www.newelectronics.co.

uk
www.newelectronics.co.uk

0828May
June2018
2016

STORAGE
SPACE
ROVERS ELECTRONICS
TECHNOLOGY
EXCEED • SAFETY
EXPECTATIONS ••REDEFINING
CONTRACT
& SECURITY
MANUFACTURING
• ENCLOSURES
HARDWARE IN REAL TIME ••CATCHING
EMBEDDED
• EMC SHIELDING
DESIGN
UP WITH CLOCKS

THE REAL DEAL?

Are augmented and virtual reality

SO WHERE
finally set to become the next big
computing platforms?
ARE THE
WOMEN?
Why is it that so few
women consider
engineering to be
a viable career
option?
 New Electronics’ weekly eZine features the
latest blogs, news, articles and more.
To register for your copy, go to
www.newelectronics.co.uk CONTENTS VOL 51 NO 9

18 23

14 20 29
Cover Illustration: Elly Walton

COMMENT 5 INTERVIEW 12 EMBEDDED DESIGN 23

The UK Government confirms A strategy bearing fruit Measuring battery life
plans for a British Global January saw Victor Peng replace Moshe Poor battery life is impacting too many devices,
Navigation Satellite System Gavrielov as president and CEO of Xilinx. Here, so how can power be analysed in enough detail
are under consideration he talks strategy with Neil Tyler to ensure they meet consumer expectations

NEWS COVER STORY 14 OPERATING SYSTEMS 26

Ekioh launches a multithread So where are all the women? The right approach
embedded browser designed Why are so few women considering engineering Choosing between Embedded Linux or a
specifically for multi-core as a viable career and what can be done to proprietary real-time operating system requires
processors 7 address it? Bethan Grylls investigates many considerations to be taken into account

Southampton University is STORAGE TECHNOLOGY 18 SYSTEM DESIGN 29

awarded multi-million pound When flash failure is not an option The cost of learning
programme to develop The tension between cost and reliability is How are embedded system designers
innovative nanotech 8 driving the development of flash-memory addressing the immense complexity associated
controllers that can support critical embedded with machine learning? By Chris Edwards
Silicon Labs acquires applications
mesh technology boosting
its presence in the IoT smart CONTRACT MANUFACTURING 20
home market 9 How ‘smart’ is the smart factory?
How is the move to digitisation changing the
UK companies may be face of electronics manufacturing? Frank
MISSION STATEMENT
shut out of the Galileo Zycinski looks at its impact on the factory ‘New Electronics keeps designers and
programme, raising serious managers abreast of the latest developments
concerns for the space in the world’s fastest moving industry’
industry 10

www.newelectronics.co.uk 08 May 2018 3

 COMMENT INDEPENDENT
SATELLITE SYSTEM

Keeping your options open

THE UK GOVERNMENT HAS CONFIRMED PLANS FOR A BRITISH GLOBAL
NAVIGATION SATELLITE SYSTEM ARE UNDER CONSIDERATION

T
he UK Government has confirmed that the UK Space Agency will be leading
work on options for the development of an independent British Global
Navigation Satellite System.
Apparently, a taskforce of Government specialists and industry will work to
develop options that will provide both civilian and encrypted signals and be compatible
with the GPS system.
The move comes as the UK and the EU finds themselves in an escalating row over
UK participation in the Galileo satellite programme, the EU’s Global Navigation Satellite
System (GNSS).
Although developed by the non-EU European Space Agency (ESA), it is funded and
owned by the EU and supervised by the European Commission, and only member states
will have an automatic right to the encrypted elements of the system.
While the UK will be able to use Galileo’s open signal, its armed forces and
emergency services were expecting to have access to the encrypted system when it “Many UK
becomes fully operational.
companies,
According to the Government, if the UK is unable to collaborate on an equal basis,
which means having access to security-related information, it may end its participation as well as
in the project – hence the decision to consider its own GPS system. the European
Many UK companies, as well as the ESA, are preparing for a ‘no deal’ scenario and Space
the likely ending of UK participation – all of this will hang on the broader negotiations
Agency, are
around Brexit and the kind of future relationship that is established.
The UK has a thriving space industry and certainly has the skills and capacity to
preparing for
develop its own programme, but any project will be extremely costly to undertake. a ‘no deal’
Reaching an agreement with the EU on continued participation would surely be more scenario and
convenient and a failure to establish a privileged position could see the UK pushed into the likely
closer collaboration with the US.
ending of UK
But while the UK would benefit from a deal, so would the EU, because if the EU
was to lose access to the UK and its companies’ expertise in this field, the costs of the
participation”
project could increase substantially and delay Galileo’s roll-out well beyond 2020.
The growing row between the UK and EU over Galileo highlights the complexity
associated with Brexit.
Surely, it’s not beyond the two sides to agree a security of information agreement,
because if no deal is the outcome of negotiations, then we are going to be spending
billions replicating Galileo so we can have the secure access to satellites we
already have!

Neil Tyler, Editor (neil.tyler@markallengroup.com)

www.newelectronics.co.uk 8 May 2018 5

 SMARC Computer on Modules
SMARC is the first COM standard to be built specifically for
modern ARM-Cortex system on chips (SoCs), aiming to
efficiently pass along ARM benefits such as low power
consumption to COTS designs.
SMARC-FiMX7
The new standard for ultra low-power COMs was developed
specifically for new modules with ARM and SoC processors and
is characterized by the extremely flat build of its form factor.
It uses a 314-pin connector that has a construction height of just
4.3 millimeters (the MXM 3.0) with an optimized ARM/SoC pin-out
definition. This connection method allows robust and cost-
effective designs that have an extremely thin construction height
SMARC-FiMX6
There are 4 modules available:
SMARC-FiMX7
NXP i.MX7 Cortex-A7 Solo 800Mhz and Dual Core 1GHz
1GB (Dual) or 512MB (Solo)
SMARC-FiMX6 SMARC-T335X
Freescale i.MX6 Cortex-A9 Solo, Dual-Lite, Dual and Quad
SMARC-T335X
TI Sitara AM3354 @600MHz, 800MHz or 1GHz
SMARC – T4370
Ti Sitara AM4378 @ 800mhz or 1Ghz SMARC – T4370

6 8 May 2018 www.newelectronics.co.uk

 NEWS MULTITHREAD BROWSER

Editor Neil Tyler

Go with the Flow
EKIOH LAUNCHES MULTITHREADED EMBEDDED BROWSER. BETHAN GRYLLS REPORTS
neil.tyler@markallengroup.com

Deputy Bethan Grylls A multithreaded HTML browser specifically designed for multi-core processors, known as ‘Flow’, has
Editor bethan.grylls@markallengroup.com been developed by Cambridge software company, Ekioh.
Contributing Chris Edwards Traditional browsers are single-threaded, meaning that they are only capable of using one of the
Editors John Walko four cores within a multi-core processor. Ekioh has created a solution that utilises all four cores to
editor@newelectronics.co.uk increase performance speed and user experience.
Art Editor Andrew Ganassin “Ekioh had to go back to the drawing board and create a solution completely from scratch,”
andrew.ganassin@markallengroup.com explained Stephen Reeder, Commercial Director.
Illustrator Phil Holmes To achieve this, it broke the layout tasks – that a browser would usually process all at once –
into sections and processed them on individual threads, meaning that all four cores could be used.
Sales Manager James Creber
james.creber@markallengroup.com
As a result, Flow’s layout and animation performance is more than double the speed of other
browsers on multi-core silicon.
Publisher Peter Ring
peter.ring@markallengroup.com Moreover, rather than painting on the CPU, Flow handles all rasterisation on the GPU. The result
is a “significantly faster” product and even more CPU power available for tasks such as layout.
Managing Jon Benson
Director jon.benson@markallengroup.com
Together, with multithreaded layout, these developments provide “dramatically improved” HTML
performance without the need for content re-authoring.
Production Nicki McKenna “Our architecture is designed to work alongside the evolution of multi-core silicon,” Reeder
Manager nicki.mckenna@markallengroup.com
concludes, “so as the chips get better, so does our technology.”
New Electronics editorial advisory panel
Trevor Cross, chief technology officer, Teledyne e2v
Pete Leonard, electronics design manager, Renishaw
Pete Lomas, director of engineering, Norcott Technologies
3D printer can place electronics directly on skin
Neil Riddiford, principal electronics engineer, Cambridge A customised, low-cost 3D printer that One of the key innovations of this
Consultants can print electronics on to skin has been technique is that the printer can adjust to
Adam Taylor, embedded systems consultant developed by researchers at the University of small movements of the body during printing.
ISSN 0047-9624 Online ISSN 2049-2316 Minnesota. Temporary markers are placed on the skin,
Annual subscription (22 issues): Associate Professor Michael McAlpine which is scanned, the printer then uses
UK £108. Overseas; £163. Airmail; £199. of the University of Minnesota believes the computer vision to adjust to movements in
New Electronics, incorporating Electronic Equipment News
and Electronics News, is published twice monthly by
technology could be used by soldiers on the real-time. The printer also uses a specialised
MA Business, Hawley Mill, Hawley Road, Dartford, DA2 7TJ. battlefield to, for example, print temporary ink made of silver flakes that can cure and
T: 01322 221144 E: ne@markallengroup.com sensors on their bodies to detect chemical conduct at room temperature.
Moving on? or biological agents, or solar cells to charge
If you change jobs or your company moves, please contact
circulation@markallengroup to continue receiving your essential electronics.
free copy of New Electronics The researchers also managed to
successfully print biological cells on the
skin wound of a mouse, and said that this
technique could lead to medical treatments
for wound healing and the direct printing of
grafts for skin disorders.

