Elektor USA - January, February 2017

Edition 1/2017 | January & February 2017
www.elektormagazine.com
LEARN DESIGN SHARE
Sand Clock
Cool Power
with D-Watt
A high-performance
amplifier with lots of
output power
a real
CAD for PCBs
eye-catcher
Free and non-free
PCB layout software Analog Random Numbers • AxiCat • BrainBox Arduino • Build a Chipamp!
• CAD for PCBs • Catdoku • Color LEDs • Connect Objects with Genuino
101 • Cool Power with D-Watt • DAB+ Antenna Diplexer • Debugging
the Arduino Zero & M0 Pro • electronica Fast Forward Award 2016 •
Elektor Labs Pipeline • Elektor SDR Reloaded (4) • Err-lectronics • The
Fetron • Retronics: From Verobox to Heavy Metal • Internet Radio with
Fluorescent Display (2) • IoT Shield for Arduino • LED Stairway Lighting •
LEGO Control Board for the Raspberry Pi • MAXREFDES99# Display Driver
Shield • The Mendocino Motor • Opamp Power Supply Connexions •
AxiCat
Link an IC to a Programming Adapter for USBasp • Sand Clock • Sensors make Sense (2)
Linux or Windows PC • Simple Interfacing to Analog and Digital Position Sensors • Simulation
with SystemVision • Super Simple Dice • SUPRA 2.0 MC Version
► More than 45 years
of experience
► 24-hour shipping
► More than 70,000 products
THE REICHELT
ADVERT
rch.lt/vuk
Photo : Foto- und Bilderwerk
Arduino microcontrollers FIND OUT
make it easy for you to turn MORE ►
your ideas into reality!
The most popular Arduino!

ARDUINO UNO REV. 3
ATMega 328
14.25
14 digital I/Os (6 as PWM)
6 analog inputs ARDUINO UNO
USB port
Small but not to be underestimated!

ARDUINO™ MICRO
Just connect and you’re ready to go!
ARDUINO™ MEGA 2560 A LL
ARDUINO
ATMega 32u4
ATMega 2560 PRODUCTS 12 analog inputs
54 digital I/Os (14 as PWM) Micro USB port
16 analog inputs
USB port
rch.lt/yb
ARDUINO MICRO 13.52
2
Small but mighty!

ARDUINO™ NANO
ATMega 328
8 analog inputs
Mini USB port
ARDUINO MEGA 25.50 ARDUINO NANO 16.73

Daily prices! Price as of: 01. 12. 2016 Prices in £ plus statutory VAT, plus shipping costs · reichelt elektronik, Elektronikring 1, 26452 Sande (Germany) Onlineshop languages:
www.reichelt.co.uk
PAYMENT METHODS:
SHOP CONVENIENTLY ONLINE! ORDER HOTLINE: +49 (0)4422 955-360

Elektor Magazine
Edition 1/2017
Volume 43, No. 481 & 482
January & February 2017
ISSN 1757-0875 (UK / US / ROW distribution)
www.elektor.com
www.elektormagazine.com
Elektor Magazine, English edition

is published 6 times a year by
Creative repair work
Elektor International Media
78 York Street
London W1H 1DP I would argue with my learned colleagues in the labs that repairing or restoring a circuit
United Kingdom can be as much fun as creating one from scratch, and the activity overall is equally edu-
Phone: (+44) (0)20 7692 8344 cational potentially. Both activities are fraught with pitfalls and frustration but eventually
should have the same moment of triumph as there is little difference between “hey it
Head Office:
works” and “hey it works again”, and I’ve been either way.
Elektor International Media b.v.
PO Box 11 The one comforting and reassuring thought when repairing a complex piece of equip-
NL-6114-ZG Susteren ment like a Tek 2235 ‘scope that’s as dead as a doornail must be this: it has worked for
The Netherlands years and now there’s one or two parts in it thwarting its normal operation and hiding
Phone: (+31) 46 4389444 out for me. Which one is it? While hunting for that one flakey part (of 1202 in principle)
Fax: (+31) 46 4370161
I have to keep in mind all the time that the electronic design of the ‘scope will have
imperfections (even the Tek), but not so many that it has never worked.
Memberships: By contrast, sitting in front of his empty (CAD) screen, the creator of a new circuit has
Please use London address no other beacons to navigate by than his/her own intuition or strong conviction that the
E-mail: [email protected] circuit will work. Some day — some year — possibly helped by others who were there
www.elektor.com/memberships
before, successfully or not.
Both the creator and the repairman have to exercise their creative minds to overcome
Advertising & Sponsoring: one obstacle after another and remain on the (hopefully) straight path to victory. After
Johan Dijk
all, the relevant activities and skills are complementary rather than incongruent. Like so:
Phone: +31 6 15894245
many repair fans create fantastic workarounds to replace dead, obsolete parts and even
E-mail: [email protected]
manage to enhance existing circuitry; and the creative genius has to repair his/her proj-
www.elektor.com/advertising
ect silently after a less fortunate choice of, say, a zener voltage or being “misguided” on
Advertising rates and terms available on
request. the pinout of a BS170P.
Which leads me to the circuits published in this first edition of 2017: perfectly presented
as they may seem — they are forever subject to improvement. It’s meant to be that
way. Let me know how you fared.
Copyright Notice
Happy hooraying,
The circuits described in this magazine are for domestic
and educational use only. All drawings, photographs, Jan Buiting, Editor-in-Chief
printed circuit board layouts, programmed integrated
circuits, disks, CD-ROMs, DVDs, software carriers, and
article texts published in our books and magazines
(other than third-party advertisements) are copyright
Elektor International Media b.v. and may not be repro-
duced or transmitted in any form or by any means,
The Circuit
including photocopying, scanning and recording, in
whole or in part without prior written permission from Editor-in-Chief: Jan Buiting
the Publisher. Such written permission must also be Deputy Editors: Thijs Beckers, Clemens Valens
obtained before any part of this publication is stored
Translators: David Ashton, Jan Buiting, Martin Cooke,
in a retrieval system of any nature. Patent protection
may exist in respect of circuits, devices, components
Ken Cox, Arthur deBeun, Andrew Emmerson,
etc. described in this magazine. The Publisher does not Tony Marsden, Mark Owen, Julian Rivers
accept responsibility for failing to identify such pat- Membership Manager: Raoul Morreau
ent(s) or other protection. The Publisher disclaims any
International Editorial Staff: Thijs Beckers, Mariline Thiebaut-Brodier
responsibility for the safe and proper function of read-
er-assembled projects based upon or from schematics,
Denis Meyer, Jens Nickel
descriptions or information published in or in relation Laboratory Staff: Ton Giesberts, Luc Lemmens,
with Elektor magazine. Clemens Valens, Jan Visser
Graphic Design & Prepress: Giel Dols
© Elektor International Media b.v. 2017
Publisher: Don Akkermans
www.eimworld.com
Printed in the Netherlands
www.elektormagazine.com January & February 2017 3

this edition COOL POWER with D-WATT
74
Volume 43 – Edition 1/2017
No. 481 & 482 January & Februrary 2017
21 electronica 2016
Overview of what’s hot
24 electronica Fast Forward Award 2016

Who won the € 75,000 media budget plus a
booth at electronica 2018?
26 ElektorBusiness: Simple Interfacing to Analog

and Digital Position Sensors
For industrial drive control systems
65 The Elektor Community

112 Elektor Store
42
128 Elektor World News
130 Play & Win: Catdoku
To kick in the New Year our traditional
Hexadoku got a feline makeover. ELEKTOR SDR RELOADED:
STANDALONE SOLUTIONS
Can we also use the SDR Shield
with just an Arduino and no
LEARN DESIGN SHARE PC at all? Certainly we can.
It’s entirely feasible (with a
6 Welcome to the LEARN section few restrictions) because all
of the vital control functions
7 Peculiar Parts, the series are available on the Elektor
The Fetron, a Solid-state Tube
Extension Shield. The only task
8 Color LEDs that remains then is processing
Past, present and future of these colorful the IQ signals.
indicator lamps that have no filament or gas
filling.
14 Tips & Tricks:

Programming Adapter for USBasp
A DIY adapter to program an ATtiny when it’s
mounted in a prototyping plug board.
46 SUPRA 2.0 MC Version
15 CAD for PCBs
15
With some simple changes this design can
An overview of a limited selection of currently
be adapted for use with moving-coil (MC)
available free PCB CAD programs.
cartridges.
50 Connect Objects with Genuino 101

Make any device talk to your mobile.
LEARN DESIGN SHARE 56 BrainBox Arduino

A sturdy Arduino with screw terminals, an on-
board buzzer and a motor-driver IC.
30 Welcome to the DESIGN section
60 LEGO® Control Board for the Raspberry Pi
32 Sand Clock
With 4 motor-control outputs and 16 digital
This nifty gadget built around an Arduino Uno
I/O channels.
uses servos and a pantograph mechanism to
write the time in a sand bed. 66 Debugging the Arduino Zero & M0 Pro
Delving deeper into the world of Arduino.
40 Super Simple Dice
A cheap and microcontroller-free project 71 Analog Random Numbers
designed especially for beginners! Using an Arduino and a noise source.
42 Elektor SDR Reloaded (4) 74 Cool Power with D-Watt

Can we also use the SDR Shield with just an A high-performance amplifier with lots of
Arduino and no PC at all? Certainly we can! output power.
4 January & February 2017 www.elektormagazine.com

Elektor has a long history and a good
reputation with high-end, yet DIY,
audio projects. This time we proudly
present a new power amplifier aimed magazine
at all audio enthusiasts, built around a
digital audio driver IC and operating
in class-D. This has the advantage
LEARN DESIGN SHARE
that the amplifier can deliver a lot
114 Welcome to the SHARE section
of output power (like 200 watts
into 8 ohms) with very low heat 115 Simulation with SystemVision
A free web-based tool.
dissipation.
Say goodbye to big heatsinks! 118 Opamp Power Supply Connexions
Capricious or logical?
119 Err-lectronics
Corrections, Updates and Feedback to
published articles.
94
120 The Mendocino Motor
It spins and sways… by solar power alone.
122 Build a Chipamp!

Great sound with only a few components.
AXICAT, 124 Elektor Labs Pipeline

A constant stream of projects and activities in
A VERSATILE USB the Labs.
DEVELOPMENT TOOL 125 Retronics: From Verobox to Heavy Metal
If you are developing an app for
Elektor lab instruments 1980s and 1990s style.
an integrated circuit and you want
a fast and easy way to link it to
a Linux or Windows PC, we have
next edition
the answer: AxiCat. This multi-
protocol USB adapter can be used
for communicating with ICs with an
I2C or SPI interface, and you don’t
have to write any code to use it.
Heating Monitor using ESP8266
This heating monitor logs essential operating parameters,
shows them on its built-in display and stores them on a USB
stick for analysis later. The data is sent via Wi-Fi, so you check
85 Sensors make Sense (2)
CAD
it out from anywhere around the home or via the Internet.
Some sensors have dual outputs, one analog
and one digital. Swiss Pi Extensions
for PCBs 92 MAXREFDES99# Display Driver Shield

An easy-to-use 256-pixel solution capable of
The Elektor ‘Swiss Pi’ expansion board from the September
& October 2016 edition provides lots of useful functions for
handling even the most demanding characters. the popular Raspberry Pi. In this follow-up article we take a
closer look at the actual use of the Swiss Server and provide
94 AxiCat
detailed descriptions of demo programs in Python and PHP.
A versatile USB development tool to link an
integrated circuit to a Linux or Windows PC.
3D Printer Head Temperature Controller
98 LED Stairway Lighting Every type of 3D printing filament requires a different extru-
One controller for a whole bunch of LEDs. sion temperature to obtain the best results. This small device
101 Internet Radio with an OLED display is an accurate extruder head tempera-
with Fluorescent Display (2) ture controller controlling the mat temperature as well.
The firmware for the display and the code for
the Raspberry Pi.
106 IoT Shield for Arduino

Lets You Do Your (Internet of) Thing. Elektor Magazine edition 2 / 2017 covering March and April is published
on 10 February 2017. Delivery of printed copies to Elektor Gold
108 DAB+ Antenna Diplexer Members is subject to transport.
Add digital reception to your car radio. Contents and article titles subject to change.

Welcome to the LEARN section
Jens Nickel
SHARE By
We take a trip
The journey from Cologne to electronica in Munich takes a few
hours by train. As usual the on-board Wi-Fi was distinctly flakey
and there wasn’t a reliable cell phone signal. It’s at times like
these you really appreciate the benefits of hard copy. There
was plenty of time to browse the electronics magazines I had
tucked into my
bag before I left. A
common thread in
many of the arti-
cles was of course
the IoT in general
and IoT security in
particular, which
are also the two main themes of electronica this year.
I lost count of the number of development systems and
boards from different manufacturers available — most
were preconfigured to operate with one of the Cloud ser-
DESIGN
vice providers. One name that always seems to crop up is

Amazon Web Services provided by the leading Cloud ser-
vice provider Amazon. After registration (restricted basic
service is free) you can upload data from a sensor boards
into the Cloud using a protocol such as MQTT. The board
is represented by a virtual version or ‘device shadow’ in the Cloud where the data is stored. Now even
when the sensor board is offline, the user still has access to all its stored data by using their smart
device to connect to the shadow device.
My own experiences of developing just such an IoT application are detailed in my blog ‘My Journey into
the Cloud’ which you can read at www.elektormagazine.com by typing the title into the Search box.
So much to see
We certainly didn’t have any spare time on our hands at the Elektor stand during the Munich trade
show. My first priority was to get better acquainted with the entrants taking part in the Fast Forward
Start-up Award and explore their prototypes more closely (I wasn’t entirely successful at this, there
were just so many interesting things to see, the photo only shows a few of the entrants).
I also didn’t get a chance to visit the stands of the major semiconductor companies. Luckily one of
our freelance contributors and utter microcontroller fan Viacheslav Gromov was able to do that job
for me and you can read his report in the LEARN section.
One I was determined not to miss was the Microchip/Atmel stand. While there I spied an interesting
‘Power’ debugger and also a kit, which – you guessed it – has been specifically designed for secure
LEARN
access with the Internet of Things (www.atmel.com/tools/at88ckecc-aws-xstk.aspx). On board there

is a dedicated Atmel crypto engine preconfigured for generating signer certificates and register-
ing them with — tadaa! — Ama-
zon Web Services. I must admit to
suffering a certain level of informa-
tion overload when discussing the
board with the Atmel experts, but
there’s no shame in admitting that,
learning something new is half the
fun. I already have a to-do list of
topics including (Root-) certifica-
tion, ECC and the entire login and
authentication process required for
using the AWS; it’s going to be
interesting...
(160244)

basics training q &a tips & tricks software
The Fetron,
a Solid-state Tube
Peculiar Parts, the series
It’s shiny and tubular, 25 mm tall and 15 mm in diameter, with seven gold-plated pins at one end. It’s
clearly an active device but is it hollow-state or solid-state? The answer is both: it’s a tube and it’s solid-
state, because it’s a Fetron.
By Andrew Emmerson (United Kingdom) and drew zero filament current was an attractive proposition.
Telephone companies took a particular shine to them and the
Actually it’s a tube substitute, entirely solid-state but in the examples seen in Figure 1 came from New Zealand, where they
form factor of the tube that it replaces. It uses cascoded JFETs were used in the launch amplifiers for the sub-sea telephone
(junction field-effect transistors) to duplicate the transfer char- cables between that country and Australia. Other applications
acteristics of specific types of vacuum tubes (mainly pentodes were in high-spec. broadcast equipment and guitar amplifiers
but some triodes were also made). Introduced by Teledyne (!), whilst Teledyne marketed special tube conversion kits for
Semiconductor in 1972 and intended for military applications upgrading vacuum tube voltmeters and oscilloscopes by Hewlett
and other ‘high utilization’ equipment (not consumer electron- Packard and Tektronix to give these the same stability, low-noise
ics), the Fetron was able to extend the service life of expensive characteristics and reliability as high-quality bipolar transistors
tube apparatus, deferring the cost of its replacement. Because (Figure 2). When the time came for users to finally replace their
Fetrons were entirely solid-state, they consumed less than half equipment with solid-state, the need for Fetrons vanished to leave
the power of vacuum tubes and slashed the cost of air condi- only a few memories of this ingenious but short-lived device.
tioning. The need for frequent adjustment and periodic tube (160265)
replacement was eliminated and despite their eye-watering price,
For further reading
Fetrons could repay their cost within six months of installation.
Teledyne developed the Fetron for the direct plug-in replace- https://en.wikipedia.org/wiki/Fetron
ment of several types of existing vacuum tubes, but herein lay http://hrsasa.asn.au/docs/Fetron.pdf
their weakness. The connoisseur of Fetrons, Dr Hugo Holden of www.philipstorr.id.au/radio/eleven/fetron.htm
Queensland, Australia, explains that marketing Fetrons as tube
www.worldphaco.com/uploads/WORLDFETRON.pdf
substitutes was a fatal error. In many applications they were not
at all ideal, for example in circuits such as local oscillators that www.elektormagazine.com/magazine/elektor-201609/39810
rely on grid current automatic bias. A Fetron malfunctions in this
application and draws gate current, he states, causing device
saturation and introducing severe harmonics into the output. Please contribute your Peculiar Parts article,
Nevertheless, a ‘tube’ that generated no heat, would not drift, email [email protected]
Figure 1. Fetrons from Down Under. Figure 2. Period product advertising.

learn design share
Color LEDs
Past, present and future
By Dr Thomas Scherer (Germany)
We have known for a hundred years how to get certain inorganic materials to emit light when a current
is passed through them, and LEDs themselves have existed for some fifty years. These colorful indicator
lamps that have no filament or gas filling have been gradually refined and have become available in more
varied and vivid colors. In recent years their efficiency has improved to the point that LED technology is
replacing even fluorescent lighting. The most recent developments include LEDs that can replace LCDs.
After the invention of the crystal rectifier being insensitive to vibration and having
by Karl Ferdinand Braun in 1874 it was a much longer life. Neon lamps, with their
another 33 years before Henry Joseph working voltages of 70 V to 100 V, also
Round discovered that the point of contact went the way of the vacuum tube: LEDs,
between the metal wire and the crystal operating from voltages of 1.2 V to 4 V
could be made to emit light, yellow, red (see the text box Voltage and Color)
or even blue depending on the applied worked much better in transistor circuits.
voltage. Round was therefore the orig- Also in the 1970s seven-segment dis-
inal inventor of the LED and the elec- plays began to appear using (red) LED
Figure 3. From the antiques collection: a red troluminescent effect in semiconductors technology: see Figure 3. Some readers
seven-segment LED display with a character that he discovered has been named the will be able to remember the first pocket
height of 10 mm. ‘Round effect’ in his honor. And so our calculators like the one shown in Fig-
story begins. ure 4, which used tiny red displays that
often consumed more power than the chip
A slow evolution doing the actual calculating. My TI-59
The first modern-style LED was made at from 1977, one of the first programma-
RCA. It produced invisible infrared light. ble calculators, had many more functions
That was in 1955, and the inventor was than that example, but still only had the
called Rubin Braunstein. Texas Instru- same type of minuscule display. And it
ments (TI) has the honor of being the took such a long time for this schoolboy
first mass producer of LEDs, starting in to save up for it!
1962. TI’s infrared LED had an efficiency Three different colors of LEDs were
of a mere 1.1 %! The first visible red LED available in the 1970s: red, yellow and
was made by General Electric (GE), also green. The upper row in Figure 5 shows
in 1962; there was a surprisingly long (ignoring the blue LED on the right for
gap of ten years before the first success- the moment) some antiques dating from
Figure 4. The first mass-market pocket
ful yellow LED was made, again at GE. this period. They were in general not par-
calculators used tiny red seven-segment LED
Through the 1970s red LEDs gradually ticularly bright, the shades of red and
displays. (Source: Erhaka, Wikimedia Commons).
improved. Efficiencies approached 5 % yellow light emitted varied from device
and they started to be used as indicator to device, and the green color was not
lights. Devices typically came in round exactly convincing. Figure 6 shows the
3 mm or 5 mm plastic packages and brightness and color variation in green
used only about 40 mW of power. They LEDs from that time: all four LEDs are
had many further advantages over bulbs wired in series, with a current of 5 mA
with glowing tungsten filaments, including flowing through them. The LED at the
Figure 5. A LED miscellany. Top row, from left to right: red, yellow and four different green LEDs
of 1970s vintage, compared to a modern blue LED which is so bright that its light saturates the
camera sensor. Bottom row: modern green LEDs in 10 mm and 20 mm packages, and RGB, white
and infrared LEDs.

Voltage and Color

Inorganic LEDs are based on doped
semiconductors, and often contain ele-
ments such as aluminum, arsenic, gal-
lium, indium, phosphorus and nitrogen.
More recently research has turned to
LEDs employing carbon (in the form of
diamond), silicon and zinc. The compo-
sition, concentrations and construction
affect not just the efficiency of the LED,
but also have a direct effect on the color
Figure 1. The relationship between an LED’s Figure 2. Spectra of various colors of LED. The
of the emitted light. The energy of the
forward voltage (bandgap) and light color. shorter the wavelength, the more broadband the
photons emitted by a material depends Voltage and wavelength are in inverse relation. spectrum. Below, for comparison, the spectrum
on what is called the bandgap, which is of sunlight. (Source: Wikimedia Commons).
the energy difference between the con-
duction band and the valence band of
the semiconductor in question. The big- with increasing temperature: this is not lengths, LEDs are reasonably narrow-
ger the bandgap the higher the energy a surprise, as the forward voltage drops band. The spectrum of blue LEDs, on
of the photon and the shorter the wave- with increasing temperature. If a partic- the other hand, is somewhat broad-
length of the light. The bandgap has a ular application requires a stable color, er. This characteristic is used in white
big effect on the threshold voltage, which the die temperature of the LED must be LEDs that consist of blue LEDs plus a
is the voltage that has to be exceeded kept in mind. phosphor layer: the result is a more
before current will flow and before elec- The purity of color of the light from uniform spectrum and hence a more
trons will jump between the conduction LEDs lies between that of (colored) in- natural (that is, sunlight-like) light.
and valence bands and so emit photons. candescent bulbs and lasers. In other The quality of light from a white LED
The theoretical relationship is shown in words, LEDs are not as narrowband as is characterized by its ‘color rendering
Figure 1. It is now clear why the forward lasers, which emit light of practically a index’, or CRI, Ra. Values are specified
voltage of a blue LED is around twice as single wavelength, nor as wideband as relative to sunlight, which has a CRI
high as that of a red LED, as the frequen- incandescent bulbs. Figure 2 shows of 100; white LEDs manage CRI values
cy of the light is proportional to the pho- that, particularly at longer wave- of between 75 and 95.
ton energy.
Table 1 shows which semicon-
Table 1
ductor materials are used to pro-
Wavelength Voltage
duce which colors, along with the Color Material
(nm) (v)
corresponding wavelength ranges GaAs gallium arsenide
and the forward voltage you can infrared over 760 less than 1.6
AlGaAs aluminum gallium arsenide
expect. Now, the actual voltage AlGaAs aluminum gallium arsenide
you might measure across the GaAsP gallium arsenide phosphide
red 610 to 760 1.6 to 1.9
terminals of a real LED depends AlGaInP aluminum gallium indium phosphide
of course not just on the bandgap GaP gallium phosphide
of the materials used but on many GaAsP gallium arsenide phosphide
other details including the resis- orange 590 to 610 AlGaInP aluminum gallium indium phosphide 1.8 to 2.2
tance of the materials and bonding GaP gallium phosphide
wires and the semiconductor junc- GaAsP gallium arsenide phosphide
tions. The voltage across the LED yellow 570 to 590 AlGaInP aluminum gallium indium phosphide 2.0 to 2.4
will also increase with increasing GaP gallium phosphide
InGaN indium gallium nitride
current. The reason that the tran-
GaN gallium nitride
sition between current not flowing
green 500 to 570 GaP gallium phosphide 2.2 to 2.7
and current flowing is not perfect-
AlGaInP aluminum gallium indium phosphide
ly sharp is that in the transition
AlGaP aluminum gallium phosphide
region electrons behave statisti-
ZnSe zinc selenide
cally, having a certain probability
blue 450 to 500 InGaN indium gallium nitride 2.6 to 3.3
of overcoming the bandgap which
SiC silicon carbide
rises with increasing voltage. The
violet 400 to 450 InGaN indium gallium nitride 3.2 to 3.6
spectrum of the light from an LED
AlN aluminum nitride
also moves slightly towards lon-
ultraviolet 230 to 400 AlGaN aluminum gallium nitride 3.5 to 4.2
ger wavelengths (‘warmer’ colors)
AlGaInN aluminum gallium indium nitride

learn design share
see a blue LED in a transparent plastic

package, operating at 5 mA. It is so bright
that it saturates the camera’s sensor and
so misleadingly appears bluish-white in
color. At the bottom left are green LEDs
in 10 mm- and 20 mm-diameter pack-
ages. The larger of these is a multi-chip
LED containing four LED chips connected
in series. Next to it on its right is a com-
mon-anode RGB LED which therefore has
four leads. The three primary-color LED
chips in this device can be driven in such
Figure 6. Four different green LEDs: each emits Figure 9. More than an RGB LED: this OSRAM
a way as to create a mixture of practically
a different light. The LED at the top left is a OSTAR LED in an SMD package contains a white
modern ‘superbright’ device, which saturates LED chip in addition to chips in the primary any desired color. The transparent LED
the camera sensor. The other 5 mm LEDs are the colors. to its right is a modern white device, and
samples from Figure 5. on the far right is an infrared LED. The
package of this device is transparent to
infrared light.
upper left is a modern high-efficiency type From the meager levels of the seventies
for comparison purposes: it is so bright the light output of LEDs began to increase
that it saturates the sensor on the cam- rapidly, especially at the beginning of this
era. One of the old-school LEDs is more millennium, reaching efficiency levels of
yellow-green than true green. Red and 100 lm/W. This puts white LEDs on a
yellow LEDs suffered from similar weak- level where they can compete with fluo-
nesses. Fortunately modern LEDs in 5 mm rescent lamps, matching them for light
or SMD packages have beautifully deep output while offering the advantages of
colors and are so efficient that, when used a longer service life and zero mercury
as indicators, the current through them content. In the laboratory, meanwhile,
must be carefully limited to avoid daz- white LEDs have been made with output
zling the user. efficiencies exceeding 300 lm/W, just a
Figure 7. Maximum theoretically possible light
few percent short of the theoretical limit
output from an LED as a function of color. Note
that the lumen is a unit which takes into account Status quo of 350 lm/W at a color temperature of
the sensitivity of the human eye to different There was an even longer gap before 6600 K. Figure 7 shows how the max-
colors. LEDs in the previously missing blue color imum theoretically possible light output
arrived. They first came on the market depends on the color of the LED. There
only in the 1990s, produced by Nichia. is a nice paradox here: although the blue
The development of the blue LED was LEDs on which white LEDs are based have
judged worthy of a Nobel prize for its an efficiency of no more than 200 lm/W,
inventors Shuji Nakamura, Isamu Akasaki the white LEDs offer a higher efficiency.
and Hiroshi Amano in 2014, a huge honor The explanation for this is that the way
for such a small device. The award of the the ‘lumen’ is defined takes into account
prize recognized the fact that high-ef- the varying sensitivity of the human eye
ficiency blue LEDs are fundamental to to different colors: green light at a given
making white LEDs, where a phosphor power looks brighter than other colors.
layer is added to the LED which absorbs And so, although when generating yellow
a portion of the higher-frequency blue light using the phosphor a little energy
light and converts it into yellow light. The is lost (in fact, converted into heat), the
combination of this yellow light with the white light looks brighter to us than the
remaining blue light yields white light blue light directly from the chip under
through additive mixing. This idea was the phosphor would.
developed by Jürgen Schneider at the It is worth noting, then, that the potential
Fraunhofer Institute and now forms the for further refinements in the lab is almost
basis for almost all LEDs used in lighting exhausted as there is very little scope
applications. Today white LEDs have a for further efficiency gains. There is still
slightly greater market share than color progress to be made, however, in convert-
LEDs. You can read more about white ing these prototypes into real products.
LEDs in my article ‘Let There Be LED’ There is plenty of room for improvement
Figure 8. An LED ‘candle lamp’. The LED strips in the January/February 2016 issue of in mass production processes to enable
in this E14 bulb mimic traditional filaments. Elektor [1]. the manufacture of devices matching the
(Source: OSRAM). At the upper right of Figure 5 you can performance achieved in the lab at rea-

sonable prices. This will no doubt happen use in specialist lighting applications such
within the next decade, and then we will as providing precise illumination for surgi-
have at our disposal a wonderfully ener- cal operations and other uses where color
gy-efficient lighting technology based on accuracy is critical. Despite its small size,
modern semiconductors which, thanks to this SMD device can output over 500 lm
its excellent service life, will also have a of light at its nominal current of 1.4 A.
relatively small carbon footprint. Readers who like to strut their stuff
occasionally will already be aware of the
Color LEDs widespread usage of LEDs in discos and
In recent years LEDs have reached such clubs. Stage spotlights, or ‘PAR cans’
high efficiency and, thanks to mass pro- as they are sometimes called, used to
duction, low prices, that they have by be incandescent lamps rated at several
and large completely superseded filament hundred watts equipped with colored fil-
Figure 10. Cameo type CLP56RGB05PS PAR stage
lamps. And LED lamps have moved on ters. Nowadays they have been largely
effect light with 151 colored LEDs and a power
from the ungainly units seen in the past LEDified with all-electronic control (see rating of around 30 W at 230 V. Not only does the
containing a large number of SMD chips: Figure 10). Although the LED versions light sport a DMX interface for remote control,
these days you can buy bulbs whose inter- are not yet quite the equal of the old it also has a built-in microphone to enable ‘light
nal LED structures mimic the filaments technology in all respects, the availabil- organ’ effects.
of traditional slender candle-shaped fila- ity of ever brighter LEDs is closing the
ment bulbs (see Figure 8). So now it is gap. Service life, robustness, much lower
possible to populate your chandelier with heat generation, reduced risk of injury
modern energy-efficient bulbs while still (stage spotlights are prone to explosion)
remaining faithful to its style. LED lighting and instant electronic control all weigh in
is now available in practically any shape, favor of the LED versions.
size and color you can imagine: we shall Colored LEDs don’t have to be used in
return to this topic later. conjunction with music, of course. At
Infrared LEDs have of course been used an early stage Philips realized that even
for many years in remote controls, as well Joe and Janet Bloggs would be keen to
as more recently in providing unobtrusive have configurable mood lighting in their
night-time lighting for security cameras home. Under the ‘LivingColors’ brand the
or for military purposes. Another inter- Dutch electronics giant has been mar-
esting application is the use in industry keting a range of table and wall lights Figure 11. Philips LivingColors: this lamp from
the highly successful range contains remotely-
of infrared lamps consisting of hundreds for a few years now (see Figure 11).
controllable colored and white LEDs in a plastic
of SMD LEDs with a total output power of These remotely-controlled LED lamps can
enclosure.
over 100 W as a component in projectors be adjusted over a wide range of colors.
to shine a grid of infrared illumination Each lamp contains a number of color and
on one part of a production line or on an white LEDs that can produce static mixed
assembly station. The idea behind this is colors or flashing patterns, and the lamps
that, unlike a human operator’s eyes, a can also be dimmed. The success of this
robot’s cameras can easily see the pro- range led Philips to produce the ‘Hue’
jected grid and hence orientate itself in range (Figure 12). This brand includes
space and better estimate the position not only lamps, but also LED strips and
of units on the production line. Humans ‘normal’ colored bulbs for E27 fittings,
working alongside the robots or monitor- all remotely controlled using a ‘bridge’. A
Figure 12. The Hue series from Philips includes
ing their operation are not disturbed by range of lamps with built-in lithium bat-
colored LED lamps with E27 fittings and a
the infrared light. teries is also available under the ‘Hue Go’ ‘bridge’ for remote control
White LEDs are not the only way to gen- brand. These lighting systems have been
erate white light: a more expensive, but highly profitable for Philips and so doubt-
equally good, if not better, alternative, less further variants will also appear on
is to use RGB LEDs. These consist of LED the market. If you are looking for a more
chips in (at least) the three primary colors economical alternative, you will not be
mounted in a single package. By passing disappointed: the DIY sheds have spot-
different currents through the different ted the opportunity and offer shelfloads
chips the outputs from the individual LEDs of inexpensive color LED lighting of all
can be mixed at will to produce practi- kinds (Figure 13), costing just a fraction
cally any desired color. Figure 9 shows of the price of the big-brand products. For
an example, an OSTAR SMD LED, which an even cheaper solution, you can import
includes not only chips with red (625 nm), directly from China using eBay or Alib- Figure 13. These days you can find a huge
green (530 nm) and blue (453 nm) LEDs, aba. In this case, of course, you have to selection of LEDs strips, bulbs and effect lamps
but also a white LED. It is designed for reckon with the complications of dealing with or without remote control in any DIY shed.

learn design share
boards or building ‘black lights’. Another environment, but that is now changing.
application for ultraviolet LEDs is illumi- OLEDs can very easily be used, for exam-
nating the water flowing through a trans- ple, to make displays, as printing tech-
parent pipe, with the aim of reducing the nology allows the organic light-emitting
growth of algae in aquaria and swimming pixels to be manufactured along with
pools. Some varnishes can be cured using their interconnect wiring. RGB OLED dis-
ultraviolet light, and the LEDs can also be plays are bright and contrasty. Similar
used to test the resistance of products to approaches to manufacturing displays
exposure to ultraviolet light. with semiconductors are much more
complicated and decidedly more pricey.
OLEDs OLED displays are increasingly compet-
Organic LEDs (OLEDs) were discovered ing with LCD displays and in principle the
much later than inorganic LEDs, and so OLED beats its liquid-crystal counterpart
far cannot compete with them in terms of in virtually all departments. There is no
Figure 14. Google Nexus 6p: the OLED display light output or long-term stability. How- backlight or any of its attendant prob-
on this reasonably-priced smartphone has a
ever, their star is in the ascendant: the lems, as the OLED pixels are themselves
resolution of 2560x1440 pixels.
advantages of OLEDs are low-cost raw the source of light. Current consumption
materials and, in principle, lower cost of is also lower, as much of the light from
with possible customs charges, and of manufacture, as they do not require such an LCD’s backlight is lost in the diffuser
course always bear in mind that cheaper strict ‘clean room’ conditions. OLEDs can and in the polarization filters. Colors that
does not necessarily mean better value also be made in flat sheets: this is an change with viewing angle are a thing
(see the text box Feel the Quality). advantage in many lighting applications, of the past with OLEDs. And last but not
Naturally ultraviolet LEDs also have their as the sheets do not have to be rigid. least the contrast ratio of an OLED display
uses. As well as specialist applications Flexible lamps are therefore a possibility. cannot be beaten: when an LED is off, it
there are uses such as exposing the For a long time it has proved difficult to is off. Blacks really are black.
photosensitive lacquer on printed circuit make OLED products outside a laboratory Given all these wondrous advantages you
Feel the Quality Correcting for this effect, a lamp consisting of ten LEDs might
have a service life of only 15000 hours rather than the hoped-for
LED lamps, containing a large number of individual LEDs to pro- 50000 hours. And, if in the real-world vehicle environment, the
vide sufficient light output, have been available for some time. life of an individual LED falls to only 10000 hours, one LED in the
Since the mid-1990s car makers have routinely used a string of ten-LED lamp might easily fail after only 3000 hours. But that is
red LEDs for the ‘third’ brake light, even on mid-range vehicles: not the end of the story: in vehicle lights that use LEDs designed
the red LEDs of the time were already bright enough and cheap for a 12 V supply, the LEDs are often wired in series strings of two
enough. Customers were happy too, as the light did not need or three devices each, and so the failure of one LED leads to the
replacing as often over the life of the vehicle as an incandescent loss of light from one or two more. Another problem is that the
version. Brake lights are an important part of vehicle safety, and LEDs in vehicle lights are often hermetically sealed and so are not
so not failing after a few hundred operating hours is a significant replaceable. Instead of just replacing a two-dollar bulb the whole
benefit. All very warm and fluffy, but cracks start to appear in the lamp assembly has to be taken out and replaced.
story when it comes face to face with reality. These are not mere academic considerations, but reflect my own
More specifically, two phenomena come into play, which togeth- personal experience: the brake light on my 1996-model Fiat failed
er make the situation far from ideal. The LED brake lights do in- in just the way described above after only three years, and a re-
deed have a huge ‘theoretical’ service life of up to 50000 hours. placement part plus labor was not a cheap proposition. And in
But in the vehicle environment that does not represent much of case you are thinking ‘well, that’s Fiats for you’ I should say that,
a guarantee, what with vibration, bad weather and high humidity despite the alleged reliability of Toyota vehicles I had an LED fail
levels (which can lead to corrosion of printed circuit board tracks, in exactly the same way on my Prius. In this case it was part of
for example), and extreme temperature variations. These lead to the left-hand rear light cluster, which contains a large number of
mechanical stress on the LED packages, chips and bond wires. As integrated LEDs all firmly fixed in position. At eleven years (in
a result the quoted service life of 50000 hours is a pipe dream. fact, at the same time as the brake disks needed to be replaced)
The car’s bodywork can get very hot in sunshine, and the heat one of the lights failed. Fortunately a third-party replacement part
dissipated by the LEDs adds to this to generate very high oper- was within my means, but still not cheap. If the LEDs in the head-
ating temperatures. The consequence is that LEDs in automotive light on a modern vehicle should fail, you are certainly looking at
applications do indeed last longer than incandescent bulbs, but a tall bill from your garage. Modern technology comes at a price!
not by the large factor that you might have hoped for. In addi- This disadvantage of multi-LED clusters becomes a significant
tion to this there is the ‘multiple LED effect’: even under optimal problem when they are used for domestic lighting. Although hu-
conditions, when many LEDs are in use it is statistically inevitable midity and temperature fluctuations play a lesser role in this en-
that the time for which all LEDs will be working will be reduced. vironment, there is still the question of the quality of the com-

would be forgiven for asking whey OLED being increased, however, and the rumor This is because total operating costs are
displays are not more widespread. The mill has it that there may be sufficient much lower than other technologies, not
problem lies in the extremely demand- capacity by the end of 2017 for an OLED just because of their greater energy effi-
ing requirements of the manufacturing panel to appear at least on the higher-end ciency, but because they require less fre-
process. The many LEDs in a single dis- versions of the iPhone 8. quent maintenance as a result of their
play must all be equally bright (within longer service life. Street lighting is one
rather narrow limits), and must stay that The future application where soon there will be no
way over time. The consequence is that Besides the rumors of an iPhone 8, OLED alternative to LEDs. LEDs could allow the
as little as two years ago even small- panels will shortly be found in all kinds construction of intelligent car headlights
screen TVs were unrealistically expensive. of gadgets and appliances that need to that automatically adjust the shape of
However, towards the end of 2016 LG, display something. It is no exaggeration the light cone away from oncoming vehi-
the market leader in OLED TVs, intro- to say that pretty much every other kind cles to avoid dazzling their drivers; per-
duced a model with a 55-inch diagonal of display technology will soon have had haps manually dipping and undipping the
for a reasonably affordable US$1500. its day. OLEDs will also make inroads into headlights will no longer be necessary.
Smartphones with OLED displays have lighting applications: when that starts We should be so lucky!
been around for rather longer, and the to become economically feasible, many (160246)
price difference for small displays com- lighting manufacturers will jump on the
pared to LCDs is now relatively small: bandwagon. Lamps for conventional fit-
the moderately-priced Google Nexus 6p, tings will remain the domain of semicon- Web Links
produced by Huawei (Figure 14) conductor LEDs, since for the foreseeable [1] ‘Let There Be LED’:
tains a 2560x1440 pixel panel with some future OLEDs do not offer the light den- www.elektormagazine.com/150577
11 million OLEDs. You might wonder why sity required to provide adequate illu- [2] www.youtube.com/
no iPhone has been produced using an mination from a small bulb. Inorganic watch?v=Oj8RIEQH7zA
OLED display: the answer is that, as of LEDs will also supersede practically all
2016, there is insufficient manufacturing other lighting technologies except in niche
capacity in the world. Capacity is steadily applications with specialist requirements.
ponents used. Low-cost LED strings from of the same color from a spare strip and
the Far East used for Christmas decora- mended the original strips, but after an-
tions or for fun are of course fine, but in other two months two more LEDs were
many cases it is hardly possible to believe pushing up the daisies (Figure 15), and
the price of a product on offer. A whole no doubt this is not the end of my woes.
string can often be sold for less than the Regular LED lighting can also often have
total cost of the LEDs it contains. How do shorter life than expected. LED bulbs have
they do it? The easiest reply to this ques- not been around for all that long, but nev-
tion is usually ‘at the expense of quali- Figure 15. A failed LED strip. It is perhaps partly ertheless I have had four bulbs with E27
ty’. And aesthetic considerations are no my fault that LED L30 has gone to join the choir fittings and two GU10 spotlights bite the
determinant of when an LED becomes invisible: because of the kink in the strip it was dust. In the case of the E27 bulbs it was
not being properly cooled, and it has failed to a
an ex-LED [2]; even my Christmas dis- the switching supply circuit whose coil
short-circuit rather than an open-circuit. Other
plays, which are of course done in the proved all too mortal, whereas in the case
LEDs in the same strip, however, seem to have
best possible taste, have on occasion suf- of the GU10 spotlights the LEDs them-
met their demise without such a good excuse.
fered from LED strings (or at least parts selves became history after exposure to
of them) going to meet their maker. excessive temperature. In the case of two
Low-cost LED strips with a self-adhesive LED tubes bought as replacements for flu-
backing, which are powered indirectly orescent lights the engineers at LG had
from a low-voltage supply and which are put a fuse with too low a rating in the in-
often used for illumination in furniture and tegrated switching power supply circuit:
similar applications, suffer from the same uprating the fuse fixed the problem and
problem. My personal tribulations involve the tubes still work today. Also, my home-
two LED strips, each 2 m long, stuck to made LED lamp (Figure 16), which uses
a length of aluminum section: a total of 13 high-quality LEDs, is still going strong
120 LEDs. My estimate for the time to after more than a decade.
the first failure had been of the order
of 4500 hours, but after just six months
Figure 16. The first ‘Scherer-brand’ home-made
and perhaps 600 hours of operation eight
LED lamp was made in 2005, and continues
LEDs had already kicked the bucket! I to run to this day, after over 10000 hours of
desoldered some replacement SMD LEDs operation.

learn design share
Tips and Tricks

From readers for readers
Here is another neat solution sure to make the life of any electronics enthusiast
that much easier.
Programming adapter for USBasp programmer to power the board). These are
From Antonello della Pia then soldered to the corresponding pins on the
8-way socket. The pin header is connected to the
other end of the cable.
I think it’s a great idea to provide a forum where we can
share ideas with the rest of the electronics community (not Now, where does this go? The cables cannot pass under the socket
forgetting the chance to win £50!). One of my all time favor- or cross so I made up another cable fitted with a female header
ite microcontrollers is the ATtiny85, which I usually program using to make the necessary connections to the USBasp. Check out the
a simple and low-cost USBasp programmer.
Sadly though it is not so simple to connect

the pins from programmer to the ICSP pins
of the ATtiny when it’s mounted in a proto-
typing plug board.
Try this adapter I made to make the job eas-

ier. I started with a turned-pin 8-way DIP IC
socket and a SIL 6-way male-/female header
connector; I then separated a six wire-wide
strip from an old 9-way ribbon cable (orig-
inally used for the serial interface). The six
wires supply the signals MISO, MOSI, SCK,
RESET, GND and VCC (when you want the
Table 1. Programming pin assignment pin number correspondence between the 8-way socket and the
(ICSP) for the ATtiny85. USBasp header (Tables 1 and 2)!
Signal Pin Pin Signal
RESET 1 8 VCC It should be clear from the photo how it all goes together. I also
found another use for the adapter: when there’s no space or you
nc 2 7 SCK
don’t want to fit an ICSP header to a board just for programming,
nc 3 6 MISO
this adapter can be used as an intermediate socket between the
GND 4 5 MOSI
ATtiny and its socket on the board for programming purposes.
Using the same principle, it should be possible to make a similar

Table 2. USBasp connector pin assignment. sort of adapter for other makes of controller also.
Signal Pin Pin Signal (160277)
MOSI 1 2 VCC
GND 3 4 TXD
RESET 5 6 RXD
SCK 7 8 GND
MISO 9 10 GND
Have you come up with an inspired way of solving a really challenging problem? Or found an ingenious
but ‘alternative’ way of using some component or tool? Maybe you’ve invented a better or simpler way of tackling
a task? Do write in – for every tip that we publish, we’ll reward you with UKP40 (or local equivalent)!

