INTELLIGENT LED CONTROL RAPID PROTOTYPING FOR IOT WIRELESS PROTOCOLS

WINTER 2014 NEWS FOR THE ELECTRONICS INDUSTRY ISSUE 2

TECH JOURNAL

EXPLORING THE
INTERNET
OF THINGS
›LIFE ON THE EDGE OF
THE INTERNET OF THINGS
ULTRA-LOW POWER
›A CLASH OF
ARCHITECTURES
ARM vs X86
+
PLUS
n LAMP SERVERS FOR IOT
n ARM SYSTEM IP
n FPGA BASICS

WWW.ELEMENT14.COM/TECHJOURNAL element14.com
 A Place Engineers
Call Home
The element14 Community

Connect with 275,000+ members

element14
globally at the world’s largest community
DESIGN CENTER BETA
For the very latest Embedded & Analog development tools
START HERE
Research > Design > Production for design engineers. Discover the latest
development kits, single-board computers
www.element14.com/designcenter
and hot new products that you can
test & review for free at the element14
Design Center! Enjoy cutting-edge webinars,
participate in design challenges and more.

Join for FREE today

at element14.com

COMMUNITY
 Welcome
Issue 2 WINTER 2014

4 Exploring the Internet of Things

EXPLORING THE IOT FROM END TO END
IoT: we’ve all heard of it. Some say it’s here,
8 Rapid prototyping for IoT it’s coming, or that it has already come and
RAPID PROTOTYPING THE INTERNET OF THINGS
gone. Whatever view you hold, there is no
Life on the edge of IoT disputing the fact that the amount of information
10
UNDERSTANDING ULTRA-LOW POWER that we gather and have access to is evolving
exponentially. Embedded systems are key
16 Wireless protocols enablers of the Internet of Things. The question
WIRELESS PROTOCOLS EXPLAINED
is, how do we use these powerful embedded
22 The IoT cloud and analog products to gather, utilise and
SHINING A LIGHT THROUGH THE IOT CLOUD
disseminate all the types of information in
order to make IoT a reality?
24 LAMP servers for IoT
DEPLOYING A LAMP SERVER INTO THE IOT
In this edition of the element14 Tech Journal, we will look at
the core components that make up the Internet of Things.
30 Intelligent lamp control In Exploring IoT from End to End, we review what is needed
INTELLIGENT LAMP CONTROL WITH XMC1000 to make the Internet of Things a reality. One of the key
enablers are edge nodes or ‘things’ that take input from all

36 Efficient and intelligent lighting

around us. In Life on the Edge of IoT we explore how the
usage of ultra low power edge node devices are at the heart
USING INFINEON’S RGB LED LIGHTING SHIELD
of these end points and how engineers can better understand
the energy usage of processors in their applications. Where
38 Energy efficient white light this information runs through or is stored is the subject of
THE BIRTH OF ENERGY EFFICIENT WHITE LIGHT Shining a Light on the IoT Cloud where we look at three
models for cloud based services. One of the key methods of

40 ARM vs x86
transmitting data from the edge nodes is through wireless
communication. In Wireless Protocols Explained, we look the
A CLASH OF ARCHITECTURES
most common protocols as well as the parameters needed
to help you decide what works best for your application.
44 System IP Additionally, we have included information on the systems
ARM TECHNOLOGY PART 2 and additional technologies that are available to help you
better understand and bring your IoT designs to market faster.
We hope you enjoy this edition of element14 Tech Journal
and welcome your comments and suggestions. Please feel free
EDITOR-IN-CHIEF CLIFF ORTMEYER to drop us a note.
MANAGING EDITOR ANKUR TOMAR

© Premier Farnell Corp 2015. All rights reserved. No portion of this publication, whether in whole or in part, can be reproduced
without the express written consent of Premier Farnell Corp. All other registered and/or unregistered trademarks displayed in this
publication constitute the intellectual property of their respective holders. Errors and omissions in the printing of this magazine David Shen Chief Technology Officer, Premier Farnell
shall not be the responsibility of Premier Farnell Corp. Premier Farnell Corp reserves the right to make such corrections as may
be necessary to the prices contained herein. Email: editor-TJ@element14.com

WINTER 2014 ELEMENT14 3

EXPLORING THE
INTERNET OF THINGS
FROM END TO END
By STEVE NELSON / EXECUTIVE DIRECTOR, FREESCALE
In the recent past, devices were singular
entities going about their tasks in silent
anonymity. Sure, some industrial
systems and computers within
networked environments collected
data and created a community of sorts,
but nothing like what we have today.
4 ELEMENT14 WINTER 2014
WE NOW HAVE CONSTANT CON-
nectivity and devices that trade mas-
sive amounts of data; data that holds
secrets we are only beginning to tap.
Transportation, healthcare, agricul-
ture, machine-to-machine communica-
tions, and entertainment are segments
that are already reaping an informa-
tion harvest that was unimaginable
10 years ago. The glue that holds the
promises of this technology together
is the Internet of Things (IoT), a con-
federacy of devices, networks, and
processing power that has vast poten-
tial to make life better by address-
ing truly global problems. Freescale
has been a pioneer at every level of
the IoT phenomenon and offers all of
the fundamental IoT building blocks
under one roof.

Edge Nodes PAN/LAN
Connectivity
MCUs
Energy
Sensors
Connectivity
●●INFRASTRUCTURE OF THE INTERNET OF THINGS
Edge nodes
Sensors measurements and monitoring
The story of IoT begins at the
edge node, the entry point for all
information. This information
collection is the primary role of
the edge node and the possibilities
for measurement are limited only
by the imagination. Temperature,
pressure, motion, position and many
other factors can be measured,
processed and transmitted.
EDGE NODE
SOLUTIONS
IOT GATEWAY
SOLUTIONS
Gateway WAN Cloud
CLOUD
Connectivity Kinetis SOLUTIONS
microcontrollers
Design potential.
Application/Action
Big Data
Realized.
i.MX applications
Industry’s most scalable
ultra-low-power, processors QorIQ processors built on
mixed-signal MCU
solutions based on the Your interface Layerscape Architecture
ARM Cortex-M and to the world Accelerating the network’s IQ
Cortex-M0+
architectures. Industry’s most versatile Industry’s first software-aware,
solutions for multimedia core-agnostic networking system
and display applications, architecture for the smarter,
with multicore scalability more capable networks of
Freescale sensing and market-leading tomorrow, end to end.
Intelligent contextual power, performance
and integration.
sensing
The right combination
of intelligent
integration, logic and
customizable software
on the platform to
deliver smarter,
more differentiated
applications.
●●FREESCALE IOT
SOLUTIONS MAPPING
We can easily recognise many edge battery replacement. Their structures
node devices as integral parts of our must provide low complexity for inter-
lives, and the many forms of edge node operability and must be robust and
devices are leading the growth of the industrial grade for prolonged use
IoT space. Smart phones, thermostats, and reliability. These devices must
entertainment and health tracking be unobtrusive so that they can be
devices are a few familiar manifesta- incorporated into a small form factors
tions. While these applications are well and low cost for rapid development
understood, there are other devices and implementation.
that will soon impact life as we know There are two primary classes of
it. For example, automobiles monitor edge node devices: those that are bat-
road conditions and informatics around tery powered and those that receive a
driving. Cities are beginning to imple- steady stream of energy. The device’s
ment integrated traffic, weather and requirements will form the design
infrastructure monitoring systems parameters and as a result the power
that offer the potential of improved source. Is an edge node in a remote
safety and transportation efficiency. location where power will not be avail-
Healthcare providers are using monitor- able? Is the device mission critical in
ing devices to collect vital information such a way that its data collection
such as blood pressure and blood sugar could have an impact on safety or a
levels. These devices need the ability company’s bottom line? Is the reporting
to sense information, process it and of information only required at inter-
maintain connectivity with gateways. mittent cycles or is a steady stream of
The endpoint in edge nodes is usu- data communication required? Those
ally composed of sensors, an embed- that operate on a battery must push the
ded processor (typically an MCU), a absolute limits on power management
connectivity method and an energy technologies in order to both collect
source. The demands of endpoint data and then transmit this information
devices are familiar. They are ideal- on a regular basis, and also maximise
ly low power in order to operate for the time between battery charging
long periods between charging or or replacement. E
WINTER 2014 ELEMENT14 5
 With the proliferation of intelligence and
connectivity in previously unimagined
applications, time-to-market pressures will
intensify in the IoT space. Freescale is the only
supplier that delivers IoT solutions from the
edge node, to the network to the cloud. We
are focused on providing application-specific
reference designs and platforms that empower
developers with real-world implementations
and enable rapid design prototyping. n

Freescale offers a wide array

●●KINETIS KL03 MCUS FOR EXTREMELY LOW
POWER, SMALL FORM FACTOR APPLICATIONS
PROCESSING
The processing at the end node level is gaining
importance as devices have more sophisticated
processing power and hence the ability to com-
municate important trends or differentials. Once
data has been gathered, sensors usually conduct
basic algorithms such as threshold detection and
simple data analysis. This is important as the IoT
landscape is moving from higher-level, cloud-
based processing to a distributed intelligence
model in which data-driven decision making is
migrating toward the edge nodes. The ultimate
goal is that they will begin to learn, adapt and
act independently in a predictive manner. For
example, in a healthcare application the process-
ing might involve triggering communication if
blood pressure is above an advisable level, but
a constant transmission of blood pressure val-
ues that are within limits and are not changing is
unnecessary. In a smart city application the sens-
ing could involve monitoring and making sense
of temperature, moisture wind or infrastructure
changes. The key is that these devices operate
based on the exception, not the rule. Constantly
reporting a non-changing value is of little impor-
tance. That the value is changing, and maybe by
how much and how quickly, is much more rel-
evant information.
Freescale’s embedded processor solutions
deliver optimised performance and power to
match virtually any IoT application, from battery-
powered consumer devices to cars, homes, civil
engineering and industrial automation. Freescale
offers the smallest package for an MCU, the CSP
that measures only 2x2mm with the full capa-
bility of a 32-bit microcontroller and the low-
power performance of the ARM Cortex-M0+,
that not only will improve the accuracy of the
acquired data, but also improve the battery life
and achieve the smallest form factor so that the
patient can forget he is wearing a device at all.
●●KINETIS KW2X
MCUS FOR EDGE
NODE APPLICATIONS
CONNECTIVITY
The key to IoT is the connection of disparate
devices with multiple hardware and software
approaches. Without dependable connectivity
the web of developing information is dead in the
water. A product can be designed to reliably con-
nect to a certain set of gateways or other devic-
es. But, in the real world, a device may connect
to any number of unknown products, which is
why standards and compliance are critical.
Freescale has worked hard in developing and
engaging in partnerships and important stand-
ards bodies. Freescale connectivity solutions
work seamlessly with a number of protocols. On
the consumer side Freescale has been a partner
and developer with consortiums such as Wi-Fi,
Bluetooth Smart, ZigBee, 6LoWPAN and HPGP.
On the industrial side Freescale has been instru-
mental in Wireless Hart, ISA100, EtherCAT, and
Modbus. Freescale’s networking processors are
designed with a systems view to deliver high-
performance connectivity that scales with the
needs of the emerging infrastructure demands of
the IoT market.
The future of connectivity will involve the forward
looking partnerships such as Thread. These part-
nerships will allow a host of devices to not only
network but create mesh environments where
information moves in a web across devices. This
type communication requires processors that can
meet demanding throughput requirements with
robust real-time, point-to-point communication.
of software-operating systems,
protocol stacks, middleware and
application software that assist
in overcoming development
hurdles, and we engage the right
partners to further streamline
development and reduce risk.
●●QORIQ LS1
TOWER SYSTEM
MODULE FOR GATEWAY
DESIGNS (TWR-LS1021A)
GATEWAY
As with many elements of IoT, the gateway is
also a shape-shifting component, free from form
factor constraint. Gateways are not required to
assume a standard form and can manifest in dif-
ferent ways. Even a cell phone can act as a gate-
way in some instances. The gateway is not a one
system fits all, there are a range of possibilities,
depending on the additional functions that they
gateway may perform.
The gateway allows for the transferal of data
through a variety of protocols. As the gateway
appliance takes information, it applies a minimal
level of processing and further conveys data on
its way to further processing. This ecosystem of
the gateway is composed of a number of com-
plex relationships including software, processors,
security and protocols.