Next generation electronics

www.markallengroup.com
© 2018. All rights reserved. No part of New Electronics may
be reproduced or transmitted in any form, by any means,
electronic or mechanical, including photocopying, recording or
any information storage or retrieval system, without permission
in writing from the Publisher.The views expressed do not
necessarily represent those of the editor of New Electronics.
Advertisements in the journal do not imply endorsement of the
products or services advertised.
Origination: CCM, London. Printed by Pensord.
Prototype devices made of zirconium tritelluride (ZrTe3) that can conduct a current density 50
times greater than conventional copper interconnect technology have been demonstrated by
engineers at the University of California (UC).
As the demand for smaller components increases, so does the search for a material that can
sustain a high current density.
Engineers at the university discovered that ZrTe3 nanoribbons delivered this capability to a level
that far exceeds any conventional (polycrystalline) metals like copper, pushing research from 2D to
1D materials.
“Polycrystallines have grain boundaries and surface roughness, which scatter electrons,”
explained Professor Alexander A. Balandin of UC. “Quasi-1D materials, such as ZrTe3, comprise
single-crystal atomic chains in one direction. They don’t have grain boundaries and often have
atomically smooth surfaces after exfoliation. We attributed the high current density in ZrTe3 to the
single-crystal nature of quasi-1D materials.”

www.newelectronics.co.uk 8 May 2018 7

 NEWS
UK invests £1bn in AI
The UK Government and industry plan to
invest £1billion in artificial intelligence (AI),
with the aim of putting the UK at the forefront
of the AI industry.
The arrangement follows the
announcement of the Government’s
Industrial Strategy in which AI was outlined
as one of the UK’s four ‘grand challenges’,
and includes almost £300million of private
sector investment.
According to Business and Energy
Secretary, Greg Clark, “This deal with
industry will ensure we have the right
investment, infrastructure and highly-skilled
workforce to establish the UK as a driving
force in the development and commercial
use of artificial intelligence technologies.”
Investments will include Japanese
venture capital firm, Global Brain, opening
its first European headquarters and investing
£35m in UK tech start-ups; a £10m AI
supercomputer from the University of
Cambridge; and Vancouver-based capital
firm, Chrysalix, opening a European
headquarters and investing £110m in AI and
robotics.
The deal will also see the Alan Turing
Institute and Rolls-Royce join forces with
several research projects exploring data
science and AI; as well as an investment into
training 8,000 specialist computer science
teachers, and 1,000 AI PhDs by 2025.
The government also intends for £20m to
fund the UK’s services industries in a hope
to identify how AI can enhance operations in
other sectors, including law and insurance.
High growth into regional tech hubs
across the country will also come from a
£21m investment into what the Government
is calling, ‘Tech Nation’, which hopes to
create a tech network for entrepreneurs.
8 8 May 2018 www.newelectronics.co.uk
NANOTECHNOLOGY
University awarded
nanotechnology programme
UNIVERSITY OF SOUTHAMPTON IS SET TO LEAD RESEARCH INTO INNOVATIVE
NANOTECHNOLOGY. NEIL TYLER REPORTS
The University of Southampton has been awarded a multi-million pound programme to lead the
development of innovative nanotechnology that could open the door to a new generation of
electronics.
Professor Themis Prodromakis from the University of Southampton is the principal
investigator of the predominantly Engineering and Physical Sciences Research Council (EPSRC)
funded programme, which, along with industrial contributions, exceeds £11million.
Working with Imperial College London and the University of Manchester, as well as industrial
partners, the project will centre on memristors and their ability to enable electronics systems to
be configured with increased capability, as opposed to transistors.
Commenting, Prof. Themis Prodromakis said: “Memristor technologies bring great prospects
for next-generation chips, which need to be highly reconfigurable yet affordable, scalable and
energy-efficient, not to mention secure.”
The project will involve assembling some of the UK’s best academics and industrialists to
develop the core technology as well as the required tools for demonstrating the benefits of the
technology in real-working services and products.
While the size of transistors has reduced to meet the increasing demands of technology they
are now reaching their physical limit, and memristors are seen as possibly holding the key to a
new era in electronics, being both smaller and simpler in form than transistors. Their ability to
retain data by ‘remembering’ the amount of charge that has passed through them could result
in computers that switch on and off instantly and never forget.
The University of Southampton has previously demonstrated a new memristor technology
that can store up to 128 discernible memory states per switch, almost four times more than
previously reported.
Prof. Themis Prodromakis added: “This unique programme of activities will allow us to
develop reconfigurable electronic systems that are at the forefront of innovation through being
embedded almost everywhere in our physical world; within vehicles and infrastructure, or even
within the human body.”
Intel delays 10nm TI retains top spot
Intel has reported that its 10nm process will According to IC Insights, Texas Instruments
start volume production next year. Despite (TI) retained top spot in the analogue
the delay, which means that four generations market with sales of $9.9billion and an 18%
of new product have been made in 14nm, marketshare in 2017.
the company’s CEO Brian Kzranich said The company’s analogue sales increased
that Intel had been able to increase 14nm by $1.4bn, rising by 16% from 2016’s figures,
process performance by 70% since it was first and were more than twice that of second-
introduced in 2014. ranked ADI. TI’s 2017 revenue represented
While the company said that it was 71% of its $13.9bn total semiconductor
currently shipping ‘some’ 10nm product, revenue, based on Insights’ estimates.
Kzranich admitted that the 10nm shrink – The top 10 suppliers held 59% of the
implementing a 2.7x density increase over the worldwide analogue market last year, with ON
14nm process – had been too ambitious. Semiconductor showing the strongest growth,
The industry average increase is 1.5-2x, with revenues increasing 35% to $1.8bn,
according to Kzranich, and Intel will look representing a 3% share of the market.
to increase density 2.4x in the 10nm-7nm The leading ten suppliers generated
transition. $32.3bn in analogue IC sales in 2017,
Despite these problems, Intel reported a compared to $28.4bn in 2016, with 8 of the
strong first quarter with revenues of $16.8 10 suppliers exceeding the 10% growth rate of
billion up 16% year-on-year. the total analogue market.
 NEWS Z-WAVE

Silicon Labs completes acquisition Materials for sustainable,

low-cost batteries
of Z-Wave business Scientists from ETH Zurich and Empa, led by
Maksym Kovalenko, Professor of Functional
MESH TECHNOLOGY WILL HELP TO STRENGTHEN SILICON LABS’ POSITION IN THE IOT
Inorganic Materials, have identified two new
HOME MARKET. NEIL TYLER REPORTS
materials that could bring about key advances
Silicon Labs has acquired Sigma Designs’ Z-Wave business, including a team of approximately in the development of aluminium batteries.
100 employees, for $240 million in an all-cash transaction. The first involves titanium nitride, a
Z-Wave is a mesh networking technology for the smart home and over 2,400 certified, ceramic material that exhibits sufficiently high
interoperable Z-Wave devices are currently available from the Z-Wave Alliance, which comprises conductivity. “This compound is made up of the
of more than 700 manufacturers and service providers worldwide. highly abundant elements titanium and nitrogen,
The acquisition complements Silicon Labs’ existing wireless hardware and software portfolio and it’s easy to manufacture,” explains
for the smart home, which includes Wi-Fi, Zigbee, Thread, Bluetooth and proprietary protocols, Kovalenko.
providing smart home developers with access to a varied network of ecosystems and to a full Scientists have successfully made
range of end-node technology options. aluminium batteries with conductive parts made
“Adding Z-Wave to Silicon Labs’ IoT connectivity portfolio will allow us to deliver a unified of titanium nitride in the laboratory – it can be
vision for the wireless technologies underpinning the smart home market,” said Tyson Tuttle, easily produced in the form of thin films and as
CEO of Silicon Labs. a coating over other materials.
“Our vision for the smart home is one in which multiple technologies will be able to work According to Kovalenko it should be
securely together,” said Daniel Cooley Senior Vice President and General Manager, IoT possible to manufacture the conductors from a
Products, Silicon Labs. “This acquisition will enable us to offer a wider range of connectivity conventional metal and coat them with titanium
options, all of which will easily be able to join a home network. It will also enable security nitride, or even to print conductive titanium
updates and feature upgrades to occur automatically.” nitride tracks on to plastic.
The second new material, polyprene, is used
for the positive electrode (pole) of aluminium
New galvanic isolation technology batteries. Whereas the negative electrode in
these batteries is made of aluminium, the
X-FAB Silicon Foundries, a specialist couplers for a wide range of applications. positive electrode is usually made of graphite.
analogue/mixed-signal and specialty Galvanic isolation electrically separates Polypyrene, is a hydrocarbon with a chain-like
foundry, has developed a galvanic isolation two circuits in order to improve noise (polymeric) molecular structure.
process technology that enables the immunity, remove ground loops, increase “A lot of space remains between the
fabrication of robust and reliable high common mode voltage and safely isolate molecular chains allowing relatively large ions of
voltage signal isolation solutions. high voltages. Optocoupler technology has the electrolyte fluid to penetrate and charge the
According to the company, the traditionally been used to provide single- electrode material easily,” Kovalenko explained.
process achieves best-in-class isolation package signal isolation between two
performance. Customers will be able to
design their own capacitive or inductive
circuits in a system. For high temperature,
high speed, high reliability or multi-channel
Graphene specialist
systems, however, chip-level closes seeding round
galvanic isolation based Paragraf, a graphene technology spin out
on capacitive or inductive from the University of Cambridge, has closed
coupling is more appropriate a £2.9 million seed round to support the
and cost-efficient. development of its first major products.
Both packaged single- Paragraf is focused on the production
channel inductive and of ‘two-dimensional’ materials, primarily
capacitive coupler samples graphene, and the development of electronic
are available. The capacitive devices based on these materials. The
coupler test chip, designated company’s technology is the first ever
G3-C1, with an isolation layer commercial-scale method validated to
thickness of 11µm is able to reproducibly deliver functionally active
withstand up to 6,000Vrms, graphene with properties targeted to its final
the maximum limit of the device-specific application.
test setup. Isolation layers The company has already produced layers
thicker than the test chip’s with electrical characteristics optimised
11µm are available if a for producing very sensitive detectors at
higher withstand voltage is commercial scale and improved efficiency
required. contact layers for common technologies such
as LEDs.

