EMBEDDED DEVELOPMENT

AUGUST 2014
circuitcellar.com ISSUE 289
CIRCUIT CELLAR | ISSUE 289 | AUGUST 2014

Protect Your Electronics Explore how electromagnetic shielding

can defend against RF radiation
Tips for shield experiments,
3-D printed shields, and
using magnetic shields

EIM Bootcamp | Q&A: Embedded Systems Training Expert

circuitcellar.com

Digital Pulse Generator | Battery Charge Control | MCU-Based ECG System

Linux Application Design | Implement a Soft-Core Processor |
Fixed Resistors | IR Transmission & Formats Self-Reconfiguring Robotics
2 CIRCUIT CELLAR • AUGUST 2014 #289

EDITOR’S LETTER

Issue 289 August 2014 | ISSN 1528-0608

CIRCUIT CELLAR® (ISSN 1528-0608) is published monthly by:
Circuit Cellar, Inc.
111 Founders Plaza, Suite 300
East Hartford, CT 06108
Periodical rates paid at East Hartford, CT, and additional offices.
One-year (12 issues) subscription rate US and possessions
$50, Canada $65, Foreign/ ROW $75. All subscription orders
payable in US funds only via Visa, MasterCard, international
postal money order, or check drawn on US bank.
SUBSCRIPTIONS
Circuit Cellar, P.O. Box 462256, Escondido, CA 92046
E-mail: circuitcellar@pcspublink.com
Phone: 800.269.6301
Internet: circuitcellar.com
Address Changes/Problems: circuitcellar@pcspublink.com
Postmaster: Send address changes to
Circuit Cellar, P.O. Box 462256, Escondido, CA 92046
ADVERTISING
Strategic Media Marketing, Inc.
2 Main Street, Gloucester, MA 01930 USA
Phone: 978.281.7708
Fax: 978.281.7706
E-mail: circuitcellar@smmarketing.us
Advertising rates and terms available on request.
New Products:
New Products, Circuit Cellar, 111 Founders Plaza, Suite 300
East Hartford, CT 06108, E-mail: newproducts@circuitcellar.com
HEAD OFFICE
Circuit Cellar, Inc. 111 Founders Plaza, Suite 300
East Hartford, CT 06108
Phone: 860.289.0800
COVER PHOTOGRAPHY
Chris Rakoczy, www.rakoczyphoto.com
COPYRIGHT NOTICE
EE PROJECTS, TIPS, & INSIGHT
I’m pleased to note that this issue is one of our most diverse to date. We cover a wide
range of technologies and topics from pulse generation to FPGAs to embedded Linux to self-
reconfiguring robots at MIT. After you get through this issue, I encourage you to share your
thoughts with our staff and followers via social media (Twitter: @circuitcellar and Facebook:
www.facebook.com/circuitcellarmagazine).
On page 8 we run an interview with Victoria, BC-based professional engineer Jason Long,
who founded Engenuics Technologies in 2010. His innovative contribution to the industry is
an embedded systems hardware training program, which he describes in his interview.
Interested in a digital pulse generator? Brian Beard explains how he built a microcontroller-
based fully digital pulse generator that’s accurate, programmable, and stable (p. 20).
Turn to page 28 for the second part of Steve Hendrix’s “Personal Solar Power Setup”
series. In Part 1, he introduced his system. This article focuses on battery charge control.
We’re always on the lookout for innovative healthcare-related electronics projects. On
page 36 we present one of the best projects we’ve seen in several months. Sean Hubber and
Crystal Lu present their microcontroller-based electrocardiography (ECG) system.
Check out Bob Japenga’s article on Linux system configuration if you want a break from
hardware design projects—or if you’re a Linux buff (p. 42). Bob looks a different ways to
configure a Linux system. He covers BusyBox, web servers and server-side support, and
more.
On page 46, Colin O’Flynn touches on the topic of implementing a soft-core processor
inside an FPGA. He notes that recent releases have made developing and customizing
these processors easier than ever.
You know everything there is to know about resistors, right? We think there’s always
more to learn. This month, George Novacek tackles the subject of fixed resistors (p. 56).
Circuit protection is an important topic for engineers and seasoned pros alike. As you
know, electromagnetic shielding protects against RF radiation. But adding shielding isn’t
always easy. Robert describes some shielding experiments and shares his results (p. 60).
Flip to page 66 to read Part 1 of Jeff Bachiochi’s new series on “IR Remotes.” This month
he introduces IR formats and then focuses on the topic of remote transmission.
We wrap up this issue with Kyle Gilpin’s presentation on self-reconfiguring robotic
systems and M-Blocks (p. 80). He describes some of
the amazing technologies engineers are working on
in MIT’s Distributed Robotics Lab. The nearby photo
shows two different views of M-Blocks.

Entire contents copyright © 2014 by Circuit Cellar, Inc. All

rights reserved. Circuit Cellar is a registered trademark of
C. J. Abate
Circuit Cellar, Inc. Reproduction of this publication in whole
or in part without written consent from Circuit Cellar, Inc. is cabate@circuitcellar.com
prohibited.

DISCLAIMER

Circuit Cellar® makes no warranties and assumes no

responsibility or liability of any kind for errors in these
programs or schematics or for the consequences of any
THE TEAM
such errors. Furthermore, because of possible variation in
the quality and condition of materials and workmanship of EDITOR-IN-CHIEF PRESIDENT FOUNDER
reader-assembled projects, Circuit Cellar® disclaims any
responsibility for the safe and proper function of reader- C. J. Abate Hugo Van haecke Steve Ciarcia
assembled projects based upon or from plans, descriptions, or
information published by Circuit Cellar®. MANAGING EDITOR PUBLISHER PROJECT EDITORS
The information provided by Circuit Cellar® is for educational
Mary Wilson Don Akkermans Chris Coulston,
purposes. Circuit Cellar® makes no claims or warrants that
Ken Davidson,
readers have a right to build things based upon these ideas ASSOCIATE EDITOR ASSOCIATE PUBLISHER
under patent or other relevant intellectual property law in
Nan Price Shannon Barraclough David Tweed
their jurisdiction, or that readers have a right to construct or
operate any of the devices described herein under the relevant
patent or other intellectual property law of the reader’s ART DIRECTOR COLUMNISTS CUSTOMER SERVICE
jurisdiction. The reader assumes any risk of infringement KC Prescott Debbie Lavoie
liability for constructing or operating such devices. Jeff Bachiochi, Ayse K.

© Circuit Cellar 2014 Printed in the United States ADVERTISING COORDINATOR Coskun, Bob Japenga, Robert
Kim Hopkins Lacoste, Ed Nisley, George
Novacek, Colin O’Flynn
 circuitcellar.com 3

US/UK
Don Akkermans
+31 46 4389444
OUR NETWORK
d.akkermans@elektor.com

ELEKTOR LABS
Wisse Hettinga
+31 46 4389428
w.hettinga@elektor.com

GERMANY
Ferdinand te Walvaart
+49 241 88 909-17
f.tewalvaart@elektor.de

FRANCE
Denis Meyer
+31 46 4389435 SUPPORTING COMPANIES
d.meyer@elektor.fr

NETHERLANDS 12th Int'l SoC Conference 77 Ironwood Electronics 79

Harry Baggen
+31 46 4389429
h.baggen@elektor.nl 2015 International CES 53 Labcenter Electronics 1

SPAIN
Eduardo Corral Accutrace 7 Lauterbach GmbH 27
+34 91 101 93 95
e.corral@elektor.es
Advanced Micro Circuits 78 MaxBotix, Inc. 79
ITALY
Maurizio del Corso
+39 2.66504755 All Electronics Corp. 78 microEngineering Labs, Inc. 79
m.delcorso@inware.it

SWEDEN CadSoft Computer GmbH 39 Mouser Electronics, Inc. 17

Wisse Hettinga
+31 46 4389428
w.hettinga@elektor.com
Custom Computer Services 78 MyRO Electronic Control Devices, Inc. 78
BRAZIL
João Martins
+31 46 4389444 Elektor 22, 23 NetBurner, Inc. C2
j.martins@elektor.com

Elektor 71 Pico Technology Ltd. C3

PORTUGAL
João Martins
+31 46 4389444
j.martins@elektor.com EMAC, Inc. 19 R.E. Smith, Inc. 49

INDIA
Sunil D. Malekar ExpressPCB 55 Reach Technology, Inc. 78
+91 9833168815
ts@elektor.in
Flash Memory Summit 59 Saelig Co., Inc. 19
RUSSIA
Nataliya Melnikova
+7 965 395 33 36 Hot Chips Symposium 45 Technologic Systems 9, 11
elektor.russia@gmail.com

TURKEY Humandata, Ltd. 33 TechTools 79

Zeynep Köksal
+90 532 277 48 26
zkoksal@beti.com.tr IAR Systems 31 Triangle Research International, Inc. 79

SOUTH AFRICA
Johan Dijk Imagineering, Inc. C4
+31 6 1589 4245
j.dijk@elektor.com

CHINA
Cees Baay NOT A SUPPORTING COMPANY YET?
+86 21 6445 2811
ceesbaay@gmail.com Contact Peter Wostrel (circuitcellar@smmarketing.us, Phone 978.281.7708, Fax 978.281.7706)
to reserve your own space for the next edition of our members’ magazine.
4 CIRCUIT CELLAR • AUGUST 2014 #289

CONTENTS

AUGUST 2014 • ISSUE 289

EMBEDDED
DEVELOPMENT

DIY DIGITAL PULSE GENERATOR

CC COMMUNITY FEATURES
06 : CC WORLD 20 : Build a Digital Pulse Generator
By Brian Beard
08 : QUESTIONS & ANSWERS An accurate DIY microcontroller-based digital pulse
Embedded Systems Training Expert generator for your workbench
By Nan Price
An interview with an engineer who launched an 28 : Personal Solar Power Setup (Part 2)
embedded systems hardware training program Battery Charge Control
By Steve Hendrix
Battery charge control options for sharing the load
INDUSTRY & ENTERPRISE between a utility and the solar power system
16 : PRODUCT NEWS
36 : Hands-On Electrocardiography
19 : CLIENT PROFILE By Sean Hubber & Crystal Lu
Total Phase (Sunnyvale, CA) An ECG system to extract a heart’s electrical output

COLUMNS
42 : EMBEDDED IN THIN SLICES
Linux System Configuration (Part 2)
Linux Application Design
By Bob Japenga
Options for selecting the packages to install on your
system and more

46 : PROGRAMMABLE LOGIC IN PRACTICE

Build a SoC Over Lunch
By Colin O’Flynn
Getting started with a simple soft-core processor
ARM-BASED DEVELOPMENT BOARD
 circuitcellar.com 5

CONTENTS

DIY SHIELDING FOR ELECTRONICS

MICROCONTROLLER-BASED ECG SYSTEM

56 : THE CONSUMMATE ENGINEER

The Humble Resistor (Part 1)
TESTS & CHALLENGES
Fixed Resistors 74 : CROSSWORD
By George Novacek
An introduction to fixed resistors, resistor selection, 75 : TEST YOUR EQ
and power dissipation
TECH THE FUTURE
60 : THE DARKER SIDE
Shielding Experiments 80 : Self-Reconfiguring Robotic Systems
By Robert Lacoste and M-Blocks
How shielding protects against RF radiation By Kyle Gilpin
A look at the future of self-reconfigurable robot
66 : FROM THE BENCH development and applications
IR Remotes (Part 1)
Remote Transmission
By Jeff Bachiochi
Accumulate and recognize an IR remote transmission

SELF-RECONFIGURING ROBOTS

FIXED RESISTORS 101 @editor_cc

@circuitcellar circuitcellar
 6 CIRCUIT CELLAR • AUGUST 2014 #289

CC WORLD

EIM “BOOTCAMP”
By C. J. Abate

Several weeks ago, Elektor International Media (EIM)

staffers from around the globe met up in the Netherlands
for a corporate “bootcamp.” The agenda: take a day to step
away from engineering, circuit testing, editing, and writing
to assess the company’s current offerings (magazines,
books, kits, etc.), discuss the needs of members, and plan
for the future. While every EIM staffer was not present, a
few dozen were in attendance. Thus, the event comprised
staffers from Elektor magazine, Circuit Cellar magazine,
Elektor.Labs, audioXpress magazine, Voice Coil magazine,
EIM management, and EIM’s customer service teams. What
follows are some key conclusions taken from the event.

MEMBER FOCUS
Each year, EIM runs comprehensive member surveys will work to engage the
COMMUNITY

in order to gauge what readers and customers want. We’ll next generation of young
continue to do this, but also engage our community more engineers more actively
often in a variety of places: online via social media and in than ever via social
person at industry events. The goal: more interaction with media, video, and more.
members and thus enhanced offerings in the near future.
E-LEARNING
SHARE KNOWLEDGE & EXPERTISE The message from
EIM has electrical engineering, audio engineering, and our community is clear:
programming experts around the globe working on content offer more e-learning
for Circuit Cellar magazine, Elektor magazine, Elektor. opportunities. We intend
Labs, Voice Coil magazine, and audioXpress magazine. to do this. In addition to
We currently share content between brands (e.g., we our currently scheduled
occasionally publish relevant electronics tips from Circuit webinars and technical
Cellar in Elektor.) Moving forward, we’ll work even harder videos, we will focus on
to share relevant content between brands. The goal is to developing a more robust
enrich the experiences of our members diverse content. e-learning offering. In
fact, we’ve already begun!
REACH & INSPIRE Stay tuned.
An important short-term goal is to reach and inspire
young engineers. Many of our current members have been
reading EIM’s magazines, such as Circuit Cellar, Elektor, and
audioXpress, since their early teens. Moving forward, we
 8 CIRCUIT CELLAR • AUGUST 2014 #289

QUESTIONS & ANSWERS

Embedded Systems
Training Expert
An Interview with Jason Long
Professional engineer Jason Long worked as an embedded systems designer
for more than a decade. In 2010 he founded Engenuics Technologies. Jason
lives in Victoria, BC, where he continues growing his company alongside the
MicroProcessor Group (MPG) embedded systems hardware teaching program
he developed in 2000.—Nan Price, Associate Editor
COMMUNITY

This ARM-based development board is
made by Jason’s company, Engenuics
Technologies.
NAN: In 2010 you founded your company
Engenuics Technologies (www.engenuics.
com) based on the success of the
MicroProcessorGroup (MPG) program. Give
us a little background. How did the MPG
begin?
JASON: MPG started way back in 2000 at the
University of Calgary when I was doing my
undergraduate studies. I figured out that
embedded systems was exactly what I wanted
to do, but struggled to find enough hands-
on learning in the core curriculum programs
to satisfy this new appetite. I was involved
in the university’s Institute of Electrical and
Electronics Engineers (IEEE) student branch,
where someone handed me my first Microchip
Technology PIC microcontroller and ran a few
lunchtime tutorials about getting it up and
running. I wanted more, and so did other
people.
I was also very aware that I needed to
drastically improve my personal confidence
and my ability to speak in public if I was going
to have any luck with a career outside of a
cubicle, let alone survive an interview to get a
job in the first place. The combination of these
two things was the perfect excuse/opportunity
to start up the MPG to ensure I kept learning
by being accountable to teach people new stuff
each week, but also to gain the experience of
delivering those presentations.
I was blown away when there were almost
30 people at the first MPG meeting, but I
was ready. Two things became very clear
very quickly. The first was that, to be able to
teach, you must achieve a whole new level of
mastery about your subject, but it was also
okay to say, “I don’t know” and find out for
next week. The second was that I could, in
fact, get my nerves under control as long as I
was prepared and didn’t try to do too much.
I’m still nervous every time I start a lecture,
even 14 years later, but now I know how to
use those nerves! The best part was that
people really appreciated what I was doing
and perhaps were a bit more tolerant since
MPG is free. I found a love for teaching that I
didn’t expect, nor did I get how rewarding the
endeavor would be.
When I was wrapping up the ninth year of
the program, I considered giving it one more
year and then calling it quits. I took a moment
to look back at what the program was when
I started and where it had come to—it had
indeed evolved a lot, and I figured I had put
in about 2,000 h by this point. It seemed like
a waste to throw in the towel. I also looked
at the relationships that had come from the
program, both personally and professionally,
and realized that the majority of my career
and who I had become professionally had
really been defined by my work with MPG. But
the program—even though it was still just in
 DESIGN YOUR SOLUTION TODAY
CALL 480-837-5200

TS-4900 Computer Module Touch Panel Computers

1.2GHz i.MX6 w/ WiFi & Bluetooth Panel Mount or Fully Enclosed

pricing
starts at
series
$134 starts at

qty 1 $ 369
$ 99 qty 100
qty 100
$ 409
qty 1
1.2 GHz Single or Quad Core Cortex A9 ARM CPU
Up to 2 GB DDR3 RAM and 4 GB eMMC Flash
Wireless 802.11 b/g/n and Bluetooth 4.0
1x Gigabit Ethernet, 1x PCI Express Bus
1x mSD slot, 1x SATA II, 1x USB Host, 1x USB OTG
70x DIO, 4x I2C, 1x I2S, 2x SPI, 2x CAN
-40 to +85C Industrial Temperature Range
Features can include:
Runs Linux 3.10, Debian, Ubuntu, Yocto, QT, OpenGL
5-inch, 7-inch and 10-inch touchscreens
Coming Soon: QNX, Android and Windows Support
Fanless operation from -20 to +70C
Up to 1.2GHz ARM CPU
Computer-on-Modules
Up to 2GB RAM, 4GB eMMC Flash
State of the Art Embedded Design
2x microSD with DoubleStore
TS-4740: High Capacity FPGA and Gigabit Ethernet 2x Ethernet, 2x USB Host
TS-4600: 450MHz low cost w/ 2 Ethernets CAN, RS-232, SPI, I2C, DIO
TS-4710: Up to 1066MHz PXA168 w/ video 1x RS485 Two-Wire Modbus
TS-4720: Like TS-4710 + 4GB eMMC Flash & 2 Ethernets Optional cellular, WIFI & XBEE radios
TS-4200: Atmel ARM9 w/ super low power Supports Android and Linux w/ fast boot times
TS-4800: 800MHz FreeScale iMX515 w/ video Headphone connector and speaker

We’ve never Embedded Support every Unique embedded

discontinued a systems that step of the way solutions add
FE

L
D

NA
N
LI

GE

PE

GI

product in 30 are built to with open value for our

NG

RI
RU
LO

years endure source vision customers

 10 CIRCUIT CELLAR • AUGUST 2014 #289
QUESTIONS & ANSWERS
COMMUNITY
MicroProcessorGroup (MPG) Level 2
focuses on application development.
Calgary—was too big to keep as a side project.
I had $10,000 in inventory to support the
development boards, and although all monies
stayed in the program, there were thousands
of dollars exchanging hands. This was a
business waiting to happen, though I had
never thought of myself as an entrepreneur. I
was just doing stuff I loved.
Around the same time I discovered
SparkFun Electronics, and more importantly, I
discovered the story of how the company got
started by Nathan Seidie. That story begins
almost exactly how MPG began, but clearly
Nathan is a lot smarter than I am and has built
an amazing company in the same time it took
me to get to this point. I feel quite disappointed
when I think about it that way, but thankfully
I don’t think it’s too late to do what I should
have done a long time ago. I hope to meet
Nathan one day, but even if I don’t, I consider
him a mentor and his story provides validation
that the MPG platform and community may be
able to grow and be sustainable.
I think MPG/Engenuics Technologies can
find similar success as SparkFun. We can do
that without ever having to compete against
SparkFun because what we do is unique
enough. There might be a bit of overlap, but
I’m always going to try to complement what
SparkFun does rather than compete against
it. We simply become another resource to
feed the voracious and infinite appetite for
information from students, hobbyists, and
engineers. Win-win is always the way to go.
I decided I should grow the program
instead of ending it, so I started Engenuics
Technologies, which would be built on the
decade of MPG experience plus the decade
of embedded design experience I had from
the industry. It seemed like a pretty solid
foundation on which to start a company!
Surely I could promote all of the content
and find students of the same mindset I was
in when I started MPG? They could lead the
program at different universities and develop
those infinitely valuable communication and
leadership skills that MPG fosters, except
they’d have the advantage of not having to
put in hundreds of hours to write all of the
material. Even if groups of people weren’t
playing with MPG, individuals could make use
of the technical resource on their own and we
could have a solid online community. I also
wanted to keep students engaged beyond the
single year of their engineer degrees in which
MPG existed.
I didn’t fully know how I was going to
execute this plan, but perhaps I had earned a
bit of karma because a few serendipitous things
happened just three weeks after I incorporated
Engenuics Technologies. My phone rang, and
it was Dynastream Innovations, a Calgary
company that developed the ANT+ protocol that
competes with Bluetooth Low Energy (BLE).
They had just interviewed three students who
brought their MPG development boards with
them to show off the ANT radio that was on the
board that they knew how to run. Dynastream
had just discovered MPG, and they were blown
away.
To wrap up this long story, Dynastream
understood the objectives and the potential of
MPG immediately and wanted more. I figured
out how to achieve my plan of growing MPG into
a resource that students would have through
their entire degree program and shared that
with Dynastream. Dynastream became my
first customer that sponsored a very intensive
four months of my time to develop MPG Level
2 in time for the upcoming fall school term.
Now MPG had the continuity it was looking
for and Dynastream had access to a trusted
resource of students who proved that they had
incredible interest, ability, and commitment to
working with embedded systems.
These are exactly the people who
Dynastream wants to hire, and MPG exposes
them without ever having to go through the
regular (and horrible) process of filtering
resumes and spending days in job interviews
trying to figure out if there is a match.
Dynastream’s continued support of the
program four years later tells me it’s working
for them. There are nearly a dozen companies
in Calgary now that sponsor MPG to support the
initiative and take advantage of the resource.
There’s no reason why this can’t be replicated
in every city in North America, right?
NAN: What other products/services does
Engenuics Technologies provide?
JASON: I describe Engenuics Technologies
as a four-tier company as there are three
 DESIGN YOUR SOLUTION TODAY
CALL 480-837-5200

TS-7670 Industrial Computer TS-7250-V2 Embedded Board

GPS & Cellular Modem to Track Assets High Performance & Industrial Grade
pricing starts at
Backwards Compatible with TS-72xx
$168 $129
qty 1 qty 100
pricing
shown w/ starts at
optional
micro $199
SD Card qty 1

$165
qty 100

low cost plastic

enclosure available

Features: Features:
454MHz ARM CPU 1x USB Host Up to 1GHz ARM CPU 2x USB Host
Up to 256MB RAM 4x DIO, 2x CAN 512MB RAM 1x USB Device
2GB NAND Flash 2x COM, 1x RS-485 4GB eMMC Flash 6x Serial Ports
2x mSD Card Socket 1x Battery Backed RTC 2x SD Card Socket 75x DIO, 1x CAN

1x 10/100 Ethernet 1x Temperature Sensor 2x 10/100 Ethernet 1x PC/104 Connector

Benefits: Benefits:
Low power with 10mW sleep state Hardware Flexibility with On-board FPGA

-40 to +85C, 100% soldered-on components Several control I/O interfaces

Easy development w/ Debian and Linux 2.6 Launches your application in under a second

Boots quickly to your Embedded Application Easy development w/ Debian and Linux 2.6

Guaranteed available until 2025 High Data Reliability with DoubleStore

Coming Soon: -40 to +85C Industrial temperature range
TS-7680 like TS-7670 w/ WiFi & Bluetooth

We’ve never Embedded Support every Unique embedded

discontinued a systems that step of the way solutions add
FE

L
D

NA
N
LI

GE

PE

GI

product in 30 are built to with open value for our

NG

RI
RU
LO

years endure source vision customers

 12 CIRCUIT CELLAR • AUGUST 2014 #289
QUESTIONS & ANSWERS
This instant serial cable tester is one
of Engenuics Technologies’s designs—
COMMUNITY
and a favorite of Jason’s.
significant aspects of the
business in addition to MPG. The
main purpose of the company is
to fill a gap in the industry for
specific training in embedded
systems. There is very little
formal training to be found for
low-to-mid-level embedded
hardware and firmware
development and quality/value is
often hit or miss. From teaching
for 10 years while being an
embedded designer for the same
amount of time, I felt like I had
the right skills to create great
training. I had already created a
LabVIEW course that I delivered
internally for a company while
I worked there, and people were blown away
by the quality and content. I saw a huge need
to develop embedded-specific training to help
new graduates transition to the industry as
well as junior engineers who were lacking in
some fundamental engineering knowledge.
We have an embedded boot camp course
that is about 20% hardware and 80% firmware
focused, which I think is essential for new
engineering graduates getting into embedded
design. Though the course is based specifically
on a Cortex-M3 development board, we ensure
that we focus on how to learn a processor so
the knowledge can be applied to any platform.
Engenuics Technologies has several courses
now and we continue to offer those periodically
though never as often as we would like, as we’ve
become too busy with the other parts of the
company. We finally got an office last August
with an onsite training room, which makes the
logistics much easier, and we’re ramping up
the frequency of the programs we offer.
I would love to establish the trust in the
industry so Engenuics Technologies would be
the first place that an engineer would look
if they wanted to learn a new processor or
specific technology (e.g., ultra-low-power ANT
radios). I’m not talking a day-long marketing
session of fluff. We do week-long, intensive,
hands-on development that will just about
make your brain explode. Our classes are never
more than six people, so it is very personal.
We recognize that getting an industry
training company stabilized is going to take
some time. So meanwhile, we are paying the
bills by doing contract design. Through a great
network and trusted relationships, we work on
some pretty great projects for companies. This
started with our immediate contacts, but has
spread nicely through referrals and even some
random people who have found us online.
There seems to be an infinite number of
people or small companies that want to create
an embedded design but need some help to
do it. We specialize in getting those to market
quickly but also training these people to be
able to do it themselves. Our culture is very
adept to these kinds of relationships. We’ve
taken dozens of products from conception
through to production.
We have also commercialized some little
gadgets and tools that have come out of
necessity in our own designs. My personal
favorite is an instant serial cable tester that
lets you check if you’ve got a null modem cable
and also verifies all the conductors. You’ll really
appreciate this one if you’ve ever wasted an
afternoon debugging an UART driver that won’t
transmit only to discover you unknowingly
grabbed a null modem cable.
MPG has led us to scour the globe to find
great quality bench tools that students can
afford and we’re happy to pass that educational
pricing on to everyone. For too long we didn’t
even advertise this stuff outside of MPG. One
day I looked at my list of finished goods and
realized our product line had close to 50 items,
many of which a few people have discovered
and absolutely love us for providing. Did I
mention I’m not a very good entrepreneur?
We add new stuff all the time, and we’re also
open to suggestions on handy little tools that
may exist as breadboard hacks on your bench
but would benefit thousands of engineers who
need to solve the same problem.
All of this makes up what I consider to be a
very solid business plan (as every entrepreneur
surely says about their own business plan).
Finally we’re taking the time to properly spread
the word and offer what we do to the huge and
hungry market out there.
NAN: You’ve run an MPG course every year
at the University of Calgary. Do you consider
this course introductory or is it designed for
more experienced engineering students?
JASON: I consider MPG as the start of the
bridge to the industry. I assume the second-
year students who start MPG Level 1 barely
know what a resistor actually does and I ensure
that they get that kind of information. I have
interviewed new EE grads who couldn’t tell me
what a pull-up resistor is, and I’ll bet 95% of
them couldn’t explain a decoupling capacitor.
That’s just wrong. There are endless online
discussions about this from people all over
North America. The industry keeps calling on
academia to fix this, but an engineering degree
is already packed as it is. Plus, not every student
cares about embedded systems or how to
solder. I think it really comes down to industry
stepping up to offer the resources (not just
money—they need to get their people involved)
to make niche programs like MPG—but that’s a
whole other story and maybe a TED talk.