CAD for PCBs

Free and non-free PCB layout software
By Harry Baggen (Elektor Netherlands) and Dr. Thomas Scherer (Germany)
Hardly anyone who works with electronics, whether professional or enthusiast, is content to leave their
design at the breadboard stage. Particularly with SMD components or relatively complex circuits, this is
anyhow difficult, and prototyping boards are not the ideal solution. There is no getting around a real PCB in
such cases. For that you need suitable software, and to choose the right software, you need information.
PCBs are essential nowadays, even for very small production

runs or one-offs for the lab in a company of any size or on the
workbench of a hobbyist, not only because they look nice but
also because the stability and reliability of electronic circuits
are strongly dependent on how well the circuits are built.
Unfortunately, it costs time and effort to learn how to make
PCBs with a CAD program. In addition, the files these pro-
grams generate for schematics and PCBs are often specific to
the program concerned and difficult or impossible to import
into other programs, although this does not apply to files in
standard production formats such as Excellon. Different pro-
grams also often offer very different features and functions. A
schematic editor and layout program are indispensable basic
components, but some packages also feature integrated sim-
ulation or 3D visualization of the PCBs, which can be helpful
for choosing an enclosure. The program resources are also sig-
nificant, especially the number of components in the libraries.
Although you can always create your own components, that
takes a lot of time and effort. The CAD packages intimately
associated with component distributors score especially well
in that area.
Below we provide an overview of a limited selection of cur- shield, to your own PCB.
rently available PCB CAD programs, to make it easier for you Unfortunately the program crashed while we were trying it
to choose a particular package or to whet your curiosity. out, causing an error report to be sent to the manufacturer.
Importing dfx files generated by Eagle (version 6.4) sometimes
Pad2Pad (version 1.9.111) caused problems. There is clearly room for improvement here.
Pad2Pad is a PCB manufacturer which specializes in supplying After registering as an Pad2Pad user, you receive an email
custom PCBs via the Internet. With their free Pad2Pad soft- every day with a link to a tutorial video. We found this help-
ware you can lay out PCBs fairly easily and send the data to ful. All of the documentation is available online, including the
the manufacturer for PCB production, but unfortunately there tutorial videos.
is no support for schematic generation. The manufacturer is constantly improving the program, which
After starting the program, you first have to enter the PCB means that you receive updates at regular intervals with new
attributes, including the number of layers and the anticipated functions and bug fixes. Pad2Pad runs under Windows XP or
number of boards. Then you start the layout process by plac- later.
ing the components. The program has a fairly large library for
this, and if a particular component or pad footprint is missing, gEDA
you can create it yourself. Unfortunately, you cannot switch As the name suggests, gEDA is a collection of electronic design
between metric and inch units on the fly; this must be speci- automation tools. It is published under a GPL license and is
fied in advance for virtually every window. available for Linux (SUSE and Debian) as well as OS X. At first
There are also quite a few templates available. You can use glance it looks very promising, but the program versions we
them to correctly number a connector in one go or copy a good tested (gschem 1.8.2 and PCB 20140316) have a fair number
layout for a board type you already know, such as an Arduino of small problems. As a result, we often had to use the Help

learn design share
command in the Terminal window, with ”project” replaced by

the name of the actual project.
KiCad
KiCad is also an open source electronic design automation
suite, with the relatively mature version number 4.0.4. The
high level of user community support for this software project
can be seen from the periodic software updates as well as the
fact that finished distributions are available not only for Win-
dows and in some cases OS X, but also for the Linux deriva-
tives Ubuntu, Debian, Mint, Arch, Fedora, openSUSE, Snappy
and Gentoo. This means that if you opt for KiCad, you can be
fairly certain that you will not suddenly find yourself stuck with
files that are no longer readable or editable because current
software is not available.
The KiCad suite consists of the project manager kicad, the

applications eeschema (schematic) and pcbnew (PCB), and
Figure 2. gEDA running under X11 on a Mac.
the tools pcb_calculator, pl_editor, bitmap2component and
gerbview. These are stand-alone applications whose data con-
menu to consult the documentation, the FAQs and the Wiki. sistency is ensured by the project manager. Along with being
However, once we learned how to use it and became familiar mature and very extensive (it takes up more than 1 GB on
with the keyboard shortcuts, we were able to generate sche- the hard disk), KiCad allows you to view finished PCBs in 3D
matics quite nicely with the editor. Here we should mention because the extensive libraries contain the 3D data of the
that the suite does not run under the normal GUI on the Mac, components. On the Mac the applications (/Programs/) and
but instead under the X11 X-window system, which detracts the libraries and templates (/Library/Kicad/...) are stored in
from more than just the look and feel. different places. There’s nothing wrong with that, but some-
times it is inconvenient.
Unfortunately, things go downhill from there. To generate a As the “professional” amongst open source CAD software, KiCad
PCB from the schematic, you first have to manually link each may not have the latest look and feel and is perhaps not as
component to its package. That is possible with the schematic nicely integrated as packages, but it makes up for that with
editor, but not with the attribute editor. It also helps if you many good and useful functions, including 3D visualization.
know all the package names by heart, because there is no The large number of features make this package somewhat
visual support for this. unintuitive in use, and it has a steep learning curve. However,
the documentation and help files are available in various lan-
After all the packages are linked, you can import the schematic guages, which makes it easier for users who are not native
into the PCB editor. There all the components land in a heap. English speakers.
You then create the desired PCB layout by distributing them
to where they belong. If you subsequently change a compo- Fritzing
nent package in gschem, you can use the command gsch2pcb Fritzing is an open-source program available for Windows, OS X
project to import the change into the PCB editor. You enter this and Linux in 32-bit and 64-bit versions, without any limitations.
What sets it apart is that it is an integrated PCB package with
a schematic view, a PCB layout view, and — very special — a
breadboard view. This means that after you draw your circuit
diagram with the program you can first try it out and debug it
on a breadboard before you cast it in stone on a PCB. That’s
especially nice for the maker scene.
As typical with open source software, the Fritzing download

is still beta, which means not yet version 1.x. However, the
current version (0.9.3b) looks perfectly okay, and community
support is in place. There is an online user guide and online
help, as well as examples on the Web. The breadboard view
is quite realistic. A second special feature is that Fritzing also
has a code editor view, so you can write and manage code for
microcontroller systems with this package. That makes it truly
ideal for makers, Arduino projects and the like. On top of that,
the software is localized in virtually all European languages as
well as Japanese, Chinese, Russian and Turkish.
Figure 3. All in one: the KiCad open source package. Once you have ironed out all the bugs in the circuit, everything

runs as it should on the breadboard and the PCB layout is fin-

ished, you can order a board directly from Fritzing Fab, the PCB
fabrication service behind Fritzing, without leaving the program
environment. However, at 70 eurocents per square centime-
ter the boards are on the expensive side – a garden-variety
Arduino shield would cost you a tidy 29 euros. Of course, you
do not have to use this fab service. The software can output
PCBs in image, PDF, SVG or Extended Gerber format, and it
can export XML and Spice net lists as well as components lists.
CometCAD
Some freeware is good, and some is not. CometCAD unfortu-
nately falls in the latter category. Although this simple com-
bination of a schematic editor and a layout program can do
forward annotation and you can make usable PCBs with it, it’s
not exactly easy.
In terms of appearance the program is rather plain, and soft-

ware updating is not its strong suite. The current version (1.09)
dates from December 2015, and a version specifically adapted Figure 4. Flashy and trendy for the maker scene: Fritzing.
to Windows 10 is not available. However, the program ran
under Windows 10 without any problems. There are no ver- closer look at EasyEDA. This is an extensive online package
sions for other operating systems. The free version (L1) is which runs in virtually every browser and is obviously cloud-
severely limited, with a maximum of two sheets per schematic based. It certainly has something to offer – along with the
and a maximum of 50 symbols per sheet, which means less schematic editor and PCB layout tool, it has an integrated
than 50 components. The maximum board size is a measly mixed-mode simulator that allows you to switch between sche-
102 x 102 mm, with a maximum of 250 pins. What’s worse,
the maximum library capacity is 2,000 components. All in all,
other packages offer a lot more and are more modern. It is
unlikely that the free version will encourage users to pay $67
for the L2 version or $134 for the L3 version, which allow more
components and larger boards.
Osmond PCB
Osmond PCB is similar to CometCAD. There is a limited free
version and a payware version. The limitations here are not
as severe; the free version is only limited to a maximum of
700 pins if you want to output the board data in a file (Excellon,
etc.) or print the layout. There are two significant differences
with respect to CometCAD: Osmond PCB is Mac only, which
means it is only available for OS X, and there are periodic bug
fixes. The current version (1.1.33) dates from August 2016.
However, that’s the end of the good news. The look and feel of
the program have not changed since the original design, which
means OS X 10.5. Since then the Mac world has changed a lot,
with OS X now at version 10.12. The user interface is simple, Figure 5. It works, but on the thin side and limited: CometCAD.
and the features are equally simple. One of the biggest short-
comings is the components library, which is a key resource for
PCB design. The library contains a total of 130 components,
which is pitifully few.
Support for scripting and data output in all major file formats
is not enough to make things right. In light of what other
packages offer, it is difficult for small software developers to
compete, and users have to consider whether they want to put
the effort into learning how to use niche software.
EasyEDA
If you are not a fan of big, feature-rich CAD packages from
major vendors with long pedigrees, or you only want to make
occasional “quick and dirty” PCBs, you might want to take a Figure 6. Mac only, and also thin and limited: Osmond PCB.

learn design share
matic and PCB views within the simulation session. The idea
appears to be that modern electronic designers don’t bother
with breadboarding, since they can simulate the circuit and
then go straight to the PCB.
An online package that runs in a browser has a number of

advantages. You don’t have to install it, the software is always
up to date, and it is totally independent of any operating sys-
tem. It is also easy to share schematics or PCB layouts. Of
course, you can save your designs on your own computer. It’s
hard to find any serious disadvantages compared to regular
offline software. EasyEDA is easy to use, smooth and modern.
The available library has a good selection of connectors and
transistors, but it’s a bit thin with ICs. However, it supports
schematic and PCB file import from the professional packages
Figure 7. As a browser-based package, EasyEDA is platform independent,

and it features integrated simulation.
Other options
Along with open source programs and free versions of commer- limited to Eurocard format (100 x 160 mm). A prerequisite is
cial CAD packages (usually with limitations), there is a third a valid .edu email address. The Express version is available to
category of CAD software which is fostered by major electronics everybody. Although it also has an autorouter, the schematic
distributors. Some examples of this are EAGLE from Farnell/ is limited to two sheets and the maximum board size is half
element14, DesignSpark PCB from RS Components, and Mul- as large (100 x 80 mm). That makes the board size the key
tiSIM Blue from Mouser Electronics. Informative articles on all difference. Interestingly enough, EAGLE is available not only
of these programs have been published in previous issues of for Windows, but also for OS X in 32-bit and 64-bit versions.
Elektor. Their key features are briefly described below.
A control panel provides access to the various modules, includ-
EAGLE ing the schematic editor, PCB layout and component libraries.
EAGLE is probably the best known PCB layout program amongst EAGLE is fully sufficient for most purposes. The Elektor Store
hobbyists. One reason for this is that along with paid versions, offers several good books on EAGLE, to help you get started
it has always been available in completely free versions, mainly quickly, correctly and comprehensively.
limited with respect to board size. Students are especially fre- EAGLE was recently acquired by Autodesk. The new owner has
quent users of the free versions. EAGLE consists of two modules: already announced that the free versions will remain available.
one for drawing schematic diagrams and the other for laying
out PCBs. They communicate with each other through direct MultiSIM Blue
forward and backward annotation. There are two different free This is a limited version of the expensive professional MultiSIM
versions: Educational and Express. The Educational version is CAD package from National Instruments. Mouser collaborated
intended for schools and universities and has minimal limita- with NI to make a free version available. Of course, the free
tions. It has an autorouter and allows up to 99 sheets per sche- version has several limitations, but fortunately (and for obvi-
matic. Boards can have up to six layers, with the maximum size ous reasons) it comes with a library of more than 100,000
components from the Mouser catalog. Although that falls well
short of the full Mouser product portfolio, it is gradually being
expanded. You can even put together a components list for a
schematic and get the Mouser prices for the components. That
is very practical, of course, but if there is a component that
Mouser does not carry and you want to add it from another
supplier, you have a problem.
The limitations are unfortunately severe: at most six user-gen-

erated components, with a maximum of 65 components and
one schematic per board. However, the board size is unlimited.
The MultiSIM package consists of the MultiSIM program for
Figure 8. EAGLE with its red and green circuit diagrams, familiar to every schematic diagrams and Ultiboard for PCB layout. As a spe-
electronics adept. cial feature compared to the other packages described in the

Altium Designer and EAGLE. It can also import net lists from Web Links
LTspice. And last but not least, it can import libraries from KiCad.
Pad2Pad: www.pad2pad.com/General/Software.html
EasyEDA comes from a Chinese company that also sells PCBs,
gEDA: www.geda-project.org
which can be ordered directly from the program. At about $17
for a single double-sided board measuring 100 x 100 mm, they KiCad: http://kicad-pcb.org
are reasonably priced. This online package is certainly worth Fritzing: http://fritzing.org/home
bookmarking as good backup CAD solution. CometCAD: www.cometcad.com
Osmond PCB: www.osmondpcb.com
What else?
The selection of packages described here is far from complete, EasyEDA: https://easyeda.com/editor
but it gives you a good idea of what is currently available in EAGLE: https://cadsoft.io
the form of slimmed-down free versions or completely open- MultiSIM Blue: www.mouser.co.uk/multisimblue
source programs with no limitations. Wikipedia has a good
DesignSpark: www.rs-online.com/designspark
tabular overview of electronic design automation (EDA) soft-
ware, including custom and niche products. If the web links CAD overview: https://en.wikipedia.org/wiki/
listed below are not enough for you, the Wikipedia overview Comparison_of_EDA_software
is a good source of additional leads.
(160176-I)
Mechanical as a conventional mechanical CAD program. For

that reason, the original PCB layout version has been renamed
DesignSpark PCB.
DesignSpark PCB has a modern look and feel and runs smoothly.
It supports real-time design rule checking and forward/back-
ward annotation between the schematic and the PCB. The pro-
gram has a reasonable component library with about 80,000
components, and it can automatically output a components list
which can be checked for availability and prices through RS.
DesignSpark PCB can import EAGLE files, and it can even gen-
erate Spice net lists, which can be used to run simulations with
LTspice, TINA or other programs. Unlike nearly all of its com-
petitors, this package can visualize PCBs with their components
in 3D. As far as we know the program is a full version with no
limitations, but it is only available for Windows (7, 8 or 10).
Figure 9. An amplifier simulation in MultiSIM Blue, the slimmed-down

professional package from National Instruments for distributor Mouser.
overview, as the name suggests it includes schematic simu-

lation capability.
The user interface and operation of both programs are excel-
lent. However, the package consists of two separate modules.
Although there is button in MultiSIM for forward annotation,
in the Blue version it is disabled for no apparent reason. The
software runs under Windows XP and later versions.
DesignSpark
DesignSpark from RS Components is a version of the Easy-PC
PCB CAD program from UK software developer Number One
Systems, specifically adapted for RS. Two additional versions
have been released in recent years: DesignSpark Electrical Figure 10. DesignSpark provides 3D visualization of finished PCBs and is
for drawing electrical system diagrams, and DesignSpark free from distributor RS Components.

ew17_185x254_GB_Elektor_2ew17P.indd 1 07.12.16 14:40
electronica 2016
Overview of what’s hot
By Viacheslav Gromov (Germany) ([email protected])
When 2,913 exhibitors and around 73,000 visitors from 88 different countries all get together you can
be sure there’s lots to talk about and marvel at. From passive components via new types of special
connectors and new sensors with pattern-recognition software the range of innovation was exhilarating.
Here we present just a tiny selection of some of the intriguing items that caught our eye at the electronica
trade show in Munich, Germany, last November.
Semiconductor design and manufactur- point at the show and Cypress already kit built around the SHT31 humidity and
ing company Cypress Semiconductor has much to offer in this area. New temperature sensor. This new sensor is
were showcasing new products in two solutions for the technology include the available in a 2.5 mm square package
different areas of interest. Firstly they EZ-PD CCG3 (48 MHz, 64 KB dual flash), and communicates via an I²C interface.
have a new addition to the established with a 32-bit ARM Cortex M0 core which It samples the temperature and humid-
can fully support a USB-C port includ- ity (with 2% or 0.3°C accuracy) at a rate
ing all the interfaces with power deliv- defined by the user. At one measure-
ery. The evaluation kit to go with this is ment per second the sensor draws just
the CY4531 EZ-PD CCG3 Evaluation Kit 2 µA. The board shows the measured
priced at $250. values which can also be read via Blue-
tooth on smart gadgets running an app.
Also from Cypress is the new CY4500 It also serves as a reference design for
EZ-PD Protocol Analyzer board which designers planning to use the sensor. The
looks like a very interesting tool for board is powered by a button cell and has
USB-C developers. The board has a an LCD screen and an 8-bit MC9S08LL-
USB-C port and a USB-C receptacle 8CGT MCU, a pushbutton and the Nordic
connector allowing it to be inserted, nRF51822 BLE module (Bluetooth Low
for example between a computer and Energy). Connections to the sensor, MCU
a peripheral device. It’s transparent to debug pins and the BLE module are easily
the passage of this USB-C data. A third accessible for development purposes. All
connector type USB micro-B hooks up to information for the hardware and software
a host computer for power and for anal- is also available from GitHub [3].
ysis of the USB data. The analyzer costs
around $200 but looks like an indispen- Infineon certainly had much to show
sible tool for developers of USB-C appli- and tell. They have serious numbers of
cations, it allows you to test programmed Arduino shields fitted out with their latest
communication between two devices as ICs. One of their popular small USB-stick
PSoC MCU family called the PSoC Ana- well as the cable and wiring [2]. sized evaluation boards the ‘XMC 2Go’ for
log Coprocessor (48 MHz, 32 KB Flash). the ARM based XMC1100 (M0+, 48 MHz,
This 32-bit MCU, based on the ARM-Cor- 64 KB) processor has now spawned a
tex-M0+ core forms an interface between family of three similar sized boards using
the analog and digital world. On board the same MCU plus sensors on board. The
there are ADCs, DACs, opamps, compar- SPI-current sensor board ‘Current Sensor
ators, analog filter, capacitive-sensing
touch interface und much more. The $49
CY8CKIT-048 PSoC Analog Coprocessor
Pioneer Kit has been developed to allow
designers to explore the possibilities of
this flexible PSoC technology. The board In addition to their advanced gas flow
has many analog sensors on-board suit- sensors Sensirion were showcasing their
able for applications ranging from tem- small but sophisticated SHT31 Smart
perature measuring to inductive proximity Gadget Development Kit priced at €30.
sensing and PIR motion detection [1]. This Smart Gadget dev kit is a simple
USB-C technology was also a big talking reference design board and development

News
F7 board already in production is the Texas Instruments. The new MCU
STM32F769 Discovery Kit at $48 fitted CC1350 SimpleLink ultra low-power dual
with a STM32F769NIH6 (216 MHz, 2 MB band wireless MCU has created a lot of
Flash) with impressive graphics capabil- interest. Apart from its 48 MHz 32-bit
ities. With its in-built Chrom-ART accel- ARM Cortex M3 core, it has 128 KB flash
erator (DMA2D) and other peripherals it memory and two RF peripherals. This
produces a 720p resolution video out- allows the MCU to communicate in sub
put signal with a frame rate of 30 Hz. In 1 GHz bands and also in the 2.4-GHz band
addition to the features on the standard compatible with Bluetooth Low Energy
4.2 spec. The sub 1 GHz offers the best
2Go’ is fitted with a TLI4970. The I²C 3D possible RF range while the Bluetooth
magnetic field sensor board ‘3D Magnetic low energy provides the link via apps on
Sensor 2Go’ is fitted with a TLV493D- BLE enabled smart devices. This combi-
A1B6 chip while the SPI-H bridge module nation gives the advantage of intelligent
‘H-Bridge Kit 2Go’ uses the IFX9201 chip. communications within a network. In an
In addition to all using the same XMC- IoT or industrial 4.0 context the need
based MCU they also have GPIOs and for network interoperability is becom-
two user LEDs for use. Another feature ing increasingly important. The CC1350
of the boards is the sensor end can be sub 1 GHz (actually 868 MHz in the EU)
snapped off leaving solder pads for easy channel for example would give the range
hook-up to any systems. The magnetic
field sensor also includes a small joystick F7-ST Discovery boards the STM32F769
so you build a smooth joystick controller. Discovery Kit also gives the possibility to
The boards are priced at around €20 [4]. add a TFT touch display, a DVI to HDMI
The theme of data security was never or a WiFi adapter if required for devel-
far away and Infineon was showing an opment purposes [6].
interesting TPM product (Trusted Plat-
form Module, an international standard
for a secure cryptoprocessor) called the Würth Elektronik has expanded their
SLB9670, compliant with TPM 1.2/2.0. range of integrated power modules with
This module has an SPI interface (also the addition of the MagI³C-FISM (Fixed
available is the SLB9645 compliant to Isolated SIP Module) and MagI³C-FDSM
TPM 1.2 with an I²C interface) and is (Fixed Step Down Regulator SIP Mod-
suitable for integration into existing MCU ule) series. The MagI³C-FDSM is a step
architectures (or a whole PC) for security. down regulator with a fixed output volt-
Up to eight 2048-bit keys can be stored age level (5 V or 3.3 V at 1 A max.) and
internally. These keys never leave the IC a wide input voltage range from 7 to 42 V
and all data encryption and decryption is (boasting an efficiency of up to 93%!). to link actuators and sensors in a local
performed internally, transparent to the The MagI³C-FISM power modules feature domestic network while the BLE channel
system processor. Normally the crypto galvanic isolation. The four variants are can provide the link to the user’s smart
keys would be stored on something like for use with an input voltage rail at 3.3, device. As you would expect TI have a
an SD card but this makes them vulner- 5, 12 or 24 V. All these variants supply a development board for this new wireless
able to theft or attack. The IC has been fixed 5 V output at 1 W (with up to 80% microcontroller: the CC1350 LaunchPad,
shown working in Linux-based systems; efficiency!) and offer an input to output which can be yours for $29 [9].
one interesting application is in conjunc- isolation good for 1 kV. They are all sim-
tion with a Raspberry Pi. A special board ple to use and very robust [7][8]. For the first time this year I detected a
has been developed for this and will be trend away from capacitive and towards
appearing shortly. These products are You certainly don’t need to hang around inductive proximity sensing technology.
also supported by extensive online doc- too long waiting for new products from Many manufacturers are already supply-
umentation and shows, explaining for ing MCUs incorporating interfaces sup-
example how you can implement a more porting this technology. Its advantage
secure SSL encrypted link between an is a greater immunity to environmental
RPI and a PC [5]. contaminants such as oil, water and dirt
etc. TI has also developed large and small
STMicroelectronics have added to their IC solutions for this switch technology
32-bit ARM Cortex-M7 family of MCUs where a simple pot is all you need to
with new devices that are not yet in full adjust sensitivity. One of the smaller ICs
production. They already have an impres- is the LDC0851; it needs very few exter-
sive selection of MCUs and their asso- nal components and uses a detection coil
ciated development boards. The latest which can be printed on the PCB. A handy

detection, and face detection. This $200 configuration of embedded software pro-
board runs Analog Devices own 400 MHz jects. The development board for this new
32 bit ADSP-BF707 processor with lots of family of ATtinys is the USB-powered
interfaces including an on-board camera. ATtiny817 Xplained Mini Board (retail-
Also with the kit for application developing at $10 (£7)) [13].
ment is the CrossCore Embedded Studio The PIC18F family also has a new sub
IDE and VOS 3.2.0 software libraries for family called the K40. These devices
indoor and outdoor applications plus full range in pin count from 28 to 64 with
user guide and schematics [12]. a 16 KB to 128 KB flash memory and a
clock frequency up to 64 MHz. Highlights
dev board for this chip has the part num- include the Analog to Digital Converter
ber LDC0851EVM and costs just €18. The with Computation (ADC²), which can go
board has a simple integrated inductive about its business without troubling the
touch button and is powered via its micro CPU. It can for example calculate aver-
USB connector or from an on board coin age values of input signals or compare
cell battery. The PCB has two snap off values (useful for touch applications).
sections allowing the sensor and/or the The peripherals also provide added secu-
sensor coil to be separated and used for rity for safety critical applications (CRC
other applications [10]. / memory scan, windowed watchdog
timer and hardware limit timer). Many
With a series of lectures Renesas were
plugging their new RPBRX71M Revela- Last but not least is news for the Atmel
tion Kit priced at €59 which should prove or should we say Microchip community.
useful if you are thinking of developing Good to their promise to maintain and
applications using the RX700 family of expand 8-bit PIC and AVR support, Micro-
microcontrollers. The Revelation Kit hard- chip had their new 8-bit ATtiny817, 816,
ware includes an RX71M MCU with 4 MB 814 and 417 on show. The four devices
Flash, 552 KB RAM and a 320 x 240 QVGA range in pin count from 14 to 24 pins
TFT LCD with touch screen (which looked and 4 KB to 8 KB of Flash. These are the
to be around 4″), allowing the user to first tinyAVR microcontrollers to feature
implement a simple, highly-integrated Core Independent Peripherals (CIPs)
HMI (Human Machine Interface) solu- which include a Peripheral Touch Con-
tion on a single chip microcontroller. The troller (PTC), an Event System (ES), a
32-bit 240 MHz RX71M has a good range Real Time Clock (RTC) plus additional members of the K40 family are avail-
of interface capabilities and comprehen- user-programmable logic blocks (LUTs). able in DIP outline so they can be used
sive security features including built-in These CIP blocks operate almost inde- with the already familiar and universal
pendently, freeing up the main proces- Curiosity Development Board (priced at
sor to take care of more important time €30) [14].
critical events. The devices are supported (160270)
by START, an online tool for graphical
Web Links
[1] www.cypress.com/products/psoc-analog-coprocessor
[2] www.cypress.com/products/ez-pd-ccg3-type-c-port-controller-pd
AES, DES, SHA and RNG for IoT applica- [3] https://www.sensirion.com/en/products/humidity-sensors/development-kit/
tions. Test and calibration functions for [4] http://goo.gl/o9hhGd
the internal peripherals (e.g. ADC and [5] http://goo.gl/Xu5Lkr
GPIOs) are also implemented. [11]. [6] www.st.com/en/evaluation-tools/32f769idiscovery.html
[7] http://katalog.we-online.de/en/pm/MAGIC-FISM
Amid all the new innovation in the truly
[8] http://katalog.we-online.de/en/pm/MAGIC-FDSM
analog world Analog Devices were
demonstrating the capabilities of their dis- [9] www.ti.com/tool/launchxl-cc1350
tinctly digital but no less impressive ADZS- [10] www.ti.com/tool/ldc0851evm
BF707-BLIP2 (Blackfin low-power imaging [11] www.renesas.com/en-eu/solutions/key-technology/human-interface/rx71m-revelation.html
platform) board. This was running some [12] Http://goo.gl/LpdZig
software libraries with multiple functional
[13] www.atmel.com/tools/ATTINY817-XMINI.aspx
profiles for applications such as intelligent
[14] www.microchip.com/promo/pic18f67k40
motion sensing, people counting, vehicle

electronica Fast
Who won the € 75,000 media budget

The “electronica Fast Forward Award, the Start-up Platform categories: Ideas, Prototypes and Start-ups. On the last day of
powered by Elektor”, as it is officially branded, turned out the show, the first place winners of each category went on to
a huge success not only for the three winners of the grand compete for the grand prize, a magnificent PR and marketing
prizes, but for everyone involved. Staged at the electronica budget valued at 75,000 euros plus an exhibition stand at
2016 trade show in Munich, Germany, 35 contestants from 15 electronica 2018 thrown in.
countries pitched their projects to a jury consisting of industry
experts from STMicroelectronics, Conrad, Würth Elektronik and, The contest, sponsored by STMicroelectronics (Platinum),
of course, your own Elektor. After three days of presentations, Conrad (Gold), Würth Elektronik (Gold) and Trinamic (Bronze),
all showing excellent work, rankings were established in three turned out highly interesting with projects covering applications
The results Overall Winners

Ideas nevisQ, intelligent fall prevention and 1. € 75,000 media budget + booth at
detection electronica 2018: Mowea, modular
1. Artem Kuchukov (Germany) – Kewazo,
3. Peter Wasilewsky (Poland) – nWatch, wind energy appliances for off-grid
scaffolding assembly robot
wearable MCU development platform areas
2. Jonas Galle (Belgium) – Valcun, safe
low-cost desktop metal 3D printer 2. € 50,000 media budget: BotFactory
Start-ups Squink, desktop PCB Printer
3. Michael & Andrey Shustov (Russia) –
Baristor, a barrier resistor 1. JF Brandon (USA) – BotFactory Squink, 3. € 25,000 media budget: Kewazo,
desktop PCB Printer scaffolding assembly robot
Prototypes 2. André Kholodov (Germany) – eCozy,
1. Till Nauman & Lara Obst (Germany) smart connected thermostat Special Price “Tech for Good”
– Mowea, modular wind energy 3. Milan Simek (Czech Republic) – € 5,000 euro media budget: Len Williams
appliances for off-grid areas Sewio, precise object tracking and (Australia) – Every Drop Counts, water
2. David Link & Christian Kind (Germany) – visualisation system consumption surveillance system

Forward Award 2016
plus a booth at electronica 2018?

as diverse as you can imagine, from IoT security to monitoring meet these inspired people and see their work.
systems for the elderly and disabled, from state-of-the-art
3D printers to RGB lights controlled by plants, and from audio Being a contest it was not possible unfortunately to award the
systems to precision measurement instruments and scaffolding Grand Prize to every participant, but none left empty handed.
building robots. Contenders from as far as India, central Russia, Although some got more than others, all were awarded a lifelong
the US of A and even Australia, but also from numerous Elektor Hero Membership and we hope to hear from them soon
“nearby” European countries were ready to spend a week in about the progress of their projects. Will you join us next time?
Munich because they believe in what they have created. We (160278)
were honored and proud that we were given the chance to
The Mowea modular wind energy appliances for JF Brandon from BotFactory came all the way Artem Kuchukov (second from the left) shined a
off-grid areas project was unanimously selected from Long Island City (USA) to present his happy smile when he received the Third Prize for
as the Overall Winner of the Fast Forward Award Squink desktop PCB printer. It was definitely the Kewazo scaffolding assembly robot project.
2016. Till Nauman (second from the right) worth the effort as he flew home with the
received the Grand Prize from electronica CEO Second Prize.
Falk Senger (right), assisted by our sponsors
(from left to right) Shawn Silberhorn (Conrad),
Alexander Gerfer (Würth Elektronik) and Jacky
Perdrigeat (STMicroelectronics).

learn design share
Simple Interfacing to Analog

and Digital Position Sensors
for industrial drive control systems
By Brian Fortman, Industrial Drives and Automation Marketing, C2000™ Microcontrollers, Texas Instruments
In industrial drive control system designs, “glue” elements like control and connectivity building blocks
pose many challenges such as lengthier development cycles, a larger board area or a higher BOM
cost. As a result, developers are often
unable to concentrate
on differentiating
features like enhanced
performance, greater
precision and
improved control
loops. Here, some
new supportive
technologies are
discussed to make
life easier.
A particular example of this is the task of interfacing micro- mentation would add cost to the system by increasing the sys-
controllers (MCUs) to position sensors. These sensors can be tem’s electronic BOM, impacting the necessary board space and
linear, angular or multi-axis and typically are used to sense the requiring lengthy development cycles. Moreover, developers
relative or absolute position of a mechanical system in motion, also have to complete extensive compliance testing to certify
propelled by a motor. The sensed position is then converted conformance with industry standards.
to an analog or digital electrical signal for transmission to the This situation begs for a solution that would simplify the inter-
controlling circuit. facing of position sensors to control elements in industrial
Historically, interfacing a position sensor to an MCU could be drive systems and thereby free designers to concentrate on
a time-consuming task that often involved the integration of features and functionality that would make their systems truly
the communication protocol into a field programmable gate distinctive, as well as more competitive, in the marketplace.
array (FPGA) or the programming of an additional MCU with
the decode protocols. In addition, this situation is exacerbated Integrating position feedback
by the fact that there are multiple encoder protocols available, Building on the C2000™ Delfino™ MCU portfolio, Texas Instru-
each suited to certain types of functionality and subsystems. ments provides a comprehensive platform for industrial drive
The system design team might be forced to develop several and control systems. Starting with the processing capabilities
protocol-specific FPGAs which would not scale effectively from required by sophisticated and precise control systems, the
one application to another. Of course, this type of FPGA imple- C2000 Delfino F28379D and F28379S MCUs are equipped with

a full complement of on-chip resources, including DesignDRIVE

Position Manager technology supporting today’s most popu-
lar off-the-shelf analog and digital position sensor interfaces.
This relieves system designers from many of the more basic,
repetitive tasks, saving design time.
TI has extensive expertise with interfacing position sensors to
digital controllers. Beginning with standalone interface solu-
tions for resolver-to-digital solutions, such as the TMDSRSLVR,
TI has continued to add to its position feedback interface sup-
port. Expensive resolver-to-digital chipsets have been replaced
by C2000 MCU on-chip capabilities, leveraging high-perform-
ance analog-to-digital converters (ADCs) and digital-to-ana-
log converters (DACs). Moreover, the powerful trigonometric
math processing of C2000 MCUs is particularly well-suited to
the additional processing needed to calculate the angle, and
extract high-resolution speed information from a resolver’s
amplitude modulated sinusoidal signals.
Many C2000 MCUs support enhanced quadrature encoder pulse
(eQEP) modules that are capable of interfacing with linear or
rotary incremental encoders. These encoders count pulses to
obtain position (once an index is known), direction and speed
information from rotating machines used in high-performance
Figure 1. A closer look at the C2000™ dual-core F28379D MCU with
motion and position control systems. In addition, the eQEPs
DesignDRIVE Position Manager technology.
can be employed to interface to pulse train output (PTO) sig-
nals generally output by a programmable logic controller (PLC)
in industrial automation for motion control.
Also, eQEPs can interface to clockwise/counter clockwise (CW/
CCW) signals. CW/CCW signals are typically used in conjunc-
tion with stepper or servo drives for controlling motors or other
motion-based hardware. The C2000 F28379 MCUs (Figure 1)
support up to three eQEP modules.
Resolver and QEP capabilities provide fast, efficient and inte-
grated solutions for effectively interfacing position sensors with
C2000 Delfino MCUs. The next step has been to extend that
support with complementary solutions that would allow the
MCU to connect directly to more advanced digital and analog
Figure 2. DesignDRIVE Position Manager technology supports the leading
position sensors. analog and digital position sensors.
DesignDRIVE Position Manager technology

Available through TI’s DesignDRIVE platform, Position Manager
technology takes advantage of the on-chip hardware resources
of the C2000 Delfino F28379S and F28379D MCUs to inter-
face to the most popular digital and analog position sensors.
Already incorporating support for incremental encoders (eQEP),
CW/CCW communications and standalone resolver solutions,
Position Manager adds solutions for analog position sensing,
integrating both resolver excitation and sensing, as well as a
SinCos transducer manager.
Unique to C2000 MCUs, Position Manager combines the ana-
log sensor support with the popular digital absolute encoders,
EnDat 2.2 and BiSS-C (Figure 2), giving system designers a
wide range of position sensor types to choose from.
This integrated Position Manager technology offers system
designers a real opportunity to accelerate development cycles Figure 3. EnDat 2.2 solution example: Stackup vs. FPGA.
and reduce BOM costs by eliminating the need for an FPGA
to interface a specific encoder to the MCU or by drastically
reducing the size of the FPGA that may still be needed for as any custom hardware and logic that previously may have
other functions. Figure 3 demonstrates how Position Man- been implemented on an external FPGA. In addition, example
ager technology relieves system designers from the burden closed-loop, position-sensor-based control projects downloaded
of developing the high- and low-level software drivers, as well from DesignDRIVE can be modified for integration into cus-

learn design share
New position sensor interfacing capabilities

TI has been able to expand its position sensor interface solu-
Cable Signal Conditioning Board tions with enhanced capabilities and performance. The follow-
Encoder ing are several recent additions to TI’s capabilities through
Power
Supply DesignDRIVE Position Manager technology.
Motor Connector F28377D
Analog Sub-System
A SinCos
PM
B SinCos SinCos is a feedback methodology which is incorporated into
Library
encoder interfaces like Hiperface® as well as other proprietary
I
SinCos
interfaces. These so-called sinusoidal absolute encoders typ-
position encoder 160221-14 ically offer much higher position and speed resolutions than do
resolver or incremental encoders. In conventional quadrature
encoders, angle information is obtained by counting the edges
Figure 4. Industrial servo drive with SinCos position encoder interface.
of a pair of quadrature pulses.
Angular resolution is fixed by the number of pulses per mech-
anical revolution. However, in SinCos transducers, precision of
the angular measurement is increased by computing the angle
8-Wire cable,
Data+
Data–
between edges using the relationship between a pair of sine
single shielded Clock+
and cosine outputs from the sensor. Effectively, an interpola-
Clock–
Up (Power) tion between edges is made to obtain a ‘fine’ angle. The fine
Un (Ground)
Up* (Power) angle is computed using an arctangent of the two sinusoidal
Un* (Ground)
inputs. For this computation to be valid, both inputs must be
Encoder sampled simultaneously.
EnDat 2.2 Master
TMS320F28379x
Connector Supply
Motor M12, 8-pin
Connector Typically, several thousand electrical revolutions of the sinus-
RS485 oidal signals occur during each mechanical revolution of the
RS485
encoder shaft.
The internal analog sub-system of the F28379 Delfino MCUs
SinCos is ideal for interfacing to SinCos transducers. The presence of
position encoder 160221-15
multiple ADCs, which can be triggered from the same source,
allows simultaneous measurements of both input channels. In
Figure 5. Industrial servo drive with EnDat 2.2 position encoder interface. addition, the F28379 MCUs include a native ARCTAN instruction
as part of the Trigonometry Math Unit (TMU) which means the
angle calculation can be done in as little as 70 nanoseconds!
Another consideration is the high motor shaft speed state
(Figure 4). In this case, there is no longer a need for precise
angle information and the measurement algorithm only needs
to count the number of complete sinusoidal revolutions to deter-
Encoder mine a “coarse” angle measurement. Typically, this is done
EnDat 2.2 Master
TMS320F28379x
Supply
using a pair of analog comparators which compare the incom-
RS485 ing sinusoids with a threshold representing the zero crossing
point. The comparator outputs correspond to the sign of each
RS485
sinusoid and the resulting digital signals are similar to those
produced by a quadrature encoder. On the F28379 MCUs, there
160221-15
are up to eight pairs of analog comparators, each with its own
programmable threshold voltage. These allow the quadrature
Figure 6. Industrial servo drive with BiSS-C position encoder interface. pulses to be generated which are then fed internally to one of
the on-chip quadrature encoder peripheral (QEP) modules for
coarse angle and speed measurements.
tomer projects. The lower system layers are provided on-chip EnDat
or through reference designs and a ready-to-use library of EnDat is a digital bi-directional four-wire interface developed
application programming interface (API) modules. by the German company, HEIDENHAIN. A sensor with an EnDat
In addition to reducing development time, Position Manager encoder can communicate position values, transmit and update
technology also decreases the compliance and interoperabil- information stored in the encoder, or save the information. Data
ity testing that system manufacturers have undertaken in the is sent along with clock signals. The C2000 MCU can select
past. The Position Manager technology is fully tested across the type of data the encoder will transmit, including positon
a variety of sensors. Please see the user’s guides for details values, parameters, diagnostics and others.
on the testing results. Moreover, future revisions and updates Position Manager technology interfaces the C2000 F28379 MCU
to the applicable standards will also be supported by Position directly to the EnDat encoder (Figure 5). The only compon-
Manager technology. ents external to the MCU are two RS-485 transceivers and the

encoder power supply circuit. The EnDat Master is implemented face can be adjusted to feature improved control of modular
using the C2000 MCU’s configurable logic block, where the functions and timing by transmitting position information from
communication protocol is handled. encoders every control cycle.
Position Manager technology has been tested against a range
of rotary, linear and multi-turn encoders from HEIDENHAIN Industrial drive control systems-on-chip
and across resolutions from 13 bits to 35 bits at distances of Powerful and programmable MCUs like TI’s C2000 Defino
70 meters or more. F28379 MCUs represent the next step toward industrial drive
control systems-on-chip (SoC). They empower more effective
BiSS-C and efficient system architectures by eliminating the need for
The open source BiSS (bi-directional/serial/synchronous) digital an external FPGA for ancillary processing requirements or by
interface is based on a real-time communications protocol. reducing the size of the FPGA significantly.
The original specification was developed by iC-Haus GmbH of By enabling a direct connection between a C2000 MCU and a
Germany. position sensor, Position Manager technology frees developers
BiSS-continuous mode (BiSS-C) is employed in industrial appli- from the more mundane tasks of device connectivity so they
cations. The specification has its roots in the Synchronous can focus on the features and capabilities that will make their
Serial Interface (SSI). The BiSS-C interface consists of two system solutions truly distinctive in the marketplace with sig-
uni-directional or bi-directional lines for the clock and data. nificant competitive advantages.
As with all interfaces supported by Position Manager technology, (160221)
a BiSS-C master running on a C2000 F28379 MCU can connect
directly to a BiSS-C encoder slave on a position sensor (Fig- The platform bar, C2000 and Delfino are trademarks of Texas Instruments. All
ure 6). The interface transmits position values and additional other trademarks are the property of their respective owners.
information directly from the encoder to the MCU.
The MCU is able to read and write directly to the encoder’s
internal memory. TI’s Position Manager technology includes a
feature-rich BiSS-C library of capabilities, which system develop-
ers can readily draw on for their development projects. For
example, clock frequencies of 8 MHz are supported on cables
up to 100 meters long. In addition, the C2000 MCU BiSS inter-
Advertisement
REDCUBE Terminals are the most reliable high-power con-

tacts on the PCB level. Low contact resistance guarantees
minimum self-heating. Four different designs cover all leading
processing technologies and offer a wide range of applications.
www.we-online.com/redcube
 Flexibility in processing and connection technologies

 Highest current ratings up to 500 A
 Board-to-Board and Wire-to-Board solutions
 Extremely low self-heating
 Robust mechanical connection
#redCUBE
REDCUBE PRESS-FIT REDCUBE PLUG REDCUBE SMD REDCUBE THR

Welcome to the DESIGN section
By Clemens Valens, Elektor Labs
SHARE Heinrich Rudolf Hertz
(1857 – 1894)
The German city of Hamburg has fathered two famous concepts:
the hamburger sandwich better known as The Burger, and the
hertz, the SI unit of frequency. Both date back to the second
half of the nineteenth century, but it is the second concept that
retains our attention in this bimonthly column because of the
importance of frequency (symbol: f) in electronics. So who was
the man who invented frequency?
Heinrich Rudolf Hertz was born in a wealthy family, his father

was a barrister and at some point even Senator of
Hamburg. Heinrich quickly showed a keen inter-
est in science and languages and went to good
universities. He obtained his PhD in Berlin where
he studied under Hermann von Helmholtz and
Gustav Kirchhoff, two scientific heavyweights
well known to electronics engineers. Heinrich
DESIGN
did not invent frequency, of course. He started to work on electromagnetism when

Helmholtz suggested that he had a shot at proving Maxwell’s the-
ory. Although initially he did not think it was feasible he returned to
the subject years later when he accidentally invented the tools that
made it possible. With Maxwell’s theory proven he let it rest as he
did not think that there would be much use for it.
Not a one-trick pony, Heinrich also laid the foundations for the cur-
rent field of contact mechanics when he solved the contact problem
of two elastic bodies with curved surfaces. With his observation that
a charged object loses its charge faster when illuminated by ultra-
violet radiation he helped to establish the photoelectric effect. His experiments
even place him at the beginning of X-rays. That Heinrich lived at the edge of
knowledge was proven for the last time when he died from a rare disease that
was only discovered officially some 40 years later.
Although he died too young, Heinrich did manage to pass some of his genius
on to his offspring, allowing his youngest daughter Mathilde to become a
world-renowned biologist and psychologist. His brother’s son, Gus-
tav Ludwig, won the Nobel Prize for Physics in 1925. Comparing
photos of Heinrich and his clever nephew, the resemblance is so
LEARN
striking that one wonders that maybe his nephew’s daddy wasn’t
his nephew’s daddy, but his nephew’s daddy didn’t know?
The SI unit for frequency became the hertz (Hz) in 1960 when it
replaced the cycle per second (cps). One hertz is defined as 1/sec-
ond or s-1, exactly as the becquerel (Bq), however one hertz means
one event per second with the events spaced exactly one second
apart, whereas one becquerel means one event per second on average.
Where Joseph Henry has a mountain range named after him, Heinrich
has a crater on the dark side of the moon that caries his name. Searching
the Internet I did manage to find a recipe for a Hertz burger, nicely com-
bining the two great nineteenth century Hamburger concepts into one.
(160247)

labs project reader’s project
Smart Power Bar

230V / 115V
By Hans-Norbert Gerbig (Germany)
BTA 08-600S
MASTER A1
G
TRI
A2
You can make a smart power F

SLAVE
bar (also called ‘block’; ‘strip’) < 6A3
with one single semiconductor

1
component: a triac!
It works like this: the full AC grid voltage directly to the AC grid with no galvanic The type of triac used here depends on
(120 V or 230 V) is applied to connection isolation, it is imperative to ensure that the amount of current drawn by the mas-
A1 of the triac and the master equipment the circuit is fully enclosed to avoid any ter and slaves. A BTA08-6008 (600 V/8 A)
live feed connects to the gate (G) lead. possibility of accidental contact with any triac requires a very small gate current
Live supply to the slave equipment comes part of the circuit during use. The Pro- and low trigger voltage.
from the A2 lead on the triac (Figure 1). tective Earth (US: Ground) lead should (130419)
remain intact to all the sockets.
When the master equipment is switched
on, its supply voltage flows between A1
and G. The triac conducts and provides
a live supply to all the slave equipment
from its A2 pin. When the master equip- K3 K4 K5
F
SLAVE SLAVE
ment is turned off the gate current drops
< 6A3 SLAVE
to zero, the triac turns off and power to
the slave equipment also turns off. A2 BTA 08-600S 230V~ 230V~ 230V~ / 115V~
TRI
K1 A1 G
It also functions if the gate and A1 con-
nections are switched so that the AC input
AC power *
connects to the gate and A1 provides 230V / 115V
power to the master socket. In this con-
K2
figuration however combined power to the
master and slave will now pass through MASTER
the gate instead of A1, see Labs com-
ment below. 230V~ / 115V~
Figure 2 shows how the triac is wired 2

to a 4-outlet power bar to provide three
slave outlets. The triac can be soldered to
a small piece of stripboard together with
Elektor Labs feedback:
a three-way terminal strip. In the power
bar cut the Live and Neutral between the This is a neat trick we haven’t seen before. It does raise the question whether a
first outlet socket and the remaining three triac can be safely used in this way. Looking at a typical triac die layout the gate
sockets. The triac ‘module’ can then be region is relatively small compared to the A1 and A2 connections. You can’t expect
wired according to the diagram using the gate to pass the same amount of current. Just how much current the gate can
short lengths of wire. handle is unclear. Most of the triac data sheets we looked at from many different
manufacturers didn’t specify a maximum value. One data sheet however gave a
For protection it’s recommended to con- maximum gate current of 500 mA. Bearing this in mind, be aware that the master
nect a slow-blow fuse in series with the equipment must not draw too much current.
slave outlets. As the circuit connects

learn design share

labs project reader project
Sand Clock
A real eye-catcher
By Ilse Joostens & Peter S’heeren (The Netherlands)

learn design share
This nifty gadget built around an Arduino Uno uses some servos and a pantograph mechanism to write
the time in a sand bed. After a configurable time interval the sand is smoothed out by a pair of vibration
motors and the cycle starts over again. Along with the automatic time mode, the clock software has a
command mode which allows you to control the pen using simple commands.
Figure 1. The kit contains all the parts (including the AC adapter) necessary Figure 2. The positioning of the two vibration motors beneath the sand bed
to build the Sand Clock. However, you have to program the Arduino yourself is critical for shaking the sand surface smooth.
using the free download software.
The original idea for this project came Most of the mechanical parts of the Sand age. Another relative from a mechani-
from one of our friends at the German Clock are made from 3-mm extruded cal perspective is the Autopen signing
magazine Make:, who in turn was inspired PMMA (acrylic) sheet, also known by the machine, which uses a pantograph mech-
by the plotter clock built by FabLab in brand names Perspex and Plexiglas. Along anism to duplicate handwritten signatures
Nuremberg (Germany). During the first with being readily available, this material on paper. These machine are still used by
hardware team meeting with the folks at can be cut very easily with a laser cut- high-ranking politicians and public figures.
Make:, the authors selected a number of ting machine, resulting in nicely finished They are also sometimes used in fund
projects that appeared to be suitable for edges. The thermoplastic properties of raising campaigns because an Autopen
adaptation and modification by Elektor the material are an added bonus. signature looks more authentic than a
Labs to produce DIY kits for retailing. Here we use three model servos to drive printed signature. Many US presidents
The first project tackled by the Elektor the clock mechanism, the same as in the have been well-known Autopen users,
Labs is the Sand Clock described here. original plotter clock. The torque neces- including Barak Obama, who used the
sary for this is very small, so the require- machine to sign a number of laws while
We started by modifying the mechanical ments on the servos are fairly relaxed. visiting France or on vacation in Hawaii.
components to make them suitable for Due to the simple drive shaft mounting The “Inverse Kinematics” inset gives
a kit, and easy to assemble. Then we arrangement for this application we opted an overview of the equations which form
designed an Arduino shield for the addi- for servos with metal gears, which also the basis for the operation of the Sand
tional electronics, and finally we rewrote keeps the mechanical play reasonably Clock, and which were also used to pro-
the original software to make it easier small. duce the software for the project.
to understand and easier to use, and to The sand bed is shaken smooth by two
add more features. All of that together 6-mm vibration motors mounted at a 45° Vibrating sand bed
resulted in the project described here. angle below the sand bed and secured The most difficult part of the project was
with plastic cable clamps. not the dense mathematics described in
A kit with all the parts, including an the inset, but instead the design of the
Arduino Uno and a fully assembled Pantograph mechanism sand bed. We tried out a lot of prototypes
shield, is available in the Elektor Store The key mechanical components of the before arriving at a satisfactory solution.
(see Figure 1). If you have access to a Sand Clock are the two servos (left and The vibration motors are supposed to
laser cutting machine and are not afraid right) and the acrylic arms, which posi- shake the sand, but not the entire clock,
of a bit of SMD soldering, all of the CAD tion the drawing pen. This arrangement so the energy should be confined to the
files and Gerber files are available as is called a pantograph due to its similarity sand bed as much as possible. For this
free downloads [1]. to pantograph drawing tools of a bygone purpose the sand bed is mounted loosely

on four supports and is able to move to

+5V_2 D1 D2
a certain extent in the horizontal plane. M+
D3 C4
The biggest challenges were to get the 3x
S2J-E3
sand smooth enough within about five IC1
100n
M–
R-785.0-1.0 9V / 1A
seconds to allow the time to be clearly +9V +5V_2 R4
T1
written in the sand, and to keep the sand 1k8

R10
from piling up in specific locations over C6 C5
BCX56
100k
the course of time. After experimenting DC IN USB
22u 10u
with different configurations and a variety MOD1 SCL
of vibration motors, it became clear that ARDUINO SDA

AREF
the desired results could not be achieved +5V_2
NC AGND
+3V3
with a single vibration motor. In the end IOREF D13
SERVO_R
RESET D12
we obtained good results with a pair of +5V_1 3V3 D11/PWM
1
2
6-mm vibration motors. The positioning +3V3 5V D10/PWM
3
POWER
GND D9/PWM
of the motors is crucial, and they have to
DIGITAL INPUT/OUTPUT
GND D8
+9V
be oriented at a 45° angle to the edge of C1 VIN
SERVO_L
D7 R3
the sandbox base plate (see Figure 2). 100n +5V_1 D6/PWM 220R
1
2
20 AN0 D5/PWM R2
R8 R9 3
VDD AN1 D4 220R
ANALOG INPUT
The type of sand is also important. The IC2 AN2 D3/PWM
4k7
4k7
R1
19 7
VBAT CLKOUT AN3 D2 220R
flow characteristics of sand depend on 2
SDI SDA/CE
4
AN4 D1/TX R5 R6 R7
SERVO_Z
3 1 1
the grain size (from coarse to fine) and B1 SDO SCL AN5 D0/RX
4k7
4k7
4k7
6 5 2
C2 TS IFS
the shape of the sand grains (sharp or 17
INT BBS
18 3
round). We obtained good results with 100n PCF2129A C3

CR2032 VSS
fine white sand and a grain size of 0.1 8 100n
to 0.3 mm. Due to the layered structure