6 ELEMENT14 WINTER 2014

 Intelligent design,
Incredible scale

Make it with Kinetis Microcontrollers

If you can imagine It, you can build it with Freescale Kinetis MCUs,
the world’s most scalable, energy-efficient MCU portfolio.
Make it smarter — Create smaller, cheaper, greener and longer-lasting products
with the world’s most energy-efficient portfolio of ARM Powered® MCUs.
Make it faster — Reuse your code from prior designs—regardless of size or
power—instead of starting from scratch every time.
Make it smaller — Enable intelligent processing and connectivity in mobile
and other compact designs with the industry’s smallest form factor MCUs.
Make it easier — Access the industry’s easiest-to-use, most widely supported
development tools to reduce costs and speed your time to market.
Whatever your design needs, wherever you are in the world, we’re here to
help you make it better. Learn more at freescale.com/Kinetis

Freescale, the Freescale logo and Kinetis are trademarks of Freescale Semiconductor, Inc., Reg. U.S. Pat. & Tm. Off. ARM is the registered trademark
of ARM Limited. ARM Powered is a trademark of ARM Limited. All other product or service names are the property of their respective owners.
© 2014 Freescale Semiconductor, Inc.
 RAPID PROTOTYPING THE
INTERNET OF THINGS
O HELP DEVELOPERS
T take advantage of the green
field that is IoT, semicon-
ductor vendors are developing low
cost, easy to use baseboards and
By ADRIAN FERNANDEZ / PRODUCT MANAGER, TEXAS INSTRUMENTS leaving the extended functions like
wireless connectivity, LED and LCD
controls and the like, to be integrat-
ed onto add-on cards that designers
can then add to their system if need-
ed. An example of this type of base-
board and accessory board system
that is designed for rapid prototyp-
i n g i s t h e L a u n c h Pa d a n d
BoosterPack ecosystem from Texas
Instruments. These development

Rapid Prototyping.
Quicken time to market.
Sink or swim.
The onset of the Internet
Micr
ophone senso
r RF B
ooster & Potenti
om
of Things is exciting
r&
ste
ete
and developers are
oo

hard at work to add

RF B

cloud‑connectivity to
existing applications
or inventing new
solutions all together.
From wearables to
home automation to
the tweeting fridge, the
cloud is poised to add
ode
CC

00
kN
32

or
+C et w new intelligence and
RF B

C110 N
L ( R F ) C e n tral
oos

value-add to a broad
or

&H
ter

se
ns

um IR
i d it y
sensor R F B o oster &
P set of solutions.

RF r RF r
Boos nso Boos nso
ter & Lig ht s e ter & R a n g e s e OVERVIEW OF A CLOUD-CONNECTED WIRELESS
SENSOR NETWORK THAT WAS RAPIDLY PROTOTYPED
USING OPEN SOURCE MODULAR HARDWARE.

8 ELEMENT14 WINTER 2014


VISUALISING THE SENSOR DATA IN OUR SUBGHZ RF STAR NETWORK THAT WE ARE AGGREGATING
USING THE CC3200 WI-FI LAUNCHPAD, WHICH SERVES AS OUR SUBGHZ-TO-WI-FI GATEWAY.
tools are open source, modular and
affordable, which enables developers
to plug together various sub-circuitries
to build complete systems. Where addi-
tional functionality is known to be need-
ed, such as in the case of low-power
wireless connectivity to enable IoT
endpoint communications, the base
boards may integrate additional func-
tionality such as the TI CC3200 Wi-Fi
LaunchPad, which is a complete devel-
opment kit that features the CC3200,
a single chip microcontroller + Wi-Fi
network processor. Many of the base-
boards also include an on-board debug-
ger/programmer as well as a stand-
ardised interface for accepting their
accessory boards or plug in modules.
Some manufacturers such as TI have
developed their own standardised type
of plug in modules called BoosterPacks.
The benefit of this type of system is
that a designer is assured to have a
tested accessory solution that they can
get up and running with little effort.
Many manufacturers have chosen to
use add-on boards with footprints that
might be compatible with a wide range
of accessories from other manufactur-
ers (such as Arduino Shields) which
can potentially broaden the field of
applicable accessories, with the limi-
tation however that not all accessories
may work with the baseboard without
additional software/driver porting.
An example of how this type of rapid
prototyping system can work is where
designers may need to develop an
application that is out of their normal
expertise, such as the development
of a complete wireless sensor net-
work. Here we have started with the
TI CC3200 Wi-Fi LaunchPad which will
serves as a sub-1GHz-to-Wi-Fi gate-
way. By adding a CC110L SubGHz
RF BoosterPack on top of the CC3200
Wi-Fi LaunchPad, we are able to eas-
ily create a cloud-connected gateway
that operates as the centre of a sub-
1GHz RF star network. In this exam-
ple, each sensor node is made up of
a Texas Instruments Microcontroller
LaunchPad kit + CC110L SubGHz RF
Wireless BoosterPack + a sensor. In
this configuration, each node sends its
sensor data to the CC3200 LaunchPad
via SubGHz RF. The CC3200 receives
the data, encodes the datapoints into
a JSON object and publishes it to the
cloud over Wi-Fi via MQTT, a light-
weight publish/subscribe communi-
cation protocol.
Once the data is in the cloud, we
can visualise the real-time sensor data
on a cloud-side dashboard as shown
above. In this example, we were able to
Texas Instruments
CC3200 LaunchPad
element14.com/LaunchPad
Created for the Internet of Things, the
CC3200 LaunchPad is a wireless MCU that
integrates a high‑performance ARM Cortex-M4.
CC3200 Wi-Fi application processor, Standalone
development platform featuring sensors, LEDs and
push-buttons, Supports 4 wire JTAG and 2 Wire
SWD, GDB (GNU Debugger) support over OpenOCD.
quickly create a public web dashboard
that enables anyone anywhere in the
world to visualise and monitor the data
in the star network that we created.
While modular, open source hardware
is important to enable efficiency in a
fast paced industry, open source and
easy-to-use software examples are just
as critical.
To create the system above, we used
a tool called Energia (energia.nu).
Energia is an open source, community
driven software development envi-
ronment that is a fork of the popular
Wiring/Arduino framework. Through
software abstraction, Energia offers
intuitive APIs and libraries that are at
the functional level, enabling develop-
ers to focus on what they want to do
as opposed to spending time figuring
out how to actually do it.
With the help of these rapid prototyp-
ing tools, developers are well-equipped
to solve the problems of a connecting
world. Through open source method-
ologies, developers are able to freely
share and collaborate to develop the
building blocks we need to build the
IoT. Hardware and software develop-
ers are able to benefit from these tools
that bring powerful technology to a
growing developer community and
lower the barrier of entry. n
WINTER 2014 ELEMENT14 9
Powering loT for Next-Gen Designs
Atmel® gives Makers and designers all the basic building blocks for Internet of Things (loT) applications. This includes everything from
embedded processing and connectivity to sensors, security, and software. We tie it all together with a rich ecosystem of design tools
and development partners to bring innovation, ease-of-use, and integrated solutions to your next big loT design.

Embedded Processing Security

Atmel AVR and Atmel | SMART ARM -based MCUs and
® ®
Atmel CryptoAuthentication™ devices provide protected
MPUs deliver a unique combination of performance and power hardware key storage and offer the ideal way to ensure
efficiency. Flexible and highly integrated, Atmel devices are products are genuine and that the data transferred back and
designed to optimize your system control, connectivity, and forth is confidential and has not been tampered with. Atmel
user interface management, all while delivering low power takes care of all the ultra-secure crypto engineering, and
and ease-of-use. delivers tiny protected hardware devices that come complete
Featured Products: with all the tools needed to make it easy to add security into
any digital system, without the need to be a crypto expert.
• Atmel | SMART SAMA5 ARM Cortex®-A5 series of MPUs
feature high-performance, market leading low-power, and Featured Product:
advanced security for cost-sensitive industrial and consumer • ATECC108 provides a robust,
applications cost-effective, and easy-to-use
• Atmel | SMART SAM D ARM Cortex-M0+ based MCU hardware security that offers
series sets new benchmarks for flexibility and ease-of-use best-in-class feature sets and is
while combining performance and energy efficiency supported by industry-leading
personalization services for added
• megaAVR 8-bit MCUs deliver low power, additional analog value and trust
functionality, pin-to-pin compatibility, and easy-to-use
interface for any smart and connected design
Communication
Sensing Atmel offers SmartConnect solutions in SoC (systemon-chip)
Onboard sensors are an essential part of the loT, giving and modules that provide self-contained, low-power, and FCC
connected devices the ability to track and respond to certified products, bringing wireless Internet connectivity to
environmental conditions. Atmel offers a complete system any embedded design, without you needing 802.11, IP stack
solution running on the Atmel platform allowing you to focus or RF design experience.
on differentiating your product features, enhancing the user Featured Product:
experience, and providing faster time-to-market.
• SAM W25 Wi-Fi module is a
Featured Product: fully-integrated single-source
• Atmel | SMART SAM G MCU + IEEE 802.11 b/g/n Wi-Fi
ARM Cortex-M4 based MCUs are solution. Use Atmel Studio 6
optimized for ultra-low-power and development tools to provide
high-performance applications. They battery-powered endpoints that
are available in small form factors with a last for years
highly-efficient set of serial peripherals
By MARKUS LEVY / PRESIDENT, EEMBC
LIFE ON THE
EDGE OF IoT
UNDERSTANDING
ULTRA-LOW POWER
THE SIMPLIFIED STRUCTURE OF THE INTERNET
OF THINGS CONSISTS OF THE CLOUD, THE HUBS
(COLLECTORS OR AGGREGATORS) AND THE EDGE
NODES (OTHERWISE KNOWN AS THE ‘THINGS’)
ANY ‘THINGS’ LIV-
M
ing on the edge must
utilise ultra-low
p owe r m i c ro c o n -
trollers to handle the
user interfaces, collect and trans-
mit sensor data, provide security
functions, and manage other tasks.
One issue faced by ‘thing’ design-
ers is ensuring that these micro-
controllers are really optimised to
meet the performance needs of
their applications and to enable the
long battery lives that are expected.
Edge nodes can be grouped more
or less arbitrarily according to their
application domain. For example,
home automation encompasses
anything that is used to control or
monitor the home or office systems
and devices, such as lighting or
environmental control, appliances
(e.g. freezer, washing machine, cof-
fee maker, or fire alarm. On the other
hand, ‘wearable’ or ‘portable’ is
anything that is worn or carried on
the person while in use. Examples
include smart watch, smart glasses,
heart rate monitor, pedometer, GPS
tracking device, blood sugar moni-
tor, music or video player, and wire-
less headset/microphone. There are
also categories for health, environ-
mental, and others.
Many edge nodes in the ‘weara-
ble’ domain, as well as other domains
can be considered ultra-low power
applications. These applications
are typically running with battery-
powered operation and implement
short, occasional periods of activity
interspersed with long periods of
inactivity, and possibly infrequent
human intervention. It’s this latter
point that highlights energy efficien-
cy as a key performance criterion
for such devices, and dictates bat-
tery lives of weeks, months, years,
or even decades (i.e. a smartphone
is not an ultra-low power device,
although some companies would
choose to market it as such). E
WINTER 2014 ELEMENT14 11
 WHAT IS ULTRA-LOW POWER?
What is this ‘holy grail’ of ultra-low power? At a fundamental level,
ultra‑low power implies different things to different applications.
The lowest active current is required when the power source is
severely limited (e.g. energy harvesting). Alternatively, the lowest
sleep current is required when the system spends most of its time
in standby or sleep mode, waking up infrequently (periodically
or asynchronously) to process some workload. Furthermore,
ultra-low power can also imply great energy efficiency, whereby
the most work is performed in a limited time period.

IoT applications can be designed variety of things. Other factors

utilising a combination of the above
attributes, but it requires careful
analysis and balancing of trade-
offs. Which brings up the ‘shocking
truth’ that an increasing number of
microcontroller vendors are ‘claim-
ing’ ultra-low power capabilities
and application developers should
not rely on datasheet parameters
alone. The ultra-low power claims
can often be connected to reality
due to features supported by mod-
ern-day microcontrollers. For exam-
ple, devices can support super-opti-
mised power modes: although some
devices have such elaborate power
modes it takes great effort to deter-
mine the optimal approach (it’s a
balance of the application’s active-
mode duty cycle and the ability to
execute the application correctly;
in other words, ensuring that the
device returns from its low power
mode in the required latency).
Modern-day microcontrollers
also sport a variety of intelligent
peripherals that serve to offload
the CPU. These peripherals perform
functions such as analog-to-digi-
tal conversion (ADC), pulse-width
modulation, and communication
protocols. In IoT applications, the
ADC could be used for sensors
handling pressure changes, moni-
toring temperature, or an infinite
that help to yield ultra-low power
claims include integrated memory
subsystems, extensive clock gat-
ing, and of course, advanced pro-
cess technologies.
From the discussion in the last
few paragraphs, you might see
why developers should not rely
on datasheet parameters alone to
compare and select an appropriate
microcontroller or to determine the
microcontroller’s real-world func-
tionality. The bottom line is that
‘energy efficiency’ is a nebulous
phrase. Most vendors utilise differ-
ent specifications when quantify-
ing parameters. What workload
should be used for the analysis?
It could be something as simple
as a few lines of code running a
‘while’ loop. Or perhaps it’s some-
thing a bit more real-world. Besides
the workload determination, what
should the conditions and the physi-
cal setup be for the device under
test? What duty cycles should be
used to represent the transitions
from active to low power state?
How should the clock source be
utilised? Should the workload run
from flash or RAM? What is the
input voltage? All of these details
must be specified and utilised by
all vendors in order to allow you to
make apples-to-apples comparisons.

12 ELEMENT14 WINTER 2014

The path to standardised
datasheet parameters
Atmel
In 2013, vendors including Analog Devices, ARM, SAMA5D3 Xplained
Atmel, Cypress, Freescale, Microchip, Renesas, element14.com/SAMA5D3
Silicon Labs, Spansion, STMicroelectronics, and The SAMA5D3 Xplained is a rapid
prototyping and evaluation platform
TI, began work on an industry standard method for microprocessor‑based design.
to quantify energy efficiency for ultra-low power ■■
SAMA5D36 ARM Cortex-A5 Microprocessor
microcontrollers. The outcome of this is the EEMBC ■■
2GBit DDR2 – Micron, 2GBit Flash – Micron

ULPBench, a benchmark providing a consistent ■■

Gigabit Ethernet, 10/100 Ethernet
USB Device connector, 2 USB Host connectors
method to measure energy efficiency, demonstrating
■■

■■
Arduino R3 compatible headers
both active power and idle (sleep) power.

