£9.95 | €10.95 | $12.

95

514 JULY & AUGUST 2022

ELEKTORMAGAZINE.COM
S

*
1

*
IN
CE 196

AUDIO TESTER
AN AFFORDABLE SOLUTION
Using PC-Based Software
and a USB Audio Interface

p. 06

FOCUS ON

Test &
Measurement
p. 14 AC Grid Frequency Meter
Monitor Mains Frequency and Voltage

p. 30
Get Started With
p. 18
Modest Inductance Meter Your Oscilloscope
A DIY Instrument for Your Find Your Way Through the Knobs
Workbench and Buttons

R2
C2
4k64

C1

R1

+5V
C3
8 1 D2
DRAIN U1 BP
7 2 AMS1117-3.3
DRAIN GND
6 3
F1 D1 1mH DRAIN FB
L1
5 4
L1 DRAIN C/S 2 U2 +3V3
L2
10 / 0W5 R19 R21
HP-01 2
R20

C8 C9 D3 C15 C7 C4 C5
MOV1

4µ7 4µ7 470µ

U 400V 400V 16V

220170-016

Disruption in Test and
Measurement Equipment
p. 48
Alternatives to the Typical
Benchtop Devices
CO2 Traffic Light with Sigfox Smart Plug Teardown
No Wi-Fi Network Needed! p. 74 Which Ones Are Hacker- p. 90
Friendly?
 e-zine
Your dose of electronics

Every week that you don’t subscribe to Elektor’s e-zine

is a week with great electronics-related articles and
projects that you miss!

So, why wait any longer? Subscribe today at

www.elektor.com/ezine and also receive free
Raspberry Pi project book!

What can you expect?

Editorial
Every Friday, you’ll receive the best
articles and projects of the week.
We cover MCU-based projects, IoT,
programming, AI, and more!
Promotional
Don’t miss our shop promotions,
every Tuesday and Thursday we
have a special promotion for you.
Partner mailing
You want to stay informed about
the ongoing activities within the
industry? Then this e-mail will give
you the best insights. Non-regular
but always Wednesdays.
 colophon editorial

Elektor Magazine,
English edition
Jens Nickel
Edition 5/2022 International Editor-in-Chief, Elektor Magazine
Volume 48, No. 514
July & August 2022

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

www.elektor.com
www.elektormagazine.com Measuring Does Not Have to be Expensive
If you measure, you measure crap: Every student of a technical or scientific discipline experi-
Elektor Magazine, English edition ences this at the latest during his or her first relevant internship — and that’s how it should
is published 8 times a year by be. The experience of the user and sound background knowledge cannot be replaced by any
Elektor International Media instrument.
 
Head Office:
Elektor International Media b.v. That’s why we’re once again offering the typical mixture of practice and theory on this import-
PO Box 11 ant topic. My colleague Clemens Valens introduces beginners to the use of oscilloscopes. In
6114 JG Susteren his article “Inexpensive Audio Measurement Station,” Alfred Rosenkränzer combines his
The Netherlands professional specialization in measurement technology with his passion for audio. A good
Phone: (+31) 46 4389444
USB audio interface, a PC and inexpensive measurement software are enough to get a more
Memberships:
than usable solution for measurements on hi-fi systems, (do-it-yourself) speakers and other
E-mail: service@elektor.com equipment. The engineer, who also proves his good judgement with various measurements,
www.elektor.com/memberships only wishes for a notch filter in front of the input of the interface. This will be added in one
of the next issues!
Advertising & Sponsoring:
 
Raoul Morreau
Phone: +31 (0)6 4403 9907
Beyond that there are of course again a lot of projects with practical use. In this issue you will
E-mail: raoul.morreau@elektor.com find a measuring device for the mains frequency as well as one for the series resistance of
capacitors. Plus, an inductance meter that’s easy on the wallet, and much more!
www.elektor.com/advertising
Advertising rates and terms available on request.

Kurt, We Will Miss You!

Copyright Notice It came as a shock to all of us that our long-time colleague
The circuits described in this magazine are for Kurt Diedrich passed away unexpectedly in April. Kurt was an
domestic and educational use only. All draw- editor at Elektor in the eighties; since then, he has remained
ings, photographs, printed circuit board layouts,
programmed integrated circuits, disks, CD-
loyal to us as a contributor for all these decades. A geology
ROMs, DVDs, software carriers, and article texts graduate, he wrote books and articles, translated entire issues
published in our books and magazines (other (most recently our sister magazine MagPi), and brought
than third-party advertisements) are copyright
schematics and other drawings into the famous Elektor style.
Elektor International Media b.v. and may not be
reproduced or transmitted in any form or by any In his spare time, the music lover built analog synthesizers and
means, including photocopying, scanning and picked up the guitar. Kurt, we will miss you!
recording, in whole or in part without prior writ-
ten permission from the Publisher. Such written
permission must also be obtained before any
part of this publication is stored in a retrieval
system of any nature. Patent protection may
exist in respect of circuits, devices, components
The Team
etc. described in this magazine. The Publisher
International Editor-in-Chief: Jens Nickel
does not accept responsibility for failing to
identify such patent(s) or other protection. The Content Director: C. J. Abate
Publisher disclaims any responsibility for the International Editorial Staff: Eric Bogers, Jan Buiting, Stuart Cording, Rolf Gerstendorf,
safe and proper function of reader-assembled Alina Neacsu, Dr Thomas Scherer
projects based upon or from schematics,
Laboratory Staff: Mathias Claussen, Ton Giesberts, Luc Lemmens, Clemens Valens
descriptions or information published in or in
Graphic Design & Prepress: Giel Dols, Harmen Heida
relation with Elektor magazine.
Publisher: Erik Jansen

© Elektor International Media b.v. 2022

Printed in the Netherlands Elektor is a member of FIPP, an organization that has “grown over almost 100 years to
include media owners and content creators from across the world.”

Elektor is a member of VDZ (Association of German Magazine Publishers), which

“represents the common interests of 500 German Consumer and B2B publishers.”

lektor July & August 2022 3

 THIS EDITION
Volume 48, No. 514
July & August 2022

Low-Cost Audio Tester

Using PC-Based Software and
a USB Audio Interface

Regulars FOCUS
3 Colophon
86 Low-Budget Tablet Oscilloscope ADS1013D
25 Starting Out in Electronics Good Value for the Money?
Rectifiers
100 Skin Impedance and Skin Capacitance
28 electronica fast forward 2022 Small Experiments

68 Peculiar Parts FOCUS

Reed Relays 106 Pokit Meter Review
A Swiss Army Knife of Test Gear
84 Ethics
Women in Tech

Industry
104 From Life’s Experience
No Local Business
114 Hexadoku
FOCUS
The Original Elektorized Sudoku
48 Disruption in Test and Measurement Equipment
Innovation from the Smaller Players

Features
22 Acoustic Wave Hovering
A Look at the Makerfabs Acoustic Levitation Kit
FOCUS
30 Get Started With Your Oscilloscope
Find Your Way Through the Knobs and Buttons
FOCUS
54 Infographics
Facts and Figures About Test & Measurement
56 Inspiration, That’s What It’s All About
Interview with Entrepreneur Walter Arkesteijn
of InnoFaith Beauty Sciences
58 Minimizing EMC Interference from Storage Chokes

FOCUS
42 Moisture Sensors for Watering Systems Low-Budget Tablet
Automatic Watering
Oscilloscope
61 GUIs with Python (Part 5) ADS1013D
Tic-Tac-Toe

86
4 July & August 2022
 Get Started With
R2
C2

4k64
C1

Your Oscilloscope
R1

Find Your Way Through the Knobs and Buttons 8

DRAIN U1 BP
1
C3
D2
7 2 AMS
DRAIN GND
6 3
F1 D1 1mH DRAIN FB
L1
5 4
L1 DRAIN C/S 2
L2
10 / 0W5 R19 R21
HP-01 2

Smart Plug
R20

C8 C9 D3 C15 C7 C4
MOV1

Teardown
30 90
4µ7 4µ7 470µ
U 400V 400V 16V

N Which Ones Are Hacker-Friendly?

Projects Next Edition

FOCUS Elektor Magazine Summer Circuits 2022 Edition
6 Low-Cost Audio Tester (August & September​2022)
Using PC-Based Software and a USB Audio Interface In the tradition of the Summer Circuits Guide, next edition will be
extra thick, filled with more than 50 DIY-projects, background articles
FOCUS
about PCB design, tips and tricks and much more!
14 AC Grid Frequency Meter
Monitor Mains Frequency and Voltage
From the contents:
18
FOCUS
A Modest Inductance Meter
> Battery-Free Radiator Sensor
An Affordable Solution for Your Workbench
> Current Source for LEDs
> Tiny Frequency Reference
34 Raspberry Pi Pico Makes an MSF-SDR > Sound Activated Switch
Decode a Time Signal with a Raspberry Pi Pico SDR > Low Noise ADC Calibrator
FOCUS
> Power Semiconductor Tester
70 Simple Analog ESR Meter
> DC/DC Boost Converter
With Moving-Coil Meter Precision
> Digital 220-V AC Dimmer with Arduino
And much more!
74 Sigfox CO2 Traffic Light
No Wi-Fi Network Needed!
Elektor Magazine Summer Circuits 2022 edition will be published
90 Smart Plug Teardown around August 4, 2022. Arrival of printed copies with Elektor Gold
Which Ones Are Hacker-Friendly? Members is subject to transport. Contents and article titles subject
to change.

Acoustic
FOCUS ON
Wave
Hovering
Test &
22 Measurement
lektor July & August 2022
 analog

Low-Cost Audio Tester

Using PC-Based Software and a USB Audio Interface
By Alfred Rosenkränzer (Germany)

Professional-grade equipment for characterizing and developing audio equipment is usually

beyond the means of a hobbyist lab. However, it is possible to obtain measurement software
and a high-quality audio interface for relatively little cash, and in conjunction with a PC
this can create a more than adequate solution for measurements on Hi-Fi devices and other
audio hardware. This article looks at how that works out in practice.

Figure 1: Spectrum of a 1 kHz sinewave at 1 VRMS (0 dbV), generated Figure 2: Compare with Figure 1: spectrum of a 1 kHz sinewave at
and measured using the Audio Precision APx555. 1 VRMS using the Steinberg UR22.

The inspiration for this project began after
I had spent over 20 years working in a
high-tech design company surrounded by
top-notch test equipment, including for audio
measurements. Suddenly this period came to
an end and my options for making a quick test
or measurement became severely limited. I
was rudely separated from the professional
equipment that I had been used to, not just by
the distance between home and workplace
but also by the effects of the coronavirus.
Considerations
A search on eBay for second-hand profession-
al-level test equipment quickly revealed the bad
news. Even units more than ten years old, long
out of manufacturer’s support, were going for
over a €1,000. New equipment with the specifi-
cations I was looking for started at a little under
€10,000, which seemed totally out of propor-
tion to what I was trying to achieve with them. I
therefore had to try to come up with a different,
and above all more affordable, solution.
Most professional test equipment of the early
third millennium more or less takes the form
of a black box, with no user controls. The
unit is controlled and its results processed
using software running on a PC. Why should
things be any different in the hobby world? So
I started off by looking for suitable software
for making audio measurements.
 
My requirements for the software were that
it should work with both internal hardware

6 July & August 2022 www.elektormagazine.com


Figure 3: Front panel of the Steinberg UR22C.
Figure 4: Compare with Figures 1 and 2: spectrum of a 1 kHz sinewave
at 1 VRMS using the Steinberg UR22C.
and with external hardware connected over
USB. I wanted it to be able to measure total
harmonic distortion (THD) and signal-to-noise
ratio (SNR) as well as frequency response.
Furthermore, the signal generator part of
the system should at a minimum be able
to produce sine, square and noise signals.
It must be possible to sweep the sine wave
output in frequency. Other waveforms such
as triangle and pulse were not a sine qua non,
but nevertheless nice to have. These options
are straightforward to implement in software
and so add little to its complexity or cost.
 
The software must be able to acquire a signal
from the audio input and display it. The avail-
able frequency range must cover at least the
audio band from 20 Hz to 20 kHz, and a wider
bandwidth is desirable. Digitization should be
possible at sample rates of 44.1 kHz, 48 kHz
and their multiples up to at least 192 kHz.
Software
During my search, the software package
AudioTester V3.0 caught my eye. It runs under
Windows XP and later versions, and can be
downloaded for free at [1] along with its
manual, available in both English and German.
The free trial version offers all the features of
the software for one month, the only limitation
being that the program automatically termi-
nates a measurement after ten minutes. To
continue, the software must be restarted. Since
it remembers its most recent settings, you can
carry on immediately from where you left off.
 
A licence costs a reasonable €39. After experi-
menting for a couple of days, I bought a licence
from the developer and received a key file to
unlock the software. At this point my search for
a suitable software solution was over. Perhaps
you will find other useful programs? If so, feel
free to get in touch with me via e-mail.
Hardware
For my first tests, I used my PC’s internal
audio hardware in loopback: I connected a
mini-jack cable directly from the audio output
to the line input (not the microphone input).
Since this was using the audio circuitry built
into the PC’s motherboard, my expectations
were not exactly high. The results, however,
were not too bad, although not good enough
for my purposes.
Next, I started to look for a higher-quality
external audio interface with a USB connec-
tion. I wanted a device with XLR connec-
tors for differential signals (‘balanced’) and
phono or jack sockets for signals referenced
to ground (‘unbalanced’). The BNC connec-
tors typically found on instrumentation are
rather rare on this kind of interface. That is
not an obstacle, however, as suitable adapt-
ers are available. A headphone output makes
it easy to verify the audio signal. Input and
output levels must be controllable either on
the interface itself or at a minimum using the
accompanying software.
Here we at once see what is perhaps the
biggest disadvantage of this type of solution
compared to its professional counterparts,
whose input and output levels are calibrated
in V or dBV. If we set a given level in the
software then that is the level that comes out
of the sockets of the generator. The software
controls the device to generate the desired
level at the best possible quality. The input can
be configured either automatically or accord-
ing to the expected input level.
Levels
The AudioTester software allows the output
level to be set in dBFS (dB full scale) — that
is, in decibels relative to the maximum output
level of the DAC. The actual level at the output
of the interface must then be set either directly
on the interface itself or using its accompany-
ing software. That requires a way to measure
the level externally (for example, with an
oscilloscope). Unfortunately, the AudioTester
software cannot set the output gain of the
interface directly.
 
This difficulty also affects the input. Towards
the top of the window, there is a kind of drive
level indicator for the ADC. By adjusting the
gains on the interface or in its software, we
can make sure that the ADC is adequately
driven, but not overdriven.
Signal Generators
I first measured the output of the internal
sound card using a professional instrument
(an Audio Precision APx555). Figure 1 shows
the output from the APx555 in purple: there
are no harmonics to be seen and the noise
floor is at around –150 dBV. The blue plot
shows the output of the sound card. The
lektor July & August 2022 7
strongest harmonic is at 3 kHz with an ampli-
tude of –98 dBV. The noise floor is some 20 dB
above that of the APx555. These results are
remarkably good for an internal sound card.
 
That statement is particularly true in compar-
ison to the Steinberg UR22 external interface.
As Figure 2 shows, the 3 kHz harmonic from
this device has a higher amplitude at –87 dBV.
Figure 5: RME Babyface Pro
 
(Photo: RME).
I borrowed the newer version of this interface,
the UR22C, for a further comparison. Figure 3
shows its front panel and the results of the
measurement can be admired in Figure 4. The
pattern of harmonics observed depends on
the sample rate chosen for the ADC. Further
tests, including on a Zoom R16 interface,
resulted in similar spectra which need not
be displayed here.
 
The main conclusion from this is that the
results from external interfaces are not exactly
convincing. I therefore consulted some techni-
cally-inclined friends, and the RME Babyface
Pro FS [2] was recommended to me. The
Steinberg device costs around €150; the
RME interface is much more expensive, at
€709. Figure 5 shows the appearance of this
interface; further information can be found,
for example, on Thomann’s website [3].
Figure 6 shows what you get for your money: Figure 6: Compare with Figures 1, 2 and 3: spectrum of a 1 kHz
the second harmonic is at –120 dBV! At a sinewave at 1 VRMS using the RME Baby Face Pro.
reduced output level of 0.25 VRMS it disap-
pears into the noise.

Audio Inputs
The second step was to connect the
high-quality output signals of the APx555 to
the audio inputs of the RME interface, display-
ing the result using AudioTester (Figure 7).

Figure 7: 1 kHz sinewave at 1 VRMS generated by

the APx555 and measured using the input of
the RME interface and the AudioTester software.
Towards the top of the window is the drive level
indicator for the ADC.

8 July & August 2022 www.elektormagazine.com


Figure 8: Loopback on the RME Babyface Pro at a sample rate of 192 kHz.
Figure 9: Loopback on the Steinberg UR22 at a sample rate of 48 kHz.
Figure 10: Frequency response of the RME Babyface Pro at a sample rate of 48 kHz.
The spectra obtained using the same test on
the internal sound card and on the Steinberg
interface are poorer and not shown here.
Next, some loopback measurements were
taken between the inputs and outputs of the
same interface. Figure 8 shows the behaviour
of the RME interface and Figure 9 that of the
Steinberg UR22. In the interest of fairness,
we should remark that the spectra change
depending on the sample rate and drive level
of the ADC, but one thing remains constant:
the RME interface performs better under all
settings.
To test the frequency response, we can
generate a white noise signal and examine
the resulting spectrum, or we can use a
sinewave frequency sweep. Figures 10,
11 and 12 show the frequency response of
the RME interface at sample rates of 48, 96
and 192 kHz, all measured using white noise.
From Figure 12, we can see that the response
is flat out to approximately 90 kHz, which tells
us that measurements well beyond the normal
audio range are possible.
Operation
The RME interface can be controlled to some
extent using its buttons and control knob.
Unfortunately, it is not possible to enable the
48 V phantom power supply or the attenuator
pad on the inputs in this way.
For full control of the interface, it is neces-
sary to install the ‘RME TotalMix’ software.
This software will automatically be launched
when the interface is connected to the PC.
This software allows us to set not just levels,
but also includes a parametric equalizer and
more besides. It is essential to read the
manual, and on YouTube there is a number
of videos explaining the fine points of the
interface.
Figure 13 shows the user interface of the RME
software. The dialogue box shown in Figure 14
lets you configure the basic settings such as
choice of input, sample rate and so on. This
latter value must agree with the sample rate
configured in the AudioTester software, or else
errors or possibly unwanted tones will appear
in the spectrum.
lektor July & August 2022 9

Figure 11: Frequency response of the RME Babyface Pro at a sample rate of 96 kHz.
Figure 12: Frequency response of the RME Babyface Pro at a sample rate of 192 kHz.
 
The RME interface can be powered over USB
or using an external 12 V power adapter. Since
the current draw over USB is around 500 mA,
the use of an external supply is recommended
when using a laptop to reduce the load on
the USB connection. If an external switching
supply is used its switching frequency should
be above the audio range.
Using the Measurement
Software
After installing and launching the software,
select 2D FFT in the upper selection box
and the screen should resemble Figure 8.
The Options menu item can now be used to
select the interface that will be used for input
and output. If the internal sound card of the
PC is not shown, then try plugging cables into
the relevant sockets. If several interfaces are
available, different ones can be chosen for
input and output.
 
Finally, under Options, you can configure
parameters such as the sample rate and
channel count for input and output. The Setup
button just to the right of Sound on can be
used to select the waveform, the DAC drive
level, the frequency and in some cases other
parameters. Sound on enables the output.
The output level can now be checked using
an oscilloscope and the level settings on the
interface adjusted as desired.
The Start button begins a measurement.
The adjacent Setup button allows a choice
of various window functions and FFT lengths.
Increasing Averages reduces noise at the
expense of increased measurement time.
In the Analyze Dialog under Analyze, you
can configure automatic measurements. For
example, you can define how many harmon-
ics are taken into account in the computation
of THD. The results of the measurement are
displayed under the graph.
We will not look into the 3D FFT Waterfall Plots
menu item. The third button enables Sweep
Measurements (for example for frequency
response measurements), with parameters
being configured under Setup. Right-clicking
on the graph will change the display.
A practical application is shown in Figure 15.
Here we are determining the frequency
response of two Butterworth filters (a 1.2 kHz
lowpass and a 2 kHz highpass). Figure 16
shows the hardware setup. The single-ended
connections on the right-hand side of the
RME interface are used.
 
The fourth menu item, Impedance Magnitude,
lets you display impedance as a function of
frequency, for example of a loudspeaker with
or without a crossover circuit. Resonances
are shown very clearly. For such a measure-
ment a power amplifier is of course required
to drive the loudspeaker and the two inputs
must be connected, along with a power

WEB LINKS
[1] Software AudioTester: http://www.audiotester.de
[2] Audio interface: https://www.rme-audio.de/babyface-pro-fs.html
[3] Information on the RME interface: https://www.thomann.de/gb/rme_babyface_pro_fs.htm

10 July & August 2022 www.elektormagazine.com

 resistor, according to the diagram in the
documentation.
 
The Thiele Small Parameter Measurement
menu item will not be discussed here. The
next item, Distortion Measurement vs.
Frequency can be used to see how distor-
tion varies as a function of frequency.
 
The next menu item is dedicated to measur-
ing distortion as a function of amplitude. Here
we need to take note that the AudioTester
software can only control the DAC drive level;
it does not have access to the level control
of the audio interface. At low levels, there-
fore, the digital resolution will be reduced,
which increases distortion. Before making
measurements on a circuit you should test
the behaviour of the system using loopback.
Alternatively, you can adjust the level manually
in the interface and keep the digital drive level
of the DAC constant.
 
As well as lacking well-defined absolute
output levels, the audio interface also omits
an important element that practically all
professional audio instrumentation includes:
an adjustable notch filter. This can be used
Figure 13: Screenshot of the user interface of the RME Babyface Pro software. to attenuate the fundamental of a signal and
thereby analyze the other parts of the signal

Figure 14: Basic settings of the Figure 15: Frequency responses of a 1.2 kHz lowpass filter and a 2 kHz highpass filter
RME Babyface Pro. (both Butterworth), logarithmic frequency axis.

lektor July & August 2022 11


(harmonics, non-harmonic tones and noise)
to greater accuracy. It also helps to reduce
the harmonics introduced by the ADC itself
when driven at a high level. An external notch
filter could be added if these advantages are
important in your application.
 
Within the confines of this article it is of
course impossible to do more than scratch the
surface of the capabilities of the AudioTester
software. To find out more you can experi-
ment or you can delve into the documenta-
tion. And you can try out everything you need
in advance, and the licence is in any case
reasonably priced, so you have nothing to lose.
Measurement Features
The AudioTester V3.0 software offers all the
measurement features that I need, and a
few more besides. It seems stable and there
appear to be few bugs. Everything can be
tried out before purchasing a licence, which
at €39 is very good value. It is compatible with
both internal sound cards and all the external
USB audio interfaces that I was able to try.
The RME Babyface Pro is indeed considerably
more costly than the other interfaces, but the
12 July & August 2022 www.elektormagazine.com
quality of its signals and of the measurements
I made with it is convincing.
200604-01
Questions or Comments?
If you have technical questions feel free
to e-mail the author at
alfred_rosenkraenzer@gmx.de
or the Elektor editorial team at
editor@elektor.com.
Figure 16: Test setup using the
RME interface with external 12 V
supply and the two filter boards.
About the Author
Alfred Rosenkränzer worked for many
years as a design engineer, originally in
the field of professional television technol-
ogy. Since the end of the 1990s he has
been developing high-speed digital and
analogue circuity for IC testers. Audio
technology is his private passion.
RELATED
PRODUCTS
> The Complete Linear Audio Library
(USB Stick) (SKU 19672)
www.elektor.com/19672
> Designing Audio Power Amplifiers
(2nd Edition) (SKU 19150)
www.elektor.com/19150
 Join the
Elektor C mmunity
Take out a
The Elektor web archive from 1974!
6x Elektor magazine (Print)
9x Digital (PDF) including
Elektor Industry (EN) magazine
A 10% discount in our web shop and
exclusive offers
Elektor’s annual DVD-ROM
Also available
The Digital G R E E N membership
membership!
www.elektor.com/member
G O LD
membership
membership!
An online Elektor LABs account, with
access to more than 1000 Gerber files
and a direct line of communication with
our experts!
Bring a project to publication or even sell
it in our shop
Access to Elektor’s web archive
10% discount in our web shop
6x Elektor magazine (PDF)
Exclusive offers
Access to more than 1000 Gerber files
 project

AC Grid Frequency Meter

Monitor Mains Frequency and Voltage

r- L a b s . c o
to
m

Figure 1: Dieter
k
Ele

Laues’s design
with its
OF grid frequency
BEST
LED bar displa
y.
m
co

le s.
E

kto
r- L a b

By Mark Stevenson (United Kingdom)
The frequency of the mains
voltage is a good indicator
of the grid’s load. This grid
frequency meter not only
monitors this important
parameter of our AC line
supply, but also the mains
voltage.
 
A few years ago, I saw an article in Elektor
by Dieter Laues which showed how you
can determine the load on the National
Grid electricity supply by measuring the
frequency of the supply [1]. In short: the
greater the load, the more it slows down the
generators and so the lower the frequency
of the supply; the lighter the load, the
faster the generators and the higher the
frequency.
The National Grid is compelled in the UK
to maintain the frequency of the supply to
±1% of the nominal frequency of 50 Hz,
these limits being 49.5 Hz and 50.5 Hz.
However, it aims to keep this to within
±0.2 Hz normally (49.8 Hz ... 50.2 Hz). A
very large power cut across parts of the UK
was caused by these limits being exceeded.
Two large generating plants failed almost
simultaneously and the sudden loss of
supply caused the remaining power plants
to become excessively loaded, lowering
the frequency. To keep within the operat-
ing limits, large parts of the country were
essentially unplugged to allow the remain-
ing plants to continue to work. Reserve
power was brought online and power was
eventually returned.
 
I built the original meter (Figure 1) when
the project was just published. Some years
later, I decided to make my own version
with a Microchip Technology PIC microcon-
troller instead of the AVR in Dieter’s project.
The LED bar from the old design is replaced
with an LCD, and my meter also displays the
mains voltage. It should be noted however
that this device shows trends of the power
grid values. It should not be regarded as a
highly accurate device measuring mains
voltage and frequency.
Hardware
The schematic diagram of the grid
frequency meter is shown in Figure 2. An
AC plug-in power pack of 9 V ... 12 V RMS
AC is connected to K1. C1 filters any high
frequency noise. D2 half wave rectifies
the supply to 5 V regulator IC1, two 470 µF
capacitors smooth the supply. C5 (and C2, C3)
supply the processor for long enough in the
event of a power failure to record the loss of
voltage in EEPROM. R5, R8 and R9 pull up
the NO momentary calibrate switch and the
120 V/220 V solder links to 5 V. If neither of
the links are soldered, the voltage scaling

14 July & August www.elektormagazine.com

defaults to 240 V. So the output voltage of
the AC power pack is used as a measure
for the mains voltage; not a highly precise
measurement, but safe and accurate enough
to monitor AC grid voltage variations.
 
Both the voltage and frequency inputs are
fed from their own half wave rectifiers, D3
and D4. D3 is connected to the AC input
through R1, which in conjunction with
D5 and R10 limits the maximum voltage
applied to the Volt_In (Pin 3) pin of the
microcontroller. Ten-turn trimmer R11
allows the voltage to be adjusted across
a reasonable range, R10 provides some
additional attenuation when the trimmer is
at its highest value. D4 connects to the base
of the pulse shaping transistor through R3.
C4 filters any noise on the AC supply, this
does cause some sloping of the pulse, but
this is compensated for within the software
by measuring both the on and off periods
of the pulse.
 
D1, R2, R4 and C8 provide a supply for the
LCD backlight. This half wave rectified
voltage can cause the backlight to flicker
when viewed close up. If this becomes Software
irritating, C8 could be increased. R2 and A new design needed new software. After
R4 should be adjusted to provide the correct some thought, I found a way to measure
current limiting for the LCD fitted. the period of a pulse using a PIC device on
  the Great Cow BASIC forum [3]. I used the
The microcontroller (IC2) is a PIC16F1825 Open Source compiler for 8-bit PICs, AVRs
running at 32 MHz. The LCD is connected in and LGTs Great Cow BASIC to write the code
4-bit mode, with R/W tied to ground perma- [4]. The source code and the HEX-file for
nently (write only mode). R13 is a trimmer programming the PIC16F1825 are available
for contrast adjustment of the LCD. for download on this project’s Elektor Labs
page [2].
Construction  
The circuit design is simplicity itself and it The period of the time that the frequency
contains standard, through-hole compo- input pin is on for is added to the period
nents only. It should not be difficult to build of time that it is measured to be off for.
on prototype board or breadboard if you The reason for this is that the transistor
don’t want to have a PCB made. Several PCB that is used to shape the half cycle derived
layouts were made for this project, most from the AC power supply gives a slightly
photos in this article and on Elektor Labs asymmetrical on pulse. By measuring the
show a version with an extra large LCD that (equally asymmetrical) off pulse and adding
I took out of a discarded device; the bottom them together corrects for this. A timeout
(copper) side of the board is used as front is provided so that, should the incoming
panel for the grid frequency meter. There supply fail due to a power cut, the meter
are Eagle design files and gerbers available does not lock up waiting for the next on
for download on Elektor Labs [2] for either or off pulse to be counted. The frequency
standard 16 x 2 LCDs or the larger LCM1602 is averaged for 20 counts. This makes the
version. displayed frequency far more stable. The

+5V Vin

DIS1 LCD 16x2

D1
Vin +5V
R13
10k
CONTR

1N4007
LED+
LED–
GND
VCC

R/W

+5V
RS

D0
D1
D2
D3
D4
D5
D6
D7

R2
E

D5
R1 R6 R8 R9 R5
47

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
3k3

10k

10k

10k
1k5

C7
100n
R4
1N4148
D3 D4 +5V
47

1 14
VDD GND
2 13
1N4148 1N4148 RA5 RA0
3 12
RA4 IC2 RA1 C8
4 11
C6 RA3 RA2
R11 5 10
RC5 RC0 470µ
10k 6 9 25V
100n RC4 RC1
7 8
RC3 RC2

T1
+5V Vin
120V 220V S1 PIC16F825 D2
R12 R3 R7 R10 7805DT 1N4007
47k

2k2
1k5

10k

IC1
C4 K1
BC547 CALIBRATE C5 C3 C2 C1
100n 9...12V AC
470µ 470µ 470µ100n
25V 25V 25V

200312-008

Figure 2: This project’s schematic diagram.

lektor July & August 2022 15


Figure 3: The measurement display under normal conditions.
software is designed to automatically detect
a nominal 60 Hz supply frequency and
calibrate the bar graph display to centre
around 60 Hz. As I have no accurate means
of generating 60 Hz, it remains untested.
 
Once the frequency has been measured, the
supply voltage is tested. The voltage applied
to pin 3 is a raw half wave, un-filtered AC
waveform. This is sampled 50 times at
1 ms intervals with the peak reading used.
The peak reading taken is then scaled
up by a factor to show the approximated
RMS supply voltage. The scaling factor for
this can be selected by soldering a pair of
jumpers as noted on the circuit diagram.
This allows for a nominal supply of 120 V,
220 V or 240 V to be catered for. A 10-turn
trimmer resistor allows for the use of an AC
power supply of 9 V ... 12 V RMS to be used.
With the correct jumpers soldered, a peak
voltage of 4 V applied to pin 3 should show
the correct supply. The voltage is averaged
for eight counts to stabilise it and give some
degree of noise immunity.
What the Display Shows
Using the most common 16 x 2 LCD, under
normal conditions, the left side of the top
line shows the grid voltage, the frequency is
displayed across the six rightmost charac-
ters, as in Figure 3. The bottom line of
the LCD displays a bar graph which shows
segments lit progressively from the centre.
Segments lit from the centre towards the
right show that the frequency is above the
nominal value, from the centre towards
the left show that the frequency is lower
than the nominal value. Each segment
represents a 25  mHz step. With eight
segments in each direction this allows a
display of ±200 mHz. When the frequency
is very close to the nominal value the centre
two elements each show a large square
block character, as the frequency moves
away from the nominal value, the left or
16 July & August www.elektormagazine.com
Figure 4: Undervoltage event display.
right of centre block changes to vertically
extended “+” character, pointing upwards
or downwards in frequency as required.
 
The centre two characters of the top line
are used for “Event” reporting. If the supply
voltage or frequency exceeds a limit (±5%
for the voltage, ±0.6% for frequency) an
event flag is shown, along with an indica-
tor to show the reason for the event. For
example, a downwards arrow is shown for
low supply voltage, an upwards arrow for
excessive supply voltage.
 
When an event has occurred, once a minute
or so the approximate time elapsed since
the event happened is displayed in the
unused portion of the bar graph display
area. This is shown as hours and minutes for
the initial 24-hours, then days and hours.
For example, the display in Figure 4 shows
an undervoltage event flag, a symbol that
mimics a plug being removed from its
socket, followed by downwards arrow. The
second line shows the event occurred one
minute ago while the grid frequency was
above nominal value: the time indication
is on the left and the bar graph occupies
the right side of the line. Had the frequency
been below the nominal, the event time
would be positioned in the right hand side
of the bar graph area.
 
The voltage events are automatically
cleared after seven days, frequency events
are cleared after ten minutes. Events can be
manually cleared by pressing the Calibrate
button briefly. The software will attempt
to store the most recent event in EEPROM.
This is most likely to be a low voltage event.
When power is removed, the electrolytic
capacitors on either side of the 5 V regula-
tor should maintain a supply to the proces-
sor for at least a few milliseconds longer
than that provided to the frequency or
voltage measurement pins of the proces-
sor. If the supply dips or drops for more than
500 ms, this event should be captured and
recorded. If the power cuts out totally, the
event should be recorded, when the power is
returned this will be shown, with the timing
starting from the time that power returned.
Calibration
The processor used has its own internal
resonator. This is reasonably accurate, but
when trying to measure frequencies down
to millihertz levels of accuracy, different
devices can show some variation. Calibra-
tion is achieved by adding a small amount
of correction to the timing obtained by
measuring the on and off times of the pulses
derived from the AC input voltage. The
correction factor is stored in EEPROM and
needs to be be adjusted only once. Calibra-
tion is done by pressing and holding the
Reset/Calibrate switch S1 down. The initial
calibration value is shown for a few millisec-
onds, then the value starts to step upwards.
The range of adjustment is from -99 through
to +99. After the calibration is increased
beyond +99 it wraps back to -99. A single
step of adjustment adds or subtracts a value
equivalent to ±250 µHz (microhertz!) the
full range of adjustment being just under
±250 mHz (millihertz). If you have access to
a stable signal generator capable of resolu-
tion (and accuracy) down into the low milli-
hertz, you could use that, with D4 removed
and the signal fed directly to pin 5 of IC2.
More practical is to wait until the early hours
of the morning (2 AM to 3 AM) and assume
that the frequency and load on the grid is
balanced and then set the meter to read
the nominal grid frequency, either 50 Hz
or 60 Hz. This is my favoured approach
and least likely to end in frustration! If set
in this fashion, I would suggest keeping an
eye on the frequency noting whether it is
predominantly above or below the nominal
frequency, making slow, small adjustments
until it is.

 
The voltage reading can be calibrated
using any reasonably accurate multime-
ter, with cautions noted regarding the
safety requirements of measuring lethal
voltages. If you are not confident in measur-
ing high voltage, you could wait until the
early hours of the morning again then set
the trimmer until the voltage displayed
shows the nominal voltage.
 