Back to MPG, I also assume that students have
taken at least one C programming course and
can tell me what 0xE is in decimal and binary or
at least use Windows calculator to figure it out.
However, if they can’t, it’s all written in my notes,
which are open source and freely available on our
MPG website.
I want everyone to feel comfortable and
to ask the “stupid questions” in the very safe
environment that MPG provides. So yes, in many
ways MPG is introductory, but what we get into
quickly surpasses the depth of what a student
will experience from their core courses because
we drill a narrow hole straight down to exactly
the information we need to make a processor
work. Since everything is written down (and
with the intent of one day being as complete
and comprehensive as the Khan Academy, which
provides “a free world-class education for anyone
anywhere”), it’s truly a resource that can be
leveraged by anyone at any time. Even I reference
my own notes when it has been a while since
I’ve worked with a particular piece of code or
hardware.
NAN: What can attendees learn from the MPG
course?
JASON: I would sum up the written course material
as everything I’ve learned in the industry for 14
years written down complimented with about
50,000 lines of commented code examples. If you
studied it all and learned everything, then you
would know everything I know about embedded
systems. Well, maybe not everything—I’m
probably about three years behind in my notes.
I always wish I could teach all of that content to
students, but as an extracurricular program, which
usually sits on top of five or six core curriculum
courses each term, what we can do in MPG is a tiny
subset. Once MPG Level 2 came along, we were able
to get much further, especially with the new Atmel
boards introduced last year that gave both Level
1 and Level 2 the same base platform and have
onboard J-Links, thanks to recently getting Segger
onboard with MPG.
We do soldering at the start of both Level 1 and
Level 2. I’m talking real soldering, not massive and
obsolete through-hole parts. The MPG development
boards come about 90% populated, but students
still put on around 40 components including 0603
passives and ICs as small as TSSOPs.
A lab tech once told me I should use bigger
parts or at least make extra-large pads to make it
easier. That is so fundamentally wrong to me and
I’ll never do it. I’ve taught 10-year olds to solder
0603s—it takes a bit of time, but then they can do
it no problem. Okay, some students never get it,
but neither do some practicing engineers and they
know it, and who cares? Knowing what we can’t
do so we can seek help is just about as good as
being able to do it. I make engineers solder during
circuitcellar.com 13
QUESTIONS & ANSWERS
interviews, by the way—that’s fun.
a)
While we’re doing the hardware, students are
learning to speak the jargon of embedded systems
and gain confidence with what they’re working on.
It is fundamentally different than a core course
because they’re doing it to feed an interest and
to see something work, not because they need
a good mark for their GPA. This transformation
from a shy, totally uniformed student who doesn’t
know what PCB stands for, to a confident, capable,
young engineer excited about blinking LEDs is one
of the most profound things I witness every year. I
wish there was a way to equate a monetary value
to that growth so students could really understand
the value of MPG. After all these years, I’ve found
that it clicks with a lot of them three to five years
after they graduate.
Anyway, after the hardware is done, Level 1 is
then all about using the main peripherals you find
on any microcontroller: general-purpose input/
COMMUNITY
output (GPIO) to work with LEDs and buttons,
PWM, UART, and a touch of I²C to get the LCD
working. We also give students some code to use
the onboard ANT radio to do neat things in the
exercises. We’re finishing up a new demonstration
project that the “hammies” out there might love:
a Morse code messenger and trainer. By the time
this is in print we should be pretty close to having
that going.
Level 2 takes all of those drivers and hammers
in the concept of an API and writing code into an
existing framework. The whole program focuses
on designing and implementing a simple version of
Pong that is played over the ANT radio. It’s pretty
challenging to build something that complex in
the time we have, so what we end up doing keeps
changing each time the program runs. We run
bare metal—no OS—so people really see what
has to happen to make a multitasking system. I
have a plan for a whole new MPG program that
incorporates FreeRTOS.
That’s the technical side of MPG, but now
that I really “get” this program I’ve realized that
the technical side is like 10% of the value that
students earn. MPG is designed now to work with
students through their entire undergraduate
degree program and then keep them engaged
(and supported) in their careers. Learning to
communicate and mentor are what it’s really all
about, but as I said earlier, this intangible aspect
of MPG takes a while to realize.
Don’t forget that this is all free for students.
We charge for the development boards to cover
their cost, though they’re discounted from what
individuals would pay because of the industry
sponsorship of MPG. As the program grows, we’ll
reduce this further and enhance things like our
bursary program and soldering competitions. It
would be really nice to actually cover some of our
labor costs in creating new content and delivering
MPG, as that’s still all out of pocket. I figure I’m in
for about $250,000 worth of time since I started
 14 CIRCUIT CELLAR • AUGUST 2014 #289
QUESTIONS & ANSWERS
COMMUNITY
Jason helped develop tracking devices
like this Loner safety-monitoring
device (SMD) when he worked at
BlackLine GPS.
MPG, but I have absolutely no regrets and no
intention on slowing down due to the intrinsic
value of the effort.
NAN: What type of work did you do prior to
Engenuics Technologies?
JASON: My entire career has been defined
by my involvement with MPG. My job out of
school was a dream-come-true as I was doing
firmware and hardware right away with a
company called BW Technologies. I got that
job because of an MPG contact and because I
actually had a solid skill set that could be put
to work immediately. The most fortunate part
of belonging to that company was establishing
my relationship with the owners and managers
who were brilliant people, technically and with
the business. They were very open about what
was going on and would talk about it if you
were interested.
I left BW Technologies when a team lead
position came up (through my network thanks
to MPG, of course) at General Dynamics (GD)
Canada, though I continued to stay in touch
with BW Technologies. I loved the opportunity
to lead a team and I loved the military aspect
of what GD did, but I didn’t like not doing
design while I was there. Fortunately, MPG
kept me engaged technically—that’s actually
when I decided to change the program from
PIC to ARM and step things up.
GD was a really great experience. It seemed
to take forever to dig in there, but eventually
my team was very successful and I loved what
we were doing. However, I made a difficult
choice to leave after just two years to go to
BlackLine GPS. BlackLine was a startup that
was discovered by some ex-BW Technologies
guys who invested and re-assembled a lot of
the core BW team. They went public, pushed
out some sweet products, and are now doing
really well. I was there for four years but was
actually working two jobs as I was ramping up
Engenuics on evenings and weekends. I felt
ready to finally give my own company a go full
time, and BlackLine was really supportive.
So just like that, a decade had gone by
where I was always doing embedded design
but also always teaching and growing MPG, and
somehow I wound up with my own business.
It seemed like an 80-h work week was pretty
common (I once did four 120-h weeks in a row),
so it’s like I earned 20 years of experience in
that time. People in the industry were always
very interested in MPG and impressed by the
effort, which really helped me grow my network
of people. Despite approaching burnout a few
times, I continue to love embedded design and
will be very surprised if I ever leave the field.
What’s most exciting now is that as the Engenuics
Technologies team grows, my very long “to-do”
list is getting done as I can hire more people
to pick away at it. I see my passion spreading
to others and of course there are thousands of
equally passionate people doing this stuff who I
can’t wait to meet, help, and learn from.
NAN: You earned your BSEE from the
University of Calgary in 2002. Can you
describe any of the projects you’ve worked
while you were there?
JASON: The professors at the U of C were a
phenomenal bunch and it was a privilege to
get to know them and work with them during
my undergraduate studies. I remain in contact
with many of them, and several are very good
friends. Aside from blinking some LEDs on
breadboards, the first complicated device I
built was an attempt at the IEEE Micromouse
competition. That proved to be a little much
and my robot never did do anything beyond go
forward, sense a wall, and then back up.
I originally thought I would base MPG
around building robots, but that proved
impossible due to cost. Building a robot is
still on my bucket list. I’ll likely get there once
my two boys are old enough to want to build
robots. I continue to fantasize about building
an autonomous quadcopter that can deliver
beer. I better get busy on that before its
commonplace!
Our IEEE student branch had a Protel 99
SE license and somehow I learned how to
design PCBs. The first board I designed was
a binary clock that I still use. I then did a
PIC programmer and later I built a combined
development board and programmer for MPG.
I also designed the PCB for our fourth-year
Capstone design project, which initially was a
very boring implementation of a phase-locked
loop, but became a lot more fun when I decided
to make it programmable with a keypad and
an LCD. I brought all these things to my BW
Technologies job interview and proudly showed
them off. For any students reading this, by
the way, landing your first engineering job is
probably 5% technical, 10% GPA, and 85%
enthusiasm and demonstrated interest and
achievement. It’s really boring to interview
someone who has done nothing extracurricular.
NAN: How long have you been designing
embedded systems? When did you become
interested?
JASON: My dad was a high school science
teacher and my mom was a nurse, so I didn’t
have a lot of technical influence growing up.
I loved talking physics with my dad, and I’m
one of the few engineers who can cook (thank
you, mom).
Aside from really liking LEGO and
 circuitcellar.com 15

QUESTIONS & ANSWERS

dismantling anything electronic (without a) b)

ever a hope of putting it back together but
always wondering what all those funny looking
components did), I barely demonstrated any
interest in EE when I was young. But somehow
I figured out in grade 12 that EE was probably
what I should study at university.
I’m sure I still had visions of being a video
game designer, but that nagging interest in
learning what those funny components did
steered me to EE instead of computer science. It d)
wasn’t until my second year at university when
someone gave me my first PIC microcontroller c)
that I really knew that embedded was where
I needed to be. That someone was a student
named Sean Hum, a brilliant guy who is now an
associate professor at the University of Toronto.

NAN: Are you currently working on or

planning any projects?

JASON: I add to my “to-do” list almost every
day. My main focus right now is to grow MPG (I
would love to talk to any student or professional
about getting this going in your community)
and the Engenuics Technologies product line.
I’m aggressively pursuing some grants to
get going on a few new products that I think
have some potential—those are secret, though
nothing that will change the world and they
are unlikely to put the valuation of Engenuics
Technologies in the billions any time soon.
We’re just wrapping up work on a customer
project for a Wi-Fi/general packet radio service
(GPRS) updateable LED message sign. These
things are very cool, and I’m going to buy a
bunch once they’re in production to be a core
part of the MPG “franchise” network since
they can be all over the world and updated in
unison over the air via a browser. There seems
to be infinite potential applications for this
thing from fast-food restaurants and lotteries,
to banks and tradeshows. I’m always excited
(and terrified) to see one of my designs come
to market. There are also more gadgets to
build and a long list of courses to finish and
start delivering.
I have six-month and 2.5 year-old boys who
are definitely projects in progress. I recently
registered a domain and a Facebook page
called embeddad.com. Now I have a whole
new to-do list of projects that let me continue
to feed my need for embedded design while
still pursing some work-life balance with my
family. I hope to share those and connect with
other engineer dads out there just to show off
some geeky stuff we do with our kids. I love to
enable people and if I can help you make your
kid’s day by putting flashing blue and red LEDs
on a toy police car, then you bet I will.
COMMUNITY
JASON: I particularly like the 2.4-GHz radio While studying at the University
technologies that hold the potential to really of Calgary, some of Jason’s first
make our environment interactive and embedded designs included a
programmable phase-locked loop
intelligent. I think the world needs more
Capstone project (a), a robot built for
intelligence to address the wasteful nature
an IEEE Micromouse competition (b),
of what we have become whether it is by an MPG development board (c) and a
actively doing something like turning the binary clock (d).
lights or heat off when we’re not around,
or by simply making us more aware of our
surroundings. I love ANT+ and am just getting
into BLE—obviously, smartphone integration
will be critical.
I think technology will drive change in
education and I hope to see (and perhaps be a
driving force behind) a more cohesive existence
between academics and the industry. I hope
MPG becomes a model to the industry of what
can be achieved with not a lot of financial
resources, but has immense payback for
employees who become mentors and students
who can connect with the industry much
earlier and thus get more from their degree Here is a larger view of the robot
programs and graduate with substantially Jason built for an IEEE Micromouse
higher capabilities. competition.

NAN: Which new technologies excite you?

 16 CIRCUIT CELLAR • AUGUST 2014 #289

PRODUCT NEWS

RN4020 BLUETOOTH SMART MODULE

Microchip Technology recently announced its first
Bluetooth 4.1 Low Energy module, the RN4020. The
integrated Bluetooth Low Energy (BTLE) stack and on-board
support for the common SIG low-energy profiles speeds
time to market, while ensuring Bluetooth compatibility,
eliminating expensive certification costs, and reducing
development risks. The module comes preloaded with the
Microchip Low-energy Data Profile (MLDP), which enables
you to stream any type of data across the BTLE link.
As a stack-on-board module, RN4020 can either connect
to any microcontroller with a UART interface or operate
standalone for basic data collection and communication,
such as a beacon or sensor. This standalone operation is
facilitated by Microchip Technology’s no-compile scripting,
which allows module configuration via a simple ASCII
command interface. Neither tools nor compiling are required.
INDUSTRY & ENTERPRISE
Microchip also recently announced the RN4020 Bluetooth
Low Energy PICtail/PICtail Plus daughterboard (part number
RN-4020-PICtail). The board enables code development via
USB interface to a PC, and over the on-board In-Circuit
Serial Programming interface for Microchip’s PICkit or
MPLAB REAL ICE tools.
The RN4020 Bluetooth Low Energy Module is available for
$6.78 each in 1,000-unit quantities. The RN4020 Bluetooth
Low Energy PICtail/PICtail Plus daughterboard costs $49.
Microchip Technology, Inc.
www.microchip.com

The designers of cost-sensitive embedded applications

are looking for turnkey solutions that make it easy to add
the low-power consumption and simplicity of Bluetooth LE
connectivity, which enables several years of operation from
a single battery and has a large installed base of compliant
smartphones, tablets and computers. The RN4020 Bluetooth
LE Smart module includes all of the hardware, software, and
certifications that you need to easily add this low-energy
connectivity to any design, while easing End Product Listing
(EPL) via QDID Bluetooth compatibility testing. All of the
programmable profiles are stored and selectable on the
module, including Microchip Technology’s flexible MLDP and
the common Bluetooth SIG low-energy profiles. In addition
to common public profiles, private services can be created
via the ASCII command interface. The RN4020 also provides
a built-in PCB antenna with 7-dBm transmit power and a
receive sensitivity of –92.5 dBm in a compact form factor of
only 11.5 × 19.5 × 2.5 millimeters.

TRACE32 TOOL QUALIFICATION SUPPORT KIT

The Lauterbach recently announced a Tool Qualification Support Kit for the TRACE32
code coverage analysis. The qualification toolkit was designed to help customers facing the
many challenges of software development for safety-related systems to
complement the tool qualification when TRACE32 is selected as preferred
solution for trace-based code coverage analysis. It aims to meet the
requirements for the development of avionic and automotive applications
in compliance with DO-178C and ISO 26262.
The TRACE32 Tool Qualification Support Kit is provided as prefabricated
package containing documents and a test suite for the validation of
TRACE32. In the upcoming months, the scope of the test suite will be
continuously extended to provide support for various platforms.

Lauterbach
www.lauterbach.com
 mouser.com
The Newest Products for Your Newest Designs®

ORE.

The widest selection

of the newest products.

Authorized distributor of semiconductors

and electronic components for design engineers.

Mouser and Mouser Electronics are registered trademarks of Mouser Electronics, Inc. Other products, logos, and company names mentioned herein, may be trademarks of their respective owners.
 18 CIRCUIT CELLAR • AUGUST 2014 #289

PRODUCT NEWS

FAST QUAD IF DAC

The AD9144 is a four-channel, 16-bit, 2.8-GSPS DAC that
supports high data rates and ultra-wide signal bandwidth to
enable wideband and multiband wireless applications. The
DAC features 82-dBc spurious-free dynamic range (SFDR) and
a 2.8-GSPS maximum sample rate, which permits multicarrier
generation up to the Nyquist frequency.
With –164-dBm/Hz noise spectral density, the AD9144
enables higher dynamic range transmitters to be built. Its
low SFDR and distortion design techniques provide high-
quality synthesis of wideband signals from baseband to
high intermediate frequencies. The DAC features a JESD204B
eight-lane interface and low inherent latency of fewer than
two DAC clock cycles. This simplifies hardware and software
system design while permitting multichip synchronization.
The combination of programmable interpolation rate, high
sample rates, and low power at 1.5 W provides flexibility
INDUSTRY & ENTERPRISE

when choosing DAC output frequencies. This is especially

helpful in meeting four- to six-carrier Global System for
Mobile Communications (GSM) transmission specifications
and other communications standards. For six-carrier GSM
intermodulation distortion (IMD), the AD9144 operates at 77
dBc at 75-MHz IF. Operating with the on-chip phase-locked
loop (PLL) at a 30-MHz DAC output frequency, the AD9144
delivers a 76-dB adjacent-channel leakage ratio (ACLR) for
four-carrier Wideband Code Division Multiple Access (WCDMA)
applications.
The AD9144 includes integrated interpolation filters with
selectable interpolation factors. The dual DAC data interface
supports word and byte load, enabling users to reduce input
pins on lower data rates to save board space, power, and cost.
The DAC is supported by an evaluation board with an
FPGA Mezzanine Card (FMC) connector, software, tools,
a SPI controller, and reference designs. Analog Devices’s
VisualAnalog software package combines a powerful set
of simulation and data analysis tools with a user-friendly
graphical interface that enables users to customize their
input signal and data analysis.
The AD9144BCPZ DAC costs $80. The AD9144-EBZ and
AD9144-FMC-EBZ FMC evaluation boards cost $495.
Analog Devices, Inc.
www.analog.com

NEW STM32 MICROCONTROLLER DESIGN TOOLS

New STM32 microcontroller design tools from STMicroelectronics bring the efficiency of the STM32Cube software
platform, and the convenience of STM32Nucleodevelopment boards, to projects targeting STM32 F3 mixed-signal and STM32
F0 entry-level devices. For STM32 F31 and STM32 F02 devices, the latest STM32CubeF3 and STM32CubeF0embedded software
provides free and ready-to-use dedicated middleware stacks, such as touch-sense and USB libraries, and a collection of
drivers for STM32 peripherals through the Hardware Abstraction Layer (HAL) that simplifies porting of applications from
one STM32 series to another. Using this embedded software with the latest version of the STM32CubeMX PC-based tool,
developers can configure the microcontroller with graphical wizards, and simply click to generate initialization C-code that
can be used directly in many popular third-party development tools.
The new software joins similar STM32Cube platforms released earlier in 2014 for STM32 F4, F2 and L0 series, which
simplify project starts and include hundreds of code-usage examples.
STM32Nucleo development boards are also now available for the STM32 F3 (NUCLEO-F302R8) and STM32 F0 lines (NUCLEO-
F072RB) to assist early application development, leveraging the large ecosystem of Arduino application shields as well as
dedicated ST extension boards. STM32Nucleo boards provide access
to all device I/Os and integrate an ST-Link debugger/programmer,
saving the need for a separate debug probe. Last but not least,
the STM32Nucleo boards are mbed-enabled, which means they
can be directly connected and run through the online development
environment at mbed.org.
The STM32Nucleo boards start at $10.32. For further information
on STM32Nucleo boards, go to www.st.com/stm32nucleo. To download
STM32Cube software free of charge, go to www.st.com/stm32cube.
.
STMicroelectronics
www.st.com
 circuitcellar.com 19

CLIENT PROFILE

Total Phase, Inc.

www.totalphase.com
735 Palomar Avenue, Sunnyvale, CA 94085

CONTACT: stephen@totalphase.com

FEATURED PRODUCTS: The Aardvark I2C/SPI Host Adapter is

a fast and powerful I2C bus and SPI bus host adapter through
USB, allowing a developer to interface a Windows, Linux, or
Mac OSX PC via USB to a downstream embedded system
environment and transfer serial messages using the I 2C and sale terminals, our users have leveraged our technology
SPI protocols. Part Number: TP240141 (www.totalphase.com/ for unprecedented visibility into closed systems for over a
products/aardvark-i2cspi/) decade. Not only do you get visibility into your system, you

INDUSTRY & ENTERPRISE

The Komodo CAN Duo Interface is a powerful two-channel USB-
to-CAN adapter and analyzer. The Komodo interface is an all-
in-one tool capable of active CAN data transmission as well
as nonintrusive CAN bus monitoring. Part Number: TP360110
(www.totalphase.com/products/komodo-canduo/)
can address a device at the chip level regardless of how it was
programmed. Once you have the information unlocked, feed
your IoT solution and complete the circuit. Close the loop by
using our programming tools to enhance your embedded IoT
solutions. Build solutions in IoT today.

WHY SHOULD CC READERS BE INTERESTED? Intelligent
visibility into embedded systems means business. Unlock your
infrastructure and stretch your investment into the Internet
of Things (IoT). Discover new ways to unlock your data on
formally closed systems. From unlocking a city bus to point of
Circuit Cellar prides itself on presenting readers with information
about innovative companies, organizations, products, and
services relating to embedded technologies. This space is
where Circuit Cellar enables clients to present readers useful
information, special deals, and more.

Wide Temp. Panel PC

PPC-090T
l Vortex86DX2 933MHz Fanless
l Up to 1GB of Onboard Memory
l Low Power Consumption
l 2 RS232/422/485 serial ports
l 3 USB 2.0 Host Ports
l 10/100 BaseT Ethernet
l PS/2 KB port, Audio Out
l Compact Flash Slot / SD Slot / SATA
l 9 inch 1024 x 600 WSVGA TFT LCD
l Resistive Touch Screen
l DC-IN +12VDC / +24VDC
l IP65 Front Panel
l Wi-Fi (Optional)
l Giga Ethernet (Optional)
l Wide Temperature (Optional) 3.xx KERNEL

The PPC-090T comes ready to run with the Operating System installed on flash
disk. Apply power and watch the Linux X-Windows desktop user interface appear
on the vivid color LCD. Interact with the PPC-090T using the responsive integrated
touchscreen. Everything works out of the box, allowing you to concentrate on your
application rather than building and configuring device drivers. Just Write-It and
Run-It... Pricing starts at $495 for Qty 1.
http://www.emacinc.com/sales/cc9

Since 1985
OVER
29
YEARS OF

EQUIPMENT MONITOR AND CONTROL

SINGLE BOARD
SOLUTIONS

Phone: (618) 529-4525 · Fax: (618) 457-0110 · www.emacinc.com

 20 CIRCUIT CELLAR • AUGUST 2014 #289

Build a Digital
Pulse Generator
FEATURES

This microcontroller-based, fully digital pulse

generator will be a welcome addition to any
electronics workbench. The device’s accuracy,
stability, and programmability are ideal for use
with many new projects.

By Brian Beard (US)

O ver the years I have designed and

built many of the instruments on my
electronics bench. Until recently, I was using
a pulse generator I built as an undergrad
decades ago. The generator used a cross-
coupled Texas Instruments 74123 dual
retriggerable monostable multivibrator
rotary switch to select the capacitors for the
time constants, and potentiometers for fine
tuning. Unfortunately, it tended to drift and had
a limited nonlinear adjustment range and noisy
switches. I decided it was time to build a new
pulse generator that solved these problems
and would be easy to use.
My design goals were to have a fully digital
pulse generator with a setting range of eight
orders of magnitude, a digital display, and
simple controls—both manual and RS-232. I
decided to stick with a 5-V logic level output
that could be externally translated to other
voltage levels, if necessary.
THE CIRCUIT
Photo 1 shows the mid-level Microchip
Technology PIC16F87 microcontroller I used
to control the pulse generator. The PIC16F87
has 4,096 words of program memory, 368
bytes of RAM, 256 bytes of EEPROM, a UART,
 circuitcellar.com 21

a 16-bit counter, and an internal oscillator.

Figure 1 shows the microcontroller and RS-232
interface schematics. The PIC16F87 oscillator
runs at 4 MHz but the time reference for the
pulse generator is an external 10-MHz crystal
oscillator stable to ±100 ppm or better. The

FEATURES
10-MHz oscillator produces a 100-ns resolution
for the pulse generator; and long pulse lengths
require a long count.
Many chips would be needed to achieve
a pulse range spanning eight orders of
magnitude—at least eight decade counters
plus latches and multiplexors for the high
(Logic-1) and low (Logic-0) count values. It
looked like it would take too many chips for
what was supposed to be a simple project.
Then I found LSI Computer Systems’s
LS7366R 32-bit quadrature counter chip
with SPI. The LS7366R (U6) counts down the
output’s Logic-0 portion and U7 counts down
the Logic-1 portion (see Figure 2).
The chips are programmed to operate in
a non-quadrature single-cycle downcounter
mode. In this mode, the chip counts down
to zero at which point it pulses the DFLAG
output low (for one 10-MHz clock cycle) then
waits until the INDEX input goes low. This
means the minimum count for each chip is
2 (1 and 0) or 200 ns. When the INDEX goes
low, the chip reloads the counter and begins
to count down.
PHOTO 1
Connecting each chip’s DFLAG output to An internal view of the assembled pulse generator
the INDEX input on the other chip ensures
that only one chip is active at any time. When
the Logic-0 downcounter (U6) reaches zero time a setting changes. Once programmed,
it pulses its DFLAG output, which sets the the counters will automatically generate the
output flip-flop (U8b) and reloads the Logic-1 output pulse train.
downcounter. The Logic-1 downcounter (U7) The pulse generator’s digital display is
then counts down to zero at which time provided by a 2 × 16 LCD module. Most small-
it pulses its DFLAG output, which clears character displays require an 8-bit parallel
the output flip-flop (U8b) and reloads the interface and some control lines. However,
Logic-0 downcounter for another cycle. The the PIC16F87 is already using most of its
PIC16F87 has to program the counters each I/O pins for RS-232 and SPI. Fortunately, an

FIGURE 1
The Microchip Technology PIC16F87
microcontroller and RS-232 interface
 BOARDS, BOOKS, DVDs AND MORE AT WWW.ELEKTOR.COM/STORE

Elektor.STORE
The world of electronics
at your fingertips! Bestseller!

W !
E
Learning to Fly With Eagle

N Eagle V6
Getting Started Guide
The book is intended for anyone who wants an
introduction to the capabilities of the CadSoft’s
Eagle PCB design software package. The reader
may be a novice at PCB design or a professional
wanting to learn about Eagle , with the inten-
tion of migrating from another CAD package.
After reading this book while practicing some
of the examples, and completing the projects,
you should feel confident about taking on more
challenging endeavors. This book is supplied
with a free copy of Eagle on CD-ROM for
MS Windows, Linux and Mac.

208 pages • ISBN 978-1-907920-20-2 • $47

ore
This book and m
are available at
m/store
The Ultimate Guide!
www.elektor.co
The LTspice IV Simulator
LTspice IV is one of the most extensive SPICE simulators currently available, with over 1.5 mil- The RPi in Control Applications
Books

lion users worldwide. It is completely free, allowing you to work with an unlimited number of
Raspberry Pi
nodes, components and simulations. So much so, that many of the features on LTspice IV con-
tinue to go unexplored and unexploited and, unfortunately, literature and technical documen-
Hardware Projects
tation are hard to come by. Recently made available in the Elektor Store, The LTspice IV This book is about the Raspberry Pi computer
Simulator book is the ultimate guide to LTspice IV. In a sturdy, hard-cover format, The LTspice and its use in control applications. Dogan
IV Simulator describes the operation of the program, all available commands, the various Ibrahim explains in simple terms, with exam-
editors, dealing with SPICE models, the use of non-linear components and more. Perhaps ples, how to configure the RPi, how to install
the most robust piece of literature on the topic, The LTspice IV Simulator is more than just a and use the Linux operating system, how to
manual. It also offers a variety of tips, methods and examples, all carefully illustrated using write programs using the Python program-
almost 500 drawings, diagrams and screenshots on high-quality paper. The book is designed ming language and how to develop hardware
so that it is suitable for both beginner and veteran SPICE users. based projects.

