Professional Documents
Culture Documents
Tech Journal Exploring The Internet of Things
Tech Journal Exploring The Internet of Things
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
COMMUNITY
Welcome
Issue 2 WINTER 2014
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
In the recent past, devices were singular WE NOW HAVE CONSTANT CON- promises of this technology together
nectivity and devices that trade mas- is the Internet of Things (IoT), a con-
entities going about their tasks in silent sive amounts of data; data that holds federacy of devices, networks, and
anonymity. Sure, some industrial secrets we are only beginning to tap. processing power that has vast poten-
Transportation, healthcare, agricul- tial to make life better by address-
systems and computers within ture, machine-to-machine communica- ing truly global problems. Freescale
networked environments collected tions, and entertainment are segments has been a pioneer at every level of
that are already reaping an informa- the IoT phenomenon and offers all of
data and created a community of sorts, tion harvest that was unimaginable the fundamental IoT building blocks
but nothing like what we have today. 10 years ago. The glue that holds the under one roof.
Application/Action
Big Data
Realized.
Energy i.MX applications
Industry’s most scalable
Sensors ultra-low-power, processors QorIQ processors built on
mixed-signal MCU
Connectivity 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.
●●INFRASTRUCTURE OF THE INTERNET OF THINGS sensing
The right combination
of intelligent
integration, logic and
customizable software
on the platform to
deliver smarter,
more differentiated
applications.
Edge nodes
●●FREESCALE IOT
SOLUTIONS MAPPING
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
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.
tools are open source, modular and may work with the baseboard without quickly create a public web dashboard
affordable, which enables developers additional software/driver porting. that enables anyone anywhere in the
to plug together various sub-circuitries An example of how this type of rapid world to visualise and monitor the data
to build complete systems. Where addi- prototyping system can work is where in the star network that we created.
tional functionality is known to be need- designers may need to develop an While modular, open source hardware
ed, such as in the case of low-power application that is out of their normal is important to enable efficiency in a
wireless connectivity to enable IoT expertise, such as the development fast paced industry, open source and
endpoint communications, the base of a complete wireless sensor net- easy-to-use software examples are just
boards may integrate additional func- work. Here we have started with the as critical.
tionality such as the TI CC3200 Wi-Fi TI CC3200 Wi-Fi LaunchPad which will To create the system above, we used
LaunchPad, which is a complete devel- serves as a sub-1GHz-to-Wi-Fi gate- a tool called Energia (energia.nu).
opment kit that features the CC3200, way. By adding a CC110L SubGHz Energia is an open source, community
a single chip microcontroller + Wi-Fi RF BoosterPack on top of the CC3200 driven software development envi-
network processor. Many of the base- Wi-Fi LaunchPad, we are able to eas- ronment that is a fork of the popular
boards also include an on-board debug- ily create a cloud-connected gateway Wiring/Arduino framework. Through
ger/programmer as well as a stand- that operates as the centre of a sub- software abstraction, Energia offers
ardised interface for accepting their 1GHz RF star network. In this exam- intuitive APIs and libraries that are at
accessory boards or plug in modules. ple, each sensor node is made up of the functional level, enabling develop-
Some manufacturers such as TI have a Texas Instruments Microcontroller ers to focus on what they want to do
developed their own standardised type LaunchPad kit + CC110L SubGHz RF as opposed to spending time figuring
of plug in modules called BoosterPacks. Wireless BoosterPack + a sensor. In out how to actually do it.
The benefit of this type of system is this configuration, each node sends its With the help of these rapid prototyp-
that a designer is assured to have a sensor data to the CC3200 LaunchPad ing tools, developers are well-equipped
tested accessory solution that they can via SubGHz RF. The CC3200 receives to solve the problems of a connecting
get up and running with little effort. the data, encodes the datapoints into world. Through open source method-
Many manufacturers have chosen to a JSON object and publishes it to the ologies, developers are able to freely
use add-on boards with footprints that cloud over Wi-Fi via MQTT, a light- share and collaborate to develop the
might be compatible with a wide range weight publish/subscribe communi- building blocks we need to build the
of accessories from other manufactur- cation protocol. IoT. Hardware and software develop-
ers (such as Arduino Shields) which Once the data is in the cloud, we ers are able to benefit from these tools
can potentially broaden the field of can visualise the real-time sensor data that bring powerful technology to a
applicable accessories, with the limi- on a cloud-side dashboard as shown growing developer community and
tation however that not all accessories above. In this example, we were able to lower the barrier of entry. n
LIFE ON THE
EDGE OF IoT
UNDERSTANDING
ULTRA-LOW POWER
M
ing on the edge must include smart watch, smart glasses,
utilise ultra-low heart rate monitor, pedometer, GPS
p owe r m i c ro c o n - tracking device, blood sugar moni-
trollers to handle the tor, music or video player, and wire-
user interfaces, collect and trans- less headset/microphone. There are
mit sensor data, provide security also categories for health, environ-
functions, and manage other tasks. mental, and others.