www.newelectronics.co.uk 8 May 2018 9

 NEWS ANALYSIS UK SPACE INDUSTRY
Worrying times for the UK’s space industry
NEWS THAT UK COMPANIES MIGHT BE SHUT OUT OF THE GALILEO PROGRAMME RAISES CONCERNS FOR COMPANIES
WORKING IN THE COUNTRY’S SPACE INDUSTRY. NEIL TYLER REPORTS
W hat next for the UK’s space industry?
It’s been a busy few weeks for the
industry what with news that the UK could
be shut out of key elements of the EU’s
Galileo satellite-navigation system on security
grounds.
In response Greg Clarke, the UK’s
Business Secretary, is not only seeking
legal advice on reclaiming money spent on
Galileo but it’s also been reported that he
is considering setting up a rival satellite
navigation system.
Last month the House of Lords’ EU
Internal Market Sub-Committee released a
letter that had been sent to Sam Gyimah,
Minister of State for Universities, Science,
Research and Innovation in which it called
for greater urgency and clarity in terms of on-
going Brexit negotiations. The letter warned
of the impact the UK’s withdrawal could have
with many UK firms losing valuable access to
EU space programmes.
The UK’s space industry is a major
success with growth averaging over 8 per
cent, year-on-year, since 2000. The sector
now employs upwards of 40,000 people.
Government has seen space activity as
an area in which ‘UK plc’ could excel so
the confusion generated by Brexit and the
possibility of the UK being cut adrift has
raised real concerns. Could it spur a flight
of high-tech space companies from the UK
undermining the government’s ambition to
take 10 per cent of the £400bn a year global
space market?
The UK has a long relationship with
10
the Galileo programme and, at present,
plays a central role in its development and
operation. Based in Guildford, Surrey Satellite
Technology Ltd (SSTL) was responsible for
building the first Galileo spacecraft and has
gone on, in partnership, to build all 22 of the
current Galileo spacecraft.
The row over the Galileo project comes
down to attempts to restrict Britain’s access
to sensitive security information after Brexit
and the European Commission is demanding
the right to cancel existing contracts, without
penalty, if a supplier is no longer based in
an EU member state. More worryingly for UK
companies already working on the project, the
Commission may also require any supplier
ejected from the programme to repay all
costs to the EU associated with finding a
replacement.
Commenting Richard Peckham, head of
UKspace, the trade association, was quoted
in the Financial Times as having said: “Since
the UK government has so far failed to make
any clear statement of intent or even a
wish to remain in these important EU space
programmes, it is not surprising that the EU
is cautious about UK industry participation.”
To date the UK has spent £1.2bn on
Galileo, Europe’s answer to the US GPS
system, hence Clark’s decision to take legal
advice on whether the UK can claim that cash
back.
A recent decision to move the back-up site
of the Galileo Security Monitoring Centre from
the UK to Spain suggests that the concerns
being voiced are real.
The UK is seen as a highly-valued and
effective member of the European Space
Agency (ESA) and while the ESA is not an
EU agency, some of the ESA’s optional
programmes are delivered in partnership with
the EU.
The UK has outstanding experience,
expertise and heritage in space and if
it is unable to participate in EU space
programmes post 2019, there are worries
that not only will it impact the UK’s
burgeoning space industry but that it could
have a negative impact on Europe’s as well.
The House of Lords’ letter to the
government urged greater clarity on the
part of the government in terms of its
Brexit negotiations and warned of the
impact continued uncertainty was having
on the UK’s space industry. It urged the
Government to confirm its position on the
UK’s relationship with current and future
EU space programmes, including those with
security implications, and went on to call for
a timetable in which an agreement could be
reached.
While the government’s planned
investment to enable new satellite launch
services and low gravity spaceflights from
UK spaceports, as detailed in the Space
Act 2018, are welcomed, they are seen as
being insufficient to compensate for a loss
of participation in these valuable EU space
programmes.
While there was always going to be a price
to pay for Brexit, is the space industry set to
be one of the first to suffer?
8 May 2018 www.newelectronics.co.uk
 INTERVIEW VICTOR PENG
A strategy
bearing fruit
Moshe Gavrielov’s decision to leave Xilinx saw
Victor Peng appointed as the company’s new
president and CEO. Here he talks to Neil Tyler
about his plans for the company
B
ack in January Moshe Gavrielov announced that he was leaving
Xilinx. He was replaced as president and CEO by Victor Peng,
who had been the company’s chief operating officer.
“Gavrielov’s decision to step down brought to end a
remarkable career, spanning 40 years in semiconductor and software
related companies,” said Peng.
Peng’s appointment, after ten years with the company, comes as
Xilinx, which invented the FPGA, hardware programmable SoCs and
now the ACAP platform, looks to embrace a new vision and strategy.
According to Peng, it will bring “new growth, technology and direction,”
to the company.
Crucially, Peng sees the company moving beyond the traditional
world of the FPGA to one that addresses and enables a “more
adaptable and intelligent world,” where Xilinx will deliver, “a new
category of highly flexible and adaptive processors and platforms.
“We want to be able to support rapid innovation across a wider array
of technologies,” Peng explained,” from the end-point to the edge.”
According to Peng three trends are affecting the market at present.
“The first is the explosion in data which is driving demand for
a different type of computing; the second is the advent of artificial
intelligence which will disrupt multiple industries.
“The third is exponential change. There’s currently no single
architecture capable of addressing the complexity that comes with that.
In truth the economics of Moore’s Law no longer work, and they haven’t
for some time.”
But, despite these trends, Peng believes that for those looking for
them, plenty of opportunities exist.
“Things are changing so dynamically that you need a vision and
strategy to manage this intelligent connected world.”
How you formulate and then deliver that strategy is crucial.
“Everything we do has to be focused on delivering a strategic plan,”
explains Peng. “The transformation of any business isn’t going to
happen overnight but we have been addressing these fundamental
challenges for some time and I’m lucky to be building on the strong
12
foundations that were left by my predecessor.
“There is a degree of continuation, in terms of our core business
where we have some real strengths, but going forward we see the
biggest opportunities coming from developments in data centres.”
According to Peng, ”When we talk about strategy, I’m talking about
a five year time line in which we will be focusing on delivering our plans
for the business.”
Strategic plan
Peng describes the company’s strategy as comprising of three core
elements.
“The first, and most important is, as previously mentioned, a new
emphasis on data centre acceleration,” Peng explains.
“At Xilinx we are looking to ramp up our efforts with key data
centre customers, ecosystem partners and software application
developers. We want to enable innovation and deployments in compute
acceleration, computational storage and network acceleration.”
Peng points to the data centre market as one in which there is rapid
growth. Customers are looking to adopt new technologies quickly as
they look to take advantage of the orders of magnitude performance
and performance per-watt improvements that will be required to enable
applications like artificial intelligence (AI) inference, video and image
processing, and genomics to be taken up.
“While we look to transform the business, there will be a degree of
continuity going forward,” Peng suggests. ”Xilinx has many thousands
of loyal customers and while we are looking to exploit new market
opportunities, and I believe that we have reached a significant inflection
point, we can’t ignore their needs either. These core markets and
customers remain central to Xilinx and we will continue to enable
innovation in these areas.”
That is the second essential pillar of the company’s strategy going
forward - accelerating growth in its existing core markets.
“These markets are areas where we have been a key technology
leader and have developed deep market traction,” Peng explains.
“Traditionally we have served multiple markets, primarily wireless
communications and infrastructure but also audio, video and broadcast;
aerospace and defence: industrial, scientific and medical and consumer
technologies.”
In time, according to Peng, “We want to make the data centre
market our number one segment and much of our focus will be on
delivering that ambition.”
Xilinx isn’t alone in targeting the data centre market. Intel PSG, for
example, now describes itself as a datacentric business and talks of
broadening its portfolio of programmable logic solutions across its
silicon, software, intellectual property and hardware platforms.
“If Intel has ‘pivoted’ to become a datacentric business so too has
Xilinx,” argues Peng. “The world is changing and data is taking centre
stage.”
Peng, however, believes that Xilinx is better positioned to address
these market changes than most of its competitors.
“We are in the end point, whether that’s in drones, cameras or
vehicles and we have developed a broad range of cost optimised, low
8 May 2018 www.newelectronics.co.uk
 Victor Peng
Victor Peng is President and Chief
Executive Officer of Xilinx. He
has over 30 years of experience
defining and bringing to market
leadership FPGAs, Programmable
SoCs, GPUs, high performance
microprocessors and chip sets,
and microprocessor IP products.
In his previous role as COO
he managed Global Sales,
Product Development and Global
Operations & Quality. He also
served as Executive Vice President
& General Manager of Products
at Xilinx, and led the definition,
development, and product
marketing of the company’s
portfolio of technologies.
Prior to joining Xilinx, Peng
held executive and engineering
leadership roles at AMD, TZero
Technologies, MIPS Technologies,
SGI and Digital Equipment Corp.