160065 - 11
of the sandbox, we do not recommend
the use of extremely fine sand.
Figure 3. Schematic diagram of the shield designed specifically for the Sand Clock. It contains a DC/
The sandbox has a rim on all four sides
DC converter for powering the servos, a real-time clock IC, a driver for the vibration motors and
with a slight overhang to prevent the sand connectors for the servos.
from spilling over the edge.
Electronics
The main electronic module of the Sand You can also control the Sand Clock manually
Clock is an Arduino Uno, which provides
enough processing power for this appli-
using commands
cation despite the many equations shown
in the inset. There are also some other
components necessary for powering the as possible, we designed an Arduino the DC/DC converter still have to be sol-
servos, driving the vibration motors and shield for the additional electronics. The dered on the board.
keeping the time correct when the power shield PCB (Figure 4) is mainly populated The clock is powered by a standard
is disconnected (see the schematic dia- with SMDs. To avoid potential soldering switch-mode AC adapter with an output
gram in Figure 3). difficulties, the board in the kit is supplied voltage of 9 to 12 V. The vibration motors
To make building the Sand Clock as easy pre-assembled. Only the connectors and and servos draw a fair amount of current,
Advertisement
Enclosures and Platforms

for Pi and Arduino
www.hammondmfg.com/1593HAM.htm
www.hammondmfg.com/1593HAMEGG.htm
01256 812812
[email protected]

learn design share
Inverse Kinematics
The objective of in- θ2 is thus:
end effector verse kinematics Θ2= α 2− β 2
for the Sand Clock
is to determine the We use the triangle P1-P3-P5 to calculate
angles θ1 and θ2 of angle θ1. The position of P5 (x5,y5) is a
the actuators (the variable which first has to be calculated.
positions of the two The triangle P4-P5-P6 is a rigid body with
servos) for a given constant sides and angles.
pen position (x6,y6).
Since the pivot In order to calculate (x5,y5), we assume
joints have only one that triangle P4-P5-P6 with side L56
degree of freedom rotates about P6 through an angle α5 as
and there are only indicated in the diagram.
two actuators, the First we determine the angle α6:
mathematical equa-
⎛ L 2+ L 2− L 2 ⎞
tions are fairly sim-
α 6= arccos ⎜ 26 46 24
⎟
ple and can be han- ⎝ 2 ⋅ L 26⋅ L 46 ⎠
dled by an Arduino
module based on We already know angle α2. Now we can
a n ATm e g a 3 2 8 p determine angle α5:
microcontroller.
α 5= π + α 2 + α 6− α 564
actuator actuator To simplify the cal-
culations, the left The position of P5 is then given by:
and right servos are
x 5= x 6 + L 56 ⋅ cos(α 5 )
A ‘kinematic chain’ in robotics is a math- assigned the same Y position: y1 = y2.
ematical model of a mechanical system The diagram shows the various positions, y 5= y 6 + L 56 ⋅ sin(α 5 )
consisting of a set of rigid bodies and lengths and angles. Some of them are
joints. The motions of kinetic chains can constant, and others are variable. Now we return to angle θ1. We derive
in turn be modelled using mathematical this angle from α1 and β1. α1 is the
equations. Constant: (x1,y1), (x2,y2), L13, L 24, L35, angle between the X axis and the line
L45, L46, L56, α564 passing through P1 (left servo) and P5:
Forward kinematics means that the con-
⎛ y − y ⎞
figuration of the kinematic chain is cal- Variable: (x5,y5), (x6,y6), L15, L26, L46, α1, α 1= arctan2 ⎜ 5 1 ⎟
culated based on the positions of the β1, θ1, α2, β2, θ2, α5, α6 ⎝ x 5 − x1 ⎠
joints. Inverse kinematics is the oppo-
site. It takes the desired configuration of First we determine the angle θ2. We We use the sides of triangle P1-P3-P5 to
the kinematic chain as the starting point derive this angle from α2 and β2. α2 is calculate angle β1. Length L35 is known.
and calculates the joint positions neces- the angle between the X axis and the We calculate length L15 as follows:
sary to achieve that configuration, usu- line passing through P2 (right servo)
ally with the objective of determining the
position and trajectory of an end effector
and P6 (pen):
L15= ( x 5− x1 )2+ ( y 5− y1 )2
⎛ y − y ⎞
α 2= arctan 2 ⎜ 6 2 ⎟
or manipulator.
⎝ x 6− x 2 ⎠ β1 is then:
Outside the domain of robotics, inverse ⎛ L 2+ L 2− L 2 ⎞
kinematics is also used in 3D animations We use the sides of triangle P2-P4-P6 to β 1= arccos ⎜ 15 13 35
⎟
and computer games. The motions of the calculate angle β2. Length L46 is known. ⎝ 2 ⋅ L15⋅ L13 ⎠
human body and of animals can also be We calculate length L26 as follows:
modelled this way. Finally there is θ1:
2 2
L 26= ( x 6− x 2 ) + ( y 6− y 2 ) Θ1= π − β 1− α 1
When we apply inverse kinematics to the
sand clock, the kinematic chain consists
of four rigid bodies and five pivot joints, Now we can determine β2: These calculations are implemented in
as shown in the accompanying illustra- the sketch in the function pen_calc.
⎛ L 2+ L 2− L 2 ⎞
tion. Two of the five pivot joints have an β 2= arccos ⎜ 26 24 46
⎟
actuator, which in this case is a servo. ⎝ 2 ⋅ L 26⋅ L 24 ⎠
The end effector here is the pen.

so it is not a good idea to use the 5 V 5-V tolerant, so they can simply be con- together with the sketch [2]), as well as
supply voltage from the Arduino board. nected to the I2C interface of the Arduino the clock functions which write the time
The 5-V voltage regulator on the Arduino without any need for level shifting. in the sand.
Uno is a low-drop linear regulator with a
maximum rated load current of 800 mA. Arduino sketch We also made several internal functions
That does not provide much margin, and The sketch for the Sand Clock is designed of the sketch accessible via commands,
due to the low efficiency things get pretty for the Arduino Uno, and it also works so that you can move the pen under man-
warm while the time is being written. with the Elektor Uno R4. The sketch ual control, draw figures in the sand,
With a USB connection between a com- includes all the functions necessary for and switch the vibration motors on and
puter and the clock, there is also a risk calibration of the Sand Clock after assem- off. For this purpose the sketch has two
that the computer’s USB supply voltage bly (described in detail in the assembly modes: autonomous and command. In
will collapse if you forget to plug in the instructions, which can be downloaded autonomous mode the Sand Clock oper-
AC adapter.
For the above reasons, we decided to
provide a separate power supply for the
motors, isolated from the supply for the Component List
Arduino. For this we use a switch-mode Resistors 2 pcs. bolt, M3x6, zinc-plated steel, Pozidrive
R1,R2,R3 = 220Ω, SMD 0805 DIN 7985A
DC/DC converter in the form of a 78xx
R4 = 1.8kΩ, SMD 0805 2 pcs. bolt, M3x8, zinc-plated steel, Pozidrive
replacement with three leads (IC1), which R5–R9 = 4.7kΩ, SMD 0805 DIN 7985A
is connected directly to the input voltage R10 = 100kΩ, SMD 0805 15 pcs. bolt, M3x10, zinc-plated steel,
Pozidrive DIN 7985A
of the Arduino board. It provides an out-
Capacitors 1 pc. bolt M4x30 plastic, Phillips (sharpen end
put voltage of 5 V at 1 A maximum with with pencil sharpener)
C1–C4 = 100nF, SMD 0805 MLCC
an efficiency of over 90%, with an input C5 = 10μF 10V, SMD 1206 MLCC
6 pcs. nut, M2, zinc-plated steel, DIN 934
7 pcs. nut, M3, steel, DIN 934
voltage in the range of approximately 9 C6 = 22μF 16V
3 pcs. locking nut, M3, zinc-plated steel, DIN
to 12 V. 985
Semiconductors
The vibration motors are driven by an 1 pc. nut, M4, polyamide
D1,D2,D3 = S2J-E3 2 pcs. washer, M3, zinc-plated steel, DIN 125A
NPN transistor (T1) under control of T1 = BCX56 4 pcs. washer, M3, plastic, DIN 125A
an Arduino I/O pin. The voltage on the IC1 = SIP3 5V/1A DC/DC converter (Würth 4 pcs. standoff, 3mm high, polyamide, for M3
Elektronik # 173 010 578)
vibration motors is reduced somewhat by 4 pcs. spacer, 25mm, M/F M3, nickel-plated
IC2 = PCF2129A brass (min. 33mm total height, e.g. TME
diodes D1 and D2 in combination with the
TFM-M3X25/DR213)
saturation voltage of the transistor (VCE- Miscellaneous 2 pcs. cable clip, polyamide, Panduit type
). This is necessary because the oper- K1 = set of SIL pinheaders, 0.1’’ pitch (1 pc. CCS25-S10-C
sat
10-pin, 2 pc. 8-pin, 1 pc. 6-pin) 4 pcs. self-adhesive rubber foot
ating voltage of the motors (according to SERVO_Z,SERVO_L,SERVO_R = 3-ppin right- 3 pcs. micro servo, Tower Pro MG90S of MG90
the data sheet) is only 3 V. angled pinheader, 0.1’’ pitch with metal cogs
The control signals for the servos are sup- B1 = CR2032 button cell with holder 2 pcs. vibration motor, round, 6mm diam.,
Arduino UNO R3 or equivalent type VM6ZK273 (JPR Electronics # 450-007)
plied directly by three I/O pins of the
Extruded (XT) PMMA, 3mm, clear, laser-cut
Arduino board. Three connectors are Mechanical parts: Fine-grain white sand (see text)
mounted on the shield for the three ser- 6 pcs. bolt, M2x10, Pozidrive/Phillips Nourishment coloring additive (optional)
6 pcs. bolt, M2.5x8, zinc-plated steel,
vos. Resistors R1, R2 and R3 are there
Pozidrive DIN 7985A A DIY kit including all parts and the SMD-
to protect the Arduino if one of the servo 6 pcs. bolt, M2.5x12, zinc-plated steel, preassembled shield is available through the
cables is accidentally plugged in with the Pozidrive DIN 7985A Elektor Store.
wrong orientation. Resistors R5, R6 and
R7 keep the levels on the data lines to the
servos stable when the Arduino micro-
controller is in sleep mode.
A real-time clock IC is included to main-
tain the correct time, so the clock does
not have to be set again every time it
is powered up. For this we chose an
PCF2129A (IC2), which has a built-in
crystal. This IC is inexpensive and very
precise, with internal temperature com-
pensation and an accuracy of 3 ppm. A
CR2032 button cell serves as a backup
battery with a lifetime of about ten years.
The RTC IC is not suitable for a 5-V sup-
ply voltage, but fortunately the Arduino
Figure 4. The shield PCB is mainly populated with SMDs, which are pre-assembled on the
board provides a 3.3-V supply voltage on
board in the kit.
the shield connectors. Equally fortunate
is the fact that the I2C pins of the IC are

learn design share
ates independently. In command mode positions of the left and right servos, connectors. Then connect the AC adapter
the sketch executes the commands you expressed in radians. This calculation is to the Arduino board and connect the
send to the sand clock. performed by the function pen_calc and USB port of the Arduino to a computer
can either succeed or fail. The calculation on which the Arduino IDE is installed.
After uploading the sketch to the Arduino, fails when the stated coordinates cannot Compile the sketch [2] and load it into
you can use the serial monitor of the be reached by the pen. If the calculation the Arduino.
Arduino IDE to send commands to the succeeds, the resulting servo positions
sand clock. Select 9600 baud and some- are converted from radians to PWM pulse After the sketch is loaded and the Arduino
thing other than “No line ending” in the widths for the left and right servos. is reset, the servos are automatically
adjacent selection box. The commands moved to their midpoint positions (pulse
consist of letters (e.g. svd), optionally The sketch uses floating point operations width 1500 ms). Disconnect the Arduino
followed by one or more parameters to calculate the inverse kinematic func- from the computer and disconnect power
(for example, ps -20.55 +55.8). Press tions. The floating-point variables have to the Arduino. Then remove the shield
the Enter key to send a command. The a width of 32 bits. The standard used and disconnect the servos so they can
Arduino returns a response for most of here is IEEE 754. All of these properties be mounted on the clock. If you have a
the commands. All available commands are determined by the AVR GCC compiler servo tester, you can use it instead to
are listed at the start of the sketch. For used to compile the sketch. set the servos to the midpoint position.
example, you can use the command pa All positions and distances are expressed Now it’s time to start the mechanical
to draw an arc. in millimeters. These values correspond assembly. Attach four rubber feet to
to the real world. The sketch includes the bottom of the base plate and mount
Now let’s get back to the original sub- definitions of the physical dimensions of the four supports for the sandbox. Next
ject. During the assembly process you run the mechanical parts of the clock. Angles assemble the base frame with the lift
through the calibration procedure. This are expressed in radians. servo, and then mount this assembly
procedure requires you to enter a series The diagram in Figure 5 shows the posi- along with the Arduino and the shield
of commands, in some cases with accom- tions (in millimeters) of the left and right on the base plate.
panying parameters. Once all the settings servos with respect to the sand bed, as
have been made (see the command sed), defined in the sketch. Now it’s time for the left and right servos
they are written to the EEPROM (com- of the pantograph mechanism. Connect all
mand sew). During start-up, the sketch During drawing in the sand the pen is the servos to the shield again and assem-
checks the validity of the EEPROM con- moved in steps with a maximum step ble the arms and pen of the pantograph.
tent. If the EEPROM content is valid, it size of 0.25 mm. Despite the many cal- When you assemble the arms of the pan-
determines the start-up mode of the culations, the processor is more than fast tograph, thread the screws directly into
clock. In most cases you will probably enough, so the code waits 5 ms between the acrylic plastic. That works quite well
set it to autonomous mode. If the content successive changes in pen position during thanks to the previously mentioned ther-
is invalid, the clock starts up in command drawing to prevent the pen from being moplastic properties of this material:
mode, the same as the first time after pushed through the sand too quickly. the heat generated by friction when the
you upload the sketch. You can interrupt autonomous mode at screws are threaded in effectively causes
Note that the EEPROM of the microcon- any time by sending an end of line (EOL) the plastic to flow around the threads.
troller is not cleared when the sketch is character over the serial interface. Press This mounting technique minimizes the
uploaded. This means that you do not the Enter button in the serial monitor mechanical play of the arms. However,
have to enter the settings again (and to switch the sketch to command mode. the mounting hole for the pen does have
save them with the sew command) every Bear in mind that the transition to com- to be threaded. If you cut your own parts
time you upload the sketch. If necessary, mand mode may take a while when the using a laser cutting machine, you have
you can use the command sec to erase clock is busy writing in the sand. You can to tap that hole with an M4 thread tap.
the EEPROM. return to autonomous mode by sending Mount the sandbox, but leave the vibra-
the command ma. tion motors aside for the time being. Now
In autonomous mode the Sand Clock peri- it’s time for the calibration. Connect an AC
odically draws the current time in the Construction adapter to the Arduino board and plug it
sand. This drawing action consists of a In this article we only provide a brief in. Then connect the Arduino board to the
number of pen motions in the horizontal description of how to build the Sand computer through a USB cable. Now you
plane, as well as up and down motion. Clock. If you want to build the project, can use the serial monitor of the Arduino
The sketch uses the current X/Y pen posi- please consult the assembly instructions IDE to send commands that set the servo
tion as the starting point for all types of [2] for a more extensive description with positions. To calibrate the servos, you
motion: drawing a straight line, tracing many detailed photos. have to determine the positions where
an arc, drawing a character, or reposi- Start by mounting the connectors and the they are exactly vertical and horizontal.
tioning the pen. DC/DC converter on the Arduino shield These positions are then stored in the
board with the pre-assembled SMD com- EEPROM. The next task is to calibrate the
The function pen_set moves the pen to ponents. Then plug the shield onto the lift servo. Put the sandbox in place and
the stated X/Y position. First the coor- Arduino board and connect the servos. position the lift servo so that the tip of
dinates are converted to the angular Pay attention to the orientation of the the pen hovers a few millimeters above

134
114
74
54
10
(0;0)
30
( 0 ; -30 )
( -13,17 ; -30 ) ( 13,17 ; -30 )
Figure 5. The positions (in millimeters) of the left and right servos relative Figure 6. The structure of the Sand Clock is clearly visible here.
to the sand bed, as defined in the sketch.
the bottom of the box, regardless of the X/Y position. You can
also adjust the distance by turning the pen higher or lower in
the mount. Once the pen is in the right position, tighten the
plastic M4 nut to hold it in place. Next you have to set the mid- Advertisement
dle position and the highest position of the lift servo. These are
not critical, and usually you can simply take the values from
the assembly instructions.
USB Add USB to your next project.
It's easier than you might think!
Write the settings to the EEPROM and then set the clock to the
right time. Now you can do a dry run with the clock (without DLP-USB1232H: USB 2.0 UART/FIFO
sand) to check its operation. If that all goes well, you can set
the clock to autonomous mode so that it automatically writes
HIGH-SPEED
the time at fixed intervals, and then disconnect it from the 480Mb/s
computer. On
Multipurpose: 7 interfaces ly $
Now disconnect the Arduino power source and mount the vibra- 28.
95!
tion motors underneath the sandbox. Twist the matching leads Royalty-free, robust USB drivers
of the two motor cables together and solder the ends to the No in-depth knowledge of USB required
solder pads on the shield. Be careful not to touch the acrylic
Standard 18-pin DIP interface; 0.6x1.26-inch footprint
parts with the hot soldering iron. Then place the sandbox on
the supports, taking care that the vibration motors do not touch
any leads and are free to turn.
DLP-IO8-G DLP-IOR4
Finally, pour the sand into the sandbox. That completes the 4-Channel Relay Cable
8-Channel Data Acquisition
construction of the sand clock. DLP-TH1b
Temp/Humidity Cable
Only
If you wish, you can give the sand a special tint. To do so, put $29.95! DLP-RFID1
the sand in a tin with a lid and add a few drops of food color. HF RFID Reader/Writer
Then shake the closed tin until the sand has a uniform color. DLP-FPGA
8 I/Os: Digital I/O USB-to-Xilinx FPGA Module
Repeat if necessary until you obtain the desired color. Let the
Analog In
sand dry completely before pouring it into the sandbox.
Temperature
(160065-I)
USB Port Powered
Single-Byte Commands
Web Links
[1] www.elektormagazine.com/160065
www.dlpdesign.com
[2] www.elektor.com/sandclock-160065-71

learn design share
Super Simple Dice

Cheap and microcontroller-free!
By Roy Aarts (Elektor Labs)
This simple electronic dice was designed specifically for beginners in electronics. The board
has a spacious layout, with only a handful of components, which are all through-hole mounted.
There is no microcontroller in this circuit; all the hard work is carried out by
two CMOS ICs.
For several decades the dice has been a

willing subject for a small electronics cir-
cuit. Over the years, Elektor has already
designed and published a large number
of circuits for a dice. It often became a
challenge to use as few components as
possible, but this sometimes resulted in
some small flaws, such as a less than
realistic representation of the spots. (The
number of LEDs lit up did give a useable
result, but the layout of the LEDs was
not the same as that of the spots on a
real dice). We’ve even managed to design
a dice using a single IC from the 4000
series, complete with a touch-sensitive
control. The only ‘disadvantage’ was the
need for a 9-V battery (fairly large and
not cheap).
The aim of the design published here was itor C1. With the values (to limit the current
to make a simple circuit with LEDs, which used here, the frequency is about through the LEDs)
didn’t have any ‘difficult’ components, 5 kHz. This signal then goes to the clock and Schottky diodes in
and which was easy to construct. A bat- input (pin 14) of decade counter IC1, a certain configuration to
tery holder with a CR2032 button cell which is a 74HC4017. By connecting the the six outputs (Q1-Q5 and CO) of IC1.
is mounted on the board, which makes Q6 output (pin 5) to the Reset input (pin The diodes prevent short circuits between
the circuit easy to carry around. This 15), the counter can only count from 1 an active output and any non-active out-
configuration also makes it very easy to to 6, when it will receive a reset pulse. puts that drive the same LEDs. Since the
replace the battery. This is a very neat This is exactly what we need for our dice. carry-out output of the IC (pin 12) is high
kit to make people interested in elec- The pulses at the clock input are only after a reset and when outputs Q0 to Q4
tronics, even if they’ve never soldered counted by IC1 when its enable input is are active, the diode matrix produces the
before. With a bit of help, anybody could low. When a pushbutton and a pull-up following numbers: pin 2 high: 3, pin 4
build this! resistor are connected to this pin, we high: 4, pin 7 high: 5, pin 10 high: 6,
can make IC2 count for as long as S1 pin 1 high: 1, pin 12 high: 2.
The circuit is pressed. When the pushbutton is
The complete circuit is shown in Figure 1. released, one of the Q1-Q5 and CO out- Since the supply voltage to the circuit is
This consists of an oscillator (IC2) and a puts will remain at a logic high level. so low (3 V), the diodes used are Schottky
decade counter (IC1), which has 7 LEDs Because the 555 operates at such a high types in order to keep the voltage drop
connected to its outputs. frequency, it’s not possible to manipulate across the diodes as small as possible. A
The well-known 555 timer was selected the selected output with S1, so the result lot of experimenting with the values of the
for IC1 It is configured here as an astable is completely random! series resistors took place in the proto-
multivibrator by connecting pins 2 and The seven high-efficiency LEDs (LED1- type, because the outputs of a 74HC4017
6 (Threshold and Trigger) together. The LED7) are of course arranged in the same can only supply a small current at that
oscillator frequency is determined by the pattern as that found on a real dice. They supply voltage, and the output voltage
values of resistors R1 and R2, and capac- have been connected via series resistors drops at greater currents. The values of

the series resistors have been adapted to

take account of the number of LEDs that
S2 D5
the outputs have to drive at the same
time, making the brightness of all the D4
R2 R3
LEDs roughly equal.
10k
10k
D6
A 3 V lithium button cell has been used 16
VCC 3
to power the circuit. It’s been mounted 4 8
Q0 D2 D1...D6 = BAT85
2
Q1
in a holder on the board to make it easy 7 RES VCC
IC1 Q2
4
DIS D1
to replace. The dice can be turned on and R1
IC2 Q3
7
3 14 10
off with the help of slide switch S2. Since OUT CLK Q4
10k
D3
6 1
THR Q5
the current through the LEDs is only sev- 2 TR TLC555 74HC4017 5
BAT1 Q6
CV GND 13 6
eral milliamps, a button cell should last 5 1
ENA Q7
9
Q8
several hours. 15 11
RST Q9
12
C1 S1 CO
GND
Construction 8
R5 R6 R8 R4 R7 R9 R10
10n
The printed circuit board shown in Fig-
100R
100R
220R
100R
220R
10R
10R
ure 2 has a spacious layout, with slightly
LED1 LED2
larger tracks and pads than usual, to
make it easier to construct for people
with less experience in soldering. For the
enthusiast, an attractively priced kit con- LED3 LED4 LED5
sisting of the board and all the compo-
nents can be obtained from the Elektor
shop.
The construction of the dice isn’t difficult, LED6 LED7
although a lot of things may not be clear

Figure 1. The circuit consists mainly of two ICs
to beginners in electronics. We won’t go and seven LEDs. 150767 - 11
into detailed descriptions here (such as
color codes for resistors), but we would
suggest that those with little experience
of electronics and soldering should con-
struct the circuit with the help of some-
body more knowledgeable.
With a bit of help, anybody could build this!
In any case, start with the mounting of

resistors, followed by the diodes (the
band on one end of the diode corresponds Component List
to the line (cathode) on the PCB), and the
capacitor. Next, the two IC sockets should Resistors
be soldered onto the board, followed by Default: 0.25W, 5%
the LEDs. The cathode (again indicated R1,R2,R3 = 10kΩ
R4,R5,R6 = 100Ω
by a line on the PCB) is denoted by the
R7,R8 = 220Ω
shorter of the two leads or by the flat side R9,R10 = 10Ω
on the casing. Finally it’s the turn of the
Capacitors
switch and the pushbutton, followed by
C1 = 10nF, 0.1’’ pitch
the battery holder. Now put the two ICs
in their sockets, making sure that the Semiconductors
indentation on each IC corresponds to D1,D2,D3,D4,D5,D6 = BAT85 Schottky diode
LED1–LED7 = LED, high-efficiency, red, 5mm
that shown on the board. And that’s it!
IC1 = 74HC4017
You should now check the board to IC2 = TLC555
make sure that every component has
been mounted correctly and is in the Miscellaneous
S1 = pushbutton, PCB mount, 6x6 mm
right place. Once you’re satisfied that
S2 = slide switch with changeover contact,
everything is ok, put the button cell PCB mount, e.g. C&K OS102011MS2QN1C)
in its holder, switch the circuit on, and BATT1 = CR2032 battery holder, PCB mount
(e.g. Multicomp CH25-2032LF)
press on the pushbutton. You’ll now be
CR2032 button cell
shown a number between 1 and 6 on PCB # 150767-1 or
the LEDs. Congratulations, your (first?) Kit of parts including PCB and all components: Figure 2. Everything’s on this board, including
Elektor Store # 150767-71 the battery and the on/off switch.
circuit works!
(150767)

learn design share
Elektor SDR Reloaded (4)

Standalone solutions
By Burkhard Kainka (Germany)
Can we also use the SDR Shield with just an

Arduino and no PC at all? Certainly we can.
It’s entirely feasible (with a few restrictions)
because all of the vital control functions
are available on the Elektor Extension
Shield. The only task that remains then is
processing the IQ signals.
Your simplest option is to mix down the signals and lis-

ten to them directly. Even headphones plugged directly
into the audio jack on the SDR Shield will produce
something audible, albeit a little bit on the quiet side. The
receiver is then operating as a direct mixer. All we need
to do then is make a suitable control for the fine tuning.
Tuning with the Arduino

On the Elektor Extension Shield (no. 140009-91, pub-
lished in July & August 2014; [1]) we have two push-
buttons and a pot. We can use these to make a conve-
nient control setup. S1 reduces the frequency, S2 raises it and
with the pot we can adjust the increment size between 20 Hz SDR Shield/Extension Shield combo (ready for free download
and 100 kHz. On the upper line the LCD indicates the current at [2]) still supports serial inputting of frequencies, so you
frequency and below this the increment we have set (Figure can still carry out tuning from your PC. You can see how a fre-
1). We can now tune to a far finer degree than the 1-kHz steps quency set up on a PC is displayed in kHz on the LCD (Figure
we had available previously. With this control we can not only 2). However, as soon as one of the two pushbuttons on the
change between frequency bands but also manage fine tuning Shield is activated, the display changes to Hertz and indicates
for CW or SSB with ease. The frequency specified initially on also the precise VFO frequency without the 12-kHz offset (see
start-up is 7000 kHz, enabling you to go immediately to the Listing 1).
40-m band and search for CW and SSB stations. You can proceed on the PC and initiate an SDR program next.
For compatibility reasons the new software for the Arduino/ After that it’s your choice entirely whether you drive the receiver
Figure 1. Display on the Elektor Extension Shield: 7001.040 kHz, adjustable Figure 2. 198 kHz, set up using the PC.
in 20 Hz steps.

Listing 1. Excerpts from the standalone software.
void loop(void) lcd.print (" "); //freq=freqHz/1000;

{ } lcd.print(freqHz);
pot = analogRead(A3); fstepOld = fstep; setfreq (freqHz);
fstep =1000000; lcd.print (" ");
if (pot > 100) fstep =1000000; if (Serial.available()) { delay (200);
if (pot > 200) fstep =500000; freq = Serial.parseInt(); }
if (pot > 300) fstep =100000; if (freq > 0){ if (digitalRead(A1)==0){
if (pot > 400) fstep =20000; lcd.setCursor(0, 0); freqHz = freqHz + fstep;
if (pot > 500) fstep =5000; freqHz=freq*1000-12000; lcd.setCursor(0, 0);
if (pot > 600) fstep =1000; lcd.print(freq); //freq=freqHz/1000;
if (pot > 700) fstep =200; lcd.print (" "); lcd.print(freqHz);
if (pot > 800) fstep =100; setfreq (freqHz); setfreq (freqHz);
if (pot > 900) fstep =20; } lcd.print (" ");
} delay (200);
if (fstep != fstepOld){ if (digitalRead(A0)==0){ }
lcd.setCursor(1, 1); freqHz = freqHz - fstep; }
lcd.print(fstep); lcd.setCursor(0, 0);
using the PC or with the controls on the Shield. But we can +5V
also do this without involving a PC at all. The simplest way is
to use the SDR Shield as a direct mixer.
220k
1k
Direct mixer
A direct mixer transposes an RF signal down to the audio
100n ½ 100R
range in one step. Of course the Shield is actually intended LM358
47u
for an IF of 12 kHz, but the IF stages provide sufficiently wide 16V
bandwidth to allow trouble-free operation down in the lower

100k
10k
ranges up to 3 kHz. The only thing else that we could ask for log.
100k
is some extra amplification.

1k
For that reason we built a small stereo headphone amplifier

100n
with presettable volume control (Figure 3). Stereo provides
a touch of special magic in combination with a basic direct
mixer. The two signals I and Q are in fact phase-shifted by
90 degrees. With a little practice, the brain can make plenty Figure 3. One channel of the headphone amplifier.
out of this combination. A bit of phase difference can in fact
be decisive, even in our spatial hearing. When you listen to
(ideally many different) CW signals for long enough, you then an op-amp in their component stores. The LM358 is prone
start to develop a spatial impression that palpably reinforces to transient distortion if driven at high level into a relatively
selective hearing. Many users declare that they can even dif-
ferentiate signals below and above the VFO frequency, thus
eliminating the key disadvantage of direct mixers: there is no
image suppression, you see.
We have now turned the SDR Shield into a veritable deluxe

direct mixer, with a stable VFO across the whole range up to
30 MHz plus fine tuning, and on top of this, a stereo IQ output.
The best thing is that the PC remains switched off and cannot
cause any interference. Of course a receiver with no ALC what-
soever is a particular blessing on CW and SSB, as the noise
level is not increased during pauses. Normally you often have
to adjust the volume control when a very loud station comes in.
The small headphone amplifier is constructed using the twin
op-amp LM358. You could of course employ fewer compo-
nents by using a special audio IC but probably everybody has Figure 4. The experimental headphone amplifier on a scrap of perfboard.

learn design share
+3V3
1k
4mH7
22n
22u Ge
I 10k 220k
16V 1n
16 kHz
Q
1n
BC547B 4mH7
22n 1n
600Ω
GND
Figure 5. A diode detector. Figure 6. Diode detector constructed on a breadboard.
low-impedance load. This problem is mitigated by providing It does, with some limitations. You must set the VFO on the
the 1-kΩ resistor from the output to the supply rail, because zero-beat position, i.e. as precisely as possible on the exact
at smaller modulation levels only the lower part of the push- carrier frequency of the transmitter. Then you can hear the
pull output is active. In series with the headphones we also modulation clearly, admittedly with an additional beat over-
have a 100-Ω resistor. On the one hand this reduces the work laid at most times. Phase-locked tuning is not possible, you
of the op-amp and on the other it eliminates any risk of dam- see. Nevertheless, receiving AM stations is certainly feasible.
aging your hearing. Limiting comes into play for the amplifier However, it will work a lot better with the AM IQ demodulator
in time to ensure that the maximum power to the headphones to be described next.
remains within bounds at all times. Incidentally, the LM358 can
be driven almost down to GND level but not quite all the way The IQ detector
up to the power rail. Consequently the open circuit voltage is Under normal circumstances the SDR software processes a
about a third of the operating voltage. 12-kHz IQ signal. The question now arises whether you can-
The amplifier can raise the signal level by up to 40 dB. This is not also ‘solidify’ signal processing into hardware format. Our
more than you need in most situations, so normally the volume goal is now to create an AM demodulator for IQ signals. This
control is throttled well back. Receiving CW stations is truly turns out to be simpler than expected.
a pleasure with this setup. Even SSB comes in well, because The first attempt (Figures 5 and 6) was based on an IF filter
the frequency can be set so conveniently and precisely. You with two fixed inductances. An operating frequency of 16 kHz
can listen for hours like this. was determined more or less by chance, because 22 nF capac-
itors were available to hand. With a coupling capacitor of 1 nF
But what’s it like when listening to AM? Actually this is not the a good band filter with suitable bandwidth was the result.
right job for a direct mixer. But does it work nevertheless? The two signals I and Q need to be mixed together with a
+9V
470R
1k
4mH7
22n
22u Si 100u
I 10k 220k 22k 100k
16V 1n 22n 16V
16 kHz
Q
1n
3k3
BC547B 4mH7 BC547B

22n 22n
32Ω
GND
Figure 7. Expansion with an IF stage.

Advertisement
mouser.com
The Newest Products for Your Newest Designs®
Figure 8. The complete circuit from Figure 7 on a breadboard.
90-degree phase shift. At 16 kHz this works out pretty accu-

rately, with the 1-nF capacitor having a capacitive impedance
of 10 kΩ here. At this operating frequency the Q channel is
shifted by 90 degrees, whereas the I channel passes through
without any phase shift. As expected, signals are amplified at
+16 kHz and canceled out at –16 kHz. Effective image sup-
pression is the result. Our first signal processing task is now
completed.
The second task is signal filtering, which is what the band filter
is for. This always worked fine in old tube radios and still does
here, only at a lower frequency. After the filter we have the AM
demodulator, in the form of a diode detector. To underscore
Over 4 Million
the historical continuity an OA61 more than 50 years old is products from over
600 manufacturers.
employed (you can also use a 1N60A or something similar of
course). Beyond this come some high-impedance headphones
or low-impedance ‘phones with a matching transformer. Already
we can hear something. The sound is very pleasant and easy on
the ear, which is probably thanks to the band filter in the main.
The first attempt worked well only for the strongest signals and
even then it produced no great volume level. For that reason
we added a further IF amplifier stage (Figure 7). Now the AM
rectifier has some bias, which is why a silicon diode also works
here. The result is good sensitivity and plenty of volume using
32-ohm headphones. Naturally the AM detector is also suitable
for connecting to a loudspeaker amplifier.
To a greater or lesser extent, the experiments just described

for standalone SDR are all simple compromises. Of course the
perfect solution would provide genuinely digital signal processing
with a DSP. But could we maybe manage this using an Arduino?
That would be a major achievement indeed. If it’s at all feasible,
it’s likely to be an Elektor reader who would take on this task.
Have fun experimenting with your SDR Shield!
(160165)
Authorised distributor of semiconductors and
Web Links electronic components for design engineers.

learn design share
SUPRA 2.0 MC Version

Getting the most out of a tiny signal
Photo: Ortofon
Following publication of the SUPRA 2.0

high-end phono preamplifier project in Elektor
edition 4/2016, many audio enthusiasts asked us whether this
design could be adapted for use with moving-coil (MC) cartridges. A reasonable request, because the noise
reduction obtained by wiring several opamps in parallel is particularly important with low input signal levels.
Here we show you the necessary changes, which are actually fairly simple.
By Ton Giesberts (Elektor Labs) ular amongst audio enthusiasts. Many steep price due to the high-quality com-
of them have taken the effort build this ponents contained in the kit.
preamp, which offers truly superior per- However, a lot of vinyl fans were holding
The SUPRA 2.0 MM/MD phono preampli- formance. There were even quite a few back and hoping for the option of con-
fier [1] has proven to be especially pop- purchasers for the parts kit, despite its necting a moving-coil (MC) cartridge to
the SUPRA 2.0 preamp. MC cartridges
are generally better than their MM or MD
Specifications & Measurements counterparts, but they require a special
preamplifier with additional gain because
Measurements made with an input signal level of 0.3 mV at 1 kHz
their output signal level is about 0.2 to
(source impedance 5 Ω).
0.5 mV. The original SUPRA was actually
Values measured with a laboratory power supply and supply voltages of ±14 V.
designed for MC cartridges, but for the
SUPRA 2.0 we decided to initially focus
• Input impedance: 100 Ω
on MM/MD cartridges and leave MC for a
• Output level with 0.3 mV input: 270 mV
bit later. Since then we have modified a
• SNR
SUPRA 2.0 in the lab and increased the
Linear: >73 dB (BW = 22 Hz to 22 kHz)
gain, and of course we checked it out on
A weighted: >79 dB (BW = 22 Hz to 22 kHz)
the test bench.
• THD+N (1 kHz): <0.02% (only noise, BW = 80 kHz)
• THD at 1 kHz: <0.0007% (from averaged FFT measurement)
The changes
• THD at 10 kHz: <0.0002% (from averaged FFT measurement)
As previously mentioned, the changes
• Deviation from RIAA curve: <0.1 dB (50 Hz to 10 kHz)
are fairly simple. If you already have
<1 dB (20 Hz)
a fully assembled SUPRA 2.0, you only
<0.15 dB (20 kHz)
need to change nine resistors in each

V+_1
1
V+_1
3 8
R16
JP1 6
P1 IC1 6k81 C4 C7 C10 C13 C20
2
1M R3 R4 5 7 100n 7 100n 7 100n 7 100n 8 100n
R5
2k2
IC1 IC2 IC3 IC4 IC5
10M
10R
4 C5 4 C8 4 C11 4 C14 4 C21
R6 C3
10R
V-_1 470p
100n 100n 100n 100n 100n
R2 1
C2
3 8
10M
MC 6
R17
V-_1 Line
IC2 6k81
220n 2
K1 R7 5 6 K3
R8 R27
1 3 7 1
2k2 C19 IC5B 47R
10R
1 5
2 R9 C6 IC5A 2
2
10R 2µ2
R1
C1 C1' 470p R26
1 C15 C16 R24 R25
100R
62k 4k7
7k5
3 8
100p
* IC3
6
R18
6k81
22n 22n C17 C18
2 R22 R23 22n 22n

5k1 470R
R10 5
not needed R11 R21
2k2
10R
1k6
R12 C9 IC1, IC2, IC3, IC4 = LT1028
10R
R20 IC5 = LM833
470p
1
110R
3 8
R19
6
IC4 6k81
2
R13 5
R14
2k2
10R
R15 C12
10R
470p
Figure 1. Schematic diagram of one channel of the SUPRA 2.0 with altered resistor values for use with MC cartridges.
channel. The portion of the schematic harm in leaving them where they are. Of Another reason for reducing the nominal
where changes are necessary is shown course, if you are building the SUPRA 2.0 output signal level compared to the MM/
in Figure 1. from scratch you can simply omit these MD version of the SUPRA 2.0 (468 mV
Most MC cartridges are specified for a capacitors. with a 2.5 mV input signal) is to keep
termination impedance of about 100 Ω. the bandwidth of the four parallel opa-
To achieve that, the resistors which deter- Next, all sixteen 47 Ω resistors in the neg- mps large enough despite the fairly high
mine the input impedance of the pream- ative feedback networks of IC1–IC4 and amplification factor of the MC version.
plifier (R1 and R30) must be replaced by IC6–IC9 (R4, R6, R7, R9, R10, R12, R13, The open-loop bandwidth of the LT1028
100-Ω resistors. However, if your car- R15, R33, R35, R36, R38, R39, R41, R42, opamps is 70 MHz. With an amplification
tridge is specified for a different input R44) must be replaced by 10-Ω resistors. factor of 221 the bandwidth is a bit more
impedance (for example, 47 Ω), then you That puts the overall amplification factor than 300 kHz. The 470 pF capacitor in
should use that value for R1 and R30. of the preamplifier at 900, with the four the feedback network limits the band-
opamps wired in parallel accounting for width of the input stage to 150 kHz. The
Capacitors C1 and C26 determine the the lion’s share (amplification factor 221). resulting effective bandwidth is approx-
input capacitance of each channel of To determine the necessary amplification imately 115 kHz.
the preamplifier. With MM and MD car- factor, we took the average output signal
tridges you can influence the high-fre- level of 26 MC cartridges, which yielded a If we increased the gain of the input stage
quency characteristic with this capaci- signal level of slightly more than 0.3 mV. even more, there would be a noticeable
tance, but with MC cartridges it has little As the line input of most audio ampli- effect on the signal amplitude at 20 kHz.
effect because their internal resistance fiers has a sensitivity of 100 or 200 mV, If more gain is needed, it would be better
is only a few ohms and their inductance we thought it prudent to choose a level to increase the gain of the second stage.
is negligible due to the small number that is not much higher, namely 250 mV, However, adjusting the gain of that stage
of turns in the coil. You can remove C1 to avoid making the amplification factor is not straightforward because it incor-
and C26 from the board, but there is no greater than necessary. porates part of the RIAA correction. It

learn design share
Summary of changes w.r.t.

original SUPRA 2.0
• R1,R30 = 100 Ω / 1% (depending

on actual MC cartridge)
• R4,R6,R7,R9,R10,R12,R13,R15,
R33,R35,R36,R38,R39,R41,R42,
R44 = 10 Ω / 1%
• C1,C26 = Not necessary, but can

be left in place if already present
• P1/R2/R3/C2/JP1 and P2/R31/R32/

C27/JP2 = Not necessary, but can
be left in place if already present
Figure 2. The PCB area with the modified components.
would therefore be necessary to change of the four opamps to be nulled if nec- leave them in place.
all the resistors in that stage. essary. Due to the low input impedance, For the MC version of the SUPRA 2.0
these networks (P1/R2/R3/C2/JP1 and we opted for the type A version of the
There are also several compensation P2/R31/R32/C27/JP2 are not necessary LT1028. It costs a lot more than the stan-
circuits on the SUPRA 2.0 PCB for each in the MC version. However, if they are dard version (which is already expen-
channel to allow the input bias currents already fitted on the PCB you can simply sive), but it has lower input offset and

bias currents. The highest output offset

measured with the type A opamps in our IC10A was approximately 100 mV. Noise considerations
prototype was just 7 mV with an ampli- We hope you enjoy listening to your vinyl
fication factor of 221. collection with the SUPRA 2.0 MC pre-
amplifier feeding the signal from your
You should bear in mind that the sum of top-end record player and MC cartridge
the four offset voltages will be amplified to the rest of your system.
by a factor of 40 in the second stage. If (160263-I)
the offset voltage at the output of IC5A Web Link
or IC10A is more than a few hundred [1] www.elektormagazine.com/150616
millivolts with the standard LT1028 opa-
mps, you can consider replacing them
with type A versions. Although that is
fairly costly (you need eight in total),
you should remember that a high offset The thermal noise of the four 6.81-
voltage at the output of IC5A or IC10A kΩ summing resistors at the LT1028
will cause more current-induced noise in outputs has virtually no effect on the
the feedback networks. In our prototype signal to noise ratio of the opamps.
he offset voltage at the output of IC5A or The noise in each feedback network
(2.2 kΩ in parallel with 10 Ω) is
0.41 nV/√Hz. The input noise of
Several characteristic curves are shown below each LT1028 is 1 nV/√Hz. If the MC
to back up the specifications. cartridge has an internal resistance
of 5 Ω, its noise contribution is
Audio Precision Chart A shows the output amplitude
+2
1.04 nV/√Hz. From all that it can be
+1.8 with the input driven by a sinusoidal
+1.6 concluded that the total input noise
+1.4
signal corrected according to the RIAA
+1.2
+1
of one opamp is 1.12 nV/√Hz, or
+0.8 standard. At 20 Hz the deviation is
+0.6 1.08 nV/√Hz with the input shorted.
d
+0.4
B +0.2
a bit less than –0.9 dB. At 20 kHz
r +0
-0.2
the deviation is –0.1 dB. A slight rise
A -0.4 At the opamp output the noise
in amplitude can be seen at 138 Hz
-0.6
-0.8
-1 will be 221 times greater, which is
-1.2
and 3.3 kHz, but it is negligibly small
-1.4
247 nV/√Hz. If we add the noise of
(0.022 dB). That is a remarkably good
-1.6
the 6.81 kΩ resistor, the output noise

-1.8
A -2
20 50 100 200 500 1k 2k 5k 10k 20k
result when you consider that all of the
Hz
is 247.2 nV/√Hz. The total noise at
components responsible for the RIAA
the summing junction is half this
correction have a tolerance of 1%.
value, which makes it 123.6 nV/√Hz.
Audio Precision
+0
This figure does not take the current
-10
noise into account.
-20
-30
Chart B shows an FFT plot with a
-40
1-kHz input signal. The displayed
-50 The noise level of the MC input
-60 spectrum runs from 10 Hz to 130 kHz,
d
B -70 stage is 3.7 times that of the MD
r
with the fundamental frequency at
input stage (with the input shorted).
-80
A
-90
1 kHz suppressed. To give a clearer
-100
However, the amplification factor is
impression, the curve is the result
-110
-120
4.62 times greater in the MC version.
-130
of averaging sixteen measurements.
-140 If we recalculate the measured
B -150
No harmonics are visible on this plot,
10 20 50 100 200 500 1k
Hz
2k 5k 10k 20k 50k 100k
noise of the MD stage with the input
from which it can be concluded that
shorted (88 dB) for a 270-mV output
the total harmonic distortion is less
signal level, the signal to noise ratio
than 0.0007%.
is 4.84 dB lower at 83.15 dB. The
Audio Precision
+0 calculated difference in noise level
-10
-20 between the MC and MD versions is

-30 Chart C shows an FFT measurement
-40 11.4 dB with the input shorted.
-50 similar to that of chart B, but with
-60
d
B an input frequency of 10 kHz. Here
From these calculations we can
-70
r
-80
A again the THD is very low — only
-90
predict that the signal to noise ratio
the second harmonic is visible at a
-100
-110
of the SUPRA 2.0 MC version will be
level of ‑114 dB, corresponding to
-120
-130
about 72 dB. That is very close to
-140
0.0002% THD.
C -150
10 20 50 100 200 500 1k 2k 5k 10k 20k 50k 100k
the actual measured value (see the
Hz
specifications).

learn design share
Connect Objects
with Genuino 101
Make any device talk to your mobile
By Clemens Valens (Elektor Labs)
The Internet of Things (IoT) and Connected Objects have been on everybody’s lips
for a couple of years now; industry analysts predict billions of connected devices
in the near future and the market will be worth tens of billions of dollars.
Impressive numbers, big money, wouldn’t it be nice to get a share of that? It
would, so here’s how to jump the bandwagon: connect an object.
When we mention connected devices and

IoT, we usually mean wireless connected
devices sending and receiving data from
the cloud. Wireless in this case can be
anything really, but Wi-Fi and Bluetooth
are today’s most commonly used tech-
nologies. Bluetooth 4.0, Low Energy or
BLE, is very well suited for connecting
(ultra) low-power devices like sensors and
wearables to a more powerful device like
a smartphone or a tablet which in turn
functions as an access point to the cloud.
Intel’s Genuino 101 board (called

Arduino 101 in the USA) (Figure 1), an
Arduino compatible board fully supported
by the popular and free Arduino develop-
ment tools, is equipped with BLE, mak-
ing it a great platform for creating your
own connected objects. Powered by an
Intel Curie module — integrating a 32-bit
Quark microcontroller, 384 KB flash mem-
ory, 80 KB SRAM, BLE, a six-axis gyro-
scope-accelerometer combo, and battery
charging circuitry — the board features
Arduino Shield compatible extension con-
Figure 1. The Intel Arduino/Genuino 101 board has a 6-axis inertial measurement unit and Bluetooth nectors giving access to up to twenty dig-
Low Energy capabilities. ital input/outputs and six analog inputs.

