipment. They are dedicated to specific tasks that contribute to the overall function-ality of the system. Depending on the nature of the system and its function, the requirements of an embedded sys-tem can differ greatly.

Embedded components in System 800xAThe embedded components that are used with System 800xA allow it to deliver many different solutions for many different requirements. These requirements can include:

Real-time execution – It is often critical that a given task is finished at a pre-dictable time, as well as being correct-

ly computed. System 800xA can meet requirements ranging from “hard” real-time, where exact timing criteria must be met, to “soft” real-time, where response is less time-critical.

Flexibility – embedded components can be dedicated to a single predefined task, or to a number of fundamentally different assignments. Compare, for ex-ample, the difference in the flexibility of an I/O (input/output) module versus that of a normal desktop PC.

Availability – Because different pro-cesses have different back-up require-ments, the redundancy level of a sys-tem must be flexible.

Cost – The acceptable unit cost for a component is often tightly linked to its required quantity. It is important to

consider whether the component will be used thousands of times in an in-stallation, or in just a single instance.

Environmental hardening – In indus-trial environments, the components, if subjected to heat, vibration and dust, must be environmentally hardened.

Distributing embedded intelligenceAs an extended automation system, the 800xA distributes intelligence and computing power to where it is most appropriate 1 . Such distribution can take the form of different types of servers, providing services to clients, and one another. On the control side, control logic can be distributed across several controllers, exchanging mea-surement and calculation values. Pre-processing can range from I/O mod-ules filtering and time-stamping data

Embedded system technologies

1 A simplified overview of a process plant built around System 800xA

System Servers

Workplaces

Remote Clients

Variable Speed Drives

Process Automation

S800 I/O

S900 I/O (Ex)

Safety

Client/Server level

Control level

Device level

Fieldbus High Speed Linking Devices (FF HSE/HI, PB DP/PA)

MCC

Process Automation and Safety

Control Network

Page 32: Abb Embedded Sym

32 ABB Review 2/2006

packets, to sensors and actuators per-forming advanced pre-processing and diagnostic functions. Input and output data from I/O buses are scanned by dedicated communication modules.

Most of the system’s components are implemented as embedded systems with a design optimized for specific needs:

I/O modules, with simple signal pro-cessing, can be implemented entirely by hardware components, with some of the logic executed in an FPGA (field-programmable gate array). More complex I/O modules, intelligent sen-sors and actuators are based on em-bedded microcontrollers that provide greater functional flexibility. Many of these use some kind of real-time operating system.

Communication modules may imple-ment a protocol stack, partly in hard-ware and partly in firmware, running on the embedded central processing unit (CPU). One way of splitting the

job may be to process acyclic messag-es with the CPU and handle cyclic messages with a direct memory access (DMA) unit, sometimes with an appli-cation-specific integrated circuit (ASIC) dedicated the task.

The Processor Module in the AC 800M uses a commercial real-time operating system and runs one of the most com-plex and flexible embedded applica-tions. Most of its functions are com-pletely defined by the user 2 .

It is often critical that a given task is finished at a predictable time, as well as being correctly computed.

Client/Server levelOn the Client/Server level, a number of software systems combine to com-prise operational functionality, eg, presenting measured values and pro-

cess status to operators. They also support engineering, commissioning and maintenance of the whole system. At this level, standard servers and PCs are built on Windows technology rather than embedded systems, but even here, special solutions are avail-able, eg, redundancy of servers and network to ensure high system avail-ability.

Controller levelThe most advanced embedded sys-tems are found at the controller level. Here, the components must be able to sustain harsh conditions such as vibration and heat. A controller should also have high flexibility, sup-porting simple functions, ranging from binary control to advanced propor-tional, integral and derivative (PID) control. ABB has a family of controller units, the most advanced of which is the AC 800M processor module 3 .

To achieve the desired flexibility of communication options, the AC 800M’s processor module has a num-ber of different communication inter-faces 4 : Two Ethernet ports allow communi-cation with the Client/Server level and other Controllers.

The ModuleBus accommodates di-rectly-connected S800 I/O modules.

The communication expansion (CEX) bus allows additional communication modules to be connected.

Embedded systems extend automation

4 The AC 800M controller mounted in a rack cabinet

Embedded system technologies

2 Extended 800xA workplace

3 The AC 800M processor module, the central unit in the controller

Redundancy Control Unit (RCU) Link Connector

Plug-in CPU unit

DIN-rail Back-plane unit

Power supply card

CPU card

Serial RS232 ports

Ethernet

Communication Expansion (CEX) bus

Page 33: Abb Embedded Sym

33ABB Review 2/2006

Embedded systems extend automation

Two RS232 ports are available for serial protocols.

A redundancy control unit (RCU) link is also available.

Eliminating moving parts, such as hard-disks and fans, ensures control unit re-liability under tough conditions. In the AC 800M processor module, program and data are stored in Flash PROM (programmable read-only memory) and RAM (random access memory), and, thanks to the energy-efficiency of the CPU, the unit is cooled by natural air-flow alone. Maintenance problems pro-hibit the use of mechanical fans.

Basing a control system’s processor module on an embedded microcon-troller reduces the number of compo-nents needed, lowering costs and power consumption. An FPGA is used for most of the additional onboard logic needed. Ethernet ports and the serial ports are implemented in the microcontroller. In addition, a number of special functions that could have been implemented in discrete hard-ware units, eg, the ModuleBus inter-face, the CEX bus and the redundancy control unit, are instead implemented as building blocks in the FPGA.

The combined abilities of the proces-sor and the real-time operating system allow the software to perform various tasks for real-time response of control loops, and timely communication with the plant operator.

The main task for the Processor Mod-ule, and therefore one with a very

high priority, is the execution of the process control logic. This is a set of calculations that defines when valves open and close, when motors start, how fast they run, etc., plus all the other actions that directly control the process. Since the calculation is based on input and output data, the process control logic is entirely dependent on the accuracy with which these data are read. The embedded system soft-ware must handle the process control logic and the I/O scanning in a way that is flexible enough to allow logic changes without losing control of the on-going production process.

Most of the system’s components are imple-mented as embedded systems with a design optimized for specific needs.

The high availability of the AC 800M is assured via redundant units for the controller CPU. The incorporation of redundancy in embedded systems is a complicated business as it requires a detailed understanding of all the dif-ferent ways in which a system could fail and a corresponding knowledge of the redundancy solutions that can handle each type of failure. Additional complications arise because some processes are more important than others, but in the AC 800M, critical failures can be detected and a back-up CPU implemented in less than 10 ms.

CommunicationSystem 800xA includes many different units that communicate via a bus or a network 5 .

The Process Automation industry em-ploys several standards for communi-cating between process controllers and peripheral units such as I/O sys-tems, intelligent sensors and actuators, and other field devices. The AC 800M controller supports a wide range of these protocols, including internation-ally standardized fieldbuses, eg, PRO-FIBUS, Foundation Fieldbus and HART, which facilitate communication with various system components, such as I/O systems, intelligent sensors and actuators.

Serial protocols, such as Modbus, and protocols that can be implemented by the user in the control logic, comprise another group of communications protocols that are supported by the AC 800M.

A third group of communications protocols supported by the AC 800M provides connectivity to other specific products, such as ABB’s motor control system INSUM, ABB’s advanced drive systems and different I/O systems using dedicated communication proto-cols.

Most of these options are implemented as dedicated communication modules that are connected on the CEX bus to the processor module. The communi-cation modules implement the proto-cols, and the exchange of process data and status, with the Processor Module through a standardized software inter-face. Data are exchanged via dual-port memory on the communication mod-ule that the processor module accesses via the CEX bus.

The requirements for real-time perfor-mance on a communication module are sometimes very complex, partly because of the large amounts of data that must be processed, and partly because the timing constraints of the protocol may be very strict. Both of these challenges justify the use of a dedicated communication module with a local, embedded CPU, instead of just adding more hardware compo-nents onto the processor module.

Embedded system technologies

5 Communication interfaces used with AC 800M

CI8

60

CI8

58

CI8

57

CI8

56

CI8

55

CI8

54

CI8

51

CI8

53

SM

810

PM

865

FF H

SE

Driv

eBus

INS

UM

S10

0 I/O

MB

300

PR

OFI

BU

S D

P

PR

OFI

BU

S D

P

RS

232

Con

trol

N

etw

ork

RS

232

Page 34: Abb Embedded Sym

34 ABB Review 2/2006

Embedded systems extend automation

Rather than using an addi-tional, dedicated module on the CEX bus, some commu-nication options are imple-mented using the Module-Bus. Certain Motor Drives can be connected directly on this bus as they use the same protocol as the S800 I/O.

HART communication for in-telligent sensors and actuators is implemented by special I/O modules that, as well as handling nor-mal process signals, handle the digital FSK (frequency shift keying) signal that is superimposed on the process signal.

Hot swapTo achieve high availability, communi-cation modules can be exchanged while the controller is running. Thus, if one communication module breaks down, it can be replaced with a spare without having to restart the controller and thereby interrupting the produc-tion process. This strategy also makes it easy to reconfigure the controller and thus change the communication options without halting the controller. Control logic and communication links that are not changed operate continu-ously during this reconfiguration. The only part of the control application affected is that that uses data from the swapped communication module.

To support this, the embedded system software that accesses the Communi-cation Modules is capable of dealing with units that suddenly fail to re-spond by configuring and restarting a healthy module.

Redundant communicationSome communication modules sup-port redundancy. Communication with units on PROFIBUS and Foundation Fieldbus HSE (High-Speed Ethernet), for example, use dual communication modules to eliminate single points of failure between the controller and the external unit.

I/O and instrumentsThe device level, which contains I/O devices and instrumentation, is found one step further down towards the pro-cess. The number of I/O units (eg, Digi-tal Input unit) in a plant is much larger than the number of controllers. Compo-

nent cost is therefore a factor to consid-er, and the reason why less advanced embedded processors are more often used here than in the controllers. A sim-ple scheduling of tasks, rather than a complete real-time operating system might also be preferable. However, real-time response is as important at this level as it is at the controller level.

Parts of the I/O system may need to be “intrinsically safe” ie, be suitable for use in hazardous environments. This can be achieved by encasing the equip-ment in an expensive housing, or, pref-erably, by using I/O units with very low power consumption such that po-tentially hazardous electrical sparks will not be generated. ABB provides a large range of I/O units for different needs, the S800 I/O system, for example 6 .

The S800 I/O comprises a substantial number of different modules of hard-ware and software solutions, each having their own specific features. For example, the hardware of the S880 safety I/O is based on an embedded microcontroller and an FPGA module. As a safety I/O module, it employs a dual solution where both the micro-controller and the FPGA execute the ModuleBus slave protocol, as well as the logic for the data input, output and diagnostics. Real-time demands on this unit are very strict. When a message is received from the control-ler, the reply must be given within 330 ms. Missing this ‘deadline’ results in the controller assuming that the I/O unit is not functioning, and continuing with the next unit. The I/O module must also handle configuration data and all possible error states.

Power supplyAnother important consideration for all embedded devices in a system with

high availability is the power supply. Units must have over- and under-voltage detection. Redundant power supplies must be carefully designed, so that they will not constitute a single point of failure.

Embedded system modules provide a high degree of flexibilityThe enormous number of em-bedded systems found in a

typical process plant provides a wide range of different hardware and soft-ware solutions. It is quite a challenge to organize these components into a single unified system, but the results are well worth the effort. As this very simplified discussion shows, different demands on different parts of a sys-tem create heterogeneous elements within a unified system. With System 800xA, ABB has brought together optimal embedded hardware and soft-ware components and integrated them to deliver a dependable system with the wide range of advanced function-ality that is needed in today’s process industries.

Top-of-the-line equipment and sys-tems, designed in consultation with end-users, will continue to improve production automation and enhance efficiency. As one of the foremost process automation companies in the world, ABB can be relied upon to provide two elements needed by every successful industry – power and productivity.

Tomas Lindström

ABB Automation Technologies AB

Västerås, Sweden

[email protected]

Lars Mårtensson

Hans Thilderkvist

ABB Automation Technologies AB,

Malmö, Sweden

lars.må[email protected]

[email protected]

Kai Hansen

ABB Corporate Research, ABB AS

Billingstad, Norway

[email protected]

6 S800 I/O

Embedded system technologies

Page 35: Abb Embedded Sym

DriveMonitorEmbedded product intelligence that enhances lifecycle management and performance in drive systemsMaciej Wnek, Michal Orkisz, Jaroslaw Nowak, Stefano Legnani

Good products offer more when they are combined with comprehensive support and maintenance packages. Optimal performance and minimal costs can be achieved through service agreements over the lifecycle of a product, but effective lifecycle management requires continuous tracking of asset history – operation, wear, damage, and maintenance. Careful monitoring of the condition and performance of assets allows the implementation of predictive maintenance programs that significantly reduce maintenance costs and the risk of failure. Without this information, performance suffers and maintenance costs rise.

ABB Medium Voltage (MV) Drives, in cooperation wih ABB Corporate Research, has developed a new customer support system – The DriveMonitorTM – a software package that allows an operator to monitor the performance of an MV drive system, collect data and store the drive’s history, all from a remote computer. The system is being tested in the Gotthard base tunnel construction site in Switzerland and offers a significant improvement in lifecycle management tools.

35ABB Review 2/2006

Page 36: Abb Embedded Sym

36 ABB Review 2/2006

DriveMonitor

The second aspect is the availability of data: From “what’s already there” to dedicated measuring systems that detect vibrations, current, corrosion etc. The third aspect relates to in-creasing levels of knowledge content and diagnostic functions: At one extreme is a simple limit threshold, at the other are advanced lifetime prediction algorithms.

In order to keep tool costs down, maintenance systems should be flexi-ble and able to accommodate a wide range of asset types. Similar assets should be treated similarly, but with individual attention dependent on their context in the system. For exam-ple, two electric motors might be iden-tical, but if one is running a ventilation fan of low importance and the other a critical fume-exhaust fan, their mainte-nance programs would be similar, but the level of investment in each would differ according to their importance.

A scalable system is not the same thing as a combination of different approaches that address different aspects of lifecycle management. To be efficient, a tool must guarantee full data interoperability, single data entry points, and unified interfacing, usage and reporting. Multiple systems can be combined in an IT integration proj-ect, but only a scalable tool can pro-vide true maintenance optimization.

In short, individual assets must be assessed to determine the level of investment that can be justified by their individual roles in a process.

A good condition assessment system is: expandable, to accommodate single or multiple asset objects

able to apply rules of varying com-plexity to the assets – vibration-based, temperature-based, electrical test-based, operation data-based, statistics- and history-based etc.

able to acquire data from various sources, eg, drive systems, control systems, vibration measuring tools, manual entries, and the asset itself.

ABB used this methodology in the development of its Asset Optimiza-tion/Asset Monitors concepts and DriveMonitorTM is a part of this truly scalable solution 1 .

ABB Drives – assets as “knowledge containers”.ABB MV Drives focuses attention on product design and development, but also on configuration and optimiza-tion in relation to customer applica-tions. A quick look “under the hood” of a drive unit will immediately show that the technological complexity of this “torque delivery plant” ranges from copper bars to electronic circuit boards. Its software ranges from

1 MV drive – an asset with a broad technology span and a rich information source

Java/.Net

Si Cu

Assembler

2 The DriveMonitorTM design principles

DriveMonitor Unit

Industrial PC

VPN

Router Firewall

Ethernet TCP/IP

Optical Fibers

ACS drivers1 ........... 5

NDBU 95

3 DriveMonitorTM – Analyzing the system’s heartbeat

Real plant systems comprise a wide variety of assets. Some are straight-

forward, simplistic even, while others are “intelligent” and capable of self-diagnosis or even self-correction. Large and critical assets often come with their own supervisory control systems, but all of the assets in a pro-cess chain are information providers – either directly, via built-in sensors, or indirectly, by reporting on other assets in the chain. All of these assets need careful monitoring.

Cost-effective data collection and processingAn efficient lifecycle management system requires scalable tools that can be adapted to the nature of an asset, its value, status, and general mainte-nance policy. The first aspect to be considered is the comprehensiveness of the system, whether it be a single asset (eg, a drive), or a whole produc-tion line, which contains many assets.

Embedded system technologies

Page 37: Abb Embedded Sym

37ABB Review 2/2006

DriveMonitor

assembler code to the newest high-level languages. To obtain the highest possible performance from such a de-vice over its entire lifetime requires some attention. However, drive units – such as the MV Drive from ABB – are represented by huge banks of data. This recorded information relates not only to the drive converter perfor-mance, but also to the driven equip-ment, and even to the whole down-stream production process. Efficient use of these drive data is the first step towards lifecycle management – ini-tially for the converter itself and ulti-mately for the whole drive-powered process.

An efficient lifecycle management system requires scalable tools that can be adapted to the nature of an asset, its value, status, and general maintenance policy.

Efficient lifecycle managementA pragmatic approach to lifecycle management issues should answer the following questions: What should be done to the asset in order to maintain the highest per-formance and the lowest costs

When should this action be taken?

Ideally the asset should be intelligent enough to provide this information to the operator. Alternatively, the intelligence can be embedded as an asset extension – intelligence that fully utilizes the data processed in the drive.

The DriveMonitorTM system is designed to meet these requirements. On the one hand it provides continuous moni-toring and analysis of the drive state and operation, supports root-cause analysis (RCA) and helps to follow predictive maintenance paths. On the other, it provides a platform upon which to offer the customer unique extension features that, by utilizing drive signals, allow the operator to visualize the whole shaft state, along with application-related KPI’s, etc. In addition, customers can rely on the ABB Support Line1) with access to experts who can remotely monitor current situations.

Intelligent, scaleable and secureThe system comprises a hardware and a software layer 2 . The hardware layer is a properly interfaced industri-al PC that is factory installed with the most powerful new ABB MV Drives (it is also available as an upgrade to existing models). The software layer automatically collects and analyses selected drive signals and parameters. The hardware is based on an industri-

al PC platform to provide the expect-ed longevity and remote accessibility. Virtual Private Network (VPN) solu-tions are used for remote access to ensure high security.

Scalability – the biggest challengeThe software layer is extremely flexi-ble with respect to the configuration of diagnostic rules, the range of the assets with which it can be used, its alarm and reporting functions, and its data intake sources. Being compatible with ABB’s Asset Monitor family, DriveMonitorTM opens the door to the whole ABB Asset Management portfo-lio, with Asset Optimizer and other Asset Monitors as optional extensions. It can be integrated easily into auto-mation systems using the ABB 800xA platform (other systems can be con-nected through OPC2) Servers). The monitor is designed for use with a single drive, and with large systems. There are possibilities for expansion to include other measurements such as corrosion, vibrations, additional temperature sensors, etc. It provides millisecond-based sampling rates with year-based scheduling, event-driven actions and alarms, and more. The various components of the system can be distributed to different computers. For instance, several monitoring units can be configured in parallel to cover larger installations and the results can be brought to a central control room PC for operator convenience.