One issue faced by ‘thing’ design- Many edge nodes in the ‘weara-
ers is ensuring that these micro- ble’ domain, as well as other domains
controllers are really optimised to can be considered ultra-low power
meet the performance needs of applications. These applications
their applications and to enable the are typically running with battery-
long battery lives that are expected. powered operation and implement
Edge nodes can be grouped more short, occasional periods of activity
or less arbitrarily according to their interspersed with long periods of
application domain. For example, inactivity, and possibly infrequent
home automation encompasses human intervention. It’s this latter
anything that is used to control or point that highlights energy efficien-
monitor the home or office systems cy as a key performance criterion
and devices, such as lighting or for such devices, and dictates bat-
environmental control, appliances tery lives of weeks, months, years,
(e.g. freezer, washing machine, cof- or even decades (i.e. a smartphone
fee maker, or fire alarm. On the other is not an ultra-low power device,
hand, ‘wearable’ or ‘portable’ is although some companies would
anything that is worn or carried on choose to market it as such). E
■■
Arduino R3 compatible headers
both active power and idle (sleep) power.
The workload of ULPBench core best to compile the code for maxi-
profile consumes 10,000-20,000 CPU mum performance (minimise the
cycles each duty cycle, depend- active cycles). Not only can this Texas Instruments
ing on the microcontroller’s archi- be accomplished by selecting the
CC3200 LaunchPad
tecture. Although the number of most optimal compiler flags, but
CPU cycles varies, each microcon- the selection of the compiler itself element14.com/LaunchPad
troller is required to perform the is very important (just like it is for Created for the IoT, the CC3200 LaunchPad
same amount of work (as opposed any code development). is a wireless MCU that integrates a
high‑performance ARM Cortex-M4.
to necessitating an arbitrary num- If the main (faster) crystal oscil-
ber of cycles). The diagram below lator is power hungry, another way ■■
CC3200 Wi-Fi application processor
portrays the duty cycle, whereby to lower energy consumption is ■■
Standalone development platform featuring
the device enters the active mode to try and run from a low power sensors, LEDs and push-buttons
once per second; the testing setup 32kHz crystal with the main crystal ■■
Supports 4 wire JTAG and 2 Wire SWD
requires a real-time clock (RTC) for oscillator switched off completely. ■■
GDB support over OpenOCD
waking up, typically with an accu- Furthermore, if the device’s leakage
racy of 20-30 ppm. current during the active state is
There are no golden rules for high, it might make sense to switch
achieve the best ULPBench results. to a faster clock during active state
Obviously, since the active mode and finish processing the workload
makes a big portion of the ener- quicker (and in turn return to sleep
gy consumed during the test, it’s mode quicker). 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.
■■
■■
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
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.
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
an industry standard should battery. Other factors, such hardware and software is cycle restrictions in the
be adopted. The frequency as systems compatibility available to you. sub-1GHz band as well. By
band must be determined, and network topology, will Each of the three most comparison, the 5GHz band
as well as the standard wire- also influence most wireless popular bands – sub-1GHz, affords designers the high-
less benchmarks of transmis- connectivity designs. 2.4GHz, and 5GHz – has est available data rate of
sion range, power consump- cer tain advantages and the three, at the expense of
tion, and data throughput. Operating band disadvantages for any given transmission range.