744 pages • ISBN 978-3-89929-258-9 • $67 290 pages • ISBN 978-1-907920-29-5 • $54

Prices and item descriptions subject to change. E. & O.E

Books

110 Elektor Editions, Over 2500 Articles New!

MIFARE and Contactless Cards in Application
RFID
MIFARE is the most widely used RFID tech-
nology, and this book provides a practical
and comprehensive introduction to it. Among
other things, the initial chapters cover phy-
sical fundamentals, relevant standards, RFID
antenna design, security considerations and
cryptography. The complete design of a read-
er’s hardware and software is described in
detail. The reader’s firmware and the asso-
ciated PC software support programming
using any .NET language.
484 pages • ISBN 978-1-907920-14-1 • $68
DVD Elektor
2000 through 2009
This DVD-ROM contains all circuits and pro-
jects published in Elektor magazine’s year
volumes 2000 through 2009. The 2500+
articles are ordered chronologically by release
date (month/year), and arranged in alphabet-
ical order. A global index allows you to search
specific content across the whole DVD. Every
article is printable using a simple print func
func-
tion. This DVD is packed with ideas, circuits
and projects that are ideal for any electronics
enthusiast, student or professional, regardless
of whether they are at home or elsewhere.
ISBN 978-1-907920-28-8 • $121
Android Breakout Board
The FTDI FT311D is a flexible bridge that
can interface your circuit to an Android
smartphone or tablet. This Elektor Android
Breakout Board offers options for seven
digital outputs, four PWM outputs, asynchro-
nous serial and I2C and SPI interfaces. The
board is compatible with Android 3.1 (Ho-
neycomb) or higher (Android Open Accessory
Mode should be supported).
Ready-built module
Art..# 130516-91 • $41

Elektor is more
than just your favorite
electronics magazine.
It’s your one-stop shop
for Elektor Books,
CDs, DVDs, E-Lock
Kits & Modules The E-Lock chip allows you to connect your
A Highly Practical Guide control system to the ‘Network of Networks’
and much more!
Process Measurements and monitor and control it from anywhere on

with C# Applications www.elektor.com/store or around the globe using your computer, ta-
Kits & Modules

blet or smartphone. This completely assem-

This book introduces PC based measurement
systems and software tools for those needing to
understand the underlying principles or apply
such techniques. C# was chosen to give readers
immediate, practical experience with a pro-
minent programming language. C-Sharp has
a wide support base and is a popular choice for
engineering solutions. The basics of measure-
ment and data capture systems are presented,
and examples of software post-processing.
144 pages • ISBN 978-1-907920-24-0 • $39
bled module is able to establish a secure
connection over the Internet (TCP/IP) using
the Transport Layer Security (TLS) encryption
protocol. The chip also offers a 7-line GPIO
(general purpose input/output), four 16-bit
ADC channels, one 12-bit DAC and one I2C bus
Elektor US that allow control and communication with
111 Founders Plaza, Suite 300 other peripheral devices.
East Hartford, CT 06108
USA Ready-built module
Phone: 860.289.0800
Art.#130280-91 • $150
Fax: 860.461.0450
E-mail: order@elektor.com
 24 CIRCUIT CELLAR • AUGUST 2014 #289
FEATURES
FIGURE 2
This schematic shows the clock reference and pulse generation circuitry in an LSI Computer Systems
LS7366R quadrature counter.
8-bit parallel output port can be connected
to the SPI bus using an STmicroelectronics
74HC4094 8-bit serial-to-parallel shift latch
register with tri-state outputs. The LCD
module interface is reduced to the already
existing SPI bus plus four control lines (see
Figure 3). The LCD’s contrast is controlled
by potentiometer VR1. The LCD module is
illuminated by LED backlights. R8 sets the
current to the LED backlights.
The PIC16F87’s I/O lines were also saved
by implementing a SPI 8-bit parallel input port
with two NAND gates and a Texas Instruments
74HC165 8-bit parallel-input/serial-output
shift register. Figure 4 shows the circuit. The
shift register’s parallel load occurs when the
FIGURE 3
The SPI-to-LCD module
P2SIR_Select signal goes low. The SPI_
Clock signal clocks data out of the shift
register. The inverted output is used because
the shift register data is inverted again by the
open-drain NAND gate (U11c). An open-drain
output is required because SPI_In is a wire-
ORed signal. Four of the eight parallel inputs
are used to read the front-panel switches.
Many of my bench instruments are
powered by an external 9-VDC, supply so I
designed the pulse generator the same way.
Figure 5 shows the Motorola 78S40 universal
switching regulator subsystem designed to
deliver 5 VDC at up to 100 mA.
PULSE GENERATOR OPERATION
Photo 2 shows the manual pulse generator
controls. The Square/Pulse toggle selects the
operating mode. When power is applied with
the toggle in the pulse position, the initial
display will be that shown in Photo 3b. This
display shows that the pulse width (Logic-1)
is 0.5 ms and the pulse spacing (Logic-0) is
0.5 ms. The LCD cursor will be on the first
digit of the time value in the line chosen by
the Line Select switch. The blinking cursor is
moved one digit to the right by pressing the
ADV (digit advance) push button. The cursor
will move from the seconds (left-most) to
the hundreds-of-nanoseconds (right-most)
position as the ADV push button is depressed.
When the cursor is on the right-most digit, the
next ADV depression will move the cursor to
the left-most digit on the same line. Toggling
the Line Select switch will move the cursor to
the same digit position on the other line.
The value under the cursor can be
incremented by pressing the INC push button.
Values wrap around from 9 to 0 without a
carry to the next higher digit. Whenever the
INC push button is depressed the output
waveform (OUT) changes to reflect the values
shown on the display. The other switches all
change the display but do not change the
output waveform. For example, a 1-ms pulse
repeated once a second would be set as 1 =
0001.0000 ms, 0 = 0999.0000 ms.
Photo 3a shows the initial display that
appears when the mode toggle is changed to
Square. This display shows that the period (P)
is 1 ms and the frequency (F) is 1,000 Hz.
The LCD cursor will be on the first digit of the
frequency value. The Line Select switch has
no effect in this mode. The ADV push button
functions the same way it does in the Pulse
mode.
Pressing the INC push button increments
the value under the cursor. Because the
Square mode produces a square wave (50%
duty factor) the minimum change in period
is 200 ns: 100 ns for Logic-1, and 100 ns for
Logic-0. Due to the discrete step size, not
 circuitcellar.com 25

every frequency value is attainable. Every

time the INC push button is depressed the
digital pulse generator will compute the
attainable frequency closest to the new
setting and update the counters and display
to the computed value. The counter values

FEATURES
that determine the output waveform are only
changed when the INC push button is pressed.
Changing from one mode to another does not
alter the output waveform. In fact, changing
modes from Pulse to Square is a convenient
way to see a waveform’s overall period and
pulse repetition frequency.
Consider the previous example of a 1-ms
pulse repeated once a second. In Pulse mode,
this would be set as 1 = 0001.0000 ms, 0
= 0999.0000 ms. Toggling to Square mode
would change the display to P = 01000.0000
ms, F = 0000001.00 Hz without changing the FIGURE 4
This SPI input port reads the front-panel switches.
output.
Host communication mode is entered
by connecting an RS-232 cable between
the pulse generator and a host computer’s
COM port. A terminal program must be
operating on the host to communicate
with the pulse generator. When the pulse
generator senses an RS-232 connection at J2
it tries to synchronize baud rates with the
host. The pulse generator waits to receive a
Return (Enter) character at 1,200 bps. If it
receives any other character (e.g., a return
at another baud rate) it doubles its own
baud rate and waits for another return at
the new baud rate. If it reaches 19,200 bps
without receiving a return it cycles back to
1,200 bps. To synchronize the digital pulse
generator with the host’s baud rate, you
simply press Return (Enter) repeatedly until
the digital pulse generator sends the menu
screen (see Figure 6).
Menu options are selected by typing the FIGURE 5
corresponding single number or character. The power supply circuitry
Any other entries will be ignored and the
menu will be displayed again. In the following
examples the text sent by the pulse generator
is italicized, the text entered by the user is
bold. Time values are entered in microseconds
or milliseconds. Microseconds are specified
by entering the number with a u or U suffix.
Milliseconds are specified with an m or M
suffix. Frequency values are entered in hertz
with an h or H suffix:

0 Set low (0) time—In this example, the

Logic-0 time is set to 9.25 ms. The output
waveform will not be changed until option 4
is selected.
?0
Low time = 9.25 m

1 Set high (1) time—In this example, the PHOTO 2

Logic-1 time is set to 750 µs. The output The pulse generator’s front panel
 26 CIRCUIT CELLAR • AUGUST 2014 #289

frequency can be seen by selecting option

a)
3. The output waveform will not be changed
until option 4 is selected.
?2
Frequency = 2000 h
FEATURES

3 Show settings—This option enables you

to see the specified settings. The settings
shown here are internally consistent but are
not related to other examples shown in this
article.
?3
Low time = 10.0000 ms
b) High time = 0.2499 ms
Period = 10.2499 ms
Frequency = 97.56 Hz

4 Update counters—This option

simultaneously updates the Logic-1 and
Logic-0 counters—or as close to simultaneous
as is possible addressing two SPI chips.
?4 Ok

5 Store powerup settings—This option

writes the current Logic-0 and Logic-1 pulse
settings to the PIC16F87’s internal EEPROM.
When power is applied to the digital pulse
PHOTO 3 generator, it reads these values from the
a—In the square mode display, the P line is period and the F line is frequency. b—In the pulse mode display, EEPROM and uses them to initialize the
the 1 line is Logic-1 time and the 0 line is Logic-0 time. counters.
?5 Ok

waveform will not be changed until option 4 6 RC servo—This option configures the
is selected. output pulse train for control of a radio control
?1 (RC) servo. RC servos rotate in response
High time = 750 u to a variable pulse width signal. The pulse
width begins at 1.5 ms, which corresponds
2 Square wave—In this example, the to the center of rotation. The minimum and
specified frequency is 2 kHz. The pulse maximum pulse widths for this routine are
generator will compute the frequency 0.86 and 2.14 ms, respectively. The Logic-0
closest to the specified frequency. The actual or pulse spacing time is fixed at 28.5 ms.
Pressing the l or L key decreases the pulse
width causing the RC servo to rotate left.
Pressing the r or R key increases the pulse
PROJECT FILES PIC16F87 Microcontroller width causing the RC servo to rotate right.
The decrease/increase step size is 10 ms.
Microchip Technology, Inc. | www.microchip.com
Pressing the escape key returns the Host
menu screen. In the example shown here the
78S40 Universal switching regulator L key was pressed two times:
subsystem ?5
Motorola/ON Semiconductor | www.onsemi.com Enter L, R, ESCape to abort.
1.50
1.49
circuitcellar.com/ccmaterials
74HC4094 Shift latch register 1.48
STmicroelectronics | www.st.com
X Exit—This option terminates the Host
SOURCES Communication mode and returns to the
74123 Rotary switch and 74HC165 shift
LS7366R 32-bit Quadrature Manual Control mode.
register
counter ?x
Texas Instruments, Inc. | www.ti.com Disconnect RS-232 cable.
LSI Computer Systems, Inc. |
C Command line—The command line
www.lsicsi.com
interface can be used in two ways. First, it
 circuitcellar.com 27

ABOUT THE AUTHOR

Brian Beard (bbbeard73@verizon.net) was just starting out as an electrical engineer when the first microprocessors were devel-
oped. He has been an electronics designer and an Assembly-level programmer ever since. Brian has worked for the government,
in industry, and in academia. Outside his full-time job near Silver Spring, MD, he designs project kits (www.lucidtechnologies.
info) and consults on programming and microcontroller systems.

FEATURES
can be used from a terminal program for
FIGURE 6
manual input without displaying the entire Digital Pulse Generator, Firmware 2.03
These are the choices in the menu
main menu. Second, it can function as the 0 Set low (0) time
screen sent by the digital pulse
control interface for a custom computer 1 Set high (1) time generator.
program on the host. 2 Square wave
Each command is specified by a single 3 Show settings
ASCII command character sent to the pulse 4 Update counters
generator following the > prompt. Command 5 Store powerup settings
characters may be lower or upper case, but are 6 RC servo
always echoed as upper case. All characters C Command line
sent to the pulse generator, except (CR), X Exit
are echoed back to the host. For commands ?
that require data (e.g., setting pulse times
or frequency), the data must follow the
command character and be terminated by The digital pulse generator’s accuracy,
a (CR). The acceptable formats for entry stability, and programmability make it ideal
of pulse time and frequency are identical to to simulate timing and stimulus signals when
those used with the main menu. prototyping new projects. The digital pulse
If the character sent to the pulse generator generator program is available on Circuit
following the prompt is not recognized as Cellar’s FTP site.
a valid command character, a (NAK)(CR)
(LF)> sequence wilfdl be sent to the host. If
an error is found in the numeric data following
a command, an error message will be sent
followed by a (NAK)(CR)(LF)> sequence. TRACE 32 ®
The command is ignored whenever a (NAK)
is returned instead of an (ACK).

PULSE GENERATOR FIRMWARE

I used the PIC Assembly language to
write the pulse generator’s firmware. Two
parts of the program were particularly
challenging. The first was parsing the input
string variations possible for milliseconds,
microseconds, and hertz. The second was
creating all the 32-bit math routines—
addition, subtraction, multiplication, division,
BCD-to-binary, and binary-to-BCD—needed
for display and to set the LS7366R counters.

AN IDEAL SOLUTION
The pulse generator has proven to be a
great addition to my electronics workbench. I
used variable RC time constants to set my old
pulse generator’s timing—switches selected
the capacitance and potentiometers set the
resistance. Whenever I needed a precise pulse
train I had to hook up an oscilloscope to
measure the pulse generator’s output, and
then it would drift over time! Thanks to the
totally digital design of the digital pulse
generator I no longer need an oscilloscope to
confirm its output and its settings don’t drift.

advertisement_arm-overview.indd 1 03.06.2014 11:48:51

 28 CIRCUIT CELLAR • AUGUST 2014 #289
Personal Solar Power Setup (Part 2)
Battery Charge Control
FEATURES
The first article in this series introduced a solar power
system that provides reliable backup and reduces utility
power usage. This article focuses on the battery charge
control and options for easily sharing the load between
the utility and solar power.
By Steve Hendrix (US)
PHOTO 1
Solar panels mounted on a south-
facing roof are shown. Note the 4“
gap under the bottom of each. This
enables rainwater, leaf litter, and balls
to pass under the panels rather than
accumulating. The upper brackets on
each panel are sized to hold the panels
at 50º to horizontal. The rows are
spaced to avoid shading one another
even during the winter solstice.
T he first part of this article series discussed
the solar panel characteristics, mounting
details, batteries, and interconnect wiring for
my solar power system (see Photo 1). This
article will provide more detail about charge
control, conversion to useful power, and safely
sharing the load between the commercial
utility power and solar power. Figure 1 shows
an overall block diagram of a typical solar
photovoltaic (PV) system for reference. Part
1 covered most of the top half of this figure;
now I’ll delve deeper into charge control and
the bottom half of the figure.
BATTERY CHARGING
The charge controller can actually
dominate the overall system efficiency. A
carefully designed charge controller can
efficiently match various combinations of
PV panels and batteries. A simple charge
controller combined with mismatched panels
and batteries can result in a lot of wasted
power and excess heat.
The commercial charge controllers I
purchased for my 12-V system are very simple
PWM controllers. During the bulk charge
phase, they simply connect the panels directly
to the batteries. A lead-acid battery acts like a
constant-voltage device. Within limits, it holds
a given voltage at its terminals. On the other
hand, the panels follow a curve that looks
like what is shown in Figure 2. If you simply
connect a panel directly to a battery, the panel
voltage will be forced to match the battery
voltage and the current at that point on the
curve will be delivered to the battery. As the
battery voltage rises, the panel walks up the
curve and the current gradually reduces.
When the voltage reaches the setpoint, the
charge controller goes into PWM, pulsing the
current on and off to hold the battery at the
soak charge setpoint. The effect is to move
the average panel current down to whatever
will maintain the desired battery voltage
while raising the panel voltage.
This particular panel’s maximum power
point (MPP) is about 16 V and 11.5 A, yielding
184 W. When you begin charging with the
battery down to around 11 V, you pull the
panel to 11 V and 12.7 A, yielding 140 W.

Photovoltaic Interconnect/
(PV) panels protection
Utility power
Thus, you give away almost 25% of the
possible power due to the mismatch. In PWM
mode, depending on the required current,
you similarly fall off well away from optimum
efficiency. This system works, but the loss of
efficiency just doesn’t sit right with me.
I have a lot of experience designing
switch-mode power supplies, especially
buck converters. Such converters efficiently
trade voltage for current, or vice-versa, so it
seemed a natural fit here. I’ll provide a quick
review to describe the basic operation.
Some think of a switch-mode converter
in terms of impedance matching as in RF;
another view is to see it as a low-pass filter for
a PWM stream. Both views are useful at times,
but I prefer to think of it in simple mechanical
terms. An inductor acts like a “flywheel” for
current. It produces “inertia” in current flow.
Current cannot change quickly in an inductor,
so you can apply pulses of power to “spin up”
the flywheel and the current flow will remain
reasonably constant.
The basic architecture can step voltage
either up or down, depending on how it’s
Photovoltaic (PV) panel V-I Curve
25 V
20 V
50 W
15 V
10 V
5V
0V
0A 1A 2A 3A 4A
circuitcellar.com 29
FIGURE 1
My personal solar power system
includes photovoltaic (PV) panels,
Interconnect/
Charge control Batteries interconnect/protection, batteries,
protection
wiring, and power.
FEATURES
Interconnect/ Interconnect/ Household wiring
Inverter
protection protection
controlled. Figure 3 shows a basic step-down
architecture. Some specific numbers will help
make the discussion more concrete. Suppose
you have 20 V coming in with 1 A available and
you wish to turn that into 10 V at 2 A. Switch
S1 is turned on for approximately 1 µs to get
the current flowing. Then it turns off for 1 µs
before repeating the cycle. Current during
the “on” time is 2 A, but the average is 1 A
due to the 50% duty cycle. But wait—current
cannot change instantly, so the current must
keep flowing. Two amps of current × the
(theoretically) infinite resistance of the switch
yields a whole bunch of volts.
In practice, this would produce arcing
or destroy the switch. Adding D1 enables
the current to be pulled from ground during
the off time. During the 50% off time, 2 A is
drawn through D1. C1 smooths out the current
bursts so the solar panels see a relatively
steady load. Thus the inductor supplies 2 A
to the load continuously at 10 V, while the
supply sees an average load of 1 A at 20 V. C2
smoothes out any voltage alterations caused
by changes in the load current relative to
200 W
100 W 150 W
FIGURE 2
The voltage-current curve for a 200-W
panel is shown. Note the open circuit
(0-A) voltage is about 21 V, the short
circuit (0-V) current is about 13 A, and
the maximum power is near 11.5 A
and 16 V. The concave curve family
5A 6A 7A 8A 9A 10 A 11 A 12 A 13 A 14 A marks lines of constant power at
50–200 W.
 30 CIRCUIT CELLAR • AUGUST 2014 #289
FEATURES
FIGURE 3
A simplified switch-mode controller uses an inductor as a “flywheel” and switches the input power at less
than 100% of the duty cycle. The result is that the output current is higher than the average input current
while the voltage is lower. Modifications of the switching and control arrangements can result in higher,
lower, or even opposite sign output voltages.
the inductor current. The feedback path in
the diagram adjusts the switching control to
maintain a particular desired voltage.
None of these components are ideal, so
there are losses. However, efficiencies of 95%
or better are feasible with careful design and
good components. In particular, D1 can be
replaced with a switch for better efficiency
but more complexity. You can modify the
switching and control arrangements to
produce a higher voltage at a lower current
than the input or a negative output voltage.
Converters are available as ICs, including
S1 as an internal transistor. For higher
power, controller chips enable you to add a
big external transistor, often with a heatsink,
so you can control substantial amounts of
current. I considered using such a chip.
The control logic in these chips is designed
to maintain a constant output voltage,
without regard to input parameters. Instead,
you want to control the conversion to hold the
solar panel at the knee of its curve (a.k.a., the
MPP). This feature is commonly known as MPP
tracking (MPPT) in many available controllers.
This type of control can be done in
firmware. I used the PWM hardware built
into a Microchip Technology PIC18F97J60
microcontroller to attempt to build this kind of
switch-mode controller. My grand ambitions
were a unit with four channels, each of which
accepts up to 15 A between 12 and 100 V and
converts that efficiently to charge 12-, 24-,
or 48-V batteries. Control logic would sense
voltage and current on the solar panel side
as well as on the battery side, expanding the
feedback path shown in Figure 3.
Most of the time, I want to pull maximum
available power from the solar panels. Control
for each channel is separate and updated in
round-robin fashion in the firmware. I simply
dither the duty cycle while monitoring the
power provided by the panels. When a dither
in one direction (increase or decrease in duty
cycle) produced a power gain in the battery,
I kept the new operating point and continued
dithering about that. If the movement
decreased the power to the batteries, I moved
back to the previous operating point.
However, the real world intruded rudely
upon my grand plans. The inductor must
conduct the full battery current essentially
continuously, so the wire in its windings must
be pretty heavy. At the same time, with a
large voltage difference between input and
output, a high inductance is needed to keep
the current running smoothly, which requires
a lot of turns. The required inductor becomes
pretty big and expensive.
The required inductance can be reduced by
increasing the switching frequency. Then the
non-ideal characteristics of switch (S1) rear
up. The best switch here is a big field-effect
transistor (FET). The FET must be physically
big to handle the amount of current and to
dissipate the resulting heat. The physical size
increases the amount of stored charge that
must be pushed into or out of the gate to
switch from on to off. Regardless of the drive
arrangement, it takes some nonzero time to
pump charge into or out of the gate. During
this transition time, the FET must spend a
measurable amount of time in an unsaturated
state This “partially on” period results in
considerable heat in the FET.
To see why, consider the “off” case.
Suppose you have a panel with a 60-V open
circuit voltage and a 10-A short circuit
current. With the switch off, the current is
zero (60 V × 0 A = 0 W). Now consider the
“on” case. Real FETs come remarkably close
to the zero resistance ideal. The ones I chose
claim a maximum 65 mΩ when saturated.
Passing 10 A, 6.5 W of heat is generated, but
only during the “on” part of the cycle. With
an appropriate heatsink, they are rated for
nearly 300 W, so there appears to be plenty
of margin.
But wait—let’s see what happens during
switching. Maximum power dissipation occurs
at the “halfway” point of the switching, when
there is 150 W (i.e., 30 V x 5 A). With a basic
driver circuit, you can get through this linear
zone in around 100 µs. So this is a trivial
amount of energy—if you only switch once.
However, if you want to switch at 5 kHz, you
actually end up spending most of your time in
the linear region, giving away nearly 150 W
as heat that must be dissipated, and not as
power going into the batteries.
Here is the fundamental tug-of-war
between the inductor, which is pulling
for a faster switching frequency and
the switch, which is pulling for a slower
switching frequency. My solution ultimately
compromised: I used three inductors in
series to get sufficient inductance along with
a beefed-up driver circuit to get through the
 EXPLORE
C-RUN FOR ARM
Runtime Analysis Simplified

C-RUN is a high-performance runtime analysis add-on

product, fully integrated with world-leading
C/C++ compiler and debugger tool suite
IAR Embedded Workbench.