In order to keep tool costs down, maintenance systems should be flexible and able to accommodate a wide range of asset types.

Scalability – hardware dimension.MV drive configuration can cover a broad range of products. Depending on the application, several rectifier and inverter units, each suitable for monitoring purposes, can be included in the set up. In order to acquire data

4 Extended support information facilitating root-cause analysis

Embedded system technologies

Footnotes1) ABB Suport Line is one of the service products

offered by MV Drives2) OPC-OLE for Process Control

Page 38: Abb Embedded Sym

38 ABB Review 2/2006

DriveMonitor

quickly and reliably, multiple monitor-ing units can be configured around a single unit that acts as the access point for all the data acquisition. The central computer can again be placed in a control room. Similar system solu-tions can be configured for multi-drive units.

ABB MV Drives focuses attention on product design and development, but also on configuration and optimization in relation to customer applications.

Application areaThe basic function of the DriveMoni-torTM is to “watch” the converter part of a drive shaft system 3 . It continu-ously monitors the drive status and responds when that status changes. Changes in drive status can be caused by drive faults (unexpected drive stop-pages), alarms (signals crossing thresh-old values), user-defined parameter changes, and higher level, DriveMoni-torTM-generated application-specific alarms. In this basic mode, when an event occurs, the software saves the current state and commences in-depth monitoring of relevant drive sub- systems 4 . These data are critical to

determining the root cause of an event. Without such a tool, by the time a service engineer arrives on site, this information is lost, and some tell-tale events (such as threshold alarms) may be ignored if they did not lead directly to a fault. Altogether, the insight gained from the monitor’s data will lead to quicker elimination of faults and quicker identification of failing components, which result in more up-time for the customer.

With extra diagnostic packages, Drive-MonitorTM can follow other shaft-train components such as the main circuit breaker, the transformer, and the driven machine. At the highest level, specialist packages directly related to specific application areas (such as rolling mills, water pumps, and com-pressors) can be integrated into the system. This kind of expansion can be done at any point in time depending on the customer’s needs. It is also possible to incorporate extra measure-ments that go beyond the drive sig-nals. In such cases, the DriveMoni-torTM system, which can already incor-porate data from several sources, can accommodate a number of off-the-shelf solutions. DriveMonitorTM-based diagnostic routines are valuable exten-sions to any plant-level Asset Manage-ment program like ABB Asset Optimi-zation solution.

Integrated in the bigger pictureABB’s Product Support organization ensures the efficient deployment of lifecycle management policies to drive products. Diagnostic tools such as DriveMonitorTM play a central role in the support system, but are part of an integrated approach to customer care that performs core maintenance func-tions, problem solving, spare part delivery and performance optimiza-tion.

DriveMonitorTM continu-ously monitors the drive status and responds when that status changes.

Concluding remarksDue to their complex role in industrial processes, drives generate and have access to large quantities of data. Though normally used to support a drive‘s controlling function, these data can also be used for diagnostic pur-poses. No additional measures are necessary as the data are already available. ABB’s drive monitoring solution exploits this opportunity to the benefits of its customers. The sys-tem is already being piloted at several industrial locations, including the Gotthard Tunnel construction site 5 , where a powerful ABB hoist machine has been installed, powered with a ACS6000 drive unit. The hoist machine is critical for the progress of the tun-nel since it removes the spoils from the tunnel level up to the surface through an 800-m long shaft. Drive-MonitorTM helps to optimize the machine’s performance and mainte-nance processes.

Maciej Wnek

Michal Orkisz

Jaroslaw Nowak

ABB Corporate Research

Krakow, Poland

[email protected]

[email protected]

[email protected]

Stefano Legnani

ABB MV Drives

Turgi, Switzerland

[email protected]

5 DriveMonitorTM connects experts to most remote locations, here the Gotthard Tunnel construction site in Switzerland

Embedded system technologies

Page 39: Abb Embedded Sym

The vision of ubiquitous or pervasive computing prescribes a paradigm shift where the computing power is embedded in our environment rather than concentrated in desktop or laptop machines. This broad vision of the future has fuelled a number of narrowly defined research areas, among them wireless sensor net-works.

A wireless sensor network (WSN) is a network of many, spatially dis-

tributed devices using sensors to moni-tor conditions at different locations, such as temperature, sound, vibration, pressure, motion, or pollutants. The devices are self-contained units com-prising a microcontroller, power source (usually, but not always, a battery), ra-dio transceiver, and sensor element 1 .

Because of the limitations on battery life, nodes are built with power con-servation in mind, and generally spend large amounts of time in a low-power “sleep” mode. The nodes self-organize their networks in an ad-hoc manner, rather than having a pre-pro-grammed network topology. Also, WSN have the ability to self-heal, ie, if one node goes down, the network will find new ways to route the data packets. This way, the network as a whole will survive, even if individual nodes lose power or are destroyed.

Although a hot research topic, this rather classical view of a WSN has few interesting applications. For example, some authors in the field mention forest fire detection as an application of WSN. For the definition of WSN to be more applicable in an industrial setting it must be somewhat loosened and extended.

WSN in the world of industrial automationIndustrial applications differ from the earlier definition in a number of re-spects. First, and maybe most impor-tantly, all sensors are crucial to the operation of the plant. This implies that losing one node is not an option, even if the overall network stays op-erational. A faulty node will have to be replaced.

Second, time is of the essence. Whereas a data packet in a standard WSN may spend an unknown time

traveling from its source to its destina-tion, an industrial application will fre-quently require hard bounds on the maximum delay allowed.

Finally, in contrast to a standard WSN, wireless solutions in industry tend to have a wired infrastructure. The data will emanate from the sensors and ripple through the network to some wired aggregation point. From here it will, in general, be transported over a high-speed bus to a controller. Apart from the classical mesh networking topology of WSN, there exist two

39ABB Review 2/2006

Wireless sensor networks New-breed networking solutions for industrial automationNiels Aakvaag, Jan-Erik Frey

1 Self-contained wireless sensor network device

SensorRadio

transceiver Power source

CPU/Memory

Page 40: Abb Embedded Sym

40 ABB Review 2/2006

Wireless sensor networks

mized. In addition, plug and produce configuration of the network enables deployment of temporary sensor net-works for maintenance or trouble-shooting purposes.

Applications and requirementsThe requirements of any WSN solu-tion will always depend heavily on the particular application. Two specif-ic Use Cases are considered next: dis-crete manufacturing and asset moni-toring.

The nodes self-organize their networks in an ad-hoc manner, rather than having a pre-programmed network topology.

Both Use Cases have low power re-quirements, although the actual ener-gy source may vary (energy storage in batteries, energy scavenging from am-bient energy sources, wireless energy transfer such as inductive coupling, etc). For both Use Cases, the unit can-not dissipate more than a maximum of a few milliwatts (mW) of average power.

In discrete manufacturing, the latency of the system is crucial. There is a hard limit on the maximum latency, beyond which the system will mal-function. This is typically a few tens of milliseconds. For asset monitoring, however, latency is much less critical. It obviously depends on the asset being monitored, but update times on

the order of minutes or even hours are commonplace.

Reliability is a third parameter of interest. Depending on the specific application, there are a number of ways to improve the odds of a mes-sage reaching its destination. One possible way is by increasing the redundancy. Several methods are available. The message can be trans-mitted along different paths (space diversity), on different frequencies (frequency diversity), several times on the same frequency (time diversity), or even be sent using different modu-lation schemes (modulation scheme diversity). The latter is a complex method that would be employed only when the requirements are extremely strict and cost is less of an issue.

Today, the office/consumer industry is the main driver for wireless technolo-gies with high volume applications, each with relatively short lifetime re-quirements of the devices. Industrial devices, however, tend to require a much longer lifetime. This means that special care must be taken when inte-grating wireless components into in-dustrial devices. A modular (hardware and software) design is crucial for en-abling effective, life-cycle maintenance of devices that are built on standard commercially available components.

Embedded development challengesAn embedded system can be defined in a number of ways. One good ex-ample is: “. . . a specialized computer system that is part of a larger system or machine” [1]. The operative word here is “specialized”. An embedded system has a single purpose and per-forms one unique task. When making dedicated systems, such as a WSN, it therefore has its own requirements, particular to the problem at hand.

The design of the embedded system includes aspects of both hardware and software. The two are intertwined and the optimum solution, if indeed one can be found, involves interaction between them.

Choosing the building blocksOne important aspect of WSN is to keep the power consumption of a node at a minimum, while at the same

more common topologies in industrial settings 2 . In the star topology, the most prevalent topology today, the wireless nodes communicate with a gateway device that bridges the com-munication to a wired network. An emerging common intermediate solu-tion of WSN is to have router devices (often mains-powered) communicat-ing with the gateway. The sensors only need to perform point-to-point communication with the routers and can therefore remain simple and low-power, while the range and redundan-cy of the network itself is improved.

Benefits of WSN The benefits of wireless communica-tion in industrial applications are numerous. Apart from increased reli-ability, the most cited advantage is the low cost of installation. Industrial sites are often harsh environments with stringent requirements on the type and quality of cabling. Avoiding the use of cables results in cheaper instal-lations. This is particularly true for retro-fit, where it may be difficult to engineer additional wires in an already congested site.

Even if the textbook definition is not directly applicable to industrial set-tings, WSN introduces new network-ing techniques that help to further reduce the cost of installing wireless sensors. The ad-hoc nature of WSN allows for easy setup and configura-tion, a task that should not be under-estimated once the network grows in size. To support plant-wide coverage of wireless sensors, manual network configuration work must be mini-

Embedded system technologies

2 Common topologies of wireless sensor networks

G

S

RG

GSS

S S

R R

R

S S

S

S S

R

S S

S S

S

S S

S

SS

S

S S S

R SG Gateway S Sensor Router Sensor with router

Page 41: Abb Embedded Sym

41ABB Review 2/2006

Wireless sensor networks

time provide the best possi-ble performance to the sys-tem users.

Designing for low consump-tion involves choosing low power components. This may seem trivial, but it is often a complex issue. The first pa-rameter to consider is the power consumption in nor-mal operating mode for the CPU, sensor, radio transceiv-er, and possibly other elements, such as external memory and peripherals. Choosing low power ele-ments usually involves com-promises on performance. Typically, a low power CPU runs on reduced clock cycle with fewer on-chip features than its more power-hungry counterparts. The trick is to choose elements with just enough performance to do the job.

It is important that the sleep-mode power consumption is low. It is often possible to switch off the power to the sensor and the transceiver com-pletely. However, the CPU will need some form of sleep mode from which it can be woken. Low consumption in this sleep mode is absolutely crucial for the overall power budget.

One aspect that is often overlooked is the time that the elements need to turn on and off. For example, the transceiver will need some minimum time for its oscillators to stabilize. While waiting, both the transceiver and the CPU burn power. This con-sumption needs to be minimized. The same obviously holds true when powering up the CPU and the sensor.

Finally, one must ensure that all the necessary elements can be controlled by the CPU. It is the master in the system and needs to have complete control over all functional blocks.

System issuesCommunication protocols are often fixed, having been chosen for a partic-ular purpose. Available resources should be used within the acceptable limits of the specification. No compo-nent that is not in use should be powered up. The task is consequently reduced to switching units, such as the

sensor, CPU and transceiver on and off, with the right timing. In the following, a node is consider as example that needs to wake up at regular intervals to transmit a sensor value, but only if the latter differs from the previous value by more than a predetermined margin. After the data has been sent over the radio channel, the unit waits for an ac-knowledgment message indicating that the packet has been correctly received. The required software behavior is best explained using what is known as a state diagram: a schematic representa-tion of the state the software is in, the events that may cause it to move from one state to another and the actions associated with each state transition 3 .

In the star topology, the wireless nodes communi-cate with a gateway device that bridges the communication to a wired network.

Note that in the system described, the units are powered up only when they are needed, thus minimizing power dissipation.

Protocol issuesIn addition to utilizing low power electronics and a clever sleep/wakeup scheme, the communication protocol has a vital impact on the final power consumption of the system.

Details in the communication protocol dictate the lower bounds of the con-

sumption. Some communica-tion protocols are notoriously inefficient and even the smartest embedded program-ming in the world cannot lower the consumption to an acceptable level. Others are designed to give low con-sumption without unduly compromising communication performance. The Wireless Interface to Sensors and Actu-ators (WISA)1) [2], [3] platform is one such low power proto-col. The high performance can be attributed to two fac-tors: single-hop and Time Division Multiplexing (TDM).

The former avoids delays in interme-diate nodes. The latter guarantees that a node will be alone on the channel, ie there will be no collisions.

The recently developed ZigBee speci-fication [4] with the underlying 802.15.4 protocol is more general, but will have lower communication performance. It specifies multi-hop, where a message can use several radio hops to get to its destination. Nodes do not have specific timeslots allocated, but have to contend for channel access. This lets more users access the wireless medium, but intro-duces uncertainty, as delay and the power consumption increase when a node has to wait its turn. In addition, intermediate nodes don’t know when they may be called upon to route packets for others. It is, therefore, advisable to have intermediate nodes, also known as router nodes, that are mains powered (see network topolo-gy 2 ).

In short, the WISA protocol is well adapted to the requirements of dis-crete manufacturing, provided the sin-gle-hop condition is met. Conversely, ZigBee is ideally suited for asset mon-itoring applications, assuming the router nodes have access to mains power.

Different hardware and software methods have a direct impact on the power consumption of the devices 4 . No effort has been made to quantify the various effects. These will depend on the particular WSN to be devel-oped.

3 Events and actions that cause transition of software from one state to another

EVENT_timer_wakeACTION_power_up_CPU

ACTION_power_up_sensor

EVENT_difference_largeACTION_power_down_sensor

ACTION_power_up_radioACTION_send_value

EVENT_acknowledge_OKACTION_power_down_radioACTION_power_down_CPU

EVENT_difference_smallACTION_power_down_CPU

ACTION_power_down_sensor

SLEEP WAIT_FOR_VALUE

WAIT_FOR_ACKNOWLEDGE

Embedded system technologies

Page 42: Abb Embedded Sym

42 ABB Review 2/2006

Wireless sensor networks

ModularityModular design is necessary in order to reuse elements. Yet it places restric-tions on the design and care must be taken to ensure that interfaces be-tween modules, hardware as well as software, are sufficiently general to allow portability.

One classical example of the separa-tion of modules is the split between the communication protocol and the application software. The latter is in-variably written by ABB, but the for-mer is frequently purchased from a third party. Embedding these two components onto the same microcon-troller can be difficult. Even more complex is handling new releases, bug fixes, and documentation, when the software running on the same pro-cessor has several sources. The risk of sub-optimizing is also high, ie the two software modules are optimized (with respect to power, performance, code size, etc) individually. This does not necessarily give a globally optimum solution.

Modularity can also be achieved at a lower level. The communication pro-tocol can be seen as consisting of sev-eral blocks, known as the Open Stan-dards Interface (OSI) layers. Given a healthy design procedure, one may be able to exchange a single layer with

one from a different source. Obvious-ly, the more the code is split up, the more modular it becomes. At the same time, the “sub-optimization” increases, yielding a less than perfect solution.

StandardizationThere are currently a number of initia-tives underway for standardizing WSN for industrial use. One of the best-known is the ZigBee standard, a low power, low cost, low data rate, wire-less specification that targets home appliances, toys, industrial applica-tions and the like. Recently, the Zig-Bee Alliance has started work on a profile for Industrial Plant Monitoring.

Another important initiative, the wire-less HART specification [5], aims to extend this well-known standard into the wireless domain, opening up the market to the large number of HART users. It will specify profiles and Use Cases that are directly applicable for wireless control.

A third on-going initiative is the ISA-SP100 [6]. Instead of standardizing all elements in the system, ISA-SP100 specifies only the upper levels in the stack, with a number of potential low-er level implementations.

In these early days it is hard to say which of these initiatives will prevail.

Eventually, the end customers will decide, based on performance and availability of products. The challenge is to adopt the dominant standard in an optimal manner, ie use as much as possible of the standard, while fulfill-ing the mission-critical requirements, and effectively maintaining/upgrading the implementation.

The advent of wireless sensor net-works brings many new and exciting technologies into the world of indus-trial automation. The key technologi-cal challenge is to keep power con-sumption of the sensor nodes to a minimum, while providing the best possible performance to the system users. The second challenge is to create a modular system design that allows devices to be maintained throughout their lifetimes, while ful-filling all mission-critical application requirements.

Niels Aakvaag

ABB Corporate Research

Billingstad, Norway

[email protected]

Jan-Erik Frey

ABB Automation Technologies

Västerås, Sweden

[email protected]

References

[1] Webopedia, http://www.webopedia.com/TERME/

embedded_system.html

[2] Jan-Erik Frey, Andreas Kreitz, Guntram Scheible;

“Unplugged but connected: Part 1 Redefining

wireless”, ABB Review 3/2005.

[3] Jan-Erik Frey, Jan Endresen, Andreas Kreitz,

Guntram Scheible; “Unplugged but connected:

Part 2 Wireless sensors and effectors in industrial

control”, ABB Review 4/2005.

[4] ZigBee Alliance, http://www.zigbee.org

[5] HART Communication Foundation,

http://www.hartcomm.org

[6] ISA-SP100, http://www.isa.org

Footnote1) WISA is an ABB proprietary protocol based on

standard low-cost hardware (2.4 GHz radio trans-

ceivers), but enhanced by a protocol that specifical-

ly addresses real-time factory automation at the

field device level.

Embedded system technologies

4 Hardware and software methods that directly impact power consumption of devices

SW Architecture

Com

mun

icat

ion

Pro

toco

l

Synchronization mechanism (polled, fixed time slots), modulation scheme, RF transmission technique, etc.....

Data package size (size of payload, size of header, CRC, etc)

Contention-free Media Access (eg., TDMA)

Contention-based Media Access (eg. CDMA)

Single Hop Multi Hop

Power-down of components during idle operation

Shutdown/Startup TimeControllable via

the CPUPower consumption during normal operationPower consumption during sleep mode

HW Components

Page 43: Abb Embedded Sym

The trend towards Ethernet in in-dustrial plants is partly motivated

by its high performance to cost ratio and its ability to support fiber optics, electrical cables and wireless technol-ogies in a single system. Another attraction is that Ethernet’s associated TCP/IP (transmission control protocol/internet protocol) technologies pro-vide a network infrastructure that can

be managed in a unified way. This streamlines infrastructure deployment and maintenance, providing savings on training and spare part supply.