This article sets out to teach connecting ers, we prefer an easy approach, and a Prerequisites
the ‘101’ to a mobile, BLE-capable device good tool for us kind of people is Evoth- • Genuino 101;
running Android 4.4 (KitKat) or higher. ings. The idea behind this tool is sim- • BME280 shield;
Officially BLE support is available since ple: provide a basic App framework to • PC (Windows, Linux or OSX) with at
Android 4.3, but because its reliability which the developer can add his own code least one free USB port, a plain text
leaves a lot to desire, it’s best to stay using JavaScript, HTML and CSS, stan- editor and, optionally, an image edi-
away from it. dard, well-known webpage development tor (I used Windows 7 Family, SP1,
languages. 64-bit);
First we will add an extension board (a • Mobile device, BLE capable, run-
Shield in Arduinospeak) with a BME280 Actually, designing an App in Evothings ning Android 4.4 or higher (I used a
sensor to capture temperature, air pres- amounts to creating a tiny website and Samsung Galaxy J5 running Android
sure and relative humidity [1]. Further- then displaying it in the framework, the 5.1.1);
more, we are going to capture data from Viewer, running on the mobile device. • Internet access for PC and mobile
the 101’s inertial measurement unit Because in Evothings the App is nothing device.
(IMU, the gyro-accelerometer combo, more than a simple website, it can be
that is) and display it in pleasant graphs changed on the fly without lengthy compi- Assemble the hardware
on the mobile device [2], and you will lation steps. All you do is write and mod- Putting together the hardware is quite
be able to read the 101’s analog inputs ify code; as soon as you save it, Evoth- simple as it consists of plugging the
and control its digital outputs from the ings will update its viewer. Instant result BME280 shield on the 101 (Figure 2).
mobile device. on your mobile — isn’t that neat? No If you want to get fancy, you can add your
risk of crashing your mobile either, the own extension board with some resistors,
In order to connect the Genuino 101 to viewer catches your mistakes. Once you LEDs and trimmers to play with the digi-
an Android device over BLE an App is are completely satisfied with your app, tal and analog I/O functionality, but this
required. Since we design the connected you can create a ‘real’ App for it and dis- is optional.
object to our own specifications, it is also tribute it through the Google Play Store.
our job to create this App. Several ways If you do, take care to avoid pin conflicts
lead to Android Apps, but some are easier Sounds good, doesn’t it? So how do we with the BME280 shield. Digital pins 0 to
than others. Being newbie App develop- go about it? 6 and analog inputs 0 to 3 are free.

learn design share
sketch in the 101, wait a few seconds

and open the Arduino IDE Serial Monitor
(Tools ¨ Serial Monitor). After a while you
should see values appear for tempera-
ture, air pressure and relative humidity.
Make sure the shield is working correctly
before continuing.
Download the Genuino 101 Explorer
sketch from [1] and upload it to the 101.
After a few seconds the Serial Monitor
will show some data indicating that the
101 is waiting for a Bluetooth connection
(Figure 3).
Set up Evothings
Evothings Studio is also called Evothings
Workbench — just launch the executable
Figure 2. Building our whatever its name is. Open the New tab
Connected Object did not require and enter “Genuino-101-Explorer” as App
any soldering. folder name. Before clicking Create, make
a note of the Parent folder where the App
will be saved. Close Evothings Workbench.
With a file manager navigate to the
Install the software • PC: Download and install Arduino App folder you just created and delete
Softwarewise there is rather a lot to install IDE (www.arduino.cc, I used version everything inside. From [1] download the
and set up before you can create and pub- 1.6.10) Genuino 101 Explorer Evothings project
lish your App in the Google Play store, • PC: Download and install Evothings and unpack it in this folder. You should
but we will start unpretentiously: Studio (https://evothings.com/, I now have some folders and files includ-
used version 2.0.0) ing index.htm and app.js. Those two are
• Mobile: Download and install Evoth- the important ones. Relaunch Evothings
ings Viewer (Google Play Store, I Workbench and open the MyApps tab.
used App version 1.4.1) You should see an entry labeled: Arduino/
Genuino 101 Explorer.
Genuino 101 On your mobile launch the Evothings
Do not connect the board to the PC Viewer App (Figure 4). It will open and
yet, you must install some drivers first. request a connect key. This key is gener-
Launch the Arduino IDE and open the ated by the Evothings Workbench: click
Boards Manager (Tools ¨ Boards ¨ the Connect tab and then the Get key.
Boards Manager…). In the top-left cor- Enter the key in the viewer and tap Con-
ner choose Arduino Certified, this will give nect. For this to work, both the PC and the
you a shorter list in which you can eas- mobile must be connected to the Internet.
ily find the Intel Curie Boards. Click on When the connection is successful the
it, and then click on the Install button. viewer will tell you what to do next. What
Depending on your Internet connection it means to say is that you have to open
the installation may take a while. When the MyApps tab in the workbench and
done, close the Manager, then go back click on the Run button of the Arduino/
to the Tools ¨ Boards menu and scroll Genuino 101 Explorer app. Do this and
down through the list until you find the observe the Loading icon appear in the
Arduino/Genuino 101. Select it. viewer, followed by the App opening a
Connect the board to the PC. A serial port few seconds later.
should become available; select it in the If the Internet connection on both your
Arduino IDE (Tools à Port). mobile and PC is good and stable, the
If you run into trouble with the installa- connection between the viewer and the
tion of the 101, please consult the online workbench will persist. However, if you
installation guide at https://www.arduino. are seeing bizarre loading errors you may
cc/en/Guide/Arduino101. try to reconnect the two with a new key.
Download the BME280 sketch from [1].
The QR code on the back of the BME280 Try it out
Figure 3. Sketch uploaded successfully and board shield contains the URL (http://bit.ly/2aN- If you had disconnected the Genuino 101
waiting for a Bluetooth connection! NDq7) that will get you there. Load the from the PC, now is the time to recon-

Figure 4. Evothings Workbench on the PC and its Figure 5. A few screenshots from the App showing the different data views.
Viewer on the mobile device.
nect it (or power it from a supply). If it Regarding the Sketch new unique ID for it. You do this by gen-
was connected already, press the Reset Now that you have a working environ- erating a Universal Unique Identifier or
button (not the Master Reset button). ment it is time to investigate the different UUID (with an online UUID generator, for
Give it ten seconds or so to get going, components that make up the system. instance). Then, for each kind of informa-
use the Serial Monitor to be sure that it is Let’s start with the Genuino 101 sketch. tion that must be received or sent, a char-
waiting for a connection, then tap the red This sketch is based on the CurieBLE and acteristic must be added to the service
Connect button in the Evothings viewer CurieIMU examples (Arduino IDE à File as an attribute, each characteristic with
on your mobile. à Examples) to which I have added the its own unique ID. Characteristics can be
BME280 driver. In this article we will con- read-only, write-only or read-write and
If Bluetooth was not yet activated on your centrate on the BLE side of things, we they may generate notifications.
mobile, it will ask you for authorization to take the hardware interfacing for granted. In theory a characteristic can convey up
activate it now. Accept this and observe The basic idea is that BLE uses services to 512 bytes, but the Genuino 101 BLE
the status messages that appear below that are identified by unique IDs. A num- library limits this to 20. Our application
the connect button. If all goes well, the ber of such services are defined by the produces three 4-byte floating point val-
button turns green after a few seconds. Bluetooth SIG (www.bluetooth.com), but ues for the accelerometer (i.e. 12 bytes
Tap the menu button in the upper-right since our combined IMU weather service in total), the same for the gyro and for
corner and select a data view (Figure 5). is a not existing one, we must create a the BME280; the analogue data fits in

learn design share
is allowed too so you can dynamically

rewrite the page), and with the get and
setAttribute methods you can query and
change the element’s attributes, like
changing its color or size.
The HTML code in return can call functions
in app.js when the user clicks or taps a
button or menu item, or toggles a check-
box by declaring a JavaScript function
for the element’s onclick attribute, for
instance onclick=”app.onStartButton()”.
The JavaScript and HTML code can nat-
urally connect to the Internet — this is
shown on the App’s ‘About’ page, where
some basic links allow the user to open
web pages. With a bit more code you can
connect to online data storage services,
Twitter and what not and publish your
data on a www scale.
Turn it into an App

Figure 6. Our own App in the Google Play Store — hey we’re in (IoT) business!
When you think that your App is ready for
transforming into a real Android App you
six 2-byte values (again 12 bytes) and About the App must compile it. Note that it is not only
the digital I/O data fits in two bytes. This On the App side things are quite similar. possible to compile the App for Android,
totals to 50 bytes which could fit in three The App’s bread-and-butter code lives but for IOS too if you prefer (but not on
characteristics, but in order to keep things in the JavaScript file app.js. The user Windows). Compiling is easy enough if
simple by avoiding complicated data mul- interface (UI) is defined by the HTML you have the Apache Cordova toolchain at
tiplexing schemes we simply define a file index.htm and the style sheet ui\ your disposition (https://cordova.apache.
characteristic for each data type. css\evothings-app.css determines just org/), but installing it requires a huge
Once the characteristics defined and how smart the HTML elements will look. amount of software:
initialized, and added as attributes to Like Genuino 101, Evothings features a
the service, the BLE peripheral can be nice BLE library that takes care of the • Node.js (I used v4.4.7);
started. As long as there are no connec- Bluetooth magic, making it easy to use • NPM (I used v2.15.8);
tions nothing much will happen, but once in our App. The unique IDs created for • Git (I used version 2.6.3.windows.1);
connected, we can start putting data in the Genuino 101 sketch must be known • Cordova (I used v6.3.1);
outbound characteristics and read data by the App in order to access the service • Java JDK (I used jdk1.8.0_101_x64);
from inbound characteristics. That is all and its characteristics. Since the charac- • Android SDK (I used v25.1.7 +
there is to it really, the library handles teristics come in asynchronously the App API 24 + API 23 + API 19).
the difficult stuff for us. activates reception notifications for them
The data from the IMU is streamed as fast to process the data as it arrives. Luckily, detailed instructions on how to
as possible to the mobile; the other data Except for the analog and digital I/O that do this can be found online at https://
is sent every 100 milliseconds. is displayed immediately, the rest of the evothings.com/doc/build/build-overview.
Reading data from a characteristic can data is stored first before being shown to html#Install so I won’t waste any space
be done in two ways, either by polling the user. Interval timers are set up for the on it here.
it using the written() function or event graphs and the BME280 data. The graphs Supposing you have Cordova working,
driven by registering a listener. Both are updated rather slowly, so rapid board here’s the recipe for building an App from
ways are illustrated for the digital I/O shaking and noise are suppressed, giving your Evothings project.
pins characteristic in the Genuino 101 them a really smooth appearance. This First you need to create a Cordova
Explorer sketch. smoothness is also due to the library used project:
to create the graphs: Smoothie Charts
The service can be given a (local) name, (http://smoothiecharts.org/). Create a folder to work in and open a
“GEN101” in our case, but care must be Communication between the UI in HTML command line interface (CLI) here.
taken to keep it short. The reason is that and the App in JavaScript is bidirectional. In this folder enter the command “cordova
it is put in the space that is left in the From app.js up to index.htm we make create [my_project] [com.elektor.
advertising buffer and that is not a lot. extensive use of the getElementById() labs.my_app] [my_app_name]” where you
If the local name does not fit, the library function. For this to work the HTML ele- replace everything inside square brack-
will make up its own name, breaking the ment to modify must have an id. The ets by names of your own devising. Note
automatic detection capability of the App innerHTML property lets you rewrite that the dotted name is supposed to be
in the process. the text inside the element (HTML code a reverse URL (you can make one up).

In the CLI, cd into the project folder you somewhere for future reference. Then run the same as in the viewer (but much big-
just created: “cd [my_project]” (again, one line command, be careful with ger). If not, try to find the reason and
Delete everything inside the subfolder the double dashes): fix it.
‘www’, but do not delete this subfolder. A final word on BLE. Handling a BLE con-
Copy your Evothings project into the cordova build android --release -- nection properly is a bit tricky as it is
folder ‘www’, i.e. all the files and subfold- --keystore=”[Path]\[Name].keystore” almost impossible to completely exit an
ers that accompany app.js and index.htm. --storePassword=[Password] App. Android keeps it running in the back-
In the CLI, enter the command “cordova --alias=[Alias] ground where it may remain (partly) con-
plugin add cordova-plugin-ble” to add nected. To get out of this kind of loopholes
the BLE library to your project. Repeat Where [Name] and [Alias] are the same kill the App in the mobile’s application
this for any other plugin that you may as before, [Path] is the path to the new manager and, just to be sure, restart the
require. keystore file, and [Password] the one connected object.
In the CLI, enter the command “cordova you were asked by keytool to create.
platform add android ”. For IOS use A tiny pop-up window will ask you for Conclusion
“cordova platform add ios”. You can your password once more, enter it and This article showed how to connect a
add more than one platform. click OK. The result is an .APK file named Genuino 101 board to a Bluetooth Low
“android-release”. Energy (BLE) capable (mobile) device like
Now you’re all set to build the project If you haven’t one already open a Google a smartphone or tablet by using open
from the CLI with these commands: Play Store publisher account (one-time source and free tools only. The board was
$25 fee), upload the APK file, go through extended with a BME280 shield to capture
cordova build android the forms, publish, and you are in busi- temperature, air pressure and relative
cordova build ios (not on Windows) ness (Figure 6)! humidity data. After defining a custom
BLE service and some custom characteris-
Normally the build should succeed with- Pitfalls tics any kind of data could be exchanged
out warnings and errors and the result Your Android App may (will?) not work between the two connected devices. This
can be found in: platforms/android/build/ as expected from what you saw in the way typical IoT data like accelerometer,
outputs/apk. To test it on your mobile viewer. The most likely reason is that gyro and weather information was trans-
device you have to copy the APK file on you did not fix all the problems that were ferred to the mobile device. A user inter-
it and install it through the mobile’s file silently handled by the Evothings viewer. face (UI) running on the mobile device as
manager. This requires your device to To find out what’s wrong, use the JavaS- a native App allowed displaying the data
allow Unknown Sources (in the security cript console in the Evothings workbench, and user interaction, and provided the
settings). on the Tools tab, and run your App in the Internet capabilities needed for transfer-
To publish your App in the Google Play viewer. Everything that’s Unreferenced ring data to and from the cloud. Finally,
Store you must recompile it in Release and Unhandled etc. must be fixed. we published the App in the Google Play
mode and sign it. For this it is necessary Keep in mind that timeouts may be too Store for our future users to enjoy.
to first create a key (one line command): short for real-life situations. Also keep in (160109-2)
mind that things work differently when
keytool -genkey -v -keystore the mobile is connected to the object. The
Web Links
[Name].keystore -alias [Alias] Bluetooth stack may be blocking some-
thing. Maybe you should close the con- [1] www.elektormagazine.com/labs/
-keyalg RSA -keysize 2048 -validity
nection first before you can do what you bme280-evaluation-board-160109-2
10000
want to do? [2] www.hackster.io/gov/
Where [Name] and [Alias] are for you What if your App doesn’t look as imu-to-you-ae53e1
to choose. Fill in the requested infor- expected? Open the file index.htm in a
mation and make sure to write it down browser on your computer, it should look
BME280 Evaluation Board

The Bosch Sensortec BME280 sensor is a high- operate on either an SPI bus or an I²C bus.
performance, high-precision device capable Elektor’s BME280 evaluation board (# 160109-2, an Arduino
of capturing temperature, air pressure compatible shield) permits the use of the sensor in SPI
and relative humidity information. or I²C mode, simply by pulling or fitting a jumper. A
Although primarily targeted second jumper permits selecting between two
at mobile devices for I²C addresses. The selected mode and I²C
face/hand proximity address can be detected by the host, as
detection and indoor well as set by suitable programming.
navigation, it also makes A user LED is available for visual
a very capable weather feedback.
station. The device can

learn design share
BrainBox Arduino
A ‘tough’ Arduino with screw terminals
By Bart Huyskens (Belgium) This rugged version of the Arduino Leonardo

has been designed to be used on a stand-
alone basis in various projects and in the
educational sector. Thanks
to its sturdy screw
terminals, a number of
different power supply
options, an on-board
buzzer and a motor-driver
IC that connects directly to
motors, there is usually no need
for breadboards, extra ICs or shields in
most applications.
Although the BrainBox ardo, the BBA also has four LEDs, These extra features make it possible to
Arduino was originally designed of which two are user programma- use this board in a range of projects,
for use in the educational sector, the ble. The differences are mainly due to without the need for spaghetti circuits
board is so versatile that it lends itself the extra features of the BBA, compared on breadboards or extra Arduino shields.
for use in many other electronics pro- to the standard Arduino boards:
jects. Anybody can program this BrainBox A buzzer with a wide frequency response. Circuit diagram
Arduino (shortened to BBA for the rest This is ideal for trying out all those ring The complete circuit diagram for the BBA
of this article) in next to no time, using tones you’ve created. is shown in Figure 1. At first sight this
their preferred programming language, appears rather complex. In view of the
since the example programs have all been A double H-bridge with four power out- educational background of the BrainBox
developed for five different development puts, which can source up to 600 mA per Arduino, it would be better if beginners
environments. pin. You can use this to directly drive 4 looked at the connection diagram in Fig-
The BBA can be obtained fully assembled, DC motors in half-bridge mode, or two DC ure 3, which clearly shows all the con-
complete with a programmed bootloader, motors in full-bridge mode. These power nections on the board with their name,
from the Elektor Store, which means you outputs can also be used to drive power as well as the functions of the main com-
don’t need any soldering experience to LEDs, to generate heat using power resis- ponents on the board.
get started with this board. Furthermore, tors or to drive stepper motors, to give a
there is also an educational kit available few examples. The user can select one of For those of you who would like to know
that contains all the hardware, software the following power sources for the four more about the ‘real’ circuit, we’ll take
and courseware to build the robot buggy power outputs, using a set of jumpers: a look at the components shown in the
shown in the inset, and to control it with 5 V, mains adapter, battery, or an exter- schematic. At the heart of the BBA is
a dedicated app. nal supply. The voltage to the power out- an Atmega32U4. This low-power 8-bit
puts must be between 4.5 V and 36 V. RISC-based microcontroller has 32 KB
Extras The power source for the processor is of flash memory, a 2.5 KB SRAM and a
The BBA is built around a powerful Leon- selected via a jumper from one of the 1 KB EEPROM. It has a built-in USB 2.0
ardo processor, the ATmega32U4, which following: USB, mains adapter or battery. full-speed interface, a 12-channel 10-bit
runs at 16 MHz. This processor can be Two connectors for servo motors. A/D converter and a JTAG interface for
programmed directly via the USB port. Extra connectors for an HC06 Bluetooth on-chip debugging. The maximum clock
Just as with the standard Arduino Leon- module, RS232 and I2C. frequency is 16 MHz, which is the speed of

6 INPUTS ANALOG & DIGITAL

MF-MSMF050-2
VUSB CONN7 CONN8 CONN9 CONN6 CONN5 CONN2
R2
C1 T1 tº C2
10k
3 2 1 3 2 1 3 2 1 3 2 1 3 2 1 3 2 1
22u 100n
+5V
CONN4
A5
A4
A3
A2
A1
A0
1
VBUS
2 R3 D–
D– 22R
3 R4 D+
HWB
AREF
UCAP
VUSB
A5
A4
A3
A2
A1
A0
RESET
D+ 22R
4 V1 V2
R10 VIN +5V
GND C13 C14
10k
1 2 3 4 5 6 1 2 3 4 6 7 8
USB-B 100n 100n
V V
CONN17 CONN18 5
V1, V2 = CG0603MLC-05E
+5V
VIN
LD1117S50
CONN21 JP1 +3V3
1 5
D3 IC2 IC3
3 3 2 R9
2 LP2985-33
1 3 4
SS16 ON/OFF BP
1k
C10 +5V
7 - 16V DC C8 C16 1 C6 C9 5V
SOURCE SELECT LED4 2
1u L1
100n 22u 100n 22u
CONN10 MH2029-300Y
D2 +5V
1 C3 C4 C12
7 - 16V DC
2 SS16 100n 100n 1u
GND
Vbatt 3
2 14 24 34 44
(6V2 max !)
UVCC VCC AVCC VCC AVCC
JP2 +5V D7 1
PE6(INT6/AIN0)
42 AREF
AREF
VPOWER D– 3 41 A5
D– PF0(ADC0)
SOURCE SELECT D+ 4 40 A4
D+ IC1 PF1(ADC1)
R5 UCAP 6 39 A3
D1 UCAP PF4(ADC4/TCK)
ARDUINO SHIELDS
VUSB 7 38 A2
10k
VBUS PF5(ADC5/TMS)
+5V RXLED 8
PB0(SS/PCINT0) PF6(ADC6/TDO)
37 A1
C7 C15 SS16
RESET SCK 9 36 A0
PB1(PCINT1/SCLK) PF7(ADC7/TDI)
100u MOSI 10 33 HWB
22u SW4 PB2(PDI/PCINT2/MOSI) PE2(HWB)
MISO 11 32 D13
P 8 16 PB3(PDO/PCINT3/MISO) PC7(ICP3/CLK0/OC4)
D11 12 31 D5
POWER VCC1 RESET PB7(PCINT7/OC0A/OC1C/RTS) PC6(PC3A/OC4A)
9 RESET 13 30 D10
3,4EN RESET PB6(PCINT6/OC1B/OC4B/ADC13)
D7 D10 15 14 D3_SCL 18 29 D9
IN4 OUT4 PD0(OC0B/SCL/INT0) PB5(PCINT5/OC1A/OC4B/ADC12)
D2_SDA 19 28 D8
U1 PD1(SDA/INT1) PB4(PCINT4/ADC11)
D9 10 11 D0_RX 20 27 D6
IN3 OUT3 PD2(RXD1/INT2) PD7(T0/OC4D/ADC10)
D1_TX 21 26 D12
PD3(TXD1/INT3) PD6(T1/OC4D/ADC9)
1 TXLED 22 25 D4
1,2EN PD5(XCK1/CTS) PD4(ICP1/ADC8)
BUZZER D6 7
IN2 OUT2
6 5
UGND ATMEGA32U4
L293D GND GND XTAL1 XTAL2 GND GND
D5 2 3
IN1 OUT1 15 23 17 16 35 43
+5V XTAL1
GND
4 5 12 13
PL3 C11 C5
1
P 16MHz
2 22p 22p
3 D0_RX
ISP +5V
4 D1_TX MISO CONN19 CONN20
SCK
HC06 BT RESET
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 10
D2_SDA
D2_SDA
D3_SCL
D3_SCL
D0_RX
D1_TX
MOSI
AREF
D10
D11
D12
D13
D4
D5
D6
D7
D8
D9
D10 OUT
+5V +3V3
RXLED
TXLED
D13
D11
D12
D8
D4
D3_SCL
D2_SDA
D1_TX
D0_RX
D5 OUT
D6 OUT
D9 OUT
SERVO
R6 R7 R8 1 2 3 4
+5V +5V
1k
1k
1k
RXLED TXLED LED3

CONN13 CONN12 CONN3 CONN11
POWER OUTPUTS NORMAL OUTPUTS COMMS POWER
4x 600mA 2x 20mA (RS232/I2C)
160001 - 11
Figure 1. The circuit diagram of the Arduino, with an Atmega32U4 at its heart.

learn design share
Young people program their robots and design electronic circuits

for the Brainbox Arduino
the crystal (XTAL1) used with this board. A buzzer is connected to pin 1 and a Pin headers JUMPER1 and JUMPER2 are
The USB connector (CONN4) is connected reset button is connected to pin 13 of used to select one of the different power
directly to the controller via protection the controller. supply options. JUMPER1 is used to select
resistors R3 and R4. The supply to this the power source for the board: USB con-
connector is fed via a resettable fuse (T1), There is an extensive power supply sec- nector, mains adapter or battery pack.
which is connected to JUMPER1 for the tion in order to cater for a wide range JUMPER2 is used to select the power
selection of the power supply source for of supply options. A mains adapter with source for the motors that are connected
the board. an output voltage between 7 V and 16 V to the motor driver IC (U1). Here you have
The Motor driver (IC U1), an L293D, con- can be connected directly to CONN21. a choice of the regulated 5 V from IC2,
tains a double H-bridge that drives the Alternatively, the 3-way screw terminal the mains adapter, or the battery pack.
four power outputs on CONN13, with a (CONN10) can be used to connect an
maximum current of 600 mA per pin. The external power supply (also 7–16 V) or The rest consists of various connec-
electronics inside the IC is powered by a battery pack with a maximum voltage tors and screw terminals. CONN17 to
voltage regulator IC2; the power drivers of 6.2 V. The 7–16 V DC voltage is fed CONN20 are available for the connec-
in the IC have a separate power connec- via reverse-polarity protection diode D3 tion of Arduino shields. There are also
tion, where the source can be selected to a 5 V voltage regulator (IC2), which six groups of 3-way screw terminals for
via JUMPER2. provides power to the microcontroller and analog and digital inputs (CONN2, CONN5
There are also four LEDs, of which two motor driver IC (U1). This is followed to CONN9), each with its own 5-V power
are used as indicators for the serial com- by a 3.3-V voltage regulator (IC3), with supply. These are all grouped on one side
munications via CONN3 (RXLED is also its output made available on the shield of the board. On the other side are the
programmable), a power indicator LED connectors and on the Power screw ter- screw terminals for the motors (CONN13),
and a freely programmable LED (LED3). minals (CONN11). two digital outputs (CONN12), serial and
Software options
Apart from the Arduino IDE, there are several other program- to the hardware. Even though this is a graphical language which
ming environments that can be used to easily program the BBA. is amply provided with libraries that make programming easier,
An important feature of this BBA project is that in many instanc- it remains the case that Flowcode is much closer to real ‘em-
es the example programs have been written for all of the five bedded programming’ than the Arduino IDE, for example. Flow-
programming environments listed below. This way you can pro- code is therefore the ideal first step on the road to Embedded C.
gram the BBA using your preferred programming environment.
Atmel Studio 7 with GCC compiler
Arduino IDE with libraries Atmel Studio is the professional development environment for
The original Arduino IDE is a simple C compiler with a large Atmel controllers, and GCC is probably the most-used free C
number of libraries. On the one hand, these libraries and the compiler for Atmel AVR controllers. You can quickly program
unique pin numbering make it possible to write fairly complex the BBA from within this environment. If you want to, you can
programs without needing much knowledge about the micro- still use the extensive Arduino libraries in this environment, to
controller. On the other hand, this means that you hardly learn make the programming of certain functions easier.
anything about the microcontroller itself.
Snap 4 Arduino
Arduino IDE without libraries With S4A it becomes possible to program the BBA with the im-
What many people don’t know is that the Arduino IDE can also mensely popular ‘Scratch puzzle blocks’. S4A uses a fairly stable
be used to program the Arduino using the real register names connection to the BBA, which requires that the ‘Firmata’ pro-
and pin numbers. Users will first have to familiarize themselves gram is loaded first. From then on, all instructions on the screen
with the microcontroller before they can start writing programs are carried out live on the hardware of the connected BBA – a
for it. It’s also possible to use a combination of both systems, bit like an emulator, really.
with and without libraries.
The beta version can translate most basic programs into real
Flowcode (7) for AVR code, which could be loaded at a later stage using the Arduino
From an educational point of view, Flowcode is without a doubt IDE, making it possible to run some programs independently.
the best way to get started with embedded programming. The It looks like a promising start for S4A, which is an ideal way
flowcharts make it very easy to see the program structure, and to impart some of our electronics enthusiasm to the younger
the clever simulator lets you try out the code before you send it generation.

I2C communications (CONN3), and finally

connector CONN11, where all the supply Connections
voltages are made available. Then there The colored pin numbers in the diagram shown here correspond to the names
is a 6-way pin header (SERVO) for driving given to the pins in the Arduino IDE and in Snap4Arduino. The gray pin numbers
two servo motors, a 6-way ISP-connector are the real pin names as shown in the datasheet, and are used in AVRStudio and
(ISP) for directly programming the micro- Flowcode.
controller and a 4-way connector (PL3),
which accepts an HC06 Bluetooth module.
We haven’t shown a parts list, since the
BBA is supplied ready-made and pre-pro-
grammed because of its educational
nature.
Overview
The layout of the Brainbox Arduino has
deliberately followed the IPO principle:
Input – Processing – Output. At the
top are all the connections for sensors,
the processing is done by the powerful
ATmega32U4 processor, and at the bot-
tom are all the connections for actuators.
The sensors and actuators are connected
directly via sturdy 5 mm screw termi-
nals. The sensor inputs can all be used
as either analog or digital inputs. Each
input has its own GND and 5 V connec-
tions. There is also a set of typical Arduino
connectors for the connection of Arduino
shields.
The BBA has a wide range of power sup- Figure 2. Connection details for the board, with brief descriptions.
ply options, which makes it suitable for
use in many applications.
On the right-hand side of the board are
the programming options. The BBA is (Dutch) teaching material and further Web Link
supplied with a pre-programmed Arduino information is available from [1].
[1] http://e2cre8.be/ (in Dutch only)
Leonardo bootloader as standard. You can (160001)
therefore immediately program the BBA
via the USB port. If you decide to pro-
gram the ATmega32U4 processor with- The Author
out a bootloader, you can do this via the
Bart Huyskens is an electronics and ICT teacher in Belgium. In addition, he
6-way ISP header.
organizes workshops and develops hardware, software and teaching material to
At the bottom-left are two connectors for
encourage inquisitive and creative STEM education.
servo motors and 4 power outputs with
PWM capabilities. These are driven by a
double H-bridge, which can supply up to
600 mA per pin.
At the bottom-right are the two most pop- Complete BrainBox Arduino kit
ular communication ports, I2C and RS232.
The BrainBox Arduino kit retailed by
These can be used to connect a huge
Elektor comes complete with all the
range of sensors, actuators and LCDs to
required sensors, actuators, wires,
the BBA. The 4-way connector at the bot-
mechanical parts, teaching material
tom-right corner is used to directly con-
and example programs to build the
nect a popular HC06 Bluetooth module.
robot buggy shown here, and control
Example programs for both the BBA and
it via a dedicated app.
the APPINVENTOR help you on your way
to use your own apps on a smartphone to
But don’t stop there, give your
communicate with your BBA. You should
creativity a free reign and use the
give it a try — you’ll be amazed how
BBA to create your own projects.
quickly you can get your app up and
running.

learn design share
LEGO Control Board ®
for the Raspberry Pi
This Raspberry Pi HAT puts at your disposal 4 motor-control outputs for LEGO EV3 motors and 16 buffered
I/O connections that can be used in combination with a powerful Raspberry Pi. Programming is possible
in languages such as C and Python. As an application example we show how this board and a few LEGO
parts can be used to build a ‘useless box’.
The board described here combines mechanical construction and the flexi- for the Raspberry Pi, you therefore only
the flexibility of LEGO with regards to bility of the Raspberry Pi with regards to need to connect a 9-V (line) power sup-
software, Internet communications and ply to this board (recommended rating
a wide range of potential sensors. 5 A; positive to center pin). From this the
When you attach this board to a Rasp- motors, the board itself and the Rasp-
berry Pi 2 or 3, you will have at your berry Pi are powered. Mobile applications
disposal 4 motor-control outputs for EV3 can use batteries, because the board uses
motors (or compatible types), which you an efficient buck converter.
can use to animate your LEGO models.
There are also 16 buffered digital inputs Since the Raspberry Pi header is con-
and outputs present, to which you can tinued through the LEGO control board,
connect sensors, LEDs, etc. You cannot, you can also connect SPI and I2C devices
however, connect LEGO sensors (in that and sensors. You can therefore combine
case you would be better served using all the projects and sensors from the
a standard EV3 controller), but you can book “Raspberry Pi – Explore the RPi in
Figure 1. The connections to the LEGO control board. connect all sorts of other sensors such 45 Electronic Projects” with your LEGO
1) 4 LEGO motor connections; 2) 16 I/O connections as infrared, ultrasonic, tilt and switches. models!
with screw terminals; 3) power supply socket, 9 V; You are limited only by your imagination. Through the Raspberry Pi itself you can
4) feedthrough of the Raspberry Pi I/O pins.
The board also contains a power supply communicate via the Internet, with your
Lego is a registered trademark of the LEGO Group

With 4 motor-control outputs and

16 digital I/Ochannels
By Bert van Dam (Netherlands)
choice of a wired or Wi-Fi connection. If nicate with your model when it is away models where you combine LEGO motors
you connect a USB/4G dongle to your from your home. and various types of sensors with the
Raspberry Pi then you can even commu- With this board you can therefore make power of the Raspberry Pi!
Table 1. Strong and weak points of the LEGO EV3 and Raspberry Pi.
LEGO — strong LEGO — weak
• The mechanical construction, which allows you to do anything • The number of connections for sensors. A reasonably serious
you like. robot would need 4 and that is only the start.
• The motors, although for serious applications you would need • Video processing and all the other things for which LEGO have
servos and hydraulics. not come up with a sensor.
Raspberry Pi — strong Raspberry Pi — weak
• Openness of the system; in combination was the easy-to- • Connection pins are mechanically weak and not electrically
learn language Python much is possible. For the enthusiast C protected.
is an even more powerful option. • Current consumption is on the high side.
• Many connectors and via I/O-expanders offers the possibility
of ‘unlimited’ expansion.

learn design share
+5V
K10
1
+5V +5V +9V
2
GND +5V
+5V
C3 C4 C5 K2 M1
C10 6
R6 R5
100n 100n 10u 5
33k 18k
K6 100n 63V 4
1 20 19 24 13 14 3
IO16 18
2
STBY
VM1
VM2
VM3
VCC
2 VCC
IO15 6 15 11 1
17 RST PWMB BO1
3 GP7 12
IO14 16 +5V BO1 NXT-Socket
GP6 8 16 7
4 15 IC3 INT BIN2 BO2
IO13 GP5 2 17 IC2 8
14 SDA
K11 BIN1 BO2 K3 M2
GP4 1 6
13 SCL R8 R7
GP3 1 2 23 5 5
1 12 MCP23008 PWMA AO2 33k 18k
IO12 GP2 5 3 4 6 4
11 A0 TB6612FNG AO2
2 GP1 4 5 6 22 1 3
IO11 10 A1 AIN2 AO1
GP0 3 7 8 21 2 2
3 A2 AIN1 AO1
IO10 9 10 1
GND
PGND1
PGND1
PGND2
PGND2
4 11 12
GND
IO9 9 NXT-Socket
13 14
18 3 4 9 10
K7 15 16
17 18
+5V
19 20
C9 21 22
23 24
+5V +9V
R18
R19
K8 100n
R4
R3
R2
R1
25 26
IO8
1 27 28 +5V
100k
100k
100k
100k
100k
100k
18
29 30
2 VCC C6 C7 C8 K4 M3
IO7 6 31 32 6
17 RST R14 R13
3 GP7 33 34 5
IO6 16 100n 100n 10u 33k 18k
GP6 8 35 36 63V 4
4 15 IC4 INT
IO5 GP5 2 37 38 20 19 24 13 14 3
14 SDA
GP4 1 39 40 2
STBY
VM1
VM2
VM3
VCC
13
GP3
SCL +5V
15 11 1
1 12 MCP23008 PWMB BO1
IO4 GP2
11 A0
5 Raspberry Pi BO1
12
NXT-Socket
2 GP1 4 16 7
IO3 10 A1 BIN2 BO2
GP0 3 17 IC5 8
3 A2 BIN1 BO2 K5 M4
IO2 6
GND R16 R15
4 23 5 5
IO1 9 PWMA AO2 33k 18k
6 4
TB6612FNG AO2
K9 22 1 3
AIN2 AO1
21 2 2
+9V AIN1 AO1
4 1
K1 +5V
PGND1
PGND1
PGND2
PGND2
T1
R10
R12
R11
FB
R9
GND
IRF9540NS +5V NXT-Socket

1 IC1 2
L1
VIN
100k
100k
100k
100k
VOUT 18 3 4 9 10
LM2596S 33uH LED1 +9V +5V
5.0V 2A1
GNDPAD ON
ON/OFF GND R17
C1 D1 C2
5 6 3 C14 C15 C11 C12 C13
330R
1000u 220u
25V MBRS540 16V 100n 100n 100n 100n 100n
159010 - 11
GND
Figure 2. The circuit of the control board comprises 2 motor driver ICs, 2 I/O expanders and a step-down voltage regulator.
+9V 4
+9V 4
+5V +5V
FB FB
L1 +5V L1 +5V
1 IC1 2 1 IC1 2
VIN VOUT VIN VOUT
LM2596S 33uH LED1 LM2596S 33uH LED1
5.0V 2A1 5.0V 2A1
GNDPAD ON GNDPAD ON
ON/OFF GND R17 ON/OFF GND R17
C1 D1 C2 C1 D1 C2
5 6 3 5 6 3
330R
330R
1000u 220u 1000u 220u

25V MBRS540 16V 25V MBRS540 16V
a GND b GND
Figure 3. Buck converter: a) cranking up the flywheel b) coasting of the flywheel.

Design considerations flywheel. The LM2596 begins by ‘turning board uses addresses 0x20 and 0x21.
The LEGO control board is certainly not up’ the voltage to ‘full’. This gets the You can, for example, add further I/O-ex-
a copy of the EV3 controller, but it does flywheel started. This does not happen tenders, D/A converters for generating
combine the strong points of LEGO with immediately, because the capacitor first analog signals, or A/D converters to sense
those of the Raspberry Pi. Table 1 lists needs to be charged via the inductor. analog signals.
the strong and weak points of each. The inductor itself builds up a magnetic
The conclusion is that this board needed field. The diode does not conduct at this Motors
to have the following characteristics (and time since the current returns via the The two TB6621FNG ICs (IC2 and IC5)
it therefore was given them): LM2596 itself. each contain two H-bridges to allow the
control of four motors in total. The cur-
• 4 LEGO EV3 motor control connec- When the voltage at the output threatens rent consumption of the LEGO motors is
tions (for LEGO 45502, 45503 or to become too high, the LM2596 switches approximately:
compatible), with all 4 of provided its output off. The magnetic field in the EV3 small EV3 large
with tacho feedback signals and 2 inductor will now collapse, which causes
free running 85 mA 80 mA
provided with PWM outputs; the voltage across it to reverse and the
slowly
• 16 buffered I/O pins (expandable diode will now conduct. This creates a 500 mA 1000 mA
forward
thanks to the I2C connection); current that circulates via the load and
• energy efficient power supply for the the diode (the path via the LM2596 is no stalled 780 mA 1800 mA
Raspberry Pi (using buck-converter); longer possible because this is switched
• free Python 2 software library; ‘off’). The capacitor also discharges. An EV3 motor can be stalled for only a
• feed-through of the Raspberry Pi When the voltage at the output threatens short amount of time, because the pro-
header (you therefore can add fur- to become too low, the LM2596 switches tection built into the motor will switch it
ther connections yourself, such as on again. off. It is nevertheless always a good idea
protected I/O or analog I/O). to prevent stalling the motor, because
The LM2596 therefore only switches on should the protection not work for some
Design result and off. The effect of this is that the effi- reason the motor could burn out.
The complete schematic for the LEGO ciency is very high and the LM2596 there-
control board is shown in Figure 1. The fore barely heats up. Because of this con- The TB6621FNG can typically handle
most relevant parts in this circuit are the tinuous on and off switching, the value 1.2 A (per channel, so per motor) with a
two motor-driver ICs type TB6612FNG of C1 is also important and for both C1 peak of 3 A; this is therefore more than
from Toshiba (which each can drive two and C2 low-ESR types need to be used. sufficient.
motors with a maximum output current
of 1.2 A) and two 8-bit I/O-expanders I/O If you intend to make a software library
type MCP23008 from Microchip with a The MCP23008 I/O extenders IC3 and IC4 for the motor controllers, then take into
high-speed I2C interface and SPI inter- are controlled via I2C and ensure that the account that the PWM pins have to be
face. We now take a look at a few inter- I/O connections of the Raspberry Pi do high in order for the TB6621FNG to oper-
esting details of the circuit. not need to be used directly. This design ate. When you apply a PWM signal then
has several advantages: it is always correct, independent of the
Power supply direction of rotation.
In order to enable everything to be pow- • The I/O connections operate at 5 V
ered a single source, the Raspberry Pi • The I/O pins have a greater current Circuit board
is powered from the control board. The capability than the Raspberry Pi pins In Figure 4 you can see the circuit board
board itself is powered from a 9-V DC • Should something nevertheless go that was designed for this LEGO control
power source. This power source can wrong, the damage will generally be board. The dimensions and connector
either be batteries or a substantial line limited to the I/O extender chip positions are obviously selected such that
power supply (recommend rating 5 A). the board will fit trouble-free on a Rasp-
To go from 9 to 5 V (the voltage required The maximum current that the MCP23008 berry Pi model 2 or 3 (using a 40-way
by the Raspberry Pi) an energy-efficient can supply is 25 mA, with a maximum per I/O connector). Since most of the com-
method needs to be used to prevent IC of 125 mA. Note that one IC can dissi- ponents are in SMD packages, you can
the batteries from draining too quickly. pate a maximum power of no more than also purchase a completely built version
The choice therefore fell on a buck con- 700 mW. This applies to each MCP23008, of the board from Elektor. Of course, it is
verter based on the step-down regula- so if you need a lot of current it would also possible to do everything yourself,
tor LM2596S-5.0V, a so-called Simple be best to spread the loads across both the circuit board layout is available from
Switcher Power Converter from Texas MCP23008s and not connect everything [1]. Note when assembling the board that
Instruments. This operates at a frequency to a single IC. the 40-way connector K11 is fitted on the
of 150 kHz and can supply 3 A. Because the Raspberry Pi headers are bottom side of the circuit board and is
The buck converter functions as follows continued through to the top of the con- soldered from the top side. The four NXT
(see Figure 3). The loop formed by the trol board you can add additional I2C com- sockets for the motors are quire rare, you
diode (D1), the inductor (L1) and the ponents yourself, provided that they have will have to search on the Internet for a
capacitor (C2) act like a kind of electronic different addresses. The LEGO control suitable supplier.

learn design share
workings; making the box we leave up

Component List to you. You can make this easily using
Resistors T1 = IRF9540NSPBF, P-channel MOSFET, cardboard or wood. With the download
R1,R2,R3,R4,R9,R10,R11,R12,R18,R19 = D2-PAK
for this article you will find a short video
100kΩ 1%, 0.1W, case 0603 IC1 = LM2596S-5.0, DC/DC buck converter,
R5,R7,R13,R15 = 18kΩ 1%, 0.1W, case 0603 TO-263-5 of this project. Both the pictures accom-
R6,R8,R14,R16 = 33kΩ 1%, 0.1W, case 0603 IC2,IC5 = TB6612, motor driver, 24-SSOP panying this article as well as the video
R17 = 330Ω, 0.1W, case 0603 (Digikey # TB6612FNGC8ELCT-ND)
show only the mechanical part, without
IC3,IC4 = MCP23008-E/SO, I/O expander,
Capacitors SOIC-18 the box. This way you can get a clear
C1 = 1000µF, 25V, aluminum, d 12.5mm, h view of its operation.
13.5mm, ELPP-CP-125-135 Miscellaneous
C2 = 220µF, 16V, aluminum, d 6mm, h K1 = DC adapter connector, 12V 3A, 1.95mm
center pin What do you need?
7.7mm, ELPP-CP-063-066
C3,C4,C6,C7,C9,C10,C11,C12,C13,C14,C15 = K2,K3,K4,K5 = NXT socket • The SD card from the book ‘Rasp-
100nF 10%, 16V, X7R, case 0603 K6,K7,K8,K9 = 4-way PCB screw terminal berry Pi – Explore the RPi in 45 Elec-
C5,C8 = 10µF, 63V, case D block, 3.81mm pitch
K10 = 2-way PCB screw terminal block, tronic Projects’;
Inductors 3.81mm pitch • The free download that goes with this
L1 = 33µH, 2.1A (e.g. Coilcraft DO3316) K11 = 40-pin (2x20) GPIO stacking header article (contains the library and the
for RPi, extra tall
PCB # 159010-1 source code for the demo project);
Semiconductors
• A LEGO motor (type 45502 ‘large’
D1 = MBRS540, 40V 5A, ELPP-DO-214AB
LED1 = LED, red, case 1206 motor) with connecting cable and a
few LEGO parts (you can build the
demo project using the LEGO Mind-
K10 K6 K7 K8 K9
storms EV3 box).
K5 • A Raspberry Pi model 2 or 3;
R16
+5V GND
• The LEGO control board;
C13
C10
• Two resistors: 1 kΩ and 10 kΩ,

C9
(C) Elektor R15

159010-1 C12
IC4
0.25W;
IC3
V2.1
K4
K1 GND 5V • An (easily operated) small in-line
IC1 C8
R14
C7
IC5
R13
switch (see Figure 4).
C1 D1 C6
C5 C15 K3
The hardware
We use a ‘large’ LEGO motor type 45502
C4
IC2
R8
C2 R7 which can be found in, for example, the
C3
LEGO Mindstorms EV3 box and the parts
R17 LED1 C14
T1
L1 K2 that are shown in Figure 5. In the down-
R6
R12
R10
R19
R18
R11
R9
R1
R3
R4
R2
R5 load you will find a short video where you

K11 1
can see how you assemble these parts
C11
40
into the motorized finger, which is shown
on the right in the figure. You now attach
Figure 4. The layout of the circuit board. The 40-way connector K11 is mounted on the bottom of the in-line switch on top of the assembly
the circuit board. in such a way that the black beam can
push the switch to its off position.
Demo project: The Useless Box ate the switch, a ‘finger’ appears from This switch is connected with two resistors
The concept of this ‘Useless Box’ is a box the box which turns the switch off again. (see Figure 6) to I/O pin 16. This could
with a switch on top. When you oper- For this project we only build the inner also be accomplished with only one resis-
Figure 5. Parts and the assembled model of the motorized finger.

tor (the 10 kΩ one), but the additional not noticeably slow down the program. As 7. Repeat step 6 as many times as you
resistor serves as protection. If you now soon as the switch is off again, the pro- like and then stop the program using
inadvertently make the input into an out- gram will turn on the brake in the motor. Ctrl-C.
put and set its level low then nothing bad The next step (Listing 1b) is to return
will happen. Without the additional 1-kΩ the LEGO finger at a slower speed This is only a simple application example.
resistor the I/O port could be damaged. (PWM 30, 30% of the maximum speed) The LEGO control board can of course be
to its rest position. For this case there is used to realize much more complicated
The software no switch at the end of the movement, applications with its four motor control
The software is surprisingly simple. (Note: so the program now uses the tacho. The connections and 16 I/O pins.
the library expects Python-2 programs!) program counts the tacho changes and (150597)
You can find more information about the once there are 100 (28% of a circle) the
library and the instructions in the manual, motor stops with the float instruction. The
Web Links
which is a free download accompanying LEGO finger is now about one centimeter
this project [1]. The very first thing we do above the surface. [1] www.elektor.com/150597
is load the rpirobot library and execute it
with the instruction ‘execfile’. The motor In the except-construct (Listing 1c) any
outputs of the LEGO control board are errors or a Ctrl-C instruction from the
now automatically configured correctly user are caught. The program gives a
and the PWM threads are running. We brief abort instruction to the Raspberry Pi
then make pin 16 of the I/O on the con- library and then terminates itself as well. Listing 1a.
trol board an input, because this is where Without this construct it is obviously still
try:
the switch is connected to. possible to exit the program using Ctrl-C,
while 1:
but then the library continues running and
# wait for the user
execfile(‘rpirobot.lib’) that is clearly not the intention.
# to push switch on
The steps if ioread(16):
# Switch on io 16, input
1. Set the LEGO finger in the rest posi- # push switch off
iodir(16,1)
tion, that means as far as possible time.sleep(0.5)
from the switch (against the bottom reverse(1)
print "Useless Box Program running"
surface for example). pwm(1,60)
2. Start the Raspberry Pi and wait until while ioread(16):
We use a try/except construct (see List- the booting process has completed. time.sleep(0.000001)
ing 1a), so that the program waits until 3. Copy the test folder from the download
brake(1)
the switch is turned on by the user or the to the Raspberry Pi (using WinSCP, for
user has quit the program using Ctrl-C. example).
Once this happens, the program waits half 4. Return to the test folder with the Listing 1b.
a second to give the user timer to remove instruction ‘cd test’.
their finger and then activates the LEGO 5. Now issue the instruction ‘python use- # retract lever
finger with a PWM value of 60 (60% of lessbox.py’. forward(1)
the maximum speed). We wait in a loop 6. Wait until the text ‘Useless Box Pro- pwm(1,30)
because of the threads that are running. gram running’ appears on the Rasp- counter1=0
Since nothing happens in the loop this is berry Pi and turn the switch on (and lastpin1=tacho(1)
not actually helpful, so we introduce a get your fingers out the way!). The while counter1<100:
short sleep instruction. This instruction LEGO finger will now return the switch time.sleep(0.000001)
takes 0.000001 seconds (1 µs), and will to its off position. if tacho(1)!=lastpin1:
counter1+=1
lastpin1=tacho(1)
S
# float motor
10k
1k float(1)
M4
+5V
GND
IO16
M3
Listing 1c.
except:
LEGO Control Board M2
LEGO 45502
# user aborts program with
M
M1 Ctrl-C
print "Program aborting"
abort()
Figure 6. Connection diagram for the switch with Figure 7. The complete model.
print "Done"
the two resistors.

learn design share
Debugging
the Arduino Zero & M0 Pro
Delving deeper into the world of Arduino
By Stuart Cording (iSystem, Germany) Atmel (now Microchip), are kitted out with a slightly different
programming interface compared to previous boards. The device
behind the USB “programming” connecter is also an Embed-
Boards like the Arduino M0 Pro and DUE provide, respectively, ded Debugger (EDBG) which can be used by many different
the same pin-out as their 8-bit Arduino Uno and Mega cous- development environments as an interface to not only program
ins, but with the advantage of significantly more SRAM, flash the board but additionally to interrogate the internal workings
memory and CPU clock speed. Perhaps the only disadvantage of the MCU at the heart of the system. With an appropriate
or challenge with these boards is the 3.3-volt restriction on the IDE, it is then possible to debug not only your sketch, but the
input and output pins. However, these advantages, combined entire Arduino core code and libraries.
with support for many of the available shields on the market One such IDE is iSystem’s winIDEA Open, the free version of
and the established base of libraries, make the ‘upgrade’ sim- their IDE targeted at MCUs that use ARM’s Cortex-M processor
ple and fast. technology. As a professional tool it is easy to get started, as,
The biggest challenge remaining for the advanced hobbyist at the most basic level, all that is needed is the output file from
or professional using Arduino for rapid prototyping is the lim- a sketch built in the classic Arduino IDE. However, as we will
ited Integrated Development Environment (IDE). For entry discover, the build manager can also be invoked to speed up
into the world of Arduino, the Arduino IDE is perfect with its the compilation of sketches. Additionally, the integrated test
simplicity and clarity. Once a sketch becomes slightly more environment testIDEA can be used to check whether any bugs
involved, however, and failures in the code cannot be found have scurried into our project during development.
with a blinking LED or serial message output alone, the sheer There are many details and nuances behind configuring and
power of a professional IDE is highly alluring. using such a tool. In order to provide both an overview for those
simply interested in this concept as well as detailed instruc-
Debug (as) a Pro, Pay Nothing tions for those who would like to go hands on, the overview in
The Arduino M0 Pro (from Arduino.org) as well as the Arduino/ this article is supplemented by a series of tutorials and code
Genuino Zero (from Arduino.cc), based on the SAM D21G from available online [1].