The workload of ULPBench core
profile consumes 10,000-20,000 CPU
cycles each duty cycle, depend-
ing on the microcontroller’s archi-
tecture. Although the number of
CPU cycles varies, each microcon-
best to compile the code for maxi-
mum performance (minimise the
active cycles). Not only can this Texas Instruments
be accomplished by selecting the
CC3200 LaunchPad
most optimal compiler flags, but
the selection of the compiler itself element14.com/LaunchPad

troller is required to perform the
same amount of work (as opposed
to necessitating an arbitrary num-
ber of cycles). The diagram below
portrays the duty cycle, whereby
the device enters the active mode
once per second; the testing setup
requires a real-time clock (RTC) for
waking up, typically with an accu-
racy of 20-30 ppm.
There are no golden rules for
achieve the best ULPBench results.
Obviously, since the active mode
makes a big portion of the ener-
gy consumed during the test, it’s
is very important (just like it is for
any code development).
If the main (faster) crystal oscil-
lator is power hungry, another way
to lower energy consumption is
to try and run from a low power
32kHz crystal with the main crystal
oscillator switched off completely.
Furthermore, if the device’s leakage
current during the active state is
high, it might make sense to switch
to a faster clock during active state
and finish processing the workload
quicker (and in turn return to sleep
mode quicker).
Created for the IoT, the CC3200 LaunchPad
is a wireless MCU that integrates a
high‑performance ARM Cortex-M4.
■■
CC3200 Wi-Fi application processor
■■
Standalone development platform featuring
sensors, LEDs and push-buttons
■■
Supports 4 wire JTAG and 2 Wire SWD
■■
GDB support over OpenOCD
E

Freescale
Freedom KE06
Approx 10k to 14k Clock cycles on Cortex-M processors (0 wait states) element14.com/FreedomBoard
Featuring Freescale Kinetis E low-power, highly
1 SECOND robust, mixed-signal 32-bit microcontroller built
on ARM Cortex-M0+ core, 48 MHz, 128k flash.

IrDA transmitter and receiver

SLEEP

■■

■■
CAN communication
■■
MMA8451Q accelerometer, thermistor
INITIALISATION MEASUREMENT ■■
Motor control function for simple BLDC
motor control on APMOTOR56F8000E
ULPBENCH RUNS WITH A DUTY CYCLE WHERE THE DEVICE IS EXPECTED TO WAKE UP ONCE PER
SECOND. THE ENERGYMONITOR CAPTURES ENERGY CONSUMPTION DATA FOR 10 SECONDS.
■■
Form factor compatible with Arduino R3 pin layout

WINTER 2014 ELEMENT14 13

THE ENERGYMONITOR GUI CONTROLS AND RECORDS THE ENERGY MEASUREMENT FOR ULPBENCH.

Achieving accurate
energy measurement
Most microcontroller vendors in the ultra-low power domain
have integrated proprietary tools for measuring energy and/or
power into their evaluation/development boards. There are
also a variety of energy measuring tools, the most obvious is
the traditional oscilloscope, but these are relatively pricey.

To support the ULPBench energy Fundamentally, the Energy-Monitor

measurement and reach a sub-$100
price point, EEMBC determined
that it was necessary to develop
a new measuring tool. Hence, the
organisation has produced the
EnergyMonitor, a USB-powered
voltage/current supply for the tar-
get device under test. It connects
to a target device through a 100mil,
2-pin header, and can measure the
energy consumption of literally
anything running on 3V and up
to 28mA. So, although the prima-
ry purpose of the EnergyMonitor
is to measure microcontroller
energy, it could also be used to
measure things such as sensors
for Internet of Things applica-
tions. However, most importantly,
it integrates directly with EEMBC
ULPBench and provides a stand-
ardised method for measuring
energy consumption.
is not an energy measuring tool, it’s
actually a specialised power supply.
Using patent-pending technology,
it ‘counts’ the number of times its
on-board capacitor is discharged,
which in turn sources current to
the device under test (this is an
over-simplified explanation).
In the ultra-low power domain
of IoT devices, battery life is one of
the key attributes. The energy con-
sumption of microcontrollers varies
tremendously, so choose carefully.
Claims of ‘ultra-low power’ must be
backed by concrete data, which
can only be obtained if measure-
ments are equitably made. The
difference of a few micro Joules/
second can mean the difference
between a device that needs a bat-
tery replaced every year versus one
that needs a battery replaced only
after 10 years. n

14 ELEMENT14 WINTER 2014

 ELEMENT14.COM/INTERNETOFEVERYTHING

The Internet of Things is here

and more engineers than ever
are optimising their designs
for an increasingly connected
world. Here we present
the webinars that have taken
place on element14 over
the last few months, where
industry experts explain
the intricacies of the IoT and
discuss the opportunities
that a connected world
can bring for EE design.

8 Useful webinars to help you

build the Internet of Things

Designing for the Home automation Opportunities and Architecture and

Internet of Things
This webinar addresses
the three main problems
with designing an IoT
solution. It divides the
focus into three areas:
cost, security, and fea-
tures. The aim of the
webinar is to help attend-
ees strike a balance.
at your fingertips
Learn how you can build
your very own open
home automation solution
using Eclipse SmartHome
and the wide range of
bindings, from EnOcean
to Philips Hue, available
from the open-source
openHAB project.
challenges with IoT
This webinar reviews the
opportunities and chal-
lenges ahead for the IoT
from a market and techni-
cal point of view includ-
ing the various connectiv-
ity technologies, micro-
controllers and processors
available with a strong
focus on applications.
protocols session
This session provides a
short overview of the IoT
architecture and its proto-
cols. This comprehensive
architecture addresses
these new concerns and
presents a meaningful
path forward for organi-
sations developing and
consuming IoT solutions.

Connecting your IoTise your The Internet of Enable your

project to the cloud
This webinar explains
how to quickly and easily
connect, control and
monitor all your projects
with the cloud via Wi-Fi,
completely free!
embedded projects!
Learn about the Eclipse
IoT initiative, and how you
can effectively use pro-
jects like Paho, Californium,
SmartHome or Kura to
build IoT applications.
Things: in production
Dr. Verena Majuntke
explains what impact the
Internet of Things will have
on manufacturing and how
it might change business
for manufacturers.
smart home
A practical introduction
on how to enable your
smart home with the
use of Raspberry Pi and
EnOcean sensors.

WINTER 2014 ELEMENT14 15

 Of the countless
design challenges
attendant with wireless
connectivity, perhaps
the most immediate is
the vast landscape of
available technologies.

I-FI, BLUETOOTH,
W Bluetooth low ener-
gy, ANT, ZigBee,
RF4CE and many others have
come to comprise a design
menu that can be rather hard
to choose from. It’s a nice

WIRELESS
problem to have – an embar-
rassment of options – but as
the number of potential solu-

PROTOCOLS
tions increases, the task of
compar ing them g rows
increasingly difficult.

EXPLAINED
Choices, choices
In selecting the appropri-
ate wireless technology, the
designer must consider, for
example, whether a propri-
By DAVID FINCH / NEWARK ELEMENT14 and DANA MYERS / TEXAS INSTRUMENTS etary solution is required or if

16 ELEMENT14 WINTER 2014


●●POINT-TO-POINT
●●MESH NETWORK
an industry standard should
be adopted. The frequency
band must be determined,
as well as the standard wire-
less benchmarks of transmis-
sion range, power consump-
tion, and data throughput.
Within these benchmarks, of
course, there are trade-offs,
especially when it comes to
power. Data throughput, for
example, impacts the over-
all power consumption; a
54Mbps Wi-Fi-based network
might last a day or two on
a single Lithium Ion source,
while an IEEE 802.15.4-based
sensor network shuttling
data at 250kbps might sur-
vive a few years on a small
●●STAR NETWORK
●●TREE NETWORK
battery. Other factors, such
as systems compatibility
and network topology, will
also influence most wireless
connectivity designs.
Operating band
The operating band is one
of the design points that
many developers struggle
with. If the various stand-
ards-based technologies
comprise a menu of design
options, then selecting an
operating band is not unlike
choosing the style of food
you want: Italian, Thai,
American (whatever that is).
Whatever band you select
helps to narrow down which
NETWORK TOPOLOGY
Network topology is another
key design factor. Starting
with a simple point-to-point
(P2P) connection, topologies
can be made more and more
complex by the addition,
arrangement, and function of
interoperable nodes.
For example, if you were
to attach additional nodes –
often referred to as endpoints
or clients – to one of the
endpoints (the hub) in a P2P
connection, the result would
be a star network topology,
or LAN. In this configuration,
the hub is responsible for
the exchange of information
between all the endpoints.
Attaching multiple star
networks forms a slightly
more complex topology
known as a tree. A WAN
(wide area network) is an
example of a tree network;
the exchange of information
is managed within each
local star cluster by their
respective hubs, and data
passes from cluster to
cluster along a bus. If an
endpoint in any cluster goes
offline – a workstation in
hardware and software is
available to you.
Each of the three most
popular bands – sub-1GHz,
2.4GHz, and 5GHz – has
cer tain advantages and
disadvantages for any given
application; it all depends
upon the benchmark you’re
designing for. For example,
range is generally reduced
by about half as frequen-
cy doubles, so long-range
applications may require
sub-1GHz operation. One
of the trade-offs with lower-
frequency operation is data
rate, which diminishes with
frequency. Designers may
run into the issue of duty
an office, for example – the
network remains alive, but
communication with that
endpoint ceases. All the
other workstations remain
functional within the
network. However, should the
central connection between
clusters fail, due for instance
to loss of wireless connection,
then the entire tree may fail.
Enter the mesh network.
Mesh networks empower
each node with transmission
and node-to-node
communication capability.
Data is transferred from one
point to another by hopping
wirelessly from node to node,
making a mesh network
not only autonomous but
inherently self-healing. When
one node in the network fails,
the other nodes communicate
with each other to determine
the best alternate route for
the data to get from Point A
to Point B. So, the more nodes
that exist in a mesh network,
the more paths there are for
data to get where it needs to
be and the more robust the
transmission system.
cycle restrictions in the
sub-1GHz band as well. By
comparison, the 5GHz band
affords designers the high-
est available data rate of
the three, at the expense of
transmission range.
Sub-1Ghz and 5GHz bands
both currently have a conges-
tion advantage over 2.4Ghz,
which has been embraced
by consumer technology and
as such has grown rather
crowded. For this reason,
some consumer applica-
tions are beginning to target
5GHz operation. But for all
its crowding issues, 2.4GHz
remains a popular choice
for wireless connectivity. E
WINTER 2014 ELEMENT14 17
 AT A GLANCE...

the STANDARD TECHNOLOGIES

●●Wi-Fi IEEE 802.11 a/b/g/n
Wi-Fi continues to dominate the
world of wireless connectivity.
Current growth projections for
the IoT alone indicate 50 bil-
lion connected devices by 2020,
and it’s no wonder. Wi-Fi offers
secure, high-bandwidth con-
nectivity and Internet access for
medical, consumer, and indus-
trial applications, while boast-
n wi-fi.org
ing data rates up to 54Mbps.
Transmission range can reach
100 meters at 2.4GHz, better
than Bluetooth and other low-
energy options, but less than
what you might get from a
Sub-1GHz radio frequency. But
this performance makes Wi-Fi
relatively power-hungry with
respect to other 2.4GHz wire-
less options: think Lithium Ion
batteries as opposed to AAAs
or coin cells for portable appli-
cations. However, some new
embedded Wi-Fi solutions, like
TI’s new SimpleLink Wi-Fi plat-
form, have lower performance
but can run on 2 AA batteries
for over a year, changing the
game where Wi-Fi can play.

●●Bluetooth
With an installed base of more
than 3 billion units, Bluetooth
is designed for lower-power
wireless connectivity between
devices. Bluetooth devices can
power down during periods of
SMART inactivity, which helps to mini-
n bluetooth.com
mise power consumption; AAA and unlicensed in most coun-
batteries are sufficient sources tries. Data throughput is less
of power for Bluetooth prod- than that offered by Wi-Fi but
ucts. Bluetooth operates in the still in the 2 Mbits/second range
industrial, scientific, and medi- to support higher data-rate
cal (ISM) band at 2.4 GHz to applications, including multime-
2.485 GHz, which is available dia products.

●●Bluetooth low energy n bluetooth.com/le

Taking a good situation and
making it better, Bluetooth low
energy was designed to suit the
needs of ultra-low-power wire-
less applications, including the
ever-growing wearable sports
and wellness markets. Compared
to classic Bluetooth capabilities,
Bluetooth low energy technol-
ogy is a connectionless protocol,
which significantly reduces the
amount of time the radio must
be on. Requiring only a fraction
of the power consumption of tra-
ditional Bluetooth technology,
Bluetooth low energy can enable
target applications to operate on
a coin cell for more than a year.