The communications needs of the office world differ from those of in-dustry, as do the needs of the embed-ded devices in different industrial applications. One typical industrial

requirement is for real-time control responses. If communication solu-tions are involved in a control loop, response time is critical. The accept-able delay in response time is deter-mined by the physical or chemical laws governing the process under control. When controlling high voltage AC currents, for example, acceptable delays could be only a few millisec-

Industrial control systems comprise a large number of different embedded devices (eg, sensors, actuators, controllers) and a number of computers

that work together to control a physical system. Such systems can control an enormous variety of installations, including process plants,

power generation and distribution systems, car manufacturing factories and air condition systems in shopping malls. ABB sup-

plies control systems and a huge number of embedded devices designed for use in such applications. While some applications require only low-tech control, based on individual components that work in isolation, more and more customers require de-vices that are able to communicate with each other, exchang-ing information and providing operators with data and status updates on demand.

Good communication solutions are as much a part of ABB devices as their ease of use and their reliability. Customers can choose their device based on the needs of their system and assume that ABB quality and efficient communications will be provided as standard. As the market moves towards an in-creased use of Ethernet to provide for its communications needs, ABB is enhancing its range of Ethernet-compatible devices.

High performance Ethernet ABB broadens its range of Ethernet-compatible devicesKai Hansen

43ABB Review 2/2006

Page 44: Abb Embedded Sym

44 ABB Review 2/2006

High performance Ethernet communication

currently in use, the most promising of which are FF HSE, PROFINET, EtherNet/IP, Modbus TCP and certain specialized solutions intended for motion control.

The theoretical limit on throughput of data using Ethernet cables and fibers is not a serious problem in most automation applications. However, the capacity of the central processing units (CPUs) of embedded devices can form a bottleneck in the flow of com-munication on the network, and this problem must be given serious atten-tion. The efficiency of stack imple-mentation in an embedded device is the single most important issue for throughput. If the limiting factor is the processor’s ability to parse a pro-tocol, upgrading a very small CPU in a device close to the field level from 10 Mbit/s Ethernet to 1 Gbit/s Ether-net might not increase throughput at all. A bandwidth of 10 Mbit/s is normally sufficient for such a device. To provide the required efficiency in stack traversal, some of the standard protocols that are typically used with Ethernet office applications must be modified or used in combination with other protocols.

The throughput and reliability of a communica-tions system are critical factors in choosing a communications solution.

A comparison of the measured delay time for UDP/IP traffic on Windows XP operating systems running on a 2.5 GHz Pentium processor 1 shows that, even with such a fast processor, the majority of time is spent handling the message in the processor. With a 1-Gbit/s Ethernet, the network delay is very short indeed.

Real-time requirementsReal-time requirements pose a particu-lar problem for “old-fashioned” Ether-net systems that are based on coaxial cables or hubs. Such systems were equipped with collision detection such that, if two devices tried to send data simultaneously (or near simulta-neously), both data packets would be lost and each device would attempt to

resend after a quasi-random waiting time. If a number of such collisions occurred consecutively, then the delay would become significant and hard to predict. New Ethernet systems, how-ever, are based on full-duplex switch technology, in which such collisions do not occur. Each device has a dedi-cated physical line to a switch, and switches will store and forward all data packets. If the port to the next switch or device happens to be in use, the switch will put the packet in a queue and send it when the port becomes available. This technology provides real-time responses that are adequate for the vast majority of in-dustrial applications. For more de-manding applications, such as motion control, it is possible to alter the Eth-ernet low-level protocol to produce a highly deterministic time-slotted sys-tem. This can be achieved using the

onds, and in mechanical motion con-trol, the tolerance could be less than a millisecond. In chemical reactions, which tend to be much slower, a one-second delay of the actuator action might be acceptable, but meeting strict deadlines is still required be-cause once it has started, a chemical reaction will never wait. Communica-tions solutions must accommodate this range of requirements, either in a sin-gle solution, or by combining multiple technologies.

Throughput and reliabilityThe throughput and reliability of a communications system are also criti-cal factors in choosing a communica-tions solution. Again, different appli-cations have different requirements. Throughput demands can influence the real-time abilities for a system, since heavy loadings can destroy real-time responses. The physical element of a communications solution defines primary design choices. Ethernet on copper cables and optical fibers is an extremely efficient system, with very little noise and small losses due to noise. Wireless communication is less reliable and a significant number of data packets can be lost. Protocol software will ensure that lost packets are re-sent, but this reduces through-put and real-time responses. If, on the other hand, the cable or fiber is seri-ously damaged, no software will get the message through. This problem can be solved only by building physi-cal redundancy into the communica-tion interfaces in the form of a second or even third cable or fiber. However, the introduction of redundancy can lead to complication of the user inter-face.

Over the past few years, the conven-tion in the automation market has been to use Fieldbus for connecting to process equipment and Ethernet for connecting terminals, servers and con-trollers. The trend now is to extend the use of Ethernet beyond control-lers, moving it closer to the field and imposing greater demands in terms of real-time requirements, reliability and safety. This requires the provision of good Ethernet-compatible embedded solutions and standardized protocols for the communication of data on Ethernet. A number of protocols are

Embedded system technologies

Page 45: Abb Embedded Sym

45ABB Review 2/2006

High performance Ethernet communication

PROFINET IRT, EtherCAT, Ethernet POWERLINK and SERCOS III technologies.

Alternatively, the stringent real-time requirements for motion control can be met by synchronizing local clocks 2 . This can be achieved using normal Eth-ernet packets, though it does present some challeng-es for implementation. One node is designated the “time master” and it provides time information to all other nodes, where it used to use to set local clocks. The dominant stan-dards for synchronization are NTP (Network Time Protocol), SNTP (Sim-ple Network Time Protocol) and PTP (Precise Time Protocol, IEEE 1588). A number of ABB products support these standards. For example, the in-dustrial DCS controller AC800M sup-ports SNTP and the PicMaster Robot supports IEEE 1588. The main source of inaccuracy in time synchronization is the jitter in the execution of the software that timestamps the arrival of an Ethernet telegram at the node. It is important to make timestamping as fast as possible. It should occur in the first interrupt routine for Ethernet, or even earlier, ie, in the hardware be-fore the operating system of the em-bedded units start. Good software im-plementation can gain a few microsec-onds in this process, while a hardware solution might bring accuracy down to 100 nanoseconds.

The efficiency of stack implementation in an embedded device is the single most important issue for throughput.

SafetyIf the system under control poses a threat to human health or the environ-ment, governments require proof that adequate safety and emergency equip-ment is in place. Such safety control systems must comply with internation-al standards, such as IEC 61508, which is based on the SIL (safety integrated level) categories for equipment and communication. SIL 2 and SIL 3 are usually demanded in chemical, petro-

chemical and off-shore oil installa-tions, and in mechanized industries.

Ethernet systems can also be certified for safety. Since it would be impracti-cal to impose the IEC 61508 safety standard on all software and hardware involved in an Ethernet system, safety certification relies on the concept of “gray” channels. This can be, for ex-ample, TCP/IP with a process-specific layer on top 3 , creating a new layer

in the communication proto-col. This “safety layer” has very high quality implementa-tion and can discover all rele-vant errors that could occur in the gray channel. For PROFI-NET, this layer is the PRO-FIsafe layer and for EtherNet/IP, this is the CIP safety layer.

ImplementationA standard Ethernet board is adequate for some products,

but in ABB devices, Ethernet is usual-ly integrated into specially designed hardware. Ethernet-compatible pro-cessors, which may be required to work at particular temperatures or under other specialist conditions, are available from a number of suppliers, eg, PowerPCs (from Motorola or IBM), ColdFire processors, and ARM-based chips. The functional requirements will determine the choice of processor – many variations are available, with differing levels of communication support. Specialized chips are now becoming available to support the special motion control variants of Ethernet. These are either an ASIC, typically with an ARM CPU built in, or an FPGA to handle the lower level Ethernet protocols.

The futureEthernet is an important emerging trend in the industrial market. It is already supported by a number of existing ABB products but, as its im-portance grows, more of ABB embed-ded devices will be developed to support this high-performance com-munications system.

Kai Hansen

ABB AS

Billingstad, Norway

[email protected]

Reference

[1] G. Prytz, S. Johannessen. “Real-time Performance

Measurements using UDP on Windows and

Linux”, ETFA 2005.

2 Stringent real-time requirements for motion control can be solved by synchronizing local clocks

Controller Receiving sensor values with

local time stamps

Ethernet

Sensor 1 Sensor 2

Sensor element

Sensor element

Local clock Local clock

1 A measurement of the delay time for UDP/IP traffic on Windows XP operating systems running on a 2.5 GHz Pentium processor [1]

μs

Windows XP 100 Mbps

Windows XP 1 Gbps

0 25 50 75 100 125

125 μs

106 μs

Stack traversal Network delay (theoretical minimum) Interrupt related latencies

Embedded system technologies

3 The layers of a typical communication protocol

Safety application

Safety layer

Process layer

TCP layer

IP layer

Physical layer

Safety application

Gray channel

Ethernet cable/fiber

Safety layer

Process layer

TCP layer

IP layer

Physical layer

Page 46: Abb Embedded Sym

The technology permitting drives to be networked with controllers is often built directly into the drive. The benefits for the customer are simplified wiring, increased reliability and lower total installation costs.

Although fieldbuses have been around for more than ten years, recent years have seen an increase in focus on application of this technology for drives. One of the problems that has impeded the rapid adoption of fieldbus technology has been the lack of standardization. In the early days, many companies offered proprietary solutions. As these proprietary solutions limidet flexibility, many customers expressed a need for a standard fieldbus. Several competing alliances were created that all strived to develop an open fieldbus that would establish itself as standard. The result is that today there exists a plethora of standards for open fieldbuses.

Manufacturers such as ABB have responded by investing in such technology. In this context, ABB uses the concept of universal connectivity. To understand what this concept implies, a closer look should be taken at fieldbus technology.

A fieldbus is a fully digital and duplex1) data transmission system

that connects intelligent field devices and automation systems to an indus-trial plant’s network. A fieldbus re-places conventional wired I/O control. It also differs from point-to-point con-nections, which allow only two partic-ipating devices to exchange data.

A fieldbus transfers information sequentially and is often referred to as a serial communication. To make sure that two devices can communi-cate over a serial link, a protocol must be agreed on that defines the meaning of each bit in a stream of data. To facilitate the description of a serial communications protocol, engi-neers often refer to an OSI model that identifies seven layers 1 . All layers together are called the communication stack. Each layer in the stack defines a set of functions.

46 ABB Review 2/2006

Fieldbuses for drives Embedded fieldbus communicationIlpo Ruohonen

Page 47: Abb Embedded Sym

47ABB Review 2/2006

Fieldbuses for drives

The second trend is the rise of indus-trial Ethernet technology. This tech-nology is relatively new, but promises a major advance in industrial commu-nications. Ethernet applies to the low-er two layers of the protocol stack shown in 1 .

Industrial Ethernet is a relatively new develop-ment but its adoption is progressing very quickly.

Most of the functionality of a fieldbus is defined in the application layer of the protocol stack, but the lower lay-ers are important for the performance. In many systems, the control loops that are closed by the fieldbus need to be fast and need to enable synchro-nized device responses as can be found in manufacturing automation. In the past this behavior was realized by implementing a physical layer that behaved in a deterministic manner. Ethernet is in principal non-determin-istic, but today it is possible to imple-ment an Ethernet protocol with bit rates of up to 1000 Mbps. This is so fast that for most practical purposes, the control loops that can be imple-mented behave in a deterministic manner.

On top of the Physical and Data Link layers one can run the TCP/IP proto-cols that are familiar from the Inter-net. The result is fieldbus that is com-patible with the control buses that are typically used in the higher levels of the control architecture.The most obvious benefit of Ethernet is that it is based on a open standard.

It will allow a standardized network architecture using components that are widely used. Another advantage is that it enables a scalable net-work architecture. This makes it easier for customers to take advantage of future technical advances compared with pro-prietary networks. Further, many people are familiar with Internet technology, so that less training will be needed and development time can be shortened.

Industrial Ethernet is a relatively new development but its adoption is pro-gressing very quickly – it will not be long before fieldbuses build on Indus-trial Ethernet will dominate the mar-ket. This is good news for customers, because it increases their manufactur-ing flexibility.

Where do these developments lead?For ABB Drives, industrial Ethernet is another important step towards the company vision of the universal con-nectivity. Shipments of industrial Ethernet have been growing at a rate of 60 percent per year and there is no abating.

Because Ethernet is so widely used in office networks, plants and factories will enjoy the high speed, low cost, wide availability, and compatibility with office networks that Ethernet offers.

Bringing Internet technology into the drive will enable many new applica-tions. Once the drive is given an IP ad-dress many functions can be performed remotely. Diagnostics are also improv-ing. This is prerequisite to further im-provements in preventive maintenance and the resulting increase in the avail-ability of plant equipment.

Ilpo Ruohonen

ABB Oy

Helsinki, Finland

[email protected]

Footnotes1) A half-duplex channel is one that can carry informa-

tion in both directions, but not at the same time.

A full-duplex channel can carry information in both

directions simultaneously.

Rather than standardizing the complete communication stack, standards are defined for each layer in the stack, or even for a specific func-tion in a layer. The partly explains the wide array of fieldbus protocols that is available today.

The lower layers of the com-munication stack, the physi-cal layer and the data link layer, are determined by the hardware. The upper layers are implemented using software alone. This distinction helps explain how universal connectivity can be achieved and also how this concept depends on recent developments in embedded control.

Universal connectivityIn the absence of a single internation-al standard for the hardware of a pro-tocol, manufacturers have standard-ized the interface to their own equip-ment and developed adapters for the different protocols that plug into this interface. Because of continuous min-iaturization, these protocol adapters have become smaller and cheaper and are now available as options that are directly built into the drive. Some standardization of the hardware has taken place, which means that differ-ent protocols can be implemented using the same hardware solution.

Recent developments in embedded control now make it now possible to implement the upper layers of the protocol stack by simply downloading different software into the field de-vice. This combination of small adapt-ers and downloadable software makes it simple for customers to obtain a drive that easily integrates into their system. ABB supports a wide array of fieldbus protocols, permitting custom-ers to choose a drive independent of the automation system.

Trends in fieldbus technologyThe first trend is the continuous rise of fieldbus use. Today around 40 per-cent of the drives use a fieldbus for remote control. This trend is driven by the falling cost of fieldbus control, as well as by the trend towards increased automation.

1 The layers of the OSI model and their places in the protocol stack

Application

Presentation

Session

Transport

Network

Data Link

PhysicalEthernet

ProfiNet Modbus/TCP

Fieldbus HSE EtherNet/IP

TCP/UDP

IP

IEEE 802.1

IEEE 802.3

TCP/IP

Embedded system technologies

Page 48: Abb Embedded Sym

No motor needs to be an island. Whereas electric motors were once considered connected when the bus bars and drive shaft were correctly attached, network connectivity is growing in importance. Communication networks are permitting advanced control, coordination, diagnosis and maintenance planning. Drawing on progress in embedded intelligence and fieldbus technology, ABB’s MNS iS motor control center is a new advanced generation for low voltage MCC applica-tions.

48 ABB Review 2/2006

In a modern industrial environment, Intelligent Motor Control System

(IMCS) technology is considered ma-ture and is widely accepted. Follow-ing the emergence of open field bus technology more than a decade ago, intelligent switchgear has rapidly gained ground in terms of user con-fidence. Communication robustness and response time suitability make the technology reliable for real-time applications.

Additionally, the ongoing pursuit of lower equipment life-cycle costs has leveraged a platform for flexible engi-neering that provides shorter commis-sioning times, more information, better diagnosis, predictive maintenance and simplified troubleshooting all leading to less downtime.

IMCS technology is, howev-er, in the midst of a further pivotal change. Customers are seeking further produc-tivity gains through better overall plant uptime and coordination of operations and maintenance. Central to achieving this is delivery of the right information to the right people at the right time. This would not be possible without the appro-priate embedded connectivi-ty. System architecture and communication configura-

tions are adapting to better meet this demand Table .

ABB was ahead of its time in recog-nizing this trend and reflecting it in its MNS iS switchgear. Significantly, ABB’s implementation offers a scal-able approach. This means customers can add, modify or enhance their system configurations at any stage of the project life cycle.

Internal Switchgear BUS: Robust and real time communication MNS iS switchgear communication is Ethernet-based yet deterministic and real time. The ability of Ethernet for speed, robust performance, simplicity of network configuration and the ability

to communicate with several starters simultaneously are fully exploited in MNS iS. However, the stigma of Ether-net being non deterministic was careful-ly avoided by embedding a Real Time Application Interface (RTAI) providing deterministic timing and fast task switching. RTnet was adopted as Ether-net network stack. RTnet implements UDP/IP1), ICMP2) and ARP3) in a deter-ministic manner. To avoid unpredictable collisions and congestions on Ethernet, an additional layer called RTmac con-trols the media access. The need for the switchgear network (Ethernet-switch-gear bus) to be separate from the pro-cess control network (Ethernet-process control) is achieved by adopting the 10Base-I4) physical Ethernet standard.

MNS iS system configuration variabilityProcess industry MCC applica-tions require various different system configurations to meet different customer plant oper-ation philosophies or site-dependent demands on infor-mation flow. In MNS iS the customer’s external control system can access:1) Motor starters via the cen-

tral communication unit: This approach allows simultaneous access to multiple control locations on different communica-tion interfaces.

Motor medical Around-the-clock screening and protection of motor healthRajesh Tiwari

Product aspects Today’s situation Trends for tomorrow

Intelligence/ Optional Embedded (In-built by communications design)Communication Single master Multiple mastershipDCS Connectivity Meaningful integrationCommunicationConfigurations Point to point Multiple combinations yet highly optimizedInformation Too much and Pertinent and precise not in context to operator needCommunication Dedicated Scalable and possible to possibilities enhance at any stage of project life cycleCommunication Any fieldbus Specific and Ethernet based

Table Market trends for low voltage switchgears

Page 49: Abb Embedded Sym

49ABB Review 2/2006

Motor medical

be taken to the electrical maintenance system or SCADA package. With this approach the information required can be presented as required and where it is needed. There is no more need for “data routing programming” in the PLC’s engendering huge engi-neering efforts or suboptimal PLC per-formance – no application programs needs to be implemented for data transfer.

MNS iS all about information flow to the right operator at the right time.

Rajesh Tiwari

ABB Switzerland Ltd

Zurich, Switzerland

[email protected]

Footnotes1) UDP (User Data Protocol) is a protocol on the trans-

port level of the communication stack (see also the

figure on page 47). It is faster than TCP but does

not offer the same level of determinism or guaran-

tee that packets are received in he order they are

sent.2) ICMP (Internet Control Message Protocol) is a pro-

tocol on the network level of the communication

stack (like IP). Its most frequent use is for sending

error messages.3) ARP (Address Resolution Protocol) is another

network level protocol. It resolves the hardware

address of a device from its protocol address. 4) 10Base-I is the physical layer of 10 Mbps Industrial

Ethernet.5) Profinet implementation is in the pipeline.6) DCS: Distributed Control System7) PLC: Programmable Logic Controller

2) Alternatively, direct fieldbus con-nection5) to the individual motor starter level: This approach allows a single control station to access a specific motor starter.