There are a few comments on Elektor Labs
regarding the accuracy or relevance of using
a processor with an internal resonator as
opposed to an external crystal. I have found
the design to be validated in usage, I have
recently been able to build my own version
of Dieter Laues’ design which uses a crystal
and ran both “mine” and “his” next to one
another for a number of months. At no
time did I feel that I needed to consider
redesigning mine to use a crystal derived
oscillator.
200312-01
Mains Grid Frequency Online COMPONENT LIST
If you are want to see or even follow the
Resistors
mains frequency without buying or building
R1,R12 = 1k5
your own meter, there are some websites R2,R4 = 47 Ω
that show interesting information about the R3 = 2k2
grid. In the UK there used to be a number R5,R8,R9,R10 = 10 k
of online resources that showed the current R6 = 3k3
frequency; however, often looking at these R7 = 47 k
would show that some would lag behind R11 = 10 k trimmer, multiturn
the other and there was almost never a R13 = 10 k trimmer
time when they would agree. Worse, most  
now seem to either have been taken down Capacitors
or rely on outdated Adobe Flash technol- C1,C4,C6,C7 = 100 nF
ogy that won’t load in a modern browser. C2,C3,C5,C8 = 470 µF, 25 V radial
When researching this article, one site was
found that seemed to be accurate in the Semiconductors
UK [5], but it doesn’t seem to automati- D1,D2 = 1N4007
cally refresh, so the page needs constant D3,D4,D5 = 1N4148
updating to get the current value. T1 = BC547
IC1 = 7805
IC2 = PIC16F1825
Miscellaneous
S1 = pushbutton NO
DIS1 = alphanumeric LCD 2 x 16
characters

Questions or Comments? RELATED

If you have technical questions or
comments on this article, feel free to
e-mail the Elektor editorial team at
editor@elektor.com.
For Europe, one was found that works well PRODUCTS
and looks to be constantly updated but we
can’t vouch for its accuracy [6].
> SmartPi 3.0 Smart Meter for
Raspberry Pi (Bundle) (SKU 19953)
www.elektor.com/19953  

> PeakTech 2715 Digital Loop / PSC

Tester (SKU 19078)
www.elektor.com/19078  

> PeakTech 3445 True RMS Digital

Multimeter with Bluetooth
(SKU 18774)
www.elektor.com/18774

WEB LINKS
[1] Dieter Laues, “Grid Frequency Monitor,” Elektor Magazine 01/2012:
https://www.elektormagazine.com/magazine/elektor-201201/19793
[2] This project’s page on Elektor Labs: https://www.elektormagazine.com/labs/electricity-grid-frequency-meter-v2
[3] W. Roth, “Measuring Pulse Width To Sub-Microsecond Resolution with PIC Capture & Compare,” Great Cow BASIC, 2015:
https://bit.ly/measuring-pulse-width
[4] Great Cow BASIC compiler: http://gcbasic.sourceforge.net/Typesetter/index.php/Home
[5] Mains frequency UK online: http://mainsfrequency.uk/fm-last10m
[6] Mains frequency EU online: https://www.gridwatch.templar.co.uk

lektor July & August 2022 17

 project

A Modest
Inductance Meter
An Affordable Solution for Your Workbench le
kto
r- L a b
s.

co
E

m
BEST
OF

Ele
to

m
r- L a b s .c o

k
Vcc
By Philippe Le Guen (France)

Need an instrument for testing R1

L1+Lx
and measuring inductors? This
do-it-yourself inductance meter
is an affordable alternative to Q1
C6

many of the expensive solutions

on the market. 
R2 R3
C2 C7

A good instrument to test and measure inductors has been on my

lab-equipment wish list for many years now, but the cost of such a 200592-30

device has been prohibitive. I therefore decided to build something

myself. The little instrument presented here cannot compete with the
complex all-singing, all-dancing instruments on the market; it is only
a modest inductance meter that enables me to obtain the value of
an unknown inductance. There are many of such devices available
online, sometimes based on Arduino.
 
Some years ago, I discovered F. Kudelsko’s website [1], which describes
a small DIY inductance meter capable of measuring inductances from
a few tens of nanohenries to approximately 10 mH. A small Windows
Figure 1: An LC tank circuit determines the
frequency of the Colpitts oscillator.
program retrieves the value of the inductor under test over USB and
displays it. Even though I found this approach quite interesting, I was
looking instead for a stand-alone device. Yet I would like to thank the
author here for sharing his work on which I based my design.

18 July & August www.elektormagazine.com

 +5V +5V

K1 LCD1
C4 R11 1 1
P1 VSS
2 2
10k

10
VDD
100n 3 3
VO
4 4
RS
+5V 20 5 5
R/W

LC DISPLAY 2x16 SIL

6 6

Principle of Operation
E
1 21 7 7
MCLR RB0 DB0
The operating principle of the inductance
22 8 8
IC2 RB1 DB1
2 23 9 9
meter is quite simple. The unknown 3
RA0/AN0
RA1/AN1
RB2
RB3
24 10 10
DB2
DB3
inductance is used to create an LC 4
RA2/AN2 RB4
25 11 11
DB4
5 26 12 12
oscillator. By measuring the oscillator’s 6
RA3/AN3 RB5
27 13 13
DB5

frequency, the value of the unknown

RA4/T0CKI RB6 DB6
7 28 14 14
RA5/AN4 RB7 DB7
inductance can be determined. 15 15
PIC18F252-I/SP 16
A
11 18 16
  12
RC0 RC7/RX
17
+5V
C
RC1 RC6/TX
There are many ways to build an LC 13
RC2 RC5/SDO
16
R18
LCD

oscillator. Here a Colpitts oscillator is

14 15
RC3/SCL RC4/SDA

220
used (see Figure 1). The circuit uses a OSC1 OSC2
+5V
common-base transistor amplifier (Q1)
8 9 10 19
X1 S1
R5 R13 R12
with the input on the emitter and the Measure

output on the collector. Without going

1k

1k

1k
C3 8MHz C1
R17
into details on how this oscillator works, LED1 LED3 LED2

10k
its output frequency is given by the
22p 22p

Zero Error Power

formula of Thomson (a.k.a. Lord Kelvin):
 
f = 1 / (2π √(LC))
  +5V +5V +5V +5V

After reworking it, we obtain: R6

16 R10
  L1
TP2
10k
VDD
C9

5k6
9
L = 1 / (4π² f²C) R1 47µH 0
C5 CTR12 7 100n
1
 
10k

11 6
CT=0 2
Here C is the value of the two capacitors
3n3 REL1 D1 5 Q3
5% 3 R8
Lx IC1 3
C6 and C7 in series: HE722 1N4148 HEF4040B
4
5
2
10k

  CT 6
4 BC547B

C = C6 × C7 / (C6 + C7) C6 13
7
TP1 10 12
+
 
8
Q1 Q2 14
100n R4 9
With the value of these capacitors being
5% 15
220k 10
1
known — I measured them with my BC547B C2 BC547B GND
11

capacitance meter for more precision in C2

R2 R3
C7 1n
R7 8

the calculations — we only have to plug

820

100
12k

C into the reworked formula (together

10µ 220n
50V 5%

with the measured frequency) to find the 200592-028

value of L. It is recommended to use 5%

MKT-types. Figure 2: A prescaler (IC1) divides the oscilator frequency set by Lx down to something the
MCU (IC2) can digest.
Schematic Diagram
The circuit of the inductance meter
is shown in Figure 2. I replaced the  
PIC18F2550 microcontroller in the origi-
nal schematic [1] with a PIC18F252, as I didn’t need a USB interface, With my component values (and Lx short-circuited by the relay,
and I added a 2 × 16 alphanumerical LCD. Some component values see below), the theoretical operating frequency of the oscillator is
were determined by what I had available. 83.821 kHz (86.488 kHz with ideal components). Measuring it with an
  oscilloscope showed a frequency of 88.652 kHz, a difference of 5.4%.
Note C5: This part was inherited from the original design [1] and has  
a slight influence on the oscillator frequency. With C5 the C to use in All the calculations are done by the microcontroller. It measures the
the calculations is given by: frequency of the oscillator via its port RC0. For this it is essential
  to divide the frequency down to values acceptable for the MCU. A
C = C5 + C6 × C7 / (C6 + C7) 4040-type binary counter (IC1) takes care of this. The division ratio
  used here is 1/32. I used an HEF4040B for IC1 because that was
C5 should be of the same precision and quality as C6 and C7. what I had in stock, but a 74HCT4040 will work too.

lektor July & August 2022 19

 R14

0 22
0W5
6 7 8
D2 D4 VCC IS DC +5V
J1
1
SWC L2
2
4x 1N4007 SWE
220µH R15
IC3
7...23V AC MC34063AP

3k
D3 D5 5 3
9...32V DC FB TC
1%
GND D6 R1
C10 C11 C12
4

1k
330µ 220p SB120 330µ
25V 25V 1%

200592-29

Figure 3: The switched-mode power supply allows for a wide input voltage range.

L1 and the Relay
To ensure the fast starting of the oscillator irrespective of the value of
the unknown inductance Lx, another inductor L1 is placed in series
with it. When taking a measurement, the frequency of the oscillator
is determined by these two inductances. However, we only want to
know the value of Lx. Therefore, the software first takes a measurement
with Lx short-circuited. This value, the zero reference, is memorized
and used later to calculate the value of Lx.
The Power Supply
The power supply is based on an MC34063 (IC3), a switching voltage
buck regulator that delivers a nice supply voltage of +5 V (Figure 3).
The power source applied to the input of the circuit is first rectified
and then filtered by capacitor C10 before entering IC3. The reason
for the rectifier D2..D5 is to allow both AC and DC power supplies
without having to worry about their polarity. This way the device
accepts 7 VAC to 23 VAC or 9 VDC to 32 VDC at its input. Its working
frequency is about 30 kHz. The power consumption of the complete
circuit is only 35 mA, not a problem for this power supply that can
deliver a maximum current of 1.2 A. The power supply connector’s
center pin is 2 mm in diameter, which allows for the use of many
standard power adapters.
Firmware
I developed a small program in mikroC [2] that, like the original design,
allows for the measurement of any inductance, and displays its value
(in nH, µH or mH) on the LCD. The measurement range selection is
automatic. Measuring the value of Lx is done in two steps:
 
1. System calibration sequence (relay closed, Figure 4).
2. Measurement sequence with display of the calculated value.
The MCU’s Timer1 is configured to start counting on the first rising
edge on RC0; it will count for one second (controlled by Timer0). After
one second, the value of Timer1 represents the frequency of the signal
present at its input. Now the processor can perform all the calculations
necessary to determine the value of either L1 (Step 1) or Lx (Step 2).
The LCD shows what is going on. During Step 1 the yellow LED (LED1,
Zero) lights up. If the inductance to be measured is disconnected, then
the message “no coil detected” is shown and the red LED (LED3, Error)
will be lit. If this happens check that the coil is correctly connected
and restart the measurement sequence by pressing the push button.
 
The use of the system is simple and comfortable as it is fully automatic.
The only thing to do is to press briefly (approx. 0.5 s) the push button
connected to RC5 to restart the procedure if needed. The display
shows an arrow to indicate a new measurement.
The Building Process
Once the prototype was validated, I designed a double-sided printed
circuit board for it [2]. It is intended to fit into a Multicomp MCRH3135
case (Figure 5). Your trained eye may have noticed the small size of
the pads. Most of them have a diameter of 1.4 mm for a 0.8 mm hole. I
therefore highly recommend using a good soldering iron for mounting
the components or reroute your board differently.
 
Even though I did my best to come up with a good PCB design, it
ended up having two issues that I did not fix:
> I forgot to check the dimensions and the shape of the pads of
power connector J1. If you use the same model as I did, you will
have to rework the PCB a bit with a Dremel-like tool (Figure 6).
As the connector is no longer correctly maintained by its
soldered connections, I glued it to the board (after soldering).
> There is no ICSP connector for in-circuit programming of the
MCU, which is rather impractical. It is, however, possible to
program the MCU by removing it from the board and placing it
on a breadboard or prototyping board.
The LCD is mounted on a single-row, 16-way connector, and it is firmly
fixed to the PCB with four M3 × 10 mm nylon spacers and suitable
bolts and nuts.
 
The ICs are mounted on sockets which are, of course, not absolutely

20 July & August www.elektormagazine.com

Figure 4: To ensure the best accuracy every measurement is done in two Figure 5: The circuit board for the inductance meter fits snuggly inside the
steps. Shown here is Step 1, the system calibration sequence. enclosure.

necessary (unless you want to reprogram the microcontroller...). The

relay is soldered directly on the PCB, mainly to avoid unwanted parasitic
capacitances and inductances. The three LEDs are soldered almost flush
with the PCB, I used light guides (optical fiber) to improve their visibility.
The binary file produced by the compiler must be programmed in the
microcontroller’s flash memory with a programmer such as the PICkit3.
 
After designing the front panel using the free Front Panel Designer
tool [2][3], I printed it on a self-adhesive aluminum foil (3M) with my
laser printer. The result is clean, but cutting the front panel properly
and sticking it at the correct position is a bit tricky.

Details and Finishing Touches

Figure 6: Errare humanum est, and so the PCB required some rework to
It is recommended to keep connections between the device and the
make the barrel jack fit on its incorrect footprint.
inductance-under-test Lx as short as possible because they can intro-
duce errors due to parasitic capacities and inductances. My test leads
are 12-cm 0.2-mm² multi-stranded wires with hooks on one end. As for
any measuring instrument, switch the device on for a few minutes to
stabilize its operating temperature before taking any measurements.
200592-01

Questions or Comments?
Do you have technical questions or comments about his article? RELATED PRODUCTS
Email the author at pleguen@gmail.com or contact Elektor at
editor@elektor.com.
> Elektor 2 MHz LCR Meter Kit (SKU 19883)
www.elektor.com/19883  

> Miniware DT71 Mini Digital Tweezers (SKU 19422)

www.elektor.com/19422  

> T. Hanna, Microcontroller Basics with PIC (Elektor, 2020,

SKU 19188)

WEB LINKS
[1] DIY inductance meter (in French): http://kudelsko.free.fr/inductance_usb/sommaire.htm
[2] This project with downloads at Elektor Labs: https://www.elektormagazine.com/labs/autonomous-inductance-meter
[3] Front Panel Designer: https://www.frontpanelexpress.com/

lektor July & August 2022 21

 review

Acoustic Wave
HH
H oovveerriinngg
Hovering
A Look at the Makerfabs Acoustic Levitation Kit

By Luc Lemmens (Elektor)

With the TinyLev project,

you can experiment with
ultrasonic standing waves
to make objects hover. The
Makerfabs kit contains all
the components needed to
make your own acoustic
levitator. But is it really that
simple to build?

In recent editions of Elektor, we have

presented three small, simple circuits for
magnetic levitation. It was noted that there
are other methods for levitating objects,
such as using acoustic waves [1]. Instructa-
bles.com describes a DIY project designed
at the University of Bristol that uses sound
waves from off-the-shelf ultrasonic transduc-
ers to achieve this apparent weightlessness
of small objects [2]. The Makerfabs Acoustic
Levitator DIY Kit, available from the Elektor
Store, contains all the parts needed to build
this so-called TinyLev project, thus saving
you the trouble of searching for the compo-
nents yourself, including the 3D printed frame  
needed to build the complete contraption. On this web page you can also find (links to) geometry of the base with the reflectors and
  background information and theory on acous- the positioning of the transducers are essen-
The kit comes without documentation, but the tic levitation; it provides good reading if you tial for the levitation effect. On the Instructa-
web link in the Elektor online store guides us want to do more than just building and playing bles site, there are two 3D designs available
to the aforementioned Instructables project with the TinyLev as a gadget. In short: the for downloading and printing the base for
page [2]. It is loaded with instructions in text, levitation relies on the standing wave pattern TinyLev. The Makerfabs Acoustic Levitator
photos and even videos, and the step-by-step of (in this case: two) arrays of ultrasonic trans- kit contains a printed specimen of version
approach will guide you through the construc- ducers, each array arranged in a curved, V1 plus the leg that must be glued the base,
tion of the kit. ball-shaped ‘reflector’. The correct shape and ready for use.

22 July & August 2022 www.elektormagazine.com

  
Start Building, but First Things The next step, Step 8 in the instructions, is to Other Hardware and Software
First wire the transducers. The kit contains a piece In the following steps the other connections
After checking the contents of the kit of thick, stranded wire with the isolation cut need to be made: between the Arduino Nano
(Figure 1), the first thing to do is to deter- open (Figure 2), every single wire from this and the driver board, the power wiring, and
mine the correct polarity of every single ultra- can be used to interconnect legs of the trans- between the driver board and the base with
sonic transducer (Step 4 … 6). Apparently, ducers in concentric circles. The longer red the transducers. The Arduino needs to be
polarity markings on ultrasonic transduc- and black wires from the kit finish the wiring of programmed of course. Download the Nano_
ers in general are not to be trusted, it is not the base, with the transducers arranged in two TinyLev.ino sketch from the Instructables site,
that the MakerFabs kits contains question- arrays of 36; the shorter wires are kept aside compile and upload it in the Arduino IDE.
able components! With 72 sounders in total, for the power supply wiring of the electronics  
this is quite a time-consuming task. In the in Step 15 and 16. Figure 3 shows what the The driver is a standard L298N dual motor
instructions, two methods how to do this are wiring of one of the reflectors will look like. driver board, in this case it provides the 40 kHz
presented, including using the Arduino Nano
board - that is also included in the Maker-
fabs kit - as a minimalistic ‘oscilloscope’.
Testing with a standard digital multimeter
is probably the easiest and fastest way to do
the job: simply mark the polarity based on
the initial potential on the transducer’s pins
when the probes are connected. However,
my auto ranging DMMs appeared not to be
usable here, you need to have a meter that
can be manually switched to the most sensi-
tive voltage range and immediately reacts
when the meter’s probes are connected to
the transducer. Eventually, I used the Arduino
Nano method. Take your time to do this, one
mistake can spoil or even completely ruin the
levitation effect. It is of the utmost impor-
tance that all transducers have the correct
phase of the sound signal.

The Transducer Arrays

Figure 1: Contents of the kit.
With all the transducers marked, it’s time to
glue them all to the 3D printed frame. The
manual on Instructables recommends the
use of hot glue; for me that doesn’t work
with relatively small parts, like these trans-
ducers. If you want me to make a mess of
my workbench, give me a hot glue gun…
I used a tube of hobby adhesive for hard
plastics. Whatever you use, make sure
that all markings you made on these parts
point to the same direction: either all to
the center or all to the outer edge of the
reflector. They should fit flush into the round
recesses/sockets in the base. I ended up
with some spare parts, four transducers were
not needed to completely fill the frame (i.e.
Figure 2: Wires
there are 76 in the kit!). Two of them may be for connecting the
needed for troubleshooting, and can be used transducers.
as sensors (microphones) to check the phase
of the sound from all transducers individually
(if things don’t work…). Figure 3: The wiring of one of both arrays.

lektor July & August 2022 23

sound signals for the two transducer arrays.
The procedure to make these interconnec-
tions will not be difficult for the experienced
tinkerer, just follow the text and photos on
Instructables which - admittedly - could have
been a bit clearer at some points. The base
board used for fixing the PCBs and power
circuit is not included in the kit, but a 10 x
10 cm or larger piece of plywood will do.

Check, Check, Double-Check…

It is advisable to do the tests described in
the instructions before the transducers are
connected, especially to check for short
circuits in the arrays, which would proba-
bly damage the driver board. Also, check
the output signals of the driver board with
an oscilloscope. Then, if you have enough
confidence in your own work, that is if you
are - as good as - sure that all transducers
are correctly connected and working, you may
save time and skip checking (the phase of )
all sounders individually. You can always do
that afterwards; if the levitation does not work
properly, there is a good chance that the fault
lies in the arrays. I had a lot of confidence in
my work and that turned out to be justified:
it worked straight away!
And Experiment…
And then, it’s time to start experimenting with
all kind of ‘things’ that can be levitated with
the TinyLev. I didn’t spend too much on this
though, I tried some rolled-up tiny pieces of
paper, small plastic beads and polystyrene
balls, and these were pretty easy to levitate
between the transducer arrays (see Figure 4).
The Instructables page also describes hover-
ing of small (dead) insects and drops of liquid.
Searching for ‘TinyLev’ on the internet, you’ll
find lots of other experiments with this project
that may be nice to try for yourself (or maybe
they can trigger new ideas).
Hovering on Acoustic Waves
Building the TinyLev is easy to do with this
MakerFab Acoustic Levitator DIY Kit. All
Figure 4: A tiny polystyrene ball hovering in the TinyLev.
essential parts are included and with all
the information that can be found on the
Instructables website and other Internet Good to hear!
sources, it will be relatively easy to start The frequency of ultrasound is above the
experimenting with acoustic levitation. Do human hearing limit, hence the name. In
not expect to finish the construction within this case, we are talking about 40 kHz,
an hour or so. I spent quite some time only about a factor of two higher than the
to check and mark the transducers and to highest tone we can perceive. At best,
solder the arrays. And with the rest, I even we will hear a click or short crackle
skipped some steps that were not absolutely when switching on the TinyLev, but other
necessary to get the levitation working on my creatures, like your pets, can hear it and
workbench. Don’t rush things, though: one may react strongly to it. The author’s cats
mistake — especially in the transducer arrays don’t seem to mind, but they only react to
— may cost a lot of time to troubleshoot and food-related sounds anyway.
fix! If you have done it right, the hard work
will be rewarded with a huge ‘wow’-feeling
when you see the first object hovering on
the TinyLev’s acoustic waves!
220062-01
Questions or Comments?
Do you have any technical questions or RELATED
comments about this article? Contact
PRODUCTS
the author at Luc.Lemmens@elektor.
> Makerfabs Acoustic Levitator DIY Kit
com or contact the Elektor team at www.elektor.com/19984  
editor@elektor.com.  

WEB LINKS
[1] Eric Bogers, “Powerful tractor beam suspends people,” Elektormagazine.com: https://bit.ly/3IzO7uu
[2] Acoustic Levitator on Instructables: www.instructables.com/Acoustic-Levitator/

24 July & August 2022 www.elektormagazine.com

 basic course

Starting Out in Electronics

Rectifiers
By Eric Bogers (Elektor)

In the May/June issue of Elektor, we welcomed the diode

as the first representative of the semiconductor family,
and you can’t think about today’s electronics without it.
Now we’re going to calculate with rectifiers. After all, we
can’t just solder parts together and hope it works.

For our convenience, we subtract 0.7 V from that peak voltage

(which is the forward voltage of the silicon diode used as a
rectifier):

U peak  15 V  2  0.7 V  20.5 V

Figure 1: The half-wave rectifier. As for the perfectionists among you, we have rounded the result
to one decimal — that is accurate enough in this case. Since the
IC needs an input voltage of at least 18 V (we can find this minimum
Let us first have a look at our half-wave rectifier (Figure 1). The input voltage in the datasheet of the IC), the voltage is allowed to
larger the capacitance of the electrolytic capacitor, the less the drop by a maximum of 2.5 V during one period. We rewrite the
voltage will drop during the negative half-periods of the AC voltage; formula for the voltage across the capacitor and add the current,
and also, as the resistance gets smaller (and thus the current drawn voltage and frequency values:
by the load gets larger), the voltage will drop more during the
negative half-periods (‘collapse’, says the electronics engineer). I I
U
 C

We can put this into a formula: f C f  U

Q Q I  T I 1A
U  U   C  8000 μF
C C C f C 50 Hz  2.5 V

We have assumed that the current is constant. Now suppose we
have a 15 V transformer, a half-wave rectifier and a load with a
current draw of 1 A. The voltage provided by the rectifier must be
stabilized by a regulator IC that requires a minimum input voltage
of 18 V to provide a stable output voltage of 15 V. (We’ll come back
to that IC in a future issue.) In that case, how large should the
capacitor be?
 
First, we have to calculate the peak value of the AC voltage; in a
previous article in this series, we saw that for the peak value applies:
 U eff  2
U peak
The next larger standard value would be 10,000 µF, and for a current
of only 1 A, that is quite a ‘bulky’ capacitor. In principle, there are
two methods of achieving a lower value (and thus a smaller and
cheaper electrolytic capacitor):
 
> We apply full-wave rectification, which means that the capaci-
tor is charged twice as often.
> We increase the transformer voltage and thus the permissible
voltage difference.
 
This full-wave rectification will be discussed later; below we will
first deal with ’increasing the transformer voltage’. If we use an
18 V transformer instead of the 15 V version in the example, we
will measure a peak voltage of about 24.7 V at the diode, and that

lektor July & August 2022 25

 is including the losses in the diode and the voltage regulator) does
not really matter, unless it forces us to use a transformer that is a
bit bigger in size.

The Full-wave Rectifier

Figure 2: Half-wave rectifier with stabiliser IC.
means a maximum permissible voltage difference of 6.7 V. The
selection of the capacitor can therefore be correspondingly smaller;
the calculated value is about 3000 µF and the next larger standard
value is either 3300 µF or 4700 µF.
 
As a famous Dutch football philosopher (Johan Cruijff) once said,
every advantage has its disadvantage: this solution involves higher
power consumption and also a higher heat dissipation in the voltage
regulator. For this, we will have a look at Figure 2.
 
The component with the part number 7815 is a fixed voltage regula-
tor. For the time being, we are not interested in how this thing
works. What matters here is that this IC turns a (within certain
limits) varying DC input voltage into an (almost) constant DC
output voltage.
The following applies to the output power in both cases (the
example with the 15 V transformer and the example with the 18 V
transformer):
We have shown the full-wave rectifier schematically in Figure 3.
Thanks to the ingenious circuit with four diodes (also called the
bridge rectifier or Graetz circuit), both half-periods of the sinusoi-
dal AC voltage at the input are now used. During the positive half
period, the charging current for the electrolytic capacitor flows
through the diodes in the upper right and lower left, and during
the negative half period through the diodes in the lower right and
upper left. The advantage is that the electrolytic capacitor is charged
twice as often and therefore (roughly estimated) only needs to be
about half as ‘thick’. This results in the Figure 4 voltage curves:
above the input voltage, in the middle the voltage across the load
without the capacitor, and below the voltage across the capacitor.
 
For the voltage variation across the buffer capacitor of the full-wave
rectifier, the following applies:
I
U 
2 f C

Pout  15 V 1A  15 W

The dissipation in the IC (the power converted into heat in the IC)
is equal to the difference between input and output voltage times
the current (in many cases — and also here — we can neglect the
IC’s own power consumption). Figure 3: The full-wave rectifier.

 U in(min)  U in (max) 
Ploss U in  U out   
I   U out   I
 2  U(t)

 18 V  20.7 V 
Ploss (15 V) 
 15 V  1A 4.35 W
 t
 2 
U(t)

 18 V  24.7 V 
Ploss (18V) 
 15 V  1A 6.35 W

 2  t

To be sure, the input voltage of the IC is not constant; in this case,

U(t)
we use the arithmetic mean of the minimum and maximum input
voltage (which is accurate enough for our purposes).
 
t

As a result, the power loss in the voltage regulator increases around

50% — that’s a lot of excess heat that has to be dissipated. The fact
that the transformer now has to supply 22 W instead of 20 W (that Figure 4: Voltage curves with the full-wave rectifier.

26 July & August 2022 www.elektormagazine.com

 (1 A at a peak voltage of approx. 20 V) — but that is for each half
period. In total, the transformer must be able to deliver at least 40 W.
 
That’s it for this time. In the next episode, we will multiply voltages.
220169-01

 
Editor’s Note: the series of articles “Starting out in electronics” is based on the
book, Basic Electronics Course, by Michael Ebner, published by Elektor. 
Figure 5: This is how we create a symmetrical supply voltage.
 

Let us now calculate our earlier example (with the 15 V transformer)
with a full-wave rectifier. Due to the extra voltage drop across the
second diode, we now have a peak voltage of:
U peak  15 V  2  1.4 V  19.8 V
Questions or Comments?
Do you have any technical questions or comments prompted
by this article? Send an email to the author or to the editor of
Elektor via editor@elektor.com.

This gives us a value for the electrolytic capacitor of:

 
1A
C  5500μF
2  50 Hz 1.8 V

The next larger standard value would be 6800 µF, but it is question- RELATED PRODUCTS
able whether this is (widely) available. If not, we have to use an
electrolytic capacitor of 10,000 µF. At a transformer voltage of 18 V,
we end up with a calculated value of 1700 µF and a standard value
> B. Kainka, Basic Electronics for Beginners,
Elektor 2020. (SKU 19212)
of 2200 µF, which is at least one size smaller. Regarding our initial www.elektor.com/13950  
expectation that the buffer capacitor should be only half the size,
the forward voltage of the rectifier diodes cancels this out. > B. Kainka, Basic Electronics for Beginners (E-Book),
  Elektor 2020. (SKU 19213)
www.elektor.com/18232
It is clear that the current through the load plays a decisive role:
the greater the current, the larger the capacitor. This is the reason
why we often find a large number of big electrolytic capacitors in
the power supply of heavy power amplifiers.

Symmetrical Supply Voltage

For circuits with operational amplifiers (‘op-amps’), we often need
a symmetrical supply voltage — in other words, we need equal
positive and negative voltages with respect to common ground. In
principle, we can achieve this with two transformers, or with a trans-
former with two secondary windings, followed by two rectifiers.
 
It can also be simplified: it is possible to derive a positive and a
negative DC voltage from a single AC voltage, as shown in Figure 5.
In fact, we use two half-wave rectifiers, one utilizing the positive
half-periods of the sinusoidal input AC voltage and the other the
negative half-periods. For the calculation of the electrolytic capac-
itors, the same equations apply as for ‘normal’ half-wave rectifiers.
  WEB LINK
However, the size of the transformer now requires a little more
[1] E. Bogers and M. Ebner, “Starting Out in Electronics:
attention. Suppose we need a symmetrical supply voltage of ±15 V
Welcome to the Diode,” Elektor 5-6/2022:
for a current of 1 A. These voltages are stabilized by voltage regula-
www.elektormagazine.com/220003-01
tors. The transformer must be capable of delivering at least 20 W

lektor July & August 2022 27

 E-FFWD
electronica fast forward
Looking Ahead Again
Together with Messe München and the industry as a whole, Elektor is
looking forward to a well-attended electronica 2022 fair in November.
We’ve been hearing from all sides about the need to meet each other
again in person, and perhaps this applies most to the creative innovators
who are pillars of the annual electronica fast forward awards (e-ffwd)!

28 July & August www.elektormagazine.com

 www.elektormagazine.com/effwd-2022
No matter how much we try not to talk directly about COViD now
that we are finally able to do some more, it is, of course, inevita-
ble to consider its impact on the upcoming edition. We are proud
that we did not have to skip an edition of the fast forward start up
awards. But, after a digital version in 2020 and last year’s Produc-
tronica edition still heavily affected by travel restrictions, it will be
clear that we expect to make a bigger impact this year.
It is therefore with great excitement that we look ahead to the
2022 edition during the largest electronics trade fair in the world
from November 15 onwards in Munich. It promises to be a great
event, which will include both the trade fair itself (almost all major
parties have announced their presence) and the electronica fast
forward, the start-up platform by electronica and Elektor. With a
leading role for innovative start-ups and scale-ups in a booth of
no less than 273 m2, we’ll more than make up for the lack of space
in recent years!
In 2022 there will be much more to discover as we will not only
be giving the stage to innovative concepts in their early stages,
but also to the players who have already leaped the first hurdles
in their entrepreneurial journey.
Best of the Best
This year we intend to really scout out the ‘crème de la crème’
of start-ups and give them a podium. We are focused on inter-
esting and innovative entrepreneurs with a lot of conviction and
boldness. In order to find them, we are going even further in 2022.
In addition to the annual call through the channels of Messe
München and Elektor, we are also actively scouting
internationally through industry collabora-
tions, platforms such as Kickstarter,
Indiegogo, and Crowd Supply,
as well as international
publications.
We are looking for
start-ups and scale-
ups in different
stages of early entre-
preneurship. With a few
up-and-coming talents and
up to three companies that have
already achieved their first business
successes, e-ffwd 2022 promises to be a
combination of inspiration and aspiration. As we
do every year, Elektor invites the most interesting and
inspiring entrepreneurs, people, and products to showcase
themselves and to tell their stories at the e-ffwd booth in front of
the entire industry.
With monthly articles/videos online — in the Elektor magazines,
e-zines and on our socials — potential candidates are introduced.
A ballot for the most interesting stories is combined with a jury
review to arrive at the verdict of who is invited to Munich. Thus,
this year’s e-ffwd is “invite only” but weighted. A total prize pool
of €150,000 in marketing exposure and one of the highly desired
invitations to electronica 2024 is available for the winners. This
way, anyone interested can also see the development with their
own eyes at the next edition of the world’s largest electronics fair.
University Team
New for 2022 is the plan to also invite a student team from a
technical university. Many universities have amazingly ambitious
projects entered in global competitions (often also sponsored by
our industry). There have been many great developments in recent
years, which to some extents have not yet received the attention
they deserve. That's why Messe München and Elektor want to
provide them with that platform in this special comeback edition
that is electronica 2022!
Again, our team is looking for a best-in-class guest from the inter-
national educational sector. Of course, we also invite our own
community to think along and nominate. This can be done via
our special fast forward 2022 website. Of course, teams can also
register themselves. Educational institutions fall outside the e-ffwd
awards competition, but there is of course no better place to show
the results of hard work to the entire international industry of which
every student dreams of becoming a member!
Ambition
Clearly, the electronica fast forward awards 2022 will be bursting
with ambition. Especially, of course, due to the ambition of the
engaged start-up entrepreneurs and students involved. But also
the e-ffwd organization itself has a renewed ambition to make this
the best edition so far. Messe München has made an impressive
stand available in the middle of Hall B4, right next to the Forum
where every student team, start-up or scale-up gets the chance to
present itself. And Elektor will use all its media to give this year's
fast forward awards guests a marketing boost.
The organization cordially invites industry partners to be part of
this special edition of the fast forward awards. Show your support
for the future of our industry and become a sponsor for this unique
event. Please contact the team at fastforward@elektor.com or ask
your direct contacts about the possibilities.
lektor July & August 2022 29
 basics
Get Started
With Your Oscilloscope
Find Your Way Through the Knobs and Buttons
By Clemens Valens (Elektor)
When you have been
tinkering with electronics
for a while and your projects
are getting more serious,
chances are that you will
need an oscilloscope.
An oscilloscope is an
instrument that displays the
evolution of one or more
voltages over time. But how
do you choose one? And how
do you use one?
  project
Watch this eo!
on vid
30 July & August www.elektormagazine.com
Of course, you already invested in a multi-
meter, which is great, but an oscilloscope
can do things you cannot do with a multime-
ter. However, oscilloscopes are also much
more complex instruments than multime-
ters, and so it takes some time and practice
to get the most out of them.
What Do You Need?
Oscilloscopes used to be expensive, but
today prices have dropped so much that
almost anyone can afford one. For most
hobbyist, tinker, and maker applications,
a basic oscilloscope is more than good
enough. If your thing is Arduino- or Raspber-
ry-Pi-based projects or building and repair-
ing audio amplifiers or musical instruments
or guitar effects, there really is no need to
invest in a four-channel (or more) oscillo-
scope with hundreds of MHz of bandwidth
and many gigasamples. A two-channel,
20-to-50-MHz oscilloscope is just fine
(Figure 1).
Not Too Small, Please
However, even though it may seem tempt-
ing, I don’t recommend spending money on
those cheap small DIY oscilloscope kits you
can find online (Figure 2). They just do not
offer the comfort and features of a low-cost
but real bench oscilloscope. They may be
practical when traveling or when you are
really tight on space, but that’s probably all
they are good for. I like my oscilloscope to
stay where it is when I pull a test lead, and
it must have knobs and buttons that I can
find quickly. To me, an oscilloscope without
knobs for the horizontal and vertical axes
is not a real oscilloscope.
Headless Oscilloscopes
Of course, there are headless oscilloscopes
with excellent specifications (Figure 3),
but they require a tablet or computer and
a mouse. Such devices are good for doing
remote measurements or for documentation
or constant monitoring, for data logging and
postprocessing (or for doing very precise
things that require a lot of fiddling). In
my lab setup, the computer is needed to
show the schematic, modify firmware, read
datasheets, and search the Internet. I don’t
want it to be an oscilloscope as well. And
I don’t want a second computer because
it takes up too much bench space. And, of
course, because I want real knobs.
 DSO, MSO, Analog?
Today’s oscilloscopes are digital, and they are
called DSOs, which stands for Digital Storage
Oscilloscope. In the past oscilloscopes were
analog. Analog oscilloscopes are bulky and
limited, and you don’t want one unless you
have very special needs. There have been
hybrid oscilloscopes that could do both, but
you don’t want one of those either. By the
way, do not confuse these with so-called
mixed-signal oscilloscopes, or MSOs, as these
are DSOs with special features for digital
signals and communication buses and things.
Let’s Try to Use It
So, now that we know a bit more about oscil-
loscopes, we can move on to learning how
to use them. If you were smart and bought a
cheap one, it will not have too many functions,
and so it is easy to get going.
 