C-RUN performs runtime analysis by monitoring application

execution directly within the development environment.
The tight integration with IAR Embedded Workbench
improves development workflow and provides each developer
with access to runtime analysis that is easy-to-use.

www.iar.com/crun
 32 CIRCUIT CELLAR • AUGUST 2014 #289
FEATURES
ABOUT THE AUTHOR
Engineering consultant
Steve Hendrix (steve
hx@hxengineering.com),
CFIAI, lives with his wife
Kathy and five children
near Cuyahoga Valley
National Park in Ohio
and runs Hx Engineer-
ing, LLC. Steve has de-
signed software and
hardware for a wide
range of projects: Space
Station Freedom, vibra-
tion analysis, IEEE-488
bus hardware, biomed-
ical sensing, industrial
instrumentation, photo-
therapy equipment, and
safety gear for facility
personnel.
linear region more quickly. I also reduced
the allowable difference between the input
and the output voltage to reduce the cost of
building the controller.
One hazard is often overlooked when
designing a switch-mode controller for low
current and voltage, but is nontrivial at
these power levels. Being the paranoid sort,
I had a circuit breaker between the controller
output and the battery. My intent was to be
able to feed 20 A per channel to the battery,
so I installed 20-A circuit breakers. I ran
out of those and installed a 15-A breaker
on one channel knowing that in my early
experiments I would only achieve about 10 A
anyway. However, one sunny day I was offsite
but monitoring the controller via its Ethernet
interface from a work site. The controller was
working superbly! I was seeing the current
ramp up to nearly 20 A per channel as the sun
moved into a good position. Then suddenly
there was no response.
Later investigation showed that the circuit
breaker had tripped. Remember that the
inductor will not permit current to change
instantaneously. Question: What voltage do
you get with 20 A flowing through the now-
open circuit breaker? Answer: A whole bunch!
Although I had foreseen this scenario and
provided a surge suppressor on the output,
it was not up to the task. The resulting
high voltage went pretty much anywhere it
wanted and destroyed the board. Any further
experiments in this direction will include
much more substantial surge protection!
My next controller iteration was less
ambitious, with eight channels but no
attempt at voltage conversion. Simple on/off
switching works to track the correct charging
voltages, but I have retained the ability to
monitor operation via the Ethernet. If you
are interested, e-mail me and I can provide
access to the controller on the Internet. As I
wrote this article, I was typing offsite, but still
monitoring the bright autumn sun via Internet.
The controller I designed and am currently
using is on a 3“ x 5” PCB with eight channels
that can handle up to 100 V in and out and up
to 22 A per channel. A Microchip Technology
PIC18F97J60 microcontroller is at the heart
of the controller. It provides an Ethernet
interface, 16 channels of analog inputs, and
a multitude of other I/O. Each channel is
individually programmable as to the battery
type and voltage (e.g., lead-acid 12 V, lead-
acid 48 V, etc.). I modified a sample application
provided by Microchip Technology to yield an
HTML page showing each channel’s voltage,
current, and charging status. I also built in the
ability to eavesdrop on an RS-485 or RS-232
line so I could pick up and display information
from my inverter, which I will describe later.
So far the webpages are display-only,
although I hope to have a password-protected
ability to adjust configuration via a webpage
soon. For the time being, there is also a Telnet
interface that enables me to not only adjust
all configurable items, but also to capture a
running text file with historical data for offline
analysis. I hope to add live graphs of current
and voltage and so forth in the future.
In summary of charging, my goal was to
have the solar panels operate at—or at least
close—to their MPP while efficiently delivering
the resulting energy into the batteries.
However, I also wanted the charge controller
to limit the delivered power to prevent
overcharging. For typical lead-acid batteries,
this requires three phases as described in
Part 1 of this article series: a bulk charge,
where all the current that can be mustered
is delivered, terminating when the battery
voltage rises to a predetermined level; a soak
charge, where the battery is held at that
voltage; and finally a float charge, where the
battery is held at some slightly lower voltage
while delivering whatever current is required
by other loads.
My system has seen three different types
of controllers. The 12-V system uses a simple
PWM architecture to load the panels down to
the battery voltage and to reduce current as
needed. The switch-mode controller is more
elaborate and runs the panels at maximum
power while delivering just the right voltage to
maximize the battery’s charge rate. The most
recent controller could be considered a very
slow PWM controller. It was also designed as a
part of the overall system, wherein the panels
are inherently matched to the batteries being
charged.
INVERTER
All that work up to now has resulted in a
lot of energy stored in batteries. So how do
you get it out and make it useful? If you’re
in a boat or motor home, you may have 12-V
loads and can use it directly. As in Part 1, I
cautioned you to make any connections to
batteries through an appropriate protective
device. Even a quick accidental touch on a
wire that can deliver a few thousand amps
can ruin your day.
In the home, 12-V loads are less common.
Back when high-brightness LEDs first came
out, I designed a PCB to use 12 V and run five
or six LEDs at a constant current. Today you
can just go to your local electronics store and
purchase them, of course. I installed them
at the top and bottom of the stairs, outside
the bedrooms, outside the bathrooms, and
so forth. Those have been off only twice in
the past dozen years or so, when I made
upgrades to the system. And I’m currently
 Easy and Quickly Mountable on 68-pin IC socket
Stamp size FPGA/CPLD Module PLCC68 Series
PLCC68 series is a very small size FPGA & CPLD Module designed to be
used with 68-pin IC socket. It's very convenient to use on universal
boards.
PLCC68 series requires only single 3.3V power supply.
The pin assignments are the same for all series. I/O power supplies are
separated in two banks (A and B, 25 I/Os each).
50 I/Os Very small size (25.3 x 25.3 [mm])
(External clock inputs are available) Separated power supply: Core, I/O
3.3V single power supply operation JTAG Signals (TCK, TMS, TDI, TDO)
Voltage converters for auxiliary All series have the same pin assignment
power supply Mountable on IC socket
[USAGE EXAMPLE]

ALTERA Series
Cyclone V PLCC68 FPGA Module
AP68-06 Series
FPGA : 5CEBA2U15C8N
5CEBA4U15C8N
FRAM
3.3 V single power supply operation
On-board oscillator, 50MHz
Configuration Device
8 Layer PCB
Mountable on 68pin IC socket
Cyclone V PLCC68 FPGA Module
AP68-06Z Series
FPGA : 5CEBA2U15C8N
5CEBA4U15C8N
3.3V single power supply operation
On-board oscillator, 50MHz
Configuration Device
8 Layers PCB
Board size : 25.3 x 25.3[mm]
RoHS compliant Mountable on 68pin IC socket
Cyclone III PLCC68 FPGA Module
AP68-04 Series
FPGA : EP3C25U256C8N
3.3V single power supply operation
On-board oscillator, 50MHz
Configuration Device
8 Layers PCB
Board size : 25.3 x 25.3[mm]
Mountable on 68pin IC socket
RoHS compliant RoHS compliant

Cyclone III PLCC68 FPGA Module MAX V PLCC68 CPLD Module MAX II PLCC68 CPLD Module
AP68-03 Series AP68-02 Series AP68-01 Series
FPGA : EP3C10U256C8N CPLD : 5M570ZF256C5N CPLD : EPM240F100C5N
3.3V single power supply operation 3.3V single power supply operation EPM570F100C5N
On-board oscillator, 50MHz Board size : 25.3 x 25.3[mm] 3.3V single power supply operation
Configuration Device Mountable on 68pin IC socket Board size : 25.3 x 25.3[mm]
8 Layers PCB 4 Layers PCB Mountable on 68pin IC socket
Board size : 25.3 x 25.3[mm] 4 Layers PCB
Mountable on 68pin IC socket
RoHS compliant RoHS compliant RoHS compliant

XILINX Series
Spartan-6 PLCC68 FPGA Module Spartan-3AN PLCC68 FPGA Module Spartan-6 PLCC68 FPGA Module
XP68-03 Series XP68-02 Series XP68-01 Series
FPGA : XC6SLX45-2CSG324C FPGA : XC3S200AN-4FTG256C FPGA : XC6SLX16-2CSG225C
3.3V single power supply operation 3.3V single power supply operation 3.3V single power supply operation
on-board oscillator, 50MHz on-board oscillator, 50MHz on-board oscillator, 50MHz
Configuration Device Board size : 25.3 x 25.3[mm] Configuration Device
Board size : 25.3 x 25.3[mm] 6 Layers PCB Board size : 25.3 x 25.3[mm]
6 Layers PCB Mountable on 68pin IC socket 6 Layers PCB
Mountable on 68pin IC socket Mountable on 68pin IC socket
RoHS compliant RoHS compliant RoHS compliant

We also have many other products Universal board for AP/ XP PLCC68 Module
All stocked items are ready to be JTAG-BUFFER Circuit / JTAG Connector
shipped immediately ZKB-092 Power input :
DC5V/2.1[mm] Jack / Terminal Block
RoHS compliant 3.3V Power Circuit
Technical documents are available for free before purchasing Power Switch / Power LED
Board size : 4.528"x 6.102" (115 x 155 mm)

See all our products, A/D D/A conversion board,

HuMANDATA LTD. boards with USB chip from FTDI and accessories at :
TEL：+81-72-620-2002（Japanese）FAX：+81-72-620-2003（Japanese/English）
E-Mail : s2@hdl.co.jp URL : http://www2.hdl.co.jp/en/ www.hdl.co.jp/CC/
 34 CIRCUIT CELLAR • AUGUST 2014 #289

building a set of LED-based window candles
to run from 12 V. I want to mount them so
I can leave them in place year-round in the
windows. They too will run directly from 12 V,
although I will have a switch for them.
What about common household appliances?
FEATURES
sinusoidal output. For example, a DC motor
could turn an alternator to produce AC. As
a bonus, it could provide complete galvanic
isolation between the DC and the AC sides.
However, such a scheme has many drawbacks
(e.g., wear of moving parts, maintenance

PROJECT FILES
circuitcellar.com/ccmaterials
My goals were to “go green” by taking some
loads off the grid and to have reliable backup
power to essential loads for outages. I grew
up in a house with a sump pump and that
was always the first “emergency” when the
power went out; all the more so because it
was usually during a storm with torrential
rains. Thus, when we built our current house,
I insisted on downhill footer drains (i.e., no
sump pump!)
For an outage of more than 1 or 2 h, my
next priorities were the refrigerator and deep
freezer, which require a solid 120 VAC at 60 Hz
and a clean sine wave. (Note: My supplemental
write up on “Experiences with Inverters” is
available on Circuit Cellar’s FTP site.)
The least expensive inverters, which are
typically rated for only a few hundred watts,
produce a square wave at 60 Hz. They use a
standard up-switcher to generate 120 VDC
and then alternate its polarity to produce a
60-Hz square wave. This output works, but it
is rich in harmonic content. The sharp edges
and harmonic content tend to make motors
buzz, generate excessive heat, and generally
sound unhappy.
Next up were the “modified sine wave”
inverters. These were only slightly better,
creating something that more closely
approximates a sine wave by producing a
higher DC voltage, but switching it positive,
then off, then negative, then off, and then
starting over. The result looks like a square
wave with an extra step every time it passes
through zero. It is better and relatively
inexpensive, but it is still not suitable for most
motorized devices.
Finally there are “true sine wave” inverters.
These are more complicated (and expensive),
but produce a waveform very close to a perfect
sinusoid and essentially indistinguishable
from good-quality commercial power. There
are several ways to convert a DC input to a
SOURCES
MS-PAE Series inverter/charger and ME-ARC
remote control
Magnum Energy, Inc. | www.magnumenergy.
com
PIC18F97J60 Microcontroller
Microchip Technology, Inc. | www.microchip.
com
requirements, and noise), so it is typically
used only in specialized situations.
A commercial inverter usually takes in 12,
24, or 48 VDC. Thus the desired output voltage
goes above and below the input voltage and
half the time is opposite polarity. Adding a few
more switches to Figure 3, along with more
complex control logic, enables this topology
to produce a voltage that is above, below, or
of opposite polarity to the input. You can then
command a target voltage that tracks the
desired waveform. With many small steps per
cycle, it’s relatively easy to filter the “jaggies”
and produce a nice clean sinusoid that should
make any AC device happy. This waveform
comes with no moving parts.
MIXING COMMERCIAL AND
SOLAR POWER
You must pay attention to safety issues
when mixing solar power and commercial
power under one roof. Building and electrical
codes vary from place to place, so please do
your own due diligence in investigating the
local rules or taking advice from a local expert.
You probably don’t want to try to power your
entire neighborhood. More importantly, if
utility power goes out, that poor utility worker
swaying on the pole up there in the driving
rain has every reason to believe that there’s
no power there. You do not want to create a
safety issue for the utility worker!
The simplest way to guarantee separation
of the two systems is physical. You can run new
wiring for 120-/240-VAC loads or you can pull
a wire out of the entrance panel and move the
entire existing circuit to a new panel installed
expressly for solar power. This approach is
clean and safe. It has the drawback of being
a bit more work to transfer that circuit back
to commercial power in case of a failure
anywhere in your solar power setup.
My approach was a little more elaborate.
I ran a couple of separate circuits that power
certain specific loads and installed a transfer
switch per circuit that enables me to choose
solar or utility power for each circuit at the
flip of a switch in the utility room. Whatever
transfer switch you use, it must meet three
specific criteria, in addition to anything else
required for your local building codes: First,
it must be rated for the voltage being used.
Second, it must be rated for the load current
or higher. Third, and most importantly, it
must be the break-before-make type. As
the switch is thrown, it must break one

contact before it connects the other. There
must never be a connection, even during
transition, between the switch’s two arms.
I personally use an ohmmeter to check this
before installing a given switch. Also note
that electrical codes in most places forbid
any device that breaks the neutral conductor,
which must run continuously at all times back
to the entrance panel. You can purchase an
automatic transfer box that will automatically
switch over to backup power, but with that
approach I couldn’t tinker with it!
You can also purchase an inverter rated
for “grid-tie” operation intended to be wired
directly into your entire home’s wiring and it
will synchronize itself to the grid. Whatever
power is produced by the inverter is subtracted
from that coming from utility power. If your
load is light and the inverter is producing more
than you are using, it will feed power onto the
grid, in theory running your meter backward.
You may need to make special arrangements
with your electric company for this type of
operation, and you may or may not be able
to actually have the power company pay you
for electricity, but you can certainly reduce
your own electric bill this way. A significant
drawback of this type, for my purposes, is
that the inverter automatically shuts off upon
loss of utility power. It must do so to meet the
safety requirements (remember that utility
worker out there climbing the pole?). This in
turn means the inverter cannot supply backup
power during an outage. Some models may
permit some sort of override, but I was not
able to find one.
My big Magnum Energy MS-PAE Series
48-V inverter has several clever features to
simplify the entire issue of utility power vs
solar. Lest I sound like an advertisement here,
I will point out that I have no relationship with
the manufacturer other than as a satisfied
customer! This inverter has served me well,
living up to its full ratings on a few occasions
when I accidentally applied more simultaneous
loads than I intended. It handled everything
gracefully without complaint, even when the
air conditioner, refrigerator, and deep freezer
were already running when the well pump
kicked on with its big surge load.
In addition to the full 120-/240-VAC output,
this inverter also features a 120-/240-VAC
input. Upon defined conditions, the inverter
can switch from using the 48-VDC battery
power to using utility or generator power. The
switch can be conditioned upon either time
or battery voltage. When it switches to utility
power, it uses a relay to feed utility power
directly through to the loads with negligible
losses. In addition, it can use whatever
power is left over, up to a predefined limit, to
recharge the batteries and can automatically
circuitcellar.com 35
switch back to battery power when enough
is available. I have it wired into its own 120-
/240-VAC 30-A circuit and I’ve programmed it
to use up to 25 A. If the battery voltage drops
below as 46 V, it cuts over to utility power,
drawing 25 A and using anything not needed
FEATURES
for the loads to recharge the batteries.
When the batteries rise to 50 V through any
combination of solar and utility power, it
returns to normal inverter operation.
For the worst-case scenario of a winter
storm that knocks out utility power for several
days while burying the solar panels under a
couple feet of snow, I still have that generator
in the garage. It will feed into the inverter
as “utility” power and can power essential
loads. It will continuously run at an efficient
full-power load, recharging the batteries as
power is available. Therefore, I can have
continuous power to the essential loads while
running the generator for only an hour or two
per day, which greatly reduces noise, odor,
and fuel consumption.
This unit also has an optional Magnum
Energy ME-ARC remote control head that
came with enough cable to just comfortably
reach my desk, so I can monitor its operation
and make adjustments without even getting
up. The remote head uses a standard RS-
485 protocol. I contacted Magnum Energy
with little hope of actually getting any
information on a proprietary protocol. Much
to my surprise, the company responded with
a full technical document I was able to use.
My charge controller is designed to eavesdrop
on the RS-485 protocol and extract complete
information on battery current and voltage,
load current and voltage, and utility current
and voltage, as well as several internal
temperatures. My charge controller then
summarizes that information and presents it
as part of a webpage.
BUILD IT YOURSELF
This two-part article has covered all
aspects of implementing a practical solar PV
power system for a home or small business.
Length restrictions have limited the amount
of detail I could include, but I hope you
now know what questions to ask yourself
and have some considerations for potential
vendors as you contemplate building your
own system.
I’m often asked how much this setup
actually reduces my electric bill. I always
answer “not enough to pay for the system for
a very long time,” at today’s utility rates.
However rates are moving inexorably upward
and I can feel that I’m doing something to
reduce fossil fuel usage. Besides, when the
power goes off for an extended time, the
system will pay for itself very quickly!
 36 CIRCUIT CELLAR • AUGUST 2014 #289

Hands-On Electrocardiography
FEATURES

This microcontroller-based electrocardiography (ECG) system enables you to view your heart’s
waveform on a mini TV. The system includes a circuit, hardware, and a voltage limiter.

By Sean Hubber and Crystal Lu (US)

W e designed
electrocardiography
and
(ECG)
built
system
that extracted the heart’s electrical output.
This hardware combined with a MathWorks
MATLAB function can be used to accurately
measure a person’s heart rate. From this
starting point, we used an Arduino Due
microcontroller board to develop a battery-
operated packaged product that provides
you with a safe and easy way to view your
heart’s waveform on a mini-TV screen (see
Photo 1). The project was done as part of
Cornell University’s Electrical and Computer
Engineering 5030 Bioinstrumentation course.
This article details the underlying circuitry,
the MATLAB function we used, and how the
Arduino and mini-TV were used to display
the waveform. Figure 1 shows the overall
system’s main components.
THE CIRCUIT
Figure 2 and Photo 2 show the system’s
main circuit, which is separated into several
stages to ensure that retrieving the signal
would be user-safe and that sufficient
an
amplification could be made to produce a
readable ECG signal. The first stage is the
conditioning stage, which ensures user safety
through DC isolation by initially connecting
the dry electrode signals directly to capacitors
and resistors. The capacitors help with DC
isolation and provide a DC offset correction
while the resistors limit the current passing
through. This input-conditioning stage is
followed by amplification and filtering that
yields an output with a high signal-to-noise
ratio (SNR). After the circuit block, the signal
is used by MATLAB and voltage limiter blocks.
Directly after DC isolation, the signal is sent
into a Texas Instruments INA116 differential
amplifier and, with a 1-kΩ RG value, an initial
gain of 51 is obtained. The INA116 has a
low bias current, which permits the high-
impedance signal source. The differential
amplifier also utilizes a feedback loop, which
prevents it from saturating.
Following the differentiation stage, the
signal is passed through multiple filters and
receives additional amplification. The first is
a low-pass filter with an approximately 16-Hz
 circuitcellar.com 37

cutoff frequency. This filter is primarily used

to eliminate 60-Hz noise. The second filter is
a high-pass filter with an approximately 0.5-
Hz cutoff frequency. This filter is mostly used
to eliminate DC offset. The total amplification
at this stage is 10. Since the noise was

FEATURES
significantly reduced and the SNR was large,
this amplification produced a very strong
and clear signal. With these stages done, the
signal was then strong enough to be digitally
analyzed. The signal could then travel to both
the MATLAB and voltage limiter blocks.

MATLAB
To use MATLAB on the signal produced by
the circuit you must first connect the device to
the computer. The packaged product contains
a phone jack that is connected to the circuit
block’s output. We used this jack to feed
the signal to the computer using a National
Instruments NI USB-6008/6009 USB device.
Once the device is connected the MATLAB
program can be used to analyze the signal.
The program starts by running a few
initialization steps before entering the data PHOTO 1
reading and display section. These initialization The black box contains the circuit and microcontroller. The mini-TV is on top, the bars on the sides are used
steps include initializing variables used in the to hold it and read the hand voltage. The power switch is on the front along with the output jack.
program, defining the input device, setting
up the GUI, and initializing the plots and data
collection.
Most of the variables are set to values
that produce the best results, while others User’s
Circuit
Voltage
Arduino Due
hands limiter
are just initialized to 0 or empty vectors. It is
important to note that the bpmstaticprev
(1/s between the previous two beats) and
bpmstring (location in time of all previous MATLAB
Mini-TV
display
beats) variables are initialized to 80. This tells
the program that initially heart rates near 80
are reasonable. If the subject has a heart rate
much less than or much greater than 80, the
program will readjust accordingly. FIGURE 1
Once the GUI is initialized, the plots of the The electrocardiography (ECG) system includes a circuit, an Arduino Due microcontroller board, a mini-TV
ECG waveform, the Pan-Tompkins plot, and display, MATLAB programming language, and a voltage limiter.

FIGURE 2
This schematic shows the conditioning
stage, the differential amplifier, and
the filtering stages.
 38 CIRCUIT CELLAR • AUGUST 2014 #289

the first plot. After this, the program begins

collecting and displaying data and results.
Until the user presses the Quit button, the
program continuously loops through a while
loop collecting and displaying data.
There are several important steps in the
FEATURES

loop. The first step is sampling the data from

PHOTO 2
The circuit block was built on a board
designed in ExpressPCB.
the threshold line are initialized. The Pan-
Tompkins plot is a processed version of the
original signal that makes it much easier to
obtain a threshold. In the last step before
the data collection and the display begin, the
program begins continuously sampling the
input device. The program initially pauses for
1 s so enough data can be collected to build
the input device. Once this sample is captured
it is placed at the end of the vector of values
to be displayed. This vector is then shifted
to the left, dropping the number of values
equal to the size of the samples that were at
the front. The vectorized signal then passes
through a low- and a high-pass filter. These
filters’ cutoff frequencies can be set in the
GUI. The filtering process produces transients,
which are removed so they do not interfere
with beats per minute (BPM) calculations.
Finally, the vector is multiplied by 10 times
the absolute value of itself, which is a manual
amplification. This amplification makes the
signal more visible on the plot, which makes
it easier to do threshold adjustments.
To determine the location in time of the
beats, the program takes the processed signal
and runs the Pan-Tompkins algorithm on it.
This requires two simple steps. The first step
is to square the signal so it consists of only
positive values. The second step is to convolve
it with a rectangle. The result of this process
is a much smoother waveform that can be
used to determine each of the heart beats’
locations. The user sets the threshold to find
each heart beat’s rising edge. It is important
that this threshold is high enough so that only
the rising edges of real heart beats will cross
it, but low enough so that “weaker” beats will
still reach it. Once the heart beats’ locations
were determined, we ran them through our
heart rate calculating algorithm to determine
the heart rate in BPM.
The heart rate calculating algorithm was
by far the most difficult piece of the program
to write. The algorithm was run only once
every 0.5 s or every fifth time through the
while loop. The first step was to compute
bpmstatic from the period of the two most

ABOUT THE AUTHORS

Sean Hubber (sjh234@cornell.edu) is from Missoula, MT. He studied Electrical and Computer En-
gineering at Cornell University and graduated in 2014. In Septemeber 2014 he plans to begin a
career in computer science as a software developer with Epic Systems in Madison, WI.

Crystal Lu (csl84@cornell.edu) graduated from Cornell University with an ME in Systems Engi-

neering (2014) and a BS in Electrical and Computer Engineering (2013). She works at Orbital
Sciences Corporation in Dulles, VA, as a systems engineer in the Fault Detection, Isolation, and
Recovery group for the Cargo Resupply Mission. Her interests include satellite systems and con-
sumer electronics.
 Get ready for

Buy now
and
upgrade for free!

Valid for new license purchases only.

Terms & Conditions: www.cadsoftusa.com/conditions-freeupgrade

w w w. c a ds o f t u s a . co m
 40 CIRCUIT CELLAR • AUGUST 2014 #289

recent beats. Once this was determined,

a) it was compared to the most recent valid
heart rate calculation, which was stored as
bpmstring(end). If it was near enough to
this value (heart rates do not change rapidly)
then it was considered valid and added to the
FEATURES

bpmstring vector, which contains all “valid”

heart rate readings. If the new bpmstatic
value was not close enough to the previous
value then the last value in bpmstring was
taken as the heart rate and added back to the
bpmstring vector. Finally, if the BPM reading
was outside the range [30,180], it was tossed.
Once a “valid” heart rate was determined
and added to the bpmstring vector, the
heart rate’s output value was calculated by
averaging the last 20 values in bpmstring.
The number of values used can be increased
to provide a more stable reading or decreased
to produce a reading more responsive to
rapid heart rate changes. Finally, the heart
rate reading is displayed on the GUI.
b) Once the heart rate has been determined,
the program plots the updated versions of the
ECG waveform, the Pan-Tompkins waveform,
and the threshold waveform. The user can
suppress this update by pressing the Pause
button. The real-time display can be resumed
by pressing the Pause button again. The
program then updates the current time
(rounded to the nearest second) during which
the program has been running.
The filter adjustments built into the GUI
enable the user to customize the filter cutoffs.
Photo 3a shows the MATLAB display with no
filter adjustment. Photo 3b shows the signal
from the same person after adjusting the
filters.
Finally, the index is increased by the size
of the sample so timekeeping is consistent.
There is a pause so each loop takes exactly
100 ms (this value is determined by the
realtime variable, which also determines
PHOTO 3
the sample size). The program will continue
a—This is the MathWorks’s MATLAB display with no filter adjustment. The red line shows the filtered
waveform signal. The blue line shows the Pan-Tompkins waveform. The dotted teal line shows the threshold.
looping through this process until the user
b—After adjusting the filter cutoff frequencies the signal is much cleaner. presses the Quit button, at which time the
program exits the loop, closes all figures, and
deletes the input device.

THE VOLTAGE LIMITER

The circuit block’s output can have values
anywhere between –9 and 9 V. The Arduino
Due only accepts a limited range of inputs.
After rigorous testing, we determined that
the Arduino Due would crash when given an
FIGURE 3
1.5 V input voltage outside the 0.5-to-1.5-V range.
Two diodes are used to limit voltage.
The voltage limiter uses two diodes to limit
Differential the voltage, one leads to a 1.5-V node, the
Arduino
Amplifier
microcontroller other leads from a 0.5-V node (see Figure 3).
Circuit
This prevents any voltages above 1.5 or below
0.5 V 0.5 V from reaching the Arduino Due. The
circuit’s resistor limits the current that is
 circuitcellar.com 41

seen by the Arduino Due. Following the block

diagram, the voltage limiter block’s output Due pin 36 -> 3k3 resistor -> Video In
becomes an input to the Arduino Due block. Due pin 37 -> 1k6 resistor -> Video In
Due pin 38 -> 820R resistor -> Video In
THE ARDUINO DUE Due pin 39 -> 390R resistor -> Video In
We used the Arduino Due to convert the Due pin 40 -> 200R resistor -> Video In

FEATURES
heartbeat waveform to an NTSC signal that Due pin 41 -> 100R resistor -> Video In
could be used by a mini-TV. The Arduino Due Due pin GND -> Video GND
continuously sampled the input provided
by the voltage limiter at 240 sps. Similar to
MATLAB, the vectorized signal was shifted FIGURE4
left to make room at the end for the most This is the connection between the Arduino Due and the mini TV.
recent sample. This provided a continuous
real-time display of the incoming signal.
Each frame outputted to the mini-TV contains
two waveforms. One has a 1-s screen width
and the other has a 5-s screen width. This
enables the user to see a standard version (5
s) and a more zoomed in version (1 s). Each
frame also contains an integer representing
the program’s elapsed time. This code was
produced by Cornell University professor
Bruce Land.

MINI-TV DISPLAY
Finally we’ve reached the end of the block
diagram. Here, the mini TV receives a signal
from the Arduino that tells it what to display.
We attached the Arduino connected to the
NTSC TV in accordance with the Arduino Due’s
video graphics array (VGA) library read me PHOTO 4
(see Figure 4). The resistors are connected to Here is the device in action. The top line (green) has a 1-s time base. The bottom line (yellow) has a 5-s
Video In via a 100-µF capacitor that provides time base.
AC coupling.