Within these benchmarks, of The operating band is one application; it all depends Sub-1Ghz and 5GHz bands
course, there are trade-offs, of the design points that upon the benchmark you’re both currently have a conges-
especially when it comes to many developers struggle designing for. For example, tion advantage over 2.4Ghz,
power. Data throughput, for with. If the various stand- range is generally reduced which has been embraced
example, impacts the over- ards-based technologies by about half as frequen- by consumer technology and
all power consumption; a comprise a menu of design cy doubles, so long-range as such has grown rather
54Mbps Wi-Fi-based network options, then selecting an applications may require crowded. For this reason,
might last a day or two on operating band is not unlike sub-1GHz operation. One some consumer applica-
a single Lithium Ion source, choosing the style of food of the trade-offs with lower- tions are beginning to target
while an IEEE 802.15.4-based you want: Italian, Thai, frequency operation is data 5GHz operation. But for all
sensor network shuttling American (whatever that is). rate, which diminishes with its crowding issues, 2.4GHz
data at 250kbps might sur- Whatever band you select frequency. Designers may remains a popular choice
vive a few years on a small helps to narrow down which run into the issue of duty for wireless connectivity. E
●●Bluetooth n bluetooth.com
With an installed base of more mise power consumption; AAA and unlicensed in most coun-
than 3 billion units, Bluetooth batteries are sufficient sources tries. Data throughput is less
is designed for lower-power of power for Bluetooth prod- than that offered by Wi-Fi but
wireless connectivity between ucts. Bluetooth operates in the still in the 2 Mbits/second range
devices. Bluetooth devices can industrial, scientific, and medi- to support higher data-rate
power down during periods of cal (ISM) band at 2.4 GHz to applications, including multime-
SMART inactivity, which helps to mini- 2.485 GHz, which is available dia products.
●●ANT n dynastream.com
ANT provides a simple, low-cost trast to BLE, ANT has a relative- able for P2P or star topologies,
and ultra-low power solution ly simple protocol and doesn’t while ANT supports the full
for short-range wireless com- require a lot of overhead. It range, including tree and mesh.
munication in point-to-point and specifies 64-bit security versus On a side-by-side, technological
more complex network topolo- 128-bit utilised by BLE. Both comparison, ANT is arguably
gies. Suitable for a wide range offer data rates of about 1Mbps, the more appealing option for
of applications, ANT is today accommodate a range of a few wearable/IoT applications, if
a proven and established tech- tens of meters, and boast years for no other reason than mesh
nology for collection, automatic of life on a coin cell battery. But support. But it’s worth noting
transfer and tracking of sensor perhaps the greatest difference once more the numbers associ-
data within sports, wellness for system-minded designers ated with Bluetooth: it’s hard
management and home health lies in the network topologies to ignore an installed base of
monitoring applications. In con- supported by each; BLE is suit- 3B units.
The platform bar is a trademark of Texas Instruments. © 2014 Texas Instruments Incorporated
E It offers a reasonably high customised specifically for proprietary option. But with
data rate, decent range, no the application, decreas- these sacrifices come the
duty cycle restrictions, and,
notably, 2.4GHz band works
STANDARDS ing the software footprint,
among other things. And the
real advantages of OEM-
independent interoperable
worldwide, making it a slam- BASED dual-edged nature of any pro- nodes – namely, a freer user
dunk for designers.
SOLUTIONS prietary design applies to
wireless connectivity: propri-
experience for consumers
and greater opportunities
Proprietary vs standard AFFORD THE etary devices may be simpler for network expansion.
The decision to develop a
proprietary solution versus
DESIGNER A to design because they only
have to operate within one
The more nodes, the big-
ger the network.
leveraging a standard tech- WEALTH OF defined network, but those
nology such as Bluetooth
often boils down to the
CONSUMER same devices won’t be able
to communicate with stand-
Honed on the range
Among the most important
basic functionality required. FRIENDLY ard consumer devices. wireless connectivity design
Highly-specialised functions
may necessitate a proprie-
CONNECTIVITY Standards-based solutions,
on the other hand, afford
parameters is transmission
range, which consists of two
tary implementation, which OPTIONS. the designer a wealth of key factors: transmission
immediately limits the num- consumer-fr iendly con- power and receiver sensitiv-
ber of solutions available nectivity options. You just ity. These are critical specs
from OEM suppliers. The have to cough up the code to consider when exploring
upside is that a proprietary space and possibly sacri- IC options from semiconduc-
wireless platform may be fice some power versus the tor OEMs.