The best of both the worlds can be achieved by combining the approach-es 1 and 2. To add to customer confi-dence for higher plant availability redundant configuration is also sup-ported.

MNS iS OPC Server capabilitiesThe OPC (OLE – Object Linking and Embedding – for Process Control) interface used in MNS iS worth men-tioning. OPC is a standardized way of handling additional information that is not mission critical, but nevertheless important for successful plant opera-

The MNS iS switchgear bus is built in. All MNS iS components on the switchgear bus are pluggable. Cus-tomers are relieved from the pains of wiring.

MNS iS offers complete communi-cation integrity with predictable behavior. The operational safety of motor is ensured against:

Breakdown in communication- MNS iS continuously monitors the communication integrity from the motor starter to the external control system (DCS: Distributed Control System) at all times. Should com-munication break down, the motor is led to a pre-defined safe state.

Unauthorized Motor Control- the MNS iS motor starter unit can be accessed from multiple control

stations. Operational safety and integrity is safeguarded and unau-thorized or unintended control op-erations prevented by controlled user access right mechanism.

MNS iS provides: DCS Communication on Industry Standard open field bus Profibus DP-V1, Modbus TCP and OPC Interface (Profinet implementation in pipeline)

Web Browser connectivity for touch panel local HMI (Human Machine Interface)

Direct field bus connectivity to motor starters on Profibus DP-V1, Device Net or Modbus RTU.*

* development pipeline

Advantages at a glance

tion and maintenance. Using OPC, customers can connect to operator stations, maintenance systems etc. directly without having to program DCS6) or PLCs7).

MNS iS all about informa-tion flow to the right operator at the right time.By using OPC Servers provided within the scope of MNS iS, additional infor-mation for operators can be added to the faceplates without routing to DCS/PLC controllers. Alarm and event han-dling is totally automated so operator stations obtain motor-starter relevant alarms and time-tagged events directly from MNS iS. Alternatively, only the maintenance relevant information can

Motor control center

MNS iS offers great ease in handling

Embedded system technologies

Page 50: Abb Embedded Sym

50 ABB Review 2/2006

An efficient communication net-work is the backbone of modern

power systems 1 . Power utility opera-tors communicate with each other to coordinate actions and exchange all kinds of operational information. The communication network conveys sig-nals for the remote control of un-manned stations, transferring data and load values from sites across the pow-er system to central control, and trans-mitting central control commands to sites. Most crucially, the communica-tion network carries many of the vital signals that are exchanged in real time between different locations to ensure optimum control and protection of the power system. In short, communica-tion networks help power utilities keep electricity flowing all the way from generator to consumer.

Traditional utility communication sys-tems were predominantly hardware-based modules, tailored to customer specifications. Today’s embedded sys-tems, such as ABB’s ETL600 Power Line Carrier (PLC) system, are based on a powerful, flexible hardware plat-form and a number of versatile soft-ware modules. This technology allows the configuration of a complex system “with a few mouse clicks”, and even to extend functionality in the future with the download of additional soft-ware modules.

Makingpower lines sing Communication keeps the power flowingStefan Ramseier, Hermann Spiess

The safe and reliable transmission of power depends on continu-ous coordination between different points in the network. From a simple telephone conversation between operators to the auto-mated control and monitoring of remote equipment, a robust and dependable communications infrastructure is a prerequisite for efficient operation. Power network operators use a broad range of communication channels – including their own power lines.

ABB draws on 64 years of experience of data transmission on power lines. The company’s latest product, the ETL600, breaks new ground in offering extensive functionality. It is easy to con-figure (taking just a couple of mouse clicks to set up) and to upgrade (through the straight-forward installation of new soft-ware) – thereby ensuring that the customer remains at the forefront of technology for years to come.

Page 51: Abb Embedded Sym

51ABB Review 2/2006

Making power lines sing

not restricted to power line routes, it can, in certain circumstances, offer an advantageous alternative, especially in difficult terrain (on mountains and islands, for instance).

Power lines are used not only to carry electrical power, but also to transmit communication signals.

Typical applications supported by utility communication systems include Local Area Network (LAN) intercon-nections, video surveillance, remote diagnostics and support, distribution automation, automatic meter reading and standard telecom services.

The main applications for “operational communications” are power system control, power line protection and operational telephone services. As the first two are most crucial to the opera-tion of the power system, they are explained in some detail here.

The availability of electric power is largely dependent on the reliability of the power control system. Accordingly, control systems, and in particular the associated communications equipment, must function reliably under worst-case operating conditions. Typical power system control applications in-clude telecontrol (supervisory control and data acquisition or SCADA) and Energy Management Systems (EMS).

Teleprotection equipment, operating in conjunction with line protection, must

be capable of reliably trans-mitting a signal to the remote end of the line, in the shortest possible time and under ex-treme interference conditions that might be caused by a fault in the power system. On the other hand, interference on the communications channel must never cause unwanted operation of the protection, for example, by simulating a tripping or blocking signal at the receiving end when no such signal has been transmit-ted at the sending end.

Power Line Carrier Power Line Carrier Systems have long been used by electric power utilities for the transmission of vital informa-tion for the operation and protection of the electric power grid, ie, voice, protection commands and control sig-nals. Thus, power lines are used not only to carry electrical power (at 50 Hz or 60 Hz), but also to transmit communication signals (typically at frequencies between 40 kHz and 500 kHz). Special coupling devices are used to connect the communication terminals to high-voltage power lines.

The use of existing power lines for communications is a meaningful choice, because these provide the most direct link for teleprotection (where speed is crucial), they are reliable and they are completely under the control of the power utility, which is impor-tant, especially in countries with de-regulated telecommunication markets. Furthermore, power lines are an ex-cellent communication medium that can bridge very long distances (sever-al hundred kilometers) without a repeater.

From valves to embedded systemsThe first PLC link was put in opera-tion by ABB in 1942 3 , and in the past 64 years, thousands of links have been installed in more than 120 coun-tries, at voltage levels of up to 1100kV AC and 500 kV DC, covering a total length of more than one million kilo-meters.

Over more than six decades, each new generation of PLC equipment has been developed using the cutting

What and how do electric utilities communicate?ABB‘s utility-communica-tions expertise is founded on experience gained from installations in electrical power utilities in over 140 countries. This experience, allied to proven solutions, is especially important in pro-tection signaling where com-munication enables protec-tion systems to clear a line fault in the shortest possible time, or to isolate primary plant components directly affected by a fault, while maintaining the availability of all other components.

The enhanced functionality and per-formance of ABB’s communication systems increase both the quantity and quality of information available for operational and management func-tions. Enabling all business units with-in a power utility to have ready access to this information means the same in-formation can be used for the remote control of substations and for evalua-tion purposes, minimizing operation and maintenance costs. For modern power utilities, powerful and reliable communications services are absolute-ly vital for the control, supervision and administration of power system operations 2 .

Rapid developments in technology in recent years, together with the con-tinuing deregulation of power mar-kets, have significantly changed the communications requirements of pow-er utilities. There are three major com-munication technologies used in the Wide-Area-Network (WAN) to meet these requirements: PLC, optical fiber and microwave radio.

Established PLC techniques play an important role owing to their high reliability, relatively low cost and long distance reach. For higher transmis-sion capacities, broadband systems based on optical fibers can handle both operational and administrative power utility data, and – depending on a utility’s strategy and on legal reg-ulations – even provide commercial telecommunications services. Because microwave radio communication is

1 Ready to talk? An overview of a communication network

Embedded system technologies

Page 52: Abb Embedded Sym

52 ABB Review 2/2006

Making power lines sing

edge technology of its day,as indeed is still happening today. Hence, many of the technological breakthroughs in electronics and telecommunications of the last few decades are reflected in the development of PLC equipment.

The first PLC systems used valves, and information was transmitted much as it is in today’s AM radio systems: Ana-log waveforms (no digital signals or bits) are modulated to the desired frequency (eg between 40 kHz and 500 kHz). The signal on the power lines appears twice – there is a mir-rored copy of the original (double side band). In the early fifties, the required frequency band – a very

scarce resource – was reduced by a factor of two, eliminating the mirrored signal (single side band, SSB). This SSB technology is still used in today’s systems, and also in some short-wave radio systems. In the mid-fifties, valves were replaced, first by germa-nium transistors, then in the early six-ties by silicon transistors, and, most recently, in the mid-seventies by inte-grated circuits. In the early nineties, it became possible for the user to tailor the PLC system to actual needs by “programming” it with switches and jumpers.

The next technological breakthrough came in the late nineties with the in-

troduction of ABB’s ETL500, the first embedded numeric PLC system. The system was no longer configured only by switches and jumpers, but mainly with a graphical user interface (GUI) running on a personal computer (PC). The signals inside the ETL500 system were no longer processed in analog waveforms, but in digital bit streams. Many of the complex analog compo-nents, such as oscillators, mixers and filters, were replaced by mathematical operations executed inside a Digital Signal Processor (DSP). Such a DSP (similar to a processor inside a PC, but designed for specific “number-crunching” applications) can perform complex operations at blazing speed.

The first PLC systems used valves, and informa-tion was transmitted much as it is in today’s AM radio systems: analog waveforms.

Another technology leap was made possible by the pioneering work in digital modulation and coding. Digital communication is now part of every-day life, be it in cellular phones, fax machines, CDs, DVDs, digital satellite or terrestrial TV and radio broadcast or MP3 players, to name but a few. In order to visualize the way in which technical advances have altered the conditions of daily life, consider how telephone lines were and are used to carry digital information with the help of so-called modems. Initially, a tech-nology called Frequency Shift Keying (FSK) was used and, in 1962, a data rate of 300 bit/s was achieved (later standardized as V.21). More than 30 years later, that speed had in-creased by more than two orders of magnitude to 56 kbit/s (V.90/V.92)! With ADSL, even higher data rates are possible, albeit requiring a much larg-er bandwidth (not otherwise used on telephone subscriber lines).

Similar progress was possible with PLC systems. Modulation and coding principles had, however, to be adapt-ed to cope with the scarce spectral bandwidth resource and difficult channel conditions of PLC systems.

Embedded system technologies

3 One of ABB’s first PLC installations, circa 1944, published in Brown Boveri Mitteilungen, the predecessor of ABB Review, in Jan/Feb 1944 (Abb. 169 & 170).

2 PLC system overview

Transport of Electrical Energy

Transmission of Data, Speech and Protection Signals

HV lineimpedance ZSubstation Line trap Line trap Substation

Coupling capacitor or CVT

Coupling capacitor or CVT

Coupling deviceMCD 80

Coupling deviceMCD 80

PLCterminal

PLCterminal

Page 53: Abb Embedded Sym

53ABB Review 2/2006

Making power lines sing

Then there was the additional hurdle of huge distances that needed to be overcome. In 1999, ABB introduced the world’s first digital PLC system with automatic speed adaptation (AMX500), allowing a data rate of up to 28.8 kbit/s in a 4-kHz bandwidth, or up to 64 kbit/s in 8 kHz. Again, this is an improvement of several orders of magnitude.

ETL600: A flexible and future-proof embedded PLC system In recent years, technology advances presented new opportunities for PLC applications, particularly those related to higher bandwidth provisioning, in-tegration into digital networks, com-bined with functional enhancements, and ease and flexibility of use. These new possibilities – taken together with the economy and reliability consider-ations for which PLC is known – have led to a remarkable revival of PLC systems worldwide.

The latest generation of ABB’s PLC equipment, ETL600 4 , is a truly em-bedded system that integrates and ex-tends many components of its prede-cessor in a most flexible way. With this new, integrated, multi-service platform it is possible to integrate all PLC applications in one single system.

The ETL600 system architecture is based on a combination of proven technology with cutting edge hard-ware and software for digital signal processing. This allows the user to configure the system with a few mouse clicks, where previously inte-gration of additional hardware mod-ules required programming with

jumpers and switches, or even solder-ing. In addition to user-friendliness and unprecedented application flexi-bility, ETL600 also guarantees uncon-ditional compatibility with legacy as well as state-of-the-art digital telecom-munication environments. ABB’s ETL600 provides data transmission speeds four-times that of other systems currently available on the market.

The novel high-speed operating mode of ETL600 paves the way for providing Ethernet/IP connectivity over high-voltage power lines.

In order to provide security and reli-ability, ETL600 incorporates extra measures for high availability and pro-tection against electromagnetic inter-ference and damage due to over-volt-age stress. Besides complying with all relevant EMC/EMI1) standards, all in-terfaces, including data ports, are electrically isolated; hence providing additional protection against over-voltages, ground potential rise and ground loops. The ETL600 also pro-vides improved reliability through built-in self-test functions and support for easy commissioning and mainte-nance.

Looking aheadEach new technology leap offers fast-er and better ways of executing rou-tine tasks. More importantly, it also opens the door to a wide variety of new applications. Traditional PLC systems were basically point-to-point links, enabled for point-multipoint connectivity through upper-layer pro-tocols of SCADA systems. With the in-troduction of digital PLC and digital multiplexers, switches or routers, mul-tiple PLC links can now be intercon-nected to form a meshed network. Such a network provides a high de-gree of resilience against link failures and supports new applications, such as wide-area monitoring, control and protection. Furthermore, voice signals, which are today still largely transmit-ted as analog signals, can be convert-ed into digital bit streams, which con-

sume less of the precious bandwidth on power lines.

The new features of digital PLC tech-nology permit the use of modern PLC systems as a reliable backup of mis-sion-critical services like SCADA and teleprotection that are normally con-veyed over broadband media. In par-ticular, the novel high-speed operating mode of ETL600 paves the way for providing Ethernet/IP connectivity (eg for LAN-LAN interconnections) over high-voltage power lines – an applica-tion that was unthinkable with tradi-tional PLC technology.

Because of the flexible and future-proof architecture of embedded sys-tems, additional functionality can later be introduced with new software releases, without needing to replace the hardware.

Although this article focuses on PLC, impressive technological advances have been made in the entire utility communication portfolio, particularly in fiber optics and microwave radio. ABB offers integrated communication solutions for mission-critical applica-tions for electric utilities, the oil and gas industry, and railways. Thanks to the latest developments, it is now possible to use a single network system for the remote management of an entire communication network.

For more information see:

http://www.abb.com/utilitycommunications

Stefan Ramseier

Hermann Spiess

ABB Utility Communication Systems

Baden, Switzerland

[email protected]

[email protected]

Footnote1) Electromagnetic compatibility (EMC) is the ability of

equipment to operate without interfering with other

devices. Electromagnetic interference (EMI) focuses

on the amount of energy that emanates from elec-

tronic equipment that can cause performance deg-

radation in nearby equipment.

4 A PLC frontrunner, the ETL600

Embedded system technologies

Page 54: Abb Embedded Sym

For many people, product development still has much to do with lonely inventors toiling away in dimly lit garages. Such clichés surround individuals ranging from Thomas Edison to recent microcomputer whizkids. For most prod-ucts, however, these methods no longer present an efficient way of developing for an increasingly demanding and dynamic market.

Development has shifted from an intuitive and empiric method where a single genius could take on the world, to a

Bright ideas Making global cooperation deliver the best productsDeia Bayoumi, Katja Rajaniemi, Eric Buchholtz

scientifically managed creation process. The impressive store of tools supporting it ranges from market analysis through risk management to the Theory of Constraints1). Numerous stakeholders are involved in the development process, spanning different views, ideas, priorities and cultures. Successful project management is about making all these groups work towards a single goal.

In this article, ABB Review looks at the development process in ABB’s Distribution Automation business.

54 ABB Review 2/2006

Page 55: Abb Embedded Sym

55ABB Review 2/2006

Bright ideas

ABB Product development organi-zations must deliver products

with innovative functionalities that meet or exceed the customers’ expec-tations. Such a product should have state of the art technology, a competi-tive price, must be easy to use, and must maintain highest levels of quality and reliability. The invention, delivery and support process presents many challenges for the company’s develop-ment organizations. The implementa-tion of common processes between R&D centers located in different parts of the world increases collaboration between these, and so raises the effi-ciency of product development. Areas of focus include project, configuration and requirements management.

ABB’s Distribution Automation (DA) business manufactures products for protection, control and monitoring in power distribution. The development organization has units in six different countries, collaborating on multiple parallel projects. Opportunities are created through the different cultures, maturity levels and processes of the local organizations. ABB is focused on creating very efficient processes to better and more fully meet its custom-er specifications and expectations while strengthening its position as global leader.

In general, there are three major op-portunities for improvement in global product development: knowledge transfer, coordination, and coopera-tion. The sharing of knowledge and the transformation of an individual’s knowledge into organizational knowl-edge is an essential ingredient to suc-cess. Lack of cooperation and coordi-nation is often caused by differing in-terests or goals, undefined roles, poor personal relationships or unfamiliar processes [1].

ABB Distribution Automation launched a project to improve process devel-opment in terms of greater quality, reliability, scalability, predictability and customer focus while reducing time to market. The areas of focus included: Knowledge Transfer: Increasing the communication between the busi-ness units by providing an environ-ment that increases the ability, de-

sire and skills to listen and share information.

Co-ordination: Defining goals and responsibility through implementing and developing a common process essential to achieving better, faster and competitive products.

Co-operation: Ensuring that all rele-vant stakeholders are involved in the process, aware of the status, risks, and issues, and committed to the defined goals and plans.

The development of a good requirements management structure is perhaps the most important of the develop-ment practices in the creation of a new product.

Process ImprovementWhen the process improvement plan was embarked upon, two different models were adopted: the Capability Maturity Model Integration (CMMI) of the Software Engineering Institute (SEI), and the IDEAL (Initiating, Diag-nosing, Establishing, Acting, and Le-veraging) model. Both models are fre-quently used to assist in the setting of process improvement objectives and priorities and provide guidance for

ensuring stable, capable, and mature processes.

The CMMI is a reference model 1 of mature practices in specified disci-plines within product development; it is used to assess a group’s capability to perform that discipline. Practices identified in the CMMI address pro-ductivity, performance, costs, and stakeholder satisfaction. Its strength lies in its integration of multiple sys-tems and software disciplines into one process improvement framework Textbox .

The IDEAL model 2 is used to guide the development of a long-range, inte-grated plan for initiating and manag-ing a process improvement program.

Initiation phaseSenior management identified the objectives and secured the commit-ment for process improvement within the organization. Based on the busi-ness objectives, an appraisal was con-ducted to identify the strengths and weakness of the existing development organizations.