●●ZigBee IEEE 802.15.4 and ZigBee RF4CE n zigbee.org

The ZigBee standard was cre-
ated to address the need for a
cost-effective, standards-based
wireless networking solution
that supports low data rates, low
power consumption, security, and
reliability. ZigBee supports self-
healing mesh networking. A tech-
nology specifically targeted for
wireless sensor networks, ZigBee
802.15.4 can be used in any
monitoring and control applica-
tion that requires a wireless link.
ZigBee RF4CE, on the other hand,
is something completely different.
RF4CE was developed as an RF
alternative to traditional line-of-
sight remote control technology,
such as IR. In addition to non-line-
of-sight control of remote devices,
RF4CE offers a longer range, long-
er battery life, and richer commu-
nication to support enhanced fea-
tures and functionality.

●●ANT
ANT provides a simple, low-cost
and ultra-low power solution
for short-range wireless com-
munication in point-to-point and
more complex network topolo-
gies. Suitable for a wide range
of applications, ANT is today
a proven and established tech-
nology for collection, automatic
transfer and tracking of sensor
data within sports, wellness
management and home health
monitoring applications. In con-
n dynastream.com
trast to BLE, ANT has a relative-
ly simple protocol and doesn’t
require a lot of overhead. It
specifies 64-bit security versus
128-bit utilised by BLE. Both
offer data rates of about 1Mbps,
accommodate a range of a few
tens of meters, and boast years
of life on a coin cell battery. But
perhaps the greatest difference
for system-minded designers
lies in the network topologies
supported by each; BLE is suit-
able for P2P or star topologies,
while ANT supports the full
range, including tree and mesh.
On a side-by-side, technological
comparison, ANT is arguably
the more appealing option for
wearable/IoT applications, if
for no other reason than mesh
support. But it’s worth noting
once more the numbers associ-
ated with Bluetooth: it’s hard
to ignore an installed base of
3B units.

18 ELEMENT14 WINTER 2014

wi-fi for All
®

connect more. anywhere. anything. anyone.

Connect more with TI’s SimpleLinkTM Wi-Fi®

CC3100 and CC3200 solutions for the
Internet of Things www.ti.com/simplelinkwifi

The platform bar is a trademark of Texas Instruments. © 2014 Texas Instruments Incorporated
E It offers a reasonably high
data rate, decent range, no
duty cycle restrictions, and,
notably, 2.4GHz band works
worldwide, making it a slam-
dunk for designers.
Proprietary vs standard
The decision to develop a
proprietary solution versus
leveraging a standard tech-
nology such as Bluetooth
often boils down to the
basic functionality required.
Highly-specialised functions
may necessitate a proprie-
tary implementation, which
immediately limits the num-
ber of solutions available
from OEM suppliers. The
upside is that a proprietary
wireless platform may be
So, which technology is right for me?
In an ideal world, technology alone would be the ultimate factor in any design decision.
But, here in the real world, technology selection is driven by a number of other factors…
Ease of design is one
such critical factor.
Especially if you’re a
smaller firm or contract
designer, it is important
to find the solution that
can get your product to
market quickly. TI’s
customised specifically for
the application, decreas-
STANDARDS ing the software footprint,
among other things. And the
BASED dual-edged nature of any pro-
SOLUTIONS prietary design applies to
wireless connectivity: propri-
AFFORD THE etary devices may be simpler
DESIGNER A to design because they only
have to operate within one
WEALTH OF defined network, but those
CONSUMER same devices won’t be able
to communicate with stand-
FRIENDLY ard consumer devices.
CONNECTIVITY Standards-based solutions,
on the other hand, afford
OPTIONS. the designer a wealth of
consumer-fr iendly con-
nectivity options. You just
have to cough up the code
space and possibly sacri-
fice some power versus the
SimpleLink portfolio, for wireless development and
example, is a platform certification by minimising
of solutions that simpli- the amount of RF exper-
fies wireless connectivity tise you need to wirelessly
design. The SimpleLink enable your application.
family provides self- So, whether you
contained solutions that choose to develop
were designed to simplify your own proprietary
proprietary option. But with
these sacrifices come the
real advantages of OEM-
independent interoperable
nodes – namely, a freer user
experience for consumers
and greater opportunities
for network expansion.
The more nodes, the big-
ger the network.
Honed on the range
Among the most important
wireless connectivity design
parameters is transmission
range, which consists of two
key factors: transmission
power and receiver sensitiv-
ity. These are critical specs
to consider when exploring
IC options from semiconduc-
tor OEMs.
?
installation or leverage
the vast landscape of
standard solutions,
when it comes to
wireless design,
you’ve got options.
And that’s a decidedly
good thing. n
WINTER 2014 ELEMENT14 21

SHINING A LIGHT
THROUGH THE
IOT CLOUD
By VANDANA LOKESHWAR / SENIOR TECHNOLOGY DEVELOPMENT MANAGER, NEWARK ELEMENT14
things
Cloud computing enables ubiquitous
network access to connected hard-
ware and software in remote locations
over the internet. The model allows
for applications, software, process
management and data to be stored
on virtualised servers. Users can then
access the data and connected devices
through mobile apps, web browsers
or lightweight desktop environments.
The shared resource structure for stor-
age, software, networks and services
allows for a scalable and lower cost
22 ELEMENT14 WINTER 2014
business operation. Cloud based ser-
vices are largely subscription based
or pay-per-use (metered). The metered
services allow for customers to use
only what they need, thereby keeping
expenses low.
The deployment models are primarily
based on the resource management,
accessibility, elasticity and most impor-
tantly the security of the cloud: pri-
vate, public or hybrid clouds. Public
clouds service providers such as AWS,
Microsoft and Google have their own
ECHNOLOGY AND SOLU-
T tions for intelligent end nodes,
wearable devices, communi-
cation gateways and data aggregators
is continuously evolving with respect
to optimising performance, integration,
energy efficiency and security. With
the help of these solutions, a huge
amount of data is being harvested from
various sources e.g. Sensor data relat-
ed to air quality, health vital signs,
diagnostics and preventive mainte-
nance information from industrial floors
etc. All of this data has to be run
through, or stored on the cloud.
THE INTERNET OF THINGS
(OR EVERYTHING), A WORLD
WITH SMART DEVICES
(THINGS) AND SYSTEMS,
NETWORKED TOGETHER
AND RUNNING ON
CLOUD BASED SERVICES
AND APPLICATIONS.
IDC PREDICTS THAT THERE
WILL BE OVER 50 BILLION
CONNECTED ‘THINGS’
BY THE YEAR 2020.
infrastructure setup within the data
centres. Private clouds are typically
within organisations, behind a firewall
and quite capital intensive. Hybrid
clouds are a cluster or combination of
private and /or public clouds, to sup-
port temporary growth requirements.
Clearly the IoT paradigm will eventu-
ally work for all of us, but not without
first addressing the challenges related
to security, data control, non-standard
infrastructure and the handling of vol-
umes of heterogeneous data. n
THERE ARE THREE KEY SERVICE MODELS FOR CLOUD COMPUTING
●●SOFTWARE AS A SERVICE
End Users have access to the database
and ‘on-demand’ application software
(front office and back office) with
a thin client via a web browser. All
the infrastructure for hardware/
software is provided by the cloud
VISUALISATION
APPLICATIONS
NETWORKING
MIDDLEWARE
RUNTIME
STORAGE
SERVERS
●●PLATFORM AS A SERVICE
DATA
O/S
Application developers are able to
design, develop, test, deploy and
MANAGED BY VENDOR host applications on virtualised
servers, with access to programming
languages, platforms and tools. It
● SOFTWARE AS A SERVICE naturally extends to running existing
applications and services including
versioning, testing and build tools.
Solutions for PaaS designed for
Internet of Things (IoT) and Machine
to Machine (M2M) applications are
being offered by various service
providers today. In most cases the
platforms provide end-end solutions
VISUALISATION
APPLICATIONS
NETWORKING
MIDDLEWARE
for support security/authentication,
networking, data communication,
RUNTIME
STORAGE
SERVERS
interoperability and device
management. XML or JSON data
DATA
O/S
from the IoT devices is acquired and
collected through RESTful APIs to the
cloud. The information thus obtained
YOU MANAGE MANAGED BY VENDOR
is then computed and applications
developed to communicate/control
● PLATFORM AS A SERVICE the devices. Real-time messaging
is supported with MQTT (Message
Queued Telemetry Transport) and
similar connectivity protocols. Data
analytics, querying and reporting
capabilities are well supported
with data visualisation tools as
well. Higher end complex machine
learning algorithms are also being
VISUALISATION
APPLICATIONS
NETWORKING
MIDDLEWARE
implemented in such models.
There is a need for scalable systems
RUNTIME
STORAGE
SERVERS
that can support the storage and
processing of these large amounts of
DATA
data. At an enterprise level, Apache
O/S
YOU MANAGE MANAGED BY VENDOR
●●INFRASTRUCTURE AS A SERVICE
● INFRASTRUCTURE AS A SERVICE
Infrastructure as a service is a
provision model for hardware
components such as storage,
servers and networking. System
administrators install operating-
system images and their application
software on physical or virtual
machines on the cloud. The cloud
VISUALISATION
user patches and maintains the
APPLICATIONS
NETWORKING
MIDDLEWARE
operating systems and the application
software. The client typically pays on
RUNTIME
STORAGE
SERVERS
a per-use basis.
DATA
The true success of the IoT revolution
O/S
is going to be dependent on the
storage, processing, interpretation and
YOU MANAGE analytics of the data acquired, such
that it can be applied effectively to
improve the environment and society
● PACKAGED SOFTWARE
SaaS
service providers, which would also
imply all maintenance and support
requirements to be handled by them.
This is primarily a metered service.
Typical examples would include
email, CRM software etc.
PaaS
Hadoop, an open source software
framework allows distributed
computing and processing of large
amounts of data; a file system that
has high throughput access to
application data. It uses a simple
programming model that allows the
use of standard servers.
Access to the embedded developers
community (including DIYers) for
such services has been facilitated by
several providers.
Xively’s public cloud and web-based
tools, help design, build and debug
connected products. It includes
enablement libraries to interoperate
with many languages, platforms,
gateways and smart devices. These
open source libraries allow for easy
integration of the Xively APIs onto
embedded products. For example, the
Xively ARM mbed library allows you
to develop on industry-standard ARM
architectures to make your code easily
portable to production-grade chips.
Similarly Plot.ly also allows sensor
data to be streamed directly from
hardware platforms like Arduino,
Raspberry Pi and also from MATLAB
data acquisition systems. The
data received can be plotted into
interactive charts using Python, R,
Excel and can be shared and edited
in a web browser. Plotly charts can be
exported to any image format through
their Workspace or APIs, including
pdf, png, eps, and jpg.
IaaS
in general. Thinking about moving?
Check the air quality and alleviate any
asthma concerns. Planning a trip to
the Caribbean? Get notifications on
any shark movements…. Thingful a
powerful search engine provides
a geographical index for all the IP
connected devices on the IoT network.
It enables people find devices (eg,
other Raspberry Pi users in the
geographic proximity), datasets and
real-time data sources by geo-location
(like air quality, seismic activity,
animals etc), and presents it using
a proprietary ranking methodology.
Users can keep track and monitor the
real-time status of these devices and
be alerted of any changes.
WINTER 2014 ELEMENT14 23
 ● ● ● ●

L A M P
LINUX APACHE MYSQL PHP

By AL FECZKO / VP OF SALES AND FIELD ENGINEERING, TIMESYS

Deploying a
purpose-built I
N THIS ARTICLE, WE’LL
ex p l o re w hy a b l e n d e d

LAMP Server
approach increasingly makes
sense and also show how it’s become
easier to implement and deploy a pur-
pose-built embedded LAMP (Linux,

into the IoT

As reported in a recent WIRED magazine article by
Cameron Graham, there are two major entry points into
IoT across industries: servers and embedded processors.
But with the increasing capabilities of modern SOCs, the
line between these two entry points continues to blur.
Apache, MySQL, and PHP) server into
the Internet of Your Things.
Most of the attention in the IoT space
these days seems to be around sensors
and edge devices, the endpoints where
‘big data’ is being generated, gathered
and then passed up to more intelligent
parts of the system. And in some ways
that is understandable, because that's
where the really big numbers are: as
many as 25-50 billion units or more
over the next few years, if you believe
what some industry analysts and