TAKE HEART
The final product of this project is a mobile
display of one’s heartbeat (see Photo 4). The
system is packaged in a box. A switch on the PROJECT FILES SOURCES
bottom is used to turn the system on and Arduino Due
off. The mini-TV is placed on the top of the
Arduino | http://arduino.cc
box with the handlebars on either side. This
enables you to easily switch the device on and
then place your hands over the bars to see ExpressPCB software
your heart reading on the screen. ExpressPCB | www.expresspcb.com
Because of the limited voltage range on
the input and substantial baseline wander in
most peoples’ signals, we were unable to circuitcellar.com/ccmaterials MATLAB programming language
accurately compute a person’s heart rate on MathWorks, Inc. | www.mathworks.com
the Arduino Due. To compensate for this, the
RESOURCES
final product contains an output port that is
B. Land, “Using the Arduino NI USB-6008/6009 Bus-powered
connected to the output of the differential
DUE,” Cornell University School multifunction DAQ USB device
amplification circuit, before it is voltage
limited. This port can be used to sample the of Electrical and Computer National Instruments Corp. | www.ni.com
signal into MATLAB, where a program we Engineering 5030, 2013.
wrote (which is available on Circuit Cellar’s INA116 Differential amplifier
FTP site) can accurately compute BPM and the S. Hubber and C. Lu., “Hands
user can further view and manipulate the Texas Instruments, Inc. | www.ti.com
On ECG,” Cornell University
signal. School of Electrical and
Computer Engineering 5030,
2013.
 42 CIRCUIT CELLAR • AUGUST 2014 #289
EMBEDDED IN THIN SLICES
Linux System
Configuration (Part 2)
Linux Application Design
The Linux system can be configured in many ways.
This article examines some options, including
COLUMNS
BusyBox, web servers and server-side support,
mail servers, time server interfaces,
secure networking, graphics libraries,
and browsers.
By Bob Japenga (US)
L et’s start with some definitions. In
Part 1 of this article series I talked
about configuring the kernel. The kernel is
the nerve center of a Linux system (i.e., the
OS). But the OS needs applications to do real
work. You need web servers, user interfaces,
I/O processing, and other logic encapsulated
in applications to do the actual work.
After you configure and build the kernel,
applications need to be selected and then built
for your particular kernel and processor. Linux
provides many open-source applications to
help fulfill your design. As a system designer,
you need to choose which applications you
want to go into your embedded system. Of
course, if your processor is infinitely fast and
has an unlimited amount of RAM and flash
memory, you can just skip this article. Others
may need to learn more.
Applications are put together in a
“package.” Packages are assembled in what
is called a “distribution.” You may have heard
of Ubuntu or Fedora. These are collections
of packages (one or more applications) and
a kernel put together as one unit called a
Linux distribution. For example, my company
recently started a project for a customer
and wanted to know what they were using
for Linux (i.e., tools being used, kernel
version, and applications running). When the
customer told us they were using an Ångström
distribution, we knew what they had (more on
Ångström next time). In this article, I want to
discuss the kinds of packages you may want
to include in your system.
WHAT ARE YOUR OPTIONS?
One of the reasons my company uses
Linux is because of the vast array of packages
available. Several times our customers have
asked: “Can you …?” You can fill in the blank.
And many times, because of the packages
available, we can say yes.
One example involved creating a virtual
display. One of our customer’s customers
wanted to control our embedded device from
a PC as if they were in front of the device’s
touchscreen. Voila. An open-source package
has already been designed to do just that.
Let’s look at some of the options you
have as a system designer in open-source
packages for embedded Linux.
BUSYBOX: THE SWISS ARMY
KNIFE OF PACKAGES
BusyBox is one package that is used in a
lot of embedded systems that are concerned
about resources. BusyBox puts a lot of (more
than 250) tiny UNIX/Linux applications into
one small executable. For example, typical
networking applications such as File Transfer
Protocol (FTP) and Telnet are included
in BusyBox. Common file manipulation
applications such as “cp” (copy a file),
“cmp” (compare), and “more” are included.
BusyBox can be configured to delete some

of these applications if space or simplicity
is appropriate. We reduced BusyBox to just
a handful of commands on one tiny Linux
system. With some distributions you can
configure BusyBox to build some “real” UNIX/
Linux applications and delete them from
BusyBox.
There is one thing you should be aware
concerning BusyBox. BusyBox commands are
not always exactly the same as the standard
UNIX/Linux documentation. Generally they
eliminate certain features rather than
changing the core operation. The first time I
used BusyBox on a project, I found that many
things I wanted to do were not present. Some
just required recompiling and rebuilding
BusyBox. Sometimes they were not present
at all. So beware.
WEBSERVERS AND SERVER-SIDE
SUPPORT
Since you are using Linux, you most likely
have a networked system. If you want to be
able to browse to the device (we usually use
this for configuring the system) or you want
to respond to HTTP queries from a host, you
will need a web server. Your system uses a
web server to deliver web content via the
Internet. As a system designer you have a lot
of options.
Apache—Apache is the grandfather of all
Linux web servers. Approximately 60% of
all web content on the Internet is handled
by Apache servers. Apache is a completely
full-featured web server with more options
than I could describe in several articles. With
this functionality comes significant resource
usage. Some estimates recommend 400 MB
of RAM and hundreds of megabytes of disk
space. It is so configurable, hard statistics
are not available about how much memory
Apache uses. We used Apache once on an
embedded system that had a gargantuan
amount of memory.
Lighttpd—We have used Lighttpd on many
of our embedded designs. Requiring less
than a tenth of the resources of Apache, it
supports almost all of the features we have
ever needed to design web-based embedded
systems.
Server-side languages—You need some
kind of server-side support language to serve
most modern webpages. If your device is going
to serve anything but a simple webpage, you
should include one or more of the following
languages: PHP, Perl, Python, Ruby, Java, or
Tcl. These languages are resource hogs on
very small embedded systems.
On one project, using PHP would have
taken 8 MB of flash memory. On this system,
which had 16 MB of RAM and 32 MB of flash
memory, we served webpages with a C
circuitcellar.com 43
language program. When we started the
project I was warned that serving webpages
in C was going to be difficult. But we just
didn’t have the flash memory resources. Now
I can truly say that the server-side software
was ugly. Most of you won’t be limited like we
were. However, if you are somewhat limited
you should choose one language you will use
to serve webpages.
MAIL SERVERS
We did a Linux project replacing an
embedded 8088 with a fax modem on an
COLUMNS
industrial control. Whenever the system
needed to sound an alert, it would fax the
details to everyone on the list. We convinced
our customer that e-mailing would be a much
better way to go. Adding e-mail support to
Linux is trivial. As with setting up a web
server, with mail servers you can go with
the standard full-featured mail server (called
the SMTP server) or there are various small-
footprint mail servers.
TIME SERVER INTERFACE
One of the common problems we face
when we create “headless” designs (i.e.,
when there is no user interface) is how to set
the time and date. There is a Linux package
that solves your problem. The Network
Time Protocol (NTP) enables your device to
continuously connect to a time server and
keep your time and date current. It works
much like what you have on a Windows PC.
SECURE NETWORKING
When we started designing networked
systems, security was not at the top of our
priority list. Now, customers no longer want
Telnet (an unsecure way of connecting to your
device). They demand Secure Shell (SSH),
which is just Telnet over a secure network. By
incorporating the OpenSSL package (which we
are all familiar with because of the HeartBleed
virus), you can add a secure layer to all of your
network communications. Generally, adding
OpenSSL is not enough since you will have
to provide some mechanism to handle the
encryption keys (encapsulated in certificates).
Again, the Linux open-source community will
provide that for you.
GRAPHICS LIBRARIES
Many Linux systems that we build
have some sort of GUI. With any graphical
interface, you need the graphical libraries
to easily design a snazzy user interface. The
granddaddy of Linux graphics libraries is the
X.Org Foundation’s X Window System. The
first UNIX system I ever worked with included
this library. It is large and sometimes painfully
slow. GTK+ is a faster and smaller-footprint
 44 CIRCUIT CELLAR • AUGUST 2014 #289
option. We have used it on several
systems and were well satisfied. QT
is a smaller and faster library, but my
company does not have any experience
using it.
We have done a few designs
where the user interface is a browser
interface. This requires the system
to have a browser and a web server.
Instead of designing widgets and
gadgets from scratch, draw upon the
vast number of webpage designs in the
world and point your browser to the
COLUMNS
device’s link local address.
ABOUT THE AUTHOR
BROWSERS
Bob Japenga has been
Whether you browse to your own
designing embedded web server (as described earlier) or
systems since 1973. In 1988, enable your user to browse externally
along with his best friend, he (we have only done this on one
started MicroTools, which spe- embedded system), you need to choose
cializes in creating a variety of at least one browser package. The
real-time embedded systems. common options we have used include
With a combined embedded Chromium (why it is not called Chrome
systems experience base of is an interesting story for another day),
more than 200 years, they love Firefox, and Midori. Chromium uses a
to tackle impossible problems lot of resources, but it is well known.
together. Bob has been award- We found Firefox to be unstable on the
ed 11 patents in many areas of one platform on which we ran it. Midori
embedded systems and motion is a fast and lightweight option. It is
control. You can reach him at part of the Raspberry Pi distribution.
rjapenga@microtoolsinc.com. We have limited experience using it.
There is a commercial browser
specifically designed for small
embedded platforms: NetFront and
Lighttpd, www.lighttpd.net.
Midori, www.midori-browser.org.
MySQL, www.mysql.com.
The Network Time Protocol (NTP), www.ntp.
circuitcellar.com/ccmaterials
org.
RESOURCES
The OpenSSL Project, www.openssl.org.
The Apache Software Founda-
tion, www.apache.org.
QT Project, http://qt-project.org.
The Ångström Distribution,
SQLite, www.sqlite.org.
www.angstrom-distribution.
org.
Ubuntu, www.ubuntu.com.
BusyBox, www.busybox.net.
Wikipedia, “Heartbleed.”
cURL, http://curl.haxx.se.
X.Org Foundation, www.x.org/wiki.
The Fedora Project, https://
fedoraproject.org.
SOURCE
NetFront
The GTK+ Project, www.gtk.
ACCESS Co., Ltd. | www.access-company.com
org.
NetFront NX. It is said to be “extremely”
optimized for embedded platforms. Nintendo
3DS and Playstation 3 Internet browsers use
this platform. Unfortunately, I do not have
licensing information or any experience
using it.
MISCELLANEOUS PACKAGES
There are a wide range of other packages
to suit many different needs. cURL is a
powerful command-line tool for sending all
kinds of data over the Internet over a variety
of protocols. We have used cURL in a several
settings where we wanted to communicate
over the Internet via a shell script rather than
programmatically.
How can I discuss packages without
talking about package managers? A
mechanism for managing the versions
of your packages is built into the fabric of
Linux. It seems as if each distribution has its
own package manager with its own nuances.
Simply stated, it is a powerful way to safely
and automatically update your applications
and packages in the field. That said, we have
never used a package manager to control
the updating of our proprietary code on
an embedded system. It seems as if our
customer’s requirements for security, the
website, version control, or something else
prevents us from using a package manager
in our embedded system.
Perhaps you have successfully used a
package manager on an embedded system.
E-mail me what you did and how you did it.
We have tried a few times to use them to
prevent us from reinventing the wheel for
software updates, but we have always been
stymied.
Another package we use on some of
our systems is a database programming
language called SQL. If your design requires a
distributed and relational database, you may
want to look into SQL. Several different SQL
packages are available. We have used MySQL,
which is a full-featured implementation and
SQLite, which is a somewhat “lite” version
of MySQL. By adding this package to your
toolkit you add all of the power of a managed
relational database.
KNOW YOUR OPTIONS
Configuring Linux involves configuring and
building the kernel, selecting the packages
you want installed in your system, and then
building them into a file system image. This
article provided you with information about
some of your options (in thin slices of course).
Part 3 of this article series will discuss the
options you have for building this vast array
into a workable Linux system.
 26
ADVANCE PROGRAM
August 10-12, 2014
A Symposium on High-Performance Chips
Flint Center for the Performing Arts-Cupertino,CA
http://www.hotchips.org
HOTCHIPS brings together designers and architects of high-performance chips,software, and systems. The tutorial and
presentation sessions focus on up-to-the-minute developments in leading-edge industrial designs and research projects
Register now by going to https://www.123signup.com/register?id=drvzv
Tutorial 1: Emerging Trends in Hardware Support for Security Organizing Committee
• Security Overview Princeton Chair
• Mobile HW Security ARM Krste Asanovic UC Berkeley
• Secure Systems Design AMD Vice Chair
Sunday
August 10

• Hardware Assists for Introspection Intel Fred Weber

Finance
Tutorial 2: Internet of Things Lily Jow HP
• Powering the Internet of Things TI Advertising
• Ultra-Low-Power Design Approaches for IoT National University of Singapore Don Draper Oracle
• Connecting the IoT Qualcomm Sponsorship
• IoT Systems Architecture: It’s not just “Embedded with a Radio” ARM Amr Zaky Invensense
Publications
High-Performance Computing Randall Neff
• SX-ACE Processor: NEC's Brand-New Vector Processor NEC Press
• SPARC64 XIfx: Fujitsu’s Next-Generation Processor for HPC Fujitsu Ralph Wittig Xilinx
• Anton 2: A 2nd-Generation ASIC for Molecular Dynamics Simulation D.E. Shaw Research Registration
Monday
August 11

Charlie Neuhauser Neuhauser

Keynote 1 Power Constraints: From Sensors to Servers Associates
Michael Muller ARM Location Services
John Sell Microsoft
Mobile Processors
Allen Baum
• NVIDIA’s Tegra K1 System-on-Chip NVIDIA Volunteer Coordinator
• Applying AMD’s “Kaveri” APU for Heterogeneous Computing AMD Gary Brown Tensilica
• NVIDIA’s Denver Processor NVIDIA Webmasters, IT
Technology Kevin Broch
• HBM: Memory Solution for Bandwidth-Hungry Processors SK Hynix Inc Production
• Improved 3D chip stacking withThruChip Wireless Connections ThruChip Communications Lance Hammond
• CMOS Biochips for Point-of-Care Molecular Diagnostics Insilixa Mike Albaugh
ARM Servers Keith Diefendorff
• The AMD Opteron “Seattle”: A 64b ARM Dense Server Processor AMD Steering Committee
• ARM Next-Generation IP Supporting LSI’s HIgh-End Networking ARM, LSI Logic Chair
Alan Jay Smith
• X-Gene2: 28nm Scale-Out Processor Applied MIcro Committee Members
FPGAs Allen Baum
• Design of a High-Density SOC-FPGA at 20nm Altera Don Draper Oracle
• A Reconfigurable Fabric for Accelerating Datacenter Services Microsoft Pradeep Dubey Intel
• Xilinx FPGAs Case Study: High capacity and Performance 20nm FPGAs Xilinx Lily Jow HP
John Mashey Techviser
• SDA: Software-Defined Accelerator for Large-Scale DNN Systems Baidu John Sell Microsoft
High-Performance ASICs Keith Diefendorff
• Hardware-Accelerated Text Analytics IBM Program Committee
• Myriad2 “Eye” of the Computational-Vision Storm Movidius Program Co-Chairs
• Goldstrike 1: A 1st Generation Cryptocurrency Processor for Bitcoin Mining Cointerra Sam Naffziger AMD
• RayChip: Real-Time Ray Tracing Chip for Embedded Applications Guri Sohi U. Wisconsin
Tuesday

Committee Members
August 12

Keynote 2 The Internet of Everything: What is it? What’s driving it? Forest Baskett NEA
What comes next? Pradeep Dubey Intel
Rob Chandhok Qualcomm John Davis Microsoft
Alan Jay Smith UC Berkeley
Dense Servers and Server Technology Steve Miller NetApp
• SCORPIO: 36-Core Shared-Memory Processor with a Coherent Mesh MIT Subhasish Mitra Stanford
• Unchaining the Datacenter with OpenPOWER: Reengineering a Server Ecosystem IBM Stefan Rusu Intel
• Oracle’s Next-Generation SPARC Cache Hierarchy Oracle Tom McWilliams BayStorage
• Atom C2000 Microserver: Power-Efficient Datacenter Processing Intel Behnam Robatmili Qualcomm
Ralph Wittig Xilinx
Big-Iron Servers Mike Taylor UCSD
• Performance Characteristics of the POWER8 Processor IBM Bill Dally NVIDIA
• Next Generation Oracle Sparc Processor Oracle Founder Bob Stewart SRE
• IvyBridge Server: Delivering Performance from Workstations to Mission Critical Intel
HOTCHIPS is accepting poster submissions by July 15 from both industry and academia. Accepted student poster
submissions will be considered for travel assistance. The best student poster will receive the Best Student Poster Award.
Warthman
Associates
Technical Writers
www.warthman.com

A Symposium of the Technical Committee on Microprocessors and Microcomputers

of the IEEE Computer Society and the Solid-State Circuits Society
 46 CIRCUIT CELLAR • AUGUST 2014 #289

PROGRAMMABLE LOGIC IN PRACTICE

Build a SoC Over Lunch

COLUMNS

Implementing a soft-core
processor inside an FPGA is not a new
concept. Recent releases have made developing and customizing these processors easier than ever, to the point
of enabling code compilation from an Arduino IDE. This article explains how you can use an FPGA to develop a
simple custom PWM peripheral for the processor.

By Colin O’Flynn (Canada)

I n a previous article, I introduced the

Xilinx’s Zynq programmable device with
dual hard-core ARM processors, which I used
to design a system-on-a-chip (SoC) solution.
(Refer to my article “All-Programmable SoC
Solution,” Circuit Cellar 285, 2014.) In this
article, I’ll introduce a much smaller and
lower-power option you can use: a simple
soft-core processor. Xilinx aficionados might
assume this will be either the MicroBlaze or
PicoBlaze devices, which are part of the Xilinx
ecosystem. But I’m actually going to discuss
another device.
Alvaro Lopes’s ZPUino is an open-source
design, which is interesting since it can be
programmed from an Arduino environment. If
you haven’t used Arduino, don’t let the hype scare
you. You are really programming in regular C/
C++ code, so if you’re already experienced with
embedded design you won’t have any problems.
One advantage of the Arduino environment is
there are many useful libraries you can utilize.
There is no need to write your own SPI, EEPROM,
or UART driver, for example. But I’m getting
ahead of myself. First let’s discuss why you want
a soft-core processor.
KILLING ME SOFTLY
Soft-core processors are convenient
because they have customizable designs. For
example, you can choose two or five UARTs,
or even interface directly to tailored FPGA

peripherals. Interfacing to custom peripherals
is one of the best aspects of adding a soft-
core processor, since some configuration
is often required for any peripheral. It can
be difficult to write a large state machine to
handle this, but it only takes a few minutes to
write a simple C program. The combination of
a processor with custom peripherals is often
referred to as a SoC.
Considering FPGAs’ (relatively) high cost,
you may question why you would not just use a
physically separate microcontroller. Powerful
microcontrollers with various peripherals cost
just a few dollars and typically have complete
development environment support. As I’ll
demonstrate, a soft-core processor is still a
useful asset since it can more tightly integrate
with FPGA logic. So let’s start by looking at
the ZPUino’s pedigree to better understand
how it was developed.
ZPU CHILDREN
In the beginning, there was the ZPU, which
is an open-source lightweight 32-bit CPU
design from Zylin Consulting. The ZPU uses
the “WISHBONE” bus standard to interconnect
with the CPU core and peripherals. For open-
source peripherals (e.g., those posted at
opencores.org), the WISHBONE standard is
probably the most widely used. The ZPU has
GCC support. If you have some experience
with embedded development you will probably
be quite happy. There is even support for the
eCos RTOS if you want a little more structure
for your program on top of the ZPU.
The ZPUino is based on the ZPU. The ZPUino
takes a version of the ZPU core, modifies it
slightly, packages it with some simple default
features and peripheral options to make life
ZPUino core
ROM System
(Instructions) memory
Bootloader mux
Bootloader Memory
code arbitrator
SDRAM
Controller
SDRAM Chip
circuitcellar.com 47
even easier, and then adds a modified version
of the Arduino IDE, which can compile code
for the ZPU. To store program code, the
ZPUino comes with a loader to program a
connected SPI flash device. Figure 1 shows an
example package of the ZPUino core as used
by Gadget Factory’s ZPUino Arduino Papilio
(ZAP) release, which I will discuss next.
The ZAP release packages the ZPUino with
a clone of Atmel’s AVR 8-bit microcontroller
and includes an interesting “graphical
configuration” utility to add peripherals
to either core. It even adds support for an
COLUMNS
open-source logic analyzer (called SUMP),
which enables you to have an integrated
logic analyzer (ILA) core inside your design!
The ZAP release targets two Gadget Factory
FPGA boards and works “out of the box” when
targeting these boards.
In this article I’ll be discussing the ZAP
release, since it provides the highest level of
integration. It’s important to realize that the
ZAP release is built on a collection of other
open-source tools. Thus you can use the ZAP
system to learn about the ZPU core. You can
later move on to your own design based on
the “original” ZPU core, if you wish.
MOPED OR MOTORCYCLE?
The ZPUino design is sufficiently speedy,
which enables you to perform some real
tasks on it. I ran my example implementation
on a Xilinx Spartan-6 LX9 MicroBoard at 96
MHz, so you certainly aren’t limited to only
simple configuration tasks. It may be easier
to perform protocol decoding on the ZPUino.
Fairly complex triggering tasks can be
done within the ZPUino. However, be aware
that the ZPU core isn’t designed for high
Stack (BRAM)
Zylin Computing
Mux ZPU 32-Bit core
Memory
arbitrator
I/O Controller
Interrupt FIGURE 1
controller
Alvaro Lopes’s ZPUino processor is
based on Zylin Computing’s ZPU
DMA Wishbone I/O Wishbone
extreme core. The ZPUino adds several
ZPUino processor useful peripherals, including the logic
to automatically load code from SPI
flash into synchronous dynamic RAM
(SDRAM) via the bootloader. This
is required for a system capable
of booting without any special
UART
SPI Driver GPIO configuration. Instructions and data
Driver
Wishbone Wishbone Wishbone are stored in the external SDRAM
slots slots slots
in this example. It’s possible to use
SPI
EEPROM internal Block RAM (BRAM) inside
Interface ports (no logic) the FPGA for code and data storage
instead.
 48 CIRCUIT CELLAR • AUGUST 2014 #289
PHOTO 1
This is Gadget Factory’s LX9 Board
(left) with the LogicStart MegaWing
(right). The two boards plug together
during operation. A Xilinx Spartan-6
LX9 along with synchronous dynamic
RAM (SDRAM), SPI flash memory, and
a USB interface is present on the main
LX9 board. The LogicStart adds the
seven-segment LED displays, LEDs,
and the ADC used in this example.
The video graphics array (VGA)
output isn’t used, but it would add
another example of a useful specialist
hardware peripheral you can include
COLUMNS
with the system-on-a-chip (SoC).
performance; instead it is intended to have
sufficient performance for most general-
purpose tasks.
If you know you’re going to be performing
tasks that require a higher-performance
core, your time would be better spent
learning a different soft-core processor
(e.g., Xilinx’s MicroBlaze, which targets
higher-performance users). As an example,
the Dhrystone MIPS (DMIPS) benchmark is
often used for embedded processors. The
MicroBlaze core on a Spartan-6 is capable
of 150 to 200 DMIPS. By comparison, my
tests showed the ZPUino clocking in around
14 DMIPS. Of course many applications don’t
need high performance and ZPUino’s lower
area and simpler configuration is of far
greater importance. So, if the ZPUino sounds
good to you, how do you get started?
SPI
ADC Logic
Driver
Eight-channel
PWM
User interface logic Seven-segment driver
GPIO
Serial port
FIGURE 2
I used a soft-core processor inside an FPGA to create this PWM system. Blocks are implemented with a
combination of hardware peripherals and software running on the processor. See text for a description of
the program flow.
PAPILIO PRO DEMO BOARD
If you want to follow along, I’m using
a different LX9 board from my previous
articles. Photo 1 shows the board I’m using
here, which is Gadget Factory’s Papilio Pro
along with its LogicStart MegaWing. The main
FPGA board has a Spartan-6 LX9, synchronous
dynamic RAM (SDRAM), a SPI flash memory
chip, and a USB interface. The LogicStart
MegaWing features I am going to use include
its eight switches, eight LEDs, four seven-
segment displays, and SPI-connected ADC.
You could modify this example to run with
another board (e.g., the Avnet LX9 Microboard,
which I used in previous articles). I’ll take full
advantage of the Gadget Factory LX9 board’s
extra I/O pins for the example application
presented here.
SEARCHING FOR A PROBLEM
Considering the number of solutions that
need problems, I decided to try and give you a
more realistic problem that is best solved with a
soft-core processor. FPGAs have an abundance
of I/O drivers, so I decided to create a very
I/O-heavy example. In particular I’m going to
use the device as a display board, since the
LogicStart MegaWing add-on has several LEDs
with additional seven-segment displays.
I’d like to drive the four seven-segment
displays on the board and drive all eight LEDs.
That alone is a simple task any microcontroller
could handle. But I also want to provide brightness
control through PWM on every channel. This is
still within the realm of microcontrollers, but I’d
be starting to tax the available number of PWM
channels on most devices.
The FPGA has no problem. I could have driven
32 or 64 separate channels if I wanted, it would
have just meant adding more PWM controllers.
Figure 2 shows a block diagram of the
system (including FPGA blocks). The SPI-
 circuitcellar.com 49

PHOTO 2
The Arduino IDE with Gadget Factory’s
ZPUino Arduino Papilio (ZAP) provides
a simple development to program the
soft-core processor.

COLUMNS
RS485/422/232/TTL
CONNECT WI TH ASC24T IBS485HV

Circuit
RS232<=>RS485 ATE Converter 5 Port Isolated
RS485 Repeater
$349.00

Cellar $45.00 board only Enclosur es, Cables,

Power Supplies and Other
Accessories

For people who are passionate

INTRODUCING THE SMFCOMX!
about hardware and software design,
embedded development, and computer • Extended Distance Units
applications. Awesome projects, • Converters • Serial to Digital I/O
industry news, trivia challenges, • Repeaters • Large Multi-Drop Networks
and more! • Multi-Repeaters
• Custom Units &
Connect. • Hubs
• Fiber Optics
Smart Units
Follow us on Twitter. Like us on Facebook.
• Isolators • Industrial, 3.0 KV Isolation
circuitcellar.com Call the RS485 Wizards
@circuitcellar 513-874-4796
@editor_cc circuitcellar www.rs485.com
 50 CIRCUIT CELLAR • AUGUST 2014 #289

library IEEE;
use IEEE.STD_LOGIC_1164.ALL;
use ieee.numeric_std.all;
use ieee.std_logic_unsigned.all;
use ieee.std_logic_arith.all;

library board;
use board.zpu_config.all;
use board.zpuino_config.all;
use board.zpupkg.all;
use board.zpuinopkg.all;
COLUMNS

entity MISC_zpuino_wb_pwm is
port (
wishbone_in : in std_logic_vector(61 downto 0);
wishbone_out : out std_logic_vector(33 downto 0);
fastclk_in : in std_logic;
pwm_out : out std_logic_vector(7 downto 0)
);
end entity MISC_zpuino_wb_pwm;

architecture behave of MISC_zpuino_wb_pwm is

component PWM is
port (
clk : in std_logic;
PWM_in : in std_logic_vector (7 downto 0);
PWM_out : out std_logic
);
end component;

signal selected: std_logic;

signal wb_clk_i: std_logic; -- Wishbone clock

signal wb_rst_i: std_logic; -- Wishbone reset (synchronous)
signal wb_dat_i: std_logic_vector(31 downto 0); -- Wishbone data input (32 bits)
signal wb_adr_i: std_logic_vector(26 downto 2); -- Wishbone address input (32 bits)
signal wb_we_i: std_logic; -- Wishbone write enable signal
signal wb_cyc_i: std_logic; -- Wishbone cycle signal
signal wb_stb_i: std_logic; -- Wishbone strobe signal

signal wb_dat_o: std_logic_vector(31 downto 0); -- Wishbone data output (32 bits)
signal wb_ack_o: std_logic; -- Wishbone acknowledge out signal
signal wb_inta_o: std_logic;

--signal PWM_Accumulator : std_logic_vector(8 downto 0);

signal PWM0_Setting : std_logic_vector(7 downto 0);
signal PWM1_Setting : std_logic_vector(7 downto 0);
signal PWM2_Setting : std_logic_vector(7 downto 0);
signal PWM3_Setting : std_logic_vector(7 downto 0);
signal PWM4_Setting : std_logic_vector(7 downto 0);
signal PWM5_Setting : std_logic_vector(7 downto 0);
signal PWM6_Setting : std_logic_vector(7 downto 0);
signal PWM7_Setting : std_logic_vector(7 downto 0);

begin
-- Unpack the wishbone array into signals so the modules code is not confusing.
wb_clk_i <= wishbone_in(61);
wb_rst_i <= wishbone_in(60);
wb_dat_i <= wishbone_in(59 downto 28);
wb_adr_i <= wishbone_in(27 downto 3);
 circuitcellar.com 51

wb_we_i <= wishbone_in(2);

wb_cyc_i <= wishbone_in(1);
wb_stb_i <= wishbone_in(0);

wishbone_out(33 downto 2) <= wb_dat_o;

wishbone_out(1) <= wb_ack_o;
wishbone_out(0) <= wb_inta_o;
-- Finish unpacking Wishbone signals.

selected <= ‘1’ when wb_cyc_i=’1’ and wb_stb_i=’1’ else ‘0’;

wb_dat_o(31 downto 8) <= (others => ‘0’);