As said at the beginning, an oscilloscope
displays the evolution of a voltage over time,
a ‘signal’, and it does this in the shape of a
two-dimensional graph with the center in
the middle of the screen. Note that I wrote
‘voltage’, and not ‘current’. An oscilloscope is
a fancy voltmeter.
Main Control Sections
Because it displays signals as graphs, an
oscilloscope has controls to adjust the
horizontal and vertical axes of the graph.
On every modern oscilloscope that I know
of, these controls are grouped together in
a section labeled Horizontal and a section
labeled Vertical (Figure 4). The horizontal
section is also known as the Time Base as
the horizontal axis usually represents time.
Then there is a third section called Trigger.
This is probably the most important section
as it determines how and when a signal is
shown. Often it is easy enough to get the
signal you are measuring to fit on the screen
by adjusting the horizontal and vertical axes,
but making the oscilloscope display the part
of the signal you are interested in can be
much more difficult. The trigger section
gives you control over this and it is there-
fore important to understand what it does
and how.
Vertical
But first let us look at the section Vertical.
This section lets you amplify or attenuate the
input voltage or signal, meaning that you can
adjust its amplitude, and you can move it up
and down. This can be set independently for
every input channel.
 
By the way, in oscilloscope language, a signal
is also called a trace. Single-trace means one
signal or channel, dual-trace means two, etc.
XY Mode
Another point of confusion can be that input
connectors besides being labeled channel
1 and 2 are sometimes also labeled X and
Y (Figure 5). This refers to a special opera-
tion mode of the oscilloscope in which the X
input controls the horizontal axis instead of
the vertical. This is the mode to create the
famous Lissajous figures with that they liked
to show in old sci-fi movies. We will not use

Figure 5. The inputs X and Y are used in

Figure 1: In most situations a simple two-channel,
Figure 4: The three main control sections of every XY-mode and let you create, for example,
20-to-50 MHz oscilloscope is all that is needed.
oscilloscope: Vertical, Horizontal and Trigger. Lissajous figures.
 

Figure 2: Even
though they
seem kind of
cool to have,
miniature
oscilloscopes
are not very
practical to use.

Figure 3: A headless oscilloscope saves bench space. Or does it? A

computer or tablet is required for viewing the signals.

lektor July & August 2022 31

Figure 6: Connect the probe’s crocodile clip to GND
(ground). The rule is to try to get it as close as possible to
the signal of interest, but often ‘anywhere’ is just fine.
XY mode in this article; for us the horizontal
axis always represents time.
Connecting the Probe
Connect a probe to one of the channel
connectors, not to the Ext, Trigger, Aux or
Z connector. Note that probes often have a
switch to choose between 1:1 or 10:1. This is
an extra attenuation option that allows large
signals to fit or to improve the precision of
measurements of sensitive signals. Some
people always use 10:1 attenuation mode
and there exist probes that are always in
10:1 mode. (Other ratios like 1,000:1 can be
had too.) Probes are also referred to as 1x
and 10x where the ‘x’ means attenuation, not
amplification. It is often possible to inform
the oscilloscope about the type of probe
you are using, so it can adapt the scales
accordingly.
 
Before connecting the tip of the probe
to the signal of interest, first connect the
crocodile clip attached to the probe to the
ground reference of the circuit under test. As
a general rule, connect it as close as possible
to the signal. However, in many cases this is
not so important as long as it is connected
to ground somewhere (Figure 6). It can
even be convenient to use the crocodile
clip of a second probe just for connecting
to ground so you can remove the crocodile
clip from the measuring probe and keep it
out of your way.
32 July & August www.elektormagazine.com
Figure 7: Always keep in mind that an oscilloscope may be
connected to earth. If it is, short circuits may arise when the
probe’s crocodile clip is connected to something else than ground.
A Word About Ground
Note that ground doesn’t have to be ground, it
can be any voltage or signal in the circuit, but
it must be the same for every channel. Also
note that probe ground is usually connected to
oscilloscope ground which can be connected
to mains ground, so if you connect the probe
ground to something other than ground, short
circuits may be created and dangerous situa-
tions may arise (Figure 7). Therefore, always
connect the crocodile clip to ground unless
you know what you are doing.
Horizontal or Timebase
With the knob in the section Horizontal, you
can zoom in and out on the time scale and
you can move the signal to the left or to the
right. Usually, the center of the screen is zero.
Triggering
Up to now, I was supposing that you did see
something on the display of the oscilloscope,
but maybe you didn’t? One reason can be
that the trace is not on the screen. Use the
vertical level control to find it. Another reason
may be that the brightness is too low. If it is,
crank it up. A third reason can be a trigger
problem. If your oscilloscope has an Auto
Setup button or something similar (Figure 8),
now is a good time to press it. After doing
so you may not see what you expected, but
you should at least see something. Use the
vertical and horizontal knobs to bring the
signal into range.
Triggering is what makes an oscilloscope
really useful as it allows you to focus on the
interesting part of a signal. A trigger is needed
to start a trace. If there is no trigger, the trace
will not start, and you will not see anything.
This is why an oscilloscope features several
options for triggering. One might even argue
that the more trigger options an oscilloscope
has, the better it is.
 
In my world, every oscilloscope features at
least automatic and normal triggering, adjust-
able trigger level, up and down or positive
and negative trigger slopes and an external
trigger input.
 
The first thing to do is select the source for the
trigger signal. Usually it is one of the signals
you want to look at, but it can be a signal
generated by another device altogether.
Automatic Triggering
Automatic triggering is easiest to use
(Figure 9). In this mode the oscilloscope
decides when to trigger, and the user can
set the trigger level and choose the slope
to trigger on. This mode is useful for quickly
seeing if something is “going on” on an input
or to simply measure a DC voltage.
Normal Mode
In Normal mode the oscilloscope only triggers
when the trigger level and slope conditions
are met or any other trigger condition that
 Figure 8: The Auto Setup button
can help you find your signals,
but the result may always not be
what you hoped for.
you may have specified. After a trigger, the
trace will run until it falls off at the right side
of the screen. A new trigger is required to
start it again. If you set the trigger level too
low or too high, the trace freezes or may not
appear as it is never triggered. This mode
is good for infrequent events or to focus on
special events.
Run/Stop & Single
It is quite probable that your oscilloscope
has a Run/Stop button and a Single button
(Figure 10). The Run/Stop button lets you
freeze the display, which is handy if you want
to study a signal in detail without allowing a
new trigger to change it or make it disappear.
Press the button again to leave this mode.
The Single button can be used when an event
happens only occasionally, for instance only at
power-on or after pressing a button, or when
it is not periodic. After the trigger happens, the
trace will run only once and then the oscil-
loscope enters Stop mode. You must press
Run or Single to restart the capture of a single
event. Press Single or Run again to go back to
normal mode (which button to press depends,
of course, on the oscilloscope).
 
Depending on the oscilloscope, there can be
more trigger options. Typical other options go
from triggering on both slopes, or on a time
delay between slopes, a pulse so to speak, to
triggering on a sequence, inside or outside a
Figure 9: Do not confuse Figure 10: The Run, Stop and Single buttons
the Auto trigger button allow capturing infrequent events or freezing
with the Auto Setup the traces for taking measurements without
button. Auto trigger mode the signal disappearing.
is useful to quickly setup
a measurement before
switching to Normal mode.
window, with delays and what not. As I said
before, the more trigger options, the better.
  Questions or Comments?
We will stop here. Your oscilloscope probably Do you have technical questions or
has several other buttons not covered in this comments about this article? Please note
article. But now that you know how to get a that we are aware that it is incomplete and
stable signal on its display, you can explore that some shortcuts have been taken. This
these functions more in depth. article is not targeted at experts. Email the
  author at clemens.valens@elektor.com or
Whatever you do, always keep in mind that contact Elektor at editor@elektor.com.
before measuring anything, you should have
an idea of what to expect so that you can
compare the result to what it is supposed to
be. If you don’t know what to expect, then you
can’t tell if it is wrong or if it’s right.
200661-01
RELATED PRODUCTS
> Owon SDS1022 2-ch Digital Oscilloscope (20 MHz) (SKU 18898)
www.elektor.com/18898  
> LabNation SmartScope USB Oscilloscope (SKU 17169)
www.elektor.com/17169  
> PeakTech 1375 All-in-one 4-ch Oscilloscope (100 MHz, 1 GS/s)
(SKU 19316)
www.elektor.com/19316   J
> JYE Tech DSO Coral 112A Oscilloscope (SKU 18319)
www.elektor.com/18319
lektor July & August 2022 33
 background

Raspberry Pi Pico
Makes an
MSF-SDR
Decode a Time Signal with a Pi Pico SDR

VLF antenna in Anthorn (Dougsim, https://bit.ly/34HXeuG)

By Martin Ossmann (Germany)

MSF is the UK’s equivalent of the

German DCF77 time signal transmitter.
This SDR project shows how a receiver
and decoder for these (and other) time
signals can be implemented quite
simply and, above all, inexpensively.
For the hardware you really won’t need
much more than a low-cost
Raspberry Pi Pico to receive, decode,
and display MSF time signal
information.
 
In Germany the DCF77 transmitter in Mainflingen transmits an encoded
long-wave time signal. Its equivalent based in the UK is the MSF signal
formerly known as “The Rugby Clock.”[1] It sends out time signals using
a 60 kHz long-wave carrier signal. In the early days, it served as a
frequency standard, and sent out a five-minute pulse train twice a day.
The “transmission protocol” of the signal has changed several times
over the decades, but it was not until 1977 that the encoding included
time-of-day and date information that could be evaluated by the receiver.
The Project
Over the years, there have been many receiver/decoder circuits
described in various Elektor articles that make use of the DCF77
signal, but this is probably the first time a design for an MSF receiver
has been featured. There was, however, an add-on circuit [2] for the
good-old 6502 Junior Computer [3] described in the English language
edition of Elektor.
 
A lot of water has flowed under the bridge since then, and technology
relating to receivers/decoders has progressed in leaps and bounds.
In this article, we will use the latest “up-to-date” concepts to build a
software-defined radio (SDR) using a small microcontroller board.
The Raspberry Pi Pico board, which uses an RP2040 CPU clocked
at 125 MHz (the in-house controller of the Raspberry Pi Foundation
is equipped with dual 32-bit ARM Cortex M0+ cores), is a suitable,
you could say, predestined piece of hardware for this application. Its
analog-to-digital converter can run at 500 ksps. All this processing
power can be purchased for an almost ridiculous €5 (see Related
Products below).
 
Here we show how to implement a complete receiver in hardware
and software for the 60 kHz MSF time signal. The entire receiver,
without a display but with an RS232 output and antenna connection,
is shown in Figure 1.

34 July & August www.elektormagazine.com

 1 USB 40
GP0 VBUS
2 39
GP2 VSYS
3 38
GND LED GND R2
4 37
GP2 3V3_EN

10k
5 36

BOOTSEL
GP3 3V3_OUT active
6
GP4 ADC_VREF
35 C1 antenna
7 34

Hardware
GP5 GP28
8 33
GND GND R1 10µ
9 32
First we will take a look at the hardware necessary to build the SDR. GP6 MOD1 GP27 10V

10k
10 31
GP7 GP26
There are just a few additional items to connect to our pico board. 11
GP8 RUN
30

 
12 29
GP9 GP22
13 28
Antenna Input: We use the analog input pin ADC2 (GPIO28, on the Pico 14
GND
GP10
GND
GP21
27

board pin 34) to receive signals from the antenna. The ADC uses the 15 26

Raspberry
GP11 GP20
16 25
internal 3.3 V as a reference voltage. This pin must therefore be biased

Pi PICO
GP12 GP19
17 24
GP13 GP18
at half the reference voltage. The two resistors in Figure 2 take care of 18
GND GND
23

this. The 10 µF capacitor C1 provides AC coupling for the incoming signal. 19
GP14 GP17
22

SWCLK

SWDIO
20 21
 

GND
GP15 GP16

RS232 Output: In its simplest form (without an LC display), the receiver 220006-004

uses a serial interface (115,200 bit/s) to output the data. The interface is Figure 2: Components required at the A/D signal input.
implemented by the circuit shown in Figure 3. We cannot use the USB
port to output the serial data because it would generate interrupts to
our software in an unpredictable way.
  +5V 1
GP0
USB
VBUS
40

PWM DACs: When no LC display is connected, it is possible to use

2 39
R1 R2 GP2 VSYS
3 38
the DACs with pulse width modulated (PWM) signals to create an easy

2k2

2k2
GND LED GND
4 37
GP2 3V3_EN
aid for debugging. We have set up two PWM DACs with the associ- 5 36

BOOTSEL
GP3 3V3_OUT
6 35
ated low-pass filters as shown in Figure 4. Using GPIO 2 and GPIO 3 D T1
7
GP4 ADC_VREF
34
RS232 GP5 GP28
as PWM outputs, for example, the demodulated signal and Bit-Timer 115200 Bd
G 8
GND GND
33

signals can be displayed on a scope (Figure 5).

9 32
S GP6 MOD1 GP27
BS170 10 31
  11
GP7
GP8
GP26
RUN
30

LCD: The 3.5-inch Arduino 8-bit module ILI9486 (non touch screen 12

GP9 GP22
29
13 28
version SKU MAR3502 [4]) can be used as the LCD. This 3.5-inch Arduino 14
GND GND
27
GP10 GP21
shield has 480x320 coloured pixels and retails for around €10. Its connec- 15 26

Raspberry
GP11 GP20
tion to the Raspberry Pi Pico is shown in Figure 6. 16 25

Pi PICO
GP12 GP19
17 24
  18
GP13 GP18
23
GND GND
The received signal is shown on the LCD together with a waveform 19
GP14 GP17
22
SWCLK

SWDIO

showing bit timing information. The received time information is shown

20 21
GND

GP15 GP16

in plain text above the waveform (Figure 7). If you do not need to display 220006-005

this information you can choose to simply omit the LCD without the Figure 3: The RS232 output from the Pico Board.
need to make any changes to the software.
 
Active Antenna: We have already covered the antenna connection;
the circuit of the active antenna can be seen in Figure 8. It is essen- C1
tially based on the LM6132 dual operational amplifier. This op-amp is
1 USB 40
GP0 VBUS
2 39
particularly suited for this application with an operating voltage of 2.7 100n GP2 VSYS
R1 3 38
GND LED GND
to 24 V, 10 MHz gain bandwidth product, rail-to-rail input and output A 1k 4
GP2 3V3_EN
37
PWM 5 36
signal capability and low current consumption of 360 µA per amplifier.
BOOTSEL

R2 GP3 3V3_OUT
B 1k 6 35
GP4 ADC_VREF
7 34
C2 GP5 GP28
8 33
GND GND
9 32
100n GP6 MOD1 GP27
10 31
GP7 GP26
11 30
GP8 RUN
12 29
cut-off GP9 GP22
13 28
frequency GND GND
approx. 14 27
GP10 GP21
1600Hz 15 26
Raspberry

GP11 GP20
16 25
Pi PICO

GP12 GP19
17 24
GP13 GP18
18 23
GND GND
19 22
GP14 GP17
SWCLK

SWDIO

20 21
GND

GP15 GP16

220006-006

Figure 1: The Raspberry Pi Pico board as a software-defined radio for MSF Figure 4: Two low pass filters for the PWM-DAC debug signals.
reception.

lektor July & August 2022 35

 3.5" ILI9486 LC Display

GND
RST

+5V
WR
CS
RS

RD
D7
D6
D5
D4
D3
D2
D1
D0
1 USB 40
GP0 VBUS
2 39
GP2 VSYS
3 38
GND LED GND
4 37 +5V
Figure 5: The PWM test signals. Top trace is the 5
GP2 3V3_EN
36

BOOTSEL
ampl-value, middle trace is the SecondTimer
GP3 3V3_OUT
6 35
GP4 ADC_VREF
showing Sampling trigger pulses and below is the 7
GP5 GP28
34

digital sigValue signal. 8

GND GND
33
9 32
GP6 MOD1 GP27
10 31
GP7 GP26
11 30
GP8 RUN
12 29
GP9 GP22
13 28
GND GND
14 27
GP10 GP21
15 26

Raspberry
GP11 GP20
16 25

Pi PICO
GP12 GP19
17 24
GP13 GP18
18 23
GND GND
19 22
GP14 GP17

SWCLK

SWDIO
20 21

GND
GP15 GP16

220006-008

Figure 7: Receiver information on an LCD. Figure 6: Connections for the 3,5“ LCD.

V1 +5V

R1 R4
8 C7 C5
L2
10k

IC1
1k

10mH
4 100n 10µ
C1
IC1 = LM6132
500p R6

L1
470

3
1 5 ca. 4V5 receiver
IC1A C6
1mA
C9
2 7
C2 IC1B
6
1µ 10µ
R3 R5
10k 10k
R2
C3 C4 C8
10k

10µ 10µ 10µ

220006-010 Figure 8: Active antenna for

L1: 500 turns enamelled copper wire 0.2 mm on 10 mm ferrite rod (length 180 mm) the 60 kHz MSF signal.
C2: as required

Isum Buffer Low pass

filter
Input
active COS
antenna ADC DDS fMix Ndown Fifo Ndown
500k 400k log√I2+Q2 + Decoding
Samples/sec
1250 Samples/sec
4
100 Bit/sec
–
SIN
0.00475

Qsum Buffer Low pass

filter Z-1 Threshold

0.995

Figure 9: Diagram showing the MSF receiver signal paths in the Raspberry Pi Pico Board. 220006-011

36 July & August www.elektormagazine.com

No doubt other op-amps would also work here but if you intend to
replace the LM6132, check carefully whether it will match the spec.

Programming the Input Mixer Figure 10: The MSF seconds pulse showing Bit A and Bit B coding.
After the hardware, we come to the programming. The analogue paths
of the SDR is structured as shown in Figure 9. The Raspberry Pi Pico
can be programmed using several different languages. For this appli-
cation we chose C using the Microsoft Visual Studio Code develop-
ment environment running on a PC under Windows 10. Let’s look at
how the different parts function.
  Year (BCD coded 0...99) Meaning
The ADC Sample Routine is triggered by the PWM and called 500,000 80 40 20 10 8 4 2 1 BCD
times per second via interrupt. The Offset ADCoffset  = 2048 is weighting
subtracted from the ADC value and the result is then multiplied by 17A 18A 19A 20A 21A 22A 23A 24A Bit
ADCscale = 10 (Listing 1).
  Month (BCD coded 1...12) Meaning
The local oscillator (LO-DDS) phase is updated and the input value is
10 8 4 2 1 BCD
multiplied by the cosine (in-phase or I signal) and the sine (quadra- weighting
ture-phase or Q signal). The products are summed over 1250 samples
25A 26A 27A 28A 19A Bit
(in Isum and Qsum). The values are ​​ then (in Listing 2) passed to a FIFO
for further processing, which then takes place at 500000/s/1250 =
Day of month (BCD coded 1...31) Meaning
400 samples/s. This sample rate is so low that all the further process-
20 10 8 4 2 1 BCD
ing can be carried out using double variable values.
weighting
 
30A 31A 32A 33A 34A 35A Bit
The values ​​are read from the FIFO and passed through a fourth-order
Butterworth low-pass filter with a cut-off frequency of 3 Hz. During
development it was found that this low cut-off frequency was necessary Day of week (BCD coded 0...6) Meaning
because the author’s antenna received strong interference signals directly 4 2 1 BCD
adjacent to the wanted signal. This is followed by another down-sam- weighting
pling, this time by a factor of 4, so that 100 samples/s are then processed. 36A 37A 38A Bit
 
The msfSample() routine in Listing 3 then calculates the carrier ampli- Hour (BCD coded 0...23) Meaning
tude ampl from the I/Q components. The logarithm of ampl is derived 20 10 8 4 2 1 BCD
and in turn stored in ampl which makes it easier to decode the bits. weighting
  39A 40A 41A 42A 43A 44A Bit
The switching level threshold is derived from ampl via a first-order
recursive filter calculation. The signal ampl is then compared with Minute (BCD coded 0...59) Meaning
the switching level threshold to determine its digital receive value 40 20 10 8 4 2 1 BCD
sigValue. Now with the analogue signal processing covered we can weighting
look at how the data is recovered from the received signal and how 45A 46A 47A 48A 49A 50A 51A Bit
this corresponds to the time-of-day information.
Minute marker Meaning
Reading the Bits 52A 53A 54A 55A 56A 57A 58A 59A Bit
The MSF transmitter sends out RF carrier pulses at each second as shown
0 1 1 1 1 1 1 0 Value
in Figure 10. At second 0 of every minute the carrier switches off for 500 ms.
The SDR uses this pulse for synchronization. The pulses emitted at each Parity bits
of the following 59 seconds contain two bits of information: A and B. At
Bit 54B with bit 17A to 24A results in odd number of bits
the start of each of these seconds the carrier is off for 100 ms (correspond-
Bit 55B with bit 25A to 35A results in odd number of bits
ing to 10 samples in our application). If Bit A = 1, the carrier remains off
for a further 100 ms, and if Bit B = 1, the carrier is off for another 100 ms. Bit 56B with bit 36A to 38A results in odd number of bits
  Bit 57B with bit 39A to 51A results in odd number of bits
In SecondTimer, a timer, synchronised to each second, runs from 0 to 99.
Figure 11: MSF time coding scheme. Source: [5].
The software decoding works as follows: In Duration, the pulse length of
the current pulse is measured. If a 0.5 s absence of the carrier is detected,

lektor July & August 2022 37

 Listing 1: A/D sampling and local oscillator Operation. Summing and downsampling with a factor of 1250 at 400 Samples/s.
 
int16_t adcv=(uint16_t) adc_hw→result;             //get ADC result
hw_set_bits(&adc_hw->cs, ADC_CS_START_ONCE_BITS);  //start ADC again
pwm_clear_irq(pwm_gpio_to_slice_num(PWM_PIN1));    //interrrupt flag
DDSp += DDSd ;                                      //increment LO-DDS phase
inputVal=ADCscale*(adcv-ADCoffset) ;               //offset and scaling
Isum += LOcosTab[DDSp>>24]*inputVal ;              //I-multiplication
Qsum += LOsinTab[DDSp>>24]*inputVal ;              //Q-multiplication
SampleTime++ ;                                     //refresh this step
if(sampleTime>=1250){                              //downsampling
      FIFO...}                                     //further steps

Listing 2: Filtering the I- and Q- values and downsampling by a factor of 4 at 100 samples/s.
 
Isample=IntFifoI[IntFifoOutPtr] ;                  //get I-signal from FIFO
Qsample=IntFifoQ[IntFifoOutPtr] ;                  //get Q-signal from FIFO
IntFifoOutPtr=(IntFifoOutPtr+1) & IntFifoMask ;    //increment FIFO pointer
IfilOut = tprun(IIfil,Isample) ;                   //lowpass filter I-signal
QfilOut = tprun(QQfil,Qsample) ;                   //lowpass filter Q-signal
kdown++ ;
      if(kdown>=4){                                //downsampling factor 4
            msfSample(IfilOut,QfilOut) ;
            kdown=0 ;
            }

Listing 3: Calculating the amplitude ‘ampl’, the switching threshold, and the pulse duration to recover bits.
 
void msfSample(double ii, double qq){
      ampl=sqrt(ii*ii+qq*qq) ;                     //get carrier amplitude
      ampl=40*log(ampl+1) ;                        //log is better!
      threshold=0.995*threshold+0.005*ampl*0.95 ;  //recursive mean as threshold
      if(ampl>threshold){                          //comparator function
            sigValue=1 ;                           //digital value = 1
            }
      else {
            sigValue=0 ;                           //digital value = 0
            }
      doScope(ampl/2.0+20,threshold/2+20,sigValue*10+10, DAC/2.0+30) ;
      if(sigValue==lastSigValue){
            Duration++ ;                           //pulse goes on
            }
      else {                                       //pulse end reached
            tt=pulseForm(lastSigValue,Duration) ;  //get pulse character
            if(tt==’z’){
                  printf("sync on z") ;            //signalize sync
                  SecondTimer=50-2 ;               //sync SecondTimer
                  }
            printf("%c",tt) ;                      //display pulse character
            Duration=0 ;                           //new pulse length starts
            lastSigValue=sigValue ;                //update lastSigValue
            }

Listing 4: Sampling of bits A and B triggered by SecondTimer.

 
IncSecondTimer() ;                         //SecondTimer runs from 0 to 99
DAC=SecondTimer ;                          //scope sawtooth signal
if (SecondTimer==5+0) { DAC=60 ; }         //scope signal pulse
if (SecondTimer==15+0) {                   //bit A sample time
   DAC=60 ;                                //scope signal pulse
   putMSFbit(Second,0) ;                   //clear bit store
   if (sigValue==0) {                      //if carrier switched off
      addMSFbit(Second,1) ;                //Bit A set true
      } ;
   }
   if (SecondTimer==25+0) {                //bit B sample time
      DAC=60 ;                             //scope signal pulse
      if (sigValue==0) {                   //if carrier switched off
         addMSFbit(Second,2) ;             //Bit B set true
         } ;
      }

38 July & August www.elektormagazine.com

 I
Low pass
filter
COS
Input ADC 60 kHz
Down-
250k
Samples/sec
sampling
SIN
Low pass
Q filter
Figure 12: Upconversion to a 100 Hz IF.
SecondTimer is set to the value 50-2=48 so that the SecondTimer timer
now runs synchronously with the second (Listing 4). At the same time,
the minute is synchronized by setting the value of the current second to
0 in doMinuteSync().
 
With the help of SecondTimer, the received signal is sampled at the mid-bit
position of Bit A and Bit B (SecondTimer==15 and SecondTimer==25) in
order to determine the values of​​ these transmitted bits. We simply output
the received digital value via GPIO-Pin 4 (Pico-Pin 6):
gpio_put(GPIO4, sigValue); // Output sigValue at
// pico GPIO4=pin 6
 
The value of SecondTimer is also output later for debugging purposes
via PWM, as are the values ​​of ampl and DAC. This is achieved with the
following two statements:
 
pwm_set_gpio_level(PWM_PIN1, ampl/5.0 ); // Output
// amplitude
pwm_set_gpio_level(PWM_PIN2, DAC );       // Output timing
Decoding Time Information
Whenever SecondTimer is synchronized by the 0.5 s gap, one minute has
passed and we can evaluate the latest time information. The received data
bits are in the values ​​MSFbits[0 to 59]. The transmitter encodes the
information listed in Figure 11 into these bits. The time and date information
is then simply reconstructed as in Listing 5 to give hours and minutes.
We also display the same information that we send out via the serial
interface as text on the LCD. This is done using the instructions given
in Listing 6. A parity check on the received information is evaluated as
in Listing 7. The monitored bits are the A-bits of the transmitted infor-
mation. The four check bits are B-bits of each corresponding seconds
pulse. Four parity checks are carried out, the integrity of up to 12 bits are
protected by one parity bit.
Debug Signals
Classic superhet receiver designs mix the incoming RF signal with a
variable frequency local oscillator signal to produce a lower intermediate
Low pass
filter
COS
100 Hz Ouput
+
–
SIN
Low pass
filter
220006-014
frequency or IF. The IF signal is then filtered with a relatively narrow
band filter. The IF signal of the MSF60 receiver can also be viewed
using an oscilloscope. Our receiver mixes the input signal down to IF =
0 Hz. If you want to observe an AC IF signal, you can upconvert the 0
IF signal to an AC IF. The circuit block diagram is shown in Figure 12.
 
The software to perform the necessary upmixing is shown in Listing 8.
A PWM output is used as the DAC. The 100 Hz AM-modulated IF
signal is shown in Figure 13.
This completes the design and construction of the MSF receiver. Only
one core of the processor is used in this application, leaving plenty
of computing power for expansion. The bit decoding, for example,
could be made more error-tolerant. A DCF77 receiver could be built
in much the same way, only the bit decoding process would need to
be adapted. The MSF signal reception here in Aachen (Germany) is
much weaker than the DCF77 signal with an equivalent SDR. This often
results in parity bits indicating errors in the received information but
sufficient error-free messages still get through frequently enough to
allow accurate time of day information to be displayed reliably.
Working with an RP2040
We have shown that with very little additional hardware, the
Raspberry Pi Pico board can be turned into a complete MSF SDR.
Figure 13: 100 Hz amplitude modulated IF Signal.
lektor July & August 2022 39
 Listing 5: Decoding and serial BCD output of hours and minutes.
 
void OutBCD2(int v){                           //issue 2 BCD digits
   uartPutc(’0’+(v>>4)) ;                      //via serial interface
   uartPutc(’0’+(v & 0xf)) ;
   }
int GetBCDbits(int StartPos , int Length){     //fetch length bits from MSFbits
   int v,k ;                                   //start at StartPos
   V=0 ;                                       //BCD coding
   for (k=0 ; k<Length ; k++ ){
      v=(v<<1) + ( MSFbits[StartPos++] & 1) ;
      }
   return v ;
   }
hours =GetBCDbits(39,6) ; OutBCD2(hours ) ;     //hours = bits 39 to 44
uartPutc(’:’) ;
minutes=GetBCDbits(45,7) ; OutBCD2(minutes) ;   //minutes = bits 45 to 51
UartBlank()

Listing 6: Output Hours and minutes information to the LCD.

 
LcdPutc(CRcode) ;                  //output carriage return
LcdPutc(LFcode) ;
LcdPutc(’0’+((hours>>4)&0xF)) ;    //MS digit of hours
LcdPutc(’0’+(( hours)&0xF)) ;      //LS digit of hours
LcdPutc(’:’) ;                     //separator
LcdPutc(’0’+((minutes>>4)&0xF)) ;  //MS digit of minutes
LcdPutc(’0’+(( minutes)&0xF)) ;    //LS digit of minutes
LcdPutc(’:’) ;                     //separator

Listing 7: Parity checking.

 
LcdPutc(CRcode) ;                          //output carriage return
int parity(int from , int to) {            //parity over A bits
int parity ;
int k ;
parity=0 ;
   for (k=from ; k<=to ; k++){
      parity ^= MSFbits[k] ;               //XOR with bits
      }
Parity &= 1 ;                              //select A bit
return parity ;
}
void parityCheck(int from , int to , int checkPosition) {
   int p ;
   p=parity(from,to) ;
   if ( (MSFbits[checkPosition] & 2)>0) {  //B bit is parity
      P ^= 1 ;                             //XOR parity bit
      }
   uartPutc(’ ’) ;
   uartPutc(’P’) ;
   uartPutc(’=’) ;
   uartPutc(’0’+p) ;                       //output parity bit
   }
   ParityCheck(17,24,54) ;                 //four parity checks
   parityCheck(25,35,55) ;
   parityCheck(36,38,56) ;
   ParityCheck(39,51,57) ;

Listing 8: Coding for software upmixing.

 
debugDDSp += debugDDSd ;              //100Hz phase update
v=IfilOut*cosTab[debugDDSp>>24]       //add I signal
   +QfilOut*sinTab[debugDDSp>>24] ;   //add Q signal
v=62+v/1024 ;                         //offset and scaling
pwm_set_gpio_level(PWM_PIN2,v );      //PWM output

40 July & August www.elektormagazine.com

 RELATED PRODUCTS

> Elektor Raspberry Pi RTL-SDR Kit (Book and

Components) (SKU 19518)
www.elektor.com/19518  
The most complex element of this build is construction of the active
antenna. Given the processing power and low cost of the board this > Elektor SDR Hands-on Kit (Book and SDR-Shield with
application shows what even a hobbyist with few resources can achieve ferrite toroid and cable) (SKU 19041)
nowadays. You often read how easy the Pico board can be programmed www.elektor.com/19041  
in Python, but in this application, it will not be able to cope with the
> Raspberry Pi Pico RP2040 (SKU 19562)
500k sample rate of the input signal. With C, however, the microcon- www.elektor.com/19562
troller hardware can be addressed more directly and programmed
efficiently. We see here that even without the benefit of a floating point
unit (FPU) the RP2040 is still more than capable of implementing a
low-pass digital filter. 
220006-01

r
this Elekto
Questions or Comments? Check out bout the
If you have any technical questions regarding this article you can video a Pi Pico! 
contact the author at ossmann@fh-aachen.de or the Elektor team Raspberry
at editor@elektor.com.

WEB LINKS
[1] “Time from NPL (MSF),” Wikipedia: https://en.wikipedia.org/wiki/Time_from_NPL_(MSF)
[2] “Time receiver for the Rugby MSF,” Elektor 9/1982: www.elektormagazine.com/magazine/elektor-198209/44950
[3] J. Buiting, “Retronics - Elektor Junior Computer,” Elektor 1/2005: www.elektormagazine.com/magazine/elektor-200501/17897
[4] 3.5” Display: www.lcdwiki.com/3.5inch_Arduino_Display-UNO
[5] NPL Time & Frequency Services, “MSF 60 kHz Time and Date Code”
www.npl.co.uk/products-services/time-frequency/msf-radio-time-signal/msf_time_date_code

Advertisement

PERFORMANCE. WE are here for you!

RELIABILITY. Join our free webinars on:
www.we-online.com/webinars
SERVICE.
Optocouplers by Würth Elektronik
With the new optocouplers, Würth Elektronik presents one of the latest additions
Optocouplers by Würth Elektronik to its optoelectronic product portfolio. The innovative design features a coplanar
structure and high-grade silicon for total internal reflection. The coplanar design
ensures the isolation gap stay fixed during the production process and provide
perfect isolation and protection for your application. The total internal reflection
provide stable CTR over the whole temperature range and high CTR even at low
current operation.

Provided in all industry standard packages. Available with all binnings ex stock.
Samples free of charge: www.we-online.com/optocoupler

• Innovative coplanar design • Stable CTR over whole

© eiPal

• High grade silicon encapsulation temperature range

• Copper leadframe for high reliability • High CTR in low current operation

lektor July & August 2022 41

 background
Moisture Sens rs for
Watering Systems
Automatic Watering
By Peter Tschulik (Austria) and
Christof Hübner (Germany)
Before you start working with
moisture sensors, consider the
following tips about sensor selection
and sensor integration. Once you
are familiar with the basics, you can
begin integrating the sensors using
the Arduino IDE for boards such as
Arduino and ESP.
For many years, I have been using automatic watering systems to make
gardening more convenient. My first system consisted of a simple
timer clock with inexpensive solenoid valves, like the ones used in
washing machines. The next generation featured an Arduino board
with a display, and it obtained its information from simple capacitive
moisture sensors, and even maintained a watering log. My current
watering system is based on an ESP32 and is fully integrated into
my home automation system. This allows me to manually switch on
the watering system, view the most recent watering times, set these
times and other parameters, and much more. I hope you can benefit
from my experience.
 
The above description is not entirely complete. For instance, there is also
an alarm function that reports things like when a watering session failed,
for example due to a broken hose. The system can also be operated from
a web browser and the software can be updated over the air.
 
The watering system is, however, not the focus of this article because
it is very much designed to meet my specific needs (my patio). Instead,
I would like to describe my experience in selecting moisture sensors,
since I have learned a lot about them over the years.
42 July & August www.elektormagazine.com
Moisture sensors have two tasks in a watering system: they inform the
system about the current watering status, and as a result they help
to make economical use of water as a resource. Quite a few plants
are equally intolerant to waterlogged or dry soil. Although there are
an enormous number of moisture sensors, it appears that only a few
really work properly.
After a brief overview of the operating principles of the most common
moisture sensors, this article focuses on sensor integration using the
Arduino IDE for commonly used boards such as Arduino or ESP. If
you are currently using a watering system without sensors or are not
satisfied with your moisture sensors, you’re in the right place here.
Measuring Soil Moisture
There are a number of different definitions of soil moisture. The volumet-
ric water content, which means the proportion of water in the total
volume of soil, is usually what is wanted. If all air spaces in the soil are
filled with water, the maximum volumetric water content in natural soil
is approximately 50 to 60%. Higher values are possible in peat and
other organic materials.
 