24 ELEMENT14 WINTER 2014

semiconductor company executives
are saying (or, perhaps hoping for?).
But in reality, for the IoT to truly
deliver on its promise, it needs to be
thought of from a system perspec-
tive, not simply as a loose collection
of autonomous, distributed devices
that are smart and connected directly
to the open Internet. In reality, most
companies building embedded devices
are also designing and managing a sub-
network to monitor and manage their
devices. While it’s fine for cell phones
and tablets to connect directly to the
Apple App Store or Google Play store,
most device manufacturers prefer to
insert their own purpose-built servers
into the device ecosystem. As such,
powerful yet efficient low-cost LAMP
servers can play this role, by collecting,
storing, processing, and analysing the
distributed device data, and also by
controlling, managing, and maintain-
ing the devices themselves.
Why deploy a purpose-built
LAMP server into IoT?
Whenever people talk about the chal-
lenges of the IoT, some common points
that seemingly come up on a regu-
lar basis are connectivity, scalability,
power management, and security/
software updates. Introducing a LAMP
Server into the IoT architecture can play
a key role in three of these four areas.
LAMP servers are widely deployed
and understood in IT circles. LAMP
servers provide many benefits, includ-
ing full control and configurability,
straightforward upgrades for new fea-
tures, modules or security fixes, high
performance and scalability, and more.
Security and software updates
Designing your devices to be secure
is important, but as evidenced by the
frequency of updates on your PC and
mobile phones, it's equally or even
more important to be able to update
the devices when new threats emerge
or existing ones evolve.
This means that an important capa-
bility for your architecture is to have
an ability to push security and soft-
ware updates out to the edge devices.
Since many embedded devices being
built and deployed today are not run-
ning iOS or Android, the App Store
and Google Play aren’t relevant. With
many low-cost edge devices being
deployed with microcon- good start towards a
trollers (MCUs) and running full LAMP server. The
either bare metal code or at
most an RTOS with a TCP/
WHILE THESE microSD card also
comes with a rich
IP stack, a custom software DAYS MOST OF assortment of other
update server is required.
That’s where the purpose-
THE ATTENTION useful software pack-
ages including Qt5
built LAMP server comes in. IN THE IOT graphical toolkit, Xfce
You benefit by having a small
but powerful connected serv-
SPACE SEEMS lightweight desktop,
dropbear SSH server,
er that can query the edge TO BE AROUND OpenSSL, GStreamer
devices to determine their
current software level and
SENSORS AND multimedia framework
and Firefox browser.
then push security fixes and/ EDGE DEVICES, While it provides a
or code updates out to the
devices when appropriate.
IT’S IMPORTANT nice out-of-the-box
demonstration and
Another benefit of building TO CONSIDER development environ-
a custom purpose-built serv-
er is that LAMP really doesn’t
THE ENTIRE ment, it’s not likely to
include the right set
have to be LAMP. Worried ARCHITECTURE. of software for your
about security issues or com- par ticular applica-
plexity in Apache or MySQL? tion req uirements.
Then substitute lighttpd and However, you can
SQLite in their place. Do your effectively create your
developers prefer Python over PHP? own purpose-built LAMP server and
Not a problem, simply build and deploy add in Apache or some other light-
Python instead. weight embedded web server such
One of the lesser known benefits as lighttpd. Using the free Timesys
recently experienced in the world of LinuxLink Cloud Factory, it’s easy to
embedded Linux is that the Shellshock reconfigure and rebuild the Gizmo 2
vulnerabilities only applied to Bash — platform software to use a different
whereas many embedded Linux devic- package set.
es are deployed with BusyBox, whose
built-in shell was not susceptible to the Leveraging a GPU on a
Shellshock vulnerability. purpose-built LAMP server
Increasingly, GPUs are being used to
Scalability and intelligence accelerate applications that have noth-
Adding processing power into the IoT ing to do with graphics. For example,
is also important, because edge devices SQL database operations and perfor-
are typically 8-bit MCUs, or increas- mance can be dramatically improved
ingly lower end 32-bit MCUs with only by implementing some parts of the
KBs of memory. A purpose-built LAMP SQL virtual machine directly on the
server — built for example with an AMD GPU. Alternatively, the increasingly
G-Series SOC containing quad or dual popular Open Computing Language
64-bit x86 cores and gigabytes of DDR3 (OpenCL) framework allows for any
RAM — can really turbocharge the intel- number of applications to programmati-
ligence of the IoT by adding much more cally access the GPU for non-graphical
processing power per watt into the parallel computing.
IoT architecture and enabling more
advanced data fusion and processing, Summary
number crunching and analytics. While these days most of the attention
in the IoT space seems to be around
Example of a purpose- sensors and edge devices, it’s impor-
built LAMP server tant to consider the entire architecture.
For an inexpensive and compact, yet A purpose-built LAMP Server can play
extremely powerful 64-bit x86 plat- an important role in the IoT, by collect-
form, consider the AMD G-Series SOC ing, storing, processing and analysing
powered Gizmo 2 platform. It ships the distributed device data as well as
with a bootable microSD card that by controlling, managing and maintain-
contains Linux, SQLite, and Python — a ing the devices themselves. n
WINTER 2014 ELEMENT14 25
 A HARDWARE OVERVIEW
INTRODUCTION TO
FPGAs
HAVE YOU WANTED TO START DESIGNING
WITH FIELD PROGRAMMABLE GATE ARRAY
BUT FOUND THE WHOLE TOPIC VERY
DAUNTING? WELL YOU ARE NOT ALONE…
By SIMON HOLT / STRATEGIC ALLIANCE MARKETING MANAGER, PREMIER FARNELL
HIS ARTICLE AIMS TO GIVE
T you a great starting point to
enter into the world of pro-
grammable logic. Once you have start-
ed to understand this technology, it is
surprisingly easy to use and will offer
you an electronics design option differ-
ent to anything else available on the
market today.
We will start by covering the basics
of FPGAs, explaining the difference
between the three most popular pro-
gramming technologies currently avail-
able. This will be followed by an over-
view of the different building blocks
that can be found within an FPGA.
An FPGA is a programmable device
that allows you to create your own
digital chip. The basic building block
of an FPGA is the humble transistor.
Depending on the density of the FPGA
device you decide to use, it can consist
26 ELEMENT14 WINTER 2014
of many thousands, even millions
of transistors. These Transistors are
grouped together to perform an array
of functions, from a simple logic gate
to a complex Clock circuitry block. The
majority of these blocks are available
to you to configure using the FPGA
software tools.
Advantages of FPGA
FPGAs are completely customisable
and as long as there are enough build-
ing blocks in the FPGA you are using,
they can be configured to perform any
digital design. Due to the layout of an
FPGA they lend themselves nicely to
parallel processing designs, aided by
the high I/O count normally found
on them. Depending on the type of
FPGA you use, they have the ability
to be re-programmed. These devices
are completely blank on power up
LARK BOARD
FPGA with Dual Core ARM
Cortex-A9 @ 800MHz
A fully featured Cyclone V SoC
evaluation board integrated
with a FPGA and ARM-based
dual‑core Hard Processor System
(HPS) in a single package and
operates at speeds up to 800
MHz. The combined technologies
help to reduce the overall system
cost and power.
FPGAS
ARE VERY
VERSATILE
DEVICES AND
THAT THE ONLY
LIMITATION IS
THE CAPACITY
OF THE DEVICE
AND YOUR
IMAGINATION.
and require the user program their
design onto it.
Disadvantages
Probably one of the biggest reasons
why people don’t want to use FPGAs
is the steep learning curve in designing
with them. It can take many hours to
learn the design tools, coding language
and debug software in order to get your
design into the device and working as
you want.
In the past the price of FPGAs has
often put designers off using them, but
in recent years the prices have dropped
to a level which is more attractive bear-
ing in mind the flexibility they offer.
The architecture of an FPGA means
that it generally consumes quite a bit
of power, although there are some
design techniques available to com-
bat this issue. E
TYPES OF PROGRAMMABLE LOGIC DEVICES
There are three main types of FPGA technology, which are SRAM, Antifuse and Flash‑based
devices. These differ by the way in which their internal routing is programmed.

SRAM technology Antifuse technology Flash technology

Most FPGAs that are used for design today are based
on SRAM configuration cells. The big advantage to
using these devices is that they can be programmed
and re-programmed multiple times. The two main
manufacturers of FPGA devices, Altera and Xilinx,
both produce SRAM based products and it is these
devices that lead the industry in terms of product
development and adopting new approaches to FPGA
design. A downside to using SRAM based devices is
that they require programming every time they are
powered-up. This operation can be performed by the
use of an external memory device or a microproces-
sor that provides the bitstream (configuration) code.
In contrast to SRAM based devices, Antifuse
FPGAs are instant-on and keep their design
code when switched off. The devices are con-
figured by blowing fuses to create the inter-
connect circuitry. This is a one-time operation
that is carried out before soldering them to
your printed circuit board in a special pro-
grammer and cannot be reversed. The advan-
tage of Antifuse devices is that they are very
secure; it is nearly possible to clone an Antifuse
FPGA. They are also ideal for harsh environ-
ments, particularly military, aerospace and
space applications.
Flash based devices are basically a
cross between the benefits of SRAM and
Antifuse technologies. They store their
configuration code in the device simi-
lar to an Antifuse device, but this can be
changed like an SRAM based device. From
a security point of view these devices
come with a user-defined key and only the
person with access to this key can read/
write data to and from the device. These
devices tend to lag behind SRAM based
devices in process node due to longer
development times.

WINTER 2014 ELEMENT14 27

 IN SIMPLE TERMS AN FPGA IS
A COLLECTION OF ELECTRONICS LOGIC BLOCK

DESIGN BLOCKS CONNECTED

TOGETHER BY MANY, MANY ROWS INTERCONNECT
AND COLUMNS OF WIRES.
Over the past few decades they have
developed from devices which contain
just a few logic cells, to ones which now
have dual ARM A9 processors, high speed PROGRAMMABLE I/O
interfaces and hundreds of thousands of
logic cells like Altera’s Cyclone V SoC devices.

The make up of an FPGA

Logic blocks
Logic blocks (these have
various names) differ slightly
from one manufacturer to
another, but generally they
contain a number of combi-
natorial logic elements and
one or more flip flops. Logic
blocks make up the largest
portion of an FPGA and often
have other design blocks
mixed in-between them to
allow for a faster, better utili-
sation of the device. The dia-
gram above shows a typical
architecture of an FPGA.
Hard Intellectual
Property (IP) Cores
In recent years there has
been an increase in the num-
ber of custom blocks being
included in FPGAs. These are
normally referred to as hard
IP cores/blocks.
These custom blocks
allow for more efficient use
of the FPGA. A very popu-
lar hard IP core being rolled
out by FPGA manufacturers
is hard processor systems.
Both Altera and Xilinx have
brought out devices with
Dual Cortex A9 processor
blocks integrated in them.
Input/Output (I/O) Blocks
The I/O Cells in an FPGA can
be found around the outer
edge of the device. They are
used to connect the internal
blocks to the physical pins on
the outside of the chip. These
blocks can be programmed
to perform a huge number of
different types of I/O stand-
ards. Some more complex
devices have I/O cells that
can process high-speed sig-
nals; these are referred to as
Transceivers blocks.
RAM Blocks
The Majority of FPGAs avail-
able today have dedicated
RAM blocks distributed
alongside the logic blocks.
These blocks are really use-
ful if your design uses RAM
as implementing these RAM
blocks in standard logic
blocks is very inefficient and
heavily reduces the number
of logic blocks left for non-
RAM functions. Knowing
the size of RAM blocks avail-
able on your chosen FPGA
is very important as you can
design with them in mind,
making the most of the
blocks available.
Interconnect/routing
Each of the logic blocks
on their own is not much
good, as they need to be
linked together to make
larger design blocks. This is
achieved by interconnecting
routing between the blocks.
There are many different
types of routing within an
FPGA. Some of the routing
is short and goes from one
block to its nearest neigh-
bour whilst other intercon-
nect lines go from the top
to the bottom of the device.
Special routing can also be
found over the whole device,
which is known as global
routing. An example of this
would be the clock circuitry.
These routing lines have low
skew rates and go to every
internal clock node on the
device, enabling synchro-
nised designs.
Programming
A Configuration block is used
to program all the different
blank blocks inside a SRAM
FPGA on power up. Boot up
time needs to be considered
and they are not instant on.
SRAM have these configura-
tion blocks whilst flash and
Antifuse do not need this
configuration block due to
their design. n
READ MORE
If reading this short introduction
to FPGAs has made you keen
to learn more, then head on
over to element14’s Community
pages. We have a dedicated
group set up to just talk
about FPGAs.
www.element14.com/
community/groups/fpga-group

28 ELEMENT14 WINTER 2014

Exclusively from
 INTELLIGENT
LAMP CONTROL
with Infineon XMC1000
By ATTILA TOMASOVICS / SENIOR ENGINEER, INFINEON

INTELLIGENT LED LIGHTING IN

OFFICES, RETAIL, HOSPITALITY AND
HEALTHCARE, AMONG OTHERS,
IS RAPIDLY GAINING POPULARITY.
AN ESSENTIAL REQUIREMENT
IS HIGH QUALITY COLOUR AND
DIMMING CONTROL. FOR ALL THESE
APPLICATIONS, THE BRIGHTNESS,
COLOUR AND DYNAMIC BEHAVIOUR
OF LEDS NEED TO BE CONTROLLED
AS ACCURATELY, PRECISELY
AND CHEAPLY AS POSSIBLE.

HIS IS ACHIEVED BY

T
modulating the current
of LEDs. An additional,
ver y challenging, re-
quirement is flicker-free
lighting. Modulated LEDs are prone to
flickering but absolutely no flicker is
allowed in these applications. Custom-
ers expect to see the colour they want
to see at the desired dimming level and
want to experience no flicker. Lighting
applications are also becoming more
complex which increases time to mar-
ket. XMC1200 and XMC1300, members
of the high-performance XMC1000 fam-
ily of 32-bit microcontrollers, have a
Brightness and Colour Control Unit
(BCCU) which is a dedicated multi-chan-
nel lighting hardware module that ena-
bles fast development of high-quality
lighting systems.