power devices such as Spartan or our Zynq SoCs. Not only that we
are well placed in communications infrastructure and to address the
acceleration that we are seeing into the Cloud.”
The third element of the company’s strategy has been the
introduction of the Adaptive Compute Acceleration Platform (ACAP).
“I would argue that this is the most significant pillar in terms of
our strategy and this new product can honestly be described as a
‘breakthrough’,” Peng contends.
ACAP is a highly integrated multi-core heterogeneous compute
platform that can be changed at the hardware level to adapt to the
needs of a wide range of applications and workloads.
“That adaptability, which can be done dynamically in milliseconds
during operation, delivers levels of performance and performance per-
watt that is unmatched by CPUs or GPUs, and is crucial,” Peng says and
it means that, “While FPGA and Zynq SoC technologies are still core to
our business, Xilinx is no longer just an FPGA company.”
The company’s strategy appears to be paying dividends.
Announced last month, the company has been able to post record
revenues of $2.54 billion for the fiscal year 2018, up 8% from this time
last year.
“Our focus on data centres; meeting the demands of new and
emerging applications and the introduction of the ACAP platform seems
to suggest that our three-pronged approach is starting to work,” Peng
concludes.

www.newelectronics.co.uk 8 May 2018 13

W
hile there has been an
increase in the number of
engineering students, there
is an estimated shortfall of 59,000
graduates each year in the UK,
according to Engineering UK, which
works with the engineering community
to promote the role of engineers and
the industry as a whole.
Building a future pipeline of talent
is proving difficult to deliver. In fact,
the Engineering UK’s Brand Monitor
(EBM) survey found that while more
than 59% of 11-14 year olds would
consider a career in engineering, by
the age of 19 that number has fallen

So where
by 20%.
Could this fall be the result of
schools failing to sustain young

are all the

people’s interest – and particularly,
female interest – in developing a
career in engineering?
The EBM found that just 34% of

women?
7-11-year-old females expressed an
interest in engineering, compared
to 59% of boys and this falls further
among 16-19 year old students.
Yet, girls recorded a higher pass
rate than boys in every GCSE STEM
subject except maths. Disappointingly,
only 27% of girls’ A-Level entries
were in STEM subjects, compared to
45.6% of boys’ entries. And at A-Level
Physics, the number of girls taking the
subject amounts to just 21.5% of total Why is it that so few women are considering engineering as a viable
entries.
In a report last year, the Institution career and what can be done to address it? By Bethan Grylls
of Engineering and Technology (IET),
found that a mere 11% of the UK’s bias among young people. Male pupils they struggle even a little, they think
engineering and technical workforce is were more likely than girls to have it’s because the subject is too difficult
female – with just 9.6% making up the a better perception and enjoyment for them.”
electronics sector. of STEM subjects; know of people
While there’s a need to encourage in STEM; regard engineering as Persistent perceptions
more students of both genders to ‘interesting’ or think that engineering Although it’s clear that there is a
entertain a career in engineering, the would work well with who they are. predetermined image of the engineer
lack of female engineers is helping to Research funded by the Institute – male – it is not going unchallenged.
exacerbate the problem and reducing of Physics (IoP) and ASPIRES project, Initiatives from the Science,
the pool of people able to address the found that a majority of girls viewed Engineering and Manufacturing
shortage of engineers. physics as a “hard subject” and that, Technologies Alliance (Semta),
“Often, I’m the only female in “hard subjects are for men”. Secondary Engineer, and the UK
meetings even in 2018,” said Ann- “Pupils don’t find all the concepts Electronics Skills Foundation (UKESF)
Marie Bayliss, a product manager at hard, it’s just an ingrained belief and are all working to increase STEM
Murata. So, where are the females? bias,” Amanda Valentine, Head of uptake among young people whether
The EBM survey points to a gender Physics at Leweston School, said. “If at school or as apprentices.