With the introduction of ARM Cortex-M-based 32-bit microcontrollers to the Arduino

portfolio, the maker has some very powerful hardware in their hands. For entry into
the world of Arduino, the simplicity and clarity of the Arduino code editor is perfect.
However, once a sketch becomes more involved, the power of a professional
development environment becomes highly alluring.
My First Project Workspace” window (Figure 1) all the files that belong to the
When starting a new project for a SAM D21 MCU in other IDEs, project, along with all the used functions, are listed in a tree
such as Atmel Studio which has been covered by Elektor before structure just like in Windows’ File Explorer. So, if you have
[2][3], you really need to start from a template and have to ever wondered what code lies behind the delay() function,
include many pre-defined library source code files. The win- simply expand the “Functions” folder, scroll down until you
IDEA Open IDE is different in that it forms the premise that find delay(unsigned long ms) and double-click. In the editor
the output of the compiler toolchain, for example the ELF window, the function will be displayed. If you are interested to
file containing the binary code of your Arduino sketch, is the see when the function is called, you can set a break-point on
most important file of your project. From this single file, the a line of code where a grey box is also shown in the editor’s
IDE can find all of your source code and the core Arduino files gutter (the grey region in the text editor to the left of the line
and libraries needed to debug the application code. Thus the numbers). Simply right-click with the mouse and select “Set
simplest method to debug an Arduino sketch is to create and Breakpoint”. Once the code is restarted, the MCU will halt when
build it in the Arduino IDE and then import and download the it reaches this breakpoint, allowing you to analyse the content
resulting ELF file into the Arduino M0 Pro using winIDEA Open. of variables, registers and memory.
In the first three projects of Tutorial 1 [1], the classic ‘Blink’ Note that, despite the Arduino environment’s apparent simplic-
sketch is created in the standard Arduino IDE. Due to slight ity, the code in its core software makes good use of many of the
differences between the Arduino M0 Pro and Arduino/Genuino advanced tricks the C programming language has to offer. As
Zero, winIDEA Open needs to be configured slightly differently. such, some ‘symbols’ (names of variables and functions) may
However, for those who just want to get started, a preconfig- appear listed in the Project Workspace window but may not actu-
ured workspace is provided allowing the developer to simply ally be associated with any code of significance in the software.
open the appropriate workspace and download the ELF file that
the Arduino IDE has created. Speeding Up the Workflow
Once the code is in the MCU, it is then possible to start exploring One of the issues with the method discussed thus far is that
the internal workings of the Arduino core code. In the “Project the original Blink.ino cannot be debugged properly. The issue

learn design share
relates back to how the Arduino IDEs compile the sketch. For
the beginner, this is really no problem, but when wanting to
go deeper into the code, it would be nice to be able to build a
sketch outside of the Arduino IDE.
There is also another benefit to be gained from this. One of
the most frustrating things about the standard Arduino IDEs is
that every compilation rebuilds the entire project from scratch.
Again, for smaller sketches this is no big deal. However, as
soon as a couple of libraries are included and you have several
source files, rebuilding and programming the board starts to
become quite a laboured activity. Here again, a professional
IDE can help by providing the means to build the project out-
side of the IDE.
In order to make the build process more intuitive and faster,

Tutorial 2 [1] uses a Makefile to build our sketch. This requires
a few changes to the creation of your sketch as follows:
• Your sketch will need to be stored in a file named <SKETCH
Figure 1. The ‘Functions’ view shows all the functions included in the NAME>.cpp and stored in a folder named ‘src’.
sketch, even those belonging to the Arduino core and library code. Your sketch will need the line of code #include
"Arduino.h" (see Figure 2) added to the top of it
(before the setup() function call).
In addition, you will need the make utility which can be installed
as part of MinGW. Detailed instructions on how to do this can
be found in Tutorial 2, available from [1].
Now, instead of building the sketch in the Arduino IDE (Fig-
ure 3) and then simply programming and debugging the code
in the winIDEA Open IDE, the whole process can be under-
taken in winIDEA Open.
As a result, we can now debug our sketch. The project work-
space works in a manner similar to that of Windows Explorer,
allowing the elements listed to be “expanded” to show their
contents in the same way folders can be expanded. In the
Project Workspace simply expand the elements listed using the
Figure 2. Blink.cpp in the winIDEA Open IDE. plus symbol to the left of each element in the following order:
sketch.elf ª Modules ª src ª Blink.cpp (Figure 4). Finally,
double-click on loop() or setup() and the editor will open the
source code at the respective line of code where the function
is implemented. As we see in the gutter, each line of code has
a grey box associated with it allowing us to set a breakpoint in
the code execution if we wish. Simply right-click on the desired
line of code and select “Set Breakpoint”.
Tips & Tricks

Now is probably a good time to note the limitations of the debug
features of the SAM D21. In total, only three breakpoints can
be set at any one time so, at some point, you will get a win-
dow pop up explaining that all breakpoint resources are used
up (Figure 5). The easiest method to work around this is to
disable a current breakpoint (right-click on a line with a break-
point and select ‘Disable Breakpoint’) and then set the new
breakpoint at the location of your choosing. It is also possible
to ‘Clear Breakpoint’, but disabling a breakpoint has the advan-
tage of leaving a red marker in the gutter of the code editor
as a reminder of where the breakpoint previously was. This
is especially helpful as you jump back and forth in the code.
In the event that you are interested in the efficiency of the code
generated by the Arduino tool chain (which incidentally is GCC),
Figure 3. The Blink sketch in the Arduino IDE. we can also open a disassembly window from the menu View

ª Disassembly. If we click inside this window (Figure 6), any

further single-stepping of the code will be executed instruction
by instruction, instead of source code line by source code line
as is the case in the editor window. In this way it is also pos-
sible to see how the CPU registers are used during the calling
of functions to pass parameters, amongst other things.
Beyond Debugging
Typically, when developing an application, specific functional-
ity will be broken out into separate functions. For small pro-
jects, where perhaps a single developer is responsible, it is
relatively easy to keep track of what is working, what isn’t
and where a bug might be when the code fails. However, for
larger, multi-programmer projects, it can help to have a bat-
tery of tests available that, when executed, ensure that the
code is still operating as originally specified. For this purpose,
winIDEA includes the Original Binary Code (OBC) unit testing
tool testIDEA. Here we will use it to develop some tests for a
function that evaluates an imaginary input value and returns Figure 4. The sketch executable can be explored as if it was a folder tree.
a new time delay for our delay() function call.
The algorithm implemented in the code is quite simple (Fig-
ure 7). If the passed value is less than 50, it returns 150. If
the passed value is between 50 and 99, it returns 1000. For
all other values, it returns 1750. Having developed the code,
it would make sense to develop a short test that could be exe-
cuted to ensure that it still works, even if someone else makes
changes and introduces a bug into the code.
Before writing tests for a function, it is worthwhile initially
considering how you would go about testing it by writing the
steps down on a piece of paper or making a small table of
Figure 5. Message when no more breakpoints are available.
input and expected output values. In the Table 1 we have
focused on input values at the extremes of possible inputs
as well as values that are close to the boundaries defined in
the algorithm (50 and 100).
In order to develop our tests, from the menu bar, we select project by clicking the refresh symbol. Once complete we
Test ª Launch testIDEA. Upon starting testIDEA, the tool sug- can select our function to be tested, evaluateNumber(),
gests creating a ‘test specification file’. This file is where our from the drop-down list.
tests will be stored and the file extension is ‘iyaml’. Here we • In the Parameters box we enter the value to be passed to
explain how to create a test: the function. For test number 1, this is 0.
• Next we enter the Expected Result, first selecting the
• From the menu, select Test ª New Test… ‘Default expression’ radio button. In the field we enter
• Now we need to define the function to be tested. Before 150, which is the expected response entered in our table.
we can do this, we have to refresh the link to our winIDEA • Click OK and the first test is complete.
Advertisement

learn design share
The uniqueness of OBC testing is that the testing itself all

occurs on the MCU target in use and not, like in some other
unit testing tools, in a simulation of the MCU. When we now
execute this test, the test will be actually executed on the MCU
target after downloading the code into the MCU’s flash memory.
This is performed by selecting Test ª Run All Tests. Hopefully,
upon executing the tests, the result will be “OK”, indicated by
a small green checkbox next to the test in the Outline panel.
In order to distinguish all the different tests from one another, we
can add some meta data to our test. In the Form panel, select
‘Meta’ and add a Test ID such as ‘Test_1’. We can now go on to
add the remaining tests to our testing plan. Upon completion, we
can run all the tests which should result in a pass in each case.
Of course, the real value is in using the tool to find any bugs that
have slipped in whilst we weren’t looking. Let us assume that
a colleague on the team misunderstands the specification and
decides that all the less-than (‘<') comparisons should actually
Figure 6. Disassembly view for the Blink sketch. be less-than-or-equal (‘<=') comparisons and changes them
in your code. The first time you become aware of the issue is
when the project as a whole does not work as expected. To
trial this potential mistake, replace the ‘<' with ‘<=' at lines 20
and 22 in the Blink.cpp source code (Tutorial 3, Project 5 [1]).
Table 1. Input values to exercise our algorithm In order to see where the bug may have occurred, simply rebuild
and the expected output values. the application in winIDEA, open up the test specification in
testIDEA and run all tests. You should see that tests 5 and 9
Test Number Input Value Expected Response
now fail and, working from that result, it should be possible
1 0 150
to determine the cause of the error.
2 1 150
3 48 150 Sounding off
4 49 150 The Cortex-M-based Arduino boards have enormous potential,
5 50 1000 especially in the area of rapid prototyping. STMicroelectron-
6 51 1000 ics recently announced a Cortex-M4 STM32-based board in
the form of the STAR Otto. Such boards will likely be used for
7 98 1000
complex prototyping with Ethernet and Wi-Fi shields. A pro-
8 99 1000
fessional debugging environment, such as winIDEA Open, will
9 100 1750 certainly provide relief when searching for why a sketch is not
10 101 1750 quite working as expected.
11 150 1750 (160228)
12 200 1750
Web Links
13 255 1750
Start
returnValue = 0
The SAMD21 is just one in the range of SAM D ARM
Cortex-M0+ based, 32-bit microcontrollers from Atmel.
These devices offer up to 256 kB of flash memory and
x < 50? returnValue = 150
32 kB of SRAM. Additionally, the devices boast a wide range
of interfaces, including full-speed USB, a real-time counter
for implementing a clock, USART, SPI and I²C interfaces,
and a 12-bit ADC. Optimal energy efficiency is also covered
x < 100? returnValue = 1000
with the power consumption lying at 70 µA/MHz. The family
also features an automotive grade version of the MCU and
some devices can be used to implement a capacitive touch
returnValue = 1750 End interface, making use of Atmel algorithms to implement
buttons, sliders and wheel user interfaces.
www.atmel.com/products/microcontrollers/arm/sam-d.aspx
Figure 7. Flow diagram of a simple algorithm.

Analog Random Numbers

Using an Arduino and a noise source
By Kurt Diedrich (Germany)
It’s easy enough these days to generate random numbers digitally using a computer program. But there is
also an analog approach: the project described here processes the noise produced by a transistor junction
using an Arduino. The random numbers so generated can be used in various interesting experiments and
will even lead us a little into the world of mysticism.
It is not only mystics who wonder whether there exists some Arduino passes the random numbers derived from the analog
universal consciousness that connects us all together and lets source to a PC over USB, where they can be displayed graphi-
us communicate subconsciously with one another. Thanks to a cally. The PC software has no pretensions to completeness and
project run from Princeton University using a collection of ran- interested readers are encouraged to extend it. Both the source
dom number generators spread all over the world, we may now code for the PC program and the Arduino sketch are available
have some signs that there may be a connection between human as a free download from the Elektor Magazine website [1].
consciousness and the behavior of electrons in a noisy pn-junc-
tion (see text box). Since I had just finished soldering together Hardware
a simple random noise source for my home-built analog syn- The circuit of the noise source is shown in Figure 1. The noise
thesizer, I hit upon the idea of trying to reproduce the Princeton originates in the reverse-biased base-emitter junction of tran-
experiment with my own hardware and software. sistor T1. This extremely weak signal has its DC component
Random numbers produced by a computer are just the results removed by C1 and is then amplified by IC1. The noise signal,
of mathematical calculations, and so are in fact deterministic, or now symmetric about zero, is taken via capacitor C2 to summing
‘pseudo-random’. Such numbers are not suitable for this project: amplifier IC2, where a voltage set by P2 is added to ensure
instead we must use an analog method to generate the numbers. that the output is always positive. This is required because
One approach is to process the noise created by the electrons the analog inputs of the Arduino can only work with positive
in the pn-junction of a reverse-biased diode. If this noise volt- voltages. Preset P1 can be used to adjust the amplitude of the
age is amplified and fed into an analog-to-digital converter, we noise signal. You can use an oscilloscope to set the average
will obtain a steady stream of unpredictable random numbers. level of the signal to about +1.4 V and the amplitude to about
1 VPP (an accurate measurement of these parameters of the
The project noise signal is not possible). The output of the noise source
The hardware I designed consists of a simple noise source can now be connected to the input of the Arduino.
whose output signal is fed into an Arduino microcontroller. The Figure 2 shows how the noise signal is fed into the Arduino:
PC via USB
+15V
R1 USB
R3
10M
10M R5 SCL
100k ARDUINO SDA
IC1
C1 R2 1 AREF
2 7 IC2
10k C2 P1 1 GND
8 6 7 NC
2
4n7 OP07 8 6 IOREF
13
3 R4 12
3u3 500k OP07 RESET
T1 4 3 Noise ~ 11
470k
C3 5 3.3V
4 Generator ~ 10
5 +15V 5V
4u7
P2 GND
~9
GND
POWER
Power GND
GND
8
BC547B –15V
DIGITAL (PWM ~ )
VIN 7
GND 100k
~6
AN0
AN0 ~5
AN1 4
C4
AN2 ~3
ANALOG IN
AN3 2
4u7
AN4 TX 1
AN5 RX 0
–15V 150831 - 11
150831 - 12
Figure 1. The circuit of the noise source uses a transistor and two Figure 2. Connections between the power supply, the Arduino and the noise
operational amplifiers. source.

learn design share
The points shown in the upper graph window represent the

random sample values received from the noise source. Every
second approximately thirty of these values are received from
the Arduino. Filling the window takes about ten seconds, cor-
responding to about 300 samples per pass that are available
for further processing.
Lower graph window

Two curves slowly build up in the lower window, one green and
one purple. Each time the upper window is filled each of these
curves is extended by one pixel. The curves are produced from
the random sample values as follows.
Purple curve (mode 2)

The program checks whether the current sample value of the
noise signal is even or odd, and respectively increments or dec-
rements a variable. Then the value of this variable is plotted.
Figure 3. User interface presented by the program, which automatically Green curve (mode 1)
starts collecting data when it is launched. If a sample is above the horizontal center line a variable is
incremented; if it is below the center line, it is decremented. It
is not easy to set the offset potentiometer in the hardware suf-
an Arduino Uno is used in this case. There are two connec- ficiently accurately, and in any case the offset will vary slightly
tions: first, the signal connection, which is taken to analog with temperature, and so a calibration process is required. The
input A0 of the Arduino; and second, the ground connection. calibration buttons help with this.
The noise source can be powered from any suitable external
2 x 15 VDC power supply. The Arduino is powered from the Calibration buttons
PC over the USB cable. These are the two buttons towards the top of the display, which
can be used to make vertical adjustments to the horizontal
Software line in the upper graph window that divides the positive sam-
As already mentioned, two separate pieces of software are ples from the negative ones. Depending on this setting, the
needed in this project: one running on the Arduino and one on green curve in the lower graph window will either run hori-
the PC [1]. The program running on the PC is written in Pro- zontally or will trend up or down. The offset potentiometer in
cessing. The Processing programming environment (which so the noise source should be adjusted so that the green curve
far I have only used under Windows XP and Windows 7) can runs as horizontally as possible when the threshold position is
be downloaded from the Internet free of charge [2]. A Pro- set to 150 using the buttons. In this case the random points
cessing file must always be placed in a directory that has the displayed in the upper graph window should appear to be per-
same name as the program file itself. fectly symmetric with respect to the center line. The threshold
value of 150 appears among the variable declarations in the
It is very important to ensure that you enter the correct COM Processing code, as follows:
port number in the setup part of the Processing code so that
the Arduino is recognized by Windows. You can find the correct int limit = 150 ;
number using the Device Manager. For example:
If a different value suits your hardware better, you can mod-
serport = new Serial(this, “COM3”, 115200); ify the code.
Unfortunately it can happen that the port number changes by Divider buttons
itself while you are experimenting with the system. The next two buttons down allow for the expansion and com-
After using the Arduino IDE to flash the sketch into the Arduino, pression of the curves in the Y direction, to allow them to be
it can be connected to the PC and then the Processing program analyzed in greater detail. The two numbers shown at the top
can be launched. After a brief delay the signal should appear on of the display are the most recent values plotted on the curves.
the screen in the upper graph window (see Figure 3, above),
and it should be symmetric about the horizontal center line. If Files and ‘LOAD FILE’
it is shifted or if the amplitude is not correct, P1 and P2 must Normally every thirty minutes a file is created storing the data
be adjusted. Below we will describe how you can fine-tune represented by the curves in the lower graph. These files are
these settings using the software. saved in the same directory as the Processing program. The
file name is determined when the program is started and con-
Operation sists of the time and the date when the program was started.
The time interval between files being saved can be adjusted
Upper graph window by modifying the variable declarations in the code: details on

Spooky action at a distance

Since the end of the 1990s a research group at Princeton balance when an event of global significance occurred that
University has been conducting a study, called the Global had an emotional effect on a large fraction of human beings:
Consciousness Project, or GCP, that collects and evaluates the events of 11 September 2001, the death of Princess
data generated by seventy random number generators Diana, or the Madrid attacks. In all these cases the readings
located all around the world. collected showed deviations from their expected properties.
In theory the distributions of numbers from these generators
should all be identical. For example, if the plot of noise Further information on the background of this research, as
sample values is shifted so that they are symmetrical about well as criticism of it, can be found at the following sites:
the zero line then occurrences of positive and negative values http://global-mind.org
will balance when analyzed over a sufficiently long time http://global-mind.org/results.html#alldata
period. The consequence of this is that the plot obtained http://global-mind.org/control.distribution.html
by accumulating a fixed number of these values should http://noosphere.princeton.edu/story.html
wobble about the zero line; if the shift is not exactly correct http://subroutine.jimdo.com (author’s website, partly in
then the plot will drift at an approximately constant rate in English, mostly in German)
one direction or the other. In rare cases the researchers at
Princeton observed significant disturbances to this delicate
how to do this are given in the program. On the other hand, you might prefer to spend your time mod-
Clicking on ‘LOAD FILE’ brings up a file selection dialog where ifying the processing program to display and analyze other
you can navigate to any previously-saved file and reload it. types of data more obviously influenced by the outside world,
such as readings from a temperature sensor!
Display current data / Display loaded file (150831)
The data plotted in the lower graph can be changed by clicking
Web Links
on the two buttons ‘Display current data’ or ‘Display loaded
file’. When the plot is changed to ‘Display current data’ it is [1] www.elektormagazine.com/150831
not updated until the plot in the upper graph reaches its right- [2] https://processing.org/download/
hand edge.
The figures below the buttons show the coordinates of the

mouse pointer when the mouse button is pressed. This is a
useful tool to have when adding new elements to the user
interface design.
Listing 1. Arduino sketch (excerpt).

Interpretation
Of course the plotted curves are open to various interpretations. void setup()
When you launch the program you will notice that, although {
there are many deviations of small magnitude, the overall Serial.begin(115200);
trend of each curve is firmly in one direction. The tiny jagged }
deviations are a natural consequence of the random nature of
the noise generator; they can also be caused by temperature void loop() // Endlosschleife
changes. However, if the curve should suddenly show a clear
{
change in direction from its previous average trend, perhaps
int sensorValue = analogRead(A0);
something is afoot! Perhaps you might want to take a quick
Serial.println(sensorValue);
look at your favorite news website to see what is going on...
Notes on the Processing code

The code might at first sight seem rather complicated. This is to the source code wherever required. Readers interested
because Processing does not have a built-in way to construct in programming in Processing will find many interesting
graphical user interfaces for programs. All buttons and text ideas in the listing, such as how buttons and text labels are
labels, and all the mouse events that connect together to programmed, how to display the contents of an array as a
form such an interface, must be written as ‘normal’ source plot in a graphics window, store them to a file and read them
code. However, the result works well and the payback for the back again, how to determine the time and how to create a
complexity of the code is that graphics handling in Processing file selection dialog.
is astonishingly quick. I have added thorough comments

learn design share
Cool Power
A high-performance amplifier
with lots of output power
By Ton Giesberts (Elektor Labs)

with D-Watt
Elektor has a long history and a good reputation
with high-end yet DIY, audio projects. This time
we proudly present a new power amplifier aimed
at all audio enthusiasts, built around a digital
audio driver IC and operating in class-D. This
has the advantage that the amplifier can
deliver a lot of output power (like 200 watts
into 8 ohms) with very low heat dissipation.
Say goodbye to big heatsinks!

learn design share
IRS2092. So why did we develop our own

design when someone else has already
FLOATING INPUT CSH done it for us? The answer, as you might
VAA VB
guess, is that a practical design turns out
to be good deal more elaborate than a
COMP HIGH
IN- VAA+VSS
SIDE
CS
simple copy of the reference design.
2
HO
OTA
GND Amplifier
HV
LEVEL VS
With an IC designed for a specific appli-
COMP SHIFT HV HV
LEVEL
FLOATING HIGH SIDE
LEVEL cation, it’s a good idea to start with the
SHIFT SHIFT
VCC standard application circuit from the man-
VSS
ufacturer. That certainly holds true for
the IRS2092. A detailed description of
this IC can be found in application note
CSD DEAD TIME
LO
AN-1138 [3].
HV SD DT
Our version is shown in Figure 2, where
LEVEL
SHIFT
COM the configuration around the driver stage
PROTECTION
CONTROL HV
LEVEL
IC2 and the dual MOSFET T1 is virtually
VREF
SHIFT LOW SIDE CS “by the book”. Transistor T1 is specifi-
OCSET
cally designed for this sort of application,
DT with two n-channel MOSFETs housed in a
single package. The IRFI4020H-117P is a
member of the Infineon family of digital
Figure 1. Block diagram of the IRS2092 digital driver IC used in this amplifier. audio MOSFETs and has a rated UDS max of
200 V, ID cont of 9.1 A and RDS(on) of 80 mΩ.
It is a perfect choice for power amplifiers
with output power of 200 to 300 W into
It has been nearly three years since we ferent components or component values 8 Ω. Thanks to the single package, this
presented our last power amplifier proj- if you wish. dual MOSFET is easy to mount on the
ect in Elektor — the Q-Watt Audio Power circuit board and on a heatsink.
Amplifier [1]. That was a standard class Design choices The feedback network between the out-
AB power amplifier with high output It is possible to build a class-D amplifier put and the input consists of R7 and R8,
power and low distortion. Since then a entirely from discrete components, but along with capacitor C6 to suppress RF
large number of audio enthusiasts have that does not make for a simple design. signals (in combination with R8). The gain
built their own copy of that amplifier. Fortunately, there are several semicon- is determined by the ratio of R7 + R8 to
With the design presented here, we are ductor manufacturers who make dedi- R5 + R6. The input impedance of this
taking the digital route for a change. In a cated ICs for class-D amplifiers. A large inverting amplifier is determined by R5
class-D power amplifier, the output tran- variety of ICs are available for low power and R6 and is fairly low (about 3.3 kΩ).
sistors operate in switching mode with amplifiers, but the choice is fairly limited if For that reason we added a buffer stage in
a very high switching frequency. Pulse you are looking for high power combined the form of opamp IC1A (an AD8031ANZ,
width modulation converts the amplitude with high quality. We found an outstand- which is a fast rail-to-rail opamp). You
of the analog input signal into a specific ing solution in the form of an IC designed can choose which of the two inputs you
ratio of positive and negative pulse widths by International Rectifier (now Infineon): want to use by fitting a jumper in the
in the output signal. A filter at the output the IRS2092. This 16-pin IC goes by the appropriate position on header JP1. The
blocks the high-frequency switching signal name “Protected Digital Audio Amplifier”, input impedance of the opamp stage is
and passes the amplified audio signal to and it integrates nearly everything you 11 kΩ. It is designed with a gain of 1,
the loudspeaker. The main advantage of need to build a relatively simple high-per- but if you wish you can increase the gain
this type of digital power amplifier is its formance class-D amplifier. You only have by fitting a resistor in the R3 position
high efficiency, which allows the heatsinks to add some power transistors. (A = R4/R3 + 1). The supply voltage
to be much smaller than with a compa- The IC contains gate drivers that can for the opamp is provided by the ±5 V
rable conventional design. deliver high currents for the connected voltage regulators, which also supply the
The key components in this amplifier are power MOSFETs, a PWM modulator with auxiliary voltages for IC2 (see below).
a driver IC in a 16-pin package and a dual built-in oscillator, bidirectional current
power MOSFET. Together with extensive protection and programmable dead
protection circuitry, they are mounted on time for driving the MOSFETs, and it can
a single PCB, so you only have to add a operate with switching frequencies up to
suitable power supply —either regulated 800 kHz (see the block diagram in Fig- Figure 2. Complete schematic diagram of the
or unregulated. There are no SMDs in ure 1). Along with the datasheet [2] and D-Watt amplifier. The amplifier circuitry is shown
the design, so building the board is very a wealth of application notes, Infineon in the top half, with the protection circuitry in
easy and you can experiment with dif- offers a lot of reference designs for the the bottom half.

77
www.elektormagazine.com January & February 2017
+VP R7 R8
120k 1k
R30 R31
R12 C6
+5V
10R
10k
IC3 +5V
1k8
2W +12V 150p +VP

7805 R10 D6 T2 TIP122G
1 3 R16
D4 4R7
IC5 +12V
10R
C17 C16 R21

R15 D7 C22 C23
MUR120G 7812
2
33k
1u 100n 1 3
10R
18V 100n 680u

1W3 50V C10 C11 BAV21 D10
VCC C20 -VP 100V C33 C32 C31
2
U 50V 10u 100n 10u R22
OCPH=19.5A
R24
50V 18V 1u 10u 100n
1 12 15 1W3 50V 50V
JP1
8k2
10k
U = unbuffered VAA VCC VB
B = buffered C4 R5 R6
U 3 16
330R 3k0 IN- CSH
+5V -5V B C7 C21 R23 -VP
10u C5 T1 +VP
C26 C27 C28
10k
2x D1 D2 +5V 1n 22u
1N4148 AD8031ANZ 1n 50V R25 D8
R9 C8 14 100n 100n 100n
B R1 HO 18R 2x
3 C2 L1
470R
1k C15 1N4007 RE1B

6 1n 4 13 K1
IC1 7
100n P1
COMP IC2 VS
R28
2
2 22µH
R2 IC1 10n C9 IRS2092PBF D9 1 LS+
C1 C3 R27 C29
10R
4 R26
reader project
R29
10k
R4 1n 11
LO 18R
1R
1n 2k 100n 1k 2 470n
2k2
GND
R3 1W -VP
-5V IRFI4020H-117P 2W
10
-5V COM
NM
C30
VCC K2
C24 C25 2
D5 R11 5
C19 C18 R14 CSD R19 -VP 100n 1 LS-
C14 100n 680u
10R VSS VREF OCSET DT
2k2
100V
1
1u
5k6
18V 100n
50V 50V 10u 6 7 8 9
1W3 2 3 R17
R13 6k8
7905 R18 R20
D3 C12 C13
IC4 -5V
1k8
NM
3k0
2W 1N4007 50V 10u 100n +VP
+5V OCPL=19.5A
-VP +VP
-VP D16
labs project
R37 K3
Ext. R43 R44 R45 +5V R49 R51 0W5 2
Test DC
1 +
10k
68V
820k
T5
15k
22k
22k
R48
2k7
R53
22k
T9
47k
JP2 D12
D13 RE1A T12
HS1 BC557C R54 K4
CSD
D11 Ext. Amp. T11 10k 2
R32 R33 1N4148 R38
+5V RT314048 D17 1 0
5520 Ω
4x R35 T7 R39 R41 1N4148 4x
4k7
15k
22k
BC547C BC547C R50 R52
T8 1N4148
6k8
22k
D14
1M
1N4148
120k
10k
Over-Voltage Protection
K5
R34 T10 2
D15
R36
C34
R40 C35 R42
IRFD 1 –
T3 T4 T6
10k
210PBF 0W5
10M
10k
4k7
2u2 68V
100u 50V R46 R47 –VP
10k 150k
Under-Voltage Protection 150115 - 11
-5V -VP
learn design share
Virtually all of the components around additional decoupling at the auxiliary sup- tive to ground, is used to drive the high-
IC2 have specific functions. The major- ply voltage pins of IC2 (VAA and VSS) is side MOSFET. This pin is connected to a
ity of the resistors connected to the IC provided by networks in series with the standard bootstrap circuit consisting of
determine the two current limit levels, supply lines (R10/C10/C11 and R11/C12/ R16, D6 and C21, along with R21 which
the dead time, the bootstrap drive volt- C13). Diode D3 protects the IRS2092 if is necessary to charge C21 before the
age for the high-side MOSFET, and the the negative supply voltage drops out. PWM oscillator starts running. Resistor
feedback from the output. The voltage on the negative supply volt- R16 limits the peak current.
IC2 requires two auxiliary voltages (+5 V age pin VCC (which is used to drive the
and –5 V). Instead of deriving them low-side MOSFET) is +12 V with respect The dead time (the time interval when
directly from the main supply voltages to the main negative supply voltage -VP. both MOSFETs are cut off to prevent large
using Zener diodes and series resistors, This pin draws somewhat more current peak currents through both devices at
we use separate voltage regulators (IC3 than the auxiliary supply voltage pins, so the same time) can be set with voltage
and IC4) to keep the auxiliary voltages a stabilization circuit consisting of volt- divider R19/R20 to one of four discrete
stable even when the amplifier is pow- age regulator IC5 in series with an exter- values: 105, 65, 40 or 25 ns. Here we
ered from an unregulated power supply. nal Darlington power transistor (T2) is selected the minimum dead time set-
There are Zener diodes (D4 and D5) at provided here. In this arrangement the ting of 25 ns by omitting R20. Note that
the inputs of the voltage regulator ICs to transistor handles the additional power increasing the dead time causes a sig-
keep their input voltages at a safe level dissipation. The VCC pin of IC2 is locally nificant increase in THD in the upper fre-
(the maximum allowable supply voltage decoupled by C20 and R15. quency range. You should bear in mind
is ±70 V) and reduce the dissipation of that MOSFETs generally need about twice
these ICs. To prevent HF switching noise The voltage on the positive supply volt- as much time (or longer) to switch off
from affecting the voltage regulators, age pin of IC2 (VB), which floats rela- than to switch on. Resistors R25 and R26
limit the charge and discharge currents
to and from the gates of the power MOS-
FETs. The short-circuit current ratings of
Measured performance the driver outputs are 1 A for sourcing
Measured with the 6/8 Ω version and ±60 V supply voltages and 1.2 A for sinking. With a value of
(Hypex SMPS400A400), BW = 22 kHz, R7 = 100 kΩ 18 Ω for R25 and R26, the peak charge
and discharge currents are limited to
• Input sensitivity: 1.3 V (189 W / 8 Ω, THD+N = 1%)
about 0.5 A.
1.25 V (174 W / 8 Ω, THD+N = 0.1%)
• Input impedance: 3.33 kΩ / 11 kΩ
Voltage divider R17/R18 determines the
• Continuous output power: 174 W into 8 Ω (THD+N = 0.1%, 1 kHz)
low-side current limit. Here the on resis-
189 W into 8 Ω (THD+N = 1%, 1 kHz)
tance (RDS(on)) of the MOSFETs is used for
224 W into 6 Ω (THD+N = 0.1%, 1 kHz)
current sensing. The positive tempera-
246 W into 6 Ω (THD+N = 1%, 1 kHz)
ture coefficient of RDS(on) has the advan-
• Power bandwidth: 5 Hz (–3 dB) to 20 kHz (–1 dB) tage that the current limit level decreases
• Slew rate: 15 V/µs (8 Ω) when the temperature of the MOSFETs
• Rise time: 4.4 µs (8 Ω) increases. The IRS2092 has an internal
• Signal to noise ratio: >77 dB (linear, BW = 22 Hz to 22 kHz) 5.1 V reference voltage (VREF on pin 7)
(reference 1 W / 8 Ω) >80 dBA for precise setting of the current limit.
• Total harmonic distortion plus noise: 0.014% (1 kHz, 1 W / 8 Ω) Unfortunately, the drain-source resis-
(BW = 22 kHz) 0.004% (1 kHz, 50 W / 8 Ω) tance of MOSFETs has a fairly high tol-
0.024% (1 kHz, 1 W / 6 Ω) erance. With a gate-source voltage of
0.0046% (1 kHz, 50 W / 6 Ω) 10 V, the RDS(on) of the MOSFETs used
• Intermodulation distortion: 0.013% (1 W / 8 Ω) here is 80 mΩ typical and 100 mΩ max-
(50 Hz / 7 kHz, 4:1) 0.023% (50 W / 8 Ω) imum (at 25°C). At 125°C, RDS(on) rises to
0.013% (1 W / 6 Ω) more than 175 mΩ. In our design we set
0.017% (50 W / 6 Ω) the current limit voltage to 1.56 V, which
• Dynamic IM distortion: 0.037% (1 W / 8 Ω) puts the current limit at 19.5 A (typical)
(3.15 kHz square wave / 0.009% (50 W / 8 Ω) at a junction temperature of 25°C. At
15 kHz sine wave) 0.043% (1 W / 6 Ω) 125°C the limit level drops to approxi-
0.008% (50 W / 6 Ω) mately 8.7 A. If the output current is too
• Damping factor: 174 (1 kHz, 8 Ω) high, the IRS2092 shuts down and the
• Efficiency at full power: 93% with 8 Ω load
LO and HO outputs are set low to protect
the MOSFETs.
(THD+N = 0.1%) 94% with 6 Ω load
• DC protection: +3 V / –4 V
The high-side current limit is set using
• DC output offset: <0.3 mV
the CSH input (pin 16). This pin has a
• Switch-on delay: 6s
fixed threshold voltage of 1.2 V relative

The main advantage of a digital power amplifier is its high efficiency
to VS (pin 13). For high-side current lim- amplifier with a 4 Ω load, it will frequently protect the connected loudspeaker. The
iting, a blocking diode (D7) is necessary go into current limiting with ±60 V sup- DC protection circuit reacts at a DC level
in addition to the voltage divider in order ply voltages. In that case it is better to of a few volts. A relatively high value was
to protect the CSH pin against high volt- modify the amplifier and use different chosen for this because there is always
age when the high side is switched off. supply voltages, as described in the “4 Ω some DC voltage present at the out-
The 0.6 V forward voltage drop over D7 Configuration” inset. put when no load is connected and the
increases the drain-source voltage of the IRS2092 is shut down. Apparently there
high-side MOSFET by 0.6 V. Resistor R24 Protection is a leakage current from the VS pin when
provides a bias voltage for D7 when the To supplement the built-in protection the IRS2092 is in shutdown mode.
high-side MOSFET is switched on. Volt- of the IRS2092, we added a number of
age divider R22/R23 reduces the voltage external protection circuits: overvolt- The DC protection circuit consists of R38,
over the high-side MOSFET and D7. All age protection for both supply voltages, R39, R40, C34, T6 and T7. Voltage divider
this means that if you want to set the undervoltage protection (also for both R39/R40 determines the trip levels. With
same limit level here as for the low side, supply voltages), and DC protection with the indicated resistor values, they are
you have to take D7 into account. That switch-on delay. Each of these protection +3 V and –4 V. Although the selected
yields a value of 10 kΩ for R23 and 8 kΩ circuits can de-energize the output relay levels are somewhat higher than usual,
(adjusted to 8.2 kΩ) for R22. See the and shut down the IRS2092. they should not be harmful to the loud-
description of this amplifier on the Elek- The DC protection circuit
tor Labs website [4] for the calculation. ensures that the output
The PWM signal at the output of the is switched off if the DC
MOSFETs is filtered by a second-order voltage at the output is
filter consisting of toroidal inductor L1 too high, in order to
and capacitor C29. It is followed by a
Boucherot network (R28/C30), which pro-
vides an output load for the amplifier at
high frequencies.
Capacitors C22–C28 decouple the supply

voltages. As with all class-D amplifiers,
there is a risk of power transfer from the
output filter to the power supply due to
the “bus pumping” effect. That mainly
occurs at high signal levels and low fre-
quencies with low load impedance. With
a regulated power supply (in particular a
switching power supply), this can desta-
bilize the feedback loop. The simplest way
to suppress a potential increase in the
supply voltage is to connect large capac-
itors in parallel with the supply lines. You
should start with a 10,000 µF, 100 V
electrolytic capacitor on each supply
line. Be careful with this, because most
switching power supplies have difficul-
ties with large capacitive loads (check
the manufacturer’s specifications).
The supply voltages for the amplifier

are ±60 V for the 6 Ω / 8 Ω ver-
sion, with 100 kΩ as the recom-
mended value for R7. The maximum
is ±70 V, but at that level the dis-
tortion increases because the PWM
component of the feedback signal
is stronger. If you want to use the

learn design share
Component List
Resistors 5x11 mm, 2mm pitch IC4 = MC7905ACTG
Default: 5%, 0.25W C15 = 10nF 10% 100V, X7R, 0.2’’ pitch IC5 = MC7812CTG
R1,R8 = 1kΩ C17,C19,C33 = 1µF 20%, 50V, 5x11 mm,
R2,R23,R24,R30,R34,R36,R37,R46,R50,R54 2mm pitch Miscellaneous
= 10kΩ C21 = 22µF 20%, 50V, 5x11 mm, 2mm pitch K1,K2,K3,K4,K5 = Faston spade connector,
R3,R20 = not mounted C22,C24 = 100nF 10%, 250V, MKT, 4x10 mm, PCB mount, dim. 6.35x0.81 mm, 0.2’’ pitch
R4 = 2kΩ 1% 7.5mm pitch (TDK type B32520C3104K000) JP1,JP2 = 3-pin pinheader, 0.1’’ pitch, with
R5 = 330Ω C23,C25 = 680µF 20%, 100V, 28mΩ @ 2.57A, jumpers
R6,R18 = 3kΩ 1% 18x40 mm, 7.5mm pitch (Rubycon type RE1 = relay, PCB mount, SPCO, 16A,
R7,R52 = 120kΩ 100ZL680MEFC18X40) 48V/5.52kΩ (TE Connectivity/Schrack type
R9 = 470Ω C26 = 100nF 10%, 200V, X7R, 0.2’’ pitch (Ke- RT314048)
R10,R11,R15,R31 = 10Ω met type C330C104K2R5TA) HS1 = heatsink, PCB mount, 2.6K/W, mount-
R12,R13 = 1.8kΩ, 2W C27,C28,C30 = 100nF 5%, 400V, poly- ing pins spaced 1 inch (Aavid Thermalloy
R14 = 2.2kΩ propylene, 15mm pitch (TDK type type 530002B02500G)
R16 = 4.7Ω B32652A4104J000) Heatsink for IC4/IC5/T2: 27x90 mm alumi-
R17,R35 = 6.8kΩ C29 = 470nF 5%, 400V, polypropylene, 15mm num sheet, 2mm thick
R19 = 5.6kΩ pitch (TDK type B32652A4474J000) Heatsink for IC3: 30 x 30 mm aluminum
R21 = 33kΩ C34 = 100µF 20%, 10V, bipolar, 2.5mm or sheet, 1mm or 2mm thick
R22 = 8.2kΩ 3.5mm pitch, diam. 8mm max. Isolation materials for T2,IC3,IC4,IC5 (TO-220
R25,R26 = 18Ω C35 = 2.2µF, 50V, 5x11 mm, 2mm pitch thermal pad + TO-220 bush)
R27 = 1Ω, 1 W 6 pcs solder pin, 1.3mm diam.
R28 = 10Ω 1W Inductor 1 pc DIP-16 IC socket
R29 = 2.2kΩ, 2W L1 = 22µH, Micrometals type T130-2 ring 1 pc DIP-8 IC socket
R32,R40 = 4.7kΩ core, max. 20x40mm. Wire: 200cm of Kit of parts incl. PCB 150115-1 (mono), excl.
R33,R43 = 15kΩ 1.5mm/AWG15 enameled copper supply: Elektor Store # 150115-71
R38,R39,R45,R48,R51 = 22kΩ PCB # 150115-1, available separately and
R41 = 1MΩ Semiconductors optionally
R42 = 10MΩ D1,D2,D11,D13,D14,D17 = 1N4148
R44 = 2.7kΩ D3,D8,D9 = 1N4007
R47 = 150kΩ D4,D5,D10 = BZX85C18 SB00018/E1, Ancillaries for Hypex SMPS400A400 PSU
R49 = 820kΩ 18V/1.3W zener diode connection
R53 = 47kΩ D15,D16 = BZX55C68-TR, 68V/0.5W zener di- 1 pc JST VHR series 5-way straight case for
P1 = 1kΩ trimpot, top screw adjustment ode (alternatively, NXP type BZX79-C68) PCB connector, 3.96mm pitch (RS Compo-
D6 = MUR120G (200V, 1A) nents # 820-1188, set of 5 pcs)
Capacitors D7 = BAV21 (250V, 250mA) 1 x JST VHR series 3-way straight case for
C1,C5,C7,C8,C9 = 1nF 1%, 63V, polystyrene, D12 = LED, green, 3mm PCB connector, 3.96mm pitch (RS Compo-
7.18mm pitch (LCR Components type EXFS/ T1 = IRFI4020H-117P nents # 820-1175, set of 5 pcs)
HR 1000pF ±1%) T2 = TIP122G 5 x JST cable set with crimp connectors at
C2,C3,C11,C13,C16,C18,C31 = 100nF 10%, T3,T4,T6,T7,T9,T10,T11,T12 = BC547C both sides (RS Components # 820-1135, set
50V, X7R, 0.2’’ pitch T5 = BC557C of 5 pcs)
C4 = 10µF 10%, 63V, MKT, 15mm pitch T8 = IRFD210PBF 3 pcs isolated Faston connector, female, 0.25’’
C6 = 150pF 5%, 500V, silver mica, 5.9mm IC1 = AD8031ANZ pitch
pitch (Cornell Dubilier type CD15FD151JO3F) IC2 = IRS2092PBF
C10,C12,C14,C20,C32 = 10µF 20%, 50V, IC3 = MC7805CTG

speaker system. When T6 or T7 starts

conducting due to excessive DC voltage,
it drives T5 into conduction, which in turn
switches on T4. That transistor discharges
capacitor C35, cutting off MOSFET T8 and
causing the relay to drop out. At the same
time, the IRS2092 is shut down by dis-
charging C14 on the CSD pin through T3,
which acts as a level adapter. Diode D11
near T5 is included to prevent T3 from
being constantly driven in to conduction
through R35.
The relay is driven by a small MOSFET
so that the value of capacitor C35 (which
Figure 4. Two of the heatsinks are home-made Figure 5. To keep the parasitic capacitance as
together with R41 provides the switch-on
from aluminum sheet metal with a thickness of low as possible, 45 turns of copper wire have to
delay) can be kept small. That also allows 1 mm or 2 mm, respectively. The small one is for be wound on the core in a specific manner.
C35 to discharge faster through T4. IC3, and the large on is for IC4, IC5 and T2.
Header JP2 is provided for testing the DC
protection circuit. For normal operation
the jumper is placed in the “Amp.” posi- the relay is de-energized immediately experiment with various modifications
tion here. If you want to try a different when the amplifier is switched off in order if you wish.
value for R40 in order to change the DC to prevent popping sounds from the loud- The bare PCB is available in the Elektor
trip level, put the jumper in the “Ext. DC speaker. Here two transistors are used for Store, but you can also buy a semi kit
Test” position so you can use an external voltage detection, instead of two Zener (order number 150115-71) containing
adjustable DC voltage source. diodes. Transistors T10 and T11 both con- the PCB and all necessary components,
duct when the two supply voltages are including the heatsinks (note that you
The maximum continuous drain current of high enough, causing T9 to be cut off. will have to drill the holes in the heat-
the IRFI4020H-117P dual MOSFET used Otherwise T9 is driven into conduction sinks yourself).
here is 9.1 A with a case temperature of and the relay is de-energized through
25°C. At a temperature of 100°C it is only T5, T4 and T8 as previously described. Not much needs to be said about mount-
5.7 A. For this reason, we added supple- Diode D14 prevents T11 from conducting most of the components; they should
mentary overvoltage protection to limit ing when T10 is fully conducting. In that not present any difficulties if you have a
the power dissipation at relatively high situation the voltage on the collector of reasonable level of soldering experience.
supply voltages. This amplifier is designed T10 is about -0.6 V, and without D14 Do not mount T1 right away.
for maximum supply voltages of ±70 V. the emitter of T11 would also be at that
The overvoltage protection is dimensioned level, so it would also start conducting Use good-quality sockets for IC1 and
to react at about ±75 V. The main com- at a very low supply voltage level. That IC2, preferably with turned pins. It is
ponents here are the two Zener diodes would impair the operation of the under- important to use capacitors of the best
D15 and D16. Depending on the values voltage protection circuit. The detection possible quality for the PWM modulator
of series resistors R52 and R53, transistor levels are determined by voltage divider circuitry and HF decoupling of the audio
T12 starts to conduct when the positive R49/R50 for the positive supply voltage input signal (C1, C5, C7, C8 and C9). For
or negative voltage is too high. It drives and voltage divider R47/R46 for the neg- that reason we provided lots of room for
transistor T5 into conduction (the same ative supply voltage. The values of R49 these components on the PCB, so you can
as for DC protection), which causes the and R47 also depend on the base current decide for yourself whether you want to
relay to be de-energized through T4 and of T11 and the emitter current of T10. mount polystyrene, polypropylene or ordi-
T8. Diode D17 is necessary for detec- With the values shown on the schematic nary polyester types. For HF decoupling of
tion of positive overvoltage. Without D17 diagram, the undervoltage protection the feedback network (C6), we specifically
the voltage on the emitter of T12 would will react when the supply voltages drop recommend using a high-quality silver
rise in an overvoltage situation, but that below about ±40 V. mica capacitor (150 pF / 500 V). Using
would not cause the transistor to start a different type here is not advisable.
conducting. Construction: For coupling capacitor C4 at the input
Undervoltage detection is a bit more com- three heatsinks and a coil we used a standard MKT type. Polypro-
plicated, and it additionally ensures that Figure 3 shows the printed circuit board pylene is not desirable here due to the
for the D-Watt amplifier. Although it may large component size, which would cause
seem fairly large, all necessary heatsinks the capacitor to pick up too much noise
are also located on the board, so there (including switching noise).
Figure 3. The entire amplifier, including the is no need to add a bulky external heat-
protection circuitry and all necessary heatsinks, sink. Furthermore, all the components Voltage regulator IC3 only needs a small
is located on a single PCB. An external heatsink are leaded types, so building the board heatsink, which you can make your-
for the output transistors is not necessary. is straightforward and you can easily self from a piece of aluminum sheet

learn design share
Some measured characteristic curves +3

Audio Precision
+2
Test equipment: Audio Precision System Two Cascade +1
Plus 2722 Dual Domain (Elektor Audio Lab) -0
-1
-2
-3
d
Plot A B
r
-4
-5
This shows the measured amplitude versus frequency with load A
-6
impedances of 4, 6 and 8 Ω. In the 20 kHz region the curves show that -7
-8
the dimensioning of the output filter is a good compromise for the different -9
-10
impedances. The deviation from 10 to 20 kHz is approximately ±1 dB. -11
With 8 Ω (cyan) the resonant frequency of the output filter is clearly A
-12
10 20 50 100 200 500 1k 2k 5k 10k 20k 50k 100k
Hz
visible. With 6 Ω (green) the curve is nearly flat, and with 4 Ω (blue) the
corner frequency is somewhat lower. Audio Precision
10
Plot B 1
0.5
THD+N at 1 W and 50 W with an 8 Ω load and measurement bandwidths 0.2
of 22 kHz and 80 kHz, respectively. Measurements with large bandwidth % 0.1
0.05
are usually not shown in data sheets for class-D power amplifiers because
0.02
the effect of the switching frequency increases with greater bandwidth. 0.01
However, the amplifier does produce output signals above 22 kHz, so 0.005
0.002
showing what happens above that frequency is only fair, even if you 0.001
B
20 50 100 200 500 1k 2k 5k 10k 20k
cannot hear it. The red curve shows the THD plus noise at 50 W, measured Hz
with a bandwidth of 22 kHz. The blue curve was measured at the same
power level but with a measurement bandwidth of 80 kHz. The cyan curve Audio Precision
+0
shows the THD plus noise at 1 W with a bandwidth of 22 kHz, and the -10
yellow curve show the THD plus noise at 1 W with a bandwidth of 80 kHz.
-20
-30
All measurements were made using a ninth-order elliptical low-pass filter -40
-50
with a corner frequency of 200 kHz ahead of the analyzer input to block -60
d
B -70
residual PWM noise from the amplifier. r
-80
A
-90
-100
-110
Plot C -120
-130
FFT of a 1 kHz signal at 1 W into 8 Ω. Hardly any harmonics are visible, -140
C
-150
although the second harmonic can be seen in the noise of you look closely. 10 20 50 100 200 500 1k
Hz
2k 5k 10k 20k 50k 100k
The TDH+N is 0.014% and consists almost entirely of noise. In this FFT
plot and the next one, the switching frequency of the Hypex power supply Audio Precision
+0
is clearly visible at 90 kHz (according to the manufacturer’s datasheet, it -10
-20
ranges from 80 to 120 kHz), but the level is so low (–82 dB) that it is not -30
a cause for concern. -40
-50
-60
d
B -70
r
-80
A
Plot D -90
-100
The same FFT measurement as plot C, but with a power level of 50 W -110
-120
into 8 Ω. Here some harmonics are visible. The THD+N is 0.004%, and -130
the switching noise from the power supply has dropped to an insignificant -140
D
-150
level of –96 dB. 10 20 50 100 200 500 1k

Hz
2k 5k 10k 20k 50k 100k
Audio Precision
Plot E 10
5
THD+N versus output power (1 kHz / 8 Ω, BW = 22 kHz). The 2
measurement bandwidth was reduced here to improve the visibility of the 1

0.5
distortion curve. The distortion gradually increases above approximately 0.2
50 W. The amplifier can comfortably deliver 200 W with the nominal % 0.1
0.05
supply voltages. 0.02

0.01
0.005
0.002
0.001
1m 2m 5m 10m 20m 50m 200m 500m 1 2 5 10 20 50 100 300
E W

with a thickness of 1 mm. An area of First check that the leads are properly filter does not cause much distortion. The
30 x 30 mm is sufficient. The metal positioned in the holes without twisting, downside of this core material is its low
mounting tab of the IC is connected to and then tighten the screw firmly. After permeability, which means a lot of turns
ground, so the IC does not need to be this, solder the MOSFET leads to the PCB. have to be wrapped on the core to obtain
insulated from the heatsink. Mount the the necessary inductance.
heatsink plate on the voltage regulator Output inductor You need 2 meters (approx. 7 feet) of
so it is a few millimeters above the sur- The most difficult component in this proj- 1.5 mm enameled copper wire (15 AWG)
face of the board. Also make sure that ect (in terms of construction) is the induc- for the winding, and you have to wind 45
the mounting hole in the corner of the tor of the output filter. You have to wind turns on the core. They will not fit in a
board remains accessible (see Figure 4). it by hand on a toroidal core with a diam- single layer on the inner circumference
IC4, IC5 and T2 are all mounted on an eter of 33 mm (1.3 inch) using 1.5 mm of the core, so the coil must be wound
aluminum plate 2 mm thick with dimen- enameled copper wire (15 AWG). The in a special way to keep the parasitic
sions of 27 x 90 mm, with one end bent core is an iron powder type from Micro- capacitance low. The turns of the sec-
at a right angle 65 mm from the end (see metals. We chose this core ond layer of the winding are placed on
Figure 4 again). To avoid contact with for its linear character- top of the turns of the first layer on the
other components, mount the heatsink istics, so that the inner circumference of the core. On the
on these three components so it is outer circumference, the turns of the sec-
at least 5 mm above the surface of ond layer are placed next to the turns
the board. All three components must of the first layer. The turns of the
be insulated, and two of them (IC5 third layer are placed next to the
and T2) are mounted opposite each turns of the first layer on the
other and secured by the same screw. inner circumference, and next
The two holes in the plate are spaced to the turns of the
4 Ω Configuration
If you want to use the amplifier with a