Cloud computing enables ubiquitous business operation. Cloud based ser- infrastructure setup within the data
network access to connected hard- vices are largely subscription based centres. Private clouds are typically
ware and software in remote locations or pay-per-use (metered). The metered within organisations, behind a firewall
over the internet. The model allows services allow for customers to use and quite capital intensive. Hybrid
for applications, software, process only what they need, thereby keeping clouds are a cluster or combination of
management and data to be stored expenses low. private and /or public clouds, to sup-
on virtualised servers. Users can then The deployment models are primarily port temporary growth requirements.
access the data and connected devices based on the resource management, Clearly the IoT paradigm will eventu-
through mobile apps, web browsers accessibility, elasticity and most impor- ally work for all of us, but not without
or lightweight desktop environments. tantly the security of the cloud: pri- first addressing the challenges related
The shared resource structure for stor- vate, public or hybrid clouds. Public to security, data control, non-standard
age, software, networks and services clouds service providers such as AWS, infrastructure and the handling of vol-
allows for a scalable and lower cost Microsoft and Google have their own umes of heterogeneous data. n
NETWORKING
MIDDLEWARE
RUNTIME
STORAGE
SERVERS
O/S
NETWORKING
MIDDLEWARE
STORAGE
SERVERS
O/S
from the IoT devices is acquired and open source libraries allow for easy
collected through RESTful APIs to the integration of the Xively APIs onto
cloud. The information thus obtained embedded products. For example, the
YOU MANAGE MANAGED BY VENDOR
is then computed and applications Xively ARM mbed library allows you
developed to communicate/control to develop on industry-standard ARM
● PLATFORM AS A SERVICE the devices. Real-time messaging architectures to make your code easily
is supported with MQTT (Message portable to production-grade chips.
Queued Telemetry Transport) and Similarly Plot.ly also allows sensor
similar connectivity protocols. Data data to be streamed directly from
analytics, querying and reporting hardware platforms like Arduino,
capabilities are well supported Raspberry Pi and also from MATLAB
with data visualisation tools as data acquisition systems. The
well. Higher end complex machine data received can be plotted into
learning algorithms are also being interactive charts using Python, R,
VISUALISATION
APPLICATIONS
NETWORKING
MIDDLEWARE
STORAGE
SERVERS
that can support the storage and exported to any image format through
processing of these large amounts of their Workspace or APIs, including
DATA
user patches and maintains the It enables people find devices (eg,
APPLICATIONS
NETWORKING
MIDDLEWARE
STORAGE
SERVERS
The true success of the IoT revolution (like air quality, seismic activity,
O/S
L A M P
LINUX APACHE MYSQL PHP
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,
INTRODUCTION TO
FPGAs
FPGAS
ARE VERY
VERSATILE
DEVICES AND
THAT THE ONLY
LIMITATION IS
HAVE YOU WANTED TO START DESIGNING THE CAPACITY
WITH FIELD PROGRAMMABLE GATE ARRAY OF THE DEVICE
BUT FOUND THE WHOLE TOPIC VERY AND YOUR
DAUNTING? WELL YOU ARE NOT ALONE… IMAGINATION.
By SIMON HOLT / STRATEGIC ALLIANCE MARKETING MANAGER, PREMIER FARNELL
HIS ARTICLE AIMS TO GIVE of many thousands, even millions and require the user program their
T you a great starting point to
enter into the world of pro-
of transistors. These Transistors are
grouped together to perform an array
design onto it.
grammable logic. Once you have start- of functions, from a simple logic gate Disadvantages
ed to understand this technology, it is to a complex Clock circuitry block. The Probably one of the biggest reasons
surprisingly easy to use and will offer majority of these blocks are available why people don’t want to use FPGAs
you an electronics design option differ- to you to configure using the FPGA is the steep learning curve in designing
ent to anything else available on the software tools. with them. It can take many hours to
market today. learn the design tools, coding language
We will start by covering the basics Advantages of FPGA and debug software in order to get your
of FPGAs, explaining the difference FPGAs are completely customisable design into the device and working as
between the three most popular pro- and as long as there are enough build- you want.
gramming technologies currently avail- ing blocks in the FPGA you are using, In the past the price of FPGAs has
able. This will be followed by an over- they can be configured to perform any often put designers off using them, but
view of the different building blocks digital design. Due to the layout of an in recent years the prices have dropped
that can be found within an FPGA. FPGA they lend themselves nicely to to a level which is more attractive bear-
An FPGA is a programmable device parallel processing designs, aided by ing in mind the flexibility they offer.
that allows you to create your own the high I/O count normally found The architecture of an FPGA means
digital chip. The basic building block on them. Depending on the type of that it generally consumes quite a bit
of an FPGA is the humble transistor. FPGA you use, they have the ability of power, although there are some
Depending on the density of the FPGA to be re-programmed. These devices design techniques available to com-
device you decide to use, it can consist are completely blank on power up bat this issue. E
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.