14 8 May 2018 www.newelectronics.co.uk


Leweston School in Dorset has
adopted a structure known as the
‘Diamond Model’, which involves
STEM lessons moving from mixed
classes in lower years, to single-sex
lessons between the ages of 13-15,
and then back to mixed sixth form
lessons.
According to data from the
Independent Schools Council,
published in 2016, the GCSE
performance for girls and boys
is higher in a Diamond Model
environment than in a fully co-ed one,
with 96.4% of girls achieving five
A*-C grades compared to 89.7% in a
co-ed environment, and 97.8% of boys
achieving five A*-C grades compared
to 87.6%.
Semta, which offers guidance in
www.newelectronics.co.uk 8 May 2018
recruitment practices and apprentice
programmes, found in a survey for
the Industry Apprentice Council (IAC)
last year that 98% of apprentices
were happy with their chosen
apprenticeship. The response from
male and female was very similar.
But despite positive comments
from female engineers, a gender gap
remains.
The EBM survey found a lack of
confidence among girls, with 30%
saying they didn’t believe they could
become an engineer even if they
wanted to. That compared to 19% of
boys.
“We’ve moved a long way from
traditional stereotypes, but there
is still a lot more to be done,”
said Kaylee Southworth, a STEM
Ambassador and TVOB operations
manager at BT. “It’s suggested that
something as simple as wording
in job adverts can put females off
a technical role, whereas a male,
lacking the necessary skills, will go for
the role regardless.”
There is also a gender bias in
parental attitudes. Male parents are
more likely than female to see an
engineering career as interesting, so
is it surprising that so few females
work in the sector?
“Girls face peer pressure,” Bayliss
said, “and need support both from
home and school to encourage them
to make the most of their talents.”
However, according to Stewart
Edmondson, CEO of the UKESF which
promotes hands-on, electronics-
based activities in the classroom,
and collaborates with companies
and universities to tackle the skills
shortage: “If it were a case of just
encouraging females we would have
solved the problem by now.
“There must be another underlying
reason for the lack of women in
electronics and I think it’s largely
down to schools.”
He continued, “We know that too
few girls take STEM subjects and
within that electronics is particularly
poorly represented. We need to
COVER STORY WOMEN IN ENGINEERING
demonstrate it’s a job for males and
females, and that women can have
demanding and satisfying STEM
careers.”
He believes “schools are the
key” in tackling the unintentional
reinforcement of stereotypes.
According to the EBM, 29% of male
teachers feel STEM careers are more
“The curriculum is for boys than girls.
already crowded, New Electronics (NE) spoke to
but we can’t drop a group of female STEM-studying
any subjects to students who reaffirmed this, saying
make more room that females were made to feel less
for electronics. capable of taking “difficult subjects
Instead, we like Science”.
have to use the “I do think that barriers are caused
opportunities we by traditional perceptions of what an
have to improve engineer should be,” Bayliss ventured.
the existing “I definitely had to work harder as a
curriculum” woman.”
Stewart The students also said they were
Edmondson “more aware of male scientists” and
had “few female role models” to
inspire them.
Edmondson agreed: “If you go
into a science department, male role
models dominate.”
Motivating students
Students often tend to see their
teachers as role models. The EBM
survey found that 61% of students
would seek a teacher’s advice, similar
to those who said they’d talk to a
parent or guardian.
Sarah Dempsey, Development
Scientist at Perateach, said
her teacher inspired her. “He
was extremely supportive and
enthusiastic,” she recalled, “and
wanted to get more women involved
in Science. He was the exception,
though. My secondary school didn’t
tend to encourage girls.”
“If you speak to people in the
industry, it’s generally a teacher
that inspired or motivated them,”
“Often, I’m the Edmondon agreed.
only female in So, is the problem that most
meetings even schools and the curriculums they
in 2018” teach fail to motivate female
Ann-Marie students?
Bayliss An IAC survey found careers advice
15
 COVER STORY
was failing. Only 22% of students
said they received “good” advice and
30% of female respondents revealed
they weren’t encouraged to take
an apprenticeship by their school/
college, almost double that of male
students. The majority, 85%, said
that they were encouraged to pursue
higher education.
“My teachers were horrified that
I wanted to do Science and not Art
subjects,” explained Bayliss. “I was
strongly encouraged to reconsider my
choices.”
The EBM survey found that
most parents and, more worryingly,
teachers believed they weren’t
equipped to give advice. Only 36%
of parents and 35% of teachers said
they had the confidence to offer
guidance about possible careers in
engineering.
Like UKESF and Semta, Secondary
Engineer looks to encourage young
people into the industry. Operated by
the Society of Operations Engineers, it
works to equip schools with the skills
to teach practical maths and science
and how they relate to engineering.
According to Secondary Engineer’s
CEO, Susan Scurlock, “If you’re
looking to make a significant change
it has to start at the core, and that’s
with teachers. You can have a big
impact on a larger number of children
if the teacher is both knowledgeable
and enthusiastic about the subject.”
Worryingly, EBM found that few
students knew the steps to becoming
an engineer, with a quarter believing
a degree was required and, among
11-14 year olds surveyed, a majority
said they knew ‘almost nothing’ about
what apprentices do and how to
become one.
“Awareness is key. If girls know
there’s an opportunity they will get
involved.” Southworth added. “Going
into schools and actually sharing
information opens up girls’ thinking
and makes them realise there are
exciting opportunities within science
and technology.”
“The idea of what women should
16
WOMEN IN ENGINEERING
and shouldn’t be doing is engrained,”
Dempsey said. “Sometimes it’s that
fear of what you don’t know that puts
you off something. Schools need to
start working closely with universities
and industry to provide students
with a better understanding. It’s not
something schools do enough.”
In fact, while there are 600
UK organisations running STEM
initiatives, the EBM survey indicated
that only 28% of young people aged
11-14 had taken part in a STEM
careers activity in the last year.
This could be due to the lack of
STEM teachers, the Department for
Education needs to recruit 2,000
additional STEM teachers each year
to hit its targets, but it could also be a
case of ‘too many cooks’.
Is there a failure to delivery STEM
in the classroom, and do STEM
initiatives need to be revisited and
restructured?
Stephen Howse, policy coordinator
at Semta, said that most educators
lack industry experience.
“Teachers don’t understand the
engineering sector. The majority have
never had a career in industry. They
have never left the world of education
and their understanding of industry,
and what it is and involves, is limited.”
“There isn’t necessarily a need
for more industry-experienced STEM
teachers,” Valentine argues. “Instead
schools need more opportunities for
STEM Ambassadors to come into the
classroom and talk about their careers
and experience.”
Whereas Scurlock thinks that,
“the metrics by which we measure
education” in the UK is unhelpful.
“Children are taught to focus on their
exams and once they’re over, forget
everything because they’ve not been
taught how to use that information.”
According to Scurlock, this focus
on passing exams means that all
children – not just females – don’t see
a true reflection of what engineering
can offer.
Edmondson agreed, suggesting
that Science is too broad a subject
and that teachers tend to concentrate
on delivering the parts of the subject
that will appear in the exam and
this can get in the way of promoting
extra-curricular/careers activities for
students.
“The curriculum is crowded as it
is, so what do you drop?” Edmondson
questioned. “The answer is nothing,
“The idea of what but instead, we use the opportunities
women should we have to improve the curriculum
and shouldn’t we’re working with.”
be doing is There has been much discussion
engrained. over the years about just how relevant
Schools need the education system is when it
to start working comes to producing engineers.
closely with Industry frequently says that students
universities lack basic skills.
and industry to Government sets educational
provide students policy, and industry has a
with a better responsibility to engage with
understanding of academia and Government to shape
engineering” that policy.
Sarah Dempsey Will the Tech or T-Levels which
the Government intends to use to
overhaul technical education help?
Between now and 2022, 15
new pathways will be developed in
15 sector areas where substantial
technical training is required
to progress into employment
and includes engineering and
manufacturing.
Should education be about training
or is ‘real’ education, as some critics
would argue, about focusing on open-
ended tasks that encourage creativity,
inventiveness and co-operation?
If education policy was to change
along those lines, would it help to
encourage more women to embark on
a career in engineering?
There certainly needs to be a
better appreciation about the role of
science in today’s society and culture,
and that will be determined not just by
the curriculum, but by the attitudes of
parents and teachers alike.
“Awareness Crucially, schools need to ensure
is key. If girls that they are offering the same
know there’s an opportunities to their female students
opportunity they as they do to males.
will get involved” After all, whatever a man can do, a
Kaylee Southworth woman can certainly do too.
8 May 2018 www.newelectronics.co.uk
When flash failure
is not an option
The tension between cost and reliability is driving the development
of flash-memory controllers that can support critical embedded
applications. The New Electronics team reports
F
lash memory has made it
possible to pack gigabytes of non-
volatile storage into a tiny space,
a factor that is making it possible
to bring advanced algorithms and
artificial intelligence to car navigation
and driver-assistance systems
and other critical applications. But
flash’s rapidly increasing density
can come with costs: reliability and
performance.
Axel Störmann, vice president
memory marketing and application
engineering division at Toshiba
Memory Europe, says: “By shrinking
devices, makers create more sensitive
products. As a result, devices are
getting slower, less reliable and less
robust.”
There are limits on how densely
the memory cells in a 2D array
can be packed, which has forced
manufacturers to look at alternatives
to the 1bit-per-cell architecture.
18
Jason Chien, product marketing
manager at Silicon Motion, says:
“While these advanced NAND Flash
technologies enable us to store
albums and photographs on a mobile
phone’s microSD card memory for
example, there’s a drawback when it
comes to miniaturisation. There are
so few electrons in each cell, that
data is prone to loss through leakage,
interference from neighbouring cells
or from corrupted read or write
operations.”
Manufacturers have looked instead
to increase the number of bits they
can stored in each location.
The first step was to move to
multiple-level cell (MLC) designs
that store two or more bits in each
physical gate. But this is much more
difficult to achieve reliably than with
the simpler single-level cell (SLC)
architecture.
Axel Mehnert, vice president of
marketing at memory-controller vendor
Hyperstone, says: “When reading data
[from a SLC array], you measure the
threshold voltage when the electrons
or current can flow and interpret it as
a binary value representing the data.
In MLC, different amounts of electrons
are ‘pushed’ per physical cell and
establish multiple different threshold
levels.
“The problem with this is that when
measuring multiple bits per cell there
is a greater likelihood of errors: the
Left: Toshiba’s threshold needs to be measured more
SSDs utilise its precisely.”
BiCS FLASH 3D Over time, in any flash device, the
flash memory, process of driving electrons onto
which is a floating gate and removing them
based on the damages the delicate interfaces.
company’s 64- Charge traps form over time that,
layer architecture eventually, cause cells to become
and uses 3-bit- unusable. The problems become more
per-cell (TLC) acute with MLC because the read
technology and write electronics are much more
sensitive to the resulting variability in
usage.
“If you are writing and erasing your
drive a lot, it will fail more quickly,”
Mehnert says.
The problem has become more
apparent in recent years because
of scaling. “Many manufacturers
are working with nodes at, or below,
20nm and, as a consequence,
the number of electrons that are
associated with each bit has fallen,”
explains Mehnert. “Due to production
variations whether it’s temperature or
some other physical effect, it’s likely
that over time cells will not represent
the correct state, leading to errors.”
“You can’t offer The reliability issues could
everything in one cause problems for mission-critical
device. Storage embedded systems. Chien points
systems will have to car information systems as an
be designed with example, which play a key role in
application at driver assistance functions where
the forefront if the loss of a single bit of data could
we are to move threaten the safety of the driver,
forwards.” passengers and other road users.
Axel Mehnert To address the need for reliable
data storage for critical applications, a
set of memory controller technologies
are now being deployed in SSDs and
8 May 2018 www.newelectronics.co.uk