30 ELEMENT14 WINTER 2014

Modulation dimming INTELLIGENT CONTROL GEAR
To control the brightness of an LED, its
current needs to be controlled. The most Driver stage
with colour and
accepted method to control the perceived dimming control
brightness of LEDs is modulation dimming.
LED driver
The LED current level is fixed (e.g. at 100 mA) and
perceived brightness is controlled by modulating LED driver
this current. Modulation basically means switch- AC in AC-DC
ing the LED on and off very quickly. If the switch-
LED driver
ing rate is high enough, then we see the lamp as a
constant light source due to persistence of vision.
Luminous intensity of the lamp is determined LED driver
by the average time the LED is switched on,
relative to the time it is switched off. e.g. an XMC1000
LED modulated by a square wave spends 50%
of the time on. Its average luminous intensity
will be half as high as when it is always on.
Dali, DMX512 Sensors
The LED current is controlled by a driver. The
desired brightness, represented by a digital value,
●●INTELLIGENT LIGHTING
is converted into an ON-OFF signal by a modu-
lator. This ON-OFF signal, which is represented
by a bit stream, enables/disables the driver.
Many different modulation methods exist, XMC1000
such as Pulse-Density Modulation (PDM), Pulse-
Width Modulation (PWM), Pulse-Amplitude BCCU
Modulation (PAM), etc. The BCCU module in PDM LED Driver
Infineon microcontrollers generates pulse-
density modulated bit streams by randomised
PDM LED Driver
first-order sigma-delta modulators.
The PDM signal contains ON and OFF pulses.
PDM LED Driver
Every pulse has the same fixed, user-config-
urable length called the bit-time. The resolu-
PDM LED Driver
tion is 12 bits so all brightness information is
contained in a frame of 4096 pulses. Due to
PDM
randomisation for better EMI behaviour, the
frames are not identical but contain the same PDM
number of ON and OFF pulses if the bright- PDM
ness value is constant. The ON and OFF bits
PDM
are distributed in the frames as uniformly as
possible to achieve the maximum number PDM
of ON-OFF and OFF-ON transitions and to
minimise the stable time between transitions. E
●●PULSE-DENSITY MODULATED BIT STREAMS●
TO CONTROL LEDS ON UP TO 9 CHANNELS

ON time OFF time

DC
PDM 0 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 1

ON pulses OFF pulses

●●LED MODULATED BY A SQUARE WAVE ●●SMALL SECTION OF A PULSE-DENSITY MODULATED BIT STREAM

BIT STREAM
BRIGHTNESS Modulator LED Driver PERCEIVED
VALUE BRIGHTNESS
N 1

●●LED BRIGHTNESS CONVERTED INTO A MODULATION SIGNAL

WINTER 2014 ELEMENT14 31

 Managing flicker
If the bit time is too long or there are too many
identical pulses next to each other, the eye may
detect the switching and distinguish between
ON and OFF levels instead of seeing the
average value, which is experienced as flicker.

The eye is particularly sensitive to flicker when

it moves. To avoid flicker, the lighting system
has to ensure that time between two transitions
is always below a certain value. This value is
commonly called the flicker fusion threshold.
This threshold is not an exact value and it
depends on the person watching the lamp,
the colour, the environment, etc. To avoid any
kind of perceptible flicker, the time between
transitions has to be kept shorter than 333
µs. Furthermore, taking videos of LED lights
with cameras may also cause flickering or
●●RECORDED LAMP REFLECTION WITH BANDING AND WITHOUT BANDING
banding in the recorded footage. The general
recommendation is that the period should
be maximum 20-100 μ s, depending on the
camera, which is an even more challenging
requirement. However, there is currently no
official standard.
For truly high quality lighting the ON and
OFF times have to be kept really short. The
OFF times are more critical because the eye
is more sensitive to flicker at low illumination
levels. Pulse-density modulation inherently
supports flicker-free designs due to its high
rate of switching, since the ON and OFF bits
are distributed as uniformly as possible. The
modulators in the BCCU module can generate
pulses shorter than 100 ns. The bottleneck is
the LED driver as it needs a minimum stable
ON time to stabilise the LED current. With
most LED driver solutions it is not possible
to have pulses shorter than 1 µ s. This cre-
ates a potential flicker problem at very low
brightness levels as the ON pulses are too few
and far between. If the OFF time is too long
between two ON pulses, the human eye may
experience flicker and camera footage will
look banded, not natural.
The only solution is to disallow these very
low brightness values. Every BCCU channel
contains a flicker watchdog that ensures that
the OFF time is never too long. If the OFF time
reaches the user- definable threshold, an ON
bit is inserted and the sigma-delta modula-
tor is reset.

32 ELEMENT14 WINTER 2014

Dimming and colour change BCCU
For intelligent lighting it is also important
to control the dynamic behaviour of
AMBER
the brightness value over time in a way INTENSITY
that appears natural and is comfortable ● AMBER
PDM LED Driver
to the human eye. BRIGHTNESS

BLUE
When the LEDs are organised into multiple INTENSITY
channels, dynamic change can be divided PDM LED Driver
● BLUE
into two distinct transitions. One is overall BRIGHTNESS

brightness change, also known as dimming.

DIMMING LEVEL
GREEN
The other is colour change, a change in rela- INTENSITY

tive intensities. The internal structure in BCCU PDM LED Driver

● GREEN
BRIGHTNESS
mimics multi-channel lamps with colour and
dimming information clearly separated. BCCU RED
INTENSITY
can automatically change both the dimming
PDM LED Driver
level and the colour of multiple lamps in a ● RED
BRIGHTNESS
natural way.
During dimming, all channels change their
brightness level but their relative intensities
remain the same, to preserve the colour of the
lamp. Due to logarithmic brightness percep-
●●DIMMING LEVEL AND CHANNEL INTENSITIES SEPARATED
tion, the human eye is more sensitive to low
levels of illumination. If the brightness level
changes linearly over time, it look non-linear.
For natural-looking dimming, it needs to be
exponential. The ability to dim along an expo-
N
nential curve is also a requirement by DALI,
one of the most popular lighting communica-
tion standards.
BCCU contains dimming engines, hardware
BRIGHTNESS

modules, which automatically perform expo-

nential dimming without CPU load. The user
only needs to select target brightness levels
and fade rates, and the dimming engines auto-
matically adjust the brightness levels along an
exponential curve. Since the dimming engines T
do not rely on interrupts, they can follow the TIME
curve very closely without additional CPU load.
The human eye is most sensitive to changes at
the lower part of the exponential curve, where ●●EXPONENTIAL DIMMING

the brightness level is very low. At low fade

rates, each individual step in brightness may
be visible. BCCU adds dither to these steps to G R
make the transition look smooth and gradual N
to the human eye.
Multi-colour lamps combine a number of
channels of different types of LEDs, each with
a different dominant wavelength. What colour B G
N/2
the human eye sees depends on the relative
intensities of these channels. The colour can
INTENSITY
CHANNEL

be changed smoothly and gradually by using

the linear walk. The user only needs to set R B
the target colour and the desired fade time.
All affected channels change their intensity Linear Walk TIME
simultaneously until the new colour is reached. E SPRING GREEN ORANGE

●●LINEAR WALK EXAMPLE IN AN RGB SYSTEM

WINTER 2014 ELEMENT14 33

 Software development
With so much hardware support,
hardly any software development is
needed for high quality LED control.

Much of what little software is necessary

can be generated by DAVE, Infineon’s free
IDE which offers an easy-to-use graphical
interface, automatic resource allocation and
code generation to allow fast time to market.

●●LIGHTING PROJECT IN DAVE

Conclusion
Infineon’s XMC1000 Cortex-M0 family is the
ideal choice for intelligent LED lighting appli-
cations. It offers the resources required for a
single microcontroller to automatically control
multiple LED channels with high dimming
and colour quality, communicate over DALI or
DMX512, with more than enough bandwidth
available for the end application. n

www.infineon.com/xmc

34 ELEMENT14 WINTER 2014

 EAGLE Version 7.2

now
available
For more information,
visit www.cadsoftusa.com
EFFICIENT &
INTELLIGENT
LIGHTING
Using Infineon’s RGB LED Lighting ■■ www.infineon.com/arduino
■■ www.infineon.com/xmc
Shield with XMC1202 for Arduino ■■ www.infineon.com/xmc-dev

36 ELEMENT14 WINTER 2014

LED lighting is gaining popularity due
to government legislation and its overall
benefits compared to traditional solutions.
EDs are used for lamps and
L light engines in offices, retail,
hospitality, street lighting, high
bay lighting, healthcare, and more.
LED lighting is efficient, allows maxi-
mum design freedom, and it can be
intelligent as well as easy to use.
Thanks to the latest RGB LED Light-
ing Shield with XMC1202 for Arduino
developers can easily explore, eval-
uate and build their next LED pow-
ered product.
The RGB LED Lighting Shield with
XMC1202 for Arduino is built on the
XMC1200 MCU series which has a
Brightness and Colour Control Unit
(BCCU). BCCU is dedicated for multi-
channel lighting engines by offering
Arduino™ Connector
Isolated DMX512
XMC1202 Control Interface
(optional)
DC Input Voltage
Power-on LED
Arduino™ Connector
RGB LED LIGHTING SHIELD WITH XMC1202 FOR ARDUINO.
full control of brightness, colour and
dynamic behaviour of LEDs as accu-
rately, precisely, efficiently and cheaply
as possible.
BCCU generates pulse-density modu-
lated bit streams with randomisation
for better EMI behaviour, and bit pack-
ing feature to be compatible with sev-
eral LED driver designs.
To offload the CPU automatic dim-
ming is provided by adjusting the
brightness levels along an exponen-
tial curve: the user only needs to se-
lect target brightness levels and fade
rates, and the dimming engines take
care of the rest.
The ability to dim along an expo-
nential curve is also a requirement by
I²C, GPIO SWD
LED String Connector
DMX512 LED Current
Control with PDM
DCDC LED Driver Stage
DALI, one of the most popular lighting
communication standards.
The little software required to con-
trol the BCCU can be created with
DAVE™ IDE. DAVE supports automatic
code generation, provides DAVE Apps
with GUI programming, and exam-
ple projects.
Infineon’s XMC1200 MCU series as
part of the XMC portfolio is the ideal
choice for intelligent LED lighting solu-
tions. XMC1200 microcontrollers pro-
vide all essentials for high quality colour,
such as full dimming control, flicker-free
light at all dimming levels with low
energy consumption and ease-of-use
thanks to the new RGB LED Lighting
Shield with XMC1202 for Arduino. n
Debug Interfaces
to XMC1202
XMC1202
WINTER 2014 ELEMENT14 37
 THE BIRTH OF

ENERGY
EFFICIENT
WHITE LIGHT
WIRE BOND
ANVIL
POST

ANODE

CATHODE

By CABE ATWELL / COMMUNITY WRITER, ELEMENT14

38 ELEMENT14 WINTER 2014


SURE, THEY’RE BLUE, BUT WE CAN’T
HAVE WHITE LIGHT WITHOUT THEM.
Scientists win Nobel Prize in physics with their creation of blue LEDs
ONS AGO (ROUGHLY
E
1907), scientists began
messing with the idea of
LEDs in the form of elec-
troluminescence using a
crystal of silicon carbide and a cat’s-
whisker detector. Apparently, Edison’s
light bulb was no longer cutting the
mustard and people wanted tiny little
lights in their radios, which is probably
why RCA was playing around with
gallium arsenide-based semiconductors
back in 1955.
Regardless, LEDs have come a long way
in the last hundred years and has become
the dominant light source on the planet
overtaking incandescent and fluorescent
lighting. While the LEDs manufactured
today are highly efficient, energy saving
and environmentally friendly, they do
have their drawbacks, in this case the
ability to generate a broad spectrum of
colours. More specifically, white light due
ANODE (p-electrode)
p-GaN
p-AIGaN
Zinc-doped InGaN
n-AIGaN
n-GaN
GaN Buffer Layer
Sapphire Substrate
CATHODE (n-electrode)
HOLE
P-LAYER
ACTIVE
LAYER
N-LAYER
ELECTRON
to the fact scientists couldn’t produce
blue light in LED form.
Confused yet? Okay, let’s break it down
in simple terms- in order to create bright,
efficient white light, green or red LEDs
need to be combined with a blue LED.
They could also be shown through a
phosphor that emits red and green light.
Sure there have been blue LEDs in ex-
istence since the 70s but the material
they are made from made them pretty
much useless in the broad spectrum
(pun intended) of practical applications.
By the late 80s, the people’s demand
for bright, shiny things continued to
grow and there still were not any de-
vices (much less radios) with tiny LEDs
that glowed bright white or blue. No-
ticing those demands, three material
scientists decided the world could wait
no longer for bright, efficient blue LEDs
and began developing their own using
gallium nitride.
IT’S ALL IN THE GaN: GALLIUM NITRIDE REVOLUTIONISED
BLUE LED LIGHTING PAVING THE WAY FOR WHITE.
In the early 90s, Professors Isamu
Akasaki (Nagoya University), Hiroshi
Amano (Nagoya University) and Shuji
Nakamura (University of California)
began working on a way to grow gal-
lium crystals big enough to be used to
emit blue light. They found that they
could do so using a specially designed
scaffold made in part with sapphire.
That’s the key, different materials emit
a different colored light when applied
to LEDs and gallium was the ticket in
emitting blue.
Gallium isn’t without its troubles as
it tends to become poisoned when ex-
posed to hydrogen. The people would
have to wait a few years more before
they would get bright, shiny blue or
white lights. To get around that issue,
the scientists ‘doped’ up the gallium
using aluminium or zinc but ultimately
went with indium, which protects the
gallium from the hydrogen used in
manufacturing the semiconductor the
LED resides on.
The hard work that all three endeav-
oured over the years to get an efficient
blue LED earned the trio a Nobel Prize
this year in physics. Not only does
their work provide bright and shiny
LEDs in our mobile devices but also
reduces the power draw on the grid
when used for lighting our homes. Not
only that, they are also better for the
environment over compact florescent
lighting (CFLs) due to the fact that the
latter is filled with mercury. The trio
received their prizes in December,
where their breakthroughs could be
seen lighting up the homes all over
the city. n
WINTER 2014 ELEMENT14 39
 >