●●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
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
DIMMING LEVEL
GREEN
The other is colour change, a change in rela- INTENSITY
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
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
EDs are used for lamps and full control of brightness, colour and DALI, one of the most popular lighting
L light engines in offices, retail,
hospitality, street lighting, high
dynamic behaviour of LEDs as accu-
rately, precisely, efficiently and cheaply
communication standards.
The little software required to con-
bay lighting, healthcare, and more. as possible. trol the BCCU can be created with
LED lighting is efficient, allows maxi- BCCU generates pulse-density modu- DAVE™ IDE. DAVE supports automatic
mum design freedom, and it can be lated bit streams with randomisation code generation, provides DAVE Apps
intelligent as well as easy to use. for better EMI behaviour, and bit pack- with GUI programming, and exam-
Thanks to the latest RGB LED Light- ing feature to be compatible with sev- ple projects.
ing Shield with XMC1202 for Arduino eral LED driver designs. Infineon’s XMC1200 MCU series as
developers can easily explore, eval- To offload the CPU automatic dim- part of the XMC portfolio is the ideal
uate and build their next LED pow- ming is provided by adjusting the choice for intelligent LED lighting solu-
ered product. brightness levels along an exponen- tions. XMC1200 microcontrollers pro-
The RGB LED Lighting Shield with tial curve: the user only needs to se- vide all essentials for high quality colour,
XMC1202 for Arduino is built on the lect target brightness levels and fade such as full dimming control, flicker-free
XMC1200 MCU series which has a rates, and the dimming engines take light at all dimming levels with low
Brightness and Colour Control Unit care of the rest. energy consumption and ease-of-use
(BCCU). BCCU is dedicated for multi- The ability to dim along an expo- thanks to the new RGB LED Lighting
channel lighting engines by offering nential curve is also a requirement by Shield with XMC1202 for Arduino. n
Isolated DMX512
XMC1202 Control Interface I²C, GPIO SWD
(optional) XMC1202
DC Input Voltage
ENERGY
EFFICIENT
WHITE LIGHT
WIRE BOND
ANVIL
POST
ANODE
CATHODE
P-LAYER
ACTIVE
LAYER
N-LAYER
ELECTRON
SURE, THEY’RE BLUE, BUT WE CAN’T IT’S ALL IN THE GaN: GALLIUM NITRIDE REVOLUTIONISED
HAVE WHITE LIGHT WITHOUT THEM. BLUE LED LIGHTING PAVING THE WAY FOR WHITE.
Scientists win Nobel Prize in physics with their creation of blue LEDs
ONS AGO (ROUGHLY to the fact scientists couldn’t produce In the early 90s, Professors Isamu
E
1907), scientists began blue light in LED form. Akasaki (Nagoya University), Hiroshi
messing with the idea of Confused yet? Okay, let’s break it down Amano (Nagoya University) and Shuji
LEDs in the form of elec- in simple terms- in order to create bright, Nakamura (University of California)
troluminescence using a efficient white light, green or red LEDs began working on a way to grow gal-
crystal of silicon carbide and a cat’s- need to be combined with a blue LED. lium crystals big enough to be used to
whisker detector. Apparently, Edison’s They could also be shown through a emit blue light. They found that they
light bulb was no longer cutting the phosphor that emits red and green light. could do so using a specially designed
mustard and people wanted tiny little Sure there have been blue LEDs in ex- scaffold made in part with sapphire.
lights in their radios, which is probably istence since the 70s but the material That’s the key, different materials emit
why RCA was playing around with they are made from made them pretty a different colored light when applied
gallium arsenide-based semiconductors much useless in the broad spectrum to LEDs and gallium was the ticket in
back in 1955. (pun intended) of practical applications. emitting blue.