memory arrays that are intended to
be used in automotive and industrial
applications.
Hyperstone and Silicon Motion have
developed techniques to cater for the
reduction in cell-level reliability. Such
memory controllers manage the flow
of data into and out of NAND Flash
arrays, using features such as error
checking and correction (ECC) on read
and write operations to maintain data
integrity.
Silicon Motion has developed
end-to-end data path protection, a
technique the company says can
guarantee the integrity of data written
to or read from an SSD.
A particular problem for the
non-volatile memory in embedded
systems compared to desktop
systems is that some locations will
receive very high write traffic as they
log real-time changes, which puts a
small proportion of cells under high
stress.
The controller can scan for these
‘at-risk’ cells and move the data to
lower-traffic regions before they fail.
“These controllers enable SSDs,
based on the latest high-density TLC
and 3D NAND memory arrays, to be
used safely and reliably in safety-and
mission-critical applications,” Chien
says.
Mehnert warns that data refresh
is a complex process. “Every refresh
will add to the wear-down. If you over
refresh you will end up destroying the
flash, so you need a fairly balanced
refresh strategy which is also tightly
coupled to the use-case.
“For example, with read-only
memory you may want to refresh in
cases where the device is
exposed to high
temperatures
or you have
to scan read
the whole drive
to figure out where
weak areas are.
If you have a use-
case that is very write
intensive, but where the
www.newelectronics.co.uk 8 May 2018
data is not stored for long, then a
different refresh strategy may have to
be considered.
“That approach requires a lot of
statistics and management of the
environmental conditions when you
devise a refresh strategy and it will
also need to be application specific.”
Hyperstone has developed a set of
algorithms which attempt to spread
the wear-down evenly over a whole
number of cells, so if a particular area
fails, it prevents the entire system
failing.
Changes to the way that flash
memories are made could provide
devices that are less prone to failure.
Toshiba is looking to 3D NAND
structures as a possible solution.
“3D NAND allows us to grow many
layers on one die, which helps
with the capacity issue,” Störmann
explains. “The big challenge for 2D
NAND is being able to address all
SECTOR FOCUS STORAGE TECHNOLOGY
these different demands. 3D can work
around these.”
Some 3D NAND approaches form
vertical strings of bits that allow tens
of bits to occupy the same location on
the die.
These devices are now moving into
production for high-density SSDs.
According to Mehnert, newer 3D
Flash architectures also comes with
new failure modes: “In 3D structures,
more cells are connected. If there’s a
shortage or strings break, the number
of cells that become defective will
be significantly higher, with the result
that bit errors and data loss could go
Above: Hyperstone’s beyond the ECC’s capability. If one cell
controllers offer a breaks you can correct it using the
turnkey platform for ECC, but, if a whole block breaks, you
the industrial space may need something like Redundant
and include features Array of Independent Disks (RAID)
such as wear- to manage data redundancy and
levelling performance improvement.”
There are other approaches
to making 3D NAND suitable for
embedded systems.
Störmann points to through-silicon
via (TSV) technology with conventional
2D NAND devices as a possible
alternative.
“This is a stacking option,” he
explains. “In one package you
can accommodate 32 dies on one
another, providing higher densities
“While advanced which means the capacity improves
NAND Flash and provides quicker speeds.”
technologies Despite the density advantages
offer a multitude associated with 3D Flash, Mehnert
of benefits, believes there is still value in using
miniaturisation 2D and that different markets will
isn’t without it’s require different levels of reliability.
drawbacks.” “We still have a lot of systems
Jason Chien which are low capacity and very
reliable. It just depends on the
application,” he contends.
He insists that the only way forward
Left: Silicon Motion when it comes to memory is to
includes data error compromise.
detection with “You can’t offer everything in one
recovery engines device. Storage systems will need
to offer data integrity to be more application specific and
throughout the entire new features will require different
Host-to-NAND-to- approaches, and that will entail
Host data path continuous innovation.”
19
How ‘smart’ is the
smart factory?
How is the move to digitisation changing the face of
electronics manufacturing? Frank Zycinski looks at the
impact it’s having on factories
I
t would be fair to say that the
electronics sector is full of
buzzwords, from the Internet of
Things (IoT), to Industry 4.0 and smart
factories. While there is a debate to
be had as to whether these are ‘new’
technologies or just processes that
are being ‘rebranded,’ the fact is
that these concepts are now taking
centre-stage within the manufacturing
landscape.
Contract manufacturing is a
market in which design, development,
manufacturing and distribution teams
are having to work closely together to
deliver new technologies, whether it’s
20
for the IoT, or using augmented reality
(AR), Li-Fi or artificial intelligence (AI).
While it’s hard to predict the speed,
scale, and scope of how these trends
and technologies will become a reality,
contract manufacturers will have a key
role in enabling them.
So, how are these trends impacting
contract manufacturers and the way
they engage with their customers?
When it comes to high-complexity
electronics, manufacturing the move
into digitisation, or smart factories,
has been building for years. But,
in reality, it is only now that the full
impact of this move is being felt
by both design and manufacturing
companies as well as their customers.
Impact of digitisation
The digitisation of manufacturing
works on two fronts. Manufacturers
are using digital technologies to
deliver greater process efficiencies,
while customers are looking to develop
smaller and more power-efficient
products at a lower cost, with the
additional ability to connect to any
kind of data infrastructure.
One of the major changes that this
is driving is that Original Equipment
Manufacturers (OEMs) are now using
Electronics Manufacturing Services
(EMS) – particularly when it comes
to getting new IoT products onto the
market. Why? Because these products
tend to often cut across traditional
sectors, pushing OEMs to merge their
traditional product lines with wireless
technologies with which they have little
experience.
It is a question of technologies and
competencies. With many, particularly
start-ups, it is that they have a product
idea that is often in an advanced
stage of planning/feasibility, but
8 May 2018 www.newelectronics.co.uk

they need a partner to engineer and
deliver it. They are looking to work
with a company that already has the
technology and competencies in place
to deliver to their requirements rather
than having to heavily invest in new
equipment and expertise themselves.
A prime example of this is
the pharmaceutical sector where
companies now pursue a different
business model to that of even ten
years ago. They are moving towards
using experts outside of their own
business to enable them to develop
and deliver new products in a cost-
effective way.
An EMS provider is no longer
considered just for its production
capabilities, but for the complete value
stream of services that it is able to
offer an OEM.
This is causing a sea-change in
the way the two parties interact. It is
very rare today that a company arrives
with a fully formed design project and
then says “make this for us”. It has
become a much more collaborative
process with the OEM selecting
a design-through-to-manufacture
company to engage with from the
earliest stages of the process. This
has the benefit of a product being
designed from the outset to maximise
its manufacturability, reducing the
opportunities for defects and delays
down the line and improving time to
market.
However, it is not just in the design
for manufacturing and engineering
stages that this collaboration is taking
place. It also extends to product
qualification and aftermarket services.
This is leading to longer relationships
between the partners and a deeper
understanding of the end customer’s
needs on both sides.
Defining the rules of engagement
This trend is also driving a need for
standardisation of processes and
terminology across the electronics
production market. The introduction of
Industry 4.0 is one example where a
‘buzzword’ is given life and becomes a
www.newelectronics.co.uk 8 May 2018
platform for key industry issues to be
discussed and then implemented.
Industry 4.0 has been referred
to as the fourth industrial revolution
through the use of cyber-physical
systems, IoT, cloud computing,
and cognitive computing. But what
does this actually mean to an EMS?
It’s a bit like the weather forecast -
without access to constant data and
measurement stations, a reasonable
prediction is impossible.
Industry 4.0 and the smart factory
enables a clearer picture of what is
happening in the factory. For example,
more facilities are using sensors that
are able to communicate what is
happening and where. This allows for
better control, planning and delivery.
This also means that the EMS
provider can confidently communicate
on progress with its customer at every
stage of the production process.
This notable change in the level of
communication within the partnership
has arisen as a direct result of the
technological developments in modern
electronic production. Today, it is
essential that effective relationships
are built with all parties at an early
stage of a product’s development.
These may be interdepartmental,
between companies or on an
international scale. Whatever
the scope, defining the rules of
SECTOR FOCUS CONTRACT
MANUFACTURING
Author Profile engagement for all parties at an
Frank Zycinski is early stage in a formal, contractual
Market Sector VP agreement is now best practice.
in EMEA, Plexus Building these networks early on
provides a common ground for the
development of trust between the
stakeholders. It allows for increased
levels of communication. These
ground rules should be laid down
at the initial proposal stage. By
describing and defining the division of
deliverables within the scope of the
project, it ensures that all participants
clearly understand the work to be
done. This provides the foundation for
a successful environment.
During the life of any project
something always ‘comes up’.
However, with a clear path of action
and defined levels for escalation
already in place, this should never
become a problem. With all these
areas covered there should be no
barrier to implementing a successful
programme for any project. Utilising
best practices means that all
Below: An EMS stakeholders will benefit from the
provider is no longer experience and capitalise on a broader
considered just base of knowledge.
for its production
capabilities, but for ‘Smart’ production
the complete value This has been a learning curve across
stream of services the whole electronics manufacturing
that it is able to offer sector. The need for wide ranging
an OEM communication has become essential
as EMSs find they are working on
increasingly complex products for
companies outside of its traditional
customer base. More OEMs are
looking to work with an EMS that can
offer both a full range of engineering
skills and close physical proximity,
so that they can keep the lines of
communication easily open.
Smart factories, Industry 4.0
and IoT may be what everyone
is talking about, but it is a high
level of communication between
all stakeholders that will ensure
that success is achieved. This is
the major change that will move
the electronics sector even further
ahead and facilitate a step change in
performance.
21