4 Ω load, we recommend using a different
MOSFET type for T1: the IRFI4212-117P. In
exactly 40 mm addition, the maximum supply voltages in that case are
apart, and IC4 ±46 V.
is located 14 mm Several other small modifications are also necessary. To start with,
from the edge of the the current limit value must be raised by setting R17 to 6.2 kΩ, R18 to
board. You will have to 3.9 kΩ and R22 to 12 kΩ. That makes the current limit level 15.33 A with a
determine the height of 3-Ω load.
these two holes by empiri- Now the 48 V relay can be connected directly to the power source, with R44
cal measurement. replaced by a wire link. The feedback network must also be modified by reducing
Next we have the heatsink for T1. The the value of R7 to 75 kΩ.
package of this dual MOSFET is com- At continuous full output power with a 3-Ω load, current limit protection will trip
pletely composed of plastic, so it does after a few seconds because the temperature of the main heatsink will rise too
not need to be insulated. First mount the much, despite the high efficiency. With this configuration it can therefore be a good
heatsink on the PCB, then apply a thin idea to make a larger heatsink yourself (no higher version is available for the type
coat of thermal grease to the rear of T1 specified in the components list).
and insert its leads into the correspond- Extensive measurement data for the 4-Ω configuration is available on the Elektor
ing holes in the board. Secure T1 to the Labs website [4]. Among other things, it shows that the amplifier can deliver
heatsink with a screw, plain washer, split 243 W into 4 Ω with 1% distortion (and even 297 W into 3 Ω).
washer and nut (screw length 12 mm).

learn design share
second layer on the outer circumference. PCB and solder it in place. usual, so we measured output voltages of
Repeat this procedure for the fourth layer. That completes the assembly of the cir- ±60 V). With this power supply the ampli-
After you have wound 45 turns on the cuit board. fier is suitable for loads from 6 to 8 Ω.
core this way, the ends of the copper For lower load impedances it advisable to
wire will be located opposite each other Power supply and enclosure modify the amplifier (see the inset), as
on either side of the core. You may have Next you have to fit the board and its otherwise the protection circuit will trip
to improvise or make a few adjustments power supply into a suitable enclosure. occasionally during high-power passages.
to get all 45 turns on the core this way, Although it is possible to use a conven-
but hopefully you understand what we tional power supply with a hefty power In Figure 6 you can see how we fit-
mean. Have a look at the picture of the transformer, bridge rectifier and filter ted two amplifier boards and two Hypex
finished coil in Figure 5. We should warn capacitors, that doesn’t really fit with a power supplies into a standard enclosure.
you that 1.5 mm copper wire (15 AWG) modern class-D amplifier. Furthermore, There is not much wiring in the enclo-
is fairly stiff and not easy to bend. Work the measured performance of the D-Watt sure because everything is located on a
carefully and wind the wire as tightly as Amp is significantly better with a regu- single PCB.
possible against the core, but do not use lated power supply. For that reason, we
any metallic tools due to the risk of dam- used switching power supplies from the We hope you enjoy building and using
aging the insulation of the wire. Dutch company Hypex in our prototype. this cool high-power amplifier. More
They are specifically designed for use with detailed information about the ampli-
When you are done, you will have a coil class-D power amplifiers. We chose the fier, along with even more measurement
with an inductance of SMPS400A400, which can supply 400 W data, is available on the Elektor Labs
22 µH and a resistance at 20 Hz with nominal output voltages website [4].
of about 20 mΩ. Now of ±64 V (the mains voltage (150115-I)
you can mount it in our lab is a bit
vertically on the lower than
Figure 6. Two amplifier boards and two switching power supplies fitted
in an enclosure.
Web Links
[1] www.elektormagazine.com/magazine/elektor-201309/23330
[2] www.infineon.com/dgdl/irs2092.pdf?fileId=5546d462533600a401535675f1be2790
[3] www.infineon.com/dgdl/an-1138.pdf?fileId=5546d462533600a40153559a077610d1
[4] www.elektormagazine.com/labs/200w-class-d-audio-power-amplifier-150511

Sensors make Sense (2)

For Arduino and more
Sensors are either analog or digital. Analog values are read in via an A-D input, whereas for digital signals
a simple Port connection frequently adequate. In some cases, however, sensors have dual outputs, one
analog and one digital.
By Burkhard Kainka (Germany) actual sensor in a voltage divider that this time is preset with a
25-turn precision trimmer pot. A second voltage divider made
up from two 100 kΩ resistors provides a reference voltage of
A glance at the sensor combo kit (available from Elektor [1]) 2.5 V for the comparator. The temperature sensor (Digital Temp)
reveals seven superficially identical PCBs equipped with differ- can now be adjusted so that the sensor voltage at the desired
ing sensors. All of them have one analog output AO and one temperature will likewise be exactly 2.5 V. A rising tempera-
digital output DO. As you know, to create a digital signal from ture at the NTC sensor makes the comparator switch on the
an analog one, we use a comparator. There is one of these digital output DO, whilst a falling temperature switches it off.
included on each of these boards, or more precisely an LM393 The second comparator connected downstream operates only
dual differential comparator. the status LED. Thanks to these LEDs we can test all seven
sensors without any need for software.
Sensors equipped with comparators
When we examine the circuit of the board in Figure 1 more The circuit does not include a bypass capacitor (see inset),
closely, its function becomes clear. Once more we find the meaning that any variations or spikes on the supply voltage

learn design share
may affect the operation. The length of the wiring and other
random factors can also have some influence. In actual fact with
the temperature sensor and the heat rising, we find a narrow
region in which the output oscillates, which is possibly caused
by parasitic capacitance on the PCB. If we fit a small capaci-
10k
100k
tor between the input and the second comparator output, we
100k DO
A1
½ LM393 can see an oscillator in action. This is easy to recognize with
½ LM393 the oscilloscope. The status LED on the sensor PCB shows it
1k
1k
+5V
as well. When things warm up slowly, it lights up initially at
half brightness (state of oscillation) and only after this does it
1k5
G
GND
reach full brilliance (stable condition). If the software behaves
AO
A2 strangely, bear this point in mind.
100k
-T NTC The other sensors using this circuit behave in more or less
the same way, even if the changes in the measured value are
Digital Temperature usually not so gradual. Most comparable with the NTC sensor
+5V is the phototransistor in the flame sensor. The dark housing
allows the longer wavelengths to pass through preferentially, in
order for flames to be detected. The Hall sensor differs in that
S N it expects an operating voltage on its third pin. The fact that
the reed switch is connected as an analog sensor may surprise
MPSA13
you. But there are two out-of-phase outputs. When a magnet
Linear Hall Reed Switch Sound Touch Flame is brought closer, one output goes on and the other goes off.
Sound and touch behave somewhat differently; normally they
deliver a squarewave signal with rapid transitions. The two
sound sensors employ one large and one small electret micro-
phone. With the pot adjusted correctly, half waves of loud sound
Figure 1. Sensors with comparators. signals appear at the output as square waves. This must be
Bypass capacitors
In many circuits we find capacitors wired between the supply

voltage (VCC) and ground (GND). They improve the stability
10k
100k
of the circuitry and help prevent radio interference. Whenever 100k DO

A1
electronic circuitry uses a lengthy cable to the power supply ½ LM393
½ LM393
it’s not just the resistance of the wires in the cable that
1k
1k
+5V
comes into play but also the inductivity. A piece of twin- Bypass
10n
1k5
G
conductor cable 1 m long can have an inductance of around GND
0.5 µH, according to the gauge (thickness) of the wires and AO

A2
the distance between them. From this arises an inductive
resistance of 3 Ω at 1 MHz or 30 Ω at 10 MHz. If your circuit
100k
-T NTC
has a varying current load, effectively you have an alternating
current in the supply lead and with this some voltage
Digital Temperature
drop. Typical problems with voltage drop of this kind are
susceptibility to radio interference and vulnerability to failures
in the passive elements of the circuit. The long (from a radio
perspective) cable can also act as an antenna. Then you have between GND and VCC, as otherwise they would not pass
radio-frequency (RF) signals radiating that potentially may their test for electromagnetic compatibility. In this case they
exceed the permitted limits. Conversely pulses of interference are usually called decoupling capacitors, because any radio-
may induce brief fluctuations in the supply voltage that might frequency currents can be diverted or decoupled by taking a
interfere with the correct functioning of a circuit. shortcut through this capacitor mostly commonly of 100 nF.
The older expression ‘blocking capacitor’ used for these
DC motors will produce a varying load on the current source components has fallen out of fashion.
and can lead to radio interference. This is why capacitors are
connected direct to the supply terminals and are generally Our sensors are all definitely not particularly susceptible to
called anti-interference or suppressor capacitors. The DC interference problems of this kind. If you do wish to employ a
motor in an RC model cannot function without these, as cable longer than 12 inches (30 cms), you should connect a
otherwise it would interfere with its own receiver onboard. bypass capacitor of 100 nF direct to the sensor between the
Microcontrollers like the Arduino also require a capacitor supply terminals (see image).

Software-based
Schmitt trigger
+5V
+5V When you have a transition
region a comparator can gen-
10k
10k
10k
erate highly volatile output
SMD RGB LED
DO DO DO
states. A Schmitt trigger helps
R
DO
91R
mitigate this by displacing
G Laser
the transition points that set
Komparator
B the switch-on and switch-off
points. An example is switch-
G
G G G ing on at 25 degrees Celsius
Sensor 1 Sensor 2 Sensor 3 and off at 20 C, as discussed
regarding the NTC sensor
(Analog Temp) in first part of
Figure 2. Direct connection to RGB LED. Figure 3. The laser on the
this series.
comparator output.
All the sensors on the compar-

taken into account during your evaluation testing. In the same ator board also have an analog output, as you know. There is
way the touch sensor generally provides a 50 Hz squarewave nothing to stop you making a comparison in the software, and
signal (60 Hz in North America) when activated. in so doing, building in some appropriate hysteresis. Because
The comparator circuit delivers a sharp switchover point, with- all the sensors are adjusted using the trimpot to a switching
out any hysteresis. This means that with a slow change of point of 2.5 V, the same software can be used for the differ-
temperature, a region can occur in which the output flutters ing sensors.
a bit. In a program you can take account of this and poll the
digital output only at longer intervals. Or you could use the
analog output signal AO direct from the voltage divider and Listing 1. A comparator with hysteresis
evaluate this, something that we’ll describe in greater detail (Comparator.bas).
below. Unlike the analog NTC sensor already described, you Do
will then have the advantage that a desired temperature can D = Getadc(2)
be set externally using the pot. If D > 514 Then Portb.2 = 0
If D > 514 Then Portb.5 = 1
It’s also interesting to note that the digital output of this board
If D < 510 Then Portb.2 = 1
can control an actuator directly, without any assistance from
If D < 510 Then Portb.5 = 0
a microcontroller. Output DO is connected to the input of the
...
relay PCB. That’s it — the temperature controller or thermo-
stat is complete. Admittedly without hysteresis, which in some Waitms 500
applications could be problematic. If you feel like connecting Loop
one of the LEDs directly to the comparator, you need to bear
in mind that a comparator has an open collector output. There-
Listing 2. The Arduino comparator.
fore it switches down actively, so that in the High state only
the pull-up of 10 kΩ supplies current, meaning that the LED //Comparator AD1
does not illuminate very brightly. Nevertheless, applications are ...
conceivable in which three separate sensors directly drive the void loop() {
three colors of an RGB LED (Figure 2), not with great intensity value = analogRead(sensorPin);
but clearly visible nevertheless. The resulting color mixture of
if (value > 514) {
pink perhaps then stands for “hot and loud”.
digitalWrite (output1, 1);
digitalWrite (output2, 0);
If you wish to control laser lights directly, you’ll need to hook
these up between DO and +5 V (Figure 3). On/Off reverses the }
situation, but you can now switch larger currents up to about if (value < 510) {
20 mA. With this arrangement an interesting experiment can digitalWrite (output1, 0);
be carried out: opto-thermal feedback. The laser is pointed digitalWrite (output2, 1);
precisely towards the NTC sensor (see header photograph). }
The switching temperature is then set exactly at the change- Serial.println(value);
over point using the potentiometer. Once the correct position lcd.setCursor(0, 0);
is found, the laser starts to flash. Here’s how: when switched lcd.print(value);
on, it warms the sensor slightly. The digital output switches on, lcd.print (" ");
which turns off the laser because of the inverted connection.
delay(50);
This gives the sensor the chance to cool down again until the
}
comparator flips and the sequence begins once more.

learn design share
from the A-D converter (as always, all of the code samples can
be downloaded from the Elektor Magazine website [3]). The
center of the measurement range represents a value of 512.
The switching threshold points in this example lie at 509 and
515. With around 5 mV per A-D step we then have a hysteresis
of 30 mV. The program also controls LED2 on the Extension
Shield. This means you can compare the switchover with that
on the sensor board yourself. The software offers two slightly
displaced switchover points as opposed to the sharply defined
switching point of the sensor-comparator. The complete pro-
gram additionally displays the analog sensor voltage, both on
the LCD of the Shield and on the terminal screen. This helps
you find the correct trimmer setting.
Figure 4. Microphone signals.

The Arduino version of the program (Listing 2) differs little
from the BASIC example. The analog signals can be displayed
without additional overheads using a serial plotter. Figure 4
In the software the two outputs B.2 (LED 2 on the Elektor Exten- shows signals from the microphone sensor. This program can
sion Shield, see Part 1 [2]) and B.5 (LED on the Arduino board) be used again for all seven sensors on the red comparator PCB.
are used as outputs. There’s a specific reason for switching
them in antiphase. You might connect the dual-color LED to both Polling contact sensors
outputs. Or you might hook up actuators of this kind between The touch sensor does not produce a slowly varying signal, but
both outputs, which in reality still require a series (dropper) rather a (normally) squarewave signal of 50 or 60 Hz. Were you
resistor but in this situation would at least see two internal to use this for controlling a relay, the result would be wavering,
resistances in series. This class of actuators also includes the noisy and inelegant. However, the signal can be improved in
SMD RGB LED and the infrared LED. The color-changing LED software (Listing 3 and Listing 4). The digital output in this
discussed later on belongs in this category too, but it includes case is connected to AD1. Yes, this is an analog input but you
a reverse-voltage protection diode, which will not tolerate the can also use it as a digital input Port. The result of polling it
inverted voltage level cheerfully. In this case you are better will be 1 or 0 and can be copied direct to an output Port. Here
off using a genuine ‘real’ dropper resistor. we chose B2 (Arduino pin 10), because it can also control the
The sample Bascom code in Listing 1 evaluates the raw data LED2 on the Shield. Additionally B5 is also controlled, as that’s
Oscillators
One of Murphy’s Laws states that when input voltage will make the output voltage
you construct an oscillator it never drop. All the same, you can build an
oscillates, whereas when you make oscillator with only one transistor. All
an amplifier it oscillates all the time. you need do is ensure that the phase is
There’s something in this observation, reversed accordingly. You can do this, for
which is why it’s always worth looking instance, using several capacitors and
closely when your circuit turns into an resistors (phase-shift oscillator). Or use
oscillator and ‘starts to hoot’ as people 0V7...5V a transformer for the feedback (Meissner
say. An oscillator consists fundamentally oscillator). If the oscillator still observes
of an amplifier and some matching Murphy’s Law and doesn’t oscillate, all
feedback from the output to the input. you normally need to do is reverse one
The signal fed back must moreover have of the two windings to get the phase
the correct phase relationship. If the right. So it’s likely that the small buzzer
voltage at the input rises directly, the is constructed in this way (image). The
same should happen at the output, only ‘loudspeaker’ has two windings, which at
correspondingly amplified. Then it’s sufficient to feed back a the same time make a transformer.
part of the output signal to the input using a capacitor — and
now you have an oscillator. If you construct an amplifier with multiple stages and higher
overall gain, it may actually be not that easy to prevent self-
An amplifier of this kind can be constructed with two oscillation. Signals from the output can sneak back to the
transistors, in which both of them rotate the phase by input via the supply voltage, which you may be able to avoid
180 degrees. Or you can use an op-amp or an integrated using a fairly large bypass capacitor. Or some small capacity
loudspeaker amplifier. may exist between the output and input wiring. In extreme
A single amplifier stage with one transistor in grounded cases the only remedy that can still solve this problem is a
emitter mode turns the phase through 180 degrees. Rising physical screening (shielding) plate.

Processing switching signals

+5V +5V Up to now all we have done with our touch sensors is achieve
flashing on the output. In reality we’d expect a contact switch
10k
to do rather more, namely switching something on. This is
entirely feasible if we write our program somewhat differently.
AC AC
10p 10p
To manage this, all we need do is make sure that the switch-
CMOS-Input
ing-on process takes precedence and switching-off takes place
after a short delay. Our Touch programs (Listing 5 and List-
ing 6) additionally employ a delay period of 10 µs and conse-
GND GND quently interrogate the Port very frequently. Once a High state
Figure 5. Contact sensor with Figure 6. Input Port used as contact Listing 3. Input and output Port in Bascom.
Darlington transistor. sensor.
Dim D As Boolean
Config Portb = Output
where the Arduino-internal LED is located. Externally you can
also hook up another LED or the laser light. Do
The program consists of a simple loop that has a waiting time Portb.2 = Pinc.1
of 21 ms built in. If you now tap the touch sensor, the output D = Pinc.1 Xor 1
flashes. What actually happens is that touching it applies a 50
Portb.5 = D
Hz interference signal to the base of the Darlington transis-
Waitms 21
tor. Therefore we get a 50 Hz squarewave signal on the digital
Loop
output, i.e. a cycle duration of 20 ms. If we now sample this
signal using a slightly different frequency, you obtain a signifi-
cantly smaller frequency. The cycle duration 21 ms indicates
Listing 4. Input and output Port using Arduino-C.
a sampling frequency of around 48 Hz. The difference 50 Hz
– 48 Hz = 2 Hz appears in antiphase at outputs B2 and B5. //Touch1 A2 > 10, 13
It may not be self-evident that we can control a transistor in #include <LiquidCrystal.h>
this manner. Usually we specify a base current and determine int input = A2;
the bias point in this way. With small signals, things would not int output1 = 10;
work without base current. The person touching the sensor is
int output2 = 13;
seen as one plate of a capacitor, with the second plate formed
by all the wires and cables located in the environment. A base
void setup() {
current would charge this capacitor so far negative that the
pinMode(output1, OUTPUT);
transistor would block completely. It works only because every
bipolar transistor contains a zener diode in the range 7 V to pinMode(output2, OUTPUT);
10 V between base and emitter (see Figure 5). The idle state }
mains hum voltage of a person is generally significantly higher void loop() {
than 10 Vpp. Therefore an alternating current flows, with its digitalWrite (output1, digitalRead(input));
positive half-wave driving the transistor. digitalWrite (output2, 1-digitalRead(input));
delay(21);
You might expect that on the collector of the Darlington tran- }
sistor we would also find an approximately equal squarewave
signal. So let’s also try coupling the analog output AO to the
digital input A1. That’s right; now the output flashes when the Listing 5. Contact sensor and touch switch
base is touched. However, the difference is that this time the (Touch2.bas).
output in idle mode is ‘on’, because the comparator on the
Dim T As Word
sensor board inverts the signal.
It even works without the sensor board. At input A1 we have Config Portb = Output
just a piece of insulated wire connected (Figure 6). If you
then touch the outside of this insulation, it still flashes! This Do
happens because the input of the controller has extremely high If Pinc.1 = 1 Then T = 50000
impedance. The wire, insulation and finger form a small cou- 'If Pinc.1 = 0 Then T = 50000
pling capacitor of a few picofarads in value. If an AC voltage If T > 0 Then T = T - 1
is applied to the input, it is limited to the input voltage range If T > 0 Then Portb.2 = 1 Else Portb.2 = 0
by the internal protection diodes. The open input does admit- If T > 0 Then Portb.5 = 0 Else Portb.5 = 1
tedly have a disadvantage compared to the correct touch sen-
Waitus 10
sor. The idle state is not unambiguous, you see, which makes
Loop
evaluation tricky.

learn design share
is detected, the program sets counter T to a value of 50000 two sound sensors in the Sensor Kit, with which, according to
and switches on the output. The counter decrements slowly the sound frequency being sampled, up to 10 kHz or more can
and only after about 500 ms without any further pulse does the appear on the output. Correct adjustment of the trimpot will
output switch off. Any further impulse that occurs will reset the give you a very usable speech switch that will also react to loud
counter High again and the timeout period is extended accord- whistling, hooting or other noises. Even gentle tapping your
ingly. Now everything functions as desired. The output goes on finger on the microphone will set off the switch. The sensitiv-
immediately if someone touches the sensor and goes off again, ity depends very much on the precise setting of the pot. You
not when the contact ceases but after some delay. The func- need to keep in mind that this type of microphone normally
tion corresponds to a retriggerable monostable multivibrator. delivers less than 1 mV. This means you must maintain the
Now our touch sensor, with its Darlington transistor, is working sensor voltage to the switching point within a few millivolts of
the way we want it to. The squarewave signal on the digital accuracy in order for sound signals to make the comparator
output has been transformed into an unambiguous switching change level correctly.
signal that will work even in environments without mains hum.
Short pulses caused by static charge are sufficient to trigger Shock sensor
the output. Certain other digital sensors, such as the Tap Module, are
appropriate for this kind of sampling. This contains a spring
There’s a special reason why we chose the loop delay of 10 µs that can close a contact when disturbed. Two very brief pulses
to be much smaller than necessary for a 50 Hz signal. The for- are produced in this process. You need to be aware that the
mat of the program makes it equally suitable for use with the module has a pull-up resistor, meaning that the output volt-
age is +5 V when not operated and the pulses are 0 V. In this
situation the sampling process needs to invert the input sig-
nal, which is marked in the software with a commented line.
Listing 6. Touch switch in Arduino-C.
Alternatively in this case you might also connect +5 V and GND
void loop() { transposed and leave the software unaltered, as the polarity
if (digitalRead(input) == 1) timeout = 50000; is of no interest to the switch and resistor.
//if (digitalRead(input) == 0) timeout = 50000;
if (timeout > 0) timeout = timeout -1; The Shock Sensor was tested in the same way, in which an
if (timeout > 0) { internal contact is closed for a short duration each time it is
digitalWrite (output1 , 1); disturbed. The sensitivity of the Shock Sensor is greater than
digitalWrite (output2 , 0); that of the Tap Module. The Tilt Switch (ball switch) can also
be sampled in this manner. At the right degree of inclination
}
a small ball rolls downwards and closes two contacts. If you
else {
shake the sensor rapidly to and fro, you can hear the ball rat-
digitalWrite (output1 , 0);
tling and watch the program triggering the output.
} In point of fact all switching sensors can be polled or sam-
delayMicroseconds(10); pled meaningfully using this method, as well as press buttons
} (Button), reed switches (Mini Switch) and the reed switch on
the comparator board. Using this type of time delay we can
achieve effective debouncing of the switches. Just about every
mechanical switch (apart from mercury switches, which are
Listing 7. Additional serial output. now banned in many territories, however, on account of the
void loop() { poisonous mercury in them) bounces momentarily one or more
if (digitalRead(input) == 1) timeout = 50;
times when operated, generating several impulses to begin
with. The software turns this into a single, elongated pulse. If
if (timeout > 0) timeout = timeout -1;
you alter the period somewhat you can make a time switch,
if (timeout > 0) {
for instance to illuminate a staircase at night. The light can
then be activated, according to the type of sensor, by loud
digitalWrite (output2 , 0); speech, touch, tapping or advancing a magnet. Even the digital
} output of the temperature sensor could be polled in this way.
else { This would simultaneously eliminate the problem of fluttering
digitalWrite (output1 , 0); transitional states between Low and High.
} With a minor alteration (Listing 7) the function of a compara-
Serial. tor sensor could also be made clearer using the serial monitor.
println(analogRead(sensorPin)+100*digitalRead For this we additionally make use of the analog signal at AD2,
(output1));
while the digital output signal of the comparator carries on with
the task of triggering the actual switching process. To display
delay (20);
the output signal with only one channel, we raise the output in
}
a High state by 100. You can then observe the original analog

signal as if raised on a pedestal. Figure 7 shows the result

for the sound sensor. You can see clearly that a sound signal
must first exceed a certain level to switch the output state.
At the end of all of the signals the software enables the laid
down delay period, after which the output drops back down.
Figure 8 portrays the work of the temperature sensor. For
this the program is slowed down to an extent by a delay of
100 ms. During the measurement time-window the sensor was
twice warmed by touching with a finger. You can spot clearly
the inversion performed by the comparator: a falling voltage
on the actual sensor switches on the comparator’s output.
Equally easy to recognize is a brief phase of oscillation in the
region around the switchover point. The software suppresses
these oscillations effectively.
Figure 7. Switching using the sound sensor.
Buzzers and other actuators

At output B2 we can again attach the relay board. But there
are yet more actuators that we can also put to use. Among
these belongs the buzzer, a small active device that is pro-
vided initially with some protective foil. At 5 V the current
consumption amounts to 25 mA. The buzzer can therefore
be attached direct to a Port. The polarity is admittedly not
completely unambiguous and depends on how the buzzer is
soldered onto the board. On the sample device the signal and
minus connections were transposed in fact and the S-pin need
to be connected to GND. A test using the lab power supply
also indicated that the buzzer begins to work already at 0.7 V
and at a different frequency if the sound opening is blocked.
Without even opening the housing, our experts were able to
figure what lay inside: a free-running oscillator with a bipolar
silicon transistor with the normal threshold voltage of 0.5 V
Figure 8. Temperature sensor in action.
to 0.7 V (naturally it insists on having the correct polarity of
supply voltage). At the same time take care that you do not
confuse the active buzzer with the passive one, which is more
comparable to a small 16-Ω loudspeaker.
Another interesting actuator is the color-changing LED (Color Controller

NC
B
Flash), in actual fact a three-color LED package with a built-in
controller. This automatic LED produces a color transition G GND (100R)
10k
12 (GND)
between red, green and blue. This is another case requiring
R
a dropper resistor, since color-change LEDs of this kind are in V
fact designed for 3 V. Incidentally, although there is indeed 10
a resistor of 10 kΩ on the board, it is connected in parallel

(Figure 9). For 5 V a dropper resistor of 100 Ω would work Color Flasher
best. Here again, however, the compromise presented in the

previous article works again: you can connect the LED between
Figure 9. Color-changing LED.
two Port pins, in order to achieve a degree of current limita-
tion using the internal resistance of the Ports. Pin 12 (B4) is
recommended for the opposite pole.
Incidentally, the internal controller of the color-change LED, like Web Links
most ICs, includes an inverse polarity-protection diode on its [1] www.elektor.com/arduino-sensor-kit
supply voltage connections and therefore reacts sulkily if you
reverse the supply connections. Without the dropper resistor
a reversed power connection could lead to destruction. This [3] www.elektormagazine.com/160173
is vitally important: GND lies in the middle of the three con-
nections. By the way, just for fun, you can also connect the
passive mini-speaker in series. Then you can not only see the
switching process but also hear it.
(160173)

learn design share
MAXREFDES99#
Display Driver Shield
256 LEDs at your command
The American Standard Code for Information

Interchange, first published in 1963 and from
then on known as ASCII, was based on the
English alphabet whose characters can be
displayed nicely with as few as 5x7 pixels
or even less. Today, electronic signs must
show anything from Arabic to Zulu and
more pixels per character are required.
The MAXREFDES99# is an easy-to-use
256-pixel solution capable of handling
even the most demanding characters.
The Shield…
… RD99 for friends, is an Arduino compatible extension
board sporting a 16x16 LED array (actually four 8x8 LED arrays) Arduino sys-
and the necessary drivers for it. The 7.7 x 7.7 cm (3 x 3”) tems, indicates the I/O voltage level
square board is intended for signage applications and its 256 of the host system. A level shifter automatically adapts
LEDs probably allow displaying any international alphanumeric to this voltage making the shield suitable for systems with I/O
character that you can think of. It can also display scrolling signal levels as low as 1.2 volts.
messages in a 5x7 font, or some other font. What exactly it Communication with the host system requires only three sig-
shows is up to you. nals: clock (rising edge on ‘D13’, H4, pin 5), data (‘D11’, H4,
The shield is supported by open-source software libraries (C++) pin 7) and a load signal (rising edge on ‘D10’, H4, pin 8).
for both the Arduino and ARM mbed rapid development plat- These signals are compatible with 16-bit SPI busses where
forms. The libraries not only take care of the low-level commu- the Slave Select (SS) signal can be used as the load signal.
nication with the display, they also offer high-level functions Add to these signals a reset line (active low, H3, pin 3), the
for displaying all printable ASCII characters from ‘space’ (32) IOREF voltage and power, and you know that the all singing,
to ‘~’ (126) and for scrolling text messages. all dancing interface needs eight connections.
The shield must be powered from 5 VDC, supplied by either the
host system or the on-board power supply, depending on the The MAX7219
position of switch SW1. The input voltage for the on-board The RD99 has four MAX7219 LED display drivers, one for each
regulator is limited by the maximum voltage allowed on the VIN 8x8 LED matrix. Each part is capable of driving up to eight
pin (connector H3, pin 8) of the host system or by the voltage 7-segment displays (with decimal dot) or up to 64 individual
rating of C1 (25 V), whichever is lowest. For an Arduino Uno LEDs. The MAX7219 includes a BCD code-B decoder, multiplex
R3 host this is 20 VDC, although it is recommended limiting it scan circuitry, segment and digit drivers, and an 8x8 bit static
at 12 VDC. The kit comes with a 9 VDC 1.3 A wall-mount power RAM to store the value for each LED. A serial interface allows
adapter with American-style pins, which is a bit of a shame programming the device’s registers and, of course, activating
since it accepts anything from 100 to 240 VAC as input. or deactivating individual LEDs. Individual digits (or groups of
The RD99 monitors the IOREF pin (H3, pin 2), the pin that, in eight LEDs) may be addressed and updated without rewriting

the entire display. The integrated BCD decoder may save some
work upstream but can be deactivated when it is not needed.
The brightness of all LEDs is controlled by a single external
resistor that sets the current flowing through them; four of
these control the four 8x8 LED matrices. Display brightness
can also be set digitally (PWM) by using the intensity register.
A convenient 4-wire serial interface permits easy connection to
a microcontroller. The serial protocol is very similar to 16-bit
SPI. If the microcontroller’s SPI peripheral is not capable of
controlling the MAX7219 (because it is 8-bit only) then it is
easy enough to bit-bang the data into the device as there are
no strict timing constraints. The only thing to keep in mind
is that data always passes through the chip’s internal shift
register no matter what the state of the LOAD pin is. A rising
edge on this pin will load the data from the shift register into
internal memory and update the display, so be careful to load
at the right moment.
Using It With Arduino

In what follows it is supposed that a working Arduino devel-
opment environment is available, including an Arduino Uno
R3 board. Before plugging the RD99 on the Uno, check if the Figure 1. Backside of the shield
shield is working. To do this, connect the power adapter and
put SW1 in the position closest to the power input socket (this
will set it to external power). A power indicator LED should light a problem with the last character in the lower right corner, the
up (irrespective of the position of SW1). Disconnect the power ‘m’. It seems as if one pixel has jumped out of line, transform-
and plug the RD99 on the Arduino Uno (without changing the ing the ‘m’ into a strange character. Inspection of the character
position of SW1), reconnect the power adapter and connect ‘m’ in the file maxrefeds99.cpp (already open in the Arduino
the Uno to your computer. IDE) shows why (line 130): the sequence “0x58, 0x44” should
From [2], from the Design Resources tab, download the firm- read “0x78, 0x04”. (Can’t figure out why? Write out all the
ware for the Arduino Platform. Unpack this file to a folder numbers as binary sequences where a ‘1’ is a dot and a ‘0’ is
where you can find it, then copy the folder MAX7219 into the not.) Apply the correction, upload the program and run demo
folder libraries of the Arduino Sketchbook location (to find number 6 once more. All should be well now.
it, from the Arduino menu select File -> Preferences). If the
libraries folder does not exist, create it first. A Simple Animation
Launch the Arduino IDE and go to File -> Examples -> MAX7219 There is a second example named MAX7219_example. Open it,
and open MAXREFDES99_example. If you don’t see this exam- upload it to the board, see what it does, and study the source
ple, then the library is not installed correctly or the Arduino code. It is rather simple, but it shows clearly how to light indi-
IDE was running while you copied the library. If it was then vidual pixels and where they are located on the grid. Inspired
close and restart the IDE. Click the Upload button to load the by this example I created a little full-screen animation that you
program into the Uno. For me uploading only worked when can download from [3]. The animation consists of a sequence
the shield was powered from the power adapter. of 11 frames displayed one after the other with an inter-frame
When uploading is finished, open the Arduino IDE’s Serial delay of 25 ms. The last frame is shown for 1.5 seconds, then
Monitor and set the baud rate to 115,200 baud. A menu will the animation starts again. A second sketch shows a way of
appear. If it doesn’t, press the Reset button on the Uno. Type scrolling large custom characters. Videos of these sketches
‘6’ in the Send box and click the Send button. At first you may are available on [3].
think that nothing is going to happen, but after about 5 seconds Having studied the provided examples you should now be
your command is suddenly acknowledged, the display comes ready to create signage applications to your own specifica-
alive and you can admire a magnificent demo. tions. Have fun!
When the demo has terminated the attentive observer will notice (160267)
As an authorised distributor for Maxim, Mouser carries Web Links

more Maxim Integrated products than any other distributor.
[1] www.mouser.com/new/maxim-integrated/
If that weren’t enough, we’re adding new products daily
maxim-maxrefdes99
so you always have access to the newest products from
[2] www.maximintegrated.com/MAXREFDES99
Maxim. Plus, Mouser backs it all with same-day shipping,
fast and accurate order deliveries combined with quality [3] www.elektormagazine.com/labs/
service and support. maxrefdes99-led-shield-experiments-160267
Find the newest Maxim products on Mouser.com

learn design share
AxiCat
A versatile USB development tool
By Ilse Joostens and Peter S’heeren (Belgium)
If you are developing an app for an

integrated circuit and you want a
fast and easy way to link it to a
Linux or Windows PC, we have
the answer: AxiCat.
an actual design. That’s easy jumper JP1. For that reason we opted for
with the AxiCat multi-protocol USB a 12 MHz clock crystal instead of a higher
adapter, and you don’t have to write any frequency, to ensure that the microcon-
code to use it. troller remains within its safe operating
Nowadays there area with a 3.3 V supply voltage.
are thousands of useful Schematic and structure The USB connection is handled by IC2,
ICs available to electronics hobbyists and The key component of the circuit, an FT245RL from FTDI. Unlike the bet-
professionals. A lot of them only come in as shown in Figure 1, is an Atmel ter known FT232RL, this device does
SMD packages, which are not very con- ATmega164A microcontroller (IC1). You not have a serial interface, but instead
venient for hobbyists, but don’t worry: can also mount other members of the an internal FIFO with an 8-bit data bus.
many well-known (and less well-known) same microcontroller family, such as the The advantage of this is that with Full
vendors who focus on the hobby sec- ATmega324A, 644A and 1284. The only Speed USB and a 12 MHz clock frequency,
tor offer a broad spectrum of breakout difference is the amount of internal mem- you can achieve a respectable data rate
boards. Most of these ICs have an I2C ory. The AxiCat normally comes with an between the USB bus and the microcon-
or SPI interface, and it would be handy ATmega164A or an ATmega324A, depend- troller (IC1).
if you could quickly connect them to a ing on availability. Thanks to the underlying layers of the
Linux or Windows PC to check out their The microcontroller is powered from 5 V USB communication protocol between
functions before you start using them in or 3.3 V, depending on the position of the driver in the host system and the
FT245RL, the data arrives in the FIFO of
the FT245RL without any bit errors. This
Features means that the data read by the micro-
controller is error-free, and the data out-
• Based on a powerful ATmega164A microcontroller clocked at 12 MHz
put by the microcontroller also arrives in
• Compatible with Windows and Linux
the host system free of errors, eliminating
• Selectable 3.3 V and 5 V logic levels
the need for error correction in the com-
• 17 bidirectional I/O lines with configurable pull-up
munication protocol. That also pays divi-
• I2C bus with pull-up resistors, master and slave mode; maximum speed 400 kHz
dends in terms of speed. IC3, a USB6B1,
• SPI master with four slave select lines; maximum speed 6 MHz
protects the bus lines against ESD, mak-
• Two independent serial ports
ing the board significantly more robust.
• 1-Wire master with strong pull-up option and built-in accelerators for
The 3.3 V supply voltage is provided by
enumeration
IC4, an LM1117-3.3 low-drop voltage reg-
• Powered over the USB bus
ulator. This voltage is available on GPIO
• Asynchronous design – all interfaces can be used simultaneously
connector K2, and along with the 5 V sup-
• Can be used as an in-system programmer for AVR microcontrollers
ply voltage from the USB port it can be
• Boot loader for easy firmware upgrade and loading your own software
used as desired to power the microcon-
• GPIO connector pinout compatible with Raspberry Pi A and B
troller and the I/O portion of IC2.

Polyfuse F1, with a rated value of 350 mA, power to the Raspberry Pi from also feed- connected to the AxiCat. However, exten-
limits the total current consumption to ing power to the AxiCat and through the sion cards for the newer Raspberry Pi
prevent overloading of the supply voltage USB port to the computer. models with additional I/O lines on the
lines of the USB port. Specifically, this GPIO connector K2 is a 26-pin header. 40-pin connector (A+, B+, 2 and 3) are
means you should keep the total current This connector and its pinout (see Fig- not compatible.
drawn from the GPIO connector (I/O lines, ure 2) were not chosen randomly; they Finally there is jumper/connector K3,
5 V supply voltage and 3.3 V supply volt- largely correspond to the GPIO connector which can be used to manually reset the
age) below 350 mA. on the early Raspberry Pi models (A and AxiCat. If jumper JP2 is kept closed at
Diode D1 in the 5 V supply line to pins B). The main advantage of this is that the same time, the internal boot loader
2 and 4 of GPIO connector K2 is there a large number of extension cards and is activated. This allows you to send your
to prevent extension cards that supply HATs for the Raspberry Pi can also be own firmware to the AxiCat.
+UB +UB
C7 C8 C9 C10
+5V +UB +UB
R5
100n 100n 100n 100n
10k
R4
C11 C6
10k
F1
5 17 38 29 27
1u 100n
VCC VCC VCC AREF AVCC
4
20 4 RESET
350mA
VCC VCCIO IC1 R1
8 1 37 19 PC0
NC D0 PA0(PCINT0/ADC0) PC0(PCINT16/SCL)
100R
19 5 36 20 PC1
RESET D1 PA1(PCINT1/ADC1) PC1(PCINT17/SDA)
3 35 21 PC2 C5
D2 PA2(PCINT2/ADC2) PC2(PCINT18/TCK)
27 11 34 22 PC3
OSCI D3 PA3(PCINT3/ADC3) PC3(PCINT19/TMS) 1
K1 IC3 28
IC2 2 33 23 PC4 10n
L1
OSCO D4 PA4(PCINT4/ADC4) PC4(PCINT20/TDO) K3
1 1 8 9 32 24 PC5
VBUS VCC D5 PA5(PCINT5/ADC5) PC5(PCINT21/TDI)
2 2 7 15 10 31 25
D– IO1 USBDP D6 PA6(PCINT6/ADC6) PC6(PCINT22/TOSC1)
3 3 6 16 6 30 26 PC7
USB
D+ IO2 USBDM D7 PA7(PCINT7/ADC7) PC7(PCINT23/TOSC2)

4 4 5
ID GND FT245RL ATMEGA164A
5 24 12 PC7 PB0 40 9 PD0
GND NC PWEN PB0(PCINT8/XCK0/T0) PD0(PCINT24/RXD0)
USB6B1 14 PB3 PB1 41 10 PD1
WR PB1(PCINT9/CLKO/T1) PD1(PCINT25/TXD0)
17 13 PB2 PB2 42 11 PD2
3V3OUT RD PB2(PCINT10/INT2/AIN0) PD2(PCINT26/INT0)
22 PB1 PB3 43 12 PD3
TXE PB3(PCINT11/OC0A/AIN1) PD3(PCINT27/INT1)
R6 26 23 PB0 PB4 44 13 PD4
C3 C4 C12 TEST RXF PB4(PCINT12/OC0B/SS) PD4(PCINT28/OC1B)
AGND
PB5 1 14 PD5
GND
GND
GND
1M
PB5(PCINT13/MOSI) PD5(PCINT29/OC1A)
4n7 10n PB6 2 15 PD6
1u 7 18 21 25 PB6(PCINT14/MISO) PD6(PCINT30/OC2B/ICP)
PB7 3 16
PB7(PCINT15/SCK) PD7(PCINT31/OC2A)
GND
GND
GND
GND
XTAL2 XTAL1
6 18 7 8 28 39
X1
JP2
IC4
LM1117-3.3 1
+5V +3V3 C1 12MHz C2
ADJ 22p 22p
C15 C13 C14 C16
10u 1u 1u 22u
+UB +3V3 +5V
D1
R2 R3
K2
4k7
4k7
1N5819
1 2
+5V +UB +3V3
PC1 3 4
PC0 5 6
PC5 7 8 PD1
3 2 1 9 10 PD0
PC4 11 12 PD5
JP1 PC3 13 14
PC2 15 16 PD3
17 18 PD2
PB5 19 20
PB6 21 22 PD4
PB7 23 24 PB4
25 26 PD6
Figure 1. Schematic diagram of the AxiCat, which is built around an
Atmega164A microcontroller.
150585 - 11

learn design share
that involve bus transactions, such as ciency of the interrupts. Instead of inter-
an SPI transfer or a 1-Wire enumera- rupt routines, we use non-blocking tasks
tion. To avoid blocking of the incoming integrated into the main loop. TIMER1,
flow of data packets, the AxiCat executes which acts as a millisecond counter, is
long-duration commands asynchronously, the only hardware function that works
which means in the background. With with an interrupt routine.
asynchronous commands, the AxiCat We chose the size of the working buf-
waits until the command has been exe- fers based on the available RAM and the
cuted before it returns a response to the code generated by the compiler. The
host. main program uses working buffers with
Asynchronous commands are stored in a size of 2n bytes and a maximum size
working buffers, which are available for of 128 bytes, which results in very effi-
Figure 2. The header pinout is designed to be
the following components: I2C master, cient and compact code for an AVR 8-bit
compatible with Raspberry Pi models A and B.
This allows a large number of Raspberry Pi HATs I2C slave Tx, I2C slave Rx, SPI master, microcontroller.
to be connected directly to the AxiCat. 1-Wire master, UART #0 Tx, UART #0 Rx, The main program supports USB remote
UART #1 Tx, and UART #1 Rx. Note that wakeup, so the AxiCat can wake the host
the UART working buffers do not retain computer when it is in sleep mode. Of
any commands; they only send data course, this is only possible if the oper-
bytes (Tx) or receive data bytes (Rx). ating system of the host also supports
K2 The working buffers enable the AxiCat to USB remote wakeup.
3.3V 1 perform bus transactions on several dif- AxiCat Bootloader is executed when the
ferent interfaces in parallel. The effective AxiCat starts up. The boot loader first
ground size of the working buffers can be con- checks the state of jumper JP2. If the
GPIO0 figured in the firmware and is primarily jumper is open, the boot loader jumps
VCC determined by how much RAM is available to the main program. If the jumper is
GND in the microcontroller. closed, the boot loader remains active
RESET and you can use the boot loader pro-
There are also autonomous response gram aximcubldr to download a new main
MOSI
ATtiny45 packets, which are called unsolicited program from the host computer to the
MISO
SCK
responses. For example, there is no com- AxiCat.
mand for reading received data bytes
from a UART. This means that the AxiCat Software
automatically sends a response packet The AxiCat Server program provides the
when a UART receives data bytes. functionality of the AxiCat via network
Figure 3. Using the AxiCat as a programmer
ports and standard I/O in the form of
connected to an ATtiny45.
Firmware a communication protocol. The server
The main program of the AxiCat is AxiCat can accept multiple client connections,
Main, which is what makes everything allowing several clients to work with the
Serial data protocol tick. The serial data protocol is imple- AxiCat concurrently. The server acts as a
The serial data protocol consists of byte mented here. The main program is written multiplexer between several clients and
packets which flow between the host and in C; we used WinAVR 20100110 for this. a single AxiCat, which creates a lot of
the AxiCat in full duplex mode. The host WinAVR is a Windows version of GCC for flexibility.
sends command packets, and the Axi- AVR 8-bit microprocessors and AVR Libc. The communication protocol consists of
Cat sends response packets. The file axi- During the development of the AxiCat we commands and corresponding responses.
cat.h, included in the download package frequently checked the generated assem- A client sends a command, and the server
from [1], contains the definitions in the bler code, and in many cases we rewrote returns a response once the command
C language. the source code to make the machine has been executed. The communication
There are two types of commands: syn- code more efficient and compact. That protocol is coded in ASCII format. The
chronous and asynchronous. They differ was necessary for two reasons: firstly syntax is designed to make the commu-
in how the AxiCat executes them after because fast iteration of the main loop is nication protocol easy to use both man-
they are received. Synchronous com- crucial for the performance of the AxiCat, ually and in software.
mands are executed immediately. If a and secondly because the main program
response package is defined for the com- has to fit in the 16 KB of flash memory The server supports a large number of
mand, the AxiCat returns it to the host in the ATmega164A. commands for working with the AxiCat.
immediately. Among other things, syn- Some examples are:
chronous commands are used to read We opted to use interrupts as little as
status data, modify settings, and access possible, because the GCC compiler often Configure GPIO pin 1 as an output, set
GPIO lines. generates interrupt routines that push a the output low and read the state:
Some commands take a relatively long lot of registers onto the stack and later Command iod 1 out
time to execute. These are commands retrieve them, which degrades the effi- Response iod ok

Command iow 1 0 AxiCat AVR ISP converts the AxiCat into

Response iow ok an in-system programmer for 8-bit AVR Component List
Command ior 1 microcontrollers. It can program flash
Response ior 01 1 0 out memory, EEPROM and fuses, and it can Resistors
Default: SMD 1206
read Intel hex files and raw files. Axi-
R1 = 100Ω
PCF2129A real time clock: activate I C 2
Cat AVR ISP works with the SPI bus (but R2,R3 = 4.7kΩ
master, write register index 0, read reg- without slave select) and a reset line. You R4,R5 = 10kΩ
R6 = 1MΩ
isters 0–9: connect the SPI bus between the AxiCat
Command ime and the microcontroller, and any desired Capacitors
Response ime ok GPIO pin to the reset pin of the microcon- Default: SMD 1206
Command imw 81 0 troller. An example is shown in Figure 3. C1,C2 = 22pF
C3 = 4.7nF
Response imw 081 00001 ack C4,C5 = 10nF
Command imr 81 10 The AxiCat Command Line tool converts C6–C10 = 100nF
Response imr 081 08h 16h 04h 30h simple text commands into byte packets C11–C14 = 1µF
C15 = 10µF 10V, tantalum, case A
22h 16h 06h 00h 08h 16h nack and sends them directly to the AxiCat. C16 = 22µF 10V, tantalum, case A
This tool is mainly useful for developing
Activate SPI master, transfer four bytes and debugging the main program for the Inductors
and use slave select #2: AxiCat. You can use it to send invalid L1 = ferrite core, 3A (Farnell #1653393)
Command sme packets to the AxiCat, which is a good Semiconductors

Response sme ok way to test the robustness of the main D1 = 1N5819, SOD123
Command smt 2 0FFh 0FFh 10100b 5 program. It is also a handy tool for any- IC1 = ATmega164A-AU, programmed*
IC2 = FT245RL
Response smt 2 000 255 032 005 one who wants to study the serial data
IC3 = USB6B1
protocol. IC4 = LM1117IMP-3.3
Many responses from the server contain
numbers. The format of these numbers Summary Diversen:
F1 = polyfuse 350mA, 1206
is fully configurable – you can chose from We mainly use the AxiCat to test our own JP1 = 3-pin pinheader, right angled,
binary, decimal and hexadecimal, with or extension boards for the Raspberry Pi 0.1’’ pitch with jumper
without leading zeros. (Swiss Pi and PiWire+) and to install firm- K1 = mini-USB connector, type B
K2 = 26-pin (2x13) pinheader, 0.1’’
ware on these boards. We also use the pitch
The communication protocol is fully asyn- AxiCat in all of our software development, K3 = 2-pin pinheader, right angled, 0.1’’
chronous, which means that a client does because writing and debugging C code is pitch with jumper
X1 = 12MHz quartz crystal
not have to wait for the response to a a lot faster and easier on a PC than on a
command before sending the next com- Raspberry Pi. At the end of the develop- *see text
mand. This allows clients to send sev- ment process, we compile and test the
eral commands in direct succession, which software on a real Raspberry Pi.
dramatically increases communication The AxiCat is available from the Elek-
efficiency. Each command can optionally tor Store as a fully assembled module
be provided with an identification num- [2]. You can find a lot more background
ber, which is sent back with the response. information and all necessary software
For each of the two UARTs, the server at [3].
offers the option of transferring data (150585-I)
bytes directly between the UART and a
network port. This option allows you to
Web Links
use a terminal emulator program such
as PuTTY to read and write data directly [1] Software download:
through the UART. www.elektormagazine.com/150585
[2] AxiCat in the Elektor Store:
The AxiCat Application Layer module pro- www.elektor.com/150585
vides a C API on top of the serial data [3] More information and software:
protocol. The C API supports the full func- www.axiris.eu/en/index.php/i-o-cards/
tionality of the serial protocol, including axicat
asynchronous transfers, so that program-
mers can manage the various interfaces
of the AxiCat optimally and operate them
in parallel. The source code of the Axi-
Cat server provides good insight into
how to use the C API. This module us
used in a variety of programs, including
AxiCat Server, AxiCat AVR ISP, I/O Card
Explorer, and 1-Wire Automation Server.