>

i v e rka>ge
Dr t Pac
/ or
BBoSarPd Supp

8 6
X

AR
M
ARM vs x86
A CLASH OF
ARCHITECTURES
DECISIONS MADE DURING UCH OF THE TIME THESE of Things (IoT) in the balance, both

DESIGN PROCESSES SHAPE M choices are not clear cut and

THE POSSIBILITIES OF SUCCESS
AND FAILURE OF A COMPONENT,
SUB‑SYSTEM OR END PRODUCT.
By MURRAY SLOVICK / COMMUNITY WRITER, ELEMENT14
few reside in the proverbial
‘no brainer’ category. For the design
engineer trying to decide whether to
select an ARM or x86 architecture for
his/her embedded project the deci-
sion, as Ringo Starr once put it, “you
know it don’t come easy.” Indeed, with
huge potential markets like the Internet
camps are going for the jugular, each
(and their respective ecosystems ) hop-
ing that their latest development will
help them sell in the embedded mar-
ket the way beer sells in a college town.
In this article we will show how and
why the decision to go with x86 or
with ARM has become more complex
and difficult than ever, as there are

40 ELEMENT14 WINTER 2014

 clear benefits available for each
architecture, depending on the
task and design requirements.
The background
But before we leap ahead of the
story – and to be properly prepared
to evaluate both contestants – a bit
of background is in order. ARM
and Intel (x86) chips have differ-
ent processor architectures and
instruction sets, so an applica-
tion compiled for the Intel archi-
tecture will not run on an ARM
device and similarly you can’t
run code compiled for ARM on
Intel devices. An ARM proces-
sor has a Reduced Instruction
Set Computer (RISC) architecture
while the X86 (for the purposes of
this article x86 will refer to both
32-bit x86 and 64-bit x64, ARM
will refer to all ARM derivatives)
has a Complex Instruction Set
> > Computer (CISC) architecture.
CISC chips have an instruction set
capable of doing complex things
with a single instruction while
BSP a RISC processor has a smaller
Boa / Dr
rd S number of more general purpose
upp
ort
iver instructions. It is important to
Pac > note, however, that the simpler
kag nature of a RISC instruction allows
e
it to be processed in usually just
one clock cycle (and often with
fewer transistors, making the
silicon less expensive and more
power efficient). Generally speak-
ing, a RISC-based ARM processor
also requires more memory than
a CISC-based processor.
Traditionally ARM’s low power
has enabled battery life that
would be better than an x86
equivalent and with its built-in
multimedia decoding capabilities
it also enables 1080p HD video
playback in a small package rela-
tive to the real estate an x86 sys-
tem would require.
KEY DIFFERENCES OF X86 VS. ARM Typically, too, for complex tasks
VA BOARD SUPPORT PACKAGE, AN such as would be needed to pro-
IMPLEMENTATION OF SPECIFIC SUPPORT cess and analyse millions of piec-
CODE – SOFTWARE – FOR A GIVEN BOARD THAT es of data, the raw performance
CONFORMS TO A GIVEN OPERATING SYSTEM. from x86 often outstrips that of an
ARM equivalent. If we want high-
IT IS COMMONLY BUILT WITH A BOOTLOADER. IN end devices to perform telemetry
EMBEDDED SYSTEMS BSP IS SYSTEM AND DEVICE and data transmission over the
DRIVERS FOR ALL THE DEVICES ON THE BOARD. Internet for critical applications,
Source: Embedded Logic for instance, x86 would still likely
be the choice, On the other hand,
if high-end functionality involves
only video and audio capabilities,
ARM would be the nominee.
Other important traditional (and
I have been using the term tradi-
tional deliberately because as we
will shortly see, things are fast
changing on both sides of the
playing field) factors are, for ARM:
■■ No BIOS (ARM uses a
bootloader) and, therefore, no
BIOS-related issues
■■ Power consumption in the
range of about 2 to 20 watt
Similarly, traditional advantages
of x86 processors include:
■■ Well established Linux support
■■ Very high computing
power and very short real-
time response.
As for run-time stability (uptime
without crashes), which can be
a key deciding factor, no clear
cut winner here: Both ARM and
x86 processors are relatively well
suited to be used in industrial
and other embedded systems
that require stable, reliable and
deterministic run-time behaviour.
Now to the changes I alluded to.
ARM and Intel chips are growing
closer to each other and the tra-
ditional lines distinguishing them
are starting to blur. Here are a few
for instances.
Intel is positioning Quark, its
smallest processor, in the IoT
device space supported by a
new VxWorks 7 version of the
widely used real time operating
system (RTOS), developed by the
Intel subsidiary Wind River, As a
counter-measure ARM is launch-
ing a new, and free low power
operating system to manage web-
connected devices that use micro-
controllers based on the compa-
ny’s 32-bit Cortex-M architecture.
ARM’s OS for this is called mbed,
Although technically not a true
RTOS (as a strict definition would
require a predictable response
time to an interrupt), ARM’s mbed
takes what the company calls an
‘event based’ approach; the pro-
cessor wakes up only if events
need to be managed. ARM says
that predictable response is not
always needed and allowing the E
WINTER 2014 ELEMENT14 41
E processor to complete one task
before switching to another allows
power savings over traditional
RTOS requirements.
Intel’s x86 chips have been
improving in power consumption.
Intel started with a more expen-
sive, high performance architec-
ture and has been lowering its
power consumption and making
its lower-end chips more price
competitive. In particular designs
based on processors such as the
Intel Atom permit small, fanless
and relatively efficient power solu-
tions. Atom microprocessors have
been used in a variety of embed-
ded applications ranging from
health care to advanced robotics
BOARD & MODULE FORM FACTORS
Boards and modules incorporating ARM processors typically run
Linux, Windows CE (Compact) or Android. Boards and modules
based on Intel and AMD x86 processors run full Windows, or
the embedded versions such as Windows Embedded Standard.
You can also run Linux on these platforms.
Recent trends in processor technol-
ogy also are affecting the newer
lightweight COM class mod-
ules such as Qseven, and Smart
Mobility ARChitecture (SMARC).
The new Qseven specification
update 2.0 allows using x86 and
ARM processors on Qseven mod-
ules and can accommodate the
new Intel E38xx Atom processors
as well as such ARM processors
as the Freescale ARM i.MX6 (both
supporting a wide variety of I/O
features). The new USB 3.0 and
UART additions are widening the
scope of this module family even
further. The devices’ strong graph-
ics performance can be used to
drive HDMI, DVI, DisplayPort and
LVDS displays. For example, the
Qseven module with Intel Atom
E38xx enables dual independent
display while the Q7-IMX6 even
supports triple independent dis-
play, a feature formerly limited to
high-end computer boards. The
majority of customers are either
42 ELEMENT14 WINTER 2014
and mobile Internet devices. At
the same time upper range ARM
processors such as Cortex A9 have
been introduced with high-speed
interfaces and multi-core tech-
nology to increase performance.
Processors such as Freescale
i.MX 6 Cortex A9 now are com-
peting with x86 platforms such
as Intel’s Bay Trail-based Atom
chips as well as AMD’s Embedded
G-Series. All of these processors
intended for embedded applica-
tions are one-chip platforms with
one, two or four CPU cores. All also
feature clock frequencies around
1GHz, DDR3 memory and high
‘‘QSEVEN
speed interfaces like PCI Express
and HDMI. SUPPORTS
ARM, INTEL AND
AMD PROCESSORS
’’
using Qseven for x86 or Qseven
for ARM exclusively; however,
there some that are moving in
both directions.
With regard to Smart Mobility
A rch i t e c t u re ( S M A RC ) , t h e
SMARC pinout is optimised to
®
support specific I/O functionality
that is more commonly identified
with ARM in mobile and handheld
devices. For example, Advantech’s
ROM-5420 is a SMARC form fac-
tor module only 82mm x 50mm
in size, (with an overall assem-
bly height of under 8mm) based
on the Freescale ARM Cortex-A9
i.MX6 dual core processor. ROM-
5420 supports triple independent
displays and a 1 x MIPI (Mobile
Industry Processor Interface)
camera input. Moreover, ROM-
5420 also supports lithium-ion
batteries that have +3 ~ 5.25V
power input design with suspend
mode under 0.3W, which makes
them an excellent choice for port-
able applications.
But SMARC is not exclusively for ARM processors
and board/module suppliers have announced
x86-based SMARC modules as well. Consider
Kontron’s latest SMARC computer-on-modules
(COMs) featuring Intel’s Atom E3800 series pro-
cessor. According to Kontron, it’s the first x86 SoC
to become available on this low-profile COM form
factor, measuring 82mm by 50 mm. The SMARC
s-XBTi COMs, compliant with the Standard Group
for Embedded Technologies (SGeT) specifica-
tion, incorporate up to 8 Gbytes of RAM and can
accommodate up to 64-Gbytes, on-board solid-
state disk to store OS and application data. The
pin-out includes three UARTs with complete GPS
function and support for the MIPI-compliant serial
camera interface. Intel Gen 7 Graphics are carried
out via HDMI 1.4 and LVDS with up to 2500 by
1600 pixels and 60 Hz to the display. Other inter-
faces include one Gigabit Ethernet LAN via Intel’s
Ethernet Controller I210, one USB 3.0 and two USB
2.0. Extensions can be implemented via two SDIO
and three PCIe x1 lanes with 5 Gtransfers/sec.
Operating temperature ranges from −40 to +85°C,
while power consumption ranges from 5 to 10 W. AMD HAS ANNOUNCED IT WILL
Adding to what is already a recipe for brain
freeze AMD has announced it will develop pro-
DEVELOP PROCESSORS, CODE NAMED
cessors, code named Mullins and Beema, which MULLINS AND BEEMA, WHICH WILL
will consist of AMD’s next generation x86-based
core, with an ARM-based core for handling secu-
CONSIST OF AMD’S NEXT GENERATION
rity functions on the same SoC. X86‑BASED CORE, WITH AN ARM‑BASED
At the same time in parallel with the company’s
next-generation of x86 processors, AMD plans to
CORE FOR HANDLING SECURITY
develop its own 64-bit ARM-based CPUs, code FUNCTIONS ON THE SAME SOC
named ‘Project SkyBridge.’ Targeting embed-
ded applications. SkyBridge’s ARM-based unit
will use low-power 64-bit ARM Cortex-57 cores,
while the x86-based chips will feature AMD’s
next-gen Puma+ central processor cores. The
clever bit is that the chips will be ‘pin-compati-
ble,’ meaning they can both use the same circuit
board, simplifying things (somewhat) for device
makers. SkyBridge chips will be manufactured
on a next-generation 20nm process and ship
sometime in 2015.
AMD’s roadmap reminds one of the 1968 film
‘Yours, Mine and Ours’ starring Lucille Ball and
Henry Fonda, in which a widower with ten children
marries a widow with eight, and the twenty of
them try to come together as one big happy family.
In summary, ARM is challenging x86 in the
embedded space and x86 is counterpunching. As
a result the decision whether to go with x86 or
with ARM has become more complex and more
difficult than ever. Two things we do know: 1)
Among the factors that you will have to take into
account are performance, price, power consump-
tion, MIPS/W, hardware and software support,
Source: AMD

the volume you have in mind and more, and 2) as

always the best place to go for parts and design
help is to your friends at element14.com. n AMD’S MULLINS PROCESSOR

WINTER 2014 ELEMENT14 43

ARM Technology Part 2
System IP
By ASHOK RAO / SENIOR SUPPORT ENGINEER, FARNELL ELEMENT14
Following up from our last article on classes of
ARM processors, their advantages, applications,
we now take a look into the other peripherals and
technologies surrounding an ARM based core.
This article focuses on a few key components
of the System IP from ARM, with a focus
around the hardware blocks inside an SoC.
44 ELEMENT14 WINTER 2014
YSTEM IP IS THAT PART OF
S an embedded system comprising of the vari-
ous modules working in the background in
order for the foreground peripherals to function as
required. This could include the inter-module communi-
cation networks, debug and trace IP, cache/DMA/
interrupt/memory controllers etc. In other words,
these are the nuts and bolts that connect the bigger
blocks together. Let us look at them to try and under-
stand in layman’s terms what these are, how they ben-
efit the larger ecosystem and where all this fits.
CoreLink 500 series
CoreLink system IP and design
tools can conveniently be used for
designing and building SoCs based
on the AMBA specifications. The
AMBA specifications maximise the
efficiency of data movement and
storage without compromising on
performance. In other words, the
latency involved in data transfer
is considerably reduced using the
AMBA specifications. The CoreLink
500 series comprises the CCN-504
(CCN: Cache Coherent Network)
and CCN-508. These provide up to
1.6 Tbps throughput. This enables
full cache coherency between up
to 16 CPUs for CCN-504 and up
to 32 CPUs for CCN-508. What do
these numbers mean? It means,
these can be paired with CoreLink
DMC-520 (Dynamic memory con-
troller) for DDR3 DRAM interfaces at
up to 3200Mbps. What this means
for you and me? Answer: Very fast
RAM. The CCN-508 is more likely
to be used in enterprise solutions
such as storage, servers and net-
work infrastructure.
The CCN-504 is the little brother
of the CCN-508. The performance
is slightly lower with scaling up to
16 CPUs and up to 1Tbps of sys-
tem bandwidth.
CoreLink 400 series
The CoreLink 400 series pro-
vides similar functionality as the
CoreLink 500 series except that
it can support up to 8 CPUs and
is optimised for mobile applica-
tions including big.LITTLE tech-
nology, automotive infotainment
and enterprise networking. The
400 series comprises of CCI-400
(Cache Coherent Interconnect) and
NIC-400 (Network Interconnect).
The NIC-400 provides a fully
configurable, hierarchical, low
latency, low power connectiv-
ity for devices implementing the
AMBA specifications.