The gravimetric water content relates to the relative mass of water in
the soil and can be determined, for example, by weighing a soil sample
before and after drying. This parameter must be measured in a lab,
so it is not suitable for the continuous monitoring of soil moisture.
 
Electrical soil moisture sensors are based on a variety of physical
measurement principles. Measuring the electrical conductivity is very
inexpensive, but unfortunately not reliable because the conductivity is
also dependent on the amount of dissolved ions. With the same water
content, measuring the conductivity can lead to completely different
results depending on the type of soil, salt content and fertilisers. The
conductivity is also strongly temperature dependent, and compen-
sating for this is virtually impossible in practice. However, the biggest
problem is corrosion of the sensor electrodes, which often last only
a few days or weeks.

Capacitive sensors with electrodes protected by an insulating layer
are much better in this regard. The measurement principle is based
on the interaction of water molecules with an electric field, since water
molecules are dipoles with positive and negative ends. If a dipole is
placed between the electrodes of a plate capacitor and a voltage is
applied to the metallic electrodes, the dipole will align itself to the
electric field (see Figure 1).
 
This polarisation effect is expressed by the dielectric constant εr. It
ranges from εr = 1 for air through εr = 3 to 8 for mineral soil particles
to εr = 80 for water [1]. The effective dielectric constant of a given soil,
consisting of a mix of air, water and soil particles, is therefore largely
determined by the water content. If a soil sample is placed between the
electrodes of a capacitor and the resulting capacitance is measured,
the associated dielectric constant can be calculated and from it the
water content. In practice, an AC voltage is applied to the electrodes
to reduce electrolysis at the capacitor electrodes, disturbances from
surface polarisation, and the influence of soil conductivity. Increasing
the frequency of the measurement signal leads to more accurate results
and increases the immunity to interference. Frequencies above 100 MHz
are ideal, but would make the circuitry significantly more expensive.
Conductivity Sensors
Measuring principle: The soil moisture is determined by measuring the
electrical conductivity of the soil between two electrodes. The sensor
is usually supplied with an analog signal processing circuit with an
adjustable threshold value (see Figure 2).
 
Disadvantages: The electrical conductivity of soil depends not only on
the water content, but also strongly on the salt content, fertilisers and
soil type, and is therefore an unreliable measure of soil moisture. The
Figure 1: Orientation polarisation of dipole water molecules in a
plate capacitor.
Figure 3: Example of a low-frequency
capacitive sensor (source: [3]).
sensor is exposed to strong corrosion in the soil and consequently has
a short lifetime. This also applies to electrodes with thin gold plating.
In addition, the terminals are usually not protected against water.
 
Advantage: Very economical.
Low-Frequency Capacitive Sensors
Measuring principle: The soil moisture is determined by measuring the
capacitance. The electronic circuit operates at low frequencies in the
range of several kilohertz to around 1 MHz. Many different versions can
be found online, often working with an NE555 timer (see Figure 3).
 
Disadvantages: Due to the low operating frequency, moisture measure-
ments are more strongly influenced by other factors such as the electri-
cal conductivity and temperature of the soil, so only limited accuracy
is possible. The electrodes are often only protected by a thin layer
of solder resist, which does not last long in the soil. The circuit and
terminals are often not sealed water-tight.
 
Advantage: Fairly economical.
VH400 High-Frequency Capacitive Sensor
Measuring principle: The soil moisture is determined by measuring
the capacitance.
 
Disadvantages: Medium price range. The influence of the soil type
on the measurement is still noticeable. The electrodes are very close
together, so the electrical measurement field barely extends beyond the
circuit board and the suitability of the sensor for large-grained soil is
limited (see Figure 4). The measurements are temperature dependent
and change when the connecting cable is touched with the hand [5].
Figure 2: A sensor for
conductivity measurement (YL69
or similar; source: [2]).
Figure 4: VH400
sensor (source: [4]).
lektor July & August 2022 43

Figure 5: CWT-SOIL-H-S
sensor (from AliExpress;
source: [6]).
Figure 6: SMT 50 high-frequency capacitive sensor
(source: [7]).
Figure 7: DAV-6440
sensor (source: [8]).
Figure 8: SMT 100 TDT
sensor (source: [9]).
44 July & August www.elektormagazine.com
Figure 9: Operating
principle of a TDT sensor.
The moisture-dependent
pulse propagation time
of the loop conductor is
converted into a frequency.
Advantage: Sufficiently accurate moisture measurement in many types
of soil.
CWT-SOIL-H-S
Measuring principle: No information available regarding the measur-
ing principle or measuring frequency.
Disadvantage: Medium price range. Frequent failures after a short
time in the soil.
Advantages: Holds the promise of additional measurement of electri-
cal conductivity; various versions (see Figure 5).
SMT 50 High-Frequency Capacitive Sensor
Measuring principle: The soil moisture is determined using a capacitive
voltage divider. The circuit operates at 16 MHz or above.
Disadvantages: Medium price range. The influence of the soil type on
the measurement is still noticeable.
Advantages: The measuring frequency is high enough for reliable
moisture measurement in many types of soil. The electrodes are embed-
ded in a multilayer circuit board, providing good mechanical protec-
tion and long lifetime in the ground. The circuitry is encapsulated in
epoxy to make it completely waterproof (see Figure 6). The 10 m long
connecting cable with a polyurethane sheath is suitable for under-
ground installation (resistant to microbes and hydrolysis).
DAV-6440 Tensiometer
Measuring principle: This Watermark sensor is an example of a tensi-
ometer, which measures the soil moisture tension (correlated to the
water content of the soil). For this the electrical resistance of a material
in hydraulic contact with the soil is determined (see Figure 7). Other
tensiometric sensors use a non-woven fabric in contact with the soil
and determine its water content by electrical heating while observing
the temperature (e.g. Gardena sensors).
Disadvantages: Tensiometers are sluggish and respond slowly to chang-
ing soil moisture. There is a hysteresis effect, resulting in different
measurements with the same soil moisture tension (with rising versus
falling water content). Under very dry conditions the sensor can lose
hydraulic contact with the soil. The stated lifetime of this sensor in the
ground is at least 5 years, which is not especially long.
Advantage: The soil moisture tension indicates the soil moisture avail-
able to the plants.
SMT 100 TDT Sensor
Measuring principle: This time domain transmission (TDT) sensor
consists of conductor loops forming a high-frequency transmission
line that is driven by pulse signals (see Figures 8 and 9). A ring oscil-
lator is implemented by feedback to an amplifier, and its frequency is
measured by a microcontroller. The capacitance between the electrodes
is influenced by the soil moisture. The higher the moisture content,
 Watering Strategy
Comparison of time-controlled and sensor-controlled watering.
 
Conventional time-controlled watering irrigates in fixed time inter-
vals, regardless of the actual need for watering. It can therefore lead
to overwatering with harmful waterlogging or to underwatering and
dry stress, depending on weather conditions. This can be mitigated
by using a rain sensor or online weather information. Sensor-con-
trolled watering is much better. This not only allows optimal water
supply to the plants, but also reduces water consumption. The
differences are shown in the chart.
 
There are various strategies for setting the threshold levels. An
experienced gardener can recognise the start of dry stress by
observing the plants (slight wilting, rolling of leaf edges). Exces-
sive watering results in saturation, with no further increase in soil
moisture. If you wait a while (around 24 hours), the excess water
will drain into the subsoil. As a rule of thumb, you can set the lower
threshold to 60% of the saturation level. During the learning phase,
observe the plants together with the moisture trend and adjust
the threshold levels as necessary to achieve long-term optimal
and economical automatic watering. It is important to position the
sensor properly in the root area of the plants. To obtain a good soil
moisture cycle with drip irrigation, the sensor should not be placed
too far away or too close to the drip location.
the longer the pulse propagation time and the lower the frequency
of the ring oscillator.
 
Disadvantage: Upper price range.
 
Advantages: The measuring frequency is over 150 MHz, enabling high
accuracy with good resolution combined with minimal influence of
the soil type on the measurement. Thanks to a multilayer circuit board
with epoxy encapsulation and a high-grade cable, the sensor achieve
a long lifetime in the soil. A large selection of interfaces is available,
including analog output, RS-485 (TBUS/ASCII and Modbus), SDI-12
and 4-20 mA current loop.
 
The SMT 100 sensor is often used for professional purposes (precision
agriculture or scientific applications in hydrology and soil science).
Integration of Sensors in Watering Systems
As previously mentioned, low-cost sensors have analog interfaces
and only higher-grade sensors have digital interfaces. Simple analog
sensors can be integrated easily in the Arduino IDE without additional
libraries. All you need is an analog input. Pin A0 of an Arduino Uno is
used in the following example program.
int SENSOR_PIN = A0;  /* select analog pin */
int SENSOR_VAL = 0;     /* variable storing sensor value */
void setup()
{
  Serial.begin(9600);    /* setup serial connection */
}
void loop()
{
  /* read value */
  SENSOR_VAL = analogRead(SENSOR_PIN);
  Serial.println(SENSOR_VAL);    /* output value */
}
 
Higher-grade sensors, such as the CWT-SOIL-H-S or SMT 100, provide
digital interfaces. In my system both sensors are connected through
a Modbus RS-485 interface. The advantages of this connection are
obvious: Long cable runs (up to 100 m from the sensor to the control
unit) are not a problem, and data transmission is immune to inter-
ference. In addition, many sensors can be connected over the same
interface.
 
The CWT-SOIL-H-S sensor basically works well, but in my system two
sensors simply stopped working after two years or started indicating
unrealistic values. In addition, the included user guide is understand-
able but the configuration tool described in the user guide is difficult
to find online and is only in Chinese. For this reason, only the integra-
tion of the SMT 100 sensor is discussed below.
 
First some remarks about RS-485 and Modbus. RS-485 is a differ-
ential serial interface with two lines that transmits data in differential
mode, which suppresses common-mode interference and improves
the signal-to-noise ratio compared to RS-232. Robust connections
up to 100 m long can be made with simple unshielded twisted-pair
cables. Modbus is a client/server communication protocol, developed
in 1979 for communication with programmable logic controllers. This
open protocol has become an accepted industry standard. All techni-
cal documents are freely available at [10]. Each node basically has a
unique address for accessing the node.
Configuring Sensors
For initial experiments with the SMT 100, it is advisable to use a USB/
RS-485 bridge or a PCIe RS-485 card so that the sensor can commu-
nicate directly with a laptop or desktop computer. I use a well-docu-
mented USB adapter based on an FTDI IC.
 
A very convenient Modbus configuration program can be downloaded
from the website of the sensor manufacturer [11]. After unpacking and
installation, proceed as follows to connect the RS-485 version of the
lektor July & August 2022 45
 Figure 10: Modbus test program for the SMT 100 sensor.
Figure 12: Output in the Arduino IDE Serial Monitor window.
sensor to the RS-485 interface: The white wire of the sensor cable is
ground, the green wire is RS-485 A, and the yellow wire is RS-485 B. A
supply voltage in the range of +4 V to +24 V (with respect to ground)
must also be connected to the brown wire.
 
After launching the Modbus tool, you first have to select the COM
port. If you are using a USB adapter and no port is shown, you need to
install the appropriate driver. If everything is okay, the window shown
in Figure 10 appears.
 
Now you can read out the individual parameters or change the Modbus
address. By default the sensor is accessible at address 253 (Broad-
cast). If you want to connect multiple sensors, each sensor must have
a unique address. In the example program the Modbus address is set
to 1 by the instruction Set new MODBUS Address. To be on the safe
side, you can check communication via address 1, for which Modbus
Address on the left must be set to 1. Note that only one sensor at a
time can be connected to the bus when setting the address.
A Test Program
Now let’s see how this sort of sensor can be integrated into a watering
system. The test setup is shown in Figure 11.
46 July & August www.elektormagazine.com
Figure 11: Test setup for connecting an SMT 100 sensor to an ESP32.
Along with the readily available ESP32-DEV-Kit, it consists of a small
board that converts the RS-485 signal from the sensor into TTL levels
suitable for the microcontroller. Terminal DI is an input for serial data
transmission from the ESP32 to the sensor. Terminal RO receives data
from the ESP32 to be sent to the sensor. Terminals DE and RE are
connected together here. When these terminals are pulled low, the
receiver in the MAX485 IC is enabled, and when they are pulled high
the MAX485 transmitter is enabled.
 
Now let’s look at the test program [12]. The first part (before the setup
routine) defines the pins and the variables for the sensors. In the setup
routine the pin for switching the transmitter/receiver is initialised, the
standard serial interface is defined as the debug interface, the serial
interface to the sensor is initialised as ‘Hardware Serial Interface 2’,
and a subroutine checks whether the sensor is connected properly.
 
In the main loop the sensor is polled every three seconds by calling the
subroutine SUB_RECEIVE_HUMIDITY, and the result is output. Querying
the sensor values is handled by three subroutines: SUB_CHECK_HUMID-
ITY checks the received sensor address for correctness to verify that
a sensor is connected, SUB_RECEIVE_HUMIDITY fetches the sensor
value, and SUB_CALCULATE_CRC calculates the CRC checksum. The
complete, extensively commented example code can be downloaded Questions or Comments?
free of charge from [12]. If the output in the Arduino Serial Monitor Do you have any questions or comments about this article? Contact
window is similar to that in Figure 12, everything is working right. the author at peter.tschulik@chello.at or the Elektor editorial team
210684-01 by email at editor@elektor.com.

WEB LINKS
[1] C. Hübner, U. Kaatze, Electromagnetic moisture measurement, Universi-
RELATED PRODUCTS tätsverlag Göttingen: https://doi.org/10.17875/gup2016-958
[2] YL69 conductivity sensor: https://bit.ly/YL69-soil-sensor
[3] Capacitive sensor: https://bit.ly/cap-sensor
> ESP-C3-12F-Kit Development Board with [4] Vegetronix VH400: https://www.vegetronix.com/Products/VH400
Built-In 4 MB Flash (SKU 19855)
www.elektor.com/19855   [5] Sensor comparison (German language): https://bit.ly/SMT50-vs-VH400
[6] CWT-SOIL-H-S sensor:
> MakePython ESP32 Development Kit https://de.aliexpress.com/item/1005001524845572.html
(SKU 20137) [7] SMT 50 sensor: https://www.truebner.de/en/smt50.php
www.elektor.com/20137   [8] DAV-6440 sensor: https://bit.ly/DAVIS6440
> Bundle: Getting Started with ESPHome + [9] SMT 100 sensor: https://www.truebner.de/en/smt100.php
LILYGO TTGO T-Display ESP32 (16 MB) [10] Modbus: https://modbus.org
(SKU 19896) [11] SMT 100 Modbus tool:
www.elektor.com/19896 https://www.truebner.de/download/ModbusTool_Installer_V3.zip
[12] Example code: http://www.elektormagazine.com/210684-01

Advertisement

Development Tools alle

an einem Ort
Tausende Tools von hunderten Wählen Sie Ihr Produkt aus
unserer breiten Palette auf
zuverlässigen Herstellern mouser.de/dev-tools

dev-tools-workbench-185x124mm-5mmbleed-DE.indd 1 17.05.22 12:04

lektor July & August 2022 47

 background

Disruption in
Test and
Measurement Equipment
Innovation from the Smaller Players
FOCUS ON

Test &
Measurement
By Stuart Cording (Elektor)

There’s nothing more satisfying than seeing a lab stocked with

test and measurement equipment. It is a sign that things are being
undertaken correctly. However, the cost of traditional tools can be
prohibitive for startups, smaller businesses, and students. So what is
being done to make measurement more accessible to the masses?

Scouring the halls and stands of an electronics exhibition such as
electronica or embedded world, it is very easy to get excited about
the rows of test and measurement (T&M) equipment emanating
from the world’s foremost suppliers. Color displays, smooth and
soft buttons, and knobs that provide a reassuring haptic bump on
rotation exude quality, reliability, and accuracy in measurement.
But this all comes at a price. Even entry-level, 50-MHz bandwidth
oscilloscopes from the big names, like Rohde & Schwarz or Keysight,
won’t leave you much change out of €1,000. Signal generators
and spectrum analyzers start at around €2,000. And, if you want
to be more mobile, perhaps with a handheld, you’re looking at
significantly more.
 
The strain on budgets quickly becomes apparent when it comes
to equipping laboratories. This is exacerbated by the use cases
of the typical education or research and development environ-
ment. While a student or engineer may spend all day at the bench,
individual items of T&M will see differing amounts of use. Some
equipment may be used the entire day, while others are used only
weekly. Oscilloscopes and spectrum analyzers are also cumbersome
items, making them awkward to keep moving around. With the
complexity of applications continuously rising, T&M equipment
is also regularly incorporated into test automation systems during
development. In such cases, a single tool may be locked into a setup
for days or weeks, stopping others from using it.
 
For some time now, the traditional T&M suppliers have been under
attack from disruptors. Startups and electronics industry veterans,
frustrated by the complexity of traditional tools, drab exteriors,
and underperforming user interfaces, have been delivering some
competent alternatives that deliver exceptional capability for a
fraction of the price of a benchtop alternative. This has changed
the way engineers equip their laboratories, and it is also changing
how students learn.
Lab in a Box
Traditionally, sourcing an oscilloscope, power supply, spectrum
analyzer, waveform generator, and voltmeter required a healthy
budget and a robust table, with each item supplied as an isolated
piece of equipment. Over the years, and linked to the growth in
automated test systems, suppliers such as National Instruments
have developed modular hardware T&M solutions, such as PXI.
These consist of a chassis into which a range of measurement cards
can be inserted, enabling the creation of custom instrumentation
solutions. But what if the hardware could remain the same, with
just the software defining the measurement capability?

48 July & August 2022 www.elektormagazine.com

 ELEKTOR INDUSTRY
Figure 1: The Moku family uses an instrument-on-chip (IoC) approach,
featuring highly capable analog coupled with a reconfigurable FPGA to
implement the various supported capabilities. (Source: Liquid Instruments)
This is the question raised by the five founders of Liquid
Instruments, a team based just north of San Diego, California. The
group of scientists, who had participated in experimental physics
and research into gravitational waves, found the existing T&M
offering to be inflexible and lacking in scalability. The availabil-
ity of powerful system-on-chip (SoC) field-programmable gate
arrays (FPGAs) made them feel there was a basis for developing a
common hardware platform that could be used in many different
ways (Figure 1). Their goal became to enable users to build measure-
ment solutions that even the team hadn’t thought of.
 
The result of their efforts is the Moku range of software-defined
T&M solutions. At the lower end of the price spectrum is the
Moku:Go (Figure 2), an engineering platform designed to fit easily
in a student’s or engineer’s backpack. Measuring just 24 × 13 cm and
just 3.8 cm high (9.4” × 5.1” × 1.5”), it features two 12-bit, 125-MSa/s
analog input channels, offers 30-MHz bandwidth, and has an input
range of ±25 V. On the output side, there are two 12-bit, 125-MSa/s
analog outputs with 20-MHz bandwidth and a ±5-V range, plus 16
channels of digital I/O. Four power supply outputs that support
between 150 mA and 1 A and -5 V and 16 V, model dependent,
round off the tool’s features.
Grab Your Tablet
Like similar tools of this kind, it relies on other equipment for the
graphical user interface (GUI). While laptops and PCs can access
the GUI through a browser, Moku is optimized for use with the
touch screen of an iPad. Wi-Fi provides the Moku:Go M0 and M1’s
connectivity, while the M2 also features an Ethernet port. Thanks
to careful consideration for the GUI design, the interface looks
clean and is intuitive.
 
Out-of-the-box, the Moku:Go [1] offers 11 different instruments,
including an oscilloscope, standard and arbitrary waveform
Figure 2: The Moku:Go is the entry-level, portable test and measurement
instrument from Liquid Instruments. (Source: Liquid Instruments)
generators, logic analyzer, and spectrum analyzer. Beyond this
are advanced instruments, such as an FIR filter builder, data logger,
20 MHz lock-in amplifier, and multi-input/output PID controller [2].
This last instrument highlights the capability of such equipment,
enabling up to two PID controllers to be implemented, configured,
and monitored. The interactive Bode plot offers real-time control
and multi-section PID builder configurations for superb insight into
how it functions for those building or teaching control systems.
 
Talking to Michi Yoneda, one of the Liquid Instruments team, he
explains that the Moku line-up frequently finds a home in university
labs and research institutes. There it is used for everything from
photonics and LiDAR, to spectroscopy and quantum computing.
The ability to change instrument on-the-fly and its programmable
interface make it a valuable alternative to costly, traditional T&M
equipment. Users can also develop their own instruments even
without access to traditional FPGA development tools. Thanks to
their Cloud Compile [3] platform, those proficient in VHDL can
create the bitstream required. Currently, this capability is limited
to the high-end Moku:Pro, but there is a chance that the feature
will be rolled out to the rest of the range.
 
One concern of such tools is their accuracy. For sure, they are
suitable for garnering a general view of signals on a printed circuit
board (PCB), but could I base promises to a customer on the value
measured? Yoneda ensures me that, yes, this is possible, thanks
to a third party who offers calibration together with a certificate.
From Frustration to Logic Analyzer
Having set up probes on a serial interface and found a suitable
decoder on your logic analyzer, one of the most frustrating things
is discovering that it requires a paid license. After all, the hardware
is already capable of the job, and you’ve paid for the tool. And other
times, it is simply a lack of access to affordable test equipment
lektor July & August 2022 49

Figure 3: The entry-level Logic 8 from Saleae offers eight channels of
mixed-signal capture in a tiny machined aluminum package that is just 5.3
cm on a side. (Source: Saleae)
which leads to the forming of businesses like Saleae, the devel-
oper and manufacturer of the popular USB-based logic analyzers.
 
Mark Garrison, Saleae’s Vice President of Engineering, explains
that the initial trigger to develop their tools came when his brother
worked on an embedded design. Back then, a benchtop logic
analyzer wasn’t financially viable, not even used on eBay, so he’d
procured a USB-based device. The only issue was that this tool
required more debugging than the product he was building.
 
Over the years, their logic analyzer line-up has developed, but
Figure 4: Logic 2 provides the user interface for Saleae’s tools. It is exceptionally easy to use, enabling edges to be located quickly and serial interfaces to be
decoded.
50 July & August 2022 www.elektormagazine.com
today it is the Logic 8 (Figure 3) that kicks off their offering [4]. It
provides eight mixed-signal inputs that can record digital, analog,
or both. With a digital sampling rate of 100 MSa/s, and 10 MSa/s
for analog, it covers the majority of use cases found in research
labs and education.
Smooth Moves
What really stands out is their Logic 2 software (Figure 4), the
user interface to the tool that runs on Windows, Linux, and Mac
OSX. The GUI is intuitive, allowing the user to enable and disable
channels, name them, and the color-coding used relates to the
colors used for the unit’s connecting cables. Decoding of standard
serial protocols is also supported, from I2C, SPI, and UART, to
MIDI, Modbus RTU, and even the HD44780 interface used on
alphanumeric LCDs. The device is also frequently used for data
logging, with developers using the Python application program-
ming interface (API).
However, compared to a traditional benchtop, the ease with which
signals can be reviewed in the software is what really blows the
user away. Garrison explains that the ease of use of the Google
Maps user interface had proven to be an inspiration at the time of
initial development. The GUI enables consecutive edges in the data
stream to be found easily, even when widely spaced. ‘Flicking’ the
data stream left or right scrolls the logic output until it slows to an
eventual stop if not held earlier. This is a vast improvement over
the knobs and buttons of traditional test equipment.
 
Today, around 80% of sales are into professional engineering teams,
many of whom value the high-resolution data capture offered in
this modest device. The remaining customer base is makers and
 ELEKTOR INDUSTRY

hobbyists, education, and startups, happy to have access to this

caliber of tool at a more acceptable price point. This is reflected
in their active user forum [5], where the community helps and
supports one another, both with the tools and developer questions.
They also create extensions, a range of Python-based packages, that
can deliver metrics on the data captured, such as signal jitter, or
perform higher-level analysis of decoded data streams.

Lab in Your School Bag

While education establishments have done an excellent job of
building on the wave of interest for all things digital by deliver-
ing embedded programmers and app developers, enthusiasm for
analog amongst students has fallen sharply. This was the experi- Figure 5: The ADALM1000 (MK1) gives electrical engineering students a full
lab experience in their dorm or home. (Source: Analog Devices)
ence of Robin Getz, Director of Systems Engineering at Analog
Devices, in discussion with a professor at MIT one decade ago. At
the time, of 200 engineering students, only around four focused
on analog; the rest concentrated on software and digital. Such
insight spurred the development of a collection of low-cost Active
Learning Module platforms designed to provide easy access to both
analog and RF technology.
 
The range starts with the ADALM1000 [6] (also known as the M1K),
a self-contained tool that enables students to explore the relation-
ship between current, voltage, and impedance (Figure 5). Powered
and configured via USB, it offers two channels for signal generation
and measurement for both voltage and current. Its 16-bit, 100 kSa/s
oscilloscopes and function generators are more than adequate
for even intermediate study of analog concepts. Supporting it is
the simple-to-use, open-source PixelPulse 2 software [7] that runs
under Windows and Mac OSX. Linux users can build the applica-
Figure 6: Now standard equipment for Analog Device’s FAEs, the
tion themselves. A more traditional user interface is provided with ADALM2000 (MK2) includes an oscilloscope, function generator, logic
the software package ALICE [8]. analyzer, and pattern generator. (Source: Analog Devices)
 
Performance is pushed up a notch with the
ADALM2000 [9] (also known as the M2K)
(Figure 6). Two analog inputs (100 MSPS,
25 MHz) and two analog outputs (150 MSPS,
30 MHz) are complemented by two variable
power supplies (0 to ±5 V at 50 mA). There are
also 16 digital I/Os and two digital triggers. This
tool uses Scopy [10], a fully-featured instrument
GUI that runs on Windows, Mac OSX, and Linux
(Figure 7). The software also works on Android,
with the tool connected via an OTG adapter to
a tablet.

Measurement on Laptop and Tablet

Support for Android tablets is intentional, says
Getz. “We notice that students follow our lab
material on their laptops while using the M2K
with their tablet in the US. However, in Asia,
most students only have a tablet.” Students are
provided with a wealth of learning material
at the Engineering University Program [11] Figure 7: The GUI is critical for the success of any headless measurement tool. Scopy provides
website. clear access to its features and settings on all operating systems and Android tablets.

lektor July & August 2022 51

 
Rather than choosing a single learning institute’s curriculum, kitted out with their tools. As long as they work within the tool’s
the team has focused on creating an amalgam of all the course limits, there are a host of design challenges they can tackle when
materials institutes cover when teaching engineering students the supporting customers in the field. And, he notes, a woodworker
basics of electronics and components. In the US, students using doesn’t just own one hammer and one saw, so why should FAEs
the Analog Devices tools typically undertake the lab work at home solely rely upon benchtop test equipment?
rather than in the lab, then discuss the results and any anomalies 220201-01
with staff back at their college or university. They can also make
use of LTspice [12] to compare simulations with practical results.
During the restrictions imposed by Covid, this has ensured that Questions or Comments?
learning can continue. But it also raises the question, when do Do you have technical questions or comments about the
students get to learn about ‘proper’ test equipment? article? Email the author at stuart.cording@elektor.com or
  contact Elektor at editor@elektor.com.
“Benchtop equipment is still needed at some point,” says Getz.
“We see ourselves providing an on-ramp to understand how to
measure and test.”
 
Students are also spending more time studying the material. Those
RELATED PRODUCTS
using the M1K and a supply of components provided by Analog
Devices at the University of Limerick, Ireland, found students
engaged more deeply and spent more time validating the concepts > PicoScope 2204A (10 MHz) (SKU 17303)
covered in the course. Professor Hayes, the head of The Electronic www.elektor.com/17303  
& Computing Engineering Department, said that he sees students
become experts more quickly, just by “learning how to recreate > ScanaQuad SQ200 Logic Analyzer & Signal Generator
their lab environment” through their practical work [13]. (SKU 18103)
www.elektor.com/18103  
Test and Measurement Changed Forever?
Benchtop T&M is not going anywhere soon. Industry needs
standardized, calibrated commercial off-the-shelf (COTS) equip-
ment that integrates easily with common software program-
ming tools like National Instruments LabVIEW or MATLAB from
MathWorks. But, when it comes to education, price, and porta-
bility, the tools offered by these disruptive players offer masses
of quality measurement capability at a fraction of the outlay for
equivalent big-name T&M suppliers. Analog Device’s Getz also
highlights that their Field Applications Engineers (FAE) are all

WEB LINKS
[1] Moku:Go Product Page: https://bit.ly/35tKNmz
[2] “Multiple-Input Multiple-Output PID Controller,” Liquid Instruments: https://bit.ly/3NF3Zin
[3] “Moku Cloud Compile: A Getting Started Guide,” Liquid Instruments, September 2021: https://bit.ly/35tWR7m
[4] Saleae Logic Analyzers Product Page: https://bit.ly/3qWyt62
[5] Saleae User Forum: https://bit.ly/376dAyf
[6] ADALM1000 Product Page: https://bit.ly/3tXoFdE
[7] Pixelpulse Software: https://bit.ly/3NRfdkj
[8] ALICE Active Learning Interface (for) Circuits (and) Electronics Software: https://bit.ly/3tYFi96
[9] ADALM2000 Product Page: https://bit.ly/36G6IaZ
[10] Scopy Software: https://bit.ly/3Jcj2Nr
[11] Analog Devices Engineering University Program: https://bit.ly/3LE3OCt
[12] LTspice Product Page: www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html
[13] “UL and Analog Devices partner on digital learning tool to recreate lab environment at home,” University of Limeric,
December 2020: https://bit.ly/3uOZVDB

52 July & August 2022 www.elektormagazine.com

November 15–18, 2022

Driving
sustainable
progress.
Discover the future today.
At electronica 2022.

Save the date!

 Elektor infographic
By Robert van der Zwan

CTM: Growth Stronger Than TM

(and Staying Strong)
The coming years, the market for Communications
Test and Measurement (CTM) will grow much 11.20 USD Billion
stronger than the market just related to Test and
Measurement (TM). This is nothing new: in 2020, Yearly growth
we reported a 9% growth rate for the CTM market rate 8%
against 4% for ‘just’ TM. However, due to Covid-19,
one can now see a little less optimism for the global
6.53 USD Billion
CTM market during the coming years. Some market
research companies think that an 8% growth rate
would be more realistic, or even want to stick to a
7% estimate. Let us go for 8%, based on various
research carried out for market size and growth rate.

(Sources: Data Bridge Market Research;

MarketsandMarkets; Research and Markets; Vantage
Market Research)
2021 2028

Growth Rate CTM Market (USD Billion)

Sensors: Yes, It Is the Supply Chain

What applies to the global market
Global Sensor Market, 2021-2026 (USD Billion) for Communications Test and
Measurement, also applies to the global
USD Billion sensor market: demand stays high
the coming years, but Covid-19 has
disrupted the supply chain. Knowing
that eventually the supply disruptions
300 249.6 will end, the high demand has the last
word, leading to a growth of about
250 r year
9.8% pe 10% in the period 2021-2026. This
200 high growth rate has to do with the
156.6 fact that there isn’t any sector that
150 is not interested in smart (or not so
smart) sensors. Applications range
100 from reducing maintenance costs to
enhancing safety levels in transport and
50 healthcare.

2023 2024 2025 2026 (Source: Allied Market Research; BCC

2021 2022 Publishing)

54 July & August www.elektormagazine.com

 Market Shares Sensors, by Type, 2019-2028

2019 2028

It Is the

en nd

nso n

se and
Radar Sensor

sor

em xim or

rs
se itio
sor

r
nso

nso

nso

nso

nso

n a enso

nso

nso

he
ts a
s
en

en

pos
se

l se

se

se

se

se
en ity

ed er

Ot
es

es

s
Bio

speomet
r

re

el

ity
nd
a
da

uc

ag

v
tic
All Around

ssu

tu

d
Le
ac ro
To

Im
Ra

mi
ra

ler
Op

Pre

spl P

tio

Hu
pe

ce
Mo
m

Ac
Te

What is and will be the most promising
measuring device this decade? Or, to
phrase it slightly differently: who would
have thought that radar technology would
become so important? Radar technology
should no longer be associated with
aviation and shipping alone. What to
di
think of autonomous cars? Radar sensors
can not only detect an object, but can
also determine its shape, its direction of
movement and even its speed. Roughly
speaking, radar sensors make up 20% of
the total global sensor market. This would
amount to around $70 billion (USD) in 2028.
(Sources: Allied Market Research; Fierce
Electronics)

Where Are They Going? Where are all these sensors going? Which
sectors make the most use of all these
sensors? One might think that IT and Telecom
Market Demand Sensors, by Sector, 2019-2028 represent the most significant market for
sensors, given the growth in broadband
2019 2028 communication. Following that line of thought,
another high-growth sector, automotive,
would probably come in second. Yes, IT
and Telecom, as well as automotive, are still
going strong globally. But it is (consumer)
electronics that most definitely takes the
ics

al

ve

nse

rs
ar
co

he
tri

oti
ron

greatest chunk out of the sensor market

fe

hc
le

Ot
us

tom
ct

De
Te

alt
Ind

between now and 2028. From washing

Ele

Au
d

He
an

an

machines to fridges to game consoles, the list

IT

ce

is endless and that is precisely the point.

pa
ros
Ae

Source: Allied Market Research)

Covid-19 is an Impetus
for Growth?! Market Size Software Testing,
2021-2027 (USD Billion)
When compared to two years ago, marketing research firm Global
Market Insights is somewhat more optimistic when it comes to the 2027
global market of software testing. There are at least two reasons for > 60 USD Billion
that. Software applications in the cloud become more popular by
the day. Given the fact that these applications are updated regularly, CAGR 2020
software testing stays at the forefront of anyone’s mind. Another > 40 USD Billion
reason has to do with the pandemic. Covid-19 provides an extra 2021 - 2027
impetus for the growth of cloud applications. That is why Global
>7%
Market Insights thinks the software testing market will grow more
than 7% between 2021-2027 instead of the 6% estimate for 2020-
2026.

(Source: Global Market Insights)

lektor July & August 2022 55

 interview
Inspiration,
That’s What It’s All About
Interview with Entrepreneur Walter Arkesteijn,
InnoFaith Beauty Sciences
By Jan Buiting (Elektor)
In electronics, ideas, crude sketches and ‘triggers’ can prove more valuable than turnkey
circuits. Walter Arkesteijn explains how he applies this philosophy within his young
company.
Jan Buiting: In the electronics industry, development and
education, we come across many readers having a “Deja-vu”
when they see Elektor (Elektuur) again, often after making a
career. But it’s different with you since you are both “in e-busi-
ness” and an active Elektor reader today.
Walter Arkesteijn: That’s right, I’m still an active reader of Elektor.
There is a difference with the old days! While I am now a nostalgic
reader, I used to be a reader of Elektor at a fairly young age. It was the
right place for me to pick up all the knowledge and feed my boyhood
dream of creating. I can truly say that Elektor has contributed to the
person I am today. Through your magazine, I was able to accumu-
late a lot of knowledge and use it to think in terms of solutions and
options. It still stimulates my creative mind and I enjoy following
the innovations and different projects that are covered.
 
Jan: What did you think and feel about the magazine? Be frank!
Walter: I found and still find the magazine a source of information.
Especially back in the day when I started with my passion for electri-
cal engineering and I still had to learn to understand each table.
 
Jan: What is your technical background and how has Elektor
helped or thwarted you? What were your favorite projects or
articles and why?
Walter: My technical background is from a higher professional
education study in electrical engineering. Elektor magazine helped
me find my passion for e-engineering, which is why I decided to
follow this program. Reading the magazine whetted my appetite for
learning more, especially within the world of electrical engineer-
ing. I’ve always been given the scope to discover everything myself
and in this respect, Elektor has helped to inspire me.
 