COLUMNS
process(wb_clk_i)
begin
if rising_edge(wb_clk_i) then
wb_ack_o<=’0’;
if selected=’1’then
if wb_we_i=’1’ then

case wb_adr_i(4 downto 2) is

when “000” =>
PWM0_Setting <= wb_dat_i(7 downto 0);
when “001” =>
PWM1_Setting <= wb_dat_i(7 downto 0);
when “010” =>
PWM2_Setting <= wb_dat_i(7 downto 0);
when “011” =>
PWM3_Setting <= wb_dat_i(7 downto 0);
when “100” =>
PWM4_Setting <= wb_dat_i(7 downto 0);
when “101” =>
PWM5_Setting <= wb_dat_i(7 downto 0);
when “110” =>
PWM6_Setting <= wb_dat_i(7 downto 0);
when “111” =>
PWM7_Setting <= wb_dat_i(7 downto 0);
when others =>
null;
end case;

end if;
wb_dat_o(7 downto 0) <= (others => ‘0’);
wb_ack_o <= ‘1’;
end if;
end if;
end process;

pwm0: PWM port map(clk => fastclk_in, PWM_in => PWM0_Setting, PWM_out => pwm_out(0));
pwm1: PWM port map(clk => fastclk_in, PWM_in => PWM1_Setting, PWM_out => pwm_out(1));
pwm2: PWM port map(clk => fastclk_in, PWM_in => PWM2_Setting, PWM_out => pwm_out(2));
pwm3: PWM port map(clk => fastclk_in, PWM_in => PWM3_Setting, PWM_out => pwm_out(3));
pwm4: PWM port map(clk => fastclk_in, PWM_in => PWM4_Setting, PWM_out => pwm_out(4));
pwm5: PWM port map(clk => fastclk_in, PWM_in => PWM5_Setting, PWM_out => pwm_out(5));
pwm6: PWM port map(clk => fastclk_in, PWM_in => PWM6_Setting, PWM_out => pwm_out(6));
pwm7: PWM port map(clk => fastclk_in, PWM_in => PWM7_Setting, PWM_out => pwm_out(7));

end behave;

LISTING 1
This VHDL code is a simple PWM module. Each of the eight PWM modules is manually instiated to demonstrate how to write to different registers. You can modify this to make the
number of PWM modules a simple configuration instead.
 52 CIRCUIT CELLAR • AUGUST 2014 #289

connected ADC is continuously sampled and

volatile unsigned int * PWM_Base; the voltage is used to generate a bar-graph
representation with the eight LEDs. Using
void setup(){ PWM provides more flexibility than using
PWM_Base = &REGISTER(IO_SLOT(8),0); eight on/off LED states. With the PWM, your
LED can be partially on (i.e., not just at full
for (int i = 0; i < 8; i++) { brightness or completely off).
*(PWM_Base + i) = i * 32; The brightness level indicates the voltage
} being measured. If the measurement range is
} 0 to 3 V, each of the LEDs represents a 0.375-V
increment. A 1.5-V input would illuminate LEDS
1 through 4. A 1.6-V input would mean LED 5
LISTING 2
is dimly illuminated, whereas a 1.88-V input
COLUMNS

This C code shows a simple example of accessing the PWM register interface. The I/O slot number is based would fully illuminate LED 5.
on the schematic shown in Figure 3, which then is used to determine the peripheral’s base address. Writing In addition, a user interface over the serial
or reading to the peripheral is trivial, using standard pointer-type access or the REGISTER() macro. port prompts you to set the seven-segment
LED display’s value and brightness. You can

CLK CLK SPI_SCK SPI_SCK

SPI_MOSI SPI_MOSI
SPI_MISO SPI_MISO ZPUino Soft Processor - Papilio Pro - Vanilla - v1.0 TXD TXD
Author: Alvaro Lopes SPI_CS SPI_CS
RXD RXD Free sesources:
Used resources: alvie.com/zpuino
11 of 32 Block RAM (BRAM) 21-Block RAM (BRAM)
External SDRAM GPIO Pins
34% of available slices 66% of slices
DRAM_DQ(15:0) DRAM_DQ(15:0) 1 PLL (clock) gpio_bus_in(97:0)
1 of 2 PLL (clock) 8 MB of Program sode space
DRAM_CAS_N DRAM_CAS_N 2 DCM (clock) gpio_bus_out(147:0)
0 of 2 DCM (clock)
DRAM_CKE DRAM_CKE (code runs out of SDRAM)
DRAM_CLK DRAM_CLK
Clocks
DRAM_CS_N DRAM_CS_N
DRAM_RAS_N DRAM_RAS_N clk_96Mhz
DRAM_WE_N DRAM_WE_N clk_1Mhz
Visit Papilio.cc for socumentation System slock runs at 96 MHz
DRAM_ADDR(12:0) DRAM_ADDR(12:0) clk_osc_32Mhz
DRAM_BA(1:0) DRAM_BA(1:0)
DRAM_DQM(1:0) DRAM_DQM(1:0)

LED LED
Wishbone Wishbone Wishbone Wishbone Wishbone Wishbone Wishbone Wishbone Wishbone
slot 5 slot 6 slot 8 slot 9 slot 10 slot 11 slot 12 alot 13 alot 14

Wishbone empty slot Wishbone empty slot Wishbone empty slot Wishbone empty slot Wishbone empty slot

fastclk_in
pwm_out(7:0) LED_OUT(7:0)

gpio_bus_in(97:0)
MISC_zpuino_wb_pwm gpio_bus_out(147:0)
7SegAN3 WING_AL0
7SegDP WING_AL1
VGA 7SegAN2 WING_AL2
VGA_Hsync 7SegE WING_AL3
VGA_Vsync 7SegF WING_AL4
VGA_Red2 7SegC WING_AL5
VGA_Red1
VGA_Red0 7SegD WING_AL6
VGA_Green2 7SegA WING_AL7
VGA_Green1
VGA_Green0 7SegAN1 WING_AH0
VGA_Blue1 7SegG WING_AH1
VGA_Blue0 7SegB WING_AH2
7SegAN0 WING_AH3
SPI_CS WING_AH4
SPI_MISO WING_AH5
SPI_MOSI WING_AH6
Seven Segment SPI_Clk WING_AH7
dot Seg7_dot
VGA_Vsync WING_BL0
segdata(6:0) Seg7_segdata(6:0) VGA_Hsync WING_BL1
VGA_Blue0 WING_BL2
extra(31:0) VGA_Blue1 WING_BL3
VGA_Green0 WING_BL4
enable(3:0) Seg7_enable(3:0) VGA_Green1 WING_BL5
VGA_Green2 WING_BL6
Seven Segment VGA_Red0 WING_BL7

VGA_Red1 WING_BH0
VGA_Red2 WING_BH1
Audio WING_BH2
SPI ADC JoyRight WING_BH3
JoyLeft WING_BH4
SPI_MISO
JoyDown WING_BH5
miso
JoyUp WING_BH6
mosi SPI_MOSI JoySelect WING_BH7
sck SPI_CLK SW0 WING_CL0
SPI SW1 WING_CL1
SPI_CS
SW2 WING_CL2
SW3 WING_CL3
SW4 WING_CL4
SW5 WING_CL5
SW6 WING_CL6
SW7 WING_CL7
Audio
LED0
clk_96Mhz Audio
LED1
raw_out(17:0) data_in(17:0) LED2
LED3
Audio passthrough audio_out LED4
LED5
Delta sigma DAC LED6
LED7

Papilio LogicStart MegaWing Pinout

FIGURE 3
This schematic is used with the Xilinx ISE software to configure the high-level blocks of the system-on-a-chip (SoC). The MISC_zpuino_wb_pwm block is the custom eight-channel
PWM engine. Listing 1 provides the source code.
IF “EXPECT THE UNEXPECTED” WEREN’T SO EXPECTED,
WE’D SAY THAT’S EXACTLY WHAT TO EXPECT.
The 2015 International CES.® Four days of business, innovation
and exactly what you didn’t expect. Register at CESweb.org.

JAN. 6-9, 2015

LAS VEGAS, NEVADA
CESWEB.ORG
#CES2015
 54 CIRCUIT CELLAR • AUGUST 2014 #289
ABOUT THE AUTHOR
C o l i n O ’ F l y n n ( c o f l y n n @ n e w a e.c o m ) h a s b e e n b u i l d i n g a n d b re a k i n g e l e c t ro n i c
devices for many years. He is currently completing a PhD at Dalhousie
University in Halifax, NS, Canada. His most recent work focuses on embed -
d e d s e c u r i t y, b u t h e s t i l l e n j o y s e v e r y t h i n g f r o m F P G A d e v e l o p m e n t t o h a n d -
soldering prototype circuits. Some of his work is posted on his
website at www.colinoflynn.com.
COLUMNS
also use this serial interface to read the state
of the eight switches on the board. With that
background in mind, let’s start looking at how
to perform this design.
THE ZAPpy IDE
One wonderful thing about the ZAP system
is how easy it is to get started. You simply
download a ZIP file with everything in it,
extract it somewhere, and run the Arduino
IDE binary. The ZIP file is less than 200 MB,
compared with the multi-gigabyte Xilinx
Software Development Kit (SDK) you need
to develop with the MicroBlaze device. If you
want to modify the FPGA portion you need
to have Xilinx ISE installed, but presumably
you’ve already committed to that during your
FPGA work!
Photo 2 shows the IDE. From this IDE you
can select which FPGA “base system” file you
wish to program, as it ships with a few that
have standard peripherals present. You’ll
probably want to modify these to add your
own peripherals, which is something I’ll be
doing in this article.
Zylin Computing, “Zylin ZPU,” http://
opensource.zylin.com.
OpenCores.org, “SoC Interconnection:
Wishbone,” www.opencores.org.
circuitcellar.com/ccmaterials
SOURCES
AVR Microcontroller
RESOURCES
Atmel Corp. | www.atmel.com
A. Lopes, “ZPUino,” www.alvie.
com/zpuino.
Papilio Pro FPGA and LogicStart MegaWing
Gadget Factory, LLC | www.papilio.cc
C. O’Flynn, “All-Programmable
SoC Solution,” Circuit Cellar
Zynq platform device, Spartan-6 LX9
285, 2014.
MicroBoard, MicroBlaze soft processor,
PicoBlaze microcontroller, and Xilinx
ProgrammableLogicInPractice.
Software Development Kit (SDK)
com.
Xilinx, Inc. | www.xilinx.com
Selecting your “board” defines the base
system file. An option called “Burn Bootloader”
programs the FPGA bitstream. By default, this
will also program it into the SPI flash on the
board (i.e., you only have to run the “burn
bootloader” once and then every time you
power on the board it will automatically load
the FPGA bitstream).
If you aren’t familiar with Arduino, the
process of compiling the source code is called
“verifying” it. Once the code compiles, you can
select the “upload” option, which programs
the code into the SPI flash memory on the
board and loads it into SDRAM.
With the ZPUino, code runs from SDRAM
on the Papilio Pro board. When the FPGA
powers on it will automatically load the code
from SPI flash memory into the SDRAM and
begin executing. The ZAP tools manage this
for you. A single SPI flash memory is used
for the FPGA bitstream and software code
storage. Now that you know how the system
works, let’s delve into the FPGA itself.
REMODELING THE FPGA
Soft-core processors are nothing new. But
the ZAP system uses Xilinx ISE’s schematic
entry mode to provide a graphical method
of adding peripherals to the base system to
configure your SoC. This provides an excellent
higher-layer view of your system, and you
can easily link a VHDL or Verilog block into a
schematic symbol. Thus the schematic entry
is only used for high-layer development. The
real work is still done with standard VHDL or
Verilog code.
Figure 3 shows a schematic of my base
system from the Xilinx ISE, which includes
a special PWM block that implements eight
PWM channels. This PWM block is connected
to a specific wishbone “slot.” You can refer to
this wishbone slot from your software code,
which then provides a simple mapping of your
peripheral to the software code.
Listing 1 shows the interior of the PWM
block. Each PWM channel has a different
register that sets the PWM duty cycle. You
can see how easy it is to add new registers to
 circuitcellar.com 55

the block. Of course you can easily interface
to a device such as a Block RAM (BRAM) by
simply deciding which high-level bits within
your peripheral address space will select
the BRAM and then connecting the low-level
address bits to the BRAM device. Because the
processor has a 32-bit data bus, the address
increases by 0x04 each time. The two low
level bits are always “00” when performing a
32-bit read or write. Now that I’ve explained
how to add a custom peripheral, let’s drive
that peripheral from the software to perform
something useful.
As always, the complete projects
including step-by-step details are available
at ProgrammableLogicInPractice.com. You’ll
also want to see details of the ZAP software
release posted on the Gadget Factory’s
website, since many useful tutorials and other
documentation are available.
Remember the best method of cementing
your knowledge is through experience, so go
forward and design your own specialist
peripherals for the soft-core processor. You
may be pleasantly surprised to discover how
easy it is and how many applications you

COLUMNS
suddenly find for your new soft-core
SOFTWARE DEVELOPMENT processor.
The custom peripherals simply get mapped
into your processor address space. This
means writing to or reading from a register is
just like accessing a peripheral in your “usual”
microcontroller or SoC device. You simply
have a base address of the peripheral and an
offset to the specific register of interest.
Listing 2 shows a code snippet that writes
a different value to each PWM register. The
REGISTER() macro provides access to a
specific register that you can read or write,
based on an I/O slot number and an offset.
You can also use this to get the base address
(as I’ve done in Listing 2) and then use
pointer math to access that.
The I/O slot number is determined
from Figure 3, which was the graphical
configuration of the system using ISE. You’ll
most likely want to write a separate class for
your peripheral, where the I/O slot number is
simply configured when you initialize the class.
The provided examples show how this is done.
The seven-segment driver is a good example
of a simple class using such a strategy.
Again, if you aren’t already familiar with
the Arduino environment, you can effectively
consider it writing regular C/C++ with only a
few minor changes. It’s hard to beat in terms
of quickly getting an example compiled and
loaded into real hardware. It only takes a few
minutes to program a FPGA base system along
with a processor program. For development
you’ll find it faster to avoid programming the
SPI flash memory with your new program.
Instead only load it into RAM, which takes just
a few seconds. This is done by holding down
shift while running the upload command.

I REJECT YOUR
MICROCONTROLLER AND
SUBSTITUTE MY OWN
If you haven’t used soft-core processors, I
hope this article has introduced an easy way
to get started. If you already have some FPGA
designs, you may find that a simple soft-core
processor is just what your design needs for
some configuration or housekeeping tasks.
 56 CIRCUIT CELLAR • AUGUST 2014 #289

THE CONSUMMATE ENGINEER

The Humble Resistor (Part 1)

Fixed Resistors

Resistors are the most common parts in

electronic circuits. This article provides an
overview of fixed resistors and touches on
COLUMNS

their values and power varieties.

By George Novacek (Canada)

E lectrical resistance is omnipresent.

Every circuit exhibits it, whether
it’s due to the presence of resistors or
inherent characteristics of components
and conductors. Negative resistance is also
recognized, but this is not a topic for this
article. Figure 1 shows the American (a) or
International Electrotechnical Commission
(IEC) (b) symbols schematic diagrams use to
represent resistors. Photo 1 shows a variety
of different through-hole fixed resistors.
When a resistor is connected to a voltage
source, a current with a magnitude related to
the voltage by Ohm’s law:
V the values are listed in Table 1. Successive
I=
R
flows through it. The current is expressed
in amperes, while voltage is a potential
difference across the resistor terminals in
volts. The resistance R is expressed in ohms.
In addition to the ohm (Ω), common
resistance units include milliohm (1 mΩ = 10 –3
Ω), kilohm (1 kΩ = 103 Ω), and megohm (1 MΩ
a) = 10 6 Ω). In circuit analyses it is sometimes
convenient to use a reciprocal of resistance
called conductance:
b) 1 series E192, 0.1%, 0.25%, and 0.5% values are
G=
R available with 192 values per decade. Unless
FIGURE 1 In Europe conductance is referred to as
These symbols denote (a) American Siemens or, mostly in English texts, as mho (1
or (b) International Electrotechnical S = 1 mho = 1 Ω–1).
Commission (IEC) resistors. Resistor values are also written differently
in American- and European-style diagrams.
In the latter, the decimal point is omitted,
such that 0.3 Ω would be written as 0R3, 3.3
Ω = 3R3, 1.5 kΩ = 1k5, 0.1 MΩ or 100 kΩ =
M1 or 100 k, 2.2 MΩ = 2M2 and so forth. (A
similar notation is used for capacitors.)
RESISTOR VALUES
Resistors come in standard values that
are related to their tolerances (custom values
could be ordered for a price). For example,
10% resistors have 12 values fitting into a
decade (e.g., 10 Ω to 100 Ω). Consequently, the
series is called E12. To refresh your memory,
values in a series are related approximately
by a factor:
n
10
where n = 12, 24, 96, or 192, respectively. The
E24 series is used for 2% and 5% resistors
(see Table 2). Table 3 lists 1% resistors,
which come in E96 series.
Most of the time digital projects can be
accomplished with 5% (i.e., E24 resistor
values), occasionally requiring 1% values). In
you work with precision analog circuits, you
are not likely to need those.
Present-day resistor manufacturing
technology is so refined and economical
that 10% tolerance resistors have almost
 circuitcellar.com 57

disappeared from the marketplace. This has

not always been so. There were times when—
mainly due to the cost—20% devices had to be
used. Because resistor manufacturers tested
and categorized their production runs into
tolerance bands, only values between ±10%
to ±20% of the nominal could be found in
a 20% resistors batch (i.e., 80 to 90 Ω and
110 to 120 Ω for the specified 100 Ω). It was
often necessary to use an ohmmeter to select
matching resistors. Paradoxically, it was
beneficial to the quality of designs. Engineers
had to work hard to make their circuits accept

COLUMNS
wide tolerances and minimize the need for
precise values as much as possible.

SELECTING RESISTORS
Many characteristics must be considered
when selecting resistors for specific functions.
The most common ones are resistance, its
tolerance, and power dissipation. I try to
minimize the use of resistors with values
that are not above 1. With the exception of
one project, I never had to exceed 10 MΩ.
High resistance values in conjunction with
input capacitances of active components
(e.g., MOSFETs) create low-pass filters and
limit bandwidth. Worse, their values are
susceptible to PCB leakage due to humidity,
moisture ingress, dirt, and so forth. If you
need resistors larger than 1 MΩ, be careful
with the layout. Make sure the PCB is well
cleaned after the assembly and conformally
coated to ensure long-term stability.
Most microelectronic circuits are quite
happy with 0.125-to-0.5-W resistor power
ratings. It is a bad idea to use resistors
rated close to their in-circuit dissipation. The
PHOTO 1
resistors become hot, their resistance may
Various fixed resistors are shown.
change, and the reliability will diminish. I
derate resistors to operate at no more than
50% of their power rating.
Engineers must also consider all the
desired characteristics such as temperature
coefficienct, working voltage, and through-
hole or surface mount device (SMD). My
favorite resistors are metal film, type ERL.
POWER DISSIPATION
Power dissipated by a resistor is calculated
as:
V2
P = I2 × R = I × V =
R
where V is the voltage across the resistor,
I is the current flowing through it, R is the

TABLE 1
10 12 15 18 22 27 33 39 47 56 68 82 Resistor values are related to their
tolerances. These are the E12
resistance series values.
10 11 12 13 15 16 18 20 22 24 27 30 TABLE 2
33 36 39 43 47 51 56 62 68 75 82 91 E24 resistance series values are
shown.

10 10.2 10.5 10.7 11 11.3 11.5 11.8 12.1 12.4 12.7 13

13.3 13.7 14 14.3 14.7 15 15.4 15.8 16.2 16.5 16.9 17.4
17.8 18.2 18.7 19.1 19.6 20 20.5 21 21.5 22.1 22.6 23.2
23.7 24.3 24.9 25.5 26.1 26.7 27.4 28 28.7 29.4 30.1 30.9
31.6 32.4 33.2 34 34.8 35.7 35.6 37.4 38.3 39.2 40.2 41.2
42.2 43.2 44.2 45.3 46.4 47.5 48.7 49.9 51.1 52.3 53.6 54.9
56.2 57.6 59 60.4 61.9 63.4 64.9 66.5 68.1 69.8 71.5 73.2
75.0 76.8 78.7 80.6 82.5 84.5 86.6 88.7 90.9 93.1 95.3 97.6

TABLE 3
These are the E96 resistance series values.
 58 CIRCUIT CELLAR • AUGUST 2014 #289
COLUMNS
ABOUT THE AUTHOR
George Novacek is a
professional engineer
with a degree in Cyber-
netics and Closed-Loop
Control. Now retired, he
was most recently pres-
ident of a multinational
manufacturer for em-
bedded control systems
for aerospace applica-
tions. George wrote 26
feature articles for Cir-
cuit Cellar between 1999
and 2004. Contact him
at gnovacek@nexicom.
net with “Circuit Cel-
lar”in the subject line.
circuitcellar.com/ccmaterials
resistance, and P is the dissipated power in
watts. The total amount of energy dissipated
over time is:
t2
W= ∫ v (t ) i (t ) dt
t1
and the average power dissipated over a
specific period is:
t2
1 and 500 ppm/°C and they have a maximum
P =
t2 − t1
× ∫ v (t ) i (t ) dt
t1
This means that many resistors can, in a
pulse, dissipate large power many times their
rating. The maximum pulse dissipation is
often specified, but it is crucial to check the
high voltage rating as well.
My engineers once substituted momentarily
unavailable carbon film resistors we used in
lightning protection circuits with metal film
resistors, without checking their maximum
voltage rating. The resulting fireworks in
the course of lightning susceptibility tests
were spectacular. The second controller
was refitted with high-voltage wire-wound
resistors. Their performance was superior
to the original carbon film, also due to their
parasitic inductance.
RESISTOR VARIETIES
Carbon composition resistors are an
old technology. Their long-term stability is
poor and, because of their noise, they are
unsuitable for low-signal applications. But
their working voltage can be around 500 V.
In my first semester at university, we were
given blank carbon resistor bodies. We used a
machine similar to a lathe to cut a helix into
their resistive surface to make resistors of
given values. Getting anywhere close to ±20%
of the desired value was quite a feat. Yet, this
is how resistors were made in factories, then
sorted and marked with values.
Carbon film resistors are made by
depositing carbon film on an insulating
substrate. Their working voltage ranges up
to about 1,000 V and their noise is much
lower than that of their carbon composition
predecessors. Because of their high pulse
stability, they are popular in transient
protection circuits.
RESOURCES
TT Electronics, plc, “Ultra High Value Resistors,”
2011.
Vishay, “Metal Film Resistors, Military/Estab-
lished Reliability, MIL-PRF-39017 Qualified,
Type RLR,” ERL (Military RLR), 2012.
Metal film resistors, available as SMDs
and through-hole, have been quite popular.
Their bodies are usually coated with nickel
chromium (NiCr) into which—similar to
carbon composition resistors—a helix is cut.
Metal film resistors are available with 0.5%,
1%, 2%, and sometimes 10% tolerances.
Metal film resistors have very low noise,
about 20 dB below that of thin film devices. A
typical temperature coefficient is between 50
100-to-500-V working voltage.
Wire wound resistors are made, as
the name implies, by winding a wire (e.g.,
nichrome) around a nonconductive body. These
resistors can operate at high temperatures
and dissipate a lot of power. Some wire
wound resistors are attached to a heatsink
for better heat dissipation (see Photo 1).
They can be built with 0.005% tolerance, a
3 ppm/°C temperature coefficient, and can
withstand 5,000 V of electrostatic discharge
(ESD). They have parasitic inductance larger
than other technologies, but this can also be
a benefit as mentioned above.
THICK AND THIN FILM
RESISTORS
In non-critical applications, carbon
resistors may be printed directly on a PCB.
Thick and thin film resistors are seen mainly
as SMDs. The chief difference between thick
and thin film resistors is the manufacturing
process. Thick film is typically 10 to 50
µm, while thin film is 10 nm to 2 µm. In
manufacturing, thick film is fabricated by silk
screening conductive and resistive materials
in a pattern on a substrate. The resistors are
designed to be on the low side and are then
laser trimmed for the required value.
Thin film resistors are fabricated by
etching conductive and resistive materials
previously vacuum deposited on a substrate
and then laser trimmed. In general, thin
film resistors perform better than their
thick film, less expensive counterparts. Their
temperature coefficient is about 5 ppm/°C
compared with 50 ppm/°C for thick film
resistors. They can withstand 2,500 V ESD as
opposed to 2,000 V for thick film resistors and
their noise is about –40 dB (i.e., 1%) of thick
film resistors. Thin film resistors can be built
with 0.05% tolerance, thick film resistors can
be built with 0.5%.
I discussed the most common resistors.
There are also many “exotic” types, such as foil
resistors, glass-encapsulated ultra-high value
resistors with values up to 100 TΩ (1014 Ω),
carbon pile , water, grid, and others, which are
too specialized to discuss at this time. In Part 2
of this article series, I’ll discuss variable
resistors.
Flash enables new designs in consumer,
computer, and enterprise products.
Are you struggling with crucial Solid State Drive (SSD) decisions? Can SSDs resolve your application bottlenecks?
How can you maximize SSD performance? Flash Memory Summit will explore new frontiers in enterprise storage
and help you make the right choices. Summit topics include:
Solid state drives • Enterprise storage • All-flash disk arrays • Controllers
PCIe SSDs/NVMe • Application acceleration • 3-D flash • Market research
Security • Standards • Interfaces • Cloud computing • Consumer/mobile applications
Embedded applications • Data recovery/erasure
Plus Keynotes by NetApp, Samsung, Microsoft, Dell, SanDisk, IBM, Diablo,
Marvell, Fusion-io, PMC-Sierra, Nimbus, Cadence

August 5 -7, 2014

and next year:
August 11-13, 2015
Santa Clara Convention Center
Santa Clara, CA
Don’t miss this important event!

REGISTER ONLINE
www.FlashMemorySummit.com
Readers of Circuit Cellar Magazine: Enter priority code SPGP for a $100 discount.
 60 CIRCUIT CELLAR • AUGUST 2014 #289
THE DARKER SIDE
Shielding Experiments
Electromagnetic shielding
against RF radiation. However, adding
shielding is not as easy as it sounds.
This article describes some
shielding experiments and
encourages readers to conduct
a few tests of their own.
COLUMNS
By Robert Lacoste (France)
protects
W elcome to the Darker Side. This column
is devoted to “black magic” subjects, but
this month I will discuss a topic even more
obscure than usual: shielding. Of course I am
talking about electromagnetic shields, nothing
related to armors or Arduino extensions.
Assume you are suffering from interference
between two electronic systems, an offender A
and a victim B. If you ask for advice you will
usually get an answer such as, “Easy, just add
some shielding!” But is it so easy?
IS A SHIELD REQUIRED?
Adding a shield on an electronic device
could be a good advice, but I encourage
you to double check all other solutions first.
Why? Because shielding is always complex
and expensive as I will explain in a minute.
Moreover, it simply may not solve the problem.
A shield protects against RF radiation.
Therefore you must verify that the
perturbation is indeed radiated from system
A to system B. This may not be the case if
there is any wire between the two systems.
For example a common power supply (e.g.,
wire) can conduct the perturbations quite
easily and would render a shield useless. Try
separating the power supplies, using batteries
if possible. Try adding some ferrite clamps
on all wires. Try adding filtering capacitors
on all interconnecting wires. Is the problem
diminished at all? If yes, it will be a good idea
to solve these conducted perturbations at
least as a first step. You may also try to shield
the interconnecting cables before blaming the
board, as a cable is a very good antenna. (Refer
to my article “Cable Shielding Experiments,”
Circuit Cellar 219, 2008.)
If you are sure that the issue is a radiated
coupling between the two systems, what
do you do next? Before dealing with board
shielding I encourage you to try to either
reduce the perturbations radiated by the
offender or to increase the victim’s immunity.
How? Try to reduce the amplitude of the
offender’s signals, as it will make them more
harmless. You can also increase the impedance
of the noisy signals still on the offender’s side.
Reciprocally, try to increase the amplitude of
the victim’s signals, as it will make them more
 circuitcellar.com 61

resistant. Also try to reduce the impedance

of all susceptible signals. Adding a 100-Ω
resistor to ground can be magical.
The last common trick is to reduce the
edge rates of all digital signals (e.g., by using a
slower digital technology), as it will drastically
limit their harmonics. Recently one of my
colleagues reduced the harmonics radiated
by a 13.56-MHz power amplifier by 10 dB by
adding a small resistor between a fast logic
driver and the power MOSFET’s gate.
Now assume that everything else failed
and you will need to add a shield on either

COLUMNS
the offender A or the victim B. Theoretically,
both solutions provide the same result.
Shielding the offender is cleaner for the other
potential victims, whereas shielding the
victim may be a better idea if other source
of perturbations can appear in the vicinity.
So this choice will be design-dependent. Of
course you can also shield both, but it will be
twice as expensive.