Regardless, LEDs have come a long way By the late 80s, the people’s demand Gallium isn’t without its troubles as
in the last hundred years and has become for bright, shiny things continued to it tends to become poisoned when ex-
the dominant light source on the planet grow and there still were not any de- posed to hydrogen. The people would
overtaking incandescent and fluorescent vices (much less radios) with tiny LEDs have to wait a few years more before
lighting. While the LEDs manufactured that glowed bright white or blue. No- they would get bright, shiny blue or
today are highly efficient, energy saving ticing those demands, three material white lights. To get around that issue,
and environmentally friendly, they do scientists decided the world could wait the scientists ‘doped’ up the gallium
have their drawbacks, in this case the no longer for bright, efficient blue LEDs using aluminium or zinc but ultimately
ability to generate a broad spectrum of and began developing their own using went with indium, which protects the
colours. More specifically, white light due gallium nitride. gallium from the hydrogen used in
manufacturing the semiconductor the
LED resides on.
The hard work that all three endeav-
ANODE (p-electrode) oured over the years to get an efficient
blue LED earned the trio a Nobel Prize
this year in physics. Not only does
p-GaN
their work provide bright and shiny
p-AIGaN LEDs in our mobile devices but also
Zinc-doped InGaN reduces the power draw on the grid
when used for lighting our homes. Not
n-AIGaN only that, they are also better for the
environment over compact florescent
n-GaN
lighting (CFLs) due to the fact that the
GaN Buffer Layer latter is filled with mercury. The trio
Sapphire Substrate received their prizes in December,
where their breakthroughs could be
CATHODE (n-electrode) seen lighting up the homes all over
the city. n
>
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
while the X86 (for the purposes of power and very short real-
this article x86 will refer to both time response.
32-bit x86 and 64-bit x64, ARM As for run-time stability (uptime
will refer to all ARM derivatives) without crashes), which can be
has a Complex Instruction Set a key deciding factor, no clear
> > Computer (CISC) architecture. cut winner here: Both ARM and
CISC chips have an instruction set x86 processors are relatively well
capable of doing complex things suited to be used in industrial
with a single instruction while and other embedded systems
BSP a RISC processor has a smaller that require stable, reliable and
Boa / Dr
rd S number of more general purpose deterministic run-time behaviour.
upp
ort
iver instructions. It is important to Now to the changes I alluded to.
Pac > note, however, that the simpler ARM and Intel chips are growing
kag nature of a RISC instruction allows closer to each other and the tra-
e
it to be processed in usually just ditional lines distinguishing them
one clock cycle (and often with are starting to blur. Here are a few
fewer transistors, making the for instances.
silicon less expensive and more Intel is positioning Quark, its
power efficient). Generally speak- smallest processor, in the IoT
ing, a RISC-based ARM processor device space supported by a
also requires more memory than new VxWorks 7 version of the
a CISC-based processor. widely used real time operating
Traditionally ARM’s low power system (RTOS), developed by the
has enabled battery life that Intel subsidiary Wind River, As a
would be better than an x86 counter-measure ARM is launch-
equivalent and with its built-in ing a new, and free low power
multimedia decoding capabilities operating system to manage web-
it also enables 1080p HD video connected devices that use micro-
playback in a small package rela- controllers based on the compa-
tive to the real estate an x86 sys- ny’s 32-bit Cortex-M architecture.
tem would require. ARM’s OS for this is called mbed,
KEY DIFFERENCES OF X86 VS. ARM Typically, too, for complex tasks Although technically not a true
VA BOARD SUPPORT PACKAGE, AN such as would be needed to pro- RTOS (as a strict definition would
IMPLEMENTATION OF SPECIFIC SUPPORT cess and analyse millions of piec- require a predictable response
CODE – SOFTWARE – FOR A GIVEN BOARD THAT es of data, the raw performance time to an interrupt), ARM’s mbed
CONFORMS TO A GIVEN OPERATING SYSTEM. from x86 often outstrips that of an takes what the company calls an
ARM equivalent. If we want high- ‘event based’ approach; the pro-
IT IS COMMONLY BUILT WITH A BOOTLOADER. IN end devices to perform telemetry cessor wakes up only if events
EMBEDDED SYSTEMS BSP IS SYSTEM AND DEVICE and data transmission over the need to be managed. ARM says
DRIVERS FOR ALL THE DEVICES ON THE BOARD. Internet for critical applications, that predictable response is not
Source: Embedded Logic for instance, x86 would still likely always needed and allowing the E
‘‘QSEVEN
ded applications ranging from speed interfaces like PCI Express
health care to advanced robotics and HDMI. SUPPORTS
ARM, INTEL AND
AMD PROCESSORS
’’
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.
System IP
By ASHOK RAO / SENIOR SUPPORT ENGINEER, FARNELL ELEMENT14
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 cessor performance through QoS tion of memory controllers. For to increase performance.