CoreLink™
CoreLink technology is a part of the system IP that enables low latency, lower power
consumption, multi channel memory controllers on the high performance Cortex and Mali
processors. It can be safely visualised as the interconnection between the core and the
peripherals. The CoreLink technology is further subdivided into interconnects, memory
controllers, security features such as TrustZone and AMBA® (Advanced Microcontroller
Bus Architecture). This family provides on-chip AMBA connectivity for the different
components implementing the AMBA specifications. Let us now discuss the various
interconnects such as CoreLink 500 and CoreLink 400 series and memory controllers.

Memory controllers
Shared off-chip memory access
plays a crucial part in influenc-
ing system performance, power
and cost. CoreLink DMC (Dynamic
Memory Controllers) deliver the
best-in-class performance and
power for AMBA based SoCs.
The DMCs are further subdivided
into the smaller 400 series and
the beefier 500 series. The 400
series significantly enhances
memory bus utilisation and pro-
cessor performance through QoS
mechanisms. These are usually
coupled with the CoreLink 400
series of interconnects as dis-
cussed above.
The CoreLink 500 series of
DMCs implements similar fea-
tures as the 400 series but is
optimised for enterprise level
solutions providing additional
features such as TrustZone and
end to end QoS. The CoreLink
DMC 520 is ARM’s 5th genera-
tion of memory controllers. For
any enterprise level SoC, as the
number of components increases,
so does the demand for data. To
satisfy this need for fast data,
efficient access to the DRAM
is required and is critical to the
performance of the SoC. The
CoreLink 520 has advanced QoS
based scheduling and arbitration
algorithms that handle read/write
access to the memory. A multi
level parallelism is also adopted
to increase performance.
CoreLink 400 is the little broth-
er of CoreLink 520 that can be
used in conjunction with the 400
series of interconnects as dis-
cussed previously. There is also
an older generation of DMCs, the
32x series that was optimised for
DDR, DDR2, SDR and other such
older generation of memory tech-
nologies. These are still in use but
their popularity has considerably
reduced over the years. E

WINTER 2014 ELEMENT14 45

 Debug and Trace IP
ARM offers one of the best on-chip debug and real-time trace
solutions for the entire SoC. These make ARM based SoCs some
of the easiest to debug. Technologies such as CoreSight™ and
Trace Macrocells work in parallel to help developers design, debug
and optimise code in the fastest and easiest ways possible.

CoreSight ■■ For application engineers, appli- real time visibility of their code exe- ■■ PTM: is used for program
CoreSight as the name suggests cations can be optimised even cution. An advantage of using the flow trace specifically with the
is something that looks inside the at assembly level using system TMC is that traditional designs tend Cortex-A9.
core. CoreSight technology enables trace and performance counters. to generate dynamic bandwidth ■■ ITM: used for high level software

developers to develop and optimise It can also prove beneficial for of trace. This usually varies with view (the C/C++ layer).
software for ARM based devices with power and performance analysis time, the type of application and ■■ HTM: is used for performance

fewer pins on the SoC itself. For the
Cortex-A series of processors the
debug can be performed with only
two pins using the Serial wire tech-
nology or alternatively using JTAG.
Some of the other advantages of
using CoreSight are listed below:
■■ Using the CoreSight DAP and
embedded cross triggering, it
is possible to perform multicore
debug and run time control while
also collecting time stamped CPU
or system trace to analyse, opti-
mise and debug software and
hardware interaction.
■■ CoreSight trace macrocells enable
developers to examine how the
software behaves on the tar-
get platform.
■■ CoreSight also provides provisions
for secure debug and trace. It also
allows designers to implement a
debug and trace solution with very
low power consumption.
■■ CoreSight is highly re-usable
enabling SoC developers reduce
design and development time.
Some of the real world examples
where CoreSight proves advanta-
geous are listed below:
■■ For embedded software/middle-
ware developers, device driver
development can be faster by
observing the CPU trace while
executing software in real time.
on the SoC.
■■ For product engineers, it can be
useful in failure analysis.
The CoreSight technology has
also been applied not only to
the Cortex-A series but also the
Cortex-R and Cortex-M series of
processors as well. On the Cortex-R
series, CoreSight provides non-
intrusive trace that helps in under-
standing and optimising how real
time software operates on the tar-
get. This is achieved using a set of
Trace macrocells that are discussed
in the later part of this article. On
the Cortex-M series, CoreSight
provides a low cost debug and
trace functionality using a whole
gamut of macrocells, breakpoints
and watchpoints.
CoreSight components
The CoreSight components along
with the trace macrocells pro-
vide the necessary infrastructure
required for effective development
and debugging. Two key compo-
nents comprise the CoreSight: the
trace memory controller and the
debug access port (SWD/JTAG).
The TMC (Trace Memory Controller)
provides low cost real time trace
all along the development cycle
(including production) for third party
software developers giving them
occasionally causes peaks leading
to loss of trace data. Whereas the
TMC uses a FIFO model enabling
trace over a longer period of time
thus reducing risk of overflows.
This in turn also allows for a small-
er trace port reducing the cost of
implementation of a trace port on
the SoC.
The Serial Wire Debug is a widely
known debug technology used
mostly as an alternate to JTAG. It
needs just two pins and is avail-
able as part of the CoreSight DAP
(Debug Access Port).
Trace macrocells
Trace macrocells provide non-intru-
sive visibility of the components in
the SoC. These are individual blocks
that help capture trace data based
on their particular function. There
are four different types of trace
macrocells viz. Embedded Trace
Macrocell (ETM), Program Trace
Macrocell (PTM), Instrumentation
trace Macrocell (ITM) and AHB
Trace Macrocell (HTM) (AHB:
Advanced Hardware Bus). The
usage of each of these is briefly
outlined below:
■■ ETM: used for non-intrusive, cycle
accurate program and data trace
for ARM processors. This occurs
at the assembly instruction level.
and functional debug at the hard-
ware level.
What is the advantage of hav-
ing all of these macro cells? Well,
conventional debugging requires
breakpoints/watch points to halt the
processor and use a debug con-
nection from there to read, examine
and modify register contents. This
method is intrusive, requires the
processor to stop, is non real-time
and does not give an indication of
true performance.
Whereas, using trace capture,
developers can capture real-time
data including instruction level
data, compress it and deliver it
either on-chip or off-chip. This
leads to real-time data analysis,
is non-intrusive and does not
halt the processor and is much
more accurate for performance
analysis. The various trace mac-
rocells mentioned above are
all essential components of the
entire debug and trace ecosys-
tem for a SoC.
A latest development in the
series of macro cells is the STM
(System Trace Macrocell) that
offers developers printf (the
well-known C language output
statement) style debugging giv-
ing visibility of the core from the
application layer itself.

46 ELEMENT14 WINTER 2014

System controllers
System controllers are the peripherals that work in
conjunction with all the other controllers discussed
above. These involve the cache controller, DMA
controller, Interrupt controller, TrustZone™ and the
system MMU (Memory Management Unit).

Cache controllers
Cache controllers play a critical part
in CPU to on-chip data access. They
reduce the latency of off-chip data
access by about 10-25%. This reduc-
tion in CPU demands for off-chip data
means it can now be used for other
tasks. Lower demand for off-chip data
also reduces power consumption.
Thus cache controllers contribute to
power efficiency and performance.
DMA controllers
DMA control plays an important part
in reducing the data transfer load on
the CPU. It takes the task of mov-
ing in and out large blocks of data
from peripherals to memory and/or
from memory to memory. This data
movement activity by a separate unit
greatly reduces the load on the CPU.
ARM provides two DMA controllers:
DMA-330 and DMA-230. The DMA-
330 is a highly flexible, configur-
able DMAC and is mainly targeted at
higher end SoCs. The DMA-230 is a
simpler version of the DMA-330 and
is ideal for use with Cortex-M series
designs. Both of these support trans-
fer modes such as Scatter-Gather, cious software attacks? Answer: controller, PS/2 interface, GPIO con-
Ping-Pong and Primary-Alternate. Very important. ARM provides trollers to name a few. Detailed dis-
TrustZone technology to prevent cussion of these is beyond the scope
Interrupt controllers access to selected memory regions of this article.
As the complexity increases with and peripherals such as screens
multicore CPUs and their handling an and keypads. TrustZone extends the Where does all this fit?
ever increasing number of interrupts security to both on-chip and off-chip SoC design is a very complex task
from peripherals, so does the require- memories. The technology works by and various factors need to be con-
ment for interrupt controllers. ARM partitioning the available memory sidered to achieve the required per-
provides many GICs (Generic Interrupt into ‘N’ number of secure regions, formance. This even applies to the
Controllers) to suit various needs. The access to which is restricted. For extent of using the optimum length
GIC-500 is the latest addition and on-chip security, the TrustZone Boot of the metal for interconnections on
supports the ARMv8 architecture. It ROM provides 8-16KB for signature the chip though micro/nano meters
can handle up to 128 CPUs and 480 code check. It also provides on-chip long. At this level of required accu-
shared peripheral interrupts. The GIC- memory for secure data and on-chip racy, every single block matters to
400 is the younger sibling that can RAM for secure code and data. For achieve maximum performance.
handle up to 8 CPUs and 480 inter- off-chip security, it provides options Nobody likes to wait forever for
rupts. The GIC-390 is based on vector to partition the available RAM into software to load or a video to start
addressing and stores the CPUs vec- up to 16 secure regions making playing. DMA controllers, cache
tor addresses in memory. The PL-190 it immune to software attacks. All controllers, interrupt controllers
can handle up to 32 interrupt sources this is controlled by the TrustZone etc, all play a vital part in bring-
and 16 vectored addresses while the Address Space Controllers TZC-380 ing everything together to help SoC
PL-192 supports up to 32 vectored and TZC-400. designers build the ultimate chip
interrupts with programmable priority for any dedicated task. Software
levels and masking. CoreLink peripherals and application engineers then use
A number of APB (Advanced these chips to generate software
TrustZone controllers Peripheral Bus) are provided from that efficiently utilises all the avail-
Question: How important is security ARM to ease building complex SoCs. able resources on the chip to create
for your code and stopping mali- Some of these are the UART, SPI immersive and useful applications.

CONCLUSION
The performance of an ARM based processor is not the sole responsibility
of only the core. There are a number of peripherals acting around
the core that influence the performance and efficiency of any ARM
based processor. There will also be a number of factors affecting the
communication speed between the core and peripherals which need
to be optimised. All this processing power and speed also needs to be
channelled properly to the outside world for developers to optimise their
designs and software. All the peripherals discussed in this article are the
skeletal connecting blocks, in other words the threads that hold the fabric
of the processor together and provide the required throughput. In the next
part of this series, we will discuss some of the other technologies from
ARM such as security on ARM based devices and a few other software
tools required for an effective design and development of an ARM based
embedded system. Until the next edition of element14 Tech Journal... n

WINTER 2014 ELEMENT14 47

Designing with Cortex-M MCUs?

Get the same power and spend less, with

KEIL DEVELOPMENT TOOLS
Discover the same functionality found in the Standard Edition – yet supporting only Cortex-M devices. It’s a complete
software development environment for Cortex™-M, processor-based devices, specifically designed for microcontroller
applications. It’s also easy to learn and use, yet powerful enough for the most demanding embedded applications.

THE NEW KEIL-MDK CORTEX-M EDITION OFFERS THE SAME FUNCTIONALITY NOW WITH ARM
CORTEX-M7
AS THE KEIL-MDK STANDARD AT A FRACTION OF THE COST! SUPPORT

Interested in ARM/Keil development tools?

Speak to our ARM certified FAEs: ARMSupport@farnell.com

Developing on an ARM-based processor?

SAVE A BUNDLE!
DSTREAM + DS-5 BUNDLES
We’ve combined DSTREAM hardware and DS-5 software tools into a complementary solution-
package to save you money. Check out both the node locked and floating license versions.

Our ARM Bundles include:

DSTREAM STREAMLINE
The ARM DSTREAM high performance, debug and trace unit enables The ARM StreamlineTM Performance Analyzer is the Linux and
powerful software debug and optimization on any ARM proces- Android performance analysis tool in DS-5.
sor-based hardware target.

ARM C/C++ COMPILER DS-5 DEBUGGER

The ARM Compiler in DS-5 Professional Edition is the only commer- The DS-5 Debugger uses the most advanced ARM technologies to equip
cial compiler co-developed with the ARM processors and specifically developers with a complete debug solution for software development
designed to optimally support the ARM architecture. projects, from hardware bring-up to application development and tuning.

Get yours today! For more information on ARM embedded products and resources please visit
www.element14.com/ARMPromo