Here my fascination goes mainly to the process of fantasy becom-
ing reality with the help of technology. If I had to mention an
56 July & August 2022 www.elektormagazine.com
Elektor product, it would be the collection of audio amplifiers
featured in the magazine. What fascinates me is how you can
specialize a lot within the same objective: audio amplification.
Through technology, there are multiple dimensions to the same
product. Here I find the conceptual insights and technical insights
very interesting.
 
Jan: You started your own company called InnoFaith. Briefly
describe how that went. What were your main motivations?
Hopefully not Elektor’s Corrections & Updates section?
Walter: I started InnoFaith Beauty Sciences [1] out of entrepre-
neurial ambition. I felt the personal urge to jump into the “oppor-
tunity hole” within the Aesthetic Medicine industry. I found out
that there was a lot to be gained when I was a teenager accompa-
nying my mother to a trade show — she is employed within the
beauty industry. I wanted to create the tool to help the industry
interpret and treat both skin problems and skin aging. This is how
Sylton was formed, the brand under which we sell these devices.
Elektor’s corrections were not an issue here. To me, Elektor really
is a breeding ground for imagination and knowledge.
 
Jan: Elektor as a magazine often propagates that it prefers to
inspire rather than instruct. How did you experience that?
What was your personal trigger condition, so to speak, to read
an article or undertake a project?
Walter: I experienced Elektor as inspirational and not ped-
antic. I still experience it as such. I didn’t do much building from
Elektor articles; the fun was quickly gone when it was already
thought out by someone else. To me, the main trigger con-
dition is this: after reading an article, a technological fascination
grows in me. Seeing different perspectives feeds my creativity;
it enables me to see opportunities that make me want to think
in new ways.
Partner Content

 
Jan: What does InnoFaith look like in terms of the organiza-
tion, and what is your product palette?
Walter: At this time, InnoFaith Beauty Sciences consists of a team
of 23 employees. We develop the software and hardware ourselves
within our team. Our main product, the Observ 520x, is sold through
distributors in different countries around the world.
 
It is vital to our product that it is easy to operate while technically
it is very sophisticated and intricate. Digital technology bridges
the world between our instinctive and analytical abilities to look
at beauty in a way that is visually intelligible and compelling in a
stimulating and inspiring way.
Jan: What electronics technologies are key at InnoFaith?
Walter: At InnoFaith Beauty Sciences we work with various items
where signal processing is especially important. But in addition,
software development and device design. We are always trying to
develop ourselves further within image processing and imaging.
For example, we work with different light modes and developed
an application required when using the Observ.
Partner Content
Jan: How would you outline the industry that InnoFaith is in?
What are the opportunities and challenges?
Walter: InnoFaith Beauty Sciences finds itself in a very interest-
ing industry where the end-user of the product is not a techie. So,
with us, it is very important the product, in addition to striving to
be innovative, should also be easy to handle.
I especially see a lot of opportunities within image processing
systems. The sector in which InnoFaith Beauty Sciences is present
is fueled enormously by the availability of new technology. In this,
the challenge is to collect a lot of image data in a short period of
time, combined with the different types of exposures. We want
to be able to capture increasing amounts of information from the
skin in order to properly understand the ratio between perspec-
tive and visual perception. We want to be at the cutting edge of the
perception of beauty and make this visible and understandable
through digital technology.
Jan: On the subject of technology availability, to what extent is
InnoFaith suffering from the crisis in the component market,
especially chips?
Walter: We, too, are facing a market with chip shortages, long
delivery times of electronic components, and sharp increases in
component prices, in addition to an overall disruption in the supply
chain. Fortunately, we have been able to minimize the impact of the
shortages by doing small redesigns, ordering on time and in large
quantities, making an accurate production forecast, and sourcing
difficult but critical chips on the open market through chip brokers.
Jan: The Netherlands wants to be a manufacturing country
again. What is your view on that?
Walter: I also see this trend, particularly in the production of serial
industrial technical products. In my view, the current labor short-
age is an inhibiting factor in this development. We have therefore
chosen to carry out the high-tech or critical production/control
in-house and to carry out the large, more labor-intensive serial and
space-consuming assembly steps in a partnership with a Slovakian
production company.
 
Jan: Thanks for the interview and we will be reading and
hearing more from you and your company soon, here in Elektor
Mag and on our YouTube video channel [2].
220257-01
WEB LINKS
[1] Innofaith Beauty Sciences: https://innofaith.com/
[2] Elektor TV: www.youtube.com/elektorim
lektor July & August 2022 57
 background
Minimizing EMC Interference
from Storage Chokes
By Ranjith Bramanpalli, Würth Elektronik
Normally, switching regulators cannot function without power inductors. However, if you
want to enhance their influence on EMC behavior, there are several places to start, including
shielding efficiency, winding start and the switching transitions.
DC/DC switching regulators are critical in
energy management, as they enable the
implementation of efficient switching
power supply for example. In this context,
storage inductors are critical components
of these switching regulators, even though
the development process often will only
focus on the pure electrical properties, such
as RDC, RAC or core losses. On the other hand,
electromagnetic radiation properties are
frequently overlooked. Figure 1 depicts
a typical switched DC/DC converter with
switches S1 and S2.
Power Inductors in Switched-
Mode Power Supplies
In switched-mode power supplies, power
inductors can be designed and assembled
using a variety of different core materials
and winding types. Additionally, power
inductors can be classified as unshielded,
semi-shielded, or shielded. Each type of
shielding has distinct advantages and
disadvantages that determine its applica-
tion areas.
 
A switching power supply generates an
alternating current (AC) voltage in the
inductor as a result of the switching
processes. Since an inductor can operate
as a loop antenna from a purely practical
standpoint, electromagnetic radiation is
dependent on a number of parameters.
These include the core and shielding
material as well as the start of the winding.
 
The electromagnetic radiation emitted
by a power inductor due to the switching
58 July & August 2022 www.elektormagazine.com
Figure 1: Typical switched-mode DC/DC
converter with switches S1 and S2.
Figure 2: Propagation spectrum in the near and
far field with a transition at λ/2π.
frequency and its harmonic waves in the
lower frequency range from 100 kHz to
30 MHz is not only dependent on the shiel-
ding of the coil, but also on the winding
properties. In contrast, the electromagne-
tic radiation’s dependency on frequency in
the upper range (30 MHz to 1 GHz), where
emissions are caused by harmonics and
their harmonic waves, is more dependent
on the shielding properties of the core
material, the switching frequency, and the
basic design.
Effects of Electromagnetic Field
When power inductors operate in DC/DC
converters, they produce undesired side
effects similar to those of a loop antenna.
AC voltage and current in the inductor
generate an electric field (E-field) and a
magnetic field (H-field). These two fields
propagate from the source at right angles
in opposite directions.
The properties of the E and H fields close
to this loop antenna (source) are dictated
by the source characteristics (switching
frequency, transitions, etc.). However, the
further away one moves from the source, it’s
the transition medium that determines the
field attributes. As a result, these distinct
yet connected phenomena can be classified
into two domains: the near field and the far
field (Figure 2). The near field is defined as
the area within λ/2π of the source, and the
far field is defined as the emissions beyond
this area.
The fields E and H must be evaluated
separately in the near field because their
ratio, denoted as the E / H characteristic
impedance, is not constant. However, in
the far field, these fields combine to form
a plane wave. As a result, the electric field
E and the magnetic field H are described
only in terms of the near field. When a
source is operated at a high current and a
low voltage, the magnetic field is dominant,
whilst when a source is operated at a low
current and a high voltage, the electric field
is dominant.
 
Partner Content
Figure 3: Test setup for the measurement of the
electric field in a DC/DC switching regulator
(DUT).
When a loop antenna is used, the magne-
tic field is very strong in the vicinity of the
source, resulting in a low characteristic
impedance in that region. With increa-
sing distance from the source, the magne-
tic field weakens and simultaneously
generates an electric field perpendicular to
the field H’s propagation direction. When
moving further away from the source,
the magnetic field weakens by a factor of
1 / r3 and the electric field weakens by a
factor of 1 / r2 (where r = radius denotes
the distance).
The characteristic impedance of a straight
wire antenna is high due to the strength
of the electric field in the source region.
The properties of attenuation are the polar
opposite of those of a loop antenna.
Radiation Behavior of Coils
As previously stated, electromagnetic
field radiation from storage inductors
in DC/DC converters is not negligible.
This is especially true when the type
and spacing of adjacent components are
considered along with their susceptibi-
lity to magnetic coupling. As engineers’
awareness of this potential EMC issue has
grown, component manufacturers have
responded by expanding their portfolio to
include shielded and semi-shielded coils
in addition to standard unshielded coils.
Shielded coils are manufactured in such
a way that the winding is totally encased
in a molded part made entirely of magne-
tic shielding material. In unshielded coils,
the coil windings are typically exposed and
there is also no magnetic shielding. Due
to the electromagnetic fields’ unrestricted
propagation, these are typically the most
powerful sources of EMI interference.
Magnetic materials are frequently applied
to the exposed windings of semi-shielded
coils using epoxy resin. The test setup for
measuring the electromagnetic field in a
Partner Content
Figure 4: H-field measurement results of unshielded, semi-shielded and shielded coils.
Figure 5: Comparative saturation behavior of a shielded (gray), semi-shielded (black) and unshielded
(red) coil.
DC/DC switching regulator (DUT) is depic-
ted in Figure 3.
 
As previously stated, each type of coil offers
both advantages and disadvantages. The
primary advantage of a shielded coil is
that its electromagnetic emissions are
relatively weak compared to semi-shielded
and unshielded coils. Figure 4 illustrates
the basic emission behavior of these three
fundamental types.
 
As many electrical engineers know,
designing a new circuit is always a delicate
act of balance. Exacerbating a desirable
quality can frequently result in amplifi-
cation of undesirable characteristics, the
maximum values of which are ultimately
constrained by project’s overall require-
ments. Inevitably, one of these constraints
is size. Shielded coils have a lower induc-
tance and magnetic saturation, as well as
greater manufacturing costs, as compared
to unshielded examples of comparable
dimensions. Obviously, designers with
less experience will be tempted to use an
unshielded coil due to its smaller size,
lower cost, and higher saturation currents.
However, this choice results in a slew of
EMC problems that are difficult to resolve
after the design phase.
Würth Elektronik is one of only a few
manufacturers to offer semi-shielded
coils capable of successfully bridging the
gap between space requirements, electrical
properties and EMC. These are particularly
well suited for applications for which the
components adjacent to the coils are not
highly sensitive to radiation.
 
The outstanding saturation properties of
the WE-LQS semi-shielded storage induc-
tor, size 8040 (744 040 841 00), are illus-
trated in Figure 5 in comparison with a
shielded coil from the WE-PD series, size
7345 (744 777 10) and an unshielded coil
from the WE PD2 series, size 7850 (744 775
10).
Effect of the Winding Start
An critical feature of EMC, which is
frequently overlooked, is the orientation of
the start of the winding, which is indicated
by a “dot” on the coil (Figure 6). It is impor-
tant to connect the coil’s side marked with
this dot as close as possible to the switch
node, as this side has the highest dU/
dt value and thus the most interference.
This way, the outer windings shield the AC
current flow from the switch node during
switching. If the unmarked end is connec-
ted to the switch node, the alternating
lektor July & August 2022 59
Figure 6: Coil types WE-XHMI and WE-PD2 Figure 7: E-field of the winding start connected to the switch node, and vice versa.
showing the “dot” that indicates the start of the
winding.

current forward voltages occur at the outer

winding. This can result in unacceptably
strong electrical or capacitive connections.
 
Magnetically shielded coils effectively
shield the dominant H-field radiation,
but are not always capable of shielding
the dominating E-field radiation. The
efficiency of E-field shielding depends
on the material properties and magnetic
permeability of the core material: the stron-
Figure 8: H-field of the winding start connected to the switch node and vice versa.
ger and more magnetically conductive the
core material, the more efficient the coil’s
E-field shielding.
  effect of various materials in the near and
The E-field emissions of a shielded power far fields is provided in the article titled
inductor manufactured by Würth Elektro- “The behavior of electromagnetic radia-
nik were measured as an example. The tion of power inductors in power manage-
switching regulator’s transistor switched at a ment” [1].
frequency of 400 kHz, generating the funda- 220295-01
mental resonance and subsequent harmo-
nics. The spectrum clearly demonstrates that
when dot-marked end of the coil is connec-
ted to the switch node, the E-field emissions About the Author
are significantly reduced (Figure 7). As a Ranjith Bramanpalli graduated from the Univer-
result, the correct orientation of the coil is sity of Massachusetts at Lowell in 2008 with two
applied. By contrast, the orientation of the master’s degrees in electrical engineering and
inductor has virtually no effect on the H-field computer science. He has since worked in the field
emissions (Figure 8). of power electronics, with a particular emphasis on
  research, development and application. He is currently
A detailed examination of the electro- employed with Würth Elektronik eiSos as a Product
magnetic interference signals caused by Application Engineer.
switching transitions and the shielding

WEB LINKS
[1] Würth Elektronik, “The Behavior of Electromagnetic Radiation of Power Inductors in Power Management”:
www.we-online.com/ANP047

60 July & August 2022 www.elektormagazine.com Partner Content

 tutorial
Free Download
for all Elektor
members: 60-p
with Parts 1 to age-PDF specia
10 of l
www.elektorma this series. Se
gazine.com/ne e
ws/GUI-PDF-Spe
cial

Create GUIs with Python:

Part 05

Tic-tac-toe
Use your GUI to control a simple game

N
0w that you have learnt how to make a basic
GUI, let’s add some more programming
logic behind the scenes to make your GUI
work as the means of controlling a game of tic-
tac-toe (also known as noughts and crosses).
Create a new file with the following code:

Laura Sach # Imports ---------------

MAKER

from guizero import App

Laura leads the
A Level team at
the Raspberry # Functions -------------
Pi Foundation,
creating resources # Variables -------------
for students
to learn about
Computer Science. # App -------------------
@CodeBoom app = App("Tic tac toe")

app.display()

Figure 1 A typical game of tic-tac-toe

Create the board
Let’s begin by creating the widgets which will make
up the game board. A traditional tic-tac-toe board
looks like the one shown in Figure 1.
You’ll use buttons to represent each of the
positions on the board, so that the player can click
Martin on one of the buttons indicating where they would
MAKER
Figure 1
If you run your program at this point, you won’t
see anything on the screen because the Box itself is
invisible.
Now let’s create the buttons to go inside it.
You will need nine buttons in total, so instead of
creating them individually, you can use a nested

O’Hanlon like to move. To be able to lay out the buttons on a loop to generate them all automatically and give
grid, let’s create a new type of guizero widget called them co-ordinates. First, add PushButton to
Martin works in
a Box. your list of widgets to import and then add this
the learning team
at the Raspberry A Box is a container widget. This means that it code immediately after the code for the board
Pi Foundation, is used for containing other widgets and grouping you just created.
where he creates
them together. Add it to the imports at the top of
online courses,
projects, and your code: for x in range(3):
learning resources. for y in range(3):
@martinohanlon from guizero import App, Box button = PushButton(
board, text="", grid=[x, y],
Set the Box to have a grid layout and add it to your width=3
app – before the app.display() line, as with all )
widgets.
Notice that there are two loop variables: x from 0
board = Box(app, layout="grid") to 2 and y from 0 to 2. As we iterate and generate
buttons, each button will be added to the board,

lektor July & August 2022 61

 which is the Box container you created earlier. The
button will be given the grid co-ordinates x,y,
meaning that each button is neatly placed on a
grid at a different position!
Your code should now look like tictactoe1.py.
The result of running it is shown in Figure 2.

Figure 2 Underlying data structure

Figure 2 A grid of nine buttons to play tic-tac-toe You might notice that when you create the buttons
using a loop, you are creating nine buttons
automatically and every single one is called

tictactoe1.py DOWNLOAD button. How will you be able to refer to each of

THE FULL CODE: these buttons in the program?
The answer is that you need an underlying data
> Language: Python 3 magpi.cc/guizerocode structure to hold a reference to each button, and
for this you will use a two-dimensional list.
001. # Imports --------------- Let’s create a function which we can call to clear
002. from guizero import App, Box, PushButton the board. It is a good idea to do this in a function
003.
004. # Functions ------------- so that you can reuse the code once the game has
005. been played to reset the board and allow the player
006. # Variables ------------- to begin a fresh game.
007.
008. # App ------------------- In the functions section, add a new function
009. app = App("Tic tac toe") called clear_board.
010.
011. board = Box(app, layout="grid")
012. for x in range(3): def clear_board():
013. for y in range(3):
014. button = PushButton(
board, text="", grid=[x, y], width=3) Your first job inside this function is to initialise
015. the data structure for the board. Let’s assume at
016. app.display() this point you have not created any buttons, so you
can initialise each position on the board as None
– the element in the list now exists but does not

tictactoe2.py yet have a value. Add the following line, indented,

to your function.

> Language: Python 3 new_board = [[None, None, None], [None,

None, None], [None, None, None]]
001. # Imports ---------------
002. from guizero import App, Box, PushButton Next, move the nested loop code from your app
003. section into the clear_board function. Make sure
004. # Functions -------------
005. def clear_board(): the indentation is correct.
006. new_board = [[None, None, None], [None, None, None], [ Inside the inner (y) loop, add a line of code
None, None, None]] to store a reference to each button at its x,y co-
007. for x in range(3):
008. for y in range(3): ordinate position within the two-dimensional list
009. button = PushButton( so that you can refer to it later.
010. board, text="", grid=[x, y], width=3)
011. new_board[x][y] = button
012. return new_board new_board[x][y] = button
013.
014. # Variables -------------
015. Finally, after the loops end, return the new_board
016. # App ------------------- you have just created. Your function should look like
017. app = App("Tic tac toe") this:
018.
019. board = Box(app, layout="grid")
020. board_squares = clear_board() def clear_board():
021.
new_board = [[None, None, None],
022 app.display()

62 July & August 2022 www.elektormagazine.com

tictactoe3.py Reset the game
> Language: Python 3 At the start, you wrote a function called clear_board.
This may have seemed unnecessary at the time, but
in actual fact it was thinking ahead to when the game
001. # Imports --------------- has ended. Since tic-tac-toe is quite a short game, it is
002. from guizero import App, Box, PushButton, Text
likely that someone might want to play more than one
003.
004. # Functions ------------- game in a row.
005. def clear_board():
006. new_board = [[None, None, None], [None, None, None], [None, Can you add a reset button to your game, which only
None, None]] appears once either someone has won the game,
007. for x in range(3): or the game was a draw? The button should call the
008. for y in range(3):
009. button = PushButton(board, text="", grid=[x, y], clear_board function and reset the turn variable as
width=3, command=choose_square, args=[x,y]) well as the message reporting whose turn it is.
010. new_board[x][y] = button
011. return new_board Hint: You will need to check the guizero documentation
012. to find out how to hide and show widgets, so that your
013. def choose_square(x, y):
button is not visible all of the time during the game.
014. board_squares[x][y].text = turn
015. board_squares[x][y].disable()
016. Hint: Create a new function which takes care of
017. # Variables ------------- everything you need to do to reset the game, and call
018. turn = "X" that function when the reset button is pressed. Don't
019.
forget that in your function you'll need to specify some
020. # App -------------------
021. app = App("Tic tac toe") variables as global.
022.
023. board = Box(app, layout="grid")
024. board_squares = clear_board() and run the program and you should see an identical
025. message = Text(app, text="It is your turn, " + turn)
026. result to the one you had at the end of the last step,
027. app.display() but now you have a hidden two-dimensional list
data structure to let you reference and manipulate
the buttons.
[None, None, None], If you want to see what your 2D list looks like, you
[None, None, None]] could add a print command to print the board_
for x in range(3): squares list: print(board_squares). You should then
for y in range(3): see nine lots of [PushButton] object with text ""
button = PushButton( appear in the shell.
board, text="", grid=[x,
y], width=3 Make the buttons work
) At the moment, your buttons don’t do anything
new_board[x][y] = button when you press them. Let’s make a function to
return new_board attach to the button, so that when it is pressed, the
button displays either X or O depending on which
In the app section, initialise a list called board_ player chose it.
squares and set it to call the new function First, create a variable in the variables section to
you just created. record whose turn it is. You can choose to start with
either player, but we will choose to start with X.
board_squares = clear_board()
turn = "X"
This variable will be assigned the value of the new_
board you created within the function, which should This now means that you need to display on the GUI
be a blank board with nine buttons. Make sure that whose turn it is (Figure 3) so the players don’t get
you create this variable after the code for creating confused. Add Text to your list of widgets to import:
the Box, otherwise you will be trying to add buttons
to a container that does not yet exist. from guizero import App, Box, PushButton,
Your code will now resemble tictactoe2.py. Save Text

lektor July & August 2022 63

 tictactoe4.py
> Language: Python 3

001. # Imports ---------------

002. from guizero import App, Box, PushButton, Text
003.
004. # Functions ------------- Move to the functions section and create a new
005. def clear_board():
006. new_board = [[None, None, None], [None, None, None], [None, function called choose_square.
None, None]]
007. for x in range(3):
def choose_square(x, y):
008. for y in range(3):
009. button = PushButton(board, text="", grid=[x, y],
width=3, command=choose_square, args=[x,y]) You will notice that this function takes two
010. new_board[x][y] = button
011. return new_board arguments – x and y. This is so that you know which
012. square on the board has been clicked.
013. def choose_square(x, y): Add the following code (indented) inside the
014. board_squares[x][y].text = turn
015. board_squares[x][y].disable() function to set the text inside the button that was
016. toggle_player() clicked to the symbol of the current player, and then
017.
disable the button so it cannot be clicked on again.
018. def toggle_player():
019. global turn
020. if turn == "X": board_squares[x][y].text = turn
021. turn = "O"
022. else: board_squares[x][y].disable()
023. turn = "X"
024. message.value = "It is your turn, " + turn Finally, connect this function to the button. Find
025.
026. # Variables ------------- this line of code inside your clear_board function:
027. turn = "X"
028.
button = PushButton(board, text="",
029. # App -------------------
030. app = App("Tic tac toe") grid=[x, y], width=3)
031.
032. board = Box(app, layout="grid")
033. board_squares = clear_board() Modify it so that it looks like the line below:
034. message = Text(app, text="It is your turn, " + turn)
035. button = PushButton(board, text="",
036. app.display()
grid=[x, y], width=3, command=choose_square,
args=[x,y])

You have added two things here. Firstly, you

are attaching a command, just as before. When
Then add a new Text widget in the app section to the button is pressed, the function with this name
display the turn. will be called. Secondly, you are also providing
arguments to this function, which are the co-
message = Text(app, text="It is your turn, ordinates x and y of the button which was pressed,
" + turn) so that you can find that button again in the list.
Figure 3 Let your
Your program should now look like
players know tictactoe3.py. Save and run it. You will now be
whose turn it is
able to click on a button and it will change to an
X. Unfortunately, in this version of the game it is
permanently X’s turn!

Alternating between players

Once one player has taken their turn, the turn
variable should toggle to be the other player. Here
is a function which will toggle from X to O and vice
versa.

def toggle_player():
global turn
if turn == "X":
Figure 3 turn = "O"

64 July & August 2022 www.elektormagazine.com

tictactoe5.py
> Language: Python 3

001. # Imports --------------- 041. ) and board_squares[2][2].text in ["X", "O"]:

002. from guizero import App, Box, PushButton, Text 042. winner = board_squares[2][0]
003. 043.
004. # Functions ------------- 044. # Horizontal lines
005. def clear_board(): 045. elif (
006. new_board = [[None, None, None], [None, None, 046. board_squares[0][0].text ==
None], [None, None, None]] board_squares[1][0].text == board_squares[2][0].text
007. for x in range(3): 047. ) and board_squares[2][0].text in ["X", "O"]:
008. for y in range(3): 048. winner = board_squares[0][0]
009. button = PushButton(board, text="", 049. elif (
grid=[x, y], width=3, command=choose_square, 050. board_squares[0][1].text ==
args=[x,y]) board_squares[1][1].text == board_squares[2][1].text
010. new_board[x][y] = button 051. ) and board_squares[2][1].text in ["X", "O"]:
011. return new_board 052. winner = board_squares[0][1]
012. 053. elif (
013. def choose_square(x, y): 054. board_squares[0][2].text ==
014. board_squares[x][y].text = turn board_squares[1][2].text == board_squares[2][2].text
015. board_squares[x][y].disable() 055. ) and board_squares[2][2].text in ["X", "O"]:
016. toggle_player() 056. winner = board_squares[0][2]
017. check_win() 057.
018. 058. # Diagonals
019. def toggle_player(): 059. elif (
020. global turn 060. board_squares[0][0].text ==
021. if turn == "X": board_squares[1][1].text == board_squares[2][2].text
022. turn = "O" 061. ) and board_squares[2][2].text in ["X", "O"]:
023. else: 062. winner = board_squares[0][0]
024. turn = "X" 063. elif (
025. message.value = "It is your turn, " + turn 064. board_squares[2][0].text ==
026. board_squares[1][1].text == board_squares[0][2].text
027. def check_win(): 065. ) and board_squares[0][2].text in ["X", "O"]:
028. winner = None 066. winner = board_squares[0][2]
029. 067.
030. # Vertical lines 068. if winner is not None:
031. if ( 069. message.value = winner.text + " wins!"
032. board_squares[0][0].text == 070.
board_squares[0][1].text == board_squares[0][2].text 071. # Variables -------------
033. ) and board_squares[0][2].text in ["X", "O"]: 072. turn = "X"
034. winner = board_squares[0][0] 073.
035. elif ( 074. # App -------------------
036. board_squares[1][0].text == 075. app = App("Tic tac toe")
board_squares[1][1].text == board_squares[1][2].text 076.
037. ) and board_squares[1][2].text in ["X", "O"]: 077. board = Box(app, layout="grid")
038. winner = board_squares[1][0] 078. board_squares = clear_board()
039. elif ( 079. message = Text(app, text="It is your turn, " + turn)
040. board_squares[2][0].text == 080.
board_squares[2][1].text == board_squares[2][2].text 081. app.display()

else:
turn = "X"
Add the code in your functions section. Notice the
first line in the function: global turn. You need to
specify this so that you are allowed to modify the
global version of the turn variable, i.e. the one you
already created. If you don’t specify this, Python will
create a local variable called turn and modify that
instead, but that change won’t be saved once the
function exits.
You also need to make sure that the Text widget
accurately reports the current player’s turn. After
the if/else statement in the toggle_player function,
update the message like this:
message.value = "It is your turn, " + turn
Go back to your choose_square function and call the
toggle_player function – with toggle_player() –
once you have set the text and disabled the button.
Your code should now resemble tictactoe4.py. Save
and test the program again and you should find
that you can click squares and they will alternately
be designated either X or O.
Do we have a winner?
Finally, you need to write a function which will check
whether there is a row, column, or diagonal of three Xs
or Os, and if so will report the winner of the game.
Although it seems very inelegant, by far the easiest
way to check if someone has won is to hard-code the
checks for each vertical, horizontal, and diagonal line
individually.

lektor July & August 2022 65


The following code is for one vertical line, one
horizontal line, and one diagonal – can you add the
rest?
def check_win():
winner = None
# Vertical lines
if (
board_squares[0][0].text == board_
squares[0][1].text == board_squares[0][2].text
) and board_squares[0][2].text in ["X", "O"]:
winner = board_squares[0][0]
# Horizontal lines
elif (
board_squares[0][0].text == board_
squares[1][0].text == board_squares[2][0].text
) and board_squares[2][0].text in ["X", "O"]:
winner = board_squares[0][0]
# Diagonals
elif (
board_squares[0][0].text == board_
squares[1][1].text == board_squares[2][2].text
) and board_squares[2][2].text in ["X", "O"]:
winner = board_squares[0][0]
Notice that the function begins by creating a
Boolean variable called winner. If by the time the
long if/elif statement has been executed, the value
of this variable is True, you know that someone has
won the game.
After adding the remaining winning line checks,
add some code at the end of the function to change
the display message if there has been a winner:
if winner is not None:
message.value = winner.text + " wins!"
You now need to make sure that this function
is called each time an X or O is placed, which
corresponds to any time a button is pressed. Add
a call to check_win at the end of the choose_square
function, just in case the square that was chosen
was the winning square.
Your program should now look like
tictactoe5.py. Run it and test the game. If you
wrote the tests in the check_win function correctly,
you should find that the game detects correctly
when a player has won.
Draw game
At the moment, the game will allow you to continue
playing even after it has been won, until all of the
squares are selected. It will also not tell you if the
66 July & August 2022 www.elektormagazine.com
Global variables
It is arguably a bad idea to use global variables because if you have many
functions in a large program, it can become extremely confusing as to which code
modifies the value of a variable and when. In a small program like this, it is not too
difficult to keep track.
Remember that it is possible to read and use the value of a global variable from
inside a function without declaring it global, but in order to modify its value you
will need to explicitly declare this. The functions in this program (and most GUI
programs in this tutorial series) are actually modifying the values of your widgets
as global variables. For example, when someone wins the game, you set the value
of the message to display who won:
message.value = winner.text + " wins!"
In this example, message is a global variable. So how can we modify its value
without declaring it as global? The answer is because we are using a property
of the message widget, the property called value. Essentially what this code is
saying is “Hey Python, you know that widget over there called message? Well,
could you modify its value property please?” Python will allow modification
through object properties in the global scope, but it won't allow you to directly
modify the value of a variable without declaring it global.
game was a draw. You could stop at this point, but if
you really want to put the icing on the cake, adding a
few more little touches could make your game more
polished.
First, let’s add some code to detect whether the
game is a draw. The game is a draw if all of the
squares contain either an X or an O, and no one has
won. In the functions section, create a new function
called moves_taken:
def moves_taken():
Create
You’re going to use this function to count the number
Graphical
of moves which have been made, so let’s start a
User
variable to keep count, initially beginning at 0.
Interfaces
with Python
def moves_taken():
moves = 0 For further
tutorials on how
Now, remember when we created the to make your
board_squares, we used a nested loop to create all of own GUIs with
guizero, take
the squares on the grid? We’re going to need another
a look at our
nested loop here to check each and every square and book, Create
determine whether it has been filled in with an X or Graphical User
O, or whether it is blank. Add this code for a nested Interfaces with
loop to the moves_taken function: Python. Its
156 pages are
packed with
for row in board_squares:
essential info
for col in row:
and a range
of exciting
Inside the loop, we need to check whether that projects.
particular square is filled in with an X or an O. If it is, magpi.cc/
add 1 to the moves variable to record that square has pythongui
been counted.
 Editor’s note: This article originally appeared in MagPi 9-10/2021 (German edition).
Elektor publishes the Dutch, French, and German editions of MagPi.

06-tictactoe.py
> Language: Python 3

001. # Imports ---------------

002. from guizero import App, Box, PushButton, Text board_squares[1][0].text == board_squares[2][0].text
003. 047. ) and board_squares[2][0].text in ["X", "O"]:
004. # Functions ------------- 048. winner = board_squares[0][0]
005. def clear_board(): 049. elif (
006. new_board = [[None, None, None], 050. board_squares[0][1].text ==
[None, None, None], [None, None, None]] board_squares[1][1].text == board_squares[2][1].text
007. for x in range(3): 051. ) and board_squares[2][1].text in ["X", "O"]:
008. for y in range(3): 052. winner = board_squares[0][1]
009. button = PushButton(board, text="", 053. elif (
grid=[x, y], width=3, command=choose_square, 054. board_squares[0][2].text ==
args=[x,y]) board_squares[1][2].text == board_squares[2][2].text
010. new_board[x][y] = button 055. ) and board_squares[2][2].text in ["X", "O"]:
011. return new_board 056. winner = board_squares[0][2]
012. 057.
013. def choose_square(x, y): 058. # Diagonals
014. board_squares[x][y].text = turn 059. elif (
015. board_squares[x][y].disable() 060. board_squares[0][0].text ==
016. toggle_player() board_squares[1][1].text == board_squares[2][2].text
017. check_win() 061. ) and board_squares[2][2].text in ["X", "O"]:
018. 062. winner = board_squares[0][0]
019. def toggle_player(): 063. elif (
020. global turn 064. board_squares[2][0].text ==
021. if turn == "X": board_squares[1][1].text == board_squares[0][2].text
022. turn = "O" 065. ) and board_squares[0][2].text in ["X", "O"]:
023. else: 067. winner = board_squares[0][2]
024. turn = "X" 068.
025. message.value = "It is your turn, " + turn 069. if winner is not None:
026. 070. message.value = winner.text + " wins!"
027. def check_win(): 071. elif moves_taken() == 9:
028. winner = None 072. message.value = "It's a draw"
029. 073.
030. # Vertical lines 074. def moves_taken():
031. if ( 075. moves = 0
032. board_squares[0][0].text == 076. for row in board_squares:
board_squares[0][1].text == board_squares[0][2].text 077. for col in row:
033. ) and board_squares[0][2].text in ["X", "O"]: 078. if col.text == "X" or col.text == "O":
034. winner = board_squares[0][0] 080. moves = moves + 1
035. elif ( 081. return moves
036. board_squares[1][0].text == 082.
board_squares[1][1].text == board_squares[1][2].text 083.
037. ) and board_squares[1][2].text in ["X", "O"]: 084. # Variables -------------
038. winner = board_squares[1][0] 085. turn = "X"
039. elif ( 086.
040. board_squares[2][0].text == 087. # App -------------------
board_squares[2][1].text == board_squares[2][2].text 088. app = App("Tic tac toe")
041. ) and board_squares[2][2].text in ["X", "O"]: 089.
042. winner = board_squares[2][0] 090. board = Box(app, layout="grid")
043. 091. board_squares = clear_board()
044. # Horizontal lines 092. message = Text(app, text="It is your turn, " + turn)
045. elif ( 093.
046. board_squares[0][0].text == 094. app.display()

if col.text == "X" or col.text == "O": if winner is not None:

moves = moves + 1 message.value = winner.text + " wins!"

Finally, once the loops have finished, add a return
statement to return the number of moves taken.
return moves
Now, call this function inside the check_win
function, to check for a draw. Add this code after the
code that checks for a winner:
# Add this code
elif moves_taken() == 9:
message.value = "It's a draw"
Your code should resemble 06-tictactoe.py. When
run, the game will now check whether nine moves
have been taken; if they have, it will change the
message to report that the game was a draw.

lektor July & August 2022 67

 components

Reed Relays Peculiar Parts, the series

By David Ashton (Australia)

While the semiconductor

industry has some great
innovations, sometimes
you can’t beat a real
switch! Here we delve
into the world of reed
relays, devices that were
extensively used in the
telephone exchanges of
yesteryear.

Reed relays are versatile components that
can be used for applications where the true
isolation offered by a physical switch is
demanded. They are based upon a reed
switch, a device consisting of two lengths
of magnetic alloy, commonly 50:50 nickel
iron. The contacts are held in a glass tube
(the “reed”) such that they almost touch,
and their ends are plated with a contact
material. When a magnetic field is
brought close to the reed, the two lengths
of magnetic material become magne-
tized, their ends attract, and, bingo, they
make contact. When the magnetic field
is removed, they open again (Figure 1).
 
Reed relays generate the magnetic field
from a coil and are typically faster to
Figure 1: At the core of a reed relay is a reed switch that uses a magnetic field to
operate it. (Source: Pickering Electronics)
switch than mechanical relays. Although
their switching is considered bounce free,
under certain conditions they can still
suffer from contact bounce, especially if
the coil is driven from a voltage higher
than specified.
 
The coil of a reed relay is usually imple-
mented with a fine wire wrapped
around the reed. A typical reed requires
50 Ampere-turns to operate. So, 50 turns
of wire cause it to operate if 1 A is passed
through the coil, while 5000 turns cause
operation at 10 mA. At such current levels,
they can be constructed to be reasonably
sensitive.
 