Measuring battery life
Poor battery life is affecting the take-up of too many devices. How can power be analysed in
enough detail to ensure products live up to consumer expectations? By Neil Tyler
O
ne of the technical reasons
holding back the uptake
of Internet of Things (IoT)
products is poor battery life. So, how
can the power consumption of an
embedded system be analysed and
optimised in enough detail to deliver
‘reasonable’ battery life?
According to Vanja Samuelsson
founder of Qoitech, a Swedish start-up
owned by Sony: “There’s a gap in the
market for an affordable development
tool that’s suitable for the IoT. The
IoT is extremely complex and many
of the people creating solutions are
inexperienced.
“While power has always been
crucial, from a hardware perspective
these devices require layers of
software to operate. This can mean
that choosing the wrong protocol or a
failure to take into account the impact
of software updates can help to ruin
the overall user experience.”
Power measurement tool
“Looking for greater simplicity, Qoitech
has developed Otii which allows quick
and simple power measurement
analysis and comparison, providing a
clear link between measurements and
code execution. It offers an integrated
method for performing automated
measurements and testing,” explains
Samuelsson.
Otii, with its hardware unit Arc and
associated multi-platform desktop
application, enables embedded
software developers to power-optimise
their application, while hardware
developers can make informed
choices during their power conversion
and battery management component
selection.
“In low-power embedded
www.newelectronics.co.uk 8 May 2018
EMBEDDED DESIGN
systems the choice of SoC or MCU
is perhaps the most significant,
and semiconductor vendors have
taken many approaches to saving
power. Sometimes peripherals,
coupled with a direct memory access
(DMA) controller, move data to and
from SRAM without the processor
needing to be active. But from the
microcontroller’s datasheet alone that
is hard to assess,” says Samuelsson.
Another challenge surrounds
oscillator stabilisation after wake-up.
A clock peripheral using a quartz
crystal will need time to stabilise
before the processor can start
processing instructions after being
woken up. If a phase-locked loop
(PLL) is included, perhaps offering
higher clock frequencies, this also Above: The Otii
needs time to lock, taking additional from Qoitech
time. Different microcontrollers allows quick and
offer different solutions and the one simple power
that truly delivers the lowest power measurement
consumption is normally found through analysis and
part experience, part luck. comparison
The Otii Arc is a single piece
of hardware that delivers a
programmable power source, wide
POWER
dynamic range current measurement,
and additional side measurement and
data capture channels.
By providing control over the supply
voltage while measuring the current
consumption, various power source
strategies can be evaluated. If the
impact of sub-circuitry needs to be
understood, the differential analogue
input on the expansion connector,
together with a small resistance in the
supply line, allows the power impact to
be clearly separated out from the total
power consumption.
The second key element, according
to Samuelsson is the serial data RX
input of the expansion port. Serial
communication data can be captured
via this interface at rates of 9600
bits per second (bps) up to 4Mbps.
Any logging messages output by the
application are then time-stamped
and synchronised with the other
time-domain measurements. A real-
time operating system (RTOS)-based
application could, together with
instrumented trace logging via a UART,
be tracked as it switched between
tasks and into and out of its idle-task
23
 EMBEDDED DESIGN
and low-power modes. The final
element is the single-ended
analogue input that is available
when differential-mode power
measurements are not being
made.
This can be used to capture
any analogue or digital signal,
Voltage, CCV
allowing a simple toggled
general-purpose input/output
(GPIO) pin to be monitored.
Analysis made simple
“The Otii software makes the
analysis of power consumption
simple. Split into three sections;
a side bar, where recordings
and visualisations can be
configured, a graph area, and a
log view,” says Samuelsson.
In order to easily understand
the relationship between power
consumption and application
execution, the graph and log views
are linked to one another. Minimum,
maximum, average and energy
measurements also automatically
update as the highlighted
measurement area or spot-point
are moved around. Searching for
key moments in the application’s
execution is simplified with the search
function within the serial output’s log
data.
Since IoT nodes cannot function
in insolation, it is often the case that
power consumption measurements
are needed for several nodes
simultaneously. In certain mesh
configurations, some devices may take
on the role of a gateway, resulting
in higher power consumption than a
device functioning simply as a node.
The Otii software can differentiate
between several connected Otii Arc
tools, even allowing the user to give
them individual names, thus, ensuring
usability when quickly switching
between the various pieces of
hardware being measured.
Other advanced features include
battery cell simulation and scripting
support. The integrated battery
simulator allows one of several
24
POWER
3.2
3.0
2.8
Pulse
2.6
2.4
2.2
2.0 IR
1.8
0 25 50 75 100 125
Capacity, mAh
different types of batteries to be
selected instead of defining a
fixed supply voltage. Each battery
selected, results in the display of
its key parameters together with
the manufacturer’s discharge
curve. Through use of a slider, the
developer can also define the level of
discharge of the cell, which is ideal
for understanding an application’s
reaction to the voltage drop that
occurs in the last 10-20% of a
battery’s life.
By using the industry standard Lua
language, serial data from log files
captured outside of the Otii software
can be imported.
“Providing these log files contain
time-stamping information, it is
possible to convert them into the
Otii format for import into the GUI,”
explains Samuelsson.
Two example scripts are provided
that handle Android ADB or Linux
Kernel logs. The scripts are also
available in the installation folder,
allowing them to be used as the basis
for modified scripts. In case the time
domain of power measurements and
the log files do not match, the log
view also provides an option to apply
either a positive or negative offset
in milliseconds, allowing data to be
quickly synchronised to power
140 measurement recordings.
Battery profiling can also be
120 undertaken using scripting.
Bknd A cell manufacturer’s
datasheet tends to provide a
100
combination of constant load
discharge, constant current
80
IR, ohms
discharge and pulsed load
discharge curves, normally at
60 or around room temperature.
Since the developer’s
40 application is unlikely to closely
fit any of these off-the-shelf
discharge characteristics,
20
a hand-crafted test can be
scripted.
0
150 175 200 225 Such tests consist of a loop
that applies different loads for
different time periods while
logging the cell’s voltage.
Above: An example “The test can be halted either upon
of pulse discharge reaching a set number of iterations, or
characteristics when the cell’s voltage has reached a
lower limit,” according to Samuelsson.
Toward the end of product
development, it is also useful to run
longer-term tests or even set up an
automated testing capability. Here,
the scripting capability coupled with
execution of the Otii software from the
command-line interface can be used.
Hand-built test harnesses or even
build and testing environments, such
as Jenkins, can be used to trigger and
monitor such test runs.
“When it comes to developing
products for the IoT, developers
need the best insights into power
consumption possible during each
cycle of the product creation process
whether that’s system design and
architecture, silicon selection,
development and testing,” says
Samuelsson.
The Otii Arc hardware, with its large
dynamic range, together with the Otii
software – featuring a simple and
clean interface – has been designed
to make it easy to link moments of
an application’s high-energy usage to
specific sections of application code,
microcontroller or SoC configuration,
or sub-system circuitry state.
8 May 2018 www.newelectronics.co.uk
ww.newelectronics.co.uk 8 May 2018 25
The right approach
When it comes to choosing between Embedded Linux or a proprietary real-time operating system,
there are many different considerations to take into account, as Marcus Nissemark explains

A
n Embedded Linux distribution
with its kernel and software
packages provides a broad
set of ready-made software, which is
attractive to managers and developers.
Although it requires considerable
knowledge of the dependencies and
configuration details, it is frequently
used in the industry. A proprietary
Real-Time Operating System (RTOS)
with a minimal code base and limited
middleware availability however, can
be just as capable of meeting the
same software requirements, but,
with a different approach. There are
many considerations when selecting
between Embedded Linux or a
proprietary operating system. Not just
cost, but also complexity; ease of use;
advanced compile and debug tools; port code to the build system. This is Figure 1. The software development process when
time to market; developer productivity; likely to cause significant unexpected Yocto Project building Embedded Linux systems.
and safety and security concepts. work as the intention was to reuse workflow It is a suitable solution for many
existing software packages. illustrating the Embedded Linux applications, as
Hardware choice RTOS vendors normally work with complex scenario long as the work associated with it is
The challenge starts at the hardware silicon vendors, enabling each other’s understood.
level. Many popular SoCs support business by supporting the SoC in The Yocto project can be tailored
Linux, but it is often limited to the question. It is important to ask them and configured, but requires significant
silicon vendors’ dedication to deliver, about their support, as they often knowledge, and is perceived, by some,
support and upstream their Linux resort to their partner network and as a bloated solution that needs a lot
kernel and user space contributions, ecosystem for supporting extended of disk storage and long build times.
effectively supporting only a subset middleware such as graphics or The default setup of a Yocto project
of the open-source components. For connectivity software. They can build, downloads a significant set of
instance, NXP supports the popular provide significant help and valuable build scripts and software, for example
NXP i.MX6 Cortex-A9 SoC through information about the use case. recipes and packages that are used
the Yocto project for a limited set of They also can provide features like for configuring, building and installing
recent Linux kernels, or through the security enhancements, functional components to your Embedded Linux
LTIB build tool for older releases. This safety compliance, or hard real-time system.
may restrict users to one Linux kernel capabilities that a Linux solution The Yocto recipes will incorporate
or middleware version, or one build cannot. a cross-compiler from a third party or
system, and the features that this The requirements lead to the build one from source. If this is not
supports. usage of the Yocto project and its managed correctly, it can create a
Depending on requirements, a build system. Yocto is an open-source configuration mismatch.
user may end up having to implement project, driven by the Linux Foundation, Additionally, the step of building
functionality in existing middleware or whose focus is on improving the a cross-compiler for the target bloats