learn design share
LED
Stairway
Lighting
One controller for a
whole bunch of LEDs
By Peter Grundmann (Germany)
If the Stairway to Heaven was fitted with lights I guess they would look something like this… Each step is
softly illuminated in sequence by a microcontroller, not just fascinating but energy efficient as well!
A friend was telling me how he had been case is seldom likely to have more than can be connected directly to the micro-
impressed by a cool stairway lighting sys- 20 steps. For inputs we need a pushbut- controller I/O pins. Looking at the sche-
tem he had seen in a high-tech house. A ton at the top and at the bottom of the matic in Figure 1 the only other items
pushbutton at the bottom of the staircase stairway to turn the lighting on and off. besides the controller and output buffer
starts a lighting sequence where each step This number of I/O lines could easily be chips are the two pushbuttons connected
beginning from the bottom is gradually lit taken care of by an ATmega8 control- via connector K2 and the power supply.
up until the topmost step is reached. At ler. Without the need of a serial port or The circuit does not contain any dan-
the top another pushbutton turns off the external crystal all the I/O pins are free gerous high voltage levels because it is
stair lighting, again in a sequence which to be used as required. powered by a mains adapter unit. The
begins from the bottom all the way up to Each step can be lit by a string of LEDs adapter should be able to supply an out-
the top step. Pressing the top pushbutton and where necessary two strings wired put voltage in the range of 12 to 20 V DC
again begins the lighting sequence but this in parallel to give more light. A standard and can, for example be salvaged from a
time from the top to the bottom. Hey, that microcontroller I/O will not be able to disused router. The processor runs with
sounds neat but after some research I supply the voltage or current required for a supply of 4 V so a simple linear voltage
found the cost of a commercially available each string so we will need some output regulator would need to dissipate con-
controller unit — at few hundred euros driver chips to drive the LEDs. siderable power. Despite the fact that
seemed a bit excessive. I have been work- for operation the controller only draws a
ing with Atmel microcontrollers for some This job can be performed by the ven- few milliamps, each active output adds
time now and wondered how simple it erable ULN2003 chip which comes in a around another milliamp so the power
would be to build my own version. DIP package and contains eight Darling- dissipated by a smaller outline L-type lin-
ton-transistor drivers able to switch up to ear regulator would quickly exceed its
First thoughts 50 V at 500 mA, much more than we need limit. We would therefore need to use a
To work out what we would need, we in this application. The inputs to each larger TO220-outline linear regulator or
can firstly assume that a domestic stair- buffer include a series resistor so they a more efficient switched-mode regulator

+4V1
K3
+12V K6
2 1 1
K4
4 3
to ATMEL 2 1
6 5
9
ISP
1
I1
CD+
O1
16 R8
47R *
+4V1
2
I2 O2
15 R9
47R *
+12V
3
I3 IC3 O3
14 R10
47R *
7 20
4
I4 O4
13 R11
47R *
C3
VCC AVCC
5
I5 O5
12 R12
47R *
R1 220n 1
PC5
28
27
6
7
I6
ULN2003
O6
11
10
R13
47R * 1
K5
C1 PC6 PC4 I7 O7
26 GND 2
PC3
1R
21 25 8 3
820u AREF PC2
IC2 PC1
24 4
6 23 5
L1 PC0
1 VCC 2 9 6
K1 SWC SWE
*
7 220uH 19 1 CD+ 16 R14 7
2 D2 IPK IC1 PB5 I1 O1 47R
1
8
DRC PB4
18 2
I2 O2
15 R15
47R * 8
1N4001 MC34063A R3 PB3

17 3
I3 IC4 O3
14 R16
47R * 9
12V
3
TC COMP
5
12K PB2
16 4
I4 O4
13 R17
47R * 10
GND
R2 PB1
15 5
I5 O5
12 R18
47R * 11
ULN2003
*
C2 4 D1 C6 2 6 11 R19 12
PD0 I6 O6 47R
*
5k1
3 7 10 R20 13
PD1 I7 O7 47R
470p 560u
1N5819 ATMEGA8-16PU GND 14
6 8 15
PD4
9 16
PB6
10 17
R6 R4 PB7
K2 4 11 9 18
5k6 5k6 PD2 PD5
3
R7 R5 PD6
12 1
I1
CD+
O1
16 R21
47R * 19
2
5k6 5k6
5
PD3 PD7
13 2
I2 O2
15 R22
47R * 20
1
PB0
14 3
I3 IC5 O3
14 R23
47R *
GND GND
4
I4 O4
13 R24
47R *
C4 C5 8 22
5
I5 O5
12 R25
47R *
6
I6
ULN2003
O6
11 R26
47R *
2u2 2u2 7
I7
GND
O7
10 R27
47R *
8
160233 - 11
Figure 1. The Atmel controller provides the muscle.
like the MC34063A which only requires in the wires. All of the outputs driving to be modified to make it work with the
a few additional external components. the LEDs are fitted with series resistors demo version of the compiler.
to limit maximum current to safe values. The port pins have been individually
Circuit building The ULN2003 driver chip has in built pro- declared and are referenced via their
The circuit was first tested using some tection diodes at each output which can alias names so that porting to a different
perfboard (a.k.a. stripboard/Veroboard). be used to suppress voltage spikes when controller or any changes to the I/O pin
The driver chips use a DIP package so the chip is used to switch inductive loads. assignments will be made easier.
they were really quick and easy to wire In this application we connect the com- Generating the pulse width modulated
up. The step-down converter chip was mon cathode pin of the three driver chips (PWM) signals to control the LED bright-
also quickly configured according to infor- together. When this connection is linked ness was a bit more challenging. The
mation from the datasheet and takes up to ground via a jumper it switches all the built-in PWM functions are only avail-
little space on the board. The connec- LEDs on together — that’s really useful able for use on a few port pins so they
tions necessary for programming the con- during installation to test the LED wiring are not of much help in this application.
troller were more difficult to arrange, a independent of the controller program. We can of course do the necessary work
double-sided board would certainly make ‘long hand’: A variable Pwm_help is used
the job easier. This PCB contains a few The Program in Bascom and continuously increments from 1 to
components not essential to the board’s The program has been written in Bascom 254 and rolls over to 1. A byte variable
function; D2 is included to prevent any (free download from [1]). I have pur- for each LED and step level defines the
damage if the input supply voltage is con- chased the full version of the compiler value of brightness. When this value is
nected the wrong way round. The push- without the code size restrictions of the smaller than Pwm_help the LED is off and
buttons to activate the lighting sequence free demo version. The controller also when it is greater the LED is switched on.
are likely to have long cable runs so they has more than enough memory for this A brightness value of 0 corresponds to
are fitted with an RC filter network to sup- application so no attempts were made at the LED being completely dark and val-
press any unwanted interference pickup code optimization. The code would need ues from 1 to 254 define the mark/space

learn design share
no significant difference in power con- You can also use continuous, adhesive
sumption. Experimenting with different LED strips cut to length. This type of
sleep modes in the controller did not pro- LED lighting strip is specified for opera-
duce significant savings in the 36 mW tion at 12 V. The strip consists of groups
consumed by the board. With the operat- of three LEDs in series together with a
ing mode triggered, current to the board series resistor and will not require any
rises to just above 14 mA and then you additional board-mounted series resistor.
need to add the current taken by the The 47 Ω resistors can, in this case be
LEDs to this figure. Here the resistors in replaced with short lengths of wire. The
series with the LEDs produce the biggest Darlingtons in the ULN driver chip have
losses. This equates to an energy loss of a saturation voltage in the range of 0.8
2.5 watts. In normal operation the LEDs to 1 V, you will need to factor this value
will not be in continuous operation so into any calculation of the LED drive cur-
these power losses can be considered rent and supply voltage to achieve the
Figure 2. The uncluttered PCB layout makes it acceptable. desired brightness.
easy to fit the components. How you hook up all the LEDs is up to
Assembly you, for the purposes of experimentation
The PCB layout is shown in Figure 2 you can use a connector but when the
ratio increasing until at 255 it is continu- and is available as a Target3001! file unit is installed its simpler to just solder
ously on. It only remains for this evalua- for download from the Elektor Magazine the wires onto the board pads.
tion process to be linked to an interrupt project page [1]. Fitting all the compo-
routine which is called at a reasonable nents to the board should be fairly sim- Why stop there?
frequency. In this case a reasonable fre- ple on this well designed board. All the The PCB can of course be used for other
quency is one that’s high enough so that components are standard and stocked projects; everything that you need to
the eye cannot detect the LED switching. by many suppliers but it will first be nec- control LEDs is on board. According to
When a pushbutton at the bottom of the essary to work out the best value and the data sheet the voltage regulator can
stairway is pressed the illumination level power rating for the LED series resistors. accept an input voltage up to 40 V. The
of the individual steps is increased from The 47-Ω 0.25-W values specified in the voltage rating of reservoir capacitor C1
the bottom step to the top step (also in parts list are designed for use with nine will however need to be increased corre-
the other direction for descending) with LEDs per step. The nine LEDs consist of spondingly. If you find your style is get-
a delay between each step. This is simply three ‘strings’ of LEDs wired in parallel ting cramped by the relatively small 8-K
controlled using the value of a few bits (each string is made up of three LEDs flash memory in the ATmega8 you could
and effective over four steps. You can wired in series). It will be useful if all the substitute the more powerful, pin-com-
alter the speed change value by chang- LEDs come from the same manufacturing patible ATmega328.
ing the values of the variables Delay1 batch so that their characteristics will be (160233)
and Delay2. similar and produce more or less equiv-
alent light output.
Web Link
Energy efficiency To work out the value of the LED series
Any lighting system such as this should resistors I first used information from the [1] www.elektormagazine.com/160233
be energy efficient. Powered from a LED data sheet. After some experimenta- (Bascom source code and
12 V supply the switch regulator alone tion with the LEDs however, it was clear Target PCB files)
draws around 2.5 mA, with the control- that their forward voltage drop was consis-
ler plugged into its socket the current tently lower than the value stated. Some
increases by a further 3 mA. Tests using experimentation may be necessary to
different versions of the ATmega8 showed arrive at the best value for these resistors.
The Author
Ever since he was young Peter Grundmann (b. 1956) has been fascinated by electronics. His first
practical experiences were connected with his interest in model trains. Much later he graduated in
business studies and now works as a software development engineer.
He has always been drawn to hobby electronics and besides analog and digital projects he is now
mainly occupied with microcontroller project design and programming. Along the way he has
developed specialized lighting effects and electromechanical actuators for model train installations.
These have also been incorporated into commercial products.
Peter is always happy to hear of your own experiences in this field of activity. You can contact him
at: [email protected].

Internet Radio with

Fluorescent Display (2)
Raspberry Pi plus ATmega plus software
By Michael Busser (Germany)
The plan is to make a kitchen radio using a good, old-fashioned, tried-and-tested fluorescent display, but with
the added twist of an Internet connection. In the first installment we looked at its functions, the hardware and
how we drive the display. But, without software, the hardware is nothing: the ATmega in the display module
needs firmware, and code is also needed to turn the Raspberry Pi into an Internet radio.
Figure 1. All the boards fitted into a clamshell enclosure and wired Figure 2. Lid on, firmware in: the kitchen Internet radio in operation. The
together. electronics is visible through the acrylic front panel.
If, with a little help from the circuit diagrams, board layouts Communication protocol
and other documentation you have built the VFD module from If two computer systems want to exchange data packets with
the first installment of this series [1], you may have discov- one another it is necessary to agree beforehand on how the
ered that (even assuming you have put everything together packets are to be interpreted: a protocol has to be defined.
correctly) not a lot happens when you power it up. Even with The homebrew protocol used here was also designed for use
all the printed circuit boards installed in the box and wired in a home automation system using a two-wire bus, and so
together (Figure 1), and with the lid closed (Figure 2) the has a few features that we will not be needing in this case.
display will still not light up as shown in the photographs. The Nevertheless, its structure follows the same lines as other
reason behind this is that the microcontroller in the display similar protocols. The firmware for the ATmega in the display
module needs firmware, so that it can understand the control unit was written entirely within the AtmelStudio development
commands sent to it. This firmware and, last but not least, environment (see screenshot in Figure 3).
the code for the Raspberry Pi, forms the topic of this second Table 1 shows the general structure of a data packet. A mes-
and final installment. sage of type TDataBuf is passed to the transport layer. The
Table 1. Data packet structure.

STX ETX
Message Container
start of transmission end of transmission
0x02 data 0x03
Table 2. The components of a message.

STX SRC DST MT MODE Payload (0 to 15 bytes) CS ETX

learn design share
two DLE characters. The following functions provide the soft-

ware interface (see serial.c and serial.h):
void initSerial(void);
void sendMsg(TDataBuf msg);
uint8_t peekMsg(TDataBuf *msg);
The function initSerial() should be called once when the main

program starts up. It initializes the UART and enables various
interrupts. The baud rate can be adjusted by modifying the
definition
#define UART_BAUD_RATE 9600UL
in serial.h. The other serial port parameters are eight data bits,
no parity, and one stop bit (‘8N1’). The function sendMsg()
takes a message of type TDataBuf, copies it to the internal
Figure 3. Screenshot of the AtmelStudio development environment showing
transmit buffer (if it is free) and then returns immediately to
the ATmega firmware for the display module.
the caller. Transmission and reception proper all happen under
interrupt control in the background. As soon as a valid mes-
transport layer starts by sending an STX character to mark the sage is received it is stored in one of the two receive buffers.
beginning of a transmission; an ETX character is sent at the A call to the function peekMsg() will then return a ‘1’ and the
end of each transmission. The STX and ETX characters should valid message will be written to msg.
not appear within the message itself: if they do, they should
be escaped by prefixing them with the data link escape (DLE, The escape mechanism is handled transparently within these
0x10) character. The receiver removes any escape character functions. There is one transmit buffer, but two receive buf-
from the data stream and appends the immediately following fers that are used alternately: this allows one message to be
character to the partially-assembled message. If it is desired processed while the next one is being received.
to transmit the escape character itself, it is necessary to send The code module is structured independently of the format
of the messages being sent, which is defined in protocol.c
and protocol.h. Two-wire bus systems such as RS-485 require
Table 3. Message codes.
an extra control signal which switches the hardware between
Field Meaning transmit and receive modes. The default mode is ‘receive’,
Source address: transmit mode being enabled only for the duration of a mes-
SRC
0 to 255 except for the values 2 and 3. sage transmission. This is the purpose of the two definitions
Destination address:
DST
0 to 255 except for the values 2 and 3. #define SENDER_OFF
Message type. The following are defined: #define SENDER_ON
0x10 StatusReq
0x11 StatusResp in serial.h. The receiver runs continuously, and so will pick up
0x12 DisplayReq any message that is sent: this can be used to detect a bus
0x13 DisplayResp collision. The only function we need in protocol.h is ownAdr(),
MT 0x14 Key_Req which returns our own address on the bus.
0x15 Key_Resp
0x16 Time_Req Table 2 shows the individual components of a packet and
0x17 Time_Resp Table 3 describes their meanings. Listing 1 shows the structure
Note that codes 0x02 and 0x03 may not be used; of a message as a C ‘struct’; the ‘union’ is a construct which
other codes are available for future expansion. describes how the TDatagramm structure and the TBuf byte
array can overlay one another at the same memory addresses.
Bit Meaning
This allows easy reading and writing of the message as a sim-
0...3: Length of payload in bytes: 0 to 15
ple array of bytes, which makes the receive and transmit code
4: 0/1 = checksum absent/
more straightforward.
checksum present
5: 0/1 = acknowledgement with ACK not
MODE ATmega communication
required/acknowledgement required
The following message types are defined.
6: 0 (reserved)
7: always 1
#define MT_STATUS_REQ 0x10
Note: values 0x02 and 0x03 may not be used. To
#define MT_STATUS_RESP MT_STATUS_REQ + 1
ensure this bit 7 is specified to always be 1.
CS Message checksum.

#define MT_DISPLAY_REQ 0x12

#define MT_DISPLAY_RESP
MT_DISPLAY_ Listing 1. Message structure.
REQ + 1 typedef struct { //construction of a message
uint8_t src; //source address (sender)
#define MT_KEY_REQ 0x14 uint8_t dst; //destination address (receiver)
#define MT_KEY_RESP MT_KEY_REQ + 1 uint8_t mt; //message type
TMode mode; //mode byte, describes structure of message
#define MT_TIME_REQ 0x16 TPayload Payload; //usage depends on application level
#define MT_TIME_RESP MT_TIME_REQ + uint8_t cs; //checksum, e.g. crc-8
} TDatagramm;
1
The DISPLAY_REQ message allows the typedef union {

Raspberry Pi to display something on TBuf Buf; //access via byte array
the VFD. KEY_REQ informs the Pi when TDatagramm Datagramm; //access via structure
a button has been pressed on the dis- } TDataBuf;
play module. TIME_REQ lets the ATmega
obtain the current time from the Pi,
which in turn obtains the time over the Listing 2. Example message function.
Internet using NTP. void sendMsgTimeReq(void) {
The ATmega can send and receive mes- TDataBuf msg;
sages over its serial interface in a speci-
fied format using simple calls to the rel- msgClear( &msg );
evant functions in protocol.c (Listing 2). msg.Datagramm.src = ownAdr();
Received messages are processed in a loop msg.Datagramm.dst = masterAdr();
in the main program (Listing 3), which msg.Datagramm.mt = MT_TIME_REQ;
continuously checks whether a new mes- msg.Datagramm.mode.bAckReq = NO_ACK_REQUIRED;
sage is available and then decides how to msg.Datagramm.mode.bCRC = NO_CHECKSUM; //WITH_CHECKSUM;
handle it. msg.Datagramm.mode.count = 0; //don't count header / trailer bytes
Let us look at an example of how these msgPrepare( &msg );
messages are used. Every 60 minutes sendMsg( msg );
the ATmega sends a message of type }
MT_TIME_REQ to the Raspberry Pi by
calling the function sendMsgTimeReq():
Listing 3. Processing a message.
if (timerFlags.flags.bMin) {
if (peekMsg(&recMsg) == 1) { //if message received: evaluate
doNTP_Sync--;
switch (recMsg.Datagramm.mt) {
if (doNTP_Sync == 0) { case MT_STATUS_REQ: break;
doNTP_Sync = NTP_SYNC; case MT_STATUS_RESP: break;
sendMsgTimeReq(); // fetch case MT_DISPLAY_REQ: i=0;
current time from Raspberry Pi b=1;
} for (i=0; i < cDISP_BUFFERSIZE; i++) {
timerClearMin(); if (recMsg.Datagramm.Payload[i] == 0) {b = 0;}
} if (b == 1) {
disp.buffer[i] = recMsg.Datagramm.Payload[i];
} else {
The Raspberry Pi replies with a mes-
disp.buffer[i] = 0;
sage of type MT_TIME_RESP, which is
}
then processed in the receive loop in the
}
main program.
disp.delay = cDISP_SHOWTEXT;
The source code is too long to reproduce
disp.mode = DS_TEXT;
here in its entirety, but it can be found in
showText();
the file Software_2.zip in the download
break;
available at [1].
case MT_DISPLAY_RESP: break;
case MT_KEY_REQ: break;
Raspberry Pi software
case MT_KEY_RESP: break;
We use version dated 09.09.2014 of the
case MT_TIME_REQ: break;
‘wheezy-raspbian’ operating system on
case MT_TIME_RESP: setTimeOnMsgReq( &recMsg ); break;
the Raspberry Pi. Detailed instructions
}
for obtaining an image, copying it to an
}
SD card, and carrying out basic config-

learn design share
you to everything you need to know.

Listing 4. The handleMsg() function.
A development environment is required
void handleMsg(int fd, TDataBuf *msg) { to get our first C program for the Rasp-
if (msg->Datagramm.mode.bCRC) { //test checksum berry Pi up and running (although in
syslog( LOG_NOTICE, "checksum ignored"); //tbd principle the command line can also
} be used). My environment of choice
if (msg->Datagramm.dst == masterAdr()) { is NetBeans, which being written in
switch (msg->Datagramm.mt) { Java also runs on my desktop com-
case MT_STATUS_REQ: onStatusReq(fd, msg); break; puter. Figure 4 shows a screenshot of
case MT_DISPLAY_REQ: onDisplayReq(fd, msg); break; the NetBeans IDE in action. A step-by-
case MT_KEY_REQ: onKeyReq(fd, msg); break; step guide to installing the NetBeans
case MT_TIME_REQ: onTimeReq(fd, msg); break; environment on the Raspberry Pi can
default: syslog(LOG_ERR, "unknown message dropped"); be found at [4].
dumpMsg(msg); Once these hurdles are overcome, it
} is possible to write C programs for
} else { //not my message the Raspberry Pi conveniently under
syslog( LOG_NOTICE, "message not for me, ignoring it"); Windows, and then compile and debug
} them on the Pi itself.
} Fortunately C code written for the
ATmega will run without significant
Listing 5. The onTimeReq() function. changes on the Raspberry Pi, as long
as you avoid using any of the special
void onTimeReq(int fd, TDataBuf *msg) {
functions and modules which it does
TDataBuf reply;
not support.
msgClear(&reply);
reply.Datagramm.src = masterAdr();
The program running on the Raspberry
reply.Datagramm.dst = msg->Datagramm.src;
Pi copies messages to a syslog server,
getTimeResp(&reply);
which helps with debugging. The sys-
//dumpMsg(&reply);
log server must be configured on the
frame_Send(fd, &(reply.Buf[0]), reply.Datagramm.mode.count + MSG_
Raspberry Pi. The function dumpMsg()
HEADER_LEN, reply.Datagramm.cs );
takes one message and outputs it to
syslog( LOG_NOTICE, "onTimeReq");
the syslog in human-readable form.
}
Inside the function main() in the Rasp-
berry Pi program we read all the char-
acters that have arrived on the serial
uration tasks can be found at [2]. port since we last looked and store them in a circular buffer:
At [3] there is a guide to configuring the Raspberry Pi as an
Internet radio using mpd (‘Music Player Daemon’) and mpc do {
(‘Music Player Client’), and there are many other similar guides len = read_serial(fd_v24, READBUFFERSIZE, readBuf);
available. A search for ‘Raspberry Pi mpd’ will magically lead RB_AddBuffer(&rb, len, &readBuf[0]);
//works even if len=0
} while ((len > 0) && !(rb.overrun));
The function frame_checkFrame() is called at regular inter-

vals. It checks whether there is a complete message in the
circular buffer. If so, it returns ‘1’, and if not, it returns ‘0’.
The parameters startindex and length give the starting offset
and length of the message.
A call to the function frame_get() returns the corresponding

message in msg. This is then processed by the function han-
dleMsg() (see Listing 4). So if, for example, a message of
type MT_TIME_RESP is received, the function handleMsg() will
pass it on to the function onTimeReq() (see Listing 5). Here a
reply to the request message is constructed and then immedi-
ately sent back to the sender using the function frame_Send().
The NTP client must be configured on the Raspberry Pi. System
Figure 4. Screenshot of the NetBeans IDE. With the software shown we can time is determined by a call to the function localtime() inside
turn the Raspberry Pi into an Internet radio. the function getTimeResp(). This function is also responsible for

assembling the reply mes-

Listing 6. Excerpt from the function onKeyReq().
sage in the TDataBuf data
structure. void onKeyReq(int fd, TDataBuf *msg) {
uint8_t keys = msg->Datagramm.Payload[0]; // define variable / check button
Internet radio if (keys & KEY_S0) {syslog( LOG_NOTICE, "onKeyReq: S0");}
The Internet radio function- if (keys & KEY_S1) {terminate = true; syslog( LOG_NOTICE, "onKeyReq: S1");}
ality itself is implemented if (keys & KEY_S2) {syslog( LOG_NOTICE, "onKeyReq: S2");}
using the mpc music player if (keys & KEY_S3) {syslog( LOG_NOTICE, "onKeyReq: S3");}
client. This allows you to con- if (keys & KEY_S4) { //next station
struct playlists from Internet if (radio.status == RADIO_ON) {
radio stations: the details are system("mpc next");}
set out at [5]. Also, the com- syslog( LOG_NOTICE, "onKeyReq: S4");
mand mpc help is available }
from the command line. Our if (keys & KEY_S5) { //radio on / off
specially-written C program if (radio.status == RADIO_OFF) { //if radio off
controls mpc by passing it system("mpc play"); //play last station
the necessary options via a radio.status = RADIO_ON;
call to system(), in just the } else //radio off
same way as it would be system("mpc stop");
used from the command line. radio.status = RADIO_OFF;
The excerpt from the func- }
tion onKeyReq() shown in syslog(LOG_NOTICE, "onKeyReq: S5");
Listing 6 is also worth a }
look. This function is called }
whenever a button on the
display module is pressed.
If mpc is called without parameters then the first line of its out- its hostname (here ‘mib2’) and the path to the project under
put is the current station and the name of the currently-playing Windows (here ‘D:/Projekte/...’).
program. The function getMPCStatus() obtains this information at However, it is easy to make this program execute automati-
regular intervals and then sends the station name to the ATmega cally when the Raspberry Pi is booted up. One description of
using an MT_DISPLAY_REQ message. The current station name how to do this can be found at [6]. If the program is started
then replaces the current date on the display. from a shell, it remains dependent on this shell; if the shell is
Each radio stream must be stored in a playlist before it can be closed, the program will also terminate. A simple ‘&’ character
used by mpd. The playlists are stored in the directory /var/ after the end of the command avoids this:
lib/mpd/playlists.
./hallowelt &
Start-up
As it stands the program on the Raspberry Pi needs to be started This makes the program execute in the background, with the
by hand, as development is not yet complete. One slight niggle shell generating a job number and a process ID. These are
with using NetBeans is that it creates very long directory paths. needed if you wish to use a command such as
If you follow the guide to installing NetBeans as the root user
exactly, executable code ends up in the directory fg %1
/root/.netbeans/remote/192.168.1.24/mib2-Windows- to bring the process back into the foreground.

x86_64/D/Projekte/RaspberryPi/RPi/HalloWelt/dist/ The program is now ready to run on the Raspberry Pi. Extensions
Debug/GNU-Linux are possible and indeed ideas for extensions are welcomed:
you can import the program directly into NetBeans from the
The part of the path printed in bold changes according to the file HalloWelt.zip. We hope you have fun experimenting!
environment, including the IP address of the Windows PC, (160207)
Internet Links
[1] Software download and first installment: https://www.elektormagazine.com/150720
[2] Raspberry Pi operating system installation: https://www.raspberrypi.org/documentation/installation/installing-images/
[3] Raspberry Pi as an Internet radio: https://www.youtube.com/watch?v=jf3M1RVpQbM (audio in German)
[4] NetBeans for the Raspberry Pi: http://bit.ly/2aBZ14A
[5] Internet radio tutorial: https://www.youtube.com/watch?v=jf3M1RVpQbM (audio in German)
[6] Automatically run a program on boot: http://pi.bek.no/autostartProgramOnBoot/

learn design share
IoT Shield for Arduino

Lets You Do Your (Internet of) Thing
This handy Arduino Uno R3 compatible shield is intended

for simple control applications in and around the
house. When the system is connected to the
Internet, the label “Internet of Things” (IoT)
becomes applicable.
Wires to and from the shield [1] can be connected

with a screwdriver – no soldering is required.
The circuit on the board proper does not use
all of the Arduino signals, and other shields can
be stacked on it as long as they do not interfere with
our board, making

ARDUINO the system very flex-
K11
10
ADA5/SCL ible and extensible. The
9
ADA4/SDA
8
board is larger than your
Aref K4
K8 GND
7 3
13
typical Arduino shield and
1 6 2
+5V JO13/SCK 12 its size allows the screw ter-
2 5 1
IOREF IO12/MISO 11
3 4
minals and trimmers to remain
K7 RESET IO11/MOSI
3V3
1 4
+3V3 IO10/SS
3
K3
accessible even when another shield is
2 5 2 3
5V +5V IO9 10 plugged onto it.
3 6 1 2
GND GND IO8 9
7 1
GND IOH 8
8
Vin
The board offers several expedient
9
POWER CD+
K6 functions:
7 10 3
K10 I7 O7 7U
• a variable voltage source for biasing
8 6 11 2
K2 IO7 I6 O6 6U
A5
3
K9 IO6
7 5
I5 IC2 O5
12 1
5U
or powering sensors;
2 6 6 4 13
A4 AD5/SCL IO5 I4 O4 • an analogue input with adjustable
1 5 5 3 ULN 14
A3 AD4/SDA IO4 I3 2003 O3 K5 gain for small signals;
4 4 2 15 3
AD3 IO3 I2 O2 4U
K1
3
AD2 IO2
3 1
I1 O1
16 2
3U
• a current and voltage limited input
3 2 2 GND 1
A2 R1 AD1 IO1/TX 2U for detecting a high voltage;
2 1 1 8
A1 1k IO0/RX
1 JP1
AD0
• six power transistor outputs for
A0 AD IOL
+5V switching for instance lamps or
relays.
P2 D1
adjust
Let’s walk through the shield’s functions
10k 3V3
step by step helped by Figure 1.
+5V
P1
100k Variable voltage source

amplify
3
8
5 Trimpot P2 provides a variable voltage
C1
JP2 1 7
IC1.A IC1 = MCP6232 IC1 IC1.B between 0 and 5 V. If jumper JP1 is
2 100n 6
4 shorted this voltage is available on the
R2 R3 Arduino’s analog input A3. In this case
it is also available at K1, pin 3, and can
100k
1k
be used to control, bias or even power

149014 - 11
an external (low-current) device like a
sensor. When JP1 is left open P2 has no
Figure 1. Just the facts, ma’am. A straightforward design without fancy tricks and hard-to-get function and this pin can be used as an
components. analog input or a digital input/output.

Small signals
IC1A, powered from the 5-V line, is a rail- Component List
to-rail input-output operational amplifier Resistors
All 5%, 0.25 W
(opamp) meaning that both its input as
R1, R3 = 1kΩ
its output cover the range of 0 to 5 V R2 = 100kΩ
(within a few millivolts). This output is P1 = 100kΩ, trimpot,
horizontal
connected to the Arduino analog input
P2 = 10kΩ, trimpot,
A0 only, hence its signal is not available horizontal
on a screw terminal block. The opamp is
Capacitors
wired as a non-inverting amplifier with its
C1 = 100nF, 0.2’’ pitch
gain is controlled by P1. The maximum
gain is about 100, the minimum gain, Semiconductors
one. The result is that the input signal D1 = BZX85C, 3.3 V, 1.3 W
Zener diode
on K1, pin 1 can either be amplified or
IC1 = MCP6232
it can simply be buffered. Although the IC2 = ULN2003
input impedance of the input is very high,
Miscellaneous
it can be fixed at 100 kΩ by shorting JP2.
JP1, JP2 = pinheader, 2-way,
Doing this is useful because it avoids the 0.1’’ mm pitch
opamp struggling with noise, hum and 2 pcs jumper for JP1 and JP2
other interference when its input is left K1,K2,K3,K4,K5,K6,K7 =
3-way PCB screw terminal
open. If, on the other hand, the source block, 5 mm pitch
connected to this input has a very high K8, K10 = 8-way stackable
IC socket, DIP-8 for IC1
output impedance, it may be better to pinheader socket, SIL, 0.1’’ mm pitch
IC socket, DIP-16 for IC2
K11 = 10-way stackable pin socket, SIL,
open JP2 to prevent squashing such a 0.1’’ mm pitch
PCB # 149014-1 v1.1
sensitive signal.
Detect high voltages max), a relay, a lamp or a motor for VIN, Reset, IOREF and AREF are only acces-
K1, pin 2 is a voltage and current limited instance, to 0 V as they can handle rela- sible on the shield connectors.
analog input. Zener diode D1 will limit tively large currents. An Arduino output (160169)
high DC voltages to 3.3 V maximum — cannot do this without the help of such
Web Links
a safe level for Arduino boards. But be a driver circuit. In case you must switch
careful, negative (AC) voltages are not more current than one channel can han- [1] www.elektor.com/
allowed, not on this input and, by the dle you can put two or more of these interfacing-screw-shield-149014-91
way, not on any other input. R1 provides outputs in parallel. If you want to switch [2] www.elektormagazine.com/160169
current limiting for both the input and inductive loads (like a solenoid, relay or
D1. Its value is a little high, meaning motor) you should add a flyback diode
that when you connect the input to, say, in parallel with the load (cathode to the
Great examples of how the shield
about 5 V, the voltage will be limited to power supply, anode to the output of IC2)
described in this article can used for
around 3 V, but with an input voltage of as the chip’s internal flyback diodes are
all sorts of applications, IoT or not,
12 V or 15 V you will measure 3.3 V here. not used on this board.
can be found in the book IoT Get U
The purpose of this input is to detect volt-
Going (Elektor Store SKU # 17460).
ages (much) higher than the Arduino can Some remarks
www.elektor.com/iot-get-u-going
accept, for instance the output voltage Power to the shield is supplied by the
of a DC adaptor or the state of a switch Arduino that carries it; do not power the
connected to a high voltage. Although shield through K7 when it is plugged onto
in theory this input can easily withstand an Arduino. This connector is intended for
voltages up to 100 V, never ever con- powering devices connected to the shield
nect it to the AC line voltage! only, not to power the Arduino.
The 3.3-V output is the Arduino’s 3.3 V
Powerful outputs output which is not very powerful, so treat
Part IC2, a ULN2003A, has a long-stand- it with loving care.
ing reputation, and a good one, which is IC1 and IC2 are mounted in sockets so
why it was used here. In it are seven pow- they can be replaced easily in case of a
erful Darlington transistors that can be mishap.
controlled digitally. Although the device
has seven channels, only six were used Most of the Arduino extension signals
on this shield. Each transistor can switch have been brought out to sturdy screw
500 mA and can stand up to 50 V. These terminals to allow easy connection of
transistors are very good to switch some- wires and cables to the board. Most, but
thing connected to a power supply (50 V not all: the serial port pins (pins 0 & 1),

learn design share
DAB+ Antenna Diplexer

Add digital reception to your car radio
If you feel like replacing the VHF-FM radio in your car with something more modern offering DAB+
reception too, you’ll discover that combination receivers have two separate antenna inputs. That’s
because two individual antennas are required. But this RF combiner makes it possible to connect both
inputs to one standard car antenna.
are provided with a glass-mounting antenna in case the car is

By Alfred Rosenkränzer (Germany)
not pre-equipped for digital radio (which was our situation).
This front windshield (windscreen) antenna is not an advantage
I began to pay attention to this remarkable type of filter when visually of course and moreover it requires making a hole in the
I considered replacing the feeble VHF radio in our car with a dashboard for the antenna cable. I wanted to avoid unpleasant
more modern set including DAB+ reception. Fortunately our car discussions with my wife in any case.
was old enough to be still fitted with a DIN slot (standardized
on many European automobiles), making the change-out rel- If you wish to use an off-the-shelf antenna, you must first estab-
atively problem-free; DIN-compliant radios are still available. lish whether it will receive signals at all in the DAB+ region;
A DAB+ car radio does of course have two antenna inputs, one this is because some active antennas have a built-in lowpass
for VHF (FM) and a separate one for DAB+. The one for VHF filter. The simplest way to find out is to connect it to a spectrum
connects to a normal car antenna, whilst some DAB+ radios analyzer (Figure 1), otherwise test it with a DAB+ receiver. As
is well known, the VHF (FM) band covers from 88 to 108 MHz,
whilst DAB in Band III goes from 174 up to 230 MHz. L-Band
in the region of 1.5 GHz (used for DAB in some territories) is
irrelevant for broadcast reception in vehicles.
The available signal needs to be split for the two antenna inputs.
In the simplest case you just connect two cables in parallel.
This of course upsets the impedance matching, leading to
signal reflection and in some cases to (partial) loss of signal.
A passive splitter (consisting of three resistors) avoids this
mismatch for sure but brings with it additional attenuation,
which you seldom want. An RF transformer is another option
for sharing the signal.
First trials
You could, however, achieve the desired separation also by
combining a lowpass and a highpass filter with suitable fre-
Figure 1. Image of the DAB+ spectrum at 178 MHz, Channel 5c. quency characteristics.

To test this out I began by measuring two commercial filters.

Figure 2 shows the frequency responses of the filters, each
measured separately with a network analyzer. The frequen-
cies illustrated are not appropriate for separating VHF and
DAB+ but both filters are commendably flat regarding the
passband region.
If you connect both filters in parallel using a T-adapter, there
is a strong mutual influence on the frequency response (Fig-
ure 3). After this I experimented further dividing the signal
with a resistive power splitter and additional 6 dB attenua-
tors after the power splitter. Although the mutual influence is
reduced, it’s still not elegant. And there is additional attenuation
of 3 or 6 dB. The low-pass filter also shows a falling frequency
response. This is not an optimal solution then.
Diplexer software
A diplexer could offer a solution here. This employs a com-
bination of lowpass and highpass filters (or bandpass filter/ Figure 2. Lowpass filter SLP-150+ and highpass SHP-250+ by Mini-Circuits,
band rejection filter). In other words a frequency divider for each swept independently with a network analyzer.
radio frequency use.
Both filters are matched to each other in such a way that they
can (and must) be driven in parallel by the source. For calcula-
tion there is the ‘Diplexer Design’ software by James L. Tonne
[1]. This program allows you to select between lowpass/high-
pass and bandpass/bandstop variants. The following descrip-
tion covers only the lowpass/highpass version.
On the Design page of this program (Figure 4) you can define
the characteristics of the diplexer. The ‘Order’ is a measure-
ment of the complexity, and hence outlay (the number of
components), and the abruptness of the filtering achievable.
‘Crossover freq’ indicates the point of intersection of the two
filters. In this VHF/DAB+ example it is roughly in the middle
between the highest VHF frequency (108 MHz) and the low-
est DAB+ frequency (174 MHz). The crossover frequency can
be fine-tuned to the extent to make the attenuation of the
unwanted band in each filter more or less equal. You can also
attempt to replace ‘awkward’ (i.e. not made commercially) coil
values calculated with readily available ones.
The lowpass in the upper section begins always with a series
inductance. When calculating an individual filter you can select
Figure 3. Both filters are connected in parallel here by T-adapter. This
whether you want to start with the series inductance (T-struc- creates a strong mutual influence on the frequency response.
ture) or a shunt capacitor (Pi-structure). Here this is not an
option. Correspondingly the highpass always begins with a
series capacitor.
‘Passband ripple’ indicates the amplitude of the periodic vari-
ation in the passband region, as in a Chebyshev filter. The
greater the ripple, the sharper the filter and vice versa.
‘System Z’ indicates the impedance, in our example 50 ohms.
When you press the ‘Plot’ button, you reach the next screen-
grab (Figure 5).
You can change the order, ripple and crossover frequency again
at this stage and observe the effects directly. You can set the
presentation style using ‘Plot Options’ and at ‘Markers’ super-
impose markers.
Next we use Simetrix to simulate the diplexer we designed
and compare it with a Chebyshev lowpass filter.
A practical design
Following this work, I developed a PCB using EAGLE. Awkward
coil and capacitor values were handled using available values Figure 4. Design page of the Diplexer Design program.

learn design share
The maximum order is 7. By omitting the capacitors and coils

(substituting wire links) you can, however, also achieve every
deeper order. The coils are available in 0603 packages.
The schematic of the practical circuit is shown in Figure 6,
with the associated PCB in Figure 7. This PCB fits in a small
plastic box and can be screwed in place (Figure 8). The input
and output cables are soldered direct to the PCB. On its upper
side there is an area of groundplane between the two filters;
the underside is entirely groundplane.
The resulting measurements in Figure 9 corroborate this design
within the scope of the component tolerances. At approximately
45 dB attenuation the lowpass goes into ‘saturation’. Although
definitely undesirable, that’s not functionally significant.
Active version
Figure 5. Plot page of the Diplexer Design program.
As the transmit power of DAB+ transmitters is rather mod-
est in comparison with analog VHF stations, I searched for an
appropriate preamplifier and struck rich with the MAX2630
wired in parallel and series. It is possible to connect capacitors from Maxim.
in parallel and install them side by side without any problems.
Connecting coils in series is another matter. The two magnetic As shown in the datasheet [2], it comes in a 4-pin SAT143
fields can interact, creating a transformer (albeit a very bad package. The only extra circuitry, besides the bypass capac-
one), and the effective inductance value is not the desired itor for the supply voltage, is one capacitor each at the input
sum of the two values. Accordingly the coils were offset by and output. Since it fits perfectly on a scrap of perfboard, no
90 degrees and, as far as the SMD layout would permit, sep- PCB had to be developed for testing. Its gain is around 15 dB
arated from one another by distance. and the noise figure is 4. The measured frequency response
L1 L2 L3 L4 L5 L6 L7 L8
SIG
68nH 22nH 100nH 15nH 100nH 4nH7 39nH 0nH
C1 C2 C3 C4 C5 C10 DAB (Coax)
39p 3p9 39p 3p3 33p 0p GND
SIG
(Coax)
GND C6 C8 C11 C13
2p7 0p 1p 10p
SIG
C7 C9 C12 C14
L16 L9 L10
12p 12p 12p 22p FM (Coax)
33nH 33nH 39nH

GND
160234 - 11
Figure 6. Schematic of the passive frequency splitter for FM/DAB+.
L3 L4 L5 L6 L7
L2
L1 C5
C3 C4 L8
C1C2 C10
L16 L9
L10
C7 C6
C8 C11 C14
C9 C12 C13
Figure 7. EAGLE layout for the circuit in Figure 6. Figure 8. Prototype of the diplexer in its plastic housing.

runs in an almost straight line up to 250 MHz (and evidently

carries on much further).
Encouraged by these results I developed an active diplexer:
see Figure 10. The input and the two outputs are each fitted
with a MAX2630.
The amplifier on the input probably makes sense only if a pas-
sive antenna is installed in the car. With the higher level from
an active antenna, this could be too much of a good thing. You
can then you can bridge the input and output of the amplifier
with a wire link or an SMD zero-ohm resistor. The same applies
on the FM output if the receiver is being overdriven. However,
the amplifier on the DAB+ output probably does make sense
in most cases.
The power is supplied by a 3.3-V voltage regulator, with each

amplifier drawing about 7 mA. The series diode at the input
prevents damage from incorrect polarity; the two resistors
should reduce the thermal power dissipated by the regulator. Figure 9. Frequency curve of the diplexer with markers.
The 12 V can be tapped off the car radio at the output provided
for powering an active antenna. The PCB of the active diplexer
(Figure 11) fits in a diecast aluminum box from Hammond
U1
(type 1550WQ).
C17
(160234)
L8 C15 C16
L3 L4 L5
Web Links C23 L2 L6 L7
U3 L1 C2 C4 C5
[1] www.tonnesoftware.com/diplexer.html C21 C1
C3
C10
C22
C6 L16 L9 L10
[2] https://datasheets.maximintegrated.com/en/ds/MAX2630- LSP1 D1
C7
C8
C11
C14
R1 R2
MAX2633.pdf C9 C12
C13
LSP2 U2 C19
C20
C18
LED1
U$1 C24
R3
Figure 11. PCB layout for the active diplexer.
+3V3
LD1117 3.3V
LSP1 D1 R1 R2 U$1
150R 150R
ES2B SMD LED1
C20 C24
+3V3
LSP2 10u 25V 10u 16V R3 C15

680R
100n
4
C22 L1 L2 L3 L4 L5 L6 L7 L8 C17 C16
3 1
SIG
+3V3 68nH 22nH 100nH 15nH 100nH 4nH7 39nH 0nH
1n 1n 1n
2
C21
C1 C2 C3 C4 C5 C10 DAB (Coax)
39p 3p9 39p 3p3 33p 0p GND

100n
4
C23
3 1
SIG
1n +3V3
2
(Coax) C18
GND
100n
C6 C8 C11 C13 4
C19
2p7 0p 1p 10p 3 1
C7 C9 C12 C14
SIG
L16 L9 L10 1n
2
12p 12p 12p 22p FM (Coax)
33nH 33nH 39nH
GND
160234 - 12
Figure 10. Schematic of the active diplexer.

welcome in your
ONLINE STORE Elektor Bestsellers
1. LEGO Control Board for
Raspberry Pi
www.elektor.com/ lego-rpi-board
EDITOR’S CHOICE
Built around an Arduino Uno, this cool gadget displays the

current time of the day, not by showing digits or changing
the relative position of a pair of hour and minute hands,
but by plotting four digits into a layer of sand. After an
adjustable delay two vibration motors flatten out the sand
layer before the write cycle begins again. Three miniature
servo motors control the arms of a pantograph assembly.
Mechanical parts of the Sand Clock (a refined project based
on an original idea
from our friends at 2. Red Pitaya for Test and
Make Magazine) mainly consist of pieces of Measurement
acrylic sheet, a material readily available www.elektor.com/red-pitaya-book
and well suited for laser cutting. The use 3. Elektor Uno R4
of metal gear servo motors provides www.elektor.com/elektor-uno-r4
high accuracy with acceptable 4 CAN Projects with ARM and
slack. Arduino
Denis Meyer www.elektor.com/can-projects
Elektor Editor 5. BBC micro:bit
www.elektor.com/microbit
6. Mooshimeter
www.elektor.com/sandclock www.elektor.com/mooshimeter
BBC micro:bit - 35 Touch Develop CAN Projects with ARM and Arduino SmartScope Maker Kit
& microPython Projects
NEW
This book is about the use of the BBC micro:bit computer This book details the use of the ARM Cortex-M family of The new SmartScope Maker Kit (available exclusively from
in practical projects. The BBC micro:bit computer can be processors and the Arduino Uno in practical CAN bus based Elektor) contains a version of the SmartScope that’s specially
programmed using several different programming languages. projects. Inside, it gives a detailed introduction to the prepared for FPGA development, with ready assembled
This book makes a brief introduction to the Touch Develop architecture of the Cortex-M family whilst providing examples internal headers and additional FPGA pins that permit reading
programming language and the microPython programming of popular hardware and software development kits. Using and writing. The remainder of the kit contents, including 2
language. It then gives 35 example working and tested these kits helps to simplify the embedded design cycle programmers and 2 probes, is also highly worthwhile.
projects using these language. Complete program listings are considerably and makes it easier to develop, debug, and test
given for each project. a CAN bus based project.
member price: £20.95 • €22.45 • US $25.00 member price: £28.95 • €33.75 • US $38.00 member price: £242.95 • €269.95 • US $300.00
www.elektor.com/microbit-book www.elektor.com/can-projects www.elektor.com/smartscope-maker-kit

shopping books cd/dvd diy projects development tools sale
Mastering Microcontrollers
Helped by Arduino
3RD revised edition
with AVR Playground and Elektor Uno R4
This third, extended and revised edition of Mastering Microcontrollers

Helped by Arduino will not only familiarize you with the world of Arduino
but it will also teach you how to program microcontrollers in general.
In this book theory is put into practice on an Arduino board using the
Arduino programming environment. Having completed this fun and
playful course, you will be able to program any microcontroller, tackling
and mastering I/O, memory, interrupts, communication (serial, I²C, SPI,
1-wire, SMBus), A/D converter, and more. This book contains two new
chapters: AVR Playground and Elektor Uno R4.
MEMBER PRICE: £32.95 • €36.50 • US $41.00

www.elektor.com/mastering-microcontrollers-3
Door Spy with Raspberry Pi Red Pitaya for Test & Measurement Elektor SDR Reloaded
This GSM HAT-board for the Raspberry Pi can be used in The Red Pitaya is a credit card-sized, open-source test This Arduino shield is a remake of our famous SDR project
combination with a Pi camera to build a hi-tech doorbell. and measurement board that can be used to replace most published in 2007. A Software Defined Radio is a universal
When a visitor rings your doorbell, the camera takes a measurement instruments used in electronics laboratories. tool in RF technology circles, one that can also be put to use
picture of him/her and sends it in an MMS-message to your This book aims to teach the principles and applications of for making measurements. The characteristics of the receiver
cell phone. You can start a telephone conversation via the basic electronics by carrying out real experiments using are defined in software, which now gives us the opportunity
built-in intercom, and you can even send an SMS to open the Red Pitaya. Many fun and interesting experiments are to use an Arduino Shield as a front-end.
your front door. included. The book also makes an introduction to visual
programming environment.
£52.95 • €58.46 • US $65.00 member price: £26.95 • €31.45 • US $36.00 member price: £98.95 • €26.96 • US $144.00
www.elektor.com/rpi-door-spy www.elektor.com/red-pitaya-book www.elektor.com/sdr-reloaded