The coil and the reed can also be hermet-
ically sealed inside a case, making them
remarkably rugged. Some types even have
metal screening inside the case to ensure
that external magnetic fields do not
result in the unwanted operation of the
device. Manufacturers also have a couple
of tricks with which reed relays can be
built as normally-closed or changeover

68 July & August www.elektormagazine.com

devices [1]. Figure 2 shows a variety of
reeds and reed relays.
 
Reed switches remain popular for a range
of applications thanks to their opera-
tional simplicity. Door switches for alarm
systems attach the reed switch to the door
frame and the magnet to the door. When
the door is closed, the magnet and switch
come into close proximity, and the reed
is closed. As the door is opened, the reed
opens, and the resultant break in the
circuit triggers the alarm. I also had a car
alarm once that used two reeds affixed
to the windscreen. It used a magnet in a
keyring fob to enter a basic code to disarm
it. For non-contact applications such as
these, reed switches are very useful.
 
They have also featured in many an
Elektor project. One that springs to mind
from years ago monitored the operation
of the light bulbs in your car. The cabling
Figure 2: A selection of reed relays. Top left is a changeover type (two contacts on the left-hand end). Bottom right is a bare
reed. The rest show the wide variety of sizes and packages available.
for the chosen lamp, such as the brakes, applications, thanks to them being physi-
was modified so that the current passed cally much smaller and capable of sensing
through a fairly thick wire wrapped around magnetic fields electronically rather
a reed switch to form a home-made reed than mechanically. However, in terms of
relay. The switch itself controlled an simplicity and isolation, reed relays win
LED synchronous to the switching of the hands down, which is why they are still
chosen bulb. If one brake light bulb burned in wide use.
out, only half the current flowed, which  
was insufficient to operate the reed. Like all If this has piqued your curiosity, suppliers
the others from Elektor’s 60-year history, such as Pickering Electronics have some
this project can be found on the Elektor great tutorials [3] and other resources [1]
website [2]. where you can learn more.
  220179-01
These days reed relays face some compe-
tition from Hall‑effect sensors in some
WEB LINKS
[1] “Reed RelayMate,” Pickering Electronics, April 2011: https://bit.ly/3qhVRKS
[2] “Car Lights Failure Indicator,” Elektor 3/1978: https://bit.ly/3tisNoq
[3] “Reed Relay Basics,” Pickering Electronics: https://bit.ly/37AU5hf
lektor July & August 2022 69
 project
Simple Analog
ESR Meter
With Moving-Coil
Meter Precision
By George Lydecker (USA)
Looking for an analog ESR meter to check
and find bad capacitors? You can build your
own. Here is one approach.  electrolytic capacitors used in power supplies to have a low ESR.
 
I was always envious of my friend’s analog ESR meter that allowed
him to check and find bad capacitors while they were still in circuit.
The particular meter he had is no longer available, which meant
doing some research to understand how they worked and to come
up with my own version.
Figure 1: Block diagram of the ESR Meter.
70 July & August www.elektormagazine.com
First, What Is ESR?
ESR stands for the equivalent series resistance of a capacitor. ESR
is frequency-dependent, temperature-dependent, and changes
as components age. It’s typically important for ‘wet’ aluminum
The typical method used for measuring ESR is to supply the capac-
itor with a known AC current (Icap) at some frequency where the
capacitive reactance of the capacitor is very low so that the ESR
dominates. By measuring the resulting AC voltage developed across
the capacitor’s terminals (Vcap), the ESR can be determined with
Ohm’s law:
ESR = Vcap/Icap
Most of the designs I found worked along the same lines as the
block diagram shown in Figure 1. Going from left to right, there
is an oscillator that supplies AC voltage to be applied to the capac-
itor. Next, the AC signal is fed into an impedance converter and
detector. The detected signal is then rectified and buffered so that
it can drive the meter on the right of the diagram.
Since the ESR meter is to be battery operated, the power supply
circuit supplies split rails for the operational amplifiers that will
be used in the ESR meter. The oscillator in most of the examples I
looked at operated at 100 kHz to 150 kHz. The driver used to reduce
the impedance of the AC signal could be anything from a transistor
current boost, transformer, or paralleled logic gates. The detector
was usually back-to-back diodes. The detected AC signal is then
rectified, amplified and fed into a DC meter.
 A
DUT B
+4V5
IC1B
R2 R7 R15
3 4 14 8 8
1 600 1M IC1 = 74HC14 10k
IC1 IC2 IC3
C2 IC2, IC3 = TL082
IC1C 7 4 4 D3 1N4148
R1 R3 R16
5 6
1k 1 600
D1
470n R11
–4V5
10k
C4 100p
4k7 R13
IC1A IC1D 6
R4 C3 R10 10k R18
D4
1 2 9 8 2 2 7
1 1 600 2 x 1N4001 1k R12 1N4148 IC3B
1 6 1 5
C1 100n IC2A 1k R14 IC3A 50k R19
IC1E 3 7 3
IC2B 1k

10k
R5
11 10 D2 5
10n 1 600 R8 R9 R17
+

10k
10

10
IC1F
R6 M1
13 12 100µA
1 600

OSCILLATOR DRIVER DETECTOR AC AMPLIFIER ABSOLUTE VALUE CIRCUIT 210017-002

Figure 2: The schematic of the ESR Meter is composed of design elements from various DIY examples found on the Internet.

My Circuit circuit because any other silicon junctions will not be forward
In the circuit I decided to build (Figure 2), I used several design biased by the relatively low AC signal. The low-level AC signal is
elements from the DIY examples I found on the Internet. For the DC decoupled with C3 and amplified by two operational amplifi-
oscillator and impedance converter, I used a single 74HC14 that ers, with a voltage gain of 4.7 for the first amplifier and a gain of
provides six inverters with hysteresis. One of the inverters acts as a 10 for the second. This provides an overall gain of 47 to the input
relaxation oscillator and the remaining five inverters operate as the of the absolute value circuit.
impedance converter. This part of the circuit came from Lawrence  
P. Glaister VE7IT. His circuit and article can be found at [1]. The absolute value circuit was taken from the Burr-Brown
  Application Bulletin “Precision Absolute Value Circuits” [2]. The
The detector portion of the ESR meter is the same as the detection absolute value circuit had plenty of drive for the 100 microamp
circuit in the commercial ESR meter built by Creative Electronics. meter I used.
Sadly, these meters are no longer made.  
  This ESR meter operates of a single 9-volt battery. The plus,
Diodes D1 and D2 clip the top and bottom of the 100 kHz AC to minus and ground voltages needed for the operational amplifier
one silicon junction drop. This allows capacitors to be tested in are derived using the opamp voltage follower with current boost
shown in Figure 3.

R22
3k3

R20 T1
R26
10k

10k

TIP41 R24
Bt1
D5

1 +4V5
10

3 7
5 1N4001
9V 6
IC4 R25
2
D6
4
–4V5
10

LF356 1N4001

T2

R21 R23

TIP42
10k

3k3

Figure 4: The oscilloscope shows the output of the oscillator as well

as the output of the ESR meter while testing a 50 µF electrolytic
Figure 3: This circuit turns a 9 V battery into a symmetric capacitor.
double 4.5 V power supply.

lektor July & August 2022 71

 Figure 6: The circuit is simple enough to be built on prototyping
board.
Figure 5: This waveform dump from the oscilloscope shows the
difference between a capacitor with a high ESR in blue and a
capacitor with a low ESR in orange.

Test and Operation

To test out the circuit, I built it up on a conventional bread board.
BNC connectors allow for easy monitoring of the various waveforms.
Initially for testing I simply looked at the signals with an oscillo-
scope instead of the 100 µA meter (Figures 4 and 5).
 
The circuit is wired on two prototyping perfboards (Figure 6). The
smaller board on the left is the circuit used to derive the plus, minus
and ground references for the op-amps. The larger perfboard is the
ESR meter circuit and is held in place by the nuts on the meter.
  Figure 7: Even when it still was a breadboarded prototye the
Even before I was able to package the meter into a case I used it to circuit proved useful when checking the motor capacitor of an air
trouble shoot our air conditioner by locating a bad motor capac- conditioner.
itor (Figure 7).
 
The completed ESR meter is sufficiently portable so that it can
be used remotely and away from the work bench. Simple Post-It
Note calibration was accomplished with a handful of 2-Ω resis-
tors (Figure 8).
 
The ESR meter works like an ohm meter. Before measuring a capac-
itor, the leads are shorted together and the knob is adjusted for
a full-scale reading (zero ESR). Rotating the knob fully counter
clockwise turns the meter off with a switch.
210017-01

Questions or Comments?
Do you have technical questions or comments about this article? Figure 8: Calibrating the meter is easily accomplished with a few
Email the author at glydeck@aol.com or contact Elektor at 2-Ω resistors and a scrap of paper.
editor@elektor.com.

RELATED PRODUCTS
WEB LINKS
[1] Circuit by Lawrence P. Glaister (VE7IT): > Elektor 2 MHz LCR Meter Kit (SKU 19883)
http://ve7it.cowlug.org/esrmeter.html www.elektor.com/19883  
[2]Burr Brown application note
> Peak Atlas ESR70 Plus Capacitor Analyser (SKU 17668)
http://glydeck.com/project/content/absolute_value_ www.elektor.com/17668
circuit.pdf

72 July & August www.elektormagazine.com

They trust us,
do you?

We love electronics and projects, and we do

our utmost to fulfil the needs of our customers.
The Elektor Store: ‘Never Expensive, Always Surprising’

Check out more reviewsw.on tor.com/TP

our Trustpilot page: ww elek
Or make up your own mind by
e, ww w.elektor.com
visiting our Elektor Stor
 project

Sigfox CO2
Traffic Light
No Wi-Fi Network Needed!

makes the CO2 traffic light system ideal for

By Peter Groppe, Frank Schleking and Bernd vom Berg operation where access to a Wi-Fi network
(Germany) is unavailable.

We have already featured a number of CO2 The Hardware

concentration meters in Elektor. Almost all Building the hardware for the CO2 traffic light
of them feature a Wi-Fi interface so that you is quite easy; nowadays there are a wide range
of very powerful microcontroller boards avail-
can check out the readings from anywhere in able off the shelf from Arduino that can be
the world. This CO2 traffic light is different; it used in this CO2 traffic light design. One such
connects to the IoT using the Sigfox network. board which includes Sigfox communication
function is the Arduino MKR FOX1200 board,
This gives the sensor significantly wider range which we have previously introduced in the
and allows it to operate where there is no Elektor article series “Sigfox and the IoT” [5].
access to a Wi-Fi network.  
Alternatively, if you do not need IoT connectiv-
ity and just want to display the CO2 measure-
ments locally, you could use a standard
Arduino Uno which also plugs onto the
A variety of studies have shown that air quality traffic light display. In the age of the IoT, most motherboard design presented here. Figure 1
decreases in unventilated public spaces; high of these systems also have a Wi-Fi interface gives the circuit diagram showing how either
CO2 concentration correlates with a higher so that measurement data can be transferred IC3 (an Arduino Uno R3) or IC1 (Arduino MKR
viral load in the air we breathe. It’s important to a cloud platform and displayed on a web FOX1200) connect to the motherboard.
to maintain a good airflow to help reduce the page viewable from anywhere in the world.  
risk of passing on airborne viruses. CO2 sensor   The microcontroller board records the
modules are widely available and can be used This CO2 traffic light system described here measured values ​​(CO2 concentration, air
with a controller to continuously monitor the connects to the Internet using the Sigfox radio temperature and humidity) from the SCD30
air we breathe. Warnings and alarm messages network rather than Wi-Fi. Sigfox is particu- CO 2 sensor at regular (adjustable) inter-
can then be issued when the measurement larly well suited to this application because vals and shows these values ​​on a 1.3” OLED
concentration level exceeds a preset threshold. we only need to send small amounts of data, display. The traffic light indicator consists of a
  and the excellent radio coverage achieved by multicolor NeoPixel LED matrix display.
That is the task of a standard CO2 measure- Sigfox gives us maximum flexibility when siting  
ment traffic light system. It provides a visual the sensor. The Sigfox network generally only The CO2 sensor SCD30 sensor [1] uses an I2C
and sometimes acoustic alarm when more requires a handful of base stations to cover bus (SDA, SCL, +5V and GND) and connects
fresh air is required. The concentration of CO2 an entire city and coverage in many countries via the space-saving double-level screw termi-
is indicated via a red, yellow or green LED around the world is already very good. This nal strip X1. If required, 4.7-kΩ pull-up resistors

74 July & August www.elektormagazine.com

 +5V JP1 7805 D1 F1
IC2
R1 X6
1N4004 800mA

1k5
C4 C2 C1 C3
VCC
220µ 220µ
GND
D2 100n 100n

+5V

R3 R2

USB

4k7

4k7
IC3 SCL
SCL
SDA
IC1
SDA
ARDUINO UNO R3 RST +5V
AREF
GND AREF 5V
A0
RESERVED 13 DAC0/A0 VIN
A1
IOREF 12 A1 VCC
A2
RESET 11 A2 GND
A3
3.3V 10 A3 RES
+5V

5V 9 A4 14 < TX
POWER

GND 8 A5 13 > RX
GND A6 12 SCL
DIGITAL (PWM)

DIO7 DIO0
VIN 7 0 11 SDA
DIO6 DIO1
6 1 10 MISO
A0 DIO5 DIO2
A0 5 ~2 9 SCK
A1 DIO4 DIO3
ANALOG IN

A1 4 ~3 8 MOSI
A2 ICSP DIO3 DIO4
A2 3 7
RESET

~4
MISO

A3 DIO2 DIO5
SCK

A3 2 ~5 6
SDA DIO1
A4 TX 1
SCL DIO0 MKR FOX1200
A5 RX 0
ANT
GND
MOSI
5V

DIO6
DIO7
DIO2
DIO7
DIO1
DIO0

DIO3
DIO5
DIO6
DIO4

A2
A0
A1
A3
SDA
SCL

SDA
SCL

R4
+5V +5V +5V +5V +5V
470

RESET 1 3 2 4 1 4 2 5 3 6 1 4 2 5 3 6 1 4 2 5 3 6 1 3 2 4

X2 C5 X4 X3 X5 X1
S1

1000µ

OLED DISPLAY LED MATRIX DIO ANALOG SCD30 200650-003

Figure 1: The CO2 traffic light circuit diagram showing both the Arduino Uno and the MKR FOX1200 board wiring.

can be connected to SDA and to SCL. The connections X3, X4 and X5. These connec-
OLED display [2] also connects via the I2C bus tors can be used to provide future additional
at the X2 double-level terminal strip. functionality to the system, for example to
  switch on air conditioning units or actuate
In addition to a power supply connection we ventilation when measured values exceed
only need a single digital port pin to operate thresholds.
the NeoPixel LED matrix display [4]. Port pin  
DIO7 is used here and the display connects Figure 2 shows the assembled circuit board
via the double-level terminal strip X4. NeoPixel onto which either the Arduino MKR FOX1200
displays with different numbers of LEDs can module or the Arduino Uno board is plugged
be driven from this single port pin. (face down) into its corresponding rows of
  connectors. Only one of these boards
Additional sensors/actuators can be should be fitted to the motherboard at any Figure 2: The finished CO2 traffic light
con­nected to the unused digital/analogue time. Do not try to mount both boards, by motherboard PCB.

lektor July & August 2022 75


Figure 3a: The Arduino MKR FOX
1200 fitted to the motherboard.
using extending headers. The board layout,
the assembly plan, a parts list and Arduino
firmware used can be downloaded from
the Elektor project page [3]. The flat Sigfox
antenna can be fixed in position using double-
sided adhesive tape. Figure 3 shows both
versions assembled.
The Power Supply
One important consideration when using
NeoPixels is their power requirements. In the
worst case when all three of the individual
LEDs in one NeoPixel is set to maximum inten-
sity with a value of 255 to produce the bright-
est white light, each LED element will draw
20 mA so a single NeoPixel requires 60 mA.
A matrix made up of 16 NeoPixels requires
16 × 60 mA = 960 mA! That’s more than can
be supplied by the 5V SMD voltage regula-
tor (without a heatsink) fitted to the Arduino
board. Powering it from the Arduino board
would overload the regulator which may shut
down or in the worst case, release its inter-
nal smoke.
 
Under normal operating conditions, you proba-
bly will not be driving the NeoPixel so hard, but
reckon on an average current consumption of
25 to 30 mA per NeoPixel, depending on the
overall colour displayed. With a 4×4 LED matrix,
you still get 400 to 480 mA, which again will
be too much for an on-board Arduino voltage
regulator to cope with.
 
Therefore, the following should be observed
when using NeoPixel displays:
 
> Install a reservoir electrolytic capacitor
directly at the NeoPixel power supply
76 July & August www.elektormagazine.com
Figure 3b: The Arduino Uno fitted face
down to the motherboard.
Figure 3c: The CO2 sensor module (left) and
various displays.
connections to mitigate supply voltage
fluctuations as the LEDs are switched.
Use an electrolytic capacitor in the
range from 470 µF to 1,000 µF for smaller
displays. The value used in this applica-
tion (C5) is 1,000 µF.
> Install a low value resistor in series with
the display control line to reduce inter-
ference. The value used here (R4) is
470 Ω.
> An external regulated power source will
be required with a high-quality voltage
regulator (possibly even cooled) and
with appropriate reservoir capacitors
(here we use a conventional 7805 linear
regulator on the motherboard powered
from a 9V/1A mains adapter). DC/DC
switching regulators can also be used
here they offer better efficiency and are
not much more expensive.
> Under no circumstances should the
uncooled on-board SMD voltage stabi-
lizer of the Arduino boards be used,
since failures can quickly occur, partic-
ularly if you are using a cloned board,
due to overheating or switching current
peaks.
> The NeoPixel LED display matrix should
only be as large as absolutely necessary
and should not be operated at maximum
brightness. The brightness of the round
7-element NeoPixel display used here is
controlled by the software using a range
of values from 4 to 16 out of 255 possible
levels.
 
If the system is operated without a NeoPixel
display for test or other purposes, and if the
total power consumption is not too high, the
supply can easily be provided via the Arduino
on-board SMD voltage regulator. In this case,
jumper JP1 on the motherboard must be
removed so that the 7805 will not become
reverse powered.
Add Some Libraries to the
Arduino IDE
The CO2 traffic light software requires some
additional libraries to operate external
components like the CO2 SCD30 sensor, 1.3”
OLED display and NeoPixel LED matrix. Two
additional libraries will also be required for
Sigfox operation with the MKR FOX1200 board
and an optional one for the on-chip RTC of
the SAMD21 microcontroller.
 
The SCD30 CO2 Sensor Library
To install the SparkFun driver library for
the SCD30 sensor, go to Tools -> Manage
Libraries... in the Arduino IDE, then look for
scd30 in the top right corner. The two most
popular SCD30 libraries will be listed, one from
Adafruit and one from SparkFun. As shown
in Figure 4, we will use the SparkFun library.
 
After installation, the library is automatically
integrated into our program in the IDE via
Sketch -> Include Library -> ... SparkFun
SCD30 Arduino Library. You can get a detailed
description of the library by clicking on More
info in the bottom left of Figure 4.
 
The 1.3” OLED Display Library
The extremely powerful and comprehen-
sive U8g2 library built by Oli Kraus is used to
drive the 128*64 pixel 1.3” OLED display. This
complete software package actually consists
of four individual libraries:
 
> U8g2: Library for graphics-capable Figure 4: Select the SCD30 library from SparkFun.

display applications using large numbers

of graphics functions and display charac-
ter sets.
> U8x8: Library for simple, purely
text-based display applications with a
reduced range of character sets.
> MUIU8g2: Special functions for the reali-
zation of interactive, graphics-capable,
Monochrome User Interfaces (MUI).
> U8log: Functions for emulating a terminal,
similar to the serial monitor feature of the
Arduino IDE. Figure 5: Select the U8g2-library to drive the 1.3” OLED display.
 
The U8g2 driver library for the OLED display
is also installed in the Arduino IDE using Tools
-> Manage Libraries...; then look for u8g2 in
the top right corner (Figure 5).
 
After the installation, the three core libraries
are automatically integrated into our program
using Sketch\Include Library\… U8g2.
 
#include <MUIU8g2.h>         // delete
#include <U8g2lib.h>         // delete
#include <U8x8lib.h>
 
We only need the text-based U8x8lib, so the
#include statements referencing the other
two libraries can be deleted.
 
The advantage of the U8g library pack is,
among other things, the large number of
supported display types which includes: Figure 6: Uncomment the correct one from this large assortment of U8x8 supported display types.
 
> LCD, OLED graphics capable displays
> Various pixel resolutions
> Control via different buses: I2C/SPI in File -> Examples -> U8g2 -> U8x8. Be careful > Display controller: SH1106
hardware or software implementation, to only select examples for the U8x8 library! > Display pixel count: 128*64
parallel buses according to 8080 and   > Display manufacturer: NONAME (we use a
6800 specification In the long list in this HelloWorld sketch, we generic unbranded display)
> Various driver components need to search for the (OLED) display used. > Hardware interface method: I2C
  You enable it by removing the // comment > Name of the created object for reference
While it’s undoubtedly useful to have an delimiter. We just copy the corresponding in the program: u8x8
overabundance of supported displays, make (commented out) line 64 of the appropriate > Expression in brackets: the display reset
sure you choose the correct one, otherwise display constructor to line 119 [3] of our CO2 pin is not used
you will run into problems. In Figure 6 you traffic light sketch:  
can see just a section of the myriad of possi-   You can get a detailed description of the library
ble supported display types commented out. U8X8_SH1106_128X64_NONAME_HW_I2C by clicking on More info in the lower left corner
  u8x8(/* reset=*/ U8X8_PIN_NONE); of Figure 5.
It therefore makes sense to first look at one of    
the finished examples that were installed with which breaks down to: The NeoPixel-LED Library
the library in the Arduino IDE — for example   The NeoPixel LED concept developed by
the HelloWorld example can be found under > Driver Library: U8X8 Adafruit makes it possible to construct large,

lektor July & August 2022 77

coloured LED strips, fields, circles and other
shapes composed of individual RGB LED
elements. All LED elements are daisy-
chained and controlled via a single digital
I/O line. The WS2812 LED controller makes
it possible to control each of the three differ-
ent coloured individual LEDs (red, green, Figure 7: Install the Adafruit NeoPixel library.
blue) with programmable 8-bit intensity, so
that almost any colour combination can be
generated.
 
The advantage of creating software for
NeoPixel applications of any size is that there
is a very convenient Arduino library from
Adafruit to provide control. This library is also
included in the Arduino program package
and is installed using the library manager you Figure 8: Select the RTC library for the SAMD21 microcontroller.
are already familiar with (search for neopixel),
as shown in Figure 7. After installation, the
library is automatically integrated into our
program via Sketch -> Include Library ->
Adafruit NeoPixel. Here, too, you can find a
detailed description of the library under by
clicking on the More info link.
 
The SAMD Microcontroller RTC Library
The SAMD21 microcontroller on the MKR
FOX1200 board has an on-chip RTC that
can be used for system-wide time and date Figure 9: Select the Sigfox-library for the MKR FOX1200 microcontroller board.
applications. This requires a backup battery
for the board so that the RTC does not lose
track of time every time the supply voltage  
is switched off. We chose not to include the Using Sigfox, the measured values of ​​ the CO2 places for debugging purposes (for the serial
backup battery in the CO2 traffic light, but traffic light system are transferred wirelessly monitor in the Arduino IDE). Once you are
a corresponding small circuit can be easily to the Internet where they can be accessed satisfied everything works as it should, these
connected to the MKR FOX1200 board. and displayed anywhere in the world using code sections can be commented out or
  a PC, laptop, tablet or smartphone. deleted.
If you want to use this RTC, there is of course    
a convenient driver library for this, which Here again we can take advantage of all the You can freely assign thresholds for the
can be installed as usual via the library ready-made functions in the Sigfox library measured CO2 concentration levels and the
manager (search for rtc, Figure 8). As usual, available in the Arduino program package. associated traffic light colour. Our suggested
the detailed description of this library can The library is installed via the library manager thresholds are:
be found under More Info. The library is just as we did with the others (search for  
integrated using Sketch -> Include library sigfox, Figure 9). Now use Sketch -> Include > 0 to 1000 ppm:       saturated green
-> RTCZero. Library -> Arduino Sigfox for MKRFox1200 to > 1001 to 2000 ppm:    yellow
install it. Click on More Info to discover more > 2001 to 5000 ppm:    orange
The MKR FOX1200 Sigfox about the available Sigfox functions. > 5001 to 12000 ppm:   saturated red.
Library  
Now we come to the connection of the CO2 Significant snippets of the annotated Operation Using the Sigfox
traffic light to the Internet of Things using Arduino program (for the MKR FOX1200 Network
the Arduino MKR FOX1200 microcontroller board) are given here in Listing 1. The In [5] we have shown in detail how the
board, which talks to the Sigfox network. For complete program in German and English parameterization and operation of the MKR
more information about Sigfox, take a look [3] is documented in detail and still contains FOX1200 module is implemented in the
at the four-part introduction in Elektor [5]. some serial output routines inserted in many Sigfox network. Here we can limit ourselves

78 July & August www.elektormagazine.com

Listing 1. Excerpts from the Arduino sketch.
void loop()
{
  unsigned char i;
  char text[15];
 
  // loop
  while(1)
  {
 
    /* The SCD30 sensor measures automatically every 2 seconds.
       If data is available, then read it out and store it in variables*/
 
    if (AirSensor.dataAvailable())      // Check if new data is available
    {
      // If yes, then read the 3 measured values
      co2 = AirSensor.getCO2();
      temperatur = AirSensor.getTemperature();
      luftfeuchte = AirSensor.getHumidity();
 
      // Increase counter
      mess_zae = mess_zae + 1;
 
      // Print values on the serial monitor
      Serial.print("Measurement-Nr: "); Serial.print(mess_zae); Serial.print("   //   ");
 
      // Print time
      print2digits(rtc.getHours());
      Serial.print(":");
      print2digits(rtc.getMinutes());
      Serial.print(":");
      print2digits(rtc.getSeconds());
      Serial.println();
 
      // print measurements
      Serial.println("CO2:      " + String(co2) + " ppm");
      Serial.println("Temp:     " + String(temperatur-temp_cor) + " °C");
      Serial.println("Hum:  " + String(luftfeuchte) + " %rh");
      Serial.println();
 
      // Processing the CO2 value for the 'LED monitor': NeoPixel display
      switch(co2)
      {
        case 0 ... 1000:  pixels.fill(gruen_satt,0,7);
                          pixels.show();
                          break;
   
        case 1001 ... 2000: pixels.fill(gelb_1,0,7);
                            pixels.show();
                            break;
   
        case 2001 ... 5000: pixels.fill(orange_1,0,7);
                            pixels.show();
                            break;
       
        case 5001 ... 12000: pixels.fill(rot_satt,0,7);
                             pixels.show();
                             break;         
      }
 
      // Output of the measured values on the OLED display 
      // Output: CO2 value
      u8x8.setCursor(6,2);
      sprintf(text,"%5d",co2);   // sprintf only works with integers
      u8x8.print(text);
 
      // delete Phantoms
      u8x8.setCursor(11,2);
      u8x8.print(" ppm");
 

lektor July & August 2022 79


Figure 10: Basic structure of the Sigfox network (Source: Sigfox).
Figure 11: The Sigfox portal (Sigfox-Backend).
Figure 12: The device message window.
to a brief description of the principle and the
necessary steps. We begin with an overview
of the Sigfox network (Figure 10).
 
The Sigfox-Objects, like our CO2 measuring
device or other sensor board, send out their
telegrams as a broadcast according to the
“fire-and-forget” principle via the license-
free 868 MHz ISM band. In addition to an
identification (sender ID), each telegram
contains a user data field with a maximum
size of 12 bytes, called the payload. A user
80 July & August www.elektormagazine.com
can transmit a maximum of 12 bytes of
measurement, status or other data with
each transmission. Since the Sigfox network
works in the license-free ISM band, to remain
within the legal regulations, each object is
only allowed a maximum of 140 transmissions
per day. Our MKR FOX1200 board is therefore
allowed to send one telegram every eleven
minutes on average.
 
Depending on the coverage, these transmis-
sions are then received by all Sigfox base
stations (Sigfox Stations) within range. These
base stations transmit all received data via
the Internet or GSM connection to the Sigfox
Cloud, from which the user can retrieve their
data and further evaluate it in their data
processing application (Customer IT).
The user account configuration interface
is called Sigfox Backend. This is where the
Sigfox Objects (devices) are registered,
groups are assigned and data forwarding
to Customer IT (via Callbacks) are set up.
First, the Sigfox device unique identifier ID
and PAC must be read out of the Sigfox
device using a small Arduino sketch [3].
These two parameters are required to regis-
ter the Sigfox device in the Sigfox Cloud
at [6].
During operation, worldwide access is
available to the telegrams. You log in to the
Sigfox backend [7] with your e-mail address
and the selected password which brings
you to the Sigfox portal start page (Sigfox
backend) in Figure 11. Here you click on the
Device tab so that a list of the active Sigfox
devices appears. If you then click on the Id
field of a device, you get to the information
page for that device. Here you now click on
MESSAGES on the left side and end up in
the window in which all telegrams are listed
that the Sigfox backend has received from
this device (Figure 12).
Creating Callbacks in the
Sigfox Backend
This raw data is not particularly meaningful.
As soon as the MKR FOX1200 sends data to
the Sigfox backend, we want it to be automat-
ically forwarded to the thinger.io dashboard
program. The Sigfox backend offers so-called
Callbacks for this purpose.
 
Such a callback is nothing more than an
automatic forwarding to the desired recipient,
which occurs immediately when the Sigfox
backend receives data from a Sigfox device —
for example, from our MKR FOX1200 module.
The creation and configuration of a Callback
was dealt with in detail in the third part of
the article series [5], so here we will restrict
our description to the essentials.
      // Output of the temperature value = float number
               oled_float(6,4,temperatur-temp_cor,1);     
 
      // Output of the humidity value = float number
      oled_float(6,6,luftfeuchte,1);     
   
      // Waiting times for measured value acquisition and transmission of the Sigfox telegram
      delay(w_zeit * 60000);            // (w_zeit * 1 Minute) wait between the measurements
      min_zae = min_zae + w_zeit;       // Minutes count until the next Sigfox transmission
      if (min_zae == SF_zyk)            // Now: send Sigfox telegram
      {
        SF_send_data();
        min_zae = 0;                    // Reset counter
} } } }     
 
/*** Send data via Sigfox ***/
 
void SF_send_data(void)
{
  Serial.print("Sigfox - Start ... \n");
 
  // Write measured values to the data structure variable
  SF_Ampel.CO2    = co2; 
  SF_Ampel.Temp   = temperatur;  
  SF_Ampel.Feucht = luftfeuchte;   
 
 
  // If required: Debug outputs
/*  Serial.println();
  Serial.print("CO2:     "); Serial.println(SF_Ampel.CO2);
  Serial.print("Temp:    "); Serial.println(SF_Ampel.Temp);
  Serial.print("Feucht:  "); Serial.println(SF_Ampel.Feucht);
  Serial.println();
*/
 
  // Activate Sigfox modem and query errors
  if (!SigFox.begin())      // Initializing the modem
  {
    Serial.println("Sigfox modem not found! - Continue with RESET!");
    while (1);  // loop
  }
  else
  {
    Serial.println("Sigfox modem initialization OK!");
  }
 
  // Enable debug LED and disable sleep modes
  SigFox.debug();
 
  // Clears all pending interrupts
  SigFox.status();
 
  // Send payload via Sigfox
  SigFox.beginPacket();               // Preparing to send a packet
 
  // Send structure variable to the Sigfox backend
  SigFox.write((char*)&SF_Ampel, sizeof(SF_Ampel));
 
  // Error checking: If endPacket() returns 'true': error
  SF_error = SigFox.endPacket()
  if(SF_error > 0)
  {
    Serial.println("Sigfox-Error !!");
  }
 
  // Sigfox end
  SigFox.end();
 
  Serial.println("Sigfox - End .... !\n"); 
}

lektor July & August 2022 81


Figure 13: Choose the desired Device Type.
Figure 14: The Callback window.
Figure 15: The large selection of different Callback types.
 
To create a callback, click on the Device Type
tab on the main page of the Sigfox Backend
and in the list that appears (with only one
entry) on the name of the Device Type, in our
82 July & August www.elektormagazine.com
case Arduino_DevKit_1 (Figure 13). In the
overview page of the Device Type, click on
Callbacks in the list on the left (Figure 14) and
in this window to the top right click on the small
button New. A list appears with all possible
types of callbacks for the most common types
of Dashboard-/Cloud-programs (Figure 15).
Here you select Custom Callback.
In the window that now appears (Figure 16),
the callback for sending the data from the
Sigfox-Backend to thinger.io is configured.
With this callback, the Sigfox backend always
sends the variables Device-ID, Telegr-Nr,
CO2, Temp and Feucht (Humidity) with their
values to
​​ the Dashboard Program immediately
after receiving the telegram from the CO2 traffic
light.
Build a Dashboard with
Thinger.io
Thinger.io is an open-source IoT visualization
platform with which you can quickly and easily
create clear and descriptive representation
of data. For smaller projects, it can be used
free of charge. The values ​​are visualized on a
dashboard, which can also be made publicly
available to web browsers.
Only a few simple steps are necessary
before you are able to develop a customized
Dashboard:
> Create a free user Account at thinger.io.
> Create a data container (Data Bucket) at
thinger.io to hold the measured values ​​
sent from the Sigfox cloud.
> Define the access point (Token) for the
Sigfox-cloud and create a receipt authenti-
cation (Access Token) at thinger.io, so that
the Sigfox-cloud receives permission to
transmit data to thinger.io.
> Configure a Callback in the Sigfox-cloud
to transfer the data from the Sigfox-cloud
to the Data Bucket in thinger.io via the
Internet.
> Design a neat dashboard on the thinger.io
site to visualize the data.
Here, too, we refer to [5], which gives a more
detailed description of the individual steps. After
creating a free account at thinger.io [8], you
log in and get to the main screen, which is the
starting point for all further actions (Figure 17).
 
To create a dashboard, click on the Dashboards
item on the left and then click on the Add
Dashboard button in the window that appears.
A large number of freely configurable widgets
(the user interface components) are available
for the individual design of a dashboard, for
example diagrams showing measurement
values over time, donut diagrams, bar charts,
analogue displays, Google Maps to show
location, images, text displays, LEDs and a
clock.
 
Figure  18 shows an example of a dashboard Figure 16: Callback configuration.
designed with thinger.io for our CO2 traffic light.
The measured values are​​ shown in the form of
a time chart. In addition to this data, the time,
telegram number and other information can be
displayed. Worldwide access to the dashboard
is possible with a few mouse clicks, so that
anyone who receives the link can then read the
dashboard and check on the current values.
200650-01

Figure 17: Account statistics on the console dashboard.

RELATED
PRODUCTS

> Arduino MKR FOX 1200 (SKU 19096)

www.elektor.com/19096  

> Arduino Uno Rev3 (SKU 15877)

www.elektor.com/15877  

> Seeed Studio Grove SCD30 CO2,

Temperature & Humidity Sensor for
Arduino (SKU 20012)
www.elektor.com/20012    

Figure 18: The CO2 traffic light dashboard.

WEB LINKS
[1] CO2-Sensor SCD30: https://bit.ly/34XbL5o
[2] 1.3“ OLED display: https://bit.ly/3fEP7AX
[3] Elektor Project Page: www.elektormagazine.com/200650-01
[4] NeoPixel LED displays Information: https://bit.ly/3qf1Y2k
[5] Frank Schleking, Bernd vom Berg, “Sigfox and the IoT (1),” ElektorMag 11-12/2019 (Parts 2 to 4 follow in the next 3 editions):
www.elektormagazine.com/magazine/elektor-114/56832
[6] Provide your dev kit details: https://buy.sigfox.com/activate/devkit/DE
[7] Sigfox backend login page: https://backend.sigfox.com/auth/login
[8] Build the dashboard using thinger.io: www.thinger.io

lektor July & August 2022 83

 ethics
Women
in Tech
“It's All About Merit
Until Merit Has Tits”
By Priscilla Haring-Kuipers (The Netherlands)
If you are wondering why Women in Tech
would need yet another article, it is because
you — my dear reader — are most likely male.
Bias Galore
Societal expectations that are based on gendered roles start early.
Clothes for toddlers come mainly in pink for a girl and astronaut
for a boy. Books for girls might focus on long-haired princesses
waiting in a tower to be rescued, while books for boys tend to have
main characters like chocolate technicians flying off in elevators.
Thousands of ‘little things’ are scaffolded into systemic barriers that
diminish the likelihood of any random human being who happened
to be born a woman from becoming an (electrical) engineer.
 