SHIELD EFFICIENCY
What can you reasonably expect in terms PHOTO 1
of noise reduction with a shield? Plenty of I designed a test PCB using only a 50-Ω microstrip track terminated on a 50-Ω resistor to ground. This
“shield efficiency calculators” are available makes a low-cost wideband test antenna.
online (see Resources). Such tools provide the
theoretical attenuation at a given frequency build my PCBs, so the track width for 50-Ω
for a given sheet of metal. Theory is great, impedance was 3 mm.
but real life is always a little trickier. Such a microstrip circuit acts as a
I will share some experiments I’ve done. very wideband antenna, even if it is a low-
The idea was to build a simple test setup to efficiency antenna. It could be used to either
measure some typical shields’ actual efficiency. generate perturbations by connecting it to
But how do you define “efficiency?” I stayed a signal source, or to measure the received
as close as possible to real-life situations
and assumed that two configurations were
representative. HP 71200C Spectrum analyzer
First, the offender and victim could be very
close to each other. For example, this appears
if the two subcircuits are on the same PCB. In
that case, the coupling is usually either purely
magnetic or is a close-field electromagnetic
coupling. Track RF
The second common configuration is when out IN

the two systems are a couple of meters away.

In that case, the coupling is usually a classic
far-field electromagnetic type and the theory
is simpler. HP8447F HP8447D
Let’s start with the close-field case. How Amplifier Amplifier

can the actual efficiency of a given shield IN1

0.1 to 1,300 MHz
26 dB OUT1 IN1
0.1 to 1,300 MHz
26 dB OUT1
in such a situation be measured? First I IN2 0.009 to 50 MHz OUT2 IN2 0.1 to 1,300 MHz OUT2
needed to experiment with a “typical” set of 28 dB 26 dB

offenders and victims. I etched two identical

test PCBs. Each circuit included a 5-cm long
50-Ω microstrip track terminated with a
50-Ω resistor to ground. Remember that a
microstrip is a track of a precise width on the
PCB’s top side and a full ground plane on the
bottom side. (Refer to my article, “Microstrip
Techniques,” Circuit Cellar 223, 2009.) I used
a standard, 1.6-mm thick FR4 substrate to
Shield under test
Test PCBs with microstrip antenna
FIGURE 1
The test setup included two test PCBs side by side. One PCB was driven by a tracking RF generator and the
other was connected to a spectrum analyzer’s input to measure the received signal strength. Using two
amplifiers improved the dynamic range.
 62 CIRCUIT CELLAR • AUGUST 2014 #289

perturbations on the victim side using a

spectrum analyzer. I added an SMA connector
on the PCB’s bottom side to easily connect to
the RF source or analyzer. Then I added some
holes to fix a shield, an extra ground plane
on the top layer everywhere except inside
the shielded area, and plenty of ground vias
(see Photo 1). I used Labcenter Electronics’s
Proteus tool suite to develop the computer-
aided design (CAD) file. (The file is available
for download on Circuit Cellar’s FTP site if you
would like to rebuild it.)
To have a complete test setup I then
COLUMNS

positioned both boards face to face, around

20 cm from each other. I connected them
to my trusty old Hewlett-Packard HP71200C
spectrum analyzer, which is available in my
company’s lab. This equipment includes a
Hewlett-Packard HP70300A tracking generator
PHOTO 2 module, which transforms it into a scalar
The test setup is shown. network analyzer. The tracking generator
provides a test carrier signal aligned with the
frequency measured by the analyzer. It may
a)
be more difficult, but you could use any RF
signal generator and a standard spectrum
analyzer to make the same measurements.
The test procedure is simple: Start with
the two boards face to face without any shield.
Sweep the offender’s signal frequency from 1
MHz or so to the desired upper frequency (e.g.,
1 GHz). Measure how much power is received
on the victim’s side at each frequency. This
is your reference. Then add a shield on one
of the boards and do the measurement
again in the same conditions. The difference
between the two measurements will give the
attenuation provided by the shield, which will
be frequency-dependent. Figure 1 shows the
overall test setup. I just added some extra
amplifiers to increase the measurement
sensitivity. Photo 2 shows the actual
experiment.

b) 3-D PRINTED SHIELDS

I recently received my first 3-D printer,
a MakerBot Replicator 2X. This experimental
printer is a dual-head unit that is optimized
for acrylonitrile butadiene styrene (ABS)
filaments. ABS is a little less common than the
usual poly lactic acid (PLA) used on low-cost
3-D printers, but I ordered this configuration
assuming it would enable me to build more
robust parts.
What about trying to use a 3-D printer to
manufacture a shield? The idea may seem
silly since ABS—like nearly all plastics—is
nonconductive, but I wanted to give it a try.
My first idea was use a standard ABS filament
PHOTO 3 to 3-D-print a shield and then use a conductive
a—Looking at a 3-D printer in action is a hypnotic experience. b—I used a MakerBot Replicator 2X paint to paint it. This type of paint is indeed
3-D printer to build this acrylonitrile butadiene styrene (ABS) shield, which is shown after I applied the often used to “shield” plastic enclosures and
electromagnetic (EMC) paint on the inside. I always wondered if it was really effective.
 circuitcellar.com 63

This test was a good way to find out.

Using the 3-D printer to design and
manufacture the ABS shield took me less than
1 h. I simply used a common CAD Drawing
Exchange Format (DXF) to export the outline
of the required shield from my PCB CAD tool.
Then I launched Punch!CAD Labs’s ViaCAD
3-D design software. I imported the DXF file,
extended it to a 3-D object, and exported the
corresponding stereolithography (STL) file.
The last steps were to import the STL file
from MakerBot’s software, click on “print,”
and voila! (The files are available at Circuit

COLUMNS
Cellar’s FTP site.)
Photo 3a shows the printer working.
I used two layers of Kontakt Chemie’s EMI-
35 electromagnetic (EMC) paint to paint the
inside of the resulting part (see Photo 3b.)
The result is pretty nice, but what about its
RF performance?
Photo 4 shows the results of the test
using the previously explained procedure. I
measured an attenuation of about 10-dB in the
PHOTO 4
lowest frequencies, up to 150/180 MHz. This is
This plot shows the measured signal strength with the conductive acrylonitrile butadiene styrene (ABS) shield
actually not too bad as a 10-dB attenuation (yellow) as well as the reference signal with no shield (blue). The shield efficiency is the difference between
means that 90% of the perturbations are no both curves. In this plot it is around 10 dB up to 150 MHz but virtually zero above that frequency.
longer transmitted. However, the attenuation
was close to 0 dB for frequencies above this
limit and up to 2 GHz at least. Why? This is ground plan (i.e., I created a continuous
probably because the EMC paint I used is not solder joint all around it). Photo 5 shows the
efficient for high frequencies. So, EMC paint results. I used the same test configuration to
can solve your problem, if you are just looking measure the shield effectiveness again (see
for a couple of decibels of improvement, but Photo 6). I measured a 20-dB attenuation in
use this paint with caution. low frequencies, 25 to 30 dB from 400 MHz to
I wanted to give 3-D printed shields another 1.5 GHz, and around 20 dB at 2 GHz.
chance, so I bought a roll of conductive ABS How do you interpret these results? First,
filaments. These are plastic with some very using a metallic shield is obviously far better
limited conductivity (some kilo-ohms per
centimeter) but I thought it could slightly
improve the shielding effectiveness. I used
these filaments to print another part, added
two layers of EMI-35 paint to be safe, and
retested it. The resulting performance
was not significantly better than with the
standard ABS filament (still around 10-dB
attenuation up to 200 MHz or so), but close
to zero above that limit. I assume this is due
to the “conductive” plastic’s high resistance. I
suppose this material could be interesting for
electrostatic dissipation, but I need to wait
for improved conductive filaments before
using them for RF.

USING METALLIC SHIELDS

What about trying a standard metallic
shield? What efficiency could you expect? I
grabbed an off-the-shelf shield and soldered
it on the victim’s test board. First, I just
soldered the shield through its six pins. The
measured attenuation, which was around 15
to 20 dB, was quite deceptive. Then I took PHOTO 5
the time to properly solder the shield to the A metallic shield can provide improved attenuation, as long as it is properly soldered.
 64 CIRCUIT CELLAR • AUGUST 2014 #289
COLUMNS
PHOTO 6
Here is the received signal strength with the metallic shield (yellow), compared with the reference signal with
no shield (pink). The blue curve shows the minimum detectable signal and was measured with a 50-Ω load in
place of the transmitting PCB. A 20-to-30-dB attenuation is measured for all frequencies up to at least 2-GHz.
PROJECT FILES
circuitcellar.com/ccmaterials
RESOURCES
The Clemson University Ve-
hicular Electronics Laboratory
(CVEL), “Shield Effectiveness
Calculator, www.cvel.clemson.
edu.
R. Lacoste, “Cable Shielding
Experiments,” Circuit Cellar
219, 2008.
———, “Microstrip Techniques,”
than using a conductive paint, especially at
high frequencies. Then even a metallic shield
is not perfect. Achieving a 20-dB attenuation
(i.e., 99% of the perturbations are blocked)
is easy when the shield is properly soldered,
but going higher is not trivial. And shield
soldering must always be done on all its
sides, not just through its legs. Why was
the measured shield efficiency lower in low
Circuit Cellar 223, 2009.
LearnEMC, “Shielding theory,” www.learnemc.
com.
SOURCES
HP71200C Spectrum analyzer HP70300A
tracking generator module
Hewlett Packard Co. | www.hp.com
(Product now obsolete, available from distrib-
utors such as eBay)
Proteus Design suite
Labcenter Electronics | www.labcenter.com
ViaCAD 3-D Design software
Punch!CAD Labs | www.punchcad.com
Replicator 2X 3-D printer
MakerBot | http://store.makerbot.com
frequencies? This was probably because the
coupling was then more magnetic than EMC,
and a non-ferromagnetic metal doesn’t stop
magnetic coupling.
FAR-FIELD COUPLING
After performing these experiments in
near-field conditions, I wanted to check if these
measurements were more or less consistent
with far-field conditions. Fortunately, I have
access to an anechoic chamber, which is ideal
for these kinds of measurements. An anechoic
chamber basically emulates an “open-field”
situation, as if you were testing an RF device in
the middle of nowhere. There are no external
perturbations and no signal reflections on
the walls, just a clean line of sight path. I
installed one of the test PCBs in the chamber,
connected it to a RF generator set to a 1-GHz
continuous wave, and measured the 1-GHz
signal received by a test antenna three meters
away (see Photo 7). This gave me a reference
level. Then I added a metallic shield on the
PCB and measured the signal’s attenuation.
I got an attenuation of about 20 dB. This
was roughly consistent with the close-field
measurements; remember that I got around
25 dB. Next I replaced the metallic shield with
the 3-D-printed conductive ABS shield. This
time I measured a 10-dB attenuation.
How does this fit with near-field
measurements? Remember that the
conductive ABS shield provided nearly no
attenuation at 1 GHz at close distance, but
it still reduced the far-field signal by 10 dB.
Why? Honestly, I don’t have a clear answer.
I guess this is due to the quite complex
behavior of near-field coupling. Perhaps this
shield still perturbates the wave propagation
through its electric component even if it
doesn’t change magnetic close-field coupling,
and this reduces the corresponding generated
far field by 10 dB? That’s my explanation, I
will be glad to hear yours!
WRAPPING UP
The experiments I presented in this article
don’t pretend to be science, they are just
experiments. Therefore they should be used
with caution. Anyway these tests show some
orders of magnitude. First a well-soldered
metallic shield easily provides a 20-to-
30-dB attenuation between two circuits,
except at very low frequencies and a close
distance where magnetic coupling is the main
contributor. Getting an attenuation higher
than 20 to 30 dB is significantly more difficult
as small leaks could happen everywhere.
Moreover, it is probably easier to work on
the circuits themselves (e.g., reducing the
signals speeds, improving the immunity of
the victim, etc.).
 circuitcellar.com 65

PHOTO 7
The test PCB is shown positioned in
an anechoic chamber (front), whereas
the receiving test antenna is at the
back.

COLUMNS
ABOUT THE AUTHOR
Robert Lacoste lives
in France, near Par-
is. He h as 25 years
of experience in em-
bedded systems,
analog designs, and
wireless telecommuni-
cations. A prize winner
in more than 15 inter-
national design con-
tests, in 2003 he start-
ed his consulting com-
pany, ALCIOM, to share
his passion for innova-
tive mixed-signal de-
signs. His book (Rob-
ert Lacoste’s The Dark-
er Side) was published
by Elsevier/Newnes in
2009. You can reach
him at rlacoste@alciom.
com. Don’t forget to put
“darker side” in the sub-
What about 3-D printers? They are great, will be a little more complex if you need 50-dB ject line to bypass spam
and I’m sure I will use my printer often for improvement. filters.
packaging, fixing parts, and so forth. However, Finally, as usual, I strongly encourage you
the use of conductive ABS filament alone to experiment. You probably don’t have an
doesn’t provide an effective shielding against anechoic chamber in your cellar, but your
RF fields, at least using the currently available neighbor’s field could be as adequate on a
filaments. Using conductive ABS filament and sunny day. If you don’t have an RF signal
conductive EMC paint enables better—but still generator you can easily build one. And if you
limited—results. With this solution you can don’t have a spectrum analyzer and don’t
expect around 10-dB attenuation, but nearly have enough spare time to build one, you can
no attenuation at a close distance above some still use a radio receiver or borrow one. You
200 MHz. The same result will be achieved have no excuse: you need to play with shields
by adding a couple of EMC paint layers in a to understand what’s going on. If not, well,
plastic casing. Don’t get me wrong: 10 dB is you will need to learn it the hard way the day
still a huge improvement and could solve your you have no other solution and a manager or
problem, but you need to know that your life customer is crying over your shoulder.
 66 CIRCUIT CELLAR • AUGUST 2014 #289
FROM THE BENCH
IR Remotes
(Part 1)
Remote Transmission
COLUMNS
By Jeff
Bachiochi (US)
W hen it comes to
audio and video
(AV) in dens, IR remote
controls still dominate.
Since most AV components
come with their own remote
control, our homes tend to
gather them like single socks on
laundry day. In fact, IR control can
be found accompanying many other
devices (e.g., lighting, ceiling fans, and air
conditioners). It’s as if someone wants us
to remain seated in our comfy chairs so we
wouldn’t miss the latest gadget being pitched
to our subconscious.
Somewhere along the line someone
suggested combining multiple controls
into a single remote. Universal remotes
have been around for years. The problem
is manufacturers continue to expand IR
codes for their equipment, which makes
older universal remotes incompatible. Yes,
some universal remote controls (URCs) are
upgradeable and, as long as the data format
can be described in established parameters,
this solves the problem.
Another problem is there are many
standards, and most manufacturers are
keeping their cards close to their vests. The
most complete information I’ve found about
IR control was collected by San Bergmans
What is the difference between
ambient IR and data-carrying IR?
This article series describes IR and
introduces IR formats. Part 1 focuses
on accumulating and recognizing an
IR remote transmission.
of the Netherlands. His website, SB-Projects,
has pages devoted to IR remote control
theory and includes an abundance of various
IR protocols. I’ll take a brief look at IR and
a simple protocol before delving into my
project, which begins by accumulating and
recognizing an IR remote transmission.
I THINK, THEREFORE IR
It may have been René Descartes who
first said this during his early work with
infrared control: Infrared is that part of the
electromagnetic spectrum (700 to 1,000 nm)
that is just above visible light (violet 380 to
red 700 nm). While we can’t see it directly, we
have sensors that can emit and detect it. LEDs,
like standard diodes, use semiconducting
materials doped with impurities to create a
p-n junction. LEDs can emit photons when
their p and n materials have a direct band
gap crystalline structure (e.g., the electron
or hole can cross the gap without a loss in
their momentum). The materials used affect
the bandgap (voltage drop) and a photon’s
wavelength (color) as shown in Table 1.
Photo diodes and photo transistors are
semiconductor devices that convert light into
current when photons are absorbed. Solar cells
are p-n photodiodes with large surface areas
and slow response times. A slightly modified
p-i-n junction improves a photodiode’s
 circuitcellar.com 67

Wavelength Voltage
Color Semiconductor material
(nm) drop (∆V)
Gallium arsenide (GaAs)
Infrared λ > 760 ∆V < 1.63
Aluminium gallium arsenide (AlGaAs)
Aluminium gallium arsenide (AlGaAs)
Gallium arsenide phosphide (GaAsP)
Red 610 < λ < 760 1.63 < ∆V < 2.03
Aluminium gallium indium phosphide (AlGaInP)
Gallium(III) phosphide (GaP)
Gallium arsenide phosphide (GaAsP)
Orange 590 < λ < 610 2.03 < ∆V < 2.1 Aluminium gallium indium phosphide (AlGaInP)
Gallium(III) phosphide (GaP)

COLUMNS
Gallium arsenide phosphide (GaAsP)
Yellow 570 < λ < 590 2.1 < ∆V < 2.18 Aluminium gallium indium phosphide (AlGaInP)
Gallium(III) phosphide (GaP)
Traditional green:
Gallium(III) phosphide (GaP)
Aluminium gallium indium phosphide (AlGaInP)
Green 500 < λ < 570 1.9< ∆V < 4
Aluminium gallium phosphide (AlGaP)
Pure green:
Indium gallium nitride (InGaN) / Gallium(III) nitride (GaN)
Zinc selenide (ZnSe)
Indium gallium nitride (InGaN)
Blue 450 < λ < 500 2.48 < ∆V < 3.7
Silicon carbide (SiC) as substrate
Silicon (Si) as substrate, under development

Violet 400 < λ < 450 2.76 < ∆V < 4 Indium gallium nitride (InGaN)

Diamond (235 nm)

Boron nitride (215 nm)
Ultraviolet λ < 400 3.1 < ∆V < 4.4 Aluminium nitride (AlN) (210 nm)
Aluminium gallium nitride (AlGaN)
Aluminium gallium indium nitride (AlGaInN)-down to 210 nm
TABLE 1
The relationship between an LED’s color, wavelength, bandgap, and materials used is shown. The IR LED’s junction has the lowest voltage drop. (Information courtesy of Wikipedia,
“Light-Emitting Diode”)

response time. When a photon of sufficient
energy strikes the diode, it creates an electron-
hole pair that separates in the material under
the effect of the depletion zone’s electric field,
thus generating a current. If the junction
exposed to light is part of a transistor, then
the current generated can be amplified by the
transistor’s beta (gain).
When an IR LED and a phototransistor are
to be used as an IR transmitter receiver pair,
they must be positioned so the transmitter’s
emitted energy is sufficient to reach the receiver
either directly or indirectly. For my purposes,
a 10’-to-20’-distance is adequate. The first
thing you may find is that a phototransistor
will pick up a wide range of frequencies from
the environment (see Figure 1). At times this
can be sufficient enough to corrupt the data
you are trying to send.
How can you discern the difference
between ambient IR and data-carrying
IR? By modulating the IR data with a
carrier frequency outside the interference
band, it takes on a unique signature. An
approximately 40-kHz modulation frequency
fills this requirement while staying under
the phototransistor’s frequency response.
Early remotes used inexpensive ceramic
resonators (i.e., 429/432/455 kHz) already
used in intermediate frequency (IF) stages of
AM radios. Since an IR controller’s modulation
was often a division of its time base, several
different carrier frequencies can be found.
While the first IR receivers were made of
discrete parts, today you can purchase IR
demodulators that contain the photo diode
amplifier (with automatic gain control) and
a demodulator in one three-pin package for
less than $1 (see Figure 2). Minimum (and
maximum) IR bursts are recommended for
these devices to operate correctly.
Return-to-zero data formats ensure that
long streams of continuous modulation are
avoided. Pulse position modulation (PPM)
 68 CIRCUIT CELLAR • AUGUST 2014 #289
FIGURE 1
This graph shows the response
of a silicon photo diode or 0.6 A/W
transistor vs wavelength of
the incident light. The device is
0.5 A/W
most sensitive in the 950-nm 0.4 A/W
IR range.
Response 0.3 A/W
0.2 A/W
0.1 A/W
0 and 5 can run directly from this 64-MHz source.
400 nm 600 nm 800 nm
Wavelength
COLUMNS
and Manchester encoding are two popular
techniques (see Figure 3). PPM uses a single
“10“ pulse for each bit. The duration of the
1 (mark or modulation ON) and 0 (space or
modulation OFF) times are the key to the bit
value. Typically, a bit value = 0 is represented
by a mark and space of equal time. A bit value
= 1 has a space time of twice the mark time.
In some cases, additional space times of three
and four times the mark time can be used to
indicate twin data bits (see Table 2).
Manchester encoded data is determined
by the state, mark, or space that is first
presented in each bit time. In Table 2, a
space preceding a mark equals a bit value = 1
and a mark before a space is a bit value = 0.
In many cases the data to be sent is
wrapped up for delivery. This wrapping is
simply some leading and trailing signal that
is distinctively different from the data. If you
could see an actual IR transmission, you would
see periods of modulated IR (mark) and periods
of nothing (space). The IR would be modulated
at an on/off rate of somewhere between 33 to
60 kHz. With a modulation rate of 40 kHz, you
might expect minimum mark and space times
of 1/10 the modulation rate or 4 kHz (250 µs).
The IR receiver/demodulator removes the 40
kHz, providing an output that is low during IR
transmission bursts (mark).
INFRA READ
So, how could you see these transmissions?
If you used one of the demodulating receivers,
you could look at the device’s reaction to
30 kΩ
FIGURE 2
Band
Many distributors sell inexpensive Input AGC Demodulator
and complete IR demodulators
pass
in single three-pin packages.
The demodulator packages are PIN
Control circuit
usually available in 30-to-56-kHz
carrier frequencies.
any IR remote sending a transmission. This
seemed like a good place to start.
I used a circuit I had previously built
consisting of a 28-pin microcontroller with all
the I/O brought out to a prototyping area. I
added an IR receiver and connected this to the
counter timer input on the microcontroller. I
used a Microchip Technology PIC18F26K22
microcontroller because it will run at 64 MHz
with its phase locked loop (PLL). Timers 1, 3,
1,000 nm Timers 2, 4, and 6 can provide a capture signal
from either an external rising or a falling edge.
This edge can also provide an interrupt trigger.
When the IR receiver changes state—a falling
edge indicating it is seeing a mark (modulated
IR signal) or a rising edge indicating a space
(loss of an IR modulation signal)—the interrupt
routine is called.
This input is either rising or falling. Once
the selected edge has triggered the interrupt
routine, one of the interrupt routine’s functions
must be to reconfigure the input to look for
the opposite edge the next time. I know an IR
transmission begins with IR modulation, so I
began looking for a falling edge (mark). The IR
receiver pulls its output low whenever it sees
an IR-modulated signal. When this happens,
the latched count (number of 64-MHz cycles
since the last interrupt) is saved.
If you reset the counter, all future counts
will be reference to a zero starting count.
Once the modulation ceases, the mark ends
and the receiver will permit its output to
return high. Since the microcontroller’s input
was reconfigured to look for a rising edge
during the last interrupt, the beginning of the
space (idle time with no modulation) is seen
as another interrupt. The interrupt routine
directs execution based on the selected edge,
so you can track the mark and space counts.
It is best to spend as little time in an interrupt
routine as possible so it doesn’t negatively
affect other interrupts. I let the main loop
process the data collected by these edge-
triggered interrupts.
STREAMING DATA
I’d like these counts delivered to my PC in
real time by sending the state and count out
through the microcontroller’s serial port. Going
by my earlier guesstimate of 250 µs per state,
I could send about four characters at 115,200
3
bps in that amount of time. That didn’t sound
good. I would need to send at least twice that.
1
So much for a real-time display.
At 64 MHz, each count would be
approximately 15 ns/count. A 250-µs burst
would therefore have a 250/0.015 count or
2 approximately 16,000. That’s a big number
just to indicate one bit time. I don’t need
anywhere near that kind of resolution.
When it
comes to
robotics,
the future
is now!
From home control systems to
animatronic toys to unmanned
rovers, it’s an exciting time to
be a roboticist. Advanced Control Robotics
simplifies the theory and best practices of
advanced robot technologies, making
it ideal reading for beginners and experts
alike. You’ll gain superior knowledge of
embedded design theory by way of
handy code samples, essential schematics,
and valuable design tips.

With this book, you’ll learn about:

• Communication Technologies
• Control Robotics
• Embedded Technology
• Programming Language
• Visual Debugging... and more

Get it today at ccwebshop.com.

 70 CIRCUIT CELLAR • AUGUST 2014 #289

FIGURE 3
a) 00000000 1 1 1 1 1 1 1 1 1 0 1 1 0 1 0 1 0 1 00 1 0 1 0
a—This message frame uses the NEC
IR transmission protocol with pulse
position encoding. b—This message
frame uses the Philips RC5 IR
transmission protocol with Manchester
encoding.
LSB LSB LSB LSB
9 ms Address Address Command Command
4.5 ms (Logical inverse) (Logical inverse)

27 ms 27 ms

67.5 ms

b)
Bit 1 Bit 2 Bit 3 Bit 4 Bit 5 Bit 6 Bit 7 Bit 8 Bit 9 Bit 10 Bit 11 Bit 12 Bit 13 Bit 14
COLUMNS

1 1 0 0 0 1 0 1 1 1 0 1 0 1

S1 S2 T MSB MSB

Start bits Toggle Address Command

bit
5.334 ms 8.89 ms 10.668 ms

24.892 ms

Choosing a slower clock rate (1 µs) will still
provide plenty of resolution (250 counts per
bit). I can use a word register to handle counts
up to 64,535 (µs). If I’m not trying to display
the data in real time, I should have plenty
of time to store it away in a buffer for later
display. The data format is 3 bytes: ASCII
“M” or “S” to indicate the state, followed by a
2-byte count.
The buffer will then have to be big enough
to store at least one complete transmission. At
6 bytes for each mark space pair transmitted,
I can store a little more than 40 bursts in a
bank of RAM (256 bytes each). I would need
to set aside a number of banks for this buffer.
There is a potential problem with this
approach. I chose to use the counter timer
for its ability to latch a timer count on edge
interrupts so I could get the fastest and
most accurate counts possible. This wasn’t a
big issue since I really didn’t need the high
resolution this provides for this particular
application. I needed to reconfigure the
peripheral after each interrupt. If I were to
Mark Space
Duration Duration
1 time unit 1 time unit
1 time unit 2 time units
TABLE 2
1 time unit 3 time units
A bit value is determined by the time
unit’s data value and mark and space 1 time unit 4 time units
durations.
miss an edge, I would really miss two edges
and not know it! So, I used two external
interrupt pins tied together instead of the
counter/timer input. I could configure the
inputs for opposite edges and therefore be
able to detect a missing count should two of
the same edges be stored in a row.
Since I plan to use the PC to display sampled
data, I might as well enable user input to begin
the sampling process. I like using the serial
port to aid in debugging, so it made sense to
provide transmit and receive buffers to send
and receive data in the background.
One of the handy aspects of this
microcontroller is the two-level priority for
interrupt sources. Since my timing information
is the most important, I don’t want any serial
port activity preventing the accurate timing
of IR data. I configured the UART interrupts
as low-level interrupts. High-priority timing
interrupts can interrupt a low-priority
interrupt. This way moving received characters
into the RXBuffer ensures no characters are
missed, while enabling important timing
Data routines to take over if necessary.
I used the single character “S” to begin
Value the sampling process. A Prompting message
is displayed; however, nothing actually begins
00
until the IR receiver sees its first IR burst
(the first falling edge is seen). At that point,
01 the timer will begin and successive edges
will store the timer count, restart the timer,
10 and enable the sample to be stored in the
SampleBuffer. The routine will continue until
11 the buffer is filled or the timer overflows (a
lack of data). A Finished message concludes
Join the Elektor Community
Take Out a GOLD Membership Now!
Your GOLD Membership Contains:
• 8 Regular editions of Elektor magazine in
print and digital
• 2 Jumbo editions of Elektor magazine in print
and digital (January/February and July/August
double issues)
• Elektor annual DVD-ROM
• A minimum of 10% DISCOUNT on all
products in Elektor.STORE
• Direct access to Elektor.LABS; our virtual,
online laboratory
• Direct access to Elektor.MAGAZINE; our online
archive for members
• Elektor.POST sent to your email account
(incl. 25 extra projects per year)
• An Elektor Binder to store these projects
• Exclusive GOLD Membership card

ALSO AVAILABLE:
The all-paperless GREEN Membership, which
delivers all products and services, including
Elektor magazine, online only.