Based on the findings, a plan was developed to define the projects for correcting the identified weaknesses. Teams were formed to implement the plans and define the new processes to

Embedded system technologies

Adoption of CMMI (Capability Matu-rity Model Integration) aids organiza-tions in achieving: a higher level of confidence of delivery of promised scope, cost, and schedule

collaboration with stakeholders to meet or exceed their expectations

competitive world-class products and services

an integrated enterprise from the business and engineering perspec-tive

proactive program management techniques

use of best practices to cope with development challenges such as changes in technology, customer requirements and markets environ-ments

optimize resources where the de-veloper works on multiple, differ-ent projects while using the same or similar processes

The resulting benefits include: commitment: understanding who the stakeholders are and achieving common understanding of the project‘s scope, time, and budget.

control: a measurement-focused process offering proactive controls throughout the program where the requirements are a fundamental basis for planning and control, and risk management is explicitly used throughout the project.

Communication: Enhancing knowl-edge sharing by building an inte-grated project team.

CMMI

Page 56: Abb Embedded Sym

56 ABB Review 2/2006

Bright ideas

a common understanding of it. The business focus of the process was strengthened through the separation of the business aspects of the market requirement specification into a new document – the “Product Business Plan”. This document links the strate-gy, product portfolio management and requirements. It defines the competi-tive environment of the product, describing the strategic aspects of why the product is needed.

To enhance the quality of the require-ment management process, multiple phase assessments and review meet-ings are held to determine the readi-ness and quality of the requirement specification. Checklists are enhanced to ensure the requirement specification contains the appropriate information and is reviewed when appropriate. Project ManagementABB Distribution Automation focused its project management resources on

two areas within project management: risk management and project estima-tion. Risk management identifies po-tential problems throughout the life of the product or project so that the ap-propriate risk-handling activities can be planned and implemented. Product development always engenders signif-icant risks. ABB DA implemented a four-phase risk management process which is iteratively looped through the project lifecycle. These phases are: identifying and classifying the risks, analyzing the risks, responding to the risks and monitoring them.

The more effort is put into identifying risks and evaluating their impact, the more accurate the project’s estimates will be. ABB DA initiated workshops to identify, analyze and classify risks associated with the project. In these, product management arranges the workshops, where it concludes the findings and discusses the market risks associated with different existing and

be approved, trained and adopted within the organization.

Improvement activitiesThere are currently ongoing process development activities in several areas. The three most significant of these are: Requirements Management, Project Management, and Configura-tion Control.

Requirements ManagementTypical obstacles to the management of requirements are posed by situa-tions where such requirements are changing or unclear. This results in in-correct facts, omissions, inconsistency and ambiguity [2]. In global develop-ment environments, the challenges identified relate to coordination and communication and can typically result in cost overruns, schedule slips, frus-trated and overworked employees, un-happy customers, and lost profitability.

The more effort is put into identifying risks and evalu-ating their impact, the more accurate the proj-ect’s estimates will be.

The development of a good require-ments management structure is per-haps the most important of the devel-opment practices in the creation of a new product. Typically the most signif-icant potential for improvement lies in: Business focus on acquiring correct data and establishing a good under-standing of customer and market needs.

Communication between different functions and across different loca-tions and cultures.

Consistency of the requirements specification.

A new requirement management system was deployed. This allows all relevant stakeholders to input their requirements and easily review those of other stakeholders. Requirement reviews are held to agree on the scope, priorities and rationale of every requirement. The participation of peo-ple from different development cen-ters (ie, sales, marketing, production and support) in the reviewing enables

1 The CMMI (Capability Maturity Model Integration) model

5Continuous

Process Improvement

Organizational Innovation

and Deployment

Causal Analysis and ResolutionOpt

imiz

ing

4 Quantitative Management

Quantitative Process Management

Software Quality Management

Qua

ntat

ivel

y M

anag

ed

3 Process Standardization

Requirements Development

Technical Solution

Product Integration

Verification

Validation

Organizational Process Definition

Organizational Training

Integrated Supplier Management

Decision Analysis and Resolution

Organizational Environment for

Integration

Def

ined

2 Basic Project Management

Requirements Management

Project Planning

Project Monitoring and Control

Measurement and Analysis

Supplier Agreement Management

Process and Product Quality

Assurance

Configuration Management

Man

aged

1 HeroicEfforts

Design

Develop

Integrate

Test

Initi

al

Productivityand

Quality

Riskand

Waste

Embedded system technologies

ResultLevel Capacity

Page 57: Abb Embedded Sym

57ABB Review 2/2006

Bright ideas

monthly meetings of the project steer-ing committee. In addition, risks are reported weekly to management. This weekly project reporting is used for the sharing of knowledge between different stakeholders, and communi-cating this knowledge to a wider group of people.

The project management assures that each project is allocated the resources it requires, as if it were in a single project environment. ABB DA uses Theory of Constrains techniques (TOC)1), among many others, to en-sure efficient project planning, moni-toring and control in the multi-project environment. When using TOC, proj-ect plans are created based on the optimistic and pessimistic estimation of each task. Critical chain2) and proj-ect buffers are created based on those estimates. At the beginning of a proj-ect, it is scheduled according to the availability of critical resources. Fur-thermore, critical resources only work on tasks where their unique ability is required – so optimizing overall per-formance of the process. To ease scheduling, each project is given a relative priority. A project with higher priority is allocated resources in pref-erence to one with lower priority. Consumption of the risk buffer and progress on the critical chain is moni-tored and reported weekly. Project managers futher collect weekly infor-

planned products: how can the estab-lished market share within a key mar-ket be maintained or improved, and what are the estimated costs associat-ed with each option. When a project is started, risk identification and analysis is more focused on technical risks, and considering whether the project can be managed according to the planned time schedule, cost and scope.

The structure of the risk workshops are Brainstorming risk sources Applying a list of potential risk sources and organization specific risks identified on the basis of lessons learnt

Prioritizing the risks (probability and severity) using risk defined categories

The idea of using risk severity catego-ries and risk sources assures that, in addition to the most probable risks coupled with small consequences, other kinds of risks are also identified, and it also avoids that risks with con-siderable consequences, but which have never yet occurred, remain unidentified. Prioritization of the risk also identifies those risks that require contingency and/or mitigation plans to reduce the impact or probability of the risk becoming a problem.

Risks are monitored and reported dur-ing phase assessments and also at

mation regarding the work remaining for each task. These activities provide the information needed for managing the whole process.

Configuration ControlAs was done in requirements man-agement, a global configuration man-agement system was deployed with a lifecycle management system. The benefits of this system are: Communication: allows the sharing information across different func-tions, locations and cultures by making the information available to all stakeholders

Control: Ensures that everyone is working from the same version of the document

Commitment: Requires that relevant stakeholders agree by approval of the documents

Quality: Forces the use of reviews to ensure that the work products are complete and accurate

Knowledge sharing: Use of a system service and an information pool

Only a properly understood and im-plemented development process can satisfy the demands of tomorrow’s market!

Deia Bayoumi

ABB Inc.

Allentown, PA, USA

[email protected]

Katja Rajaniemi

ABB Oy

Vassa, Finland

[email protected]

Eric Buchholtz

ABB Inc.

Raleigh, NC, USA

[email protected]

References

[1] Smith, 1995, Surakka, 2005, Hoopes, Postrel,

1999

[2] Hooks, Farry, 2001

Footnotes1) For more background on TOC, see also “How to

control the chain with TOC”, ABB Review 1/2006,

page 25.2) The Critical Path is the sequence of work packages

in a process with the longest overall duration,

taking resource dependencies into account.

2 The IDEAL (Initiating, Diagnosing, Establishing, Acting, and Leveraging) model

Learning

Acting

Establishing

Diagnosing

Initiating

Stimulus for

Change

Set

Context

Build

Sponsor-

ship

Charter

Infra-

structure

CharacterizeCurrent & Desired States

Develop

Recommen-

dations

Set

Priorities Develop

Approach

Plan

Actions

Create

Solution

Pilot Test

Solution

Refine

Solution

Implement

Solution

Analyze

and

Validate

Propose

Future

Actions

Embedded system technologies

Page 58: Abb Embedded Sym

Do-it-yourself robotics Embedded software allows users to program their own robot applicationIngela Brorsson, Ralph Sjöberg, Anna Liberg

58 ABB Review 2/2006

The year 2005 saw more orders being placed for ABB robots than ever before. The IRC5, ABB’s 5th generation robot controller, established itself on the world market and its sales are outstripping that of its predecessor, the S4CPlus. The IRC5 represents a landmark in robotics: its powerful MultiMove® feature sets new standards for motion control. It enables complex, coordinated patterns, in which as many as four robots (up to 36 axes) are controlled in independent or synchronized movements by a single controller module.

But it’s not just the controller that breaks new ground! State-of-the-art embedded software in the new hand-held operator unit, the IRC5 FlexPendant, now allows IRC5 users to realize the benefits of customized interfaces.

Page 59: Abb Embedded Sym

59ABB Review 2/2006

Do-it-yourself robotics

Ease-of-use and flexibility are key concepts in the IRC5. The FlexPen-

dant is central to this user-centred phi-losophy and represents a significant breakthrough in both concept and technology. It is an integral part of the IRC5, yet a complete computer in it-self, developed with Microsoft’s latest technology for embedded systems, Windows CE.NET. It has an innovative ergonomic design and fits naturally in-to either hand, leaving the other free. An eye-catching feature is the unique, ABB three-way joystick for intuitive jogging of the robot. There are only eight hard buttons, used for fast access to crucial functions, such as starting and stopping robot programs, and of course an emergency stop. Other but-tons needed for operation appear on the graphical touch screen when need-ed, eg, a soft keyboard for entering text. This is a significant improvement on the more complex key-based sys-tems of competing devices. As the FlexPendant is subject to continuous operation in harsh industrial environ-ments, the touch screen is easy to clean and resistant to water, chemicals and even accidental welding splashes.

Usability issues have guided the de-velopment of the FlexPendant through all its phases. The use of the one-fin-ger touch screen is fast and natural, and the Windows-like interface, with internationally recognizable icons, is familiar to most end-users and there-fore minimizes operator training. As

its name suggests, the FlexPendant can be adapted to end-users’ specific needs. Currently, it can be operated in 14 different languages, including Asian character-based languages such as Japanese and Chinese. Left-handed operators can adapt the device from its default setting by simply rotating the display through 180 degrees 1 . Moreover, four of the hard keys are programmable, ie, their function can be assigned by the end-user.

“The ABB FlexPendant team is very astute and has made demanding re-quests for new functionality and bug fixes in our plat-form. They have definitely helped us to continue to improve the .NET Compact Framework platform.” Richard Greenberg1)

Development of the FlexPendantDevelopment of the IRC5 FlexPendant began in earnest in late 2001. The technical requirements of the device were well suited to Microsoft’s em-

bedded operating system, Windows CE 4.0, which was specifically de-signed for intelligent hand-held devic-es. While the choice of operating sys-tem was clear, it was more difficult to decide on the most suitable program-ming model. One possibility consid-ered was to use COM/ATL as compo-nent technology with MFC to create the user interface. The main concern was the complexity of the program-ming model. The FlexPendant had to be on the market within a couple of years, and the chosen technology would later be required to provide a user-friendly software development kit, enabling third parties to add cus-tom applications to the device. It was readily understood that Micro-soft’s coming framework for embedded devices, .NET Compact Framework (.NET CF), would offer an improved programming model, less error prone and less time consuming. But adopting new technology is a risky business and unforeseen technical problems often cause delays in the time schedule. In this case, it would be necessary to use both alpha and beta releases from Mi-crosoft. However, the benefits of using .NET, in terms of quality and produc-tivity, made it a very attractive option. It would also allow ABB to realize the concept of operator-customization.

During the first year of development, ABB worked in close collaboration with Microsoft as a participant in their

1 The FlexPendant is easily adjusted to suit a left-handed user

2 Klöckner-Desma, Germany, was an early adopter of the FlexPendant SDK

Embedded system technologies

Footnote1) Richard Greenberg, Group Program Manager of

the .NET Compact Framework team, Microsoft

(April 2006).

Page 60: Abb Embedded Sym

60 ABB Review 2/2006

Do-it-yourself robotics

Early Adopter Program (EAP) for .NET CF. Microsoft support was essential to the success of the project and approv-al for the EAP ensured the necessary commitment from Microsoft for the concurrent development of the Flex-Pendant and the software platform it would use. For their part, ABB, com-mitted themselves to launch the Flex-Pendant soon after the planned release of .NET CF in 2003.

ABB was about the only EAP partici-pant exploring .NET CF on Windows CE, as most others were using Pocket-PC as an operating system. Microsoft took a genuine interest in the develop-ment of the FlexPendant, and soon began to use ABB robots in their tele-vision commercials for .NET. From the start, the business relationship with Microsoft was characterized by com-mitment and technical expertise. The partnership led to ABB launching one of the very first advanced industry products built on Windows CE.NET. The amount of code developed for the FlexPendant is substantial; there are well over 180,000 executable lines of C# code, along with 25,000 lines of C++ code, providing the communica-tion layer towards the robot controller.

Software Development KitThe FlexPendant-SDK (Software de-velopment kit) was initiated in 2003, in parallel with further enhancement of the FlexPendant base software. An EAP for ABB customers was soon ini-tiated, and Klöckner-Desma, a Ger-man company targeting the shoe man-ufacturing industry, realized innova-tive ideas on how to facilitate the task of supervising the soling process. The idea was to display the robot path, ie, the contour of the sole graphically, thus providing an easier way to tune robot positions 2 . While robots are usually delivered with a general oper-ator interface, a customized solution is clearly more desirable to the end-user. Tailored solutions are easier to oper-ate and they optimize users’ invest-ment in automation. The FlexPendant-SDK has been part of the ABB soft-ware product Robot Application Builder (RAB) 3 since 2004. This allows an end-user or a third party to develop their own robot applications. These RAB applications are integrated with the basic functionality of the

FlexPendant using the standard struc-tured menu system. RAB represents an important advancement in robot tech-nology and places ABB‘s products ahead of its competitors’.

“The FlexPendant is one of the most sophisticated applications we haveseen using the .NET Compact Framework on Windows CE.”Mike Zintel2)

The embedded software platform cho-sen for the FlexPendant means ease-of-use for RAB users. Among pro-grammers, .NET distinguishes itself by the programming model provided by the Microsoft .NET Framework. One feature is its programming language independence, leaving the choice to the RAB developer to use any lan-guage provided by the integrated development environment, Microsoft Visual Studio. Most prefer C# or Visual Basic, which offer safe, yet efficient, development. As the majority of pro-grammers already know how to pro-gram Windows platforms using Visual Studio, they do not need extensive training when moving to RAB 4 .

Advanced software productsTo further speed up the development process for the customer, Virtual IRC5 is included in the Robot Application Builder package. Using Virtual Robot

Technology, Virtual IRC5 literally puts a robot controller on the desktop, as it allows the IRC5 controller software to run on a PC. An important feature of the .NET Framework is that com-piled .NET code can run on any sup-ported platform. This has enabled the development of a virtual FlexPendant, now included in Virtual IRC5, in parallel with that of the real device. Custom applications can thus be developed and tested at the desktop. Debugging is easy, with either a virtu-al or a real FlexPendant. The user needs only attach the main process to Visual Studio, set a break point in the code and step through it while it is executing. Developing real-time applications for devices with limited process and memory resources is, nevertheless, more demanding than developing PC applications. There-fore, the user documentation empha-sizes the skills of optimizing for per-formance and memory consumption.

RAB development in China and SwedenToday, many customers benefit from using RAB. These include robot system integrators, automotive companies, and even ABB itself. In 2004, to further strengthen its position in China and the Far East, ABB established a software development team in Shanghai. Their first challenge was to use RAB to devel-op a software application for use in the plastics industry. The goal was to speed up the process of programming and operating ABB robots used in the injec-tion molding process. RAB provided the team with a clean interface toward controller functionality and proved to be a real facilitator between teams working together from different corners

3 Robot application builder (RAB) enables customers to develop their own robot applications

Embedded system technologies

Footnote2) Mike Zintel, Production Unit Manager of the .NET

Compact Framework team, Microsoft (April 2006).

Page 61: Abb Embedded Sym

61ABB Review 2/2006

Do-it-yourself robotics

of the world. The result, RobotWare Plastics, was successfully launched in 2005. Again, ease-of-use was a primary concern and end-users were closely in-volved from the early stages.

One of the first customers to be in-volved with RobotWare Plastics was the Swedish molding company, AD-Plast. This company was founded in 1963 as a tooling company, but later moved in-to injection molding, with automotive components as a key product segment. Robotization has allowed the company to achieve consistent quality levels throughout the process and to maintain competitive pricing. RobotWare Plastics

on ease-of-use and the desire to in-crease customer value. A well-designed user interface presents relevant infor-mation and functionality at the right time. This is much easier to achieve with a customized user interface 6 than with a general-purpose interface.

The futureThe embedded technology chosen by ABB for the new generation of smart devices for the robot industry has ex-ceeded expectations. Advantages such as worldwide competence of platform, lasting commitment from Microsoft to the embedded market and continuous inspiration from the PC world, all con-tribute to making Windows CE.NET a long-term technology choice. Mean-while, ABB’s standing on the robot market will be strengthened as state-of-the-art technology, flexibility and ease-of-use combine to further enhance robot automation, giving ABB and its customers a competitive advantage.

Ingela Brorsson

Ralph Sjöberg

ABB Robotics

Västerås, Sweden

[email protected]

[email protected]

Anna Liberg

ABB Robotics

Shanghai, China

[email protected]

has meant smoother and faster start-up for new production processes. Opera-tors no longer need to be robot-pro-gramming experts, as the program wizard of the graphical user interface guides the operator through the entire programming process. Another helpful feature is the graphic cell overview 5 .

RobotWare Dispense, a robot applica-tion enabling dispensing processes, such as gluing and sealing, has been part of the ABB product range for many years. In 2005, an operator inter-face, customized for the dispensing process was developed in Sweden. Again, a heavy emphasis was placed

Embedded system technologies

5 RobotWare Plastics is used to program and operate the ABB robot tending the injection molding machine. Pictures of equipment and produced items on the graphic cell overview make the interface intuitive and easy to use.

4 FlexPendant software architecture

.NETCompact Frame-work

App

licat

ion

Fram

ewor

k (C

#)

Windows CE 5.0

RAB App2C#

RAB App1Visual Basic

FlexPendant SDKuser interface controls, CAPI

C#

Class librariesC#, C++

COM-based internal APItowards robot controller

C++/COM/ATL

Product & Presentation

Logics

Data Access

RAB App – Application developed with Robot Application Builder, of which FlexPendant SDK is a part.

CAPI – Controller Application Programming Interface, public API offering robot controller services.

COM/ATL – Component Object Model, Microsoft component technology.

Active Template Libraries, set of template-based C++ classes that simplify the programming using COM.

*arrows show dependence

6 A RAB application for dispensing processes is tested at the desktop before it is down-loaded to the real device. This is made feasi-ble by Virtual IRC5.