For years now educational policies have tried to increase the low
numbers of girls choosing Science, Technology, Engineering, Math
(STEM) subjects in high-school. Such girls-in-STEM interventions
are trying to fight back at biases that have established themselves
firmly in our culture. My high school math teacher plainly told
the class that he would not be spending his time answering my
questions as “girls do not understand.” I dropped math. Twenty
years later I run statistical data analysis to establish effect research,
but the road was not a straight one.
84 July & August 2022 www.elektormagazine.com
Photo by Kelli McClintock on Unsplash
The gender equality gap is slowly getting smaller, but both in the
US and the UK the share of female engineers is only around 13%
[1]. Should you — against the odds — find yourself working as a
female engineer, you are most likely underpaid, undervalued and
underutilized compared to your male colleagues. Merit be damned.
Your work environment might range from unfemale to unsafe, and
yet you will likely be told to “lean in” or to “think like a Queen” and
be above your surroundings. Which is asking an already disadvan-
taged person to overcome real-world problems that they did not
create nor can control by having a better attitude. It is not surpris-
ing that a lot of women leave the field of engineering.
Personal Quota
It is why we need things like quotas for hiring capable female
engineers. We did not all have a supported start into STEM on equal
terms, and we are not judged in engineering on merit alone. Yes,
there are biological differences between men and women, but they
do not translate into males having math heads or a female procliv-
ity for baking. Neurologically there is no reason for the low number
 Female Factory
There are a lot of women involved in building electronics but not as the engineers.
They are the workforce on the factory floor. In the assembly houses we work with in
Shenzhen and the Netherlands, the hands on our boards are mostly female. About 70%
of the people that are carrying boards from machine to machine, doing through-hole
soldering, providing optical inspections and clamping test beds are women. They are
mostly young women and likely migratory workers in China and mostly middle-aged
women in the Netherlands. In a very practical sense women are making our electronics.

Photo by Austrian National Library on Unsplash

of women in tech. The biases that cause this situation are learned.
Which should mean that we can unlearn them. The same is true for
any other underrepresented group of human beings in engineering.
 
A very powerful part of forming your worldview is ‘normalising’. It
means that you think that whatever you are exposed to is normal.
This is why representation and role-models matter so much. So
try giving yourself a women’s quota wherever you connect to your
particular field of engineering. Make an effort to get maybe a third
of your input from female engineers. (You get only half points for
this article as I am a ‘woman in tech,’ but I am not an engineer.)
Try to fill your social media account with more women. And when
you find interesting female voices; amplify them. To start you off
on your journey of normalising women in tech, I would like to
introduce you to three of my favourite femmes:
> Jeri Ellsworth: All-round inventor. Amazing programmer
and electrical engineer. Ran the R&D team at Valve, and has
recently been busy building Tilt Five. This is an AR board
gaming system where players wear glasses with tiny projec-
tors that make holograms appear on a reflective gaming
surface. After four years, the system is currently in production
and is about to explode onto the gaming market.
www.jeriellsworth.com
 
> Naomi Wu (Sexy Cyborg): A tech reviewer and DIY maker of
all sorts in Shenzhen. Giving you a candid peek into the birth-
place of most electronics. Worked on the Creality 3D conveyer
belt printing mill and builder of the blinkini. Once made a
company uphold their code licencing by simply walking into
their offices. Called out tech facemasks for not being safe.
Advocates and opinionates loudly so that those in the back
can hear. The tag line, “It’s all about merit until merit has tits,”
is hers.
www.youtube.com/c/SexyCyborg
> Flavia Ferreira (Young Padawan): In our house there is a
group of women known as the ‘soldering ladies’ that roam the
world of audio hardware doing fantastic assembly work and
repairs. Flavia is one of them. If your circuits are built to make
sound, then these are the hands you want to work on them.
She also posts soldering tips, Let’s Talk Nerd interviews and
makes music. The force is definitely strong in this one.
www.instagram.com/imyoungpadawan
 
Please share with us which Women in Tech you would recommend
to follow.
220200-01
World Ethical Electronics
Forum 2022
In November of 2021, Elektor
launched the World Ethical Electron-
™ ics Forum (WEEF) in Munich,
Germany. The event inspired global
innovators in electronics with an
open discussion about ethics and sustainable development
goals (SDGs). In addition to Elektor engineers and editors,
the list of speakers and panelists included Dr. Stefan Heine-
mann (Professor of Business Ethics at the FOM University of
Applied Sciences), Dr. Paula Palade (PhD, Jaguar Land Rover),
Margot Cooijmans (Director, Philips Foundation), and several
other thought leaders, including Priscilla Haring-Kuipers. Visit
the WEEF webpage (www.elektormagazine.com/weef) to stay
informed about Elektor’s plans for WEEF 2022, which will take
place in November of this year.

WEB LINK
[1] Wikipedia, “Women in Engineering”: https://en.wikipedia.org/wiki/Women_in_engineering

lektor July & August 2022 85

 review

Low-Budget
Tablet Oscilloscope ADS1013D
Good Value for Money?

By Luc Lemmens (Elektor)

The time when a 100-MHz

scope would cost a fortune is
long gone. But for only €150
... too cheap to be true? Most
sellers are advertising
the ADS1013D as “Tablet
oscilloscope, 2-channels,
100 MHz bandwidth and
1Gigasamples/second.” Let’s
see what you get when you
buy one.

If you are used to working with more expen-
sive oscilloscopes from renowned brands, it
is very difficult to look at a device from the
lower market segment without prejudice.
The device I’m going to discuss here looks
good, but — with all due respect — it ‘feels’
cheap. You know in advance that in terms of
specifications, it can’t compete with devices
that cost a multiple of the ADS1013D. If you
want to do very accurate measurements,
you’d better save up and buy something
really good.
But as with so much equipment and tools,
it doesn’t all have to be top of the bill to still
be useful. Let’s be honest: how often do we
really need a high-precision oscilloscope?
It strongly depends on the kind of work
you do, of course. During testing proto-
types and repairing electronic devices, for
example, accuracy and precision aren’t
actually that important, a quick glance if
the signal-under-test has about the correct
level or waveform will do. If that is what
you need, this affordable unit
may be good enough for the
measurements you want to
take, so why spend more
money than necessary?
 
Figure 1: The box and its contents.

86 July & August 2022 www.elektormagazine.com

 Specifications > Coupling mode: AC/DC
> Analog bandwidth: 100 MHz > Trigger mode: Single, Normal, Auto
> Number of channels: 2 > Trigger edge: rising/falling edge
> Maximum real time sampling rate: 1 GSa/s > External trigger voltage: 0 to 40 V
> Vertical sensitivity: 50 mV/div to 500 V/div > Display: 7” TFT - 800*480
> Horizontal time base range: 50 S/div to 10 nS/div > Operating: capacitive touch screen + gesture
> Maximum test voltage: 40 V (1X probe), 400 V (10X probe) > Extension ports: USB picture export
> Storage depth: 240 Kbit > Power supply: 1 * Lithium battery, 6000 mAh
> Input resistance: 1 M > Item size: 184 * 124 * 50 mm/7.2 * 4.9 * 2.0 in
> ADC precision: 8 bits > Package weight: 700 g/1.5 lb

The original plan for this review was to
test a Daniu ADS1013D, but when I was
looking to buy one that I could get at an
acceptable price, at short notice, I came
across the FNIRSI brand. They sell exactly
the same oscilloscope with even the same
type number, and apparently it is also
being sold as a KKmoon and Yeapook
ADS1013D. I wouldn’t be surprised if the
product is available from other brands as
well. There are many online shops where
you can buy the ADS1013D, and with prices
ranging from a lower hundred to more
than €300, it certainly pays to search well
before you actually buy. Most sellers are
advertising the ADS1013D as “Tablet oscil-
loscope, 2-channels, 100 MHz bandwidth
and 1 Giga-samples/second.” Let’s see what
you get when you buy one.
What’s in the Box?
My FNIRSI ADS1013D is packaged in a
sturdy cardboard box with protective
foam inside, and it contains the oscil-
loscope, a 2 A USB-charger with a 1.5 m
long USB-C cable, two 100 MHz probes
and a small manual (Figure 1). To start
with the latter (you know: that booklet
that you never wanted to read anyway), I
can be very short: if you look inside, you
immediately hope that you’ll never need
to read it, with its small print, vague illus-
trations and very compact layout. I would
recommend to use the online documen-
tation instead [1].
The charger is a standard 5 V USB type, so
there’s no fast type C USB-charging on this
device. With a fully charged battery, the
oscilloscope works for about four hours,
not bad at all!
The box also contains two P6100 1:1/1:10
switchable 100 MHz probes — no high-qua-
lity stuff, but they will do the job. The hook
clips could have been a bit smaller, though.
 
And then, of course, the most important
part: the oscilloscope.
The ADS1013D Itself
As I said earlier, you shouldn’t buy an
oscilloscope in this price range if you need
highly accurate measurement equipment,
and I’m even not going to test or verify
the specifications of this one (see textbox).
Other reviews on the Internet (like [2])
show that these specs are — mildly put —
very optimistic, for example, by pointing
out that the 8-bit A/D-converters at the
input of each channel (AD9288) are by no
means fast enough for real time 1 GSa/s
sampling. Even more important: the
100 MHz bandwidth that is claimed for
both inputs is also unrealistic; in terms
of true bandwidth, 20 MHz seems to be
a more accurate value [2]. A maximum
sensitivity of 50 mV/div for the inputs is
a bit disappointing, and a maximum time
Figure 2: Power switch, connectors and LEDs.
base setting of 50 s/div will hardly ever be
of practical use.
 But instead of complaining what this oscil-
loscope does not have, let’s have a look at
what it does have to offer. First of all, a
7” touchscreen, a bright 800 x 480 pixel
colour LCD with a good viewing angle,
nothing to complain about there. With this
touchscreen, no knobs, buttons or switches
are needed to operate this instrument,
except for the good old-fashioned ON/OFF
switch. The good thing about this switch is
that it keeps the scope powered, without
any auto power off timer that switches it off
during measurements; but of course, if you
forget to switch it off without the charger
connected, the battery will probably be
drained the next time you want to use it.
 
The power switch is recessed in the top side
of the oscilloscope (Figure 2), just like the
two BNC connectors for the probes, a 1 kHz
square wave output connection for probe
calibration, a red power indicator LED
plus a green LED to signal that the battery
is fully charged, and the USB-C connector

lektor July & August 2022 87


to connect the charger or data connection
cable to a computer. On the back side of the
enclosure is a foldable tilt stand.
Working With the ADS1013D
A very good thing about this ADS1013D is
that it is very simple to control. At the top
and the right side of the screen are (menu)
buttons you can tap to open menus, trigger
functions or toggle settings. For example,
tapping the yellow CH1 button opens
the menu for the settings of channel 1
Figure 4: Attenuators’ adjustments.
88 July & August 2022 www.elektormagazine.com
Figure 3: Input channel 1
settings menu.
(Figure  3), and CTRL in the top right
corner enables control buttons to adjust
the channels 1 and 2 input attenuators
(Figure 4). There are no nested menus,
which makes it easy to quickly see how to
operate this oscilloscope, even if you only
use it occasionally. Not completely true:
the System settings in the main menu (as
I will call the menu in the top left corner)
has a sub-menu, but this doesn’t contain
options and functions for normal, common
oscilloscope operation. Silly enough, I had
Figure 5: ADS1013D connected via USB.
trouble to find how to manually set the
time base, which appeared to be a matter
of tapping either the left (increase) or
the right (decrease) side of the waveform
display area. Not the most obvious way,
in my opinion, but it works well once you
know how to do it.
 
Most of the controls are self-explanatory,
with familiar names and operation as on
other oscilloscopes. However, there are
two buttons in the bottom right corner of
the LCD that were not clear at first: Save Pic
and Save Wave, respectively. The first button
saves a snapshot (picture) of the complete
screen; the second saves the waveform.
Other than a static picture, a waveform
can be treated as a normal input signal,
i.e. the time base can still be adjusted, the
attenuator settings altered, measurements
can be made, etc. after it has been saved.
Both Pictures and Waveforms can be retrie-
ved on the display by tapping the corres-
ponding items in the main menu, in the
top left corner.
There are a few items controlled by swiping
the screen, like the vertical position of the
waveforms, trigger position and measure-
ment cursors. However, it is sometimes
uncertain what exactly will move when you
swipe the display. At the top of the screen,
there is a toggle button marked move slow/
move fast that controls the swiping sensi-
tivity, the latter setting allows for more
precise movement of waveforms and
cursors. It doesn’t work to use two fingers
on the touchscreen to zoom in and out, like
you would do on other touch screen devices
like your tablet or smartphone.
The ADS1013D can perform Fast Fourier
Transforms on both channels, but — like
with most other affordable digital oscillos-
copes — the resulting frequency spectra
are not very useful. It can display Lissajous
figures when the X-Y display mode is set;
and that’s about it. There are no mathe-
matical functions that can be performed

on the input waveforms, not even simple
operations like sum and/or difference of
the two input signals.
Connecting to a Computer
Apart from charging the internal battery,
the USB-C cable can also be used to hook
the device up to a computer. When USB
Connection is selected in the main menu,
the picture shown in Figure 5 appears on
the display and the ADS1013D is presented
as a USB drive on your computer. You can
then access the images stored on the oscil-
loscope from your computer. Unlike other
digital oscilloscopes, this one cannot be
controlled via USB, the interface is for trans-
ferring pictures only.
To Buy, Or Not to Buy?
That completely depends on what you want
and what you expect from a (simple) oscil-
loscope. If you need it for accurate measure-
ments, you shouldn’t buy this device from
FNIRSI, or any other oscilloscope in this
price range. But in many cases, when only
standard functional tests are performed
at lower frequencies and exact, accurate
measurements are of minor importance,
you don’t need to buy expensive equip-
ment and get good value for money with
this ADS1013D!
  200640-01
Questions or Comments?
Do you have questions or comments
about his article? Email the author at
luc.lemmens@elektor.com.
WEB LINKS
[1] FNIRSI online documentation: www.fnirsi.cn/productinfo/556152.html
[2] ADS1013D tablet oscilloscope tested: https://bit.ly/3Doz6Kv
Inside the ADS1013D
Since it only takes five Phillips
screws to open the case of the
ADS1013D, I could not resist the
temptation to have a look inside
the oscilloscope. Apparently,
there was an earlier model of this
oscilloscope that didn’t have any
shielding around the input stages,
and this has been improved for the
version I have. As far as I can tell
from photos I saw of the older one,
most of the PCB layout looks unchanged.
I can’t tell if the ADCs at the inputs are still the same, the type numbers are removed
from the ICs.
What’s in a Name?
I already mentioned that this is called a tablet oscilloscope, and I guess we all know what
a ‘tablet’ is, or we all have some idea what it should be or look like. If that would just be a
‘rectangular device with touch screen operation’, then this is a true tablet scope. On the
cardboard packaging it is called a Table Oscilloscope, and, in my opinion this is a more
appropriate name, but I assume that it is either a typo or a bad translation. To me, a ‘tablet’
sounds like some kind of portable device that can even be used ‘in the field’ by service
technicians or in the workshop. The ADS1013D is relatively small, weighs only about 700
grams and has a rechargeable internal battery; in all these respects it is a portable device,
but I would not recommend carrying it around too much. Even though the plastic enclo-
sure looks and feels sturdy, I don’t think the display or device itself will survive if you drop
it and there’s no case supplied to protect it during transport. That’s why I would keep it on
the table, and wouldn’t treat or use it like a tablet.
RELATED PRODUCTS
> OWON SDS1102 2-ch Digital Oscilloscope (100 MHz) (SKU 18782)
www.elektor.com/18782  
> Micsig STO1152C Plus 2-ch Oscilloscope (150 MHz) (SKU 19372)
www.elektor.com/19372
lektor July & August 2022 89
 project
b•
or la Elekt
kt

Ele

or
lab
ORIGIN
AL

Ele k

ab
rl rl

to
Smart Plug Teardown
• Ele k to
ab

Which Ones Are Hacker-Friendly?

By Mathias Claußen (Elektor)

Smart plugs or Wi-Fi-enabled sockets can be had for less than €15 these days. They
allow lamps and other consumer appliances to be controlled remotely (some units
can even record the appliance’s power consumption). For electronics engineers,
it’s not always that simple. Questions such as “I wonder what hardware they are
using there?” and thoughts like “I might be able to swap out the firmware and use
a different cloud platform to the one specified by the manufacturer” are always in
the back of our mind. To check out the internals of some of these smart sockets
and find out which ones are more hacker-friendly, I turned to my toolbox and took
out screwdrivers, a hacksaw, a camera, and a magnifying glass.
 

 
 

Attention mains voltage! The electronics

discussed here operate at peak DC
mains voltage! Do not attempt to
operate any electrical appliance outside
its protective enclosure as it could be
lethal. Any improper use will lead to
life-threatening electric shocks.
 
There are plenty of smart mains sockets
controlled by radio or Wi-Fi available these
days. With regard to their functions and the
network connection (usually Wi-Fi), the various
models often only differ in their software and
its associated cloud service. Price-wise they
are also pretty evenly matched. What’s inter-
esting is how the various manufacturers have
actually designed their products.
You only need to Google “smart plug” to see
the wide range of such devices currently avail-
able (see examples in Figure 1, Figure 2, and
Figure 3). From the outside it’s almost impos-
Figure 1: The Gosund Figure 3: Shelly Smart
SP211. (Source: Figure 2: Amazon Smart Plug. Plug S. (Source:
https://bit.ly/3wAWu6g) (Source: https://bit.ly/3wCtife) https://bit.ly/3D9TJtQ)
sible to work out what their internal hardware ble) the components used. Not only did it
might consist of. To investigate further, I chose become clear which microcontrollers were
three popular makes of smart plugs to sacri- installed, but also revealed some short cuts
fice in this teardown investigation. I was in the circuit design that have been made in
interested in the following questions: Which order to save costs.
components are used? How are these wired?
Can the unit measure power consumed and The Smart Plugs
mains voltage, and if so, how? Is there any To delve deeper I will take a closer look at
overload protection? three sockets from different manufactur-
  ers. Figure 4 shows the EU6S model from
The most interesting questions are proba- Aoycocr. Figure 5 shows the Tapo P100 type
bly which of these variants can more easily from TP-Link. Figure 6 shows the S26R2 from
be hacked and if the manufacturer’s default Sonoff. In addition to these three, I also got the
firmware could be improved upon. To answer chance to dismantle a TRÅDFRI socket from
these questions, I drew out the circuits of IKEA (Figure 7 and Figure 8) fairly recently
three units and identified (as far as possi- to use some parts for another project.

90 July & August www.elektormagazine.com

 Figure 4: The Aoycocr
EU6S.
Figure 5: The TP-Link Figure 6: The S26R2 Figure 8: The TRÅDFRI
Tapo P100. from Sonoff. Figure 7: Ikea’s TRÅDFRI. internals.

Table 1 contains the technical data of the
three sockets examined in more detail. The
manufacturers specified a maximum load
current of 10 A (2.3 kW at 230 V≈) for the
EU6S and Tapo P100 models. With the S26R2,
Sonoff pushes the limits of what is possible
and offers up to 16 A (3.6 kW at 230 V≈),
which is useful for switching larger loads such
as kettles or electric heaters.
These items are not intended to be service-
able; they rely on the integrity of the casing to
provide electrical isolation from the user. I was
only able to open (and reassemble) the EU6S
and S26R2 smart sockets without destroying
them. Unfortunately, the TP-Link Tapo P100
uses a sealed housing, while getting inside
is possible, it will be damaged in the process.
Incidentally, the TRÅDFRI socket from IKEA
was a job for the hacksaw, it could not be
reused.
Aoycocr EU6S circuit board, on which all components except
The EU6S housing is held together by four the power contacts are located. This layout
triangular socket head screws (Figure 9), makes reengineering easier.
allowing fairly easy access to the internals.  
With the cover off, the PCB is held in place I divided the PCB into three basic functional
by two screws which screw into and make groups. Figure 13 shows the sections of the
contact with the live and neutral pins of the power supply function (yellow), Wi-Fi/MCU
plug. You will need a suitable tool which can module (orange) and relay including wiring
be found in the iFixit kit from the Elektor Shop (magenta). The shunt resistor R7 and IC U3
[1] to access the screw heads. Figure 10 (Figure 14) are used here to provide a means
shows the two internal threads in the plug of measuring the current drawn by the appli-
pins which are used to fix the PCB (Figure 11). ance switched by the unit. What’s obviously
Figure  12 shows the component side of the missing in this configuration is a transformer
Table 1: Technical data of three smart plugs
Name Max load  Wi-Fi BLE  MCU/SoC
Aoycocr EU6S 2.3 kW (10 A @ 230 V≈) 2.4 GHz No ESP8266
TP-Link Tapo P100 2.3 kW (10 A @ 230 V≈) 2.4 GHz Yes* RTL8720CF
Sonoff S26R2 3.6 kW (16 A @ 230 V≈) 2.4 GHz No ESP8285
*Only used to configure Wi-Fi

Figure 9: Four screws secure Figure 10: Hard-to-reach Figure 11: Underside of the

the EU6S casing. mounting of the EU6S PCB. EU6S-PCB with fixings.

Figure 12: Component side of Figure 13: Functional areas of the Figure 14: Close up of U3 and shunt
the EU6S PCB. EU6S PCB. resistor.

lektor July & August 2022 91

 R2
C2

C1

R1

+5V
C3
8 1 D2
DRAIN U1 BP
7 2 AMS1117-3.3
DRAIN GND
6 3
F1 D1 1mH DRAIN FB
L1
5 4
L1 DRAIN C/S 2 U2 +3V3
L2
10 / 0W5 R19 R21
HP-01 2
R20

C8 C9 D3 C15 C7 C4 C5
MOV1

4µ7 4µ7 470µ

U 400V 400V 16V

+3V3 +3V3

R14 R15 R16 R4 LED1

C11
R6
10k

10k

10k

10k
N 1k

R7 R11 SW R3
1 8 10
0 1

VDD SEL
S1

680
2 7 R12
R5 VIP U3 CF1 10
3 6
N OUT 1k VIN CF2 R13
4 5 10
C12 C11 V2P GND

BL0937 LED

R8 R9
L1 1M2 1M2
R10
C14
1k

SEL
CF1
CF2

+5V

+5V
R17
1k7

Q1 1 14
R18
REL RX 2 13 TX
900
GPIO4 3 12 GPIO5
GPIO0 4 11 GPIO2
L1
GPIO13 5 10 GPIO15
Re1 D4 GPIO14 6 9 GPIO12
EN 7 8 TOUT
5V
F2
L1 OUT U4
10A

220170-015

Figure 15: Circuit of the EU6S smart plug.

that would normally provide galvanic isola-
tion in a switched-mode power supply design.
Let’s take a look at the power supply first.
 
The Power Supply
In the upper area of Figure 15 you can see
the circuit of the power supply built around
the IC U1 (HP-01 AL93PH.1, Figure 16). Unfor-
tunately, an Internet search did not bring up
any data sheet for this device. I drew out the mains signal. C8, C9 and L1 are configured
circuit showing all the components wired to to form a pi filter to smooth the resulting DC
this chip to determine the pin functions. voltage. This voltage is applied to pins 5 to 8
  of U1. The two parallel shunt resistors R19 and
Mains voltage is present on the left and enters R20, diode D3 and coil L2 together with the
via F1, a 10-Ω resistor used here as a fuse with feedback network formed by D2, C1, C2, R1
a 0.5 W power rating (Figure 17). The varistor and R2 indicate that U1 is most likely a buck
MOV1 provides overvoltage protection. Diode switching regulator providing a 5 V DC output
D1 provides a cost-effective half-wave rectified smoothed by capacitor C7 from the 325 V DC

92 July & August www.elektormagazine.com

 R2
C2

4k64
C1

R1

+5V
C3
8 1 D2
DRAIN U1 BP
7 2 AMS1117-3.3
DRAIN GND
6 3
F1 D1 1mH DRAIN FB
L1
5 4
L1 DRAIN C/S 2 U2 +3V3
L2
10 / 0W5 R19 R21
HP-01 2
R20

C8 C9 D3 C15 C7 C4 C5
MOV1

4µ7 4µ7 470µ

U 400V 400V 16V

220170-016

Figure 16: Markings on the HP-01 AL93PH.1 (U1) are more readable using false colouration.

input on capacitor C9. The circuit requires 5 V
for operation of the relay K1 which switches
power to the appliance.
 
IC U2 is the familiar AMS1117-3.3 low-drop
linear regulator that converts 5 V into a stable
3.3 V. This voltage level is required by the
Figure 17: F1 is a 10-Ω resistor.
microcontroller and Wi-Fi hardware which
cannot tolerate 5 V.
 
I have seen similar power supply configura-
tions used by standard mains powered LED
lamps with an E14 or E27 base. Even though
I couldn’t find a data sheet for this particular
IC (U1) I can deduce that it provides a similar
function to chips such as MPS MP4050A [2]
(Figure 18) or Power Integration LNK302 [3]
(Figure 19) often used in LED lamps. Their
data sheets give information about pin assign-
ments and circuit wiring indicating they are
compatible with the function of U1. Figure 20
shows the block diagram of the internal struc-
ture of the MPS MP4050A. Although GND is
the reference for the control signal and the
gate driver, it is not the reference point for the
output voltage. It is reasonable to assume that
U1 is used here in a similar way. The wiring
and layout for an application using the MPS
MP4050A is similar to how U1 is configured
in the circuit of the UE6S smart plug.
 
Caution! This uses a non-insulated
power supply without galvanic isola-
tion, the ground of the circuit board can
have the same potential as the phase,
depending on the polarity of the socket
wiring. Touching the circuit ground and
almost all components and circuit board
pads is therefore life-threatening!

Figure 18: Example application circuit using the MPS MP4050A. Source: [2].
 

lektor July & August 2022 93


Figure 19: Basic circuit using the Power Integration LNK302. Source: [3].
Figure 20: Internal Block diagram of the
MPS MP4050A. Source: [2].
The use of non-isolated DC/DC step-down
converters is a cost effective method of
supplying power to the electronics in
these smart plugs and many other electri-
cal products. The PCB and all operating
buttons and LEDs are enclosed in a plastic
housing, under normal operation and with
Figure 21: An HJL-01 chip is used in the
EU6S.
94 July & August www.elektormagazine.com
an undamaged housing there is no danger
you could accidentally touch any live parts.
All electrical products should carry markings
indicating they conform to current standards
of electrical safety. The switching regulator
ICs used in these designs offer a wide-range
input and will often work with AC voltages
Figure 22: Pin-outs and block diagram of the BL0937. Source: [4].
between 80 to 250 V (at 50 and 60 Hz) to
ensure the product is compatible with mains
supply standards around the world.
Measuring Power
This smart plug is able to record current and
thus power and probably also the mains voltage
supplied to the load. The middle left part of the
circuit in Figure 15, containing the chip U3, is
responsible for this function. The markings on
U3 can be seen in Figure 21. It’s possible to
make out “HJL-01” along with other numbers
and letters but sadly there was no sign of a
data sheet on the internet. Further searching
uncovered similar, pin-compatible, chips such
as the BL0937 or the HLW8012. Since U3 is
powered from 3.3 V, it is probably more similar
to the BL0937 chip, the HLW8012 needs a 5 V
supply. The question remains what does U3 do?
The BL0937 is a smart meter IC for sockets
and other smart applications that can measure
power in single-phase networks. The pin
assignment and block diagram of the BL0937
can be seen in Figure 22. Pin CF of this chip
outputs continuous 38 µs pulses at a repetition
rate proportional to the active power measure-
ment. Pin CF1 outputs a similar pulse train with
a repetition rate equivalent to the RMS value
of active current (when the pin SEL input is at
‘0’) or the RMS value of active voltage (when
the pin SEL input is at ‘1’).
When comparing the circuit of U3 in Figure 15
with the reference circuit of the BL0937 given
in the data sheet [4], it is noticeable that the
voltage divider resistor values​ for voltage
measurement do not correspond. The U3 uses
a divider ratio of 2,400:1 instead of 1,980:1 as in
the BL0937, which should be relevant if alter-
native firmware is contemplated.

Figure 23: The Aoycocr EU6S
uses a 10 A slow blow fuse F2.
Figure 26: The TP-Link Tapo
P100 forms a compact unit.
Current is measured by the voltage drop
across a 1-mΩ shunt resistor R7 placed in
the neutral (N) path. The voltage across R7
connects to pins 2 and 3 of U3 through R6
and R5. Together with C11 and C12, they form
a low-pass filter for removing mains-borne
interference.
 
Pins SEL, CF1 and CF use series 10-Ω resis-
tors and 10-kΩ pull-up resistors for the signals
which then connect directly to the MCU pins.
The interface is as simple and as low-cost as
possible. The HJL-01 chip is probably cheaper
than a BL0937, which retails for less than ¢60.
By comparison the HLW8012 alternative is
about 30% more expensive.
 
The Relay, LED and Push Button
Appliances plugged into the socket are
switched by relay K1. The type used is the
NHLC 932-5VCD-SL-AH with a maximum
load capacity of 10 A at 250 V≈. Relays with
this spec generally cost around ¢50. The control
can be seen in the bottom left of the circuit in
Figure 15.
 
The relay coil K1 is connected to 5 V and
switched by transistor Q1. This driver transis-
tor switches 0.4 W from 5 V through the coil
from the 3.3 V signal output by the controller.
D4 ensures the back-EMF generated when
switching the inductive relay coil is safely
discharged to the supply rail.
Figure 24: The ESP8266 Module.
Figure 27: The pins down the
middle are soldered.
The slow-blow mini-fuse F2 (Figure 23) is rated
at 10 A and protects the relay and circuit board
from damage when excessive current is drawn
by the switched load. The fuse can be easily and
cheaply replaced to get the unit working again.
LED1 uses a 680-Ω series resistor (Figure 15,
bottom middle). This low value of current can
be supplied directly from the MCU pin LED.
The push button input uses a 10-kΩ pull-up
resistor R4.
 
The Wireless Microcontroller
The MCU used in this unit is an ESP8266,
which has built in Wi-Fi compliant to IEEE
802.11 b/g/n. Together with a few other compo-
nents it is on its own small PCB (Figure 24),
fixed upright to the main PCB. Apart from a
RESET, no other test point is labelled on the
module. Next to the MCU is the ZB25VQ80A-
ITG 8  Mb capacity flash memory chip
(Figure 25). I unsoldered this small board
to find which components connect to the
ESP8266. I was then able to determine the pin
assignments of the module board and allocate
functions to them. This information is given in
the circuit of Figure 15.
 
MCUs like the ESP8266 can easily be provided
with their own firmware. It is possible, for
example to operate the MCU with ESPHome
[5] or Tasmota [6]. In addition to simply switch-
ing the load via the relay, it is also possible to
Figure 25: A ZB25VQ80AITG
8-Mb flash chip.
Figure 28: The PCB can be divided into three
main sections.
measure and read out the energy consump-
tion of the connected device. However,
the measurements from U3 should also be
calibrated so that the data collected can be
relied upon.
All the programmer interface contacts of the
MCU are on the underside of the PCB so its
necessary to take the PCB out of its enclosure
if you wanted to flash ESPHome or Tasmota to
the MCU. The screws are not so easy to reach
— in particular, assembly is more difficult than
disassembly. You can, however, open the case
without causing any damage, which is a bonus.
The TP-Link Tapo P100
The Tapo P100 socket (Figure 26) from TP-Link
is significantly more compact compared to the
product from Aoycocr. No screws are visible
from the outside, because the Tapo P100 model
has a housing whose parts are held together
with clamps or latches. Opening the case will
leave marks, even if you are careful. With the
right tools, you can get to the inner workings
quite quickly. Unfortunately, the circuit board
cannot be taken out of the housing that easily
because it is soldered to the contacts of the
plug pins (see Figure 27). I desoldered it to
take a look at the bottom of the board.
 
The PCB can be roughly divided into the three
areas of power supply, relay and MCU based
on functions, which are marked in Figure 28.
lektor July & August 2022 95
 R168
10
1W
L101 +5V
B1 U4
L BP2525X
R150 D15
4 3
3k3 DRAIN VCC
5
N DRAIN
2 C170
SEL
6 1
CS GND
C176 C171
R11 1% L10
2µ2 4µ7 1 3
400V 400V D13
R9 1% C172
1 3
470µ
10V

U1
+5V
RTL8720CF-VA
F2
30 VBAT_IN

D10 A1-1A-F
L SW1 D14
R5
100 23 GPIOA_9
L_OUT

C175 C174
24 GPIOA_10

SW1 = RJ-SS-105DM Q1
R155
500 21 GPIOA_7
R158
10k

R6
10k 31 VD33_OUT

SW3 R7
100 38 GPIOA_17

C6

220170-032

Figure 29: The Tapo P100 circuit diagram.

Figure 31: Realtek
RTL8720CF.

Figure 30: The BP2525 is mounted

underneath the PCB.

Figure 32: The thermal fuse is fitted close to

the relay housing. Figure 33: Disassembled Sonoff S26R2.

96 July & August www.elektormagazine.com

 U2
BP2905
4 3
DRAIN VCC
5
DRAIN
2
SEL C5
6 1
R1 CS GND
+5V
L1 39!
D2
AP1117-33

RV1 CE1 L1 1mH R2 L2

STE 10D471k DB1 0!33 U1 +3V3
680µH
47n R3
U 275V

100!
N D1
R5
CE3 CE2 C6 CE4 C2 C1 C3 C4 C7

2k
3µ3 3µ3 470µ
400V 400V 10V

+3V3

LED1

R10
+5V +3V3
brother, the RTL8720DN [9], also offers Wi-Fi
in the 5 GHz band in addition to the usual
1k

+3V3 R12
L1 OUT
Q2
R8
2.4 GHz band.

1k
K1 D3
 
LED2
1k

S1
Push Button, LED, Relay and Fuse
C8
L1
R11
Control of the LED in this circuit (Figure 29)
Q1
R6 looks a bit strange at first glance. Resistor R5
1k

1k

R7 REL LED1 SW1 and a two-colour LED D10 connect directly

between pins GPIOA_9 and GPIOA_10 of the
4k7

microcontroller. By generating PWM signals

220170-015
Figure 34: Circuit diagram of the Sonoff S26R2.
A chip to measure current is not available in
this design (Figure 29).
 
The Power Supply
The power supply circuit was easy to work
out because the IC labelling was legible and
datasheets could be found. Previous experi-
ence with the EU6S previously was helpful:
Here, too, a 10-Ω resistor (R168) — this time
with a 2 W rating — is installed as a fuse and
current limiter.
 
The main chip U4 is a switching regulator
type BP2525 [7] (Figure 30). Here, too, a
non-isolated step-down converter is used;
it generates a selectable 5 V or 3.3 V from
up to 500 VDC at the input. A bridge recti-
fier converts the AC mains voltage into a DC
voltage which is then smoothed with a Pi filter
made up of C176, L101 and C171.
 