Shared off-chip memory access mechanisms. These are usually any enterprise level SoC, as the CoreLink 400 is the little broth-
plays a crucial part in influenc- coupled with the CoreLink 400 number of components increases, er of CoreLink 520 that can be
ing system performance, power series of interconnects as dis- so does the demand for data. To used in conjunction with the 400
and cost. CoreLink DMC (Dynamic cussed above. satisfy this need for fast data, series of interconnects as dis-
Memory Controllers) deliver the The CoreLink 500 series of efficient access to the DRAM cussed previously. There is also
best-in-class performance and DMCs implements similar fea- is required and is critical to the an older generation of DMCs, the
power for AMBA based SoCs. tures as the 400 series but is performance of the SoC. The 32x series that was optimised for
The DMCs are further subdivided optimised for enterprise level CoreLink 520 has advanced QoS DDR, DDR2, SDR and other such
into the smaller 400 series and solutions providing additional based scheduling and arbitration older generation of memory tech-
the beefier 500 series. The 400 features such as TrustZone and algorithms that handle read/write nologies. These are still in use but
series significantly enhances end to end QoS. The CoreLink access to the memory. A multi their popularity has considerably
memory bus utilisation and pro- DMC 520 is ARM’s 5th genera- level parallelism is also adopted reduced over the years. E
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 on the SoC. occasionally causes peaks leading and functional debug at the hard-
Cortex-A series of processors the ■■ For product engineers, it can be to loss of trace data. Whereas the ware level.
debug can be performed with only useful in failure analysis. TMC uses a FIFO model enabling What is the advantage of hav-
two pins using the Serial wire tech- The CoreSight technology has trace over a longer period of time ing all of these macro cells? Well,
nology or alternatively using JTAG. also been applied not only to thus reducing risk of overflows. conventional debugging requires
Some of the other advantages of the Cortex-A series but also the This in turn also allows for a small- breakpoints/watch points to halt the
using CoreSight are listed below: Cortex-R and Cortex-M series of er trace port reducing the cost of processor and use a debug con-
■■ Using the CoreSight DAP and processors as well. On the Cortex-R implementation of a trace port on nection from there to read, examine
embedded cross triggering, it series, CoreSight provides non- the SoC. and modify register contents. This
is possible to perform multicore intrusive trace that helps in under- The Serial Wire Debug is a widely method is intrusive, requires the
debug and run time control while standing and optimising how real known debug technology used processor to stop, is non real-time
also collecting time stamped CPU time software operates on the tar- mostly as an alternate to JTAG. It and does not give an indication of
or system trace to analyse, opti- get. This is achieved using a set of needs just two pins and is avail- true performance.
mise and debug software and Trace macrocells that are discussed able as part of the CoreSight DAP Whereas, using trace capture,
hardware interaction. in the later part of this article. On (Debug Access Port). developers can capture real-time
■■ CoreSight trace macrocells enable the Cortex-M series, CoreSight data including instruction level
developers to examine how the provides a low cost debug and Trace macrocells data, compress it and deliver it
software behaves on the tar- trace functionality using a whole Trace macrocells provide non-intru- either on-chip or off-chip. This
get platform. gamut of macrocells, breakpoints sive visibility of the components in leads to real-time data analysis,
■■ CoreSight also provides provisions and watchpoints. the SoC. These are individual blocks is non-intrusive and does not
for secure debug and trace. It also that help capture trace data based halt the processor and is much
allows designers to implement a CoreSight components on their particular function. There more accurate for performance
debug and trace solution with very The CoreSight components along are four different types of trace analysis. The various trace mac-
low power consumption. with the trace macrocells pro- macrocells viz. Embedded Trace rocells mentioned above are
■■ CoreSight is highly re-usable vide the necessary infrastructure Macrocell (ETM), Program Trace all essential components of the
enabling SoC developers reduce required for effective development Macrocell (PTM), Instrumentation entire debug and trace ecosys-
design and development time. and debugging. Two key compo- trace Macrocell (ITM) and AHB tem for a SoC.