Welcome to the SHARE section
By Thijs Beckers (Elektor Netherlands)
SHARE
The following remarkable story was shared with me by one
of the designers in the lab. He had problems at home with
his hot-water heater. The following was happening: whenever
warm tap water was required, there was a brief period of hot
water, followed by a long period of cold water and finally the
water became warm again and then stayed warm… Since this
was a rented heater, there was no need to think about this
too long and a repairman was called. The fault was reasonably
quickly diagnosed. The water heater detected much too late
that warm water was required and as a result was very slow
to turn on. The first small quantity of warm water came from
a smaller heater. Anyway, a new part needed to be ordered.
But until then — and here it comes — the temporary solution:
plug the AC line plug the other way around in the wall socket!
(in Holland, earthed AC outlets are not polarized, Ed.)
This was more than a little strange, of course, because what is the difference with AC, you would think.
The repairman explained that the problem probably was related to the ionization of the part, but he
wasn’t able to provide more details. We can imagine that ionization could indeed have an influence on
DESIGN
the operation of the temperature or the flow sensor, but that turning the plug around would have any
effect we find at least a little suspicious. This leads to the conjecture that somewhere in the electronics
in the heater, part of a circuit has been economized away during the development. But that doesn’t
mean we are not interested in knowing what is happening exactly! If you know what is going on, write
to us at [email protected] and we will share it with the Elektor community. We are very curious.
Now something completely different. As the permanent Retronics editor, colleague Jan Buiting frequently
comes into contact with tubes. Old tubes. Such-thick-layers-of-dust-that-the-type-number-is-unread-
able tubes. So, what do you do? Exactly! Clean them. Only in that past the (white) ink used wasn’t the
best, so if you don’t take
care you will clean the
number off right along
with the dust!
Jan has by now devel-
oped a knack for this,
which I would like to
share with you:
Firstly: never ever wipe
a cloth over the printing.
You are sure to rub off
the printing! Wipe only
the dust off the other
LEARN
parts of the tube, this

is usually without any
problems. Now take a
new, unused paint brush
with long, soft bristles
and use this to brush
carefully back and forth
across the printing, taking very good care that you do not damage the lettering. It could easily take
50 back and forth sweeps before you see any results. Don’t be tempted to use a toothbrush or similar
for this; these are much too stiff and possibly damage the printing. A soft paint brush works the best.
If necessary you can then clean the remainder of the tube with a damp cloth, optionally with a small
amount of glass cleaner, but stay away from the printing. And done! Don’t wipe the brush across the
damp tube and the printing, otherwise there is the risk that you rub the latter off after all.
(160252)

review from the labs web scouting updates
Simulation
with SystemVision
A free web-based tool
Circuit simulation is a valuable

tool for finding errors in circuit
designs at an early stage. There
are lots of Spice-based programs
available for this purpose, but
they are usually fairly expensive.
Free versions are also available,
but most of them are rather
limited. Mentor Graphics recently
launched SystemVision Cloud as
a free browser-based simulation
tool. We took this program out
for a spin.
By Tam Hanna (Slovakia)
It’s common knowledge that the ear- who register for the service through their Twitter, for example, you see a pop-up
lier you find an error, the less it costs to Facebook, Microsoft, LinkedIn, Twitter or window on the screen where you have to
correct. Computer verification of circuit Google accounts. The actual login process enter your Twitter address to give Sys-
designs is especially suitable for circuits depends on which network you use. With temVision access to your account.
without programmable components.
SystemVision from Mentor Graphics is

a comprehensive simulation program
which has been around for a good while
already, but it is not free. SystemVision
Cloud (Figure 1) was launched a while
ago as a free version with basic function-
ality. The main difference between the
free and full versions is that components
generated with SystemVision Cloud are
automatically shared with the community.
Getting started
For our needs, that should be enough
to get started. Go to the website www.
systemvision.com and click the Log In
button to open the pop-up login window.
At the time of publication of this article,
SystemVision accepts not only traditional
user accounts, but also accounts for users Figure 1. The SystemVision workspace.

learn design share
connect them in the usual way. Now

comes the question of how you assign
individual component values in System-
Vision. The answer is that you click the
label, which opens the dialog window
shown in Figure 4. There you enter the
desired parameter values and then click
somewhere outside the dialog window to
save the entered values.
Next you enhance the circuit as shown in
Figure 2. SystemVision makes extensive use of context menus.
Figure 5 to create the conditions neces-
sary for proper simulation of the RC net-
work. Here it should be mentioned that
Due to the strong social media focus of a Design (Figure 2). SystemVision basically works with units of
SystemVision, creating a new project is SystemVision allows you to choose an ohms, farads and so on. You can change
not entirely straightforward. Click on the existing design as the next step. If you the order of magnitude by entering the
yellow taskbar at the top of the screen want to start with a completely blank following letters or letter combinations
and select the option Designs ¨ Create sheet, click My Designs instead. In the after the numerical values as appropriate:
window that appears next, click the Cre-
ate function. In response, SystemVision • f
opens the Design view, which in turn • p
fetches the wizard for opening a new • n
design on the screen. • u
• m
The program takes a while to load. • K
Depending on your system configuration, • Meg
it may go faster with Google Chrome or • G
Firefox than with the traditional Internet • T
Explorer. When you generate a design
from scratch, you have to enter the title At this point alert readers are probably
Figure 3. Components can be flipped and rotated
and select the visibility. If you select wondering why we chose a pulsed volt-
as desired.
“Show only to me” in the Visibility field, age source for this circuit instead of a
you create a private design. constant voltage source. This seemingly
illogical choice is due to a minor bug: if
The user interface of SystemVision is you power an RC network from a constant
fairly simple. On the left side there is voltage source, you disable part of the
a list of the various components avail- simulation environment, and as a result
able in SystemVision. They can be posi- you only’ see constant voltages instead
tioned on the drawing sheet by drag and of characteristic waveforms.
drop. Options for flipping and rotation are
shown in grey for currently selected com- Super simulation
ponents (Figure 3). You can use these Before you can start the simulation, you
options to adjust component orientations have to declare a ground symbol. That’s
as necessary in your circuit. because the simulation engine behind
SystemVision only works when it has
You can wire components together by been assigned a reference potential. Then
dragging and dropping lines between their you click the Simulation End Time box on
terminals. To do so, position the mouse the Emulation toolbar at the right side
cursor on a component terminal, where it of the screen and enter “5” for a simu-
changes to a crosshair. Then drag a line lation interval of five seconds. Then you
to the destination terminal and drop it click the arrow symbol at the top to start
there to complete the connection. a time domain simulation. SystemVision
can also do simulations in the frequency
As a first example, we decided to draw domain, but there is not enough space
a simple RC circuit which demonstrates here to talk about that.
the familiar gradual voltage rise after the After the simulation run has finished on
power is switched on. the Mentor Graphics computer, the pro-
gram opens the results window shown
Figure 4. The window for entering component First you drag and drop the necessary in Figure 6, where the various quanti-
attributes is hidden away. components onto the drawing sheet and ties generated during the simulation are

made in the browser window when you

click the Save button located on the pre-
viously mentioned yellow toolbar.
Social media and more

Although the simulation functions of Sys-
temVision are very attractive, its main
feature is the ability to share designs
with your friends and colleagues. When
you click the Share Design button, your
Figure 5. Our RC circuit is ready for simulation. Figure 7. You can use the pencil symbol to create
design is released for sharing and Sys-
a watcher.
temVision displays a URL (for example,
http://sysvis.io/smKBn9) which you can
listed. Any errors in the circuit model are share with other people. immense functional scope which can only
indicated by red bubbles on the screen, The free basic version of SystemVision be touched on in a single article such
and the error messages are generally allows up to five private designs. If you as this.
easy to understand. need more you have to buy the profes-
The main feature is the ability to share designs with your friends and colleagues
If you want to save yourself the trouble sional version, but you can also stay If you are interested in circuit simulation,
of constantly digging through the results below the five-design limit by deleting you should definitely check out this pro-
folder, you can use the pencil symbol to designs that are no longer needed or gram. Particularly with regard to usability,
create a watcher as illustrated in Fig- releasing them for sharing. Unfortunately, it is miles ahead of PSpice.
ure 7. A watcher is a component that clicking the Upgrade Account button only (160203-I)
hovers above the circuit and displays the leads to an email address. No price infor-
results of the current simulation run. mation can be found on the Internet.
At the time of publication of this article, If you are looking for inspiration, you can
SystemVision had an annoying bug in the browse through the circuits provided by
simulation system: when you launch a other designers under the Find a Design
second simulation, an additional folder heading. That is especially helpful when
is created in the simulation window. For you want to learn more about a particu-
this reason it is advisable to delete dia- lar component or you are looking for tips
grams and simulation components that on how to set up effective simulations.
are no longer needed, by right-clicking
them and selecting Delete in the con- Summary
text menu. Another minor irritation is Although SystemVision appears a bit com-
how changes are saved: SystemVision plicated at first glance mainly due to the
only saves changes to the circuit diagram cloud-based user interface, it offers an
Figure 6. The simulation results have arrived.

learn design share
Opamp
Power Supply Connexions
Capricious or logical?
What can go wrong when fitting a component on a circuit board? All sorts of things of course, but in this case
the cause turned out to be really strange: The power supply connections of the IC were in the wrong place!
By Luc Lemmens (Elektor Labs) to not function properly, but one of those turer of the device in hand in order to
was very remarkable. be 100 percent sure of the correct pin-
out. With ICs, the suffix after the part
During the testing of the first prototype The true and tested method of finding number is usually an indication of the
of the Tabula Tangibles (a project that such faults is very simple: Remove the specifications or the type of package,
is a collaboration between Elektor Labs components one by one from the circuit but that an R on an IC means that the
and Aachen Technical University), there board and measure the resistance across supply connections are reversed, that’s
was something seriously amiss with the the power supply after each one, until new to me! I can’t think of any plausi-
power supply: the output of the volt- the culprit has been found. Fortunately ble explanation why, as a manufacturer,
age regulator appeared to be short-cir- there are not that many components on you would produce both these variants;
cuited. The first thought is a soldering this board and I also had the good for- the datasheet doesn’t mention a single
problem. An SMD board that’s built by tune that after unsoldering the first IC word about this either.
hand could, for example, have a solder the short-circuit disappeared! This was
bridge between the pins of an IC, or a single opamp type MCP601 from Micro- As already mentioned, there was some-
something. This is generally not a big chip in a SOT23-5 package, so back to thing else wrong (incorrect step-down
deal and is easily fixed with some solder studying the datasheet once more and converter for the desired voltage range),
wick. However, a thorough visual inspec- what turns out to be the case: there are so that is why I haven’t investigated
tion of the board didn’t reveal any prob- two variants of this IC in this particu- this phenomenon any further at the
lems. An additional check of the sche- lar package — the 601 and the 601R, moment. But after I promised the Editor
matic did not bring anything suspicious with the difference that the latter has to write a From the Labs article about
to light either. Neither did another check the power supply connections VDD and it, I found it so improbable that I mea-
of the footprints and the connections of VSS the other way around than planned sured it again, just to be doubly sure.
the components. Nevertheless there was on my PCB... But it is really so: there are two vari-
a suspiciously low resistance between ants. Whoever understands the logic
the output of the voltage regulator and I have come across many surprising behind it may elucidate us.
ground, there had to be ‘something’ hap- things in electronics. For example, con- (160257)
pening here. In the end there appeared sider the BF254 or BS170P FETs, where
to be two things that caused the circuit you need to know the actual manufac-

New Precise Nixie Clock

Elektor 3/2016, p. 56 (150189)
UPDATE. A case for this unit includes six LEDs and three resistors to give cool background
illumination effects. A circuit diagram to assist in wiring the lights is given on page 4 of the
manual which can be downloaded from:
www.elektor.com/acrylic-glass-case-for-six-digit-nixie-clock-150189-72
Err-lectronics
Corrections, Updates and Feedback to published articles
LEDs replace the fluorescent tube

Elektor 1/2016, p.123 (Err-lectronics)
FEEDBACK. Further to a letter in Elektor 3/2016:
I finally got round to testing two replacement LED strip lights or
tubes in a bathroom cabinet fitting which had two fluorescent tubes
connected in tandem (in series). My conclusion is that LED tubes are
not suitable as a direct replacement in this series-connected application.
Even with the two LED tubes wired in series without a ballast they will light
but one or both tubes will have reduced light output. A competent electrician could
relatively easily rewire the light fitting so that both LED tubes are connected together
in parallel and both receive full mains voltage when the light is switched on. Mains
voltage is lethal so it’s important to ensure all necessary precautions are taken. Once
the unit has been rewired the ballast and pseudo starter can be dispensed with. Note
that conventional fluorescent tubes can no longer be used in the fitting once it has
been rewired for use with LED tubes.
Robert Schrott
ELF Receiver
Elektor 10/2014, p. 8 (140035)
FEEDBACK. In addition to my participation in lectures at the University of Jena I am also actively investigating
low frequency (electromagnetic) vibrations. So far the frequency range of my investigations has been between
3 Hz and 1 MHz.
I am also interested in lower frequency signals below 1 Hz. I went on to build the Elektor ELF receiver and the
resulting investigations are very interesting. I was surprised to detect signals in the frequency range from 0.01 to
0.02 Hz. I use a wire antenna 8 cm to 17 m in length so these signals are more likely to be electrical rather than magnetic
in nature. So far I have been unable to find any literature describing these signals. Also I have not had
any satisfactory explanation as to their origin at the University of Jena.
Maybe geological vibrations of the earth are the key but I don’t know the mechanism
involved where (mechanical) spherical or toroidal vibrations could induce these
electrical signals. The frequency range is maybe about right?
I wonder if any other reader has had a similar experience.
Walter Koch

learn design share
The Mendocino Motor

It spins and sways by solar power alone
By Manfred Klose (Germany)
At first sight it looks like something that came out of a wizard’s studio — hovering in mid-air and spinning
on its own, apparently without any source of energy. The first time you see a Mendocino motor you can’t
help being fascinated. The operating principle is really very simple, a well-designed and neatly-finished
model adds to the intrigue and makes the effect even more magical.
A Mendocino motor is a type of electric support the rotor in mid air. They are also available to order from the Elektor Store.
motor powered by energy produced by slightly offset to keep the rotor in suspen- The frame of this motor is made from
solar cells. It uses a rotor suspended on sion. According to Earnshaw’s theorem transparent acrylic sheet. A powerful neo-
low-friction magnetic bearings. The motor it is not possible to achieve static levita- dymium magnet is positioned centrally
was invented by Larry Spring who comes tion using any combination of fixed mag- beneath the rotor. Its magnetic field per-
from Mendocino County in California. nets and electric charges. To make this meates the rotor coils above it. A small
The motor consists of a rotor and frame. bearing arrangement stable, one end of current flows through the coil when light
The rotor is an eight-sided cylinder with the axle is terminated with a ball bearing falls on the illuminated solar cell. The
an axle passing through its centre. Circu- which is pushed against a plate to form Lorentz energy produced in the coil gives
lar magnets are fitted at each end of the a low-friction contact. This prevents the the rotor a pulse to make it rotate. The
axle with the same magnetization direc- rotor being pushed out of its position in arrangement of coils and solar cells on
tion as magnets fitted to the frame (Fig- the frame by the magnets. the rotor ensures that pulses of energy
ures 1 and 2). This arrangement acts To illustrate the motor principle we will will keep the rotor spinning, so long as
as the rotor bearings while the magnets take a closer look at the X-8 type of motor the light intensity falling on the cells is
exert a repelling force on one another to which was built by the author and is now sufficient.

The Lorentz force is very weak so the

low-friction magnetic bearing is neces-
Magnet on axis
sary in the design. Because the forces
Axis Solar cells & field coils
involved are so small it’s important to Glass plate
reduce any static imbalance in the rotor

also so that the mass inertia can be over-
come and the rotor can start turning on
its own when the light level is sufficient. Fmagnet Fmagnet
Construction
of the Mendocino motor
Altogether there are eight solar cells Bearing magnet Field magnet
Base plate
arranged edge to edge to form the rotor.
This gives the rotor an octagonal cross
section. One cell will be illuminated while Figure 1. The magnets on the axle are suspended Figure 2. Sketch showing the Mendocino motor
its opposite number will be in the shade. by magnets fixed in the frame. bearing arrangement.
The two cells are wired in series which
would seem to present a short circuit
to the voltage generated (Figure 3).
In practice however it doesn’t cause a
problem because one of the cells will be The motor spins from the glow of a large tea light.
in the shade whilst its partner will be in
full sun. Between the cell facing upwards
and the cell facing downwards there is a
difference in the incident light and also a
difference in the current generated. The Light
cell facing the light produces a voltage
while the cell in the shade acts as a rela-
tively high impedance. Current therefore Coil about 5
flows through the coil. After the rotor has S
Solar cells,
rotated through 180˚ the coil has turned each 0.5 V
around so that the energy pulse supplied N
by the opposite cell (which now faces the
light) is of the correct polarity to reinforce
the rotor’s spin. N
For this motor we use 65 x 20 mm sized Dark S

Magnet
solar cells which have a maximum out-
put voltage of 0.5 V. The magnetic field Figure 3. Two cells are connected in series which Figure 4. The coils are arranged inside the
strength produced by a coil can be cal- should produce a short-circuit condition but only assembly of solar cells.
culated as the product of the number one of the two cells will have light falling on it.
of turns and the current flowing in the
coil. To get a high value of current flow-
ing it’s necessary to ensure that the coil rotor produces a sequence of drive pulses from the incident sunlight by the time
does not have a high value of resistance. which are close together, making rota- the next pulse is provided by the follow-
Each coil on our motor is made up of 60 tion smoother and improving the motor’s ing cell. The rapid decrease of energy
turns of 0.3 mm diameter (AWG 28) wire. ‘willingness’ to start turning. from the preceding cell means that the
The coil winding has a DC resistance of The coils in the rotor assembly fit under- energy it produces will not be sufficient to
around 5 ohms. neath the solar cells (Figure 4) and are stop the rotor. Eight pulses per revolution
Current in the coil is given by the voltage the same width as the drive magnet in ensure that the motor will spin even with
divided by the resistance: the frame. They are positioned directly relatively low light levels. In full sun you
behind the solar cells and are therefore can expect it to get up to around 1400
I = V/R = 0.5/5 = 100 mA. positioned optimally in the magnetic field. rpm. What’s more impressive however is
In contrast to other motor designs where that the glow from an ordinary tea light
The solar cell specified here can supply the coils are positioned parallel to the (with a flame height of around 2 cm) is
almost 300 mA in full sunlight, yield- cells and are visible, the wire length in enough to get the motor turning.
ing some reserve in the system. Solar this design is reduced which has a ben- (160227)
cells show a steep drop in output volt- eficial influence on current flow.
Web Link
age and current as the incident light level The ideal number of solar cells is eight.
decreases. The use of a relatively high The cell which generated the preceding www.elektor.com/
number of small narrow solar cells for the rotational pulse will have turned 45° away search?o=mendochino&q=mendocino

learn design share
Build a Chipamp!
Great sound with only a few components
The building and tweaking of audio equipment is something that occupies many enthusiasts. It is very
satisfying and nothing beats listening to music on equipment that you have built yourself, even partially.
In this installment of WebScouting we’ll show you that you don’t need a complex design, or a large
number of components, to produce a fine sounding power amplifier.
By Harry Baggen (Elektor Labs)

The popularity of these amplifiers, which bled board for less than $10. It uses the
were really designed for general audio use LM3886 [2], which is the most powerful
It’s certain that not all audiophiles are and not specifically for Hi-Fi (let alone IC from the Overture series that can out-
techies. Most of them aspire to get as high-end) equipment, began in 1999 with put almost 70 W into 4 Ω. They also sup-
faithful a sound reproduction as possible, the introduction of the 4706 (Gaincard) ply the required power supply boards very
and technical details don’t interest them. amplifier made by the Japanese manu- cheaply. At those prices you are beginning
They’re perfectly happy to just listen and facturer 47 Laboratory [1]. The aim of to question the quality of the components
trust their ears. This has caused many 47 Labs was to create the best possi- they’ve used…
strange products to appear in the audio ble sound reproduction by using as few
world, with their effectiveness questioned components as possible in the audio sig- The advantage of such a power amplifier
by the techies, such as wooden supports nal path. This amplifier consisted of just is that it’s very easy to experiment with
for loudspeaker cables… nine components per channel, which had different types of components, and of dif-
an LM3875 at its heart (an IC costing a ferent qualities. All you need to add is a
In contrast, when you’re choosing com- few euros). In various reviews the 4706, transformer, a bridge rectifier and several
ponents for an audio amplifier you can which cost several thousand dollars, was electrolytics, and you’ll get some music.
certainly work out why they would sound praised to high heaven. As a result, sev- You can then find out what the sound
different. The choice of the amplifier eral other well-known audio amplifier is like when you use different types of
configuration also plays a big role, of manufacturers started to use this IC in electrolytics, or when you connect many
course. When I was searching the Inter- their amplifiers. smaller electrolytics in parallel, or when
net I noticed that there were many web a different input capacitor is used, etc.
sites that dealt with so-called chipamps This has encouraged many hobbyists to A good description of several experi-
or gainclones. These are simple audio experiment with the ICs from the Over- ments using this LM3886 chipamp can
power amplifiers that use a power amp ture series. There are even several spe- be found on the (Dutch) website of Dick
module. Most of these were ICs from the cialist web shops that sell only compo- van de Merwe [3]. The website of DIY
Overture series made by National Semi- nents and PCBs for these ICs. Chinese Audio Projects has several projects based
conductor, which has since been taken manufacturers have also jumped on the on chipamps, such as the DIY LM3875
over by Texas Instruments. bandwagon, and offer a fully assem- (which they call ‘The Beast’) [4] and the

DIY LM3886 Chip Amplifier (Gainclone) Web Links

Kit [5].
[1] www.sakurasystems.com/reference.html
It’s even possible to build this circuit with-
out the use of a PCB. An extensive set [2] www.ti.com/product/lm3886
of photos that show you how it’s done [3] www.audio-creative.nl/audio-creative/diy-lm3886-chipamp/
is shown on Mick Feuerbacher’ website, [4] http://diyaudioprojects.com/Chip/Beast/
Dogbreath [6].
[5] http://diyaudioprojects.com/Chip/LM3886_CA/LM3886_CA.htm
Of course, Elektor got in on the act as [6] http://dogbreath.de/

well, publishing a design for the LM3886 [7] https://www.elektormagazine.com/magazine/elektor-199807/34116
in the Bumper Summer Issue of 1998 [7],
when the IC had only
just appeared on the
market. It even included
a PCB layout!
There are many more

similar designs to be
found on the Internet,
why not have a look? The
bottom line is that all
these designs use a sin-
gle IC along with a few
passive components. It’s
hard to believe really. Oh
well, you can’t argue
about taste and sound
quality!
(160261)

learn design share
Elektor Labs
Pipeline
The New Year has begun, and of course the (stale) list of resolutions compiled a few years back got dug up
once more. Since these resolutions are never fulfilled anyway, why not replace some of them with things that
are feasible, like doing some electronics projects? Here are a few suggestions to get you started.
Build a LOUD sound player

Searching for a powerful gong or audio notifier? The “Card Sound” board presented
here consists of a TDA7266 audio amplifier featuring two 7-watt outputs, a CS4344
AD converter, an STM32F401 ARM Cortex-M4 microcontroller, and a slot for a micro
SD card. The goal is to play different (high-quality) sounds stored on an SD card
with enough volume to wake up even the deepest sleeper.
https://goo.gl/yc4or6
Let a Raspberry Pi water your lawn

A little worked up because of precious time lost on watering his lawn when he could
have been working on his electronics projects, the author started thinking of ways
to automate this tedious job. A first prototype, a 12-channel sprinkler timer/con-
troller, built with an intelligent PIC-based, touch-capable graphic display worked
well, but sadly could not be controlled remotely. The second version is technologi-
cally more advanced with a PIC32, Ethernet, Windows 10 IoT, and a Raspberry Pi. https://goo.gl/JwND75
Remote controlled speaker switch box

Some people sure would like to listen to a 5+1 surround audio system, but don’t
want to give up their stereo system. Sadly, both systems in one room would require
eight speakers, which is too much for many, and also not necessary. Since both
systems have speakers for the left and right channels the idea emerged to build
a switch box which allows configuring the system as 5+1 or as stereo while only
requiring the 5+1 system’s speakers. https://goo.gl/6h6Jos
Intelli-Cell smart battery analyzer

Here is a very smart and reprogrammable rechargeable battery analyzing instru-
ment with a special mounting system to access the terminals of any cell phone
battery (including Apple), as well as any other type of battery of any composition
or voltage from 1.25 V up to 12 V. It has automatic polarity sensing, heat sensing
both on the mounting plate and on the external cable accessory. It can be pro-
grammed easily for any composition (NiCd, NiMH, Li-Ion, Li-Fe, Lead Acid) and any
amp-hour rating from 400 mA to 5000 mA.
The mounting plate includes fully adjustable pogo contacts with slides and eleva-
tor to access any battery contacts, no matter how difficult. The external cable has
alligator clips and a sensing thermistor to reach larger batteries or for using the
Apple adapter. https://goo.gl/ar15x7
(160253)

From Verobox
to Heavy Metal
Elektor lab instruments 1980s and 1990s style
Roughly between 1984 and
1988, Elektor published
several pieces of DIY
laboratory instruments
all housed in two-color
Verobox, hard plastic cases
with aluminum front and
rear panels. A second gulf
of benchtop instruments
for many more applications
was launched in 1989 and
marked by solid, metal cases and again a sort-of uniform ‘look and feel’ to them. I found a couple of both
series in the Retronics attic and decided to take the family picture “in the box, 15 years on”.
By Jan Buiting, Editor-in-Chief tor measurement series” in fact there plans for a series of affordable, high qual-
were several. This article is my attempt ity, DIY measurement apparatus, simply
at describing two, suitably identified by stopping at the PCB was not an option.
Back in 1995, with time on hands before the boxes the Elektor lab workers agreed Like true industrial designers, they started
my return flight, I was given a private to use. Did they? with the enclosure and the ergonomics
tour of the RadioSpares laboratories in of the controls, before even going to the
Corby, Northants. That’s right, it’s todays Product engineering library and opening the RS Components
headquarters of RS Components. It was If you are among the countless readers Datasheet Collection to find an opamp or
at a time when the first CD-ROM version who started out as an electronics hobby- a connector. Some designers argued that
appeared of the RS Catalog, complete with ist and now wield spreadsheets, $K bud- the prescribed enclosure was a restrictive
a search function, which was a sensation gets and BOMs “for the industry” you will factor to their creative genius but eventu-
although it proved unable to find ‘BC547’. remember that the sheer gusto produced ally followed suit. One size fits all — these
One function of the RS lab staff in Corby by your working circuit was often so great instruments must have the Elektor look
was to supply verified engineering data as to quench any desire to build the PCB and be instantly recognizable!
on components and systems, for use in into a case decent enough to show out-
the RS Datasheet Collection RS customers side your shack. And indeed many Elek- In the beginning was Vero
could subscribe to. All on paper of course tor projects published over the years The first more or less coordinated series
and in luscious, dark red binders complete are “done, finished, ready” when the was to be housed in the two-tone (two-
with tab sheets. To my amazement all five autorouter says zero errors and the pro- color) “Verobox” sized 205 (w) x 137 (d)
lab staff in the Components section had a totype board works. The fact that these x 75 (h) mm. As an example for your
full complement of Elektor test equipment days an increasing number of projects are amusement and déjà vu experience, I
published from about 1984 onwards as microcontroller based does not seem to have pictured the Elektor Pulse Gen-
DIY projects. The instruments were easy encourage any of the programmers out erator in Figure 1. The case consist of
to recognize from their Verobox housings there to have a go at giving the project an off-white cover and a greyish bottom,
with the distinctive grey and off-white a professional look (and feel). They hate both made from ABS (hard plastic). The
color scheme! The techies at RS were nuts and bolts. front and rear panels are 1.5-mm mm
very satisfied with the equipment and said thick aluminum sheet. I remember we
it all worked okay once calibrated against A case for the case sold ready-made screen-printed self-ad-
“the unaffordable gear at head office”. Back in the 1980s when a team of Elek- hesive foils for sticking on the front panel
While it’s tempting to think of one “Elek- tor lab workers and editors outlined their (which you had to drill first). The word-

learn design share
Figure 1. To kick off in ABS: Elektor Pulse Figure 3. A rare bird and seldom used too Figure 5. Over to Telet metal cases with the HFE
Generator, mid 1980s. Sorry about the missing as nobody cares any longer about the exact Tester from 1990.
cap on the 0.1–1.0 Repetition Time variable expression of analog voltages: True RMS Meter,
control. again mid-1980s.
Figure 2. A look inside the Pulse Generator; Figure 4. Inside the True RMS Meter. Note the Figure 6. Plenty of space inside that Telet case!
note the vertically mounted boards and the two range switch connected directly to the PCB.
transformers rightly deferred to the rear panel.
ings and symbols used on the foil were Another fine specimen of the Verobox T many, Geist, while C-I Electronics served
fiercely debated between the four lan- & M series was the Elektor True RMS most non-European areas with mail order
guage departments within Elektor, and Meter I stumbled on in the Retronics only. The upshot: Elektor achieved a mas-
fortunately it was agreed to use a com- attic (Figure 3). Sorry about that LCD sive success and these instruments are
bination of (minimal) English and “fits-u- which went fuzzy over the years, I will now legendary.
all” symbols. The name of the instrument send a complaint to Sharp. In this instru-
remained a challenge though as you can ment the power transformer was mounted Italian metal is better
see in Figure 1 (top left hand corner). on a separate aluminum sheet, possibly Like just about any supplier of test and
The slots formed by pairs of protrusions because of its weight (Figure 4). measurement instruments, Elektor
at the short sides of the bottom half of Sadly I do not have the pinnacle of the worked to innovate and refresh the prod-
the Verobox were used to secure printed Elektor Verobox T & M series in my col- uct line and improve brand awareness. By
circuit boards vertically. As shown in Fig- lection, the Elektor XR2206 Function about 1988, the Verobox got a bit long in
ure 2, the slots allowed front panel con- Generator. Rest assured though that it the tooth and a new “standard enclosure”
trols such as potentiometers and switches looks just like the other two instruments was sought by the Elektor team to house
to (a) assist in securing the “controls” I was able to rescue from the skip back yet another glut of DIY instruments for
board to the front panel and (b) to be in 2006. That Function Generator is still use in the home lab. Still using the ada-
connected directly to the PCB tracks seen today on auction sites and it’s a gio: Affordable — Quality — Professional.
rather than with flying wires as in “hobby- veritable classic. Rumor has it that parts distributor Texim
ist” projects in EPE et al. LEDs and other From memory I recall several other Electronics, a fantastic source of elektor-
indicators could shine through transparent instruments in the Verobox series, like ical parts and located in a murky part of
clearances in the front panel foil. a wattmeter and a multimeter but I do Holland called Achterhoek (‘back corner’),
Everything to do with AC line voltage and not have them. “kindly advised” metal cases produced by
power transformer’ing was allocated to That Verobox was expensive and Elek- Telet, Italy. As opposed to the Verobox,
the rear panel. Note that the molded tor got many complaints especially from these Telet cases were (a) all-metal and
screw bases in the bottom half of the super thrifty Hollanders. Some advertisers (b) available in different sizes. They were
Verobox were not used — all PCB orien- were thrilled though with the T & M series accepted as the new standard.
tation was to be vertical. And note those and started to supply kits as well as the Figure 5 shows a random example of
ventilation holes carefully drilled in the infamous “elektorical” parts. In the UK, Elektor’s “new” T & M series, the HFE
top and bottom sections by Jan Visser Maplin was the main supplier. In Holland, Tester, sized 198 (w) x 180 (d) x 80
and/or Mr. Feron. DIL Elektronica ruled the roost, in Ger- (h) mm. Note the correct use of subscripts

on the front panel — a feat for Patrick

Retronics is a monthly section covering vintage electronics
Wielders on the graphics & drawing staff!
including legendary Elektor designs.
Also watch those two shades of Arduino
Contributions, suggestions and requests are welcome;
green (some say: turquoise) used for the www.elektor.tv please telegraph [email protected]
front panel foil. I hope the colors come
out true on the photos.
Within the Telet series of instruments, all of the instruments discussed here actu- ded technology projects, the articles pub-
boards were allowed to be fitted horizon- ally work and are part of the Retronics lished in Elektor Magazine on these humble
tally on standoffs, and controls secured Classic Repair initiative I started here at instruments not only explain their assem-
directly to the front panel for wiring to the Elektor House. While I realize their joint bly and practical use, but also delve into
board (Figure 6). Note that in the case of functionality can be run in twenty or so the measurement principles, and engineer-
the HFE Tester the enclosure is pretty vacant clock cycles of a Red Pitaya board, or by a ing issues spotted and vanquished ele-
as no internal PSU was incorporated and 15-year old running an Arduino or Rasp- gantly in Elektor fashion. Add to that, the
the instrument ran off a DC power adapter berry Pi with suitable extension boards, effort the lab team put into metalworking
(well done guys, Safety First!). these retro instruments are fun to use and product engineering. If you have a
What other Telet instruments from the even occasionally. For example, the Watt Verobox’ed or Telet-cased instrument built
Retronics collection? Meter is permanently inserted in the AC from an Elektor article I did not mention
• An AC line Wattmeter with LCD line supply to alert me to boatanchors here, please let me know as I am sure the
readout, switchable ×1 and ×10 consuming more than 1,000 watts or so list presented here is not exhaustive. I am
ranges and the AC line out socket from the electricity rationed to Retronics. in the ESTD 2004 box.
right on the front panel (Figure 7) In stark contrast with the latest embed- (160197)
(Telet 198 × 132 × 80 mm) [2].
• The Watt-Hour Meter from 1993 in
the largest Telet case Elektor used
in their series (Telet 297 × 180 x 80
mm) (Figure 8) [3]. Advanced for
its time, at least in DIY land, it had a
V24 serial interface to capture read-
ings as well as do the calibration.
The V24 connector is a 9-pin sub-D
located on the back panel of the Figure 7. The very plain, almost Spartan looking Figure 8. One of the largest instruments built
instrument. Watt Meter from 1991. into a Telet case was the Watt-Hour Meter from
• The TV Pattern Generator which I 1993.
still use today to check vintage oscil-
loscopes for their trigger and delay
capabilities (Figure 9) (Telet 247 ×
180 × 80 mm). Here the LEDs are
seated in holes and they protrude
slightly from the front panel. This
instrument is AC line powered.
• Finally, an Inductance Meter feign-
ing to have nanohenry ranges (Fig-
ure 10). Admittedly, in the lowest Figure 10. Inductance Meter. I fail to see why the
Figure 9. Still in frequent use at Retronics Labs:
nH range this instrument is beaten tiniest of tiny center dot (interpunct) was used
the TV Pattern Generator, albeit not to test TV
hands-down by our more recent 500 instead of a plain ‘×’ sign in front of the range
sets or monitors any longer.
ppm LCR Meter. markings.
Two other Telet-style projects are not Web Links

shown here but should be mentioned also:
Note: original articles from 198x and 199x are available as pdf files on Elektor’s
a simple 15 V, 500 mA power supply and
1980-1989 and 1990-1999 DVDs respectively.
the Advanced LCR Meter [4]. The latter I
have available as a kit originally purchased [1] HFE Tester (900078), Elektor Electronics September 1990, www.elektormagazine.
from DIL Elektronica by a deceased Elektor com/magazine/elektor-199009/32219
reader. I still have to assemble this fantas- [2] Watt Meter (910011) Elektor Electronics April 1991,
tic design by Hans Bonekamp, and pitch www.elektormagazine.com/magazine/elektor-199104/32348
it against our recent 500 ppm LCR Meter [3] Watt-Hour Meter, Elektor Electronics February & March 1993, www.elektormaga-
which costs 2.5 dB more. zine.com/magazine/elektor-199302/32797
[4] Advanced LCR Meter, Elektor Electronics April 1997,
Not forgotten
www.elektormagazine.com/magazine/elektor-199704/33817
With the exception of the True RMS Meter,

Elektor World News
Compiled by Robert van der Zwan
Fast Forward Award 2016: wind and… water

Low on Sunshine?
The winner of the 2016 electronica Fast Forward Award sure
knows how to exploit wind energy. But what about exploiting Why Not Switch to the
water with great accuracy? This
is the idea behind a novel water
meter with a level of precision
rarely seen today. The FFA Jury
considered the idea of Australian
The electronica Fast Forward Award winners can
FFA participant Lenn Williams to
step forward now! Two of them are from Ger-
be so unique that it created the
many, one from the USA. It sure was an exciting
new ‘Tech for Good’ award right
Friday morning at the electronica 2016 trade
there and then. Lenn (photo; with glasses) flew back to his
show in Munich. The excitement was in the air
home country with a 5,000 euro check to spend on PR and
until the very last moment – and naturally so!
marketing, and for advocating to the entire water sector
that it and its customers can get a fair(er) share.
The electronica Fast Forward Award ceremony
kicked off at 10 am sharp at the Elektor Booth,
the vast booth at the East entry that captured
Do not take our word for it the imagination of grown-ups and youngsters
Do not take our word for it. Falk Senger, alike. The Elektor booth was all about inno-
the Managing Director at Messe München, vation, bringing together 35 inventions from
made it very clear he was more than 16 countries. The international Jury, coor-
happy with the way the electronica Fast dinated by Elektor and also consisting of
Forward Award brought all those e-inno- renowned companies like STMicroelectron-
vations to the fore. Concluding the FFA ics, Würth Elektronik, Conrad and Trinamic,
2016 awards ceremony, he could not help had a tough time deciding which three entries
noticing that both the quality and the should win in the set categories Prototype,
diversity of the entries were outstanding. Start-up and Idea.
The Jury of the FFA, ‘powered by Elektor’,
could not concur more. After some deliberations on Friday morning,
based on the 35 presentations during the
first three days of electronica 2016, the
Jury reached its conclusion arouns 10:20:
electronica 2016 in numbers Mowea from Berlin topped in the category
Let’s have a short flashback on the electronica 2016 trade Prototype, BotFactory from New York was
show. A total of 2,913 companies from more than 50 countries first in the category Start-up and Kewazo
presented their solutions to approximately 73,000 visitors from from Garching (Germany) took pole
more than 80 countries. The main topics were security and position in the category Idea.
automotive. 99 percent of the visitors rated the Munich show
as ‘good’ or ‘excellent’. Combined with a 7 percent increase in All three were asked to respond to salient
the numbers of exhibitors since the last electronica in 2014, questions from both members of the Jury
Falk Senger, Managing Director at Messe München, thinks and the public.
exhibitors and organizers did “an impressive job”.
At 11:00, it was decision time. FFA 2016 Jury Manager Clemens
Valens announced the winner of the third prize; Kewazo with
Director of electronica München
its (yes…) scaffolding building robot. Artem Kuchukov became
Angela Marten and Publisher of
the happy owner of a check worth 25,000 euros to be spent
Elektor Don Akkermans signed
on PR and marketing.
an agreement to continue the
collaboration on the electron-
As Clemens rightly put it: “calling out number 2 is, by
ica Fast Forward Awards, the
implication, also calling out number 1”. Number 2 was
Start-up Platform powered by
BotFactory with its PCB prototype desktop printer (a complete
Elektor. The FFA will return to
pick & place machine included!). CEO J.F. Brandon was the
electronica in 2018 and be
lucky owner of a check worth 50,000 euro.
kept alive through 2017.
PEOPLE NEWS ● Tanja Pohlen can take credit for organizing the electronica Fast Forward Award, as
trade show, Tessel Renzenbrink and Jens Nickel kept online readers updated on everything they needed
anchorman Jan Buiting and cameraman Patrick Wielders helped a few camera-shy people to speak up
Sales, Client, and Member staffers Julia Grotenrath, Chantalle Reuling and Raoul Morreau spoke to

EXPERT PROFILE
Elektor works closely together with more than 1,000 experts and authors
for the publication of books, articles, DVDs, webinars and live events. In
each installment of Elektor World News we put one of them in the limelight.
Weakest of Winds?
Name: Dr. Till Naumann
Education:
PhD, Technical University Berlin
Professional interests: I love

wind and aerodynamics, maybe
because my father was a stunt
pilot
Who is Till Naumann?

I am an engineer from Berlin. I wrote my PhD thesis at the Tech-
nical University Berlin about the aerodynamics of wind turbines.
I’d like to see ‘pico wind’ made affordable, using my knowledge
about big wind turbines. I met my partner Andreas [Amberger,
Ed.] at the TU Berlin. We developed a modular wind energy sys-
tem based on 10-kW nominal power.
Where you surprised to win the electronica FFA?

Our team was very surprised! We came from Berlin by train, slipped
in one of the cheapest hotels and put all our energy in the pitch in
front of the Jury – after almost three days of no sleep. We made
it and we are very happy indeed!
Do you have a plan how to spend your 75 K euros or is it

too early to tell?
Being a start-up, we need marketing in every direction you can
imagine: corporate design, a better looking homepage, testimo-
nial videos of our prototypes and of our field tests, content for
B2B partners, presentation opportunities at relevant trade shows
and especially community building, on Facebook and other media.
We are planning a crowdfunding campaign, tentatively at the end
of 2017.
What do you hope to accomplish within the next five years?

We aim to build a fast growing, but also solid business that is able
to deliver clean, affordable off-grid energy solutions. We want to
Number 1 was… Mowea take part in the decarbonization of energy production.
with its micro wind turbines, allowing households or small
companies to combine solar energy with wind energy,
What made you start developing your wind energy plan?
and so create a stable power supply despite being off-
I love wind and aerodynamics. Because my father was a stunt pilot,
grid. The prize for CEO Till Naumann and his team: a
I spent a lot of time in the air as a small kid. Anyhow, Andreas
75,000 euro check and an exhibition stand at electronica
and I decided to take up our passion for ‘small wind’ when we
2018 included!
noticed a strong interest in small, renewable energy solutions
worldwide — but especially in countries that lack a proper energy
Photo: Dr. Till Naumann and Lara Obst of Mowea presenting their
infrastructure.
‘pico wind’ turbine.
What will be the most key electronics development in the

not too distant future?
can Angela Marten from electronica 2016 ● During the Andreas is an expert in everything to do with electronics and
software. We want to make our turbines ‘smart’ so that we can
to know from Munich ● While in Munich, Elektor TV
collect data about energy production and customer behavior. We
and share bits and bytes of extra information ● Elektor also want to integrate our turbines into micro grids. In short there
is an awful lot we want to do!
countless people at the Elektor Booth ... ... ...

jouez & gagnez
chatdoku a mind-teaser for elektornics enthusiasts

Contributed by Claude Ghyselen (France) The rules for filling in the grid and the characters allowed are
the same as for the normal Hexadoku grid.
Reminder: a Hexadoku grid uses figures from the hexadeci-
mal system, i.e. from 0 to F. You should fill it in so that all the
This New Year, our traditional Hexadoku grid is being hexadecimal figures from 0 through F (0–9 and A–F inclusive)
replaced with a Catdoku grid (the outline is shaped are used once and only once in each row, column, and block
like a cat’s head). (identified by a thicker line).
The Catdoku differs from the Hexadoku

in the following ways: • The number of boxes in each block depends on its shape
and position, but is never more than 16. So this grid can
• The grid has 16 blocks, but 20 lines (1–20) and 20 col- still be described as a Hexadoku.
umns (A–T) (see grid below).
• The total number of active boxes is 204, compared to
• Five areas are empty (white). 256 for a Hexadoku.
• The blocks are not square or rectangular, but are more As certain blocks contain fewer than 16 boxes, it’s not possi-
like pieces of a jigsaw (referred to as a hexadoku jig- ble to use all the hexadecimal characters in these ones. This
saw). is just a little extra difficulty to bear in mind when solving it.
Certain figures have already been entered in the grid to define
• Block #1 (pupils of the eyes, in bright yellow) is made the starting situation. In order to enter, there’s no need to send
up of two separate parts, one each for the left and right us the whole grid, all you need do is send us the sequence
eyes. of six figures in the grey boxes.
Boxes grouped by blocks Number of boxes per line

A B C D E F G H I J K L M N O P Q R S T
1 3 8 2
Block Number
2 3 3 8 8 4 number of boxes
3 3 3 3 8 8 8 6 1 8
4 3 3 3 3 6 6 7 7 8 8 8 11 2 11
5 3 3 3 3 6 6 6 7 7 7 8 8 8 8 14 3 16
6 3 3 6 6 6 6 6 7 7 7 8 8 8 9 14 4 13
7 4 4 4 6 6 6 6 7 7 7 9 9 9 9 14 5 13
8 4 4 4 1 6 7 7 1 9 9 9 11 6 15
9 2 4 4 1 10 1 9 9 9 9 7 13
10 2 4 4 1 10 1 9 9 11 11 10 8 16
11 2 4 4 4 1 10 10 10 1 9 9 9 11 11 14 9 16
12 2 5 5 5 10 10 10 10 10 10 10 10 10 11 11 11 16 10 14
13 2 2 5 5 5 15 15 12 12 12 12 13 13 11 11 11 16 11 12
14 2 2 2 5 5 5 15 15 12 12 12 12 13 13 11 11 16 12 15
15 2 2 5 5 15 15 15 12 12 14 14 14 12 13 4
16 5 5 15 15 15 12 12 12 12 12 14 14 12 14 9
17 15 15 15 15 15 14 14 14 14 9 15 15
18 16 16 16 16 16 16 6 16 14
19 16 16 16 16 4
204
20 16 16 16 16 4
1 3 7 16 15 13 10 14 11 13 13 13 13 16 11 13 10 6 4 2 204
n

A B C D E F G H I J K L M N O P Q R S T
1 9
2 8 0 F
3 B 7 1
4 1 6 3 0 4 A 5
5 D 5 9 B 4 2
6 0 E 3 7 F B
7 6 F 8 B 3 9
8 5 E B 0 D C
9 2 6 D A 4 E B
10 6 C 2 F 1 4 B
11 F A 8 3 5 1 D E
12 B 1 7 0 8 9 2
13 0 1 E C A
14 D 9 7 2 5
15 4 3 B 0 7 D
16 9 F D E 3 B
17 F 9 1 3 6 2
18 6 9 A 4
19 8 5 7
20 B D 0
Solve Catdoku and win!

Correct solutions received from the entire Elektor readership automatically enter
a prize draw for three Elektor Book Vouchers worth $70.00 / £40.00 / €50.00
each, which should encourage all Elektor readers to participate.
Participate!
Ultimately February 1, 2017, supply your name, street address
and the solution (the numbers in the gray boxes) by email to:
[email protected]
3 6 4 5 A 7 B F 8 D E 0 2 9 C 1
D E 1 7 C 0 2 8 F 9 3 5 A B 4 6
A F 0 2 9 D 1 5 C B 4 6 8 E 3 7
Prize Winners 9 8 B C 3 4 6 E A 1 7 2 5 D F 0
The solution of Hexadoku installment 5/2016 (November & E C 3 6 D 8 4 7 0 2 F 9 B A 1 5
December) is: 6182F.
4 1 2 8 B 5 0 3 D 6 A E C F 7 9
The €50 / £40 / $70 book vouchers have been awarded to: 5 7 9 0 F A C 2 3 8 B 1 4 6 D E
Eugene Stemple (USA); Michael Staffler (Germany); B A F D E 1 9 6 4 C 5 7 3 0 8 2
Dirk Dreesen (Belgium). C 0 E B 2 9 7 4 6 A 8 3 D 1 5 F
Congratulations everyone! 1 2 A F 5 B 3 0 7 4 9 D 6 8 E C
6 3 8 4 1 E D C B 5 0 F 9 7 2 A
7 5 D 9 6 F 8 A 1 E 2 C 0 3 B 4
F 9 5 E 0 C A 1 2 3 D 8 7 4 6 B
0 B 6 1 8 2 F D 9 7 C 4 E 5 A 3
2 D 7 3 4 6 E B 5 0 1 A F C 9 8
The competition is not open to employees of Elektor International Media, its
8 4 C A 7 3 5 9 E F 6 B 1 2 0 D
subsidiaries, licensees and/or associated publishing houses.

Elektor USA - January, February 2017

Uploaded by

Copyright:

Available Formats

Elektor USA - January, February 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Elektor USA - January, February 2017

Uploaded by

Copyright:

Available Formats

What is Catdoku and how does it differ from Hexadoku?

What is Catdoku and how does it differ from Hexadoku?

What are the rules for solving Catdoku?

What are the rules for solving Catdoku?

Edition 1/2017 | January & February 2017

LEARN DESIGN SHARE

Arduino microcontrollers FIND OUT

make it easy for you to turn MORE ►

your ideas into reality!

The most popular Arduino!

Small but not to be underestimated!

Small but mighty!

ARDUINO MEGA 25.50 ARDUINO NANO 16.73

SHOP CONVENIENTLY ONLINE! ORDER HOTLINE: +49 (0)4422 955-360

ISSN 1757-0875 (UK / US / ROW distribution)

Elektor Magazine, English edition

www.elektormagazine.com January & February 2017 3

24 electronica Fast Forward Award 2016

26 ElektorBusiness: Simple Interfacing to Analog

65 The Elektor Community

14 Tips & Tricks:

50 Connect Objects with Genuino 101

LEARN DESIGN SHARE 56 BrainBox Arduino

42 Elektor SDR Reloaded (4) 74 Cool Power with D-Watt

4 January & February 2017 www.elektormagazine.com

122 Build a Chipamp!

AXICAT, 124 Elektor Labs Pipeline

for PCBs 92 MAXREFDES99# Display Driver Shield

106 IoT Shield for Arduino

www.elektormagazine.com January & February 2017 5

vice providers. One name that always seems to crop up is

access with the Internet of Things (www.atmel.com/tools/at88ckecc-aws-xstk.aspx). On board there

6 January & February 2017 www.elektormagazine.com

Figure 1. Fetrons from Down Under. Figure 2. Period product advertising.

www.elektormagazine.com January & February 2017 7

8 January & February 2017 www.elektormagazine.com

Voltage and Color

www.elektormagazine.com January & February 2017 9

see a blue LED in a transparent plastic

10 January & February 2017 www.elektormagazine.com

www.elektormagazine.com January & February 2017 11

12 January & February 2017 www.elektormagazine.com

www.elektormagazine.com January & February 2017 13

Tips and Tricks

Sadly though it is not so simple to connect

Try this adapter I made to make the job eas-

Using the same principle, it should be possible to make a similar

14 January & February 2017 www.elektormagazine.com

CAD for PCBs

PCBs are essential nowadays, even for very small production

www.elektormagazine.com January & February 2017 15

command in the Terminal window, with ”project” replaced by

The KiCad suite consists of the project manager kicad, the

As typical with open source software, the Fritzing download

16 January & February 2017 www.elektormagazine.com

runs as it should on the breadboard and the PCB layout is fin-

In terms of appearance the program is rather plain, and soft-

www.elektormagazine.com January & February 2017 17

An online package that runs in a browser has a number of