Take Out Your Membership Now at www.elektor.com/member

Connect with us! www.facebook.com/elektorim www.twitter.com/elektor
 72 CIRCUIT CELLAR • AUGUST 2014 #289

(the first translation is mark approximately

Element Time in Microseconds 8,794 µs, so a transition is added from mark
to space)
0x4D (M) 0x22 0x55 (8,794) ¦A
1,8794<CR> ¥A
0x53 (S) 0x11 0x53 (4,435) ¦B 0,8794<CR>

(the next translation is space approximately

0x4D (M) 0x01 0x9A (410) ¦C 8,794 µs + 4,435 µs, so a transition is added
from space to mark)
0x53 (S) 0x02 0x01 (513) ¦D
0,13229<CR> ¥B
0x4D (M) 0x01 0x9A (410) ¦E 1,13229<CR>
COLUMNS

0x53 (S) 0x02 0x01 (513) ¦F
TABLE 3 the operation. Now that I have some data, I
This stored data is the result of need to display it.
performing a translation and sending
the data to the serial port for each
element.
A LOOK-SEE
In the back of my mind I kept thinking,
“comma delimited.” The plan was to make
visual sense of the sampled data by using a
tool (e.g., Excel) to graph the data. I found I
was spending a lot of time trying to create
meaningful graphs when what I really needed
was a digital storage scope display. Fortunately,
I quickly understood I needed to provide an
additional data point for each sample.
I had saved a state and time for each
element and I needed the transition between
states. The C command became a translation
from the stored data to comma-delimited
plottable points. For each stored element, I
performed a translation and sent the data
out the serial port. Table 3 shows the stored
data. It gets translated as:
(The initial transition @ 0.0 µs from space to
mark)
0,0<CR>
1,0<CR>
G. Dixon, “IRP Notation: Specification of IRP
Notation.”
SB-Projects, “ IR Remote Control Theory,” www.
sbprojects.com.
circuitcellar.com/ccmaterials SOURCES
PIC18F26K22 Microcontroller
Microchip Technology, Inc. | www.microchip.com
REFERENCE
[1] M. Bellis, “History of the
RealTerm terminal program
Television Remote Control,”
SourceForge | www.sourceforge.net
About.com.
RESOURCES
DatPlot, www.datplot.com.
(the next translation is mark approximately
13,229 µs + 410 µs, so a transition is added
from mark to space)
1,13639<CR> ¥C
0,13639<CR>
(etc.)
0,14152<CR> ¥D
1,14152<CR>
1,14562<CR> ¥E
0,14562<CR>
0,15075<CR> ¥F
1,15075<CR>
This “C”hart data can be captured into
a file by the terminal program. (I used
SourceForge’s RealTerm.) The comma-
delimited format is imported with Excel, or
you can use DatPlot, a free plotting program.
From this plot you can make out a wide header
followed by some bits and a large trailing gap
between the data and the start of what could
be another burst (see Figure 4).
What now? How do you know what to look
for and how to identify it and decode its data?
In some instances, you don’t need to know
any of that. If all you want to do is reproduce
the signal, you have enough information to do
that. But what fun would that be? Let’s take a
look at a few of the known formats and see if
this can be identified.
IR FORMATS
If you’ve ever used ultrasonics to measure
distances, this carrier frequency is probably
familiar to you. According to Mary Bellis’s
article “History of the Television Remote
Control,” in the early 1960s Zenith’s Space
Command remote control reaped the benefits
of transistor technology using ultrasonics
to create a small hand-held and battery-
operated remote control. More than 9
million ultrasonic remote controls were sold.
Ultrasonic remote controls ruled living rooms

until the early 1980s when they were replaced
by IR remotes.[1]
Changing batteries in a remote is a nuisance
every user despises. IR transmissions should
be as short as possible to conserve precious
reserves. Understandably, the first remotes
used whatever techniques would provide the
longest life. With so many formats and no real
standards, how can you keep track of what is
out there?
You may think this has already been
done. After all, universal remotes can mimic
anything, right? All you have to do is punch
in some four-digit number and your URC
knows what to do. The fact is that each URC’s
dictionary is unique. A four-digit code used in
one remote is not necessarily the right one
in another remote. However, an open solution
has been implemented.
IRP NOTATION
In 2003 the IRP notation was developed by
John Fine as a way of describing IR protocols.
The IRP notation has been illustrated in great
detail by Graham Dixon (see Resources).
The overall structure of IRP notation
is contained in as little as three sections:
GeneralSpec, Bitspec, and IRStream. If
required, a fourth section may include the
definition of any expressions previously used
in the notation. Let’s take a look at the IRP
notation for a Sharp device.
The GeneralSpec section includes the
carrier frequency (hertz default), whether
the data is presented most-significant bit or
least-significant bit (default){ first, and any
basic time unit (microsecond default). This
section is surrounded by {}:
{38k,400} ‘ 38,000 Hz, default, 400 µs
The Bitspec section describes the format
used to represent each bit mark and space,
where “–” means space. This section is
surrounded by “<>” (with each bit separated
by “|”) and follows the form:
<1,–1| ‘ bit0 =
1 × 400 µs mark and 1 × 400 µs space |
1,–3> ‘ bit1 =
1 × 400 µs mark and 3 × 400 µs space
The IRStream section describes the
data for each subsection, header, data, and
trailer. The header and trailer (when used)
are described in terms of bit(s), while the
data subsection may have a combination of
constant:bit quantity and/or expression:bit
quantity. A preceding “~” indicates the
value is complemented and “+” means the
transmission may be sent multiple times.
The bitstream is surrounded by “()”. This
circuitcellar.com 73
COLUMNS
FIGURE 4
The IR remote sampled for this plot is an Acer multimedia remote that came with my Acer tablet.
(_ D:5,F:8,1:2, 1,-165,_ D:5,~F:8,2:2,1,-165)+
l l l l l l L - + = may be sent again
l l l l l L - trailer = 1 * 400uS Mark and 165 * 400uS Space
l l l l L - data = xxxxx xxxxxxxx 10
l l l L - header = none
l l L - trailer = 1 * 400uS Mark and 165 * 400uS Space
l L - data = xxxxx xxxxxxxx 01
L - header = none
FIGURE 5
The IRP Notation’s IRStream section provides information about each subsection.
data subsection includes two transmissions
with differing data in each (see Figure 5). ABOUT THE AUTHOR
Note the data designations D:5, F:8, and J e f f B a c h i o c h i ( p ro -
1:2. The letters are an indication of some nounced BAH-key-AH-
variable or expression (:) of a particular bit key) has been writing
length. The number indicates a constant for Circuit Cellar since
value (:) of a particular bit length. The colon 1988. His background
precedes an indicator of data size (in bits). includes product design
Most protocols can be described using this and manufacturing.
notation. Even though IRP offers a way for an
You can reach him at
application to evaluate the notation, it does
jeff.bachiochi@imagine
not describe the actual original button code
thatnow.com or at www.
for each key on a remote. While the variable/
imaginethatnow.com.
expressions in the data subsection of the
previously described IRStream tell you where
that data goes, this data needs to be defined
somewhere. Again, there is no standard for
this. In fact, it may be harder for you to locate
this information than to identify the format of
a particular IR transmission.
TO DO LIST
Fortunately, with this project once you’ve
identified the IR format, you can decode this
data for various key presses. That’s what I’ll
be doing in Part 2 of this article series:
identifying, decoding, and reproducing the IR
I captured and described in Part 1.
 74 CIRCUIT CELLAR • AUGUST 2014 #289

CROSSWORD AUGUST 2014

The answers will be available at circuitcellar.com/crossword.

1 2

4 5

9 10 11 12

13
TESTS & CHALLENGES

14 15

16

17

18

19

EclipseCrossword.com

ACROSS
1. Ethernet at 100 Mbps DOWN
3. IC factory 2. Wafer or substrate
6. Logic gate 3. Antonym for FIFO
7. HDL created in the early 1980s by Goel and Moorby 4. Base-5
9. 0.001 inches = 25.4 what? 5. CODEC [two words]
10. Current flow over time [two words] 7. Gilbert’s static-detection device
13. Greek national soccer team manager with a degree in 8. Release heat
electrical engineering 11. Relates to touch
14. Quad, flat, ... [two words] 12. 1 watt second
16. Step-down 15. Process of purposely adding impurities
17. 0.000000000000001
18. Ground pin
19. Zero
 circuitcellar.com 75

What’s your EQ? The answers to the Circuit Cellar 288

EQ are posted below and at www.circuitcellar.com/
category/test-your-eq/. You can contact the quizmasters at
eq@circuitcellar.com.

ANSWER 1
The cutoff frequencies of the two filters should be the same so
TEST YOUR EQ
that the overall frequency response when the filter outputs are Contributed by David Tweed
recombined is flat and has no phase shift. For example, if you feed the
cutoff frequency into both filters and then combine the results again,
Energizer Duracell
the output will be the same level as the input (0-dB overall gain). As
If the stereo is running for 12 h on a set of batteries, based on
long as the “order” of the two filters is the same (they have the same
eyeballing these graphs, it’s probably getting about 8 A-h of capacity
roll-off slope), the gain will be flat across the entire transition band
out of one set, so it’s drawing about 660 mA on average. Putting
of frequencies.
three sets of batteries in parallel would drop the current in each set
What’s really going on is this: A filter’s –3-dB point is where the
to about 220 mA, and it will get something closer to 12 A-h from
output has half the power of the input signal, which means that the
each set.
output voltage is 1/sqrt(2) times the input voltage. The –3-dB point
In other words, if you use, say, three sets of batteries sequentially,
is also where the output signal is phase shifted by 45°. It lags by 45°
you’ll get 36 h of playing time (24 A-h total), but if you use them in
in the low-pass filter and leads by 45° in the high-pass filter. This
parallel together, you’ll get something closer to 54 h of playing time
means that the outputs of the two filters have a total phase shift of
(36 A-h total).
90° relative to each other.
When you add two sinewaves that have the same amplitude and
a 90° phase shift, you don’t get double the voltage. You get sqrt{2}
times the voltage. You also get a waveform that has a phase midway ANSWER 3
between the two signals being added. A capacitor consists of two flat conductors separated by a dielectric.
So, the final amplitude is sqrt{2}/sqrt{2} times the original input Aluminum foil is an obvious candidate for the conductors, and either
voltage, and the final phase is midway between 45° and –45°, or 0°. waxed paper or plastic food wrap would be suitable dielectrics —
In other words, you get the original sinewave back exactly. they have similar characteristics.

ANSWER 2 ANSWER 4
In general, batteries provide greater capacity at lower average You need to do some basic calculations first. The formula for
currents. This is partly due to the battery’s internal chemistry, but capacitance is:
largely due to the simple fact that less power is wasted in the internal εR ε 0 A
resistance of the battery. C=
d

TESTS & CHALLENGES

Here are two graphs taken from two different datasheets that where:
illustrate this.
εR is the relative permittivity of the dielectric
ε0 is the permittivity of free space
A is the area of one plate
d is the separation between the plates

Let’s say you want to use 1-mil (25.4 µm) waxed paper as a
dielectric. Note that this will determine the voltage rating of the
capacitor. The dielectric strength of waxed paper is about 35-40
MV/m, so this will give you a capacitor that can theoretically handle
almost a kilovolt, but be conservative in how you use it!

The relative permittivity of waxed paper is about 3.7, the

permittivity of free space is 8.854e-12 F/m. Solve for the area
required:

C×d
A =
εR ε 0
5 µF × 25.4 µm
A= = 3.8767 m2
3.7 × 8.854 × 10 12 F/m

If you get aluminum foil and waxed paper that’s about 12” (30 cm)
wide, you can probably get an overlap of, say, 25 cm, which means
that you’ll need a length of about 15.5 m to get the area you need.
If you then roll up your capacitor (using a second layer of waxed
paper), the capacitance will be doubled, or about 10 µF. Obviously,
this will be physically rather large, more than a foot long and several
inches in diameter.
Plastic food wrap has a similar dielectric constant and dielectric
strength as waxed paper, but typically comes in a 0.5-mil (12.7 µm)
thickness. A capacitor using this would have about half the voltage
rating and about half the overall volume.
 76 CIRCUIT CELLAR • AUGUST 2014 #289

CC SHOP

ADVANCED CONTROL
ADVANCED CONTROL ROBOTICS

ROBOTICS HANNO SANDER

1 ADVANCED ROBOTIC
CONTROL—NEW!
When it comes to robotics, the future
is now! This book simplifies the theory
and best practices of advanced robot
technologies by providing insight to
communication technologies, control
robotics, embedded technology,
programming, and more. Complete with
code samples, schematics, and design
tips. Great reading for beginners and
experts alike.

PARALLAX PROPELLER
Author: Hanno Sander
Item#BK-ELNL-978-0-963013-33-0

1
4

2 CC VAULT
CC Vault is a pocket-sized USB
that comes fully loaded with ev-
ery issue of Circuit Cellar maga-
zine! This comprehensive archive
provides an unparalleled amount
of embedded hardware and soft-
ware design tips, schematics, and
source code. CC Vault contains
all the trade secrets you need to
become a better, more educated
electronics engineer!
Item #: CCVAULT
2 Other projects demonstrate some of
3 ANDROID APPS:
n Schwark Stefan Schwark
4 ADUC841 MICRO-
CONTROLLER DESIGN MANUAL
This book presents a comprehen-
sive guide to designing and program-
ming with the Analog Devices ADuC841
microcontroller and other microcon-
trollers in the 8051 family. It includes a
set of introductory labs that detail how
to use these microcontrollers’ most
standard features, and includes a set
of more advanced labs, many of which
make use of features available only on
the ADuC841 microcontroller.
The more advanced labs include
several projects that introduce you to
ADCs, DACs, and their applications.
the many ways you can use a micro-
controller to solve practical problems.
The Keil µVision4 IDE is introduced
early on, and it is used throughout the

book. This book is perfect for a univer-

Android Apps PROGRAMMING STEP-BY-STEP
Stefan Schwark

sity classroom setting or for indepen-

dent study.
programming step-by-step Many smartphones and tablet com-
+ Eclipse development platform
+ JAVA object oriented programming puters are powered by an Android OS.
Author: Shlomo Engelberg
These portable devices’ speed and com-
www.elektor.com/android Item #: CC-BK-9780963013347
omputers.
puting power enable them to run appli-
hat allow
cations that would have previously re-
Android: programming step-by-step

ncreasing

quired a desktop PC or custom-designed

to be run
go. Mean-
eos – and Further information and ordering
hardware. Android Apps introduces you
f the shelf
s is easier
to the programming required to design www.cc-webshop.com
eration of
apps for Android devices. Operating the CONTACT US:
nal appli-

Android system is explained step-by- Circuit Cellar, Inc.

111 Founders Plaza, Suite 300
examples,
ight up to
n the Java step to show how personal applications
med using

can be easily programmed. East Hartford, CT 06108

USA
Author: Stefan Schwark Phone: 860.289.0800
Item #: BK-ELNL-978-1-907920-15-8 Fax: 860.461.0450
3
26.11.12 15:38
E-mail: custservice@circuitcellar.com
 12th International System-on-Chip (SoC)
Conference, Exhibit & Workshops
October 22 & 23, 2014
University of California, Irvine - Calit2

www.SoCconference.com
CALL FOR SPEAKERS AND SPONSORS. . . Don’t Miss Out!
This Year’s Theme: “Innovative SoCs Empowering the Communications Market.”

Platinum Sponsors
 FinFET Technology & Design  Complex Mixed-Signal SoCs
 Analog & Mixed-Signal Designs  SOI vs. CMOS
 Sub 14nm Designs & Beyond  RF Design
 3-D ICs Designs  FPGAs –Trends & Designs
 IC Security & Challenges  High-Speed I/Os
 Multicore Software Development  Smarter Mobile Devices
 SoC Design & Verification  Multicore SoC Platforms
 Innovative EDA Tools  Network-on-Chips (NoCs)
 Complex IP Subsystems  Informative Panels
 Low-Power Techniques  IEEE Student Design Contest
 Memory Trends & Technologies  Networking Opportunities
 Table-Top Exhibit (Free Passes)  And Much More. . .

Promote Your Technology, Products & Services at

The Most Informative, Targeted & Educational
IC & IP Design Conference, Exhibit & Workshops of the Year!

Keynote Speakers

Microsemi Skyworks Solutions

Jim Aralis, Chief Technology Dr. Peter L. Gammel,    
Officer (CTO), and Vice Chief Technology Officer
President of R&D. (CTO).

The SoC Conference brings together a very targeted and sophisticated audience from leading-edge
technology companies as well as key universities for two exciting days devoted to groundbreaking
products, technologies, and business disclosures, focused exclusively on designing complex
System-on-Chip and the related technologies driving the SoC/ASIC/ASSP/FPGA/Foundry industry.

For More Information or Questions, Please Contact the SoC Conference Organizing Committee at:
SoC@SoCconference.com or (949) 851-1714

www.SavantCompany.com & www.SoCconference.com

78 CIRCUIT CELLAR • AUGUST 2014 #289

IDEA BOX the directory of

PRODUCTS & SERVICES

AD FORMAT: The Vendor Directory at

circuitcellar.com/vendor
Advertisers must furnish digital files that meet our specifications
(circuitcellar.com/mediakit). is your guide to a variety of
engineering products and
ALL TEXT AND OTHER ELEMENTS MUST FIT WITHIN A 2” x 3” FORMAT.
services.
E-mail adcopy@circuitcellar.com with your file.

For current rates, deadlines, and more information contact Peter Wostrel
at 978.281.7708 or circuitcellar@smmarketing.us.

“Your reliable
outsourcing partner”
LOW cost
with HIGH quality
PCB, PCBA and More
INSTANT QUOTE AT:
www.myropcb.com
OR CALL:
1-888-PCB-MYRO

MyRO

LEARN FPGA PROGRAMMING

USING THE ALTERA
QUARTUS II AND VERILOG
The course comes
with a Cyclone
board, as shown,
and an USB
Blaster. Also
included is a
175-page manual
containing ten
chapters that
are full of examples and tutorials. This
in-depth course is a must for the hobbyist,
as well as, for schools teaching FPGAs. The
manual is easy to understand and correct.

$55 + $12 shipping

ADVANCED MICRO CIRCUITS
443-817-2557
www.amcmicros.com
information@amcmicros.com
 circuitcellar.com 79

Join The

Save 10%
Online Web-Order Code:
C311408
Valid thru September 1, 2014
HRXL-MaxSonar®-WR™
- High noise tolerance
- IP67 rated
- 1 mm resolution
- Multi-Sensor operation
- Calibrated beam pattern
- Starting at $109.95

XL-MaxSonar®-EZ™
- Great for UAV’s and robotics
- Incredible noise
immunity
- Small in size
- 1cm resolution
- Automatic calibration
- Starting at $39.95
Phone: 218-454-0766 Email: sales@maxbotix.com
www.maxbotix.com

Real $349.00
TRi
Programmable Logic Controllers from
Powerful & Easy Ladder

Logic Analyzers
+BASIC Programming
Ethernet integrated
MODBUS TCP/IP
DI/Os & AI/Os integrated with Decoders & Plug-in Kit included
• 9 to 36 Channels
• 100 Msps to 400 Msps
OEM Prices as low as $119
for full-feature Nano-10 PLC
• Store & Forward Architecture
• Real-time Smart Compression
tel : 1 877 TRI-PLCS • Plot & Measurement Features
• Powerfull Search Tools
web : www.triplc.com/cci.htm

• Plug-ins for full Custom

Protocol integration
TRIANGLE
RESEARCH
by
INTERNATIONAL
www.tech-tools.com/real.htm • (972) 272-9392

Connect
with engineers
microEngineering Labs, Inc.
www.melabs.com 888-316-1753
Programmers for Microchip PIC® Microcontrollers
seeking your PC-Tethered USB Model (shown):
design solutions. •Standalone software
•Command-line operation
Magazine, e-newsletters, •Hide GUI for automated use
•Override configuration with drop-downs
website, advertorials, and more.
RESERVE ADVERTISING Stand-Alone Field Programmer:
•Power from target device or adapter
SPACE TODAY!
•Program file stored on SD-CARD
Strategic Media Marketing, LLC •Programming options stored in file

978.281.7708
•Single-button operation
Starting at $79.95

Program in-circuit or use adapters for unmounted chips.

circuitcellar@smmarketing.us Zero-Insertion-Force Adapters available for DIP, SOIC, SSOP, TQFP, and more.
www.smmarketing.us PIC is a registered trademark of Microchip Technology Inc. in the USA and other countries.
 80 CIRCUIT CELLAR • AUGUST 2014 #289

Self-Reconfiguring Robotic
Systems and M-Blocks
By Kyle Gilpin
The creation of autonomous machines capable
of shape-shifting has been a long-running dream of
scientists and engineers. Our enthusiasm for these
self-reconfiguring robots is fueled by fantastic science
fiction blockbusters, but it stems from the potential
that self-reconfiguring robots have to revolutionize our
interactions with the world around us.
Imagine the convenience of a universal toolkit that
can produce even the most specialized tool on demand
in a matter of minutes. Alternatively, consider a piece
of furniture, or an entire room, that could change its
configuration to suit the personal preferences of its
occupant. Assembly lines could automatically adapt to
new products, and construction scaffolding could build
itself while workers sleep. At MIT’s Distributed Robotics
Kyle Gilpin, PhD, is a Postdoctoral
Lab, we are working to make these dreams into reality
Associate in the Distributed
through the development of the M-Blocks.
Robotics Lab at the Massachusetts
The M-Blocks are a set of 50-mm cubic modules
TECH THE FUTURE
magnetic pole. The benefit of this arrangement is that
as two modules are brought together, the magnets will
automatically rotate to attract. Furthermore, as one
and then two additional M-Blocks are added to form a 2
× 2 grid, the magnets will always rotate to realign and
accommodate the additional modules.
The same cylindrical magnets that bond neighboring
M-Blocks together form excellent pivot axes, about which
the modules may roll over and around one another. We
have shown that the modules can climb vertically over
other modules, move horizontally while cantilevered
from one side, traverse while suspended from above,
and even jump over gaps. The permanent magnet
connectors are completely passive, requiring no control
and no planning. Because all of the active components of
an M-Block are housed internally, the modules could be
hermetically sealed, allowing them to operate in extreme
environment where other robotic systems may fail.

Institute of Technology (MIT)

capable of controlled self-reconfiguration. Each M-Block
where he is collaborating with
is an autonomous robot that can not only move
Professor Daniela Rus and John
independently, but can also magnetically bond with
Romanishin to develop the
other M-Blocks to form larger reconfigurable systems.
M-Blocks. Kyle works to improve
When part of a group, each module can climb over and
communication and control in
around its neighbors. Our goal is that a set of M-Blocks,
large distributed robotic systems.
dispersed randomly across the ground, could locate one
Before earning his PhD, Kyle spent
another and then independently move to coalesce into a
two years working as a senior
macro-scale object, like a chair. The modules could then
electrical engineer at a biomedical
reconfigure themselves into a sphere and collectively
device start-up. In addition to
roll to a new location. If, in the process, the collective
working for MIT, he owns a
encounters an obstacle (e.g., a set of stairs to be
contract design and consulting
ascended), the sphere could morph into an amorphous
business, Crosscut Prototypes. His
collection in which the modules climb over one another Two different views of the M-Blocks show the internal flywheel
past projects include developing (partially obscured, left and right), which is driven by the
to surmount the obstacle. Once they have reached their
cellular and Wi-Fi devices, real- brushless motor (orange, right). A self-tightening band brake,
final destination, the modules could reassemble into a
time image processing systems, driven by a servomotor horn (white, left), quickly decelerates the
different object, like a desk.
reconfigurable sensor nodes, flywheel so that one module pivots about the other using the
The M-Blocks move and reconfigure by pivoting
robots with compliant SMA cylindrical magnets held captive in the modules’ edges. A PCB
about their edges using an inertial actuator. The energy (green, left), controls the flywheel and the brake using commands
actuators, integrated production
for this actuation comes from a 20,000-RPM flywheel received from the on-board 2.4-GHz radio. On-board lithium-
test systems, and ultra-low-
contained within each module. Once the motor speed polymer batteries (not shown) provide power to each module.
power sensors. You can contact
has stabilized, a servomotor-driven, self-tightening band
him at either kwgilpin@csail.mit.
brake decelerates the flywheel to a complete stop in 15 While we have made significant progress, many
edu or kwgilpin@xcutproto.com.
ms. All of the momentum that had been accumulated in exciting challenges remain. In the current generation
the flywheel is transferred to the frame of the M-Block. of modules, there is only a single flywheel, and it is
Consequently, the module rolls forward from one face fixed to the module’s frame, so the modules can only
to the next, or if the flywheel velocity is high enough, it move in one direction along a straight line. We are close
rapidly shoots across the ground or even jumps several to publishing a new design that enables the M-Blocks to
body lengths through the air. (Refer to www.youtube. move in three dimensions, makes the system more
com/watch?v=mOqjFa4RskA to watch the cubes move.) robust, and ensures that the modules’ movements are
While the M-Blocks are capable of independent highly repeatable. We also hope to build new varieties
movement, their true potential is only realized when of modules that contain cameras, grippers, and other
many modules operate as a group. Permanent magnets specialized, task-specific tools. Finally, we are
on the outside of each M-Block serve as un-gendered developing algorithms that will allow for the coordinated
connectors. In particular, each of the 12 edges holds control of large ensembles of hundreds or thousands of
two cylindrical magnets that are captive, but free modules. With this continued development, we are
to rotate, in a semi-enclosing cage. These magnets optimistic that the M-Blocks will be able to solve a
are polarized through their radii, not through their variety of practical challenges that are, as of yet,
long axes, so as they rotate, they can present either largely untouched by robotics.
MULTIPLE PCBA
NO MORE WAITING
LINES
ACROSS THE GLOBE!
Imagineering utilizes the Jet Printer Technology which has been proven
successful in five continents. This breakthrough innovation allows us to achieve
high-precision solder deposits at speeds of more than one million dots per hour.
High-Speed Non-Contact Jetting Nozzle • Complete Volume Control • 100% Software Driven
More Components, More Possibilities • High-Performance Platform

www.PCBnet.com
(847) 806-0003 • sales@PCBnet.com