Some of the real world examples nents comprise the CoreSight: the Trace Macrocell (HTM) (AHB: A latest development in the
where CoreSight proves advanta- trace memory controller and the Advanced Hardware Bus). The series of macro cells is the STM
geous are listed below: debug access port (SWD/JTAG). usage of each of these is briefly (System Trace Macrocell) that
■■ For embedded software/middle- The TMC (Trace Memory Controller) outlined below: offers developers printf (the
ware developers, device driver provides low cost real time trace ■■ ETM: used for non-intrusive, cycle well-known C language output
development can be faster by all along the development cycle accurate program and data trace statement) style debugging giv-
observing the CPU trace while (including production) for third party for ARM processors. This occurs ing visibility of the core from the
executing software in real time. software developers giving them at the assembly instruction level. application layer itself.
Cache controllers fer modes such as Scatter-Gather, cious software attacks? Answer: controller, PS/2 interface, GPIO con-
Cache controllers play a critical part Ping-Pong and Primary-Alternate. Very important. ARM provides trollers to name a few. Detailed dis-
in CPU to on-chip data access. They TrustZone technology to prevent cussion of these is beyond the scope
reduce the latency of off-chip data Interrupt controllers access to selected memory regions of this article.
access by about 10-25%. This reduc- As the complexity increases with and peripherals such as screens
tion in CPU demands for off-chip data multicore CPUs and their handling an and keypads. TrustZone extends the Where does all this fit?
means it can now be used for other ever increasing number of interrupts security to both on-chip and off-chip SoC design is a very complex task
tasks. Lower demand for off-chip data from peripherals, so does the require- memories. The technology works by and various factors need to be con-
also reduces power consumption. ment for interrupt controllers. ARM partitioning the available memory sidered to achieve the required per-
Thus cache controllers contribute to provides many GICs (Generic Interrupt into ‘N’ number of secure regions, formance. This even applies to the
power efficiency and performance. 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
DMA controllers supports the ARMv8 architecture. It ROM provides 8-16KB for signature the chip though micro/nano meters
DMA control plays an important part can handle up to 128 CPUs and 480 code check. It also provides on-chip long. At this level of required accu-
in reducing the data transfer load on shared peripheral interrupts. The GIC- memory for secure data and on-chip racy, every single block matters to
the CPU. It takes the task of mov- 400 is the younger sibling that can RAM for secure code and data. For achieve maximum performance.
ing in and out large blocks of data handle up to 8 CPUs and 480 inter- off-chip security, it provides options Nobody likes to wait forever for
from peripherals to memory and/or rupts. The GIC-390 is based on vector to partition the available RAM into software to load or a video to start
from memory to memory. This data addressing and stores the CPUs vec- up to 16 secure regions making playing. DMA controllers, cache
movement activity by a separate unit tor addresses in memory. The PL-190 it immune to software attacks. All controllers, interrupt controllers
greatly reduces the load on the CPU. can handle up to 32 interrupt sources this is controlled by the TrustZone etc, all play a vital part in bring-
ARM provides two DMA controllers: and 16 vectored addresses while the Address Space Controllers TZC-380 ing everything together to help SoC
DMA-330 and DMA-230. The DMA- PL-192 supports up to 32 vectored and TZC-400. designers build the ultimate chip
330 is a highly flexible, configur- interrupts with programmable priority for any dedicated task. Software
able DMAC and is mainly targeted at levels and masking. CoreLink peripherals and application engineers then use
higher end SoCs. The DMA-230 is a A number of APB (Advanced these chips to generate software
simpler version of the DMA-330 and TrustZone controllers Peripheral Bus) are provided from that efficiently utilises all the avail-
is ideal for use with Cortex-M series Question: How important is security ARM to ease building complex SoCs. able resources on the chip to create
designs. Both of these support trans- for your code and stopping mali- Some of these are the UART, SPI immersive and useful applications.
CONCLUSION channelled properly to the outside world for developers to optimise their
designs and software. All the peripherals discussed in this article are the
The performance of an ARM based processor is not the sole responsibility skeletal connecting blocks, in other words the threads that hold the fabric
of only the core. There are a number of peripherals acting around of the processor together and provide the required throughput. In the next
the core that influence the performance and efficiency of any ARM part of this series, we will discuss some of the other technologies from
based processor. There will also be a number of factors affecting the ARM such as security on ARM based devices and a few other software
communication speed between the core and peripherals which need tools required for an effective design and development of an ARM based
to be optimised. All this processing power and speed also needs to be embedded system. Until the next edition of element14 Tech Journal... n
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
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.
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.
Get yours today! For more information on ARM embedded products and resources please visit
www.element14.com/ARMPromo