26 8 May 2018 www.newelectronics.co.uk


the host installation and build process
with unnecessary files and extra time,
and causes an unnecessary risk from
a test and validation perspective. An
RTOS like INTEGRITY usually requires
a validated version of the toolchain/
compiler that is mandatory to use, and
some commercial Linux alternatives
bundle this into their Yocto support.
This helps to remove risks associated
with toolchain configurations, and
simplifies the process for the
developer.
The next step is configuration
management of the Embedded Linux
system within Yocto. This consists
of a Linux kernel, hardware related
patches and a number of open source
packages, each requiring their own
configuration setup, build process
and installation. It is easy to get
lost trying to navigate between what
the hardware vendor supports, what
the user requirements are, and the
actual available and tested software
solutions.
Understanding all of the
configuration possibilities and
package dependencies is one of
the bigger obstacles encountered
when developing an Embedded
Linux system. For instance, adding
support for a new software package
may bring in a dependency to
another software package or another
version of that package, which is not
included in the build system. This
quickly becomes hard to resolve,
and causes a cascading effort of
changes required in your system,
none of which were initially planned
for. This type of dependency dilemma
is similar to a ‘top down’ approach
to software package configuration
management. Each component or
package must be configured to require
fewer dependencies or to scale to
requirements. This can consume many
hours.
For a proprietary RTOS, there are
still dependencies to various versions
and components supported by the
vendor or their partner network, so
choice is limited to the software
www.newelectronics.co.uk 8 May 2018
EMBEDDED DESIGN
solutions they support. However,
the proprietary RTOS approach to
configuration management is normally
to include only the components
required, using a simpler ‘bottom-up’
approach. In other words, far less time
will be spent configuring components
since they are already pre-configured
for the specific use case by vendors.
This helps reduce effort and puts
focus on to the user’s own software
development.
Developer focus
Using Embedded Linux solutions
leverages existing software
capabilities and saves development
time. If developers spend time on
other activities, there is no developer
focus. Tools beyond the compiler
can address that, where powerful
debuggers, Integrated Development
Environments (IDEs), or other support
tools make developers efficient. All
Embedded Linux systems and some
proprietary RTOSes rely on the well-
performing GNU compiler gcc. The
GNU debugger requires extensive
configuration and knowledge to get
it to support debugging properly,
partly mitigated by the usage of
Eclipse based configurations and
pre-configured setups. In the end,
many developers tend to use printf()
debugging, mainly due to lack of
better tools, or lack of knowledge
or time to properly configure the
debugging solution.
Generally, RTOS vendors put
extra effort into the tools and
their debugging solutions are pre-
OPERATING SYSTEMS
Author details configured and work out of the
Marcus box. These solutions are also
Nissemark is a bundled with additional analysis
Field Applications and performance monitoring tools
Engineer at Green that aid development. Application
Hills Software development for any operating system
is a straightforward task when using
common APIs like POSIX, socket
APIs for connectivity, and OpenGL for
advanced graphics. The application
code does not depend on the choice
of RTOS. Easy access and properly
setup tools make developers more
efficient, increasing their focus and
productivity.
There can still be a need for
education and support. RTOS vendors
provide professional support services
that really can make a difference
when problems occur, and they
provide professional training courses
by their own personnel. Embedded
Linux training is available from various
stakeholders, including professional
support services, and there is online
support in abundance. Just remember
that searching for suitable support
may be a time-consuming challenge.
Conclusion
Embedded Linux is attractive at first
sight because of its low upfront costs,
but make sure you consider the
following:
Time to market – Do not
underestimate the time needed for
problem resolution, configuration
Figure 2. The MULTI management and software package
Debugger from dependencies;
Green Hills Software Developer efficiency – Efficient, easy
to use and pre-configured tools make
developers more efficient, allowing
them to find and resolve more bugs
and provide better software quality;
Engineering Time – Engineering
time is a cost related to the project;
developers should focus only on the
actual application development; and
Alternative requirements – Functional
safety, high security or real-time
characteristics can be hidden
features that strengthen your end
product.
27
 SYSTEMS DESIGN SYSTEM
ARCHITECTURES

simply collapse. Google themselves

said if everyone in the world with an
Android phone were to use their voice
assistant for just 3 minutes they
would have to double the number of
servers.”
Beyond the consumer-visible
applications of voice response and
image recognition, machine learning
is moving into deeply embedded
applications. Ceva decided to
incorporate machine learning into its
Penta-5G IP cores destined for the
next generation of mobile phones.
Emmanuel Gresset, director of
business development at Ceva, says

The cost of learning

the increases in bandwidth promised
by 5G are made possible by the
combination of techniques such as
MIMO and beamforming: “This is
How are embedded system designers addressing the incredible efficient only if link adaptation is very
well done. But it’s a complex task.
complexity associated with machine learning? By Chris Edwards If the link adaptation is not accurate
you will not achieve the throughput
and you will consume a lot more

M
achine learning has been
through numerous slumps,
each following a spasm of
technological over-enthusiasm. But it
may be the only way to deal with the
incredible complexity of situations
that embedded-systems designers
now face – something that is now
being reflected in the strategies of
embedded-processor designers.
Rene Haas, IP products group lead
Illustration: illumate/stock.adobe.com
Although people tend to associate
machine learning with image and
audio recognition through services
like Apple’s Siri, the systems already
in the market send a lot of data to
the cloud to be processed. That is
changing rapidly, says Haas.
“What is rising is the amount
of inference and training that can
take place at the edge. The level of
analytics, the level of learning, level
power.
“We looked at traditional,
more algorithmic approaches. The
[parameter] dimensions are so
large that it would require a lot of
memory to do that. The complexity
increases exponentially. That’s why
we came up with a neural network-
based approach. The AI processor
“The level of takes as input the channel conditions
analytics, of and it computes the best transmit

at ARM, says: “We believe machine
learning is one of the most significant
change that is hitting our computing
landscape. We believe that years
from now people will not be looking
at machine learning as a unique
category where computers learn but
rather it will be native to everything
that computers do.”
of sophistication performed locally learning and of parameters,” Gresset notes.
has moved much faster than I think sophistication
everyone has anticipated,” Haas performed locally Trade-offs
explains, pointing to bandwidth and has moved much The problem that faces designers
server-capacity problems as chief faster than I think of machine-learning accelerators
concerns. everyone has is the trade-off between flexibility
“Why not send the data to the anticipated.” and energy consumption. There
cloud? There isn’t enough bandwidth Rene Haas are many implementations, and
in the world; the internet would computer scientists are rapidly

www.newelectronics.co.uk 8 May 2018 29

 SYSTEMS DESIGN
adding modifications that attempt
to cater for different types of
data and application. There is no
guarantee that currently fashionable
convolutional neural networks (CNNs)
will remain dominant as researchers
work on other options ranging from
techniques based on statistical
processing such as Gaussian
processes to architectures that
borrow even more from neurology
than CNNs.
General-purpose processors offer
the ability to handle the widest range
of options but they have seen their
best days, argues Steve Mensor, vice
president of marketing at Achronix:
“As the tasks have become more
complex, clock speed have maxed
out, meaning we need more CPUs.
The only thing that keeps going up
and to the right is transistor count.
All the other functions have tapered
off. So, the CPU’s ability to keep up
with market requirements appears to
have been lost. Accelerators will be
the brawn. A lot of start-up companies
are looking at different architectures
but they are all working on highly
parallelised compute units that are
targeted for a particular application.“
To maintain a balance between
efficiency and programmability, Ceva,
Cadence Design Systems and others
favour augmenting parallelised DSPs
with specialised instructions working
alongside more specific hardware
accelerators. The accelerators home
in on frequently used operations
such as the thresholding step that
typically happens after a neuron has
incorporated data from all its inputs.
Other vendors are building AI
accelerators into programmable-logic
fabrics to take advantage of the way
that these devices make it easy
to route data between processing
elements without the energy-sapping
need to shift data to and from
memory at each stage. Having seen
its FPGA, as well as those from
chief competitor Intel PSG, move
into data-centre blades for various
acceleration tasks, including machine
30
SYSTEM ARCHITECTURES
learning, Xilinx aims to capitalise on
the shift with its forthcoming “Project
Everest” devices. These will link
the core programmable-logic fabric
to more specialised programmable
accelerators using an onchip network
that CEO Victor Peng says he expects
will be reconfigured on the fly.
Mensor argues the embedded
FPGA technology has advantages over
the off-the-shelf variety. Embeddable
IP provides the ability for customers
to add their own optimisations to the
programmable fabric. He says the
company has worked with a customer
on building specialised cores for
CNNs: “They handle different types
of object-recognition applications and
they use an array of kernels. We have
cores where we have changed the
DSP to the requirements they have,
which cuts 50 per cent of the silicon
area of those blocks. Overall, we get
a 35 per cent die-size reduction on
the final device.”
Peng says embedded-FPGAs
companies have the disadvantage
of only having access to a
programmable-logic technology and
not the various other IP cores the
standalone-device makers can wrap
around them, such as high-speed
Serdes I/O ports.
Mensor says the I/O itself
becomes less important because the
embedded-FPGA technology makes it
easier to squeeze more onto a single
die. He points to the existing data-
centre applications where, according
to his company’s estimates, as
much as 60 per cent of the FPGA
is dedicated to I/O. Very wide local-
memory interconnect cuts the delay
and power to reach locally stored
data.
However, CNNs have prodigious
appetites for data that can make it
difficult to bring everything onchip.
“People are Pulin Desai, director of AI and imaging
realising there IP product development at Cadence,
are limits to says: “People are realising there are
the amount of limits to the amount of memory you
memory you can can put onchip. So the question is:
put onchip. So, can we find other methods?”
the question is: A number of the vendors are
can we find other working on techniques to reduce
methods?” the amount of data a trained CNN
Pulin Desai needs to access. Most have already
implemented narrow datapaths on
the basis that high data resolution is
important in training – which can be
offloaded to remote servers – but for
the real-time inferencing processes,
bit widths as narrow as 8bit offer
reasonable performance. There are
indications that some of the weight
calculations performed by artificial
neurons can scale down to binary.
Often there are redundant paths that
“The CPU’s can be pruned away entirely. Another
ability to keep option is to compress data at rest
up with market and only decompress it when it is
requirements needed.
appears to “There is a lot of activity in the
have been lost. compression of weights as well as
Accelerators will pruning and decompressing on the
be the brawn.” fly,” Desai notes.
Steve Mensor Ultimately, the memory issue
in machine learning and other
data-intensive algorithms that
look to dominate computing in the
decade ahead may force a more
radical change in architecture.
Server designers are already
pushing more processing into flash-
storage subsystems. Although the
programming will be much more
difficult, processors distributed
through memory may be the only
way to deal with the need to boost
performance without suffering
a massive increase in energy
consumption.
8 May 2018 www.newelectronics.co.uk