Page 62: Abb Embedded Sym

Design patterns Co-design patterns for advanced control with AC 800PECErnst Johansen

Power electronics has, over recent decades, made great progress, not only in terms of power and speed performance, but also in the breadth of applications being catered for. Power converters are required to become ever faster, cheaper, lighter and more flexible while fitting into less space and requiring less installation and maintenance time.

The implementation of the corresponding power electronics control systems presents many tough challenges, including the magnitude of the control time-domain, which ranges from nanoseconds to seconds. Costs and risks of de-velopment can be greatly reduced through the adoption of a control platform. Drawing on tried and tested component technologies, individual systems can be developed very quickly and to high quality and performance standards. ABB’s AC 800PEC is such a platform.

62 ABB Review 2/2006

Control platforms are necessary to be able to meet the market’s

demand for faster and more cost-efficient engineering. At the same time, such a platform creates a single-point-of-failure, representing a poten-tial risk to the whole organization. Successful platform development requires striking a delicate balance between optimizing reusability (and so reducing costs) and optimizing performance (at the price of reusabili-ty and hence, potentially, at the price of quality).

The secret behind the success of the AC 800PEC control platform is a col-lection of design patterns that offers excellent testability – a key feature permitting high quality to be com-bined with reduced time-to-market.

The simulation conceptThe concept behind the PEC (Power Electronic Controller) is the develop-ment workflow in which simulation models are converted directly into code for the target controller 1 . This conversion requires no manual recod-ing. In this way, an important source of errors is eliminated and a high degree of confidence is provided in the equivalence of the behavior of the simulated and real systems.

The PEC ArchitectureIn power electronic control, the time-domain ranges from nanoseconds in the switching patterns up to seconds in the start-up sequences. A great strength of the PEC architecture lies in it covering these nine orders of magnitude in the control time-domain without compromising on simplicity or flexibility.

Page 63: Abb Embedded Sym

63ABB Review 2/2006

Design patterns

together. So how does the PEC exe-cute fast control and implement I/O connections?

The control program can be divided into two main tasks: slow control (mil-lisecond range and slower) and fast control. A classic design would utilize two different physical components for these main tasks, a CPU for slow con-trol and a DSP for fast control. By in-vestigating different use-cases it was concluded that the load distribution between fast (typically 100s) and slow (typically 10ms) control was strongly application specific. The lack of a uni-versal rule for load distribution prompted developers to use a single CPU for both fast and slow control. This decision resulted in the need for a very high-performance CPU. Besides solving the load distribution problem, this architecture greatly simplified automatic code generation.

The concept behind the PEC (Power Electronic Controller) is the develop-ment workflow in which simulation models are con-verted directly into code for the target controller.

The concept of automatically generat-ing real-time code from simulation models cannot be implemented if the simulation tool offers no auto-code capability. ABB decided to use Math-works® Matlab/SimulinkTM for the system simulation. This tool offers a powerful Real-Time-WorkshopTM (RTW) extension for target code generation.

The architecture is designed to sup-port cost sensitive small systems with local I/O only 2c , as well as very large systems requiring distributed I/O 2d using fiber-optic connections. These two system types demand a totally dif-ferent design of the I/O circuits in the controller. To offer a solution capable of covering all use-cases, a system-lev-el FPGA (Field Programmable Gate Array) was used. This is a hardware component in which the circuit itself is fully programmable. Such FPGAs are used both in the PEC controller and in the distributed I/O nodes. Besides solving the flexibility problem, the FPGA has the additional advantage of being backed up by a very mature design and simulation workflow.

Like the Matlab/SimulinkTM-based workflow for controller code develop-ment, the FPGA implementation work-flow is based on a simulator and a compiler. Even thought compilers are available that will translate some Mat-lab/SimulinkTM models into VHDL code, ABB decided not to use such tools in the PEC workflow. The reason for this is that most of the FPGA com-ponents in the PEC library are neither modeled nor verified efficiently in the Matlab/SimulinkTM language. Instead, a VHDL-based workflow was used for the digital circuits. The adopted work-flow was originally developed for ASIC design where high first-past yield1) is mandatory. Furthermore, the workflow offers excellent modeling and verification capabilities.

At the time the architecture was de-fined, however, there was one major drawback – the cost of the high-per-formance CPU and the system-level

In order to support the direct conver-sion of simulation models, the archi-tecture 2 has two major differences compared to classic control systems. No dedicated DSP (Digital Signal Pro-cessor) is provided for fast control and there is no mechanical rack where I/O modules are connected

Embedded system technologies

3 Co-design patterns defined by ML/SL (Matlab/Simulink) and VHDL models

CPU

FPGA

s

nsIO

IO

ML/SL

VHDL

4 System models are converted into the real-time domain for accelerated verification through execution on PEC hardware

Simulator

Control Model

System Model

Real-TimeController

Real-TimeSystem Model

Accelerator

Real-TimePEC

Real-TimePEC

1 The simulated model is automatically converted to executable code for the real-time domain

Simulator

Control Model

System Model

Real-TimePEC

Real System

2 A single model can flexibly be adapted to handle different control time-domains

CPU

FPGA

ms

s

ms

μs

nsIO FOIO

IO IO

a

b

cd

a mainly fast controlb mainly slow controlc Local I/Od Distributed I/O connected by fiber optics

Page 64: Abb Embedded Sym

64 ABB Review 2/2006

Design patterns

FPGA. How this problem was finally solved will be shown later in this arti-cle.

Design patterns for control and verificationA design pattern is a pre-engineered solution-template to a specific prob-lem. Design patterns are a method that has been used by software engi-neers for a long time. However in the area of hardware/software co-design, the definition of generic patterns is more difficult [1]. The AC 800PEC con-trol system makes use of the design pattern method for several design issues found in power electronic applications. A collection of reusable design patterns allows development engineers to rapidly define new sys-tems with high complexity. A system engineer can concentrate on solving his unique problem while trusting in pre-engineered patterns for imple-mentation details.

PEC systems differs from most other systems in that the design patterns in the PEC system are not pure software patterns, but reusable co-design pat-terns 3 . The motivation for using co-design patterns is to cover nine orders of magnitude in the control time-do-main (ns to s), a capability not feasi-ble using one single technology (eg, software).

Co-design is, however, a great chal-lenge for system verification. Excellent test coverage is mandatory to assure high confidence that the implementa-tion is error-free, but the simulation of a control system covering nine orders of magnitude in the time-domain is extremely slow. Simulating a complete PEC co-design system would take days and weeks to complete on a PC workstation. Such a prerequisite is simply not compatible to fast time-to-market requirements.

But the PEC concept has an intrinsic feature that can be used to solve this tricky problem very elegantly: The con-cept behind the PEC is to offer a work-flow where simulation models are con-verted directly into target controller code. This principle is not only applica-ble to the control model, but also to the model of the simulation environment used with it. By executing the control and system model on the PEC controller concurrently 4 , the verification of co-design patterns in the real-time domain is speeded up significantly.

Co-design – a real challenge for embedded system designersA signal filter can be implemented using analog electronic circuits, a digital filter in an FPGA, or as a piece of software running on a CPU. These solutions all offer identical functional-

ity, but differ totally in terms of cost and reusability. Co-design is about tak-ing the right decisions on how to map a solution to different technologies.

The invention of system-level FPGA components meant that programma-bility was no longer restricted to soft-ware. The invention permits new de-sign patterns for hardware and system design. As there is no cookbook for co-design, it remains a real challenge for the system designer.

Excellent test coverage is mandatory to assure high confidence that the imple-mentation is error-free

System simulation to explore optimal design patternsIn the process of finding optimal algo-rithms and structures, system simula-tion is applied in the evaluation and comparison of different designs. As an example of the co-design process, the Analog-Digital Conversion (ADC) cir-cuit is discussed in the following.

As the developers were required to im-prove the existing ADC design pattern in terms of cost and quality (Signal to Noise Ratio – SNR), they selected dif-ferent topologies 5 that fitted the PEC architecture. The topologies where simulated in the Matlab/SimulinkTM simulation environment and compared in terms of complexity and quality.

The developers concluded that, theo-retically, the best SNR was obtained by utilizing a combination of over-sampling and digital filters 5a (due to the noise-shaping capability of digital filters [2]). Over-sampling 5b-d utilized a much-lower cost ADC circuit than this solution, but added the need for a high-speed digital filter operating at 25x-speed. Was it feasible to imple-ment the filter? Should the filter calcu-lations be executed on the CPU or in the FPGA? Did it pay-off to increase the digital processing payload?

Direct Code GenerationThe capability to automatically con-vert simulation models into real-time control applications made it very easy to create target code for the different

Embedded system technologies

5 Analog-digital conversion co-design topologies, with different components of the task being handled by analog circuits, on FPGA and on CPU

1xADC

14-bit

25xADC

12-bit

25xADC

12-bit

25xADC

12-bit

N + +N N N+

ML/SL

VHDL

a b c d

ML/SL = Matlab/Simulink

Page 65: Abb Embedded Sym

65ABB Review 2/2006

Design patterns

topologies. As the PEC had a build-in load monitor it was easy to measure the CPU load (payload) for all topol-ogies 6 . Operating the fast filter 6b in software turned out to generate a too high CPU load and was not feasi-ble.

As Matlab/SimulinkTM offers comprehensive libraries it was actually not necessary to develop any new code for the CPU filter design.

Optimized VHDL ComponentsFor the FPGA filters, Matlab/SimulinkTM was used to eval-uate the filter topology, char-acteristic and calculate the appropriate coefficients. The imple-mentation and simulation of the filters was done in the VHDL environment.

In an FPGA circuit, the payload is measured in circuit area. Compared to a digital filter implemented on the CPU, FPGA filter design offers many more options. The precision (number of bits), the clock-frequency, the filter architecture, the throughput (samples per second), the number of MAC

(Multiply-Accumulate) operations per filter and the number of channels per filter are all programmable, offering a vast choice of design alternatives, all with different payloads. The 5c topol-ogy, with one high-speed filter operat-ing inside the FPGA and one slower filter calculated by the CPU, turned out to offer the most cost-efficient co-design solution. This was selected as the preferred design pattern for ADC conversion 7 .

Real world verificationDuring the co-design process, the real system was modeled – including expected signal noise. In many systems, the noise is unpredictable and the simulation of noise unreliable. Real world verification is therefore still important to guarantee product quality 8 .

Cost and performance – a moving target At the time of the definition of PEC architecture in 1999, the drawback of the architec-ture was the high cost of the CPU and the system-level FPGA. As these components where very expensive at that time, they where used primar-

ily in high-end applications such as flight-simulators and prototyping systems for ASIC development.

As the process technology for digital circuits improved very rapidly, the manufacturing costs of CPU and FPGA dropped dramatically – during a peri-od of five years the cost of these digi-tal circuits was reduced by more than 90 percent. As these lower-cost devic-es came onto the market, a further advantage of the architecture paid off – its excellent application portability. Today ABB is offering AC 800PEC con-trollers based on the most cost effi-cient 90 nm silicon process technology, offering customers excellent product quality at a very competitive price.

Ernst Johansen

ABB Schweiz AG

Turgi, Switzerland

[email protected]

References

[1] F. Mayer-Lindenberg, Dedicated Digital Proces-

sors: Methods in Hardware/Software Co-Design,

John Wiley & Sons (February 12, 2004),

ISBN 0-470844-44-2

[2] Walt Kester, Analog-Digital Conversion,

Analog Devices Inc. (March 2004),

ISBN 0-916550-27–3, 2.37–2.41

Footnote1) First pass yield is a ratio of the number of “good”

units (ie, not requiring rework) to the total produced.

8 Real-time verification of 12-bit / 1MSps ADC (yellow) and FPGA-filter with noise-shaping (pink)

Embedded system technologies

6 Target load evaluation of variants 5b-d

ML/SLRTW

Compiler

Target CPULoad

Monitor

dcb

Simulator

7 Optimal VHDL filter pattern (variant 5c )

VHDL

ML/SLVHDLTestbench8 Ch

IIR

1x MAC

80 MHz

35-bit

c

ML/SL = Matlab/Simulink

Page 66: Abb Embedded Sym

An industrial application can have thousands of embedded subsystems that need to communicate with their environment. Each of these requires its own data and power connection. Cabling is costly to install, it is a fre-quent source of failure and a curb on flexibility. Such applications are better served by wireless technologies.

Wireless communication in automation environments has made important a dvances in recent years [1]. However, wireless power supplies remain a chal-lenge. In 2004, ABB released a series of unique wireless products — wire-less proximity switches — in which both communication and power sup-ply are wireless. Since the introduction of these devices, the “WISA” (Wireless Interface to Sensors and Actuators) technology [2] on which they were developed has been further expanded to include new products and commu-nication profiles.

Wireless power in wireless products How to cut the power cordGuntram Scheible, Rolf Disselnkoetter

66 ABB Review 2/2006

Page 67: Abb Embedded Sym

67ABB Review 2/2006

Wireless Power in Wireless Products

The devices that benefit most from a wireless power supply are com-

ponents of distributed control and auto mation systems, typically sensors and actuators – devices that often have embedded intelligence. They are generally located in remote environ-ments, where no general power source is available, or in areas that are diffi-cult to access. They can be amongst numerous other devices, in highly mobile assemblies, such as robots, or in high-voltage environments.

In such applications, energy efficiency is a primary concern. The power con-sumption of conventional wireless technologies, as used in standard office wireless components, for exam-ple, is generally far higher than that of dedicated systems that have been designed around low-energy require-ments. Low-energy devices often make savings by using hierarchical, pulsed operation modes. Most appli-cations, such as data collection, actua-tion, processing and communication, are performed only at specific time points. Such tasks can be carried out using pulsed operation modes that are interspersed with energy-saving sleep phases.

Sources of wireless power Generally speaking, wireless power/energy can be either:

Included in the system in the form of batteries, fuel cells, etc.

Taken from the local environment in the form of light, heat, vibration, user activation, etc.

Transmitted to the system via opti-cal or radio frequencies, sound, etc.

The primary WPU100 units are tuned and controlled automatically, which means that the WISA-POWER supply system can be applied to different applications just by altering the primary coil geometries.

Although the use of battery power is considered acceptable in the consumer world, regular battery charging or replacement is not a practical option in typical industrial applications. In extremely remote areas, batteries, per-haps in combination with solar or geo-thermal power, remain the only practi-cal option. In general industrial appli-cations, however, where hundreds of devices that require constant, reliable power supplies run day and night, batteries are not an option. Their ener-gy density, around 1.2 Wh/cm3, is too

low. Fuel cells are somewhat better, but even their potential is little more than 2 Wh/cm3, and much develop-ment is still required before they can be used in everyday industrial installa-tions.

Environmental energy sources also fail to meet the needs of industry applica-tions, due to their unpredictable na-ture – both in terms of general usabil-ity and reliability; and so they fail to meet one of the most important con-siderations of all. Such solutions would also incur considerable engi-neering and design costs for every single application.

3 A “Helmholtz-like” arrangement of rectangu-lar coils integrated into an industrial applica-tion. D is the separation between coils and S is the smallest dimension (width or height).

1 ABB’s wireless power technologies are based on the well-known transformer principle

principle of a transformer

tech

nolo

gie

s

larg

e vo

lum

e w

irele

ssm

ediu

m p

ower

co

ntac

tless

high

pow

er

cont

actle

ss

single

sensors wireless I/O valves

other

(servo drives) welding

primary winding

secondary winding

Power/W (RMS!)

10 m 100 m 1 10 1 k 100 k

WISA-power

Embedded system technologies

2 WISA-POWER: Wireless power supply. A power supply a feeds primary loop b with a current at 120 kHz. Sensors c within the primary loop are equipped with second-ary coils. The right-hand schematic shows the equivalent circuit diagram with loose coupling

a a

b

cb

c

c

c

c

c

c

Page 68: Abb Embedded Sym

5 WISA-Power ring-type primary coil structure in a completely rotating cable winding ma-chine (ABB High Voltage Cable, Karlskrona, Sweden). 156 wireless proximity switches WPS with embedded electronics rotate in a complex 2D-movement inside the machine, to ensure a failure-free production.

4 WISA-POWER primary spot coil, integrated into a machine tool with receiver coil provid-ed by customer and embedded electronics (wireless tool survey system DDU WiSy Courtesy of ARTIS GmbH, Bispingen, Germany)[5]

68 ABB Review 2/2006

Wireless Power in Wireless Products

After a thorough evaluation of the various available options [3], it seems that the only viable, generally applica-ble solution is one based on long-wave radio frequencies, a form of “magnetic coupling”. ABB has a num-ber of power supply options that use magnetic coupling 1 . Depending on the transmission distance, a wide range of applications and power levels can be implemented.

The power needs of distributed elec-tronic devices, such as single sensors, and wireless I/O (input/output) devic-es, in discrete factory automation set-tings, are covered by the first genera-tion of WISA-products 1 . The wireless supply unit WPU100, together with a coil setup (“primary loop wires”), provides a low-level power supply across large distances (a few meters). This is suitable for most sensors and actuators in discrete factory automa-tion.

WISA-POWER: The “magnetic supply”The basic principle of a magnetic field-induced power supply can be described by the well-known “trans-former” principle 2 . In the case of the WPU100, the power supply feeds the primary winding 2b , a large coil, which can be arranged around a pro-duction cell, the secondary side of which corresponds to a practically unlimited number of small receiver coils 2c . Each receiver coil is equipped with a ferrite core to increase the amount of flux collected by the coil.

For this type of “transformer”, magnet-ic coupling is low. The receivable power is determined by the amplitude of the magnetic field at the location of the “receiver” (secondary) winding. However, if the primary coils are set up in a “Helmholtz-like” arrangement 3 , the field (and therefore the receiv-able power) will be fairly constant over a large volume of space.

The design rules for the number and size of the primary coils are very sim-ple: D = 0.7 × S, where S is the small-est dimension (width or height) of one coil frame and D is the separation between the coils to provide an ade-quately homogeneous field amplitude inside the arrangement 3 .

Although people will rarely work con-tinuously in such an automated pro-duction cell, the strength of the mag-netic field at all normal working posi-tions (including within such a cell) complies with international occupa-tional regulations and recommenda-tions [4]. WISA-POWER works at a similar frequency (120kHz) and in exactly the same way as many of the anti-theft and radio frequency identifi-cation systems used in shops and supermarkets.

The fundamental chal-lenges of wireless power distribution and the reli-ability of real-time-suited wireless communication have been successfully resolved.

Within this therefore limited-ampli-tude magnetic field, power levels can be scaled according to the needs of different applications by changing the size of the secondary coil. This al-lows embedded systems to be con-nected to wireless power by the inte-gration of a suitable receiver coil and circuit. A good example of this can be seen in 4 . Artis, a company, based in Bispingen (Germany), has used WISA-POWER technology to create its own secondary-side electronics, adapted to the special needs of tool-ing sensors [5].

Energy losses in such a system are surprisingly small and are mainly due to skin and eddy current effects in the coil or in nearby metallic objects. In typical factory automation environ-ments, energy losses are around 15W/m3.

Resonant, medium-frequency power supplyThese unconventional transformers are best operated in a ‘resonant’ mode. In this mode, the transformers’ relatively large leakage inductances are compensated for by a capacitance, which allows the WPU to stimulate the resonant circuit at relatively low voltages. The WISA WPU100 primary power supply must also be able to accommodate: Changes in the environment over time, eg, caused by the movement of large, mobile metallic objects such as robots.

Different ’load’ requirements, caused by differently sized primary coils (inductance values), and losses, caused by factors such as eddy cur-rents in adjacent metallic objects.

Other nearby wireless supply systems, which may couple induc-tively.

To accommodate these requirements, the WISA WPU100 contains a fast and highly accurate control unit, which compensates for such changes and automatically keeps the primary sys-tem at a fixed resonance frequency of 120 kHz. The WPU100 unit can adapt

Embedded system technologies

Page 69: Abb Embedded Sym

69ABB Review 2/2006

Wireless Power in Wireless Products

to inductive loads from 11–54 µH and supply currents of 4–24 A.

The primary WPU100 units are tuned and controlled automatically, which means that the WISA-POWER supply system can be applied to different ap-plications just by altering the primary coil geometries, eg, by using ring- or line-type coil structures or spot coils 4 5 . This is particularly useful if wire-less power is needed only in certain areas of an installation, for example in devices that move along a ring or line structure, or for bridging critical spots in a system.

Due to its unique capabilities, the WPU100 unit can also be used in applications with customized power-receiver units adapted to specific geo-metrical and application needs 4 .

Rotating fieldUnidirectional magnetic fields can be shielded unintentionally by metal objects. To avoid this, two loops can be mounted orthogonally. The loops must be fed by separate power sup-plies, whose currents are phase-shift-ed by 90 ° with respect to each other. This creates a continuously rotating, 2-dimensional field.

Omni-directional receiver structureTo achieve sufficient power output on the receiver side, the secondary coils must also be operated in a resonant mode. Further, to make the available power independent of the receiver’s

orientation with respect to the primary field vector, an orthogonal setup of three coils on a common core has been chosen. Being easily tunable to the fixed resonance frequency, this arrangement is well suited to mass production.

The generic and modular technology of the WISA family products, which started with the wireless proximity switches, is now being extended to other devices and applications.

With this technology, the available power densities for typical “worst-case” shielding conditions in real applications remain in the order of 1.2 mW/cm3. The absolute power level can be modified with the coil size and shape.

The scalability and integration of WISA-POWER receiver coils into prod-ucts has been demonstrated in differ-ent applications. Total power con-sumption of the wireless proximity switch and its electronics 6 is signifi-cantly below 10 mW – the new sensor pad, WSP100 7 , which allows the connection and supply of up to eight sensor heads, can already provide several tens of milliwatts in “worst case” conditions of shielding. Under normal conditions, at this size, the

WISA-POWER principle can provide several hundred milliwatts, and, under controlled conditions, up to 1 Watt.

Unbounded future With the introduction of the WISA power and communication technolo-gies, ABB has made significant ad-vances in the technology of wireless embedded systems. The fundamental challenges of wireless power distribu-tion and the reliability of real-time-suited wireless communication have been successfully resolved.

The generic and modular technology of the WISA family products, which started with the wireless proximity switches, is now being extended to other devices and applications. The generic WISA-POWER supply and WISA-COM communication technolo-gies are set to find their way into many further applications.

Guntram Scheible

ABB STOTZ-KONTAKT GmbH

Heidelberg, Germany

[email protected]

Rolf Disselnkoetter

ABB Corporate Research

Ladenburg, Germany

[email protected]

References

[1] Niels Aakvaag, Jan-Erik Frey: Wireless communi-

cation and sensor networks. New-breed net-

working solutions for industrial automation ABB

Review 2 /2006

[2] Jan-Erik Frey, Andreas Kreitz, Guntram Scheible:

Wireless but connected, ABB Review 3 and

4 /2005

[3] G. Scheible: Wireless energy autonomous sys-

tems: Industrial use? Sensoren und Messysteme

VDE/IEEE Conference, Ludwigsburg, Germany,

March 11–12 2002.

[4] International Commission on Non-Ionizing Radia-

tion Protection (ICNIRP): Guidelines for Limiting

Exposure to Time-Varying Electric, Magnetic, and

Electromagnetic Fields (up to 300 GHz). Health

Physics vol 74, no 4, 494–522, 1998.

[5] Berend Denkena, Dirk Lange, Dipl.-Ing. Jan

Brinkhaus: Spielraum in der Überwachung;

Fachzeitschrift mav „maschinen anlagen ver-

fahren“ Konradin Verlag Robert Kohlhammer, 2005

6 WISA-POWER “Power Cube” integration into the WISA communication module WSIX100 of a wireless proximity switch

7 WISA-POWER Integration of a scaled “Power Cube” into the WISA sensor pad WSP100 to supply eight sensor heads and their real-time WISA-COM communication

Embedded system technologies

Page 70: Abb Embedded Sym

Looking for an electronic solution to a problem can be a daunting and mind-boggling task. The speed at which technology is advancing means a cus-tomer never seems to have the most modern and efficient device. Designers of high-performance embedded sys-tems are constantly pushing to fit more processing power into the same box. This “push”, it must be said, is strongly dictated by demands for better time-to-market, increased power and volu-

Coming of age Embedded system FPGA and VHDLErik Carlson, Franz Zurfluh, Catherine Körbächer

70 ABB Review 2/2006

A field-programmable gate array or FPGA as it more commonly

known is a semiconductor device con-taining programmable logic compo-nents and programmable intercon-nects. The programmable logic com-ponents can be programmed to dupli-cate the functionality of basic logic gates (such as AND, OR, XOR, NOT) or more complex combinatorial func-tions such as decoders or simple math functions. In most FPGAs, these pro-grammable logic components (or logic blocks, in FPGA parlance) also in-clude memory elements, which may be simple flip-flops or more complete blocks of memories.

This technology was introduced sever-al years ago 1 . However cost and per-formance limitations initially meant that FPGA technology was used only for fast prototype developments. Vol-ume production was than carried out using an ASIC1) (Application Specific Integrated Circuit) design.

Today ASIC development has in-creased considerably both in time and

me efficiency, and the need to squeeze every last drop of performance out of a device. However, rather than having to buy new devices, customers mostly prefer to upgrade or add new features to their previous investments. Tradi-tionally this has been made possible by designing microprocessors and software into the products. One way of making such demands possible is to use FPGA technology to replace digital signal processors. By

introducing FPGA logic, hardware can be used as flexibly as software. In addition high speed, low power con-sumption and easy reuse of proven logic can also be achieved.

FPGA technology has been used by ABB for several years, in particular in the automation, medium and high volt-age industries. This article discusses some FPGA design aspects as well as its benefits to ABB and its customers.

Page 71: Abb Embedded Sym

71ABB Review 2/2006

Coming of age

plex a design; and they draw more power. To reduce power consump-tion, the clock drivers have been de-signed so as to ensure they are inac-tive when not in use. This feature has been optimized in several hand-held and battery operated devices where low power is essential. Unfortunately, low feature size gives large static leak-age current and must be considered when the optimum FPGA technology is selected.

For designs with lower production volumes, FPGAs are now more cost effective than an ASIC design.

Modern FPGAs also include built-in higher level functionality. These fixed hardware structures – together with the regular gate structures – not only reduce the area required but they also enhance the speed performance. Ex-amples of embedded functions in-clude multipliers, generic DSP blocks, embedded processors, high speed IO (input/output) logic and embedded memories. More importantly, ready made and tested designs for complex tasks are also available such as Fast Fourier Transforms (FFTs) for harmon-

ic analysis, CORDIC algorithms3) for vector manipulation, and high perfor-mance microprocessor cores. These blocks are implemented with normal logic cells. They are called IP (Intel-lectual property) blocks and are avail-able both from FPGA vendors and from the commercial market.

FPGA Technology – From glue logic to System-on-ChipOnce used only for glue logic, FPGAs have progressed to a point where System-on-Chip (SoC) designs can be built on a single device. Over the past 10 years, the number of gates and features has increased dramatically: FPGA capacity has increased more than 200-fold and speed has increased more than 20-fold to compete with capabilities that have traditionally been offered through ASIC devices only. Innovative architectural and circuit features are equally important , as are advancements in design meth-odology. External system clock rates now exceed 150 MHz. The cost for an FPGA with 10,000-gate functionality has decreased by a factor of over a hundred. I/Os have to be compatible with many new standards and must be able to drive transmission lines.

Device specific circuits embedded in FPGAs. Configurable Logic Blocks (CLBs) provide the functional elements for combinatorial and synchronous logic. Modern technologies contain function generators (look-up tables, shift regis-ters) storage elements, arithmetic logic gates and multiplexers.

As regards clock management circuit-ry, clock modifiers are implemented as analogue Phase-Locked-Loops (PLL)4) or digital Delay-Locked-Loops (DLL)5). PLLs and DLLs are used for clock skew compensation and clock synthesis (multiplication/division). Flexible PLL/DLL circuits are available in the newest FPGAs, and some new FPGA devices support glitch-free clock multiplexing, as well as clock shutdown for low-power applications.

Advanced interfaces such as Input/Output Blocks (I/O Blocks) are pro-grammable as input, output and bi-di-rectional, and registers are edge-trig-gered flipflops or level-sensitive latch-

Embedded system technologies

tooling costs because of the rapid in-crease in complexity. As feature sizes have shrunk and design tools im-proved, the maximum complexity (and hence functionality) possible in an ASIC has grown from 5,000 gates to over 100 million. For designs with lower production volumes, FPGAs are now more cost effective than an ASIC design. Other advantages include shorter time to market, lower non-re-curring engineering costs, the ability to re-program in the field and to add new functions or fix bugs. In addition, FPGAs contain programmable logic blocks and programmable intercon-nects that allow the same FPGA to be used in many different applications. The regular structure of the cell array lends itself to small geometries. In fact, FPGA designs have more than outperformed Moore’s law2).

Several millions gates are already available in modern day FPGA fami-lies. Interconnections can be done in up to 9 metal layers, thus allowing easy monitoring and testing during the development and debugging phas-es. Powerful clock drivers together with complex routing software enable internal clock frequencies of up to the GHz range.

Compared to their ASIC counterparts, traditional FPGAs: have been general-ly slower; they can‘t handle as com-

1 Progres in FPGAs technology is opening more and more applications for them

Per

form

ance

& D

ensi

ty

1980 1990 2000

Gates and FFs

Adders Counters

Data and Control Path Memory

PCI FFT/FIR Filters

Encription MP3 Decoder

Industrial Automation

Medical Imaging Control Systems Graphic Cards

Printers

Glue Logic

Small Systems

Integration

LSI

Application Specific

Functions

Volume Apps.

Page 72: Abb Embedded Sym

72 ABB Review 2/2006

Coming of age

es. A range of single-ended standards such as LVTTL, Peripheral Component Interconnect (PCI)6) and differential signaling are supported. The newer FPGAs are compatible with many I/O standards and I/O voltages.

Small RAM devices can be emulated with CLBs, but these are slow and consume a lot of logic. Single, two-ported, dual-ported and quad-ported block RAMs are available in many FPGAs today. They are cost and per-formance optimized RAM implementa-tions.

Other device specific circuits embed-ded in modern day FPGAs include:

CPU soft/hard cores: An FPGA device with embedded CPU hardware IP is a new category of microprocessor. Today products with 8-bit or 32-bit CPUs are available. The two basic com-ponents of a platform FPGA are the CPU core and on-chip bus architecture.

Multipliers: Some FPGA families have dedicated hardware multipliers – some that range from 8×8 bit to 18 ×18 bit – that help increase computational per-formance. Multipliers can be modeled in virtually any FPGA using CLBs, but they require a lot of logic and are slower than dedicated multipliers.

The FPGA development processDesigning an FPGA with a million gates has become both a system and an architectural problem. Modern lan-guages, such as the Unified Modeling Language (UML)7), are necessary to ensure correct specification and de-sign. Simulation and verification is often done with Matlab® or other high-level simulation tools. However, logic description is mostly done using a Hardware Description Language (HDL).

HDL describes electronics circuits in terms of the circuit‘s operation, its design, and tests to verify its opera-tion by means of simulation. In con-trast to a software programming lan-guage, an HDL‘s syntax and semantics include explicit notations for express-ing time and concurrency, which are the primary attributes of hardware. VHSIC Hardware Description Lan-guage (VHDL) is widely used as a

design-entry language for FPGAs and ASICs in electronic design automation of digital circuits.

VHDL was originally developed in the early 80’s at the behest of the US De-partment of Defense to document the behavior of ASICs that supplier com-panies were including in equipment. In other words, it was developed as an alternative to huge, complex man-uals. The key advantage of VHDL when used for systems design is that it allows the behavior of the required system to be described (modeled) and verified (simulated) before synthesis tools translate the design into real hardware. Another benefit is that VHDL allows the description of a concurrent system, and it synthesizes the detailed structure from a more abstract specification.

FPGAs are ideal as co-processors or pre-/post-processors for offloading highly computational.

Basic VHDL design processSeveral steps are required to create the structure and behavior of an ap-plication before it can be downloaded to an FPGA. 2 illustrates the basic VHDL design process.

The first step is design coding. This can be done using a hardware de-scription language such as VHDL, Ver-ilog or SystemC, or the code can be generated with so called system com-pilers (see “Designing DSP applica-tions for FPGAs” later in this article). In parallel, a test bench is developed

so that the design can be verified using a simulator – the simulator exe-cutes the design and checks that the correct results are produced. When the design is free of errors, a com-plete synthesis is done. During syn-thesis an intermediate representation of the hardware, called a netlist, is generated for the layout tool.

The next step is called layout or place-and-route. This involves mapping the logical structures described in the netlist into macrocells, interconnec-tions, and input and output pins. In other words, the netlist is fitted to the actual FPGA architecture. However, factors constraining the layout include speed and area optimization. The lay-out tool can generate another netlist with the timing delay information represented in Standard Delay Format (SDF)8). This netlist can also be used for simulation with the test bench to verify the correct timing behavior. Finally a configuration bit stream is generated. This file can be download-ed into the FPGA’s control memory or directly into the FPGA itself.

FPGA based Digital Signal ProcessingState-of-the-art FPGAs provide the ba-sic functional capability to implement signal processing functions, even in the low-cost families. These devices are ideal as co-processors or pre-/post-processors for offloading highly computational intensive functions. Implementing signal processing algo-rithms in an FPGA instead of a DSP gives the designer additional degrees of freedom. As 3 shows, high perfor-mance calculations can be implement-ed in parallel for high data through-put, and semi parallel or in series for

2 The basic VHDL design process

Design Coding

Synthesis

VHDL Code

Netlist

Layout

VHDL Netlist

VHDL Code

Simulation

SDF & Netlist

Test Bench Development

Generation of BIT File

Embedded system technologies

Page 73: Abb Embedded Sym

73ABB Review 2/2006

Coming of age

cost efficient applications. The archi-tecture can be ideally customized for different applications to enable per-formance and cost targets to be reached.

However, design tools are the bottle-neck in DSP designs implemented on an FPGA. DSP system design in pro-grammable logic devices requires both high-level algorithm and HDL devel-opment tools. Nowadays, major FPGA vendors offer DSP builder tools that help shorten DSP design cycles. These tools combine the algorithm develop-ment, simulation, and verification ca-pabilities of Matlab® and Simulink® with synthesis, simulation, and place-and-route.

Several millions gates are already available in mod-ern day FPGA families.

As an example, 4 shows how a finite impulse response (FIR) filter with

N-coefficients is implemented in a DSP (using von Neumann architec-ture)9). In this case it needs a total of n cycles to produce the output. How-ever, an FPGA can handle the multi-ply-accumulate (MAC) operations in parallel and only one clock cycle is needed for the entire computation!

Designing DSP applications for FPGAsOften the DSP part of the FPGA de-sign is just one block in a larger im-plementation where traditional FPGA design methods and tools are used. An integrated system design approach to help the simulation and develop-ment of each specific part will be-come essential in the future. There are many possible methods ranging from hand coding, model-based design approaches – C/C++ to RTL synthesis – to DSP synthesis for modelling and implementing DSP functions in FPGAs. Model-based design: Matlab® is a popular mathematical modelling en-vironment. Simulink® supports the

simulation of continuous- and dis-crete-time systems with libraries for modelling DSP and communication systems, the capabilities for data analysis, and visualization. It is therefore a suitable platform for FPGA design tools. FPGA vendors have developed tools on top of Simulink® that support system mod-elling. The tool consists of parame-terized IP models representing some DSP operations such as Fast Fourier Transforms (FFT) or FIR filter func-tions. The major problem with this technique is that the transition from the algorithmic domain and the im-plementation domain is not com-pletely automatic: many low-level aspects of the model have to be manually handled.

C/C++ to RTL: There are tools on the market allowing the synthesis of Register Transfer Logic (RTL)10) from C/C++ code. Some require addition-al architecture-specific information in the C source code to define con-currency and timing, whereas others allow the direct synthesis of RTL from ANSI C or C++.

DSP synthesis: DSP synthesis tools allow engineers to design and simu-late DSP algorithms at Simulink® level. An automated way of migrat-ing the design to implementation level (RTL) is also supported. The key feature of the solution is a

Xilinx has traditionally been the FPGA leader.

Altera is the second FPGA heavyweight.

Lattice Semiconductor focuses on low-cost, feature-optimized FPGAs and non-volatile, flash-based FPGAs.

Actel has antifuse and repro-grammable flash-based FPGAs.

QuickLogic has antifuse (pro-grammable-only-once) products.

Cypress Semiconductor Atmel focus on providing AVR Microcontrollers with FPGA fabric on the same die.

Achronix Semiconductor has very fast FPGAs in development.

FPGA manufacturers

4 Traditional DSP processor (left), FPGA solution (right) allowing parallel processing

FPGA

MAC operations in 1 clock cycle

DSP

Data in

Loop algorithm n times

Reg

MAC

Data out

x

+

x

Data in Reg 0 Reg 1 Reg 2 Reg n

x x x

+

Data out

C 0 C 1 C 2 C n

3 Implementing signal processing algorithms in an FPGA instead of a DSP gives the designer additional degrees of freedom: structure optimized for speed (extreme left) or cost (extreme right)

Semi-Parallel SerialParallel

DQ DQ

Embedded system technologies

Page 74: Abb Embedded Sym

74 ABB Review 2/2006

Coming of age

“block set” that is avail-able in Simulink. The DSP designer therefore enters the process only at the al-gorithmic level and does not have to deal with defi-nitions at low-level imple-mentation decisions. The only attributes needed are the filter coefficients and gain requirements . The hardware design engineer adds the desired sample rate, speed, and target technology of the design. The tool then generates the appropriate RTL. This method is more or less equal to model-based de-sign. However, its biggest advantage over model-based design is that the models have fever low-level param-eters and are not vendor specific: any FPGA target with DSP features can be used.

Typical ABB applicationsABB has developed its own IPs11) for using FPGA technology in power net-work monitoring and controllers. For example, a typical multiprocessor product in ABB contains a microcon-troller (MC) that handles display, de-vice configuration and communication details; while a DSP takes care of all computations. These architectures may be combined in an MC-FPGA so that the computations are done in an FPGA rather than a DSP. A successful example of this architecture is the

new SAHIB platform developed by ABB Corporate Research and the com-pany’s Automation Technologies busi-ness as a general-purpose platform for use in both the Power and Automa-tion Technologies segments.

FPGA technology is also applied in: Power electronics and medium volt-age drives

Inverter control (eg, modulators, switching logic, and protection) and communication

Motor Control (eg, modulators, direct torque control 2/3/5 level inverters)

Medium voltage products for auto-mation analogue data acquisition, down sampling, filtering, RMS calculations, and protection func-tions

High voltage products to perform system control of HV switchgear drive, as well as analogue data acquisition and down sampling A product family for ana-logue data acquisition and protection known as SLIM-LINE 5 . The FPGA is used for down sampling, filtering, and RMS calculations WISA [1] (Wireless Interface to Sensors and Actuators)

Thanks to the advent of new technologies in the field of FPGAs, designers are now provided with an option other than ASICs. FPGAs have im-

proved their capacity to build systems on a chip with more than million ASIC equivalent gates and a few megabits of on-chip RAM, making them more than suitable for low volume produc-tion

Erik Carlson

ABB Corporate Research

Billingstad, Norway

[email protected]

Franz Zurfluh

ABB Corporate Research

Baden-Dättwil, Switzerland

[email protected]

Catherine Körbächer

ABB Review

Baden-Dättwil, Switzerland

5 A typical embedded board with FPGA, xScale microprocessor and communication support, all part of ABB’s SLIMLINE project for LV protection

Footnotes

References

[1] ABB Review Issues 3 and 4 2005, “Unplugged but connected”, Parts 1 and 2.

1) An ASIC (application-specific integrated circuit) is

an integrated circuit (IC) customized for a particular

use rather than intended for general-purpose use.

For example, a chip designed solely to run a cell

phone is an ASIC.2) Moore’s law is an empirical observation that at the

rate of technological development, the complexity

of an integrated circuit, with respect to minimum

component cost, will double in about 18 months. It

is attributed to Gordon E. Moore, a co founder of

Intel. 3) CORDIC stands for COordinate Rotation DIgital

Computer) and is a simple and efficient algorithm to

calculate hyperbolic and trigonometric functions. It

is the algorithm of choice if no hardware multiplier is

available.

4) A phase-locked loop (PLL) is a closed-loop feedback

control system that maintains a generated signal in a

fixed phase relationship to a reference signal.5) A device that reduces clock skew in digital circuits.6) The Peripheral Component Interconnect standard

(PCI) specifies a computer bus for attaching periph-

eral devices to a computer motherboard.7) The Unified Modeling Language (UML) is a non-pro-

prietary, object modeling and specification language

used in software engineering. UML has its strengths

at higher, more architectural levels and has been

used for modeling hardware.8) Standard Delay Format (SDF) is an IEEE standard

for the representation and interpretation of timing

data for use at any stage of an electronic design

process.

9) von Neumann architecture refers to a computer

design model that uses a single storage structure

to hold both instructions and data. The separation

of storage from the processing unit is implicit in

the von Neumann architecture.10) Register Transfer Logic (RTL) is a description of a

digital electronic circuit in terms of data flow be-

tween registers. The RTL description specifies

what and where this information is stored and how

it is passed through the circuit during its operation.11) A well designed IP should: include a test bench;

be reusable in several products; and be easily

upgradeable to new generations of the FPGA.

Embedded system technologies

Page 75: Abb Embedded Sym

ABB Review 2/2006

Thema

Basic and advanced signal processing algorithms run in a large variety of ABB products that are equipped with embedded electronics, from small domestic motion detectors, to sophisticated control units for medium- and high-voltage switchgears. In field devices in particular, signal processing provides an opportu-nity to improve the quality of measurements and the overall functionality of instruments.

The PILD (plugged impulse line diagnostics) algorithm is an example of such an improvement. It has been developed to alert operators to blockages in the impulse lines of pressure transmitters. Such a warning system allows users to switch from preventative maintenance programs to more cost-effective, event-driven, predictive practices.

Signal Processing in Embedded Systems Signal processing as an opportunity to increase functionality of industrial instrumentationAndrea Andenna

75

Page 76: Abb Embedded Sym

76 ABB Review 2/2006

Signal Processing in Embedded Systems

Signal processing usually brings to mind audio applications, image

processing or communication technol-ogies, but a glance at ABB’s product portfolio reveals a much wider pic-ture. Signal processing applications are found in many of ABB products, both in automation and power tech-nology. Many of the applications are integrated in devices, such as control units and industrial instruments, and run on embedded platforms.

Power line communication modems, for example, use a wide variety of digital signal processing (DSP) algo-rithms. Key topics are digital modula-tion and demodulation, digital filter-ing, Fourier transforms, sampling rate conversion, frame acquisition, carrier phase and symbol timing synchroniza-tion, channel estimation and equaliza-tion, and error detection and correc-tion. The basic principles of signal processing are well established, and used in all of today’s communication systems. However, a considerable R&D investment is required to meet the increasing requirements of power-line communication systems. Increas-ing processing power will support higher data transfer rates per channel, and channel bandwidths will increase from the traditional 4 kHz to ~32 kHz. In the long-term future, single systems may offer flexible (configurable) sup-port of much higher bandwidths, up to 1 MHz. Such truly broad-band power line modems will have to implement fur-ther efficient signal process-ing algorithms.

Today’s protection and con-trol units for switchgears and circuit breakers provide a large variety of electronic protection functions for the electrical systems they su-pervise. These devices work by measuring current and voltage and then digitalizing and processing the acquired signals. This is generally achieved by means of Fouri-er analysis: the harmonics of the electrical signals are computed and become the major inputs for most of the protection functions. These include over-current, over-

voltage, and differential and distance protections.

ABB field devices and analytical in-struments are normally equipped with an electronic part that acquires signals from the sensing part of the device: pressure transmitters, for example, acquire a signal from a piezo-resistive sensor chip, magnetic flow meters read the voltage induced by the gen-erated magnetic field, temperature transmitters read the signal from a thermocouple. So, in general, inside an industrial instrument, one or more electrical signals are acquired from the sensing part (sometimes referred to as the primary part) by the elec-tronic part (the secondary part). In general, all these sensor signals need to be amplified, analogue filtered, converted from analogue to digital and then digitally processed in micro-processors or DSS. Signal processing is also important in today’s sensor systems for modelling the sensor char-acteristic curves, to compensate for non-linearity and influencing effects.

Signal processing is an opportunity for improving field devicesField devices are becoming more intel-ligent, mainly because of rapid im-provements in the semiconductor in-dustry, particularly in terms of cost and power consumption of the com-ponents. In this context, signal pro-

cessing provides an opportunity to im-prove sensor properties, in spite of the abundance of influencing effects, such as manufacturing variance, hysteresis, drift, ageing and cross sensitivity, which are unavoidable and represent a systematic source of uncertainty [2]. Additionally, customers now require industrial instruments with an extend-ed set of functions, besides the prima-ry goal of the device. Device and pro-cess diagnostic functions are particu-larly appreciated because they promise to reduce maintenance costs and to improve the general reliability of the instruments. Competitors are clearly confirming this trend and “diagnostics” is now a common keyword in the mar-ket requirement specifications of new generation instruments. Up to now, this process-supervision functionality has usually been provided at the con-trol-system level of a plant, where much higher computation power is available. But improvement of the em-bedded platforms now allows integra-tion of complex algorithms at the de-vice level, rather than in PCs and con-trol systems. In other words, the cur-rent trend is to shift intelligence from the system down to the field devices and instruments. The last part of this article discusses an example of this.

Limitations of the embedded platformsIt is well known that electronic compo-nents such as processors, memories

and chips, have been improv-ing dramatically for a number of years, increasing their per-formance and reducing their size and cost. This applies to every chip market segment, from personal computers to smaller embedded architec-tures of industrial applications. Nevertheless, in the embedded platforms typically used in ABB devices and instruments, cost and power consumption remain a challenge:

In the industrial instrumenta-tion market, price plays a very important role in the maintenance and increase of market shares. Very often, competing products are com-parable in quality and cus-tomers’ decisions are based mainly on price. However, as

Embedded system technologies

1 Differential pressure transmitter in a harsh environment: access for maintenance is hard

Page 77: Abb Embedded Sym

77ABB Review 2/2006

Signal Processing in Embedded Systems

described above, the general trend for chips is to become cheaper, and, while the cost of electronics normally represent a significant portion of the production costs for a device, the manufacturing and material costs of the instrument can sometimes be much higher. Therefore, from a cost point of view, today’s embedded ar-chitectures for industrial instrumenta-tion have the potential to improve cal-

culation power and memory: more advanced algorithms and additional intelligence can be added easily.

Many embedded architectures have limitations on the power they can con-sume. For example, battery-powered devices have specific battery life re-quirements and are, therefore, limited in terms of power consumption. There are devices that normally work with an

auxiliary power supply (110 / 220 V) but which, in case of emergency, must work without this supply, albeit with limited functionality. This is the case for many control units for circuit breakers. The solution here is either a battery or a self-supply strategy, (eg, power taken from the current flowing through the circuit breaker).

Many instruments are supplied through the 4–20 mA channel, which is also used as the main analogue input or output channel. These devices, known as two-wire instruments, can consume only a few tens of milliwatts. The in-trinsic safety provided by this low power consumption is an advantage for industrial two-wire instruments and is actually one of the key reasons that customers still strongly support this type of power supply. However, power consumption became a limiting factor for the improvement of electronics, and therefore functionality, some years ago, and it remains a particular prob-lem for two-wire devices.

Signal processing pro-vides an opportunity to improve sensor proper-ties, in spite of the abun-dance of influencing effects, such as manufac-turing variance, hysteresis, drift, ageing and cross sensitivity.

An embedded signal processing application: PILDThe PILD (plugged impulse line diag-nostics) function is a signal processing algorithm that was recently integrated in ABB differential pressure transmit-ters, one of the most commonly used field device types. This R&D project showed both the potential of signal processing for improving field devices and also the constraints imposed by their limited embedded architectures.

Differential pressure transmitters are instruments for sensing the difference in pressure between two points of a process. They can be installed in harsh environments where access for maintenance can be difficult 1 . The

2 Noise level in the differential pressure signal under various conditions in the impulse lines

0.02

0.015

0.01

0.005

0nois

e po

wer

/sig

nal p

ower

0 50 100 150 200 250 300

Time (s)

a Lines not plugged

0.02

0.015

0.01

0.005

0nois

e po

wer

/sig

nal p

ower

0 50 100 150 200 250 300

Time (s)

b Both lines plugged

0.02

0.015

0.01

0.005

0nois

e po

wer

/sig

nal p

ower

0 50 100 150 200 250 300

Time (s)

c (+) line plugged

0.02

0.015

0.01

0.005

0nois

e po

wer

/sig

nal p

ower

0 50 100 150 200 250 300

Time (s)

d (-) line plugged

Embedded system technologies

Page 78: Abb Embedded Sym

78 ABB Review 2/2006

Signal Processing in Embedded Systems

main application of this device is computing the flow rate inside a pipe-line. This is achieved by measuring the pressure drop caused by a primary element, typically a Venturi tube or an orifice plate. Through this measure-ment, and with a knowledge of the geometry of the primary element, the flow rate can be computed.

Differential pressure transmitters are connected to the process through two pipes called impulse lines. They nor-mally have a small diameter, less then 1 cm, and can be very long. During the life of a device, the impulse lines can become partially or completely clogged by solid process material (eg, sand), sediment, or deposits that build up inside the lines, or by frozen water.

Differential pressure trans-mitters that automatically identify plugged impulse lines have the potential to cut costs by reducing preventative maintenance efforts.

In contrast to most other field device malfunctions, a plugged impulse line has no impact on the device hardware and, if it goes unnoticed, the process value will remain in a valid condition. By plugging the impulse line, the cur-rent pressure state becomes trapped and decoupled from the true process state. The control system continues to use the pressure value in control

loops, not realizing that it is “frozen”. The only indication the process opera-tor has for such an event is the misbe-haviour of control loops, which could also, and is actually more likely, to be caused by valve wear.

The maintenance effort required to identify and unblock a plugged impulse line is high. Moreover, if a process fluid has a known tendency to cause plug-ging, costly preventive maintenance will usually be carried out. Differential pressure transmitters that automatically identify plugged impulse lines have the potential to cut costs by reducing pre-ventative maintenance efforts.

The PILD alogrithmThe principle of plugged impulse line detection is based on the observed characteristics of pressure signals over time. Flow processes are affected by fluctuations in the pressure value caused by other devices and ma-chines, such as pumps, that interact with the process. These fluctuations can be seen as noise in the differen-tial pressure signal. Under normal op-erating conditions, with clear impulse lines 2a , this process noise is mostly cancelled out because the device mea-sures pressure from two points that are relatively close together, normally only a few centimetres apart. If one impulse line becomes blocked 2c 2d , the pressure fluctuations are no lon-ger cancelled out and the process noise is fully apparent in the differen-tial pressure signal. If both impulse lines become blocked 2b , the process noise will be reduced almost to zero

because the pressure connection be-tween sensor and process will be lost completely.

So, the PILD function first measures the noise level in the differential pres-sure signal when the impulse lines are clear (training phase). Then, during normal device operation, it statistically compares the noise level with values stored during the training phase. If the statistical analysis shows a significant difference between the live values and those acquired during the training phase, an alarm signals that one or both impulse lines are plugged.

The training phase is a configurable time period, during which the algo-rithm “learns” the nominal process conditions so that it can later identify readings that indicate plugging of the impulse lines. A reliable and effective training is the key to the success of the PILD function. Differential pres-sure transmitters are used in very dif-ferent process conditions, in terms of media (high viscosity liquids, water, steam, gases etc) and environmental conditions (temperature from – 40 to 85° Celsius, and absolute pressure up to 600 bars). Without an automatic way of adapting the algorithm to this large variety of conditions, the PILD function would be useless.

The PILD function was developed between 2003 and 2005. The function has recently been integrated in the new release of ABB 264 Differential Pressure Transmitters with Foundation Fieldbus interface.

Andrea Andenna

ABB Corporate Research

Baden, Switzerland

[email protected]

References

[1] Hengjun Zhu, E. H. Higham, J. E. Amadi-Echendu,

Signal Analysis applied to Detect Blockages in

Pressure and Differential Pressure Measurement

Systems, IEEE Instrumentation and Measurement

Technology Conference, Proceedings Vol. 2

(1994), Pages 741–744.

[2] H. Tränkler, O. Kanoun, “Importance of Signal Pro-

cessing in Sensor Systems”, Technisches Messen

71 (2004) 3

[3] A. Andenna, G .Invernizzi, D. Eifel, “Embedded di-

agnosis to detect plugged impulse lines of a differ-

ential pressure transmitter”, ITG-/GMA Sensoren

und Messsysteme 2006, Conference Proceedings

Embedded system technologies

Page 79: Abb Embedded Sym

79ABB Review 2/2006

Editorial Board

Peter TerwieschGroup R&D and Technology

Adam RoscoeCorporate Communications

Ron PopperGroup Editorial ServicesCorporate Communications

Friedrich PinnekampGroup R&D and Technology

Nils LefflerChief [email protected]

Publisher’s officeABB Schweiz AGCorporate ResearchABB Review/REVCH-5405 Baden-DättwilSwitzerland

The ABB Review is published four times a year in English, French, German, Spanish, Chinese and Russian.

Partial reprints or reproductions are permit-ted subject to full acknowledgement. Complete reprints require the publisher’s written consent.

The ABB Review is free of charge to those with an interest in ABB’s technology and objectives. For a free subscription please contact your nearest ABB representative or the publisher’s office.

Publisher and copyright ©2006ABB Ltd. Zurich/Switzerland

PrintersVo rarl berger Verl agsan stalt AGAT-6850 Dornbirn/Austria

LayoutDAVILLA Werbeagentur GmbHAT-6900 Bregenz/Austria

DisclaimerThe information contained herein reflects the views of the authors and is for informa-tional purposes only. Readers should not act upon the information contained herein without seeking professional advice. We make publications available with the under-standing that the authors are not rendering technical or other professional advice or opinions on specific facts or matters and assume no liability whatsoever in connec-tion with their use. The companies of the ABB Group do not make any warranty or guarantee, or promise, expressed or im-plied, concerning the content or accuracy of the views expressed herein.

ISSN: 1013-3119

www.abb.com/abbreview

Preview 3/2006

As the technical review of the ABB Group, ABB Review strives to maintain good coverage of research and devel-opment activities within the group. Its articles regularly cover trends, achieve-ments and applications both from the corporate research units and from ABB’s mainstream businesses. Howev-er, besides the business areas that the group is most commonly associated with, ABB is active in several highly successful niche businesses. The next issue of ABB Review will focus specifi-cally on these less-known areas of ABB.

Some of the company’s business endeavours have achieved remarkable results by transferring knowledge and solutions gathered in one area of the company’s activities to a totally differ-

ent fields – either within or outside ABB’s main business areas. Engineer-ing groups in local ABB companies have applied the company’s product port folio and know-how in creative ways to solve issues faced by custom-ers – including some very unorthodox applications.

ABB Review hopes that by providing insight into ingenious solutions ap-plied in one market, further opportu-nities for ABB’s knowledge pool can be opened elsewhere. The examples presented in this next issue should provide inspiration in creative thinking and open fresh areas for synergy and innovation – enabling more and more industries, products and people to benefit from the company’s vast pool of knowledge and experience.

Erratum

On page 8 of the printed version of ABB Review 4/2005, we included a section of text taken from the web-site of PBS of the USA without ac-

knowledging this fact. We apolo-gize for this oversight. (http://www.pbs.org/wgbh/nova/einstein/legacy.html)

Page 80: Abb Embedded Sym

Ensuring areliable powergrid is where

we really shine.

Ensuring areliable powergrid is where

we really shine.

© 2005 ABB

Providing breakthroughs in power transmission anddistribution for the 21st century. Visit us at www.abb.com

a