Chip U4 provides an output voltage (depend-
ing on the version) up to 500 mA at 5 V.
GPIO12 GPIO13 GPIO0
from these two pins, it is possible to produce
a range of LED colours.
Transistor Q1 switches the relay and D14
serves as a freewheeling diode. GPIOA_7
controls the base of Q1 via R155. A pull-down
resistor is provided with R158 so that the
transistor remains off while the firmware is
being loaded at startup. F2 is a special feature
The wiring is easy to understand from the here (Figure 32). This non-resettable thermal
data sheet and the components are easy to fuse (AUPO A1-1A-F [10]) ruptures in the event
identify. There is not much difference to the of overheating when the relay is overloaded.
EU6S power supply. There is no 3.3-V supply This interrupts current to the coil so that the
generated in this design. The MCU used here relay drops out and cools down but the fuse
(see below) makes it clear why this is the case. does not reset. Access to the internals to
  replace the fuse is quite difficult so any P100
The RTL8720CF with Wi-Fi and BLE unit suffering this fate is most likely to join
The Tapo P100 does not use the ubiquitous the increasing mountain of electronic waste.
ESP8266 microcontroller but instead uses  
the RTL8720CF from Realtek (Figure 31). The The built-in button has a classic circuit with a
datasheet [8] indicates that it only requires a 10-kΩ pull-up resistor to 3.3 volts, a capacitor
5-V supply because it has an integrated 3.3-V for contact debouncing and a 100 Ω resistor
regulator. The maximum load on this 3.3-V connecting to GPIOA_17.
supply is 450 mA.
  The Sonoff S26R2TPF
Wi-Fi is supported in the 2.4 GHz band and By the time I got round to dismantling the
Bluetooth Low Energy 4.2 is also available. The Sonoff, I knew what to expect and was just
MCU is based on an ARM Cortex-M33 core curious as to how the various functions
(called KM4 by Realtek), which can be clocked had been implemented. The Sonoff model
at up to 100 MHz. The MCU also has 2 MB S26R2TPF (Figure 33) can switch up to 16 A
integrated flash memory and 256 KB RAM, at 250 V≈; its circuit diagram can be seen in
altogether an interesting choice. Its bigger Figure 34.

lektor July & August 2022 97


Figure 35: The BP2905 is compatible
with the BP2525 chip.
 
The Power Supply
The power supply section is built around U2,
a switching regulator IC marked as a BP2905
(Figure 35). Unfortunately, I couldn’t find a
datasheet for this. It is possible to deduce
from the package outline, pins and wiring to
external components that it is most likely a
pin-compatible relative of the BP2525 chip.
 
In this circuit again resistor R1 acts as a
fuse. The bridge rectifier DB1 is followed by
a pi filter made up of CE2, L1 and CE3. The
BP2905 produces a stable 5 V on the reser-
voir capacitor CE4. As with the EU6S, a linear
voltage regulator is used to generate 3.3 V
for U1.
 
Push Button, Relay and LED
The wiring of the button S1 is simple and only
requires a 1-kΩ pull-up resistor plus capacitor
for contact debouncing. It connects directly
to the SW1 input of the MCU without any
series resistor.
 
The anode of LED2 connects to the 3.3-V
supply while its cathode is switched via series
resistor R11 from a GPIO pin of the MCU.
 
LED1 is powered from the 3.3-V supply and
has its own driver transistor Q2 which is
switched by the same GPIO signal from the
MCU which operates the relay. It indicates
when the load is switched on.
 
The relay is connected as in previous variants
with one terminal of its coil connected to 5 V,
and the other connection switched to ground
via Q1 by a 3.3 V control signal from the MCU.
A diode (D3) across the coil again snubs the
coil’s back-EMF. As with the other two sockets,
a resistance between the base and emitter of
98 July & August www.elektormagazine.com
Figure 36: The ESP8285 module fitted in
the S26R2.
the switching transistor prevents the relay from
unintentionally switching during boot.
 
The Wireless MCU
The ESP8285 MCU (U1) is on its own mini
circuit board, mounted at right angles to
the main board (Figure 36). Thanks to its
integrated flash memory, this ESP8266 variant
does not need any additional flash chip. The
pin assignment can be found in the circuit
diagram in Figure 34.
Software to Use a Different
Cloud
While we have concentrated on the hardware
aspects of smart plugs,there are also consid-
erations when the unit is in use and its
connectivity to a cloud-based server. The
manufacturer’s software will usually connect
the socket to the home Wi-Fi and then
needs to be configured via an app so that
the socket can be used with a smartphone,
Amazon’s Alexa, Google’s Home or Apple’s
Homekit. Many smart sockets use the cloud
service of a provider who hosts their servers
somewhere in the ether which ultimately also
has access to the smart device. This config-
uration then allows loads to be switched
remotely via the Internet. Unfortunately, such
a cloud connection is not entirely reliable
since there is no guarantee that any cloud
provider will not simply discontinue its
service one day [11].
 
As already mentioned, there are alternatives
for smart plugs based on the ESP8266 by
installing the open-source firmware TASMOTA
[6] or ESPHome [5]. This replacement of the
original firmware then allows the device to
be controlled from your own LAN via MQTT,
WebUI, Alexa or other protocols.
If you are interested in using alternative
firmware on the devices a very helpful
website on this topic can be found at [12]
called Tuya-Convert. A number of these smart
sockets can even be flashed with alterna-
tive firmware without the need to physically
get inside the unit. If your model needs to
be opened most ESP8266 MCUs can be
re-flashed using the Tasmotizer [13]. Appro-
priate settings must be made according to
the model.
If you want to use ESPHome, you can install
this on some sockets without opening the
case using OTA (Over The Air) programming
via Wi-Fi [14]. With some models, however,
you will need to resort to a screwdriver and
soldering iron.
 
Second warning: As this article shows,
all sockets discussed here use non-iso-
lated power supplies. This means parts
of the circuit will be at peak line voltage
(almost 400 V in Europe). It can be lethal
if any part of your body makes contact.
The design of these smart plugs relies
on the integrity of their insulated enclo-
sures to protect the consumer from
harm. Touching any internal components
during operation is life-threatening.
This cannot be overstated: Never work
on live circuits and always follow the
applicable safety rules. In addition: The
mains-side capacitors used in power
supply units can still carry a high voltage
charge of several hundred volts for a
while after they have been switched off.
Before working on these (unpowered)
circuit boards, make sure to discharge
them to be on the safe side.
ESP8266, ESP8285, RTL8720...
If you are now thinking of buying a smart
socket with the intention of installing alter-
native firmware, note that both ESPHome and
Tasmota currently only support ESP family
chips. While there is ample documentation
for the RTL8720CF that does not mean that
alternative firmware support is available.
 
In addition to the MCUs, we found here you
might find other types such as the Realtek
RELATED PRODUCTS
RTL8710, Beken BK7231T or the XradioTech
XR809 installed in other brands of smart
plug. As ever it makes sense to do as much > Koen Vervloesem, Getting Started with ESPHome (Elektor 2021) (SKU 19738)
background research as possible before www.elektor.com/19738  
purchase. If you’re lucky, another dedicated
developer has already done the hard work > CH340 USB/TTL Converter module CH340G (3.3 V/5.5 V) (SKU 19151)
www.elektor.com/19151  
to analyze and then publish their findings on
the web. Another important aspect is that > OWON OW18B Digital Multimeter with Bluetooth (SKU 18777)
manufacturers products in the same range www.elektor.com/18777
will inevitably undergo optimization over time
so that hardware may change possibly by
installing other MCUs thereby making them
more difficult (or more interesting) to modify.

A Helpful Hack
As we discovered, all three smart plugs are
very similar. An un-isolated power supply
steps down the mains voltage to power an
MCU with Wi-Fi that controls a relay. There are
some variations in the implementation such as
protective measures, the LED control method
or the addition of a smart meter chip. Even
though IKEA’s TRÅDFRI only made a guest
appearance here, I can tell you that there is
nothing out of the ordinary in its design also.
Hacking into these products was definitely
instructive for me. I got a glimpse into some
safety aspects of the design and how “optimi-
zations” can cut costs, which is a particularly
important criteria in the design of high volume
consumer products. I can use that knowledge
to assess other products and use the circuits
as inspiration for future projects.
 
If you have already tinkered with smart
sockets, share your experiences (whether
good or bad) with us. Any unusual findings
and insights into the design of other smart
plugs are very welcome and may help
others. 
220170-02
Questions or Comments?
Do you have any technical questions
or comments prompted by this article?
Email the author at mathias.claussen@
elektor.com or contact the team at
editor@elektor.com.

WEB LINKS
[1] iFixit Pro Tech Bit Set: www.elektor.com/ifixit-pro-tech-toolkit
[2] MP4050A: www.monolithicpower.com/en/mp4050a.html
[3] Power Integrations LNK302: www.power.com/products/linkswitch/linkswitch-tn/lnk302dg
[4] BL0937: https://bit.ly/37KY7Uo
[5] ESPHome: https://esphome.io/
[6] Tasmota: https://tasmota.github.io/docs/
[7] MP2565: www.monolithicpower.com/en/mp2565.html
[8] RTL8720CF: https://bit.ly/3DbrJ8Z
[9] RTL872xD: https://bit.ly/3JESUf2
[10] AUPO A1-1A-F: www.tme.eu/en/details/a1-1a-f/thermal-fuses/aupo/
[11] N. Statt, “Nest is permanently disabling the Revolv smart home hub,” The Verge, April 4, 2016: https://bit.ly/3ICGi6O
[12] Tuya Convert: https://tasmota.github.io/docs/Tuya-Convert
[13] Tasmotizer: https://github.com/tasmota/tasmotizer
[14] ESPHome OTA Installation: https://bit.ly/3JIlps9

lektor July & August 2022 99

 background

Skin Impedance
and Skin Capacitance
Small Experiments

By Burkhard Kainka (Germany)

What do capacitors have in common with wires in water or fingers

touching metallic surfaces? Quite a lot, as the author discovered
with his own experiments and measurements. Skin impedance
can be used for a variety of interesting applications.

Many years ago, I encountered some strange
effects involving wires in water and fingers
touching metallic surfaces. At the time, I
wanted to measure wood moisture content
using stainless steel screws in the wood.
Unfortunately, the characteristics of the
contacts changed quickly, resulting in less
current flow. It was even possible to build up
a charge on the contacts, and I was able to
measure a voltage for several seconds after
the supply was disconnected. I decided to call
this a ‘wood battery’. Only later did I realise
that this involved a double layer of water
molecules forming a double-layer capacitor.
Now I have clarified this mystery with further
measurements.
Measuring Skin Impedance
The starting point for the measurements was
the development of a conductivity meter for
human skin (Figure 1) using a microcontrol-
ler, which was ultimately intended to be used
as a lie detector. I made the measurements
using an AC signal, or, more precisely, a short becomes very small and the internal imped-
positive pulse followed by an equally short ance is therefore dominant. This means that
negative pulse. The result was astonishingly electric shocks are more dangerous than you
good conductivity values. A normal multi- might think from an ohmmeter reading.
meter shows a resistance of around 1 MΩ  
between the two contact electrodes, but the The other focus of such investigations lies in
microcontroller instrument gives values of the medical domain. The aim is to find out
around 10 kΩ.
 
If you do an online search on this topic, you
will quickly discover that skin impedance
is frequency dependent. The higher the
frequency, the higher the conductivity. The
impedance also drops with increasing voltage.
Investigations like this are performed for a
variety of reasons. At one time it was about
the dangers of electricity. Researchers inves-
tigated how much current could flow through
the human body under various conditions.
This revealed that the internal impedance
from hand to hand is only a few thousand
ohms, and the skin impedance is dominant.
With high AC voltages, the contact resistance Figure 1: Measuring skin impedance.

100 July & August 2022 www.elektormagazine.com

how ECG electrodes work or what conclu-
sions can be drawn from the body impedance.
These investigations reveal that the contact
resistance can change over time during the
investigation and that it can be influenced by
a saline solution. The impression remains that
the skin is a very complex organ.
An Equivalent Circuit for the
Skin
What an electronics hobbyist or engineer
actually wants is a usable equivalent circuit
for small signals. Up to now this has usually
been assumed to be a resistance on the order
of 100 kΩ to 1 MΩ, but this is apparently only
valid for low DC voltages up to approximately
9 V. The behaviour of the skin is different with
AC voltages. So, I switched on my sinewave
generator and built a voltage divider consist-
ing of two fingers and a fixed 10 kΩ resistor.
Measurements with low voltages around 1 V
(Figure 2) showed that the current waveform
was an undistorted sinewave (Figure 3). The
strong frequency dependence could also be
confirmed. Between 1 kHz and 10 kHz the
impedance dropped by around a factor of
10. My hand with the two wire electrodes was
thus behaving more or less like a capacitor.
Comparisons with different capacitors from
the parts bin showed that a 3.3 nF capacitor
showed very similar behaviour.
 
The entire measurement process can be
simplified by using a square-wave signal
(Figure 4). For this I used the test signal from
the oscilloscope, with an amplitude of 0.2 V
and a frequency of 1 kHz. This is not an AC
voltage, but instead pulsed DC. The output
voltage is zero for 0.5 ms and then 0.2 V for
0.5 ms. This signal can be regarded as a DC
voltage of 0.1 V with a superimposed square-
wave signal having a peak amplitude of 0.1 V.
 
The result is the typical pulse waveform of
an RC high-pass filter (Figure 5). Since the
measuring signal has a DC component, it
is clear that the DC conductivity is insignif-
icant, as otherwise the output signal would be
noticeably offset in the pulse region. This also
confirms that skin with two wire electrodes
FINGER
1V
OSC
10k
200621-006
Figure 2: Measuring with AC signals. Figure 3: The voltage drop over the resistor.
FINGER
0V/0V2
1kHz
OSC
10k
200621-008
Figure 4: Measuring with square-wave signals. Figure 5: The same signal waveform as with a
high-pass filter.
behaves like a capacitor with a capacitance
of a few nanofarads. In parallel there is a high
resistance of around 1 MΩ.
An Explanation
The large capacitance of a several nanofar-
ads with the very small contact areas of the
wires gave me the idea that what is actually
formed here is a double-layer capacitor. The
skin moisture produces a water film on the
metal surface. A double layer of polarised
water molecules is then formed on the bound-
ary surface. This would be exactly the same
as the operating principle of a supercapaci-
tor, in which a graphite coating provides an
especially large boundary surface.
 
This sort of capacitor can also be formed with
two copper wires in pure water (Figure 6). The Figure 6: Experimental setup for a double-layer
measuring circuit is still the same. To obtain capacitor.
lektor July & August 2022 101
 you need more capacitance for some reason,
it’s not a problem.
 
WATER
In summary, the result is clear: the impedance
0V/0V2
of two fingers with wire contacts corresponds
to the equivalent circuit in Figure 9. The exact
1kHz result, however, can vary significantly from
OSC
one person to the next and is, of course, very
10k

strongly dependent on the contact area, the

200621-011
instantaneous skin moisture level and the
contact pressure.
 
Figure 7: The water capacitor with an applied Figure 8: Output signal of the water RC network.
square-wave signal. If we examine this capacitor in more detail,
it turns out to actually be two capacitors.
The dielectric of these capacitors is the thin
water layer between the skin and the metal
electrode. In addition there is naturally the
inner impedance of the body, for example,
10n between one finger and another. All in all, this
FINGER leads to a more extensive equivalent circuit
1M

FINGER
1k
1M

3n3 (Figure 10). The internal impedance is difficult

to determine, and for the sake of simplicity, it
10n

200621-013
200621-014
Figure 9: The simplified equivalent circuit. Figure 10: The more extensive equivalent circuit.
comparable results, the wires should only be
ATtiny85 immersed about 1 mm into the water.
B3 B4 B1  
The measurements (Figure 7 and Figure 8)
show that the glass of water with two
electrodes has more capacitance than my
hand. The capacitance can be calculated
from the waveform, but it can also be deter-
mined by comparison with other capacitors.
200621-015 The water capacitor has approximately the
same effect as a 47 nF film capacitor, despite
the very small surface area of the electrodes
Figure  11: An automatic Morse key with touch
sensors. immersed only about 1 mm into the water. If
Figure 12: The touch contacts. Figure 13: Signal on port B3.
is here assumed to be 1 kΩ.
If you repeatedly increase the frequency,
the impedance of the skin capacitor drops
accordingly. Then the internal impedance
becomes more significant. I personally know
this from painful experiences in amateur radio.
If you accidentally touch the output connec-
tor of a shortwave transmitter, you don’t get
an electric shock but instead a burnt finger.
The burn effect, however, is totally different
from what you get from touching a hot solder-
ing iron. There is no burn blister like you get
from a soldering iron, and it feels completely
different. The heating clearly occurs beneath
the epidermis in the deeper skin tissue with
higher conductivity, in particular where the
current density is highest.
Application: Touch Sensor
A possible application for the finger capacitor
is an automatic Morse key with touch sensors.
I used an ATtiny85 microcontroller (Figure 11)
together with two touch sensors from an
FM radio (Figure 12). In the radio the sensor
currents were amplified by transistors, which
means that the skin resistance was sensed.
This has the disadvantage that a certain
degree of soiling can lead to malfunctions.
An evaluation method in which the micro-
controller measures the charging time of the
capacitor formed by the finger is significantly

102 July & August 2022 www.elektormagazine.com

 Listing 1: Morse key.       waitms 1
      next n
'ELbug with touch sensors       waitms 100
     end if
$regfile = "attiny85.dat"    portb.3=0            ' discharge B3
$crystal = 8000000    ddrb.3 = 1
$hwstack = 8  
$swstack = 4    DDRB.4 = 0           ' B4 high impedance
$framesize = 4    Portb.4 = 1            ' pullup
dim n as byte    waitus 10
     if PINB.4 = 0 then   ' still low after 10 µs?
        portb.4 = 0
ddrB = &B00011010       ddrb.4 = 1
Portb = 0       for n = 1 to 150  ' output dash
        PortB.1 = 1
        waitms 1
Do       portb.1 = 0
   DDRB.3 = 0            ' B3 high impedance       waitms 1
   Portb.3 = 1            ' pullup       next n
   waitus 10       waitms 100
   if PINB.3 = 0 then   ' still low after 10 µs?    end if
      portb.3 = 0    portb.4=0            ' discharge B4
      ddrb.3 = 1    ddrb.4 = 1
      for n = 1 to 50    ' output dot    waitus 10
      PortB.1 = 1 Loop
      waitms 1  
      portb.1 = 0 End

more reliable (see Listing 1). For this a port

(B3, B4) is first configured as an output and Questions or Comments?
switched low to discharge the capacitor. Do you have technical questions or
Then the port is put into the high-impedance comments about this article? Email
state and the internal pull-up (approximately the author at b.kainka@t-online.de or
50 kΩ) is enabled. The sensor capacitor then contact Elektor at editor@elektor.com.
starts charging. The critical point is when the
port voltage rises above the threshold level
of around ½ VCC. For the task at hand, it is
sufficient to wait 10 µs and then read the port
state. If a low state is detected at that point,
the key is regarded as being pressed. In this RELATED PRODUCTS
case a square-wave signal as a feedback
tone is generated for the piezo loudspeaker
connected to B0. Figure 13 shows the situa- > B. Kainka, Basic Electronics for Beginners 
(Elektor 2020) (PDF, SKU 19213)
tion with a too-light touch. There the pulses
www.elektor.com/19213
rise to 4 V within 10 µs, so the key is regarded
as not pressed.
 
In the next edition of Elektor (the Summer
Circuits 2022 issue), I will show two other appli-
cation circuits: the finger capacitor and the
two-finger organ. Until then, I hope you have
a lot of fun with your own experiments!
200621-01

lektor July & August 2022 103

 developer’s zone
Tips & Tricks, Best Practices and
Other Useful Information

From Life’s
Experience
No Local Business

By Ilse Joostens (Belgium)

When I was a teenager, the main

hobby electronics activity was
rebuilding audio amplifiers, audio
equipment and related disco stuff,
occasionally with the more traditional
DIY lab power and measurement
equipment. Many years later, when I took the
step into self-employment, I decided to leave the
common path and focused more on niche products
Photo: Dim Dimich / Shutterstock.

with vintage components such as nixie tube clocks

that are still immensely popular. At that time, the
product range was still quite limited with only a few
foreign resellers, and I had made up my mind to just
try to influence the entire Belgian market. Unfortunately,
not much later I realised that there seemed to be no market at
all in Belgium for this kind of more technically oriented stuff.
Before we knew it, we had grown into a micro-multinational
with customers all over the world, except in Belgium.

You Can Catch More Flies with Honey Than With
Vinegar
I used to love the Piet Pienter and Bert Bibber comics. The fact that
technology plays an important role in this series by Jozef Van Hove,
alias Pom, who in real life was a qualified engineer at Blaupunkt in
Germany during the Second World War, obviously contributed to this.
If I remember correctly, one of the personalities once said, “You can
catch more flies with honey than with vinegar.” It is the real truth, with
some elementary politeness, a business-like attitude and honesty one
can succeed in a foreign country.
Somewhere in 2012, when we had established an entire range of
products using vintage technology, but we were also among the
first to offer enclosures and expansion boards for the Raspberry Pi
computer. Our website was available in both Dutch and English and
sales were made through our own web shop and eBay. However, the
orders from Belgium just didn’t come... It was quite a lot of work to
keep the website up to date in two languages, and after a while we
decided to stop the Dutch site without much ado.
 
Since then, we have served countless foreign customers. Our products

104 July & August 2022 www.elektormagazine.com

were sold in more than 25 countries, including far-off locations such thanks to the competition, there is often nothing left than simply pay
as Australia, Hong Kong and Japan. We have gained a number of a part of the shipping costs yourself, which puts a lot of pressure
good friends and collaborations on the profit margins. When shipping
over the years. Belgium, unfortu- outside Europe, there are also
nately, was a different story, and customs formalities like the notori-
all in all, we have sold maybe 20 ous CN23 form and the so-called
times to customers in Belgium, “Commercial Invoice”. Apart from
and even then, mostly to acquain- the fact that the list of TARIC codes
tances and family. And even sometimes causes me a headache,
worse, cooperating with people it is not so bad in terms of complex-
in Belgium seems not possible at ity, but it is a lot of extra adminis-
all. We have been scratching our trative effort and time that I could
heads about this phenomenon, but have spent in a better way. On top
unfortunately, there is no obvious of that, the tax authorities rejected
explanation. Is it due to the fact international phone calls outside
that Belgium lags behind in many office hours, as they do not consider
areas compared to its neighboring them as business calls. My God, have
countries? Do Belgians think that you never heard of the existence of
their own products are less good time zones?
or more expensive than those from  
abroad? Fear of the unknown? If you want to make a profit from
Are they envious? Or is it a your activities without too much
kind of technophobia? Who hassle, some local business is
knows? Apparently, we are a good thing. In Antarctica,
not the only ones because just like in Belgium, there is
a well-known Belgian no local market, and you
chocolatier and supplier have to buy and sell every-
to the Royal Court once thing from abroad. This
mentioned this phenome- is called the South Pole
Photo: Jeremy Walter / Shutterstock.
non in an article in my favor- effect, and forget about the
ite business newspaper [1]. The North Pole, which has been melting
fact is that, since we started working with Elektor, our products have away for a while now, and you won’t survive there anyway. It is not
suddenly found their way from the Netherlands to Belgium. that bad here yet.
  220198-01
The somewhat surprised reaction of one of our few Belgian custom-
ers was also typical. He thought it was fascinating but also hard to
understand that someone here, in particular in Belgium, was develop- Questions or Comments?
ing expansion boards and enclosures for the Raspberry Pi. He didn’t Do you have any technical questions or comments prompted by this
expect this at all, wished us good luck and suggested that if we really article? Send an email to the editor of Elektor via editor@elektor.com.
wanted to move our business forward, we’d better move abroad.

The South Pole Effect

Each customer is valuable and, in principle, it should not make much
difference whether they are located in Timbuktu or just around the
corner. In practice, things are not so prosperous, especially regarding
the local mail services, customs and the tax authorities are sometimes
difficult to deal with. As soon as a national border is crossed, postal
WEB LINK
rates suddenly skyrocket. The absolute distance is less important,
and sending a small parcel a few dozen miles away to a neighboring [1] Proud Mary:
country suddenly costs almost as much as sending it to the United www.tijd.be/cultuur/gastronomie/Proud-Mary/9491697
States. Customers prefer the lowest possible shipping costs and,

lektor July & August 2022 105

 review

Pokit Meter Review

A Swiss army knife of test gear

By Dr. Thomas Scherer

The Pokit Meter is a puck-sized device with extendable probes

that can function as a multimeter, oscilloscope, spectrum
analyzer or data logger and hooks up to a smart device via
Bluetooth to provide its GUI. The concept is novel, but what can
it really do and how well?

I have been looking for a fairly basic data
logging device for some time now. When I
read the spec for the Pokit Meter, it looked
as if it might just fit the bill. The design looked
rather unconventional and interesting. I didn’t
have to wait long before the small cardboard
box (Figure 1) was sitting on my bench.
Symbols on the outside of the box (Figure 2)
indicate the functions of the device inside.
On the right in Figure 2c, you can see what’s
inside. The 48 mm diameter 18 mm thick puck
shown in Figure 1 is the Pokit meter. Gripper
probes in red and black are to the left
and right and at the top right is a tiny
spare fuse (Figure 3). This miniscule
SMD component is a spare fuse on a
tiny PCB. Some versions of the packag-
ing include a pop out section which
holds the fuse securely and allows it
Figure 1: The boxed Pokit Meter.

106 July & August 2022 www.elektormagazine.com

 a b c

Figure 2: On one side it says that you can measure with it, anything and anywhere (a). Another
side is bursting with features (b). The opened packaging (c) showing what’s in the box.

to be safely stowed into one of the pockets in
the carry pouch. I took it out completely and
taped it to the instruction sheet to ensure it
doesn’t get lost.
Let’s Fire It Up
Figure 4 shows all the items spread out on the
bench. A zip up carry pouch has space for the
Pokit meter, spare fuse and gripper probes.
The 8 x 22 cm information sheet lists the basic
technical information for the device and lets
us know that the corresponding ‘pokitMe-
ter’ app is available to download for smart
devices running Android or iOS from the play
or app store. It also gives battery information
and indicates that the fuse is situated under-
neath the battery. A link to the manufacturer
or QR code would have been useful here. On
the homepage [1] of the manufacturer Pokit
Innovations there is hardly any further infor-
mation to be found; but there is a link to a
forum. At least I know now that this measur-
ing device was funded through a successful
crowdfunding campaign and that I have the
Figure 3: Keep it safe. The tiny spare SMD fuse
black version from a range of four coloured on its mini PCB.
versions and that there will soon be a Pokit
Pro version and a Smartwatch app.
 
The Pokit meter’s retractable leads are termi- supplied. The leads extend (Figure 5) to a
nated in needle point probe tips which can be maximum reach between the gripper probes
used directly for probing a circuit or can be of around 56 cm. The button on the bottom of
plugged into one of the spring loaded grippers the puck retracts the leads just like the power

Figure 4: The contents set out on the bench. Figure 5: With the two leads extended, Pokit is ready to measure.

lektor July & August 2022 107


a b
Figure 6: Don’t forget to remove the green waxed paper insulation (a) under the coin cell. And
replace the coin cell (b).
cord stow feature of some vacuum cleaners.
Overall, the construction of this unit has been
very well thought-out.
According to the instruction leaflet, we need
to fit a CR2032 type lithium coin cell but mine
came with one already installed. It was just
necessary to pop it out and remove the thin
green insulating disc of waxed paper under-
neath the cell (Figure 6). With the battery
Figure 7: The pokitMeter app, in the Apple App
Store. Here the measured values ​​shown have
three decimal places.
Figure 8: In firmware version 1.2, the display does not go to zero for alternating current. At the top left
the warning of voltages higher than 60 V. Display error: On the right you can see the bottom edge of
all the scale selection labels are truncated.
108 July & August 2022 www.elektormagazine.com
replaced we can go to the Play Store or
App store to locate and install the ‘pokit-
Meter’ app (Figure 7) on our smart device.
Curiously the image displayed on the app
page actually shows a sample measurement
reading of ‘3.222 mV’ but the app will in fact
only display a measurement resolved to two
decimal places.
From now on there were no more big
surprises. With the app running, we can pair
with the Pokit meter and interact with it via
the Bluetooth link. The first thing I noticed
was that the display remains consistently in
landscape format, no matter how you turn
the phone. That makes sense for displaying
waveforms, but for the multimeter option at
least, I would prefer portrait format; it’s easier
to hold a phone in that orientation with one
hand, and it’s almost universal for conven-
tional multimeters.
Flakiness Persists
When trying it out for the first time, it was
noticed that when operated as a multime-
ter, in some measuring ranges, neither open
nor short-circuited terminals produce a clear
reading of zero. Figure 8 shows the screen
for measuring alternating current where the
display is showing 60 µA with open termi-
nals. It looks like the zero adjustment is a bit
buggy when the measurement ranges are
switched. The device does however measure
other values correctly.
Regarding the measurement properties for
this device: Pokit can record the ‘real RMS
value’ of alternating voltages and alternating
currents which is very commendable. After
comparative measurements with calibrated
multimeters, I can confirm that claim is in
fact true!
A bit of playing around showed that when
switching from ‘continuity test’ to another
mode, the beep tone (when the terminals were
short-circuited) was often not switched off. You
had to select the other mode a second time,
then it was quiet and everything was in perfect
order. So not a major issue, just an annoying
carelessness somewhere in the code.
Autoranging is the default setting in multime-
ter mode but you can also select the measur-
ing range manually. When in oscilloscope
mode, the GUI occasionally suffers redraw
errors. The view in Figure 9 shows that the
selected time base indicator sometimes can
hardly be read. Unlike a conventional scope,
the Pokit meter oscilloscope behaves like
a single shot DSO triggered by tapping on
the white button to the right of the screen.
The trace does not free run and continuously
overwrite the screen (or scroll horizontally).
This mode of operation would be good as
an option, but it is the only option and thus
limits its usability as a scope. The more expen-
sive Pokit Pro version is still under develop- Figure 9: Display errors occasionally occur when selecting the scope time base.
ment but will be able to operate in conven-
tional scope mode. Incidentally, you can set a
cursor on the displayed waveform by tapping
it (Figure 10) and the measured amplitude is
displayed - interestingly enough, to a resolu-
tion of three decimal places. A second cursor
can also be positioned. Both should be able to
be moved horizontally with the finger on the
waveform, but I could not get that to work...
another bug?
 
At this point, an update request popped up.
Checking out my unit, I found I was using
hardware version 1.6 running firmware version
1.2. The update offered firmware version 1.5
which I accepted. After updating, the redraw
bug disappeared and the cursor could now
be moved as intended. It was reassuring to
Figure 10: Version 1.5 of the firmware allows use of the cursors.
know that software maintenance is ongoing.
The newer version didn’t cure the beeping
problem but the fault now seems to occur
less often when switching between ranges.

Measurement Accuracy and

Inconsistencies
Pokit should be able to measure voltages and
currents to an accuracy of 1% — regardless
of whether they are AC or DC levels. I can
confirm this by making comparative measure-
ments. At 1 V DC and 5 V DC the error was
≤0.2% in both cases and at 12 V DC it was
even less than 0.1%. The currents readings
also stayed at around 0.2% error. I measured
the frequency response to AC signals. In the
range up to 500 Hz, the error was less than
1%. At 5 kHz, on the other hand, it rose to
around 5%. Measurements in the upper range Figure 11: Spectrum of the mains hum picked up from my index finger. Harmonics at integer multiples
of the audio frequency band cannot really be of 50 Hz can be seen. A warning indicates a value has been clipped but where?

lektor July & August 2022 109


Figure 12: The display of a 50 kHz sine wave is acceptable at a sample rate of 1 MS/s. The display is
switched to full screen mode so none of the controls are visible.
Figure 13: Data logger plot. Seven minutes in the life of my car battery sampled at 1-s intervals.
trusted, but this is not a special drawback of
Pokit alone. You can only expect more consis-
tant values from more professional measur-
ing equipment beyond €1000. All in all, I was
quite impressed by the Pokit’s measurement
accuracy.
 
I can accept that the device lacks many of
the features of a ‘real oscilloscope’ that could
have been implemented in software at no
extra cost. You can’t really expect more from
a device in this price range. The function as
a spectrum analyzer (Figure 11) is also rather
rudimentary and an add-on.
110 July & August 2022 www.elektormagazine.com
 
As I mentioned, the function of particular
interest to me is the data logger. The hardware
technical specification indicates that 8,000
measurements can be recorded. That doesn’t
sound too generous, but the measurement
intervals can be set at: 1/2/3/4/5/10/30 s and
1/5/10 m and 1 h. Crunching the numbers,
this results in a maximum time span ranging
from 8,000 s = 2:13:20 h:m:s (at a one second
sample rate) up to 8,000 h orjust over 333 d
(i.e., about 11 months). The Pokit paperwork
suggests that the span with a 1-s interval
is shorter i.e. 103 m (= 6,180 s) and with a
1-h interval only 8 months (≈ 5,850 h). The
technical spec on the website indicates opera-
tion ‘up to 6 months’ which is most likely a
result of the 6 month maximum battery lifes-
pan. It would certainly inspire more confi-
dence if there were more consistency in the
documentation.
Also in the Elektor Store [2], we correctly
identify the A/D converter in the Pokit as
12 bit, the page integrated in the app under
‘TECH SPECS’ suggests it’s 14 bit — which
it definetely isn’t. This page also contains a
few other anomalies such as the data logger
sample interval is selectable in the range
from ‘1 s — 1 day’ and the maximum log
time is ‘1 year (at 6 hours/sample)’ — none
of which is correct. Anyway the six-month
battery life would certainly put an early halt
to proceedings.
The screenshot in Figure 12 shows the
waveform of a 50 kHz sine wave at the fastest
time base setting; you can see that at least
the specified sample rate of 1 MS/s is correct.
This means that the oscilloscope function is
good to display signals in the entire audio
spectrum. As a ‘real oscilloscope’ it would
need to be at least 10 times faster and have
a number of additional trigger capabilities;
with the Pokit you do have at least a mini
storage scope in your pocket and can monitor
changes in voltages or currents in the sub-HF
range far out in the field far away from any
power supply. That’s the environment in which
it really shines.
For the test I wanted to monitor the voltage
level of my Prius directly on the 12 V battery
every 1 s while the car is in motion. At my
first attempt, Pokit was displaying incorrect
values (battery voltage of 7.5 V). The remedy
was to cancel and start recording again. The
second attempt was successful. The logged
values are again not shown in real time. Only
at the end of the logging period we can see
a waveform made up of the logged values.
This again is less than optimal, a real time
view would be more useful.
 
Figure 13 shows car’s battery voltage level
during the seven-minute drive down a country
lane. The voltage fluctuates between 11.7 V
and 14.55 V. When first connecting Pokit, the
on-board voltage was exactly 12.23 V. At the
beginning you can see a slight drop to 12.175 V
when getting into the car, as the hydraulic
pump kicks in to build up brake pressure; it
draws up to 15 A for a few seconds when the
driver’s door is opened. When starting, the
voltage goes briefly below 12 V when all the
electrics are activated and then the battery
receives a charge. When the battery is full,
charging is terminated and the battery voltage
drops to around 13.4 V. Even if the end-of-
charge voltage is a bit high for my taste -
the curve in Figure 13 is quite informative.
The logger is very well suited to reveal such
insights and is by far accurate enough.
A Handy Solution
With a few reservations about the GUI, I found
the Pokit Meter to be quite usable and well
worth the money — just under €100. Small
inconsistencies in its operation are annoying,
but you can live with them. Version 1.5 of the
GUI is still a little flaky, but the overall concept
is good, and the company is a relatively new
upstart, so we can expect the product to
improve as more updates arrive. I can’t under-
stand why the bug with the beep has not been
resolved, at least with the Apple version of the
app; it should really have been noticed. And
why is it not possible to set the oscilloscope
time base down to a minimum of 20 µs/div?
The end is there at 1 ms per screen = 200 µs/
div. The 20 µs per horizontal division can only
be achieved by zooming out using two-fin-
gers. The hardware has the capability, but it
cannot be configured in the firmware.
A lot of thought has gone into the physical
layout of this unit to make it convenient to
use. Its physical size is just about on the limits
of what you would consider unobtrusive and
pocketable when attached as a fob to a set
of keys.
For me the data logger function is ideal, and and beginners, its capabilities are so wide
I’m sure, as I use it more and more I will come ranging they may not need to buy any other
to appreiciate its other capabilities. A little measuring device at first for low voltage, low
more storage space would be good, but on frequency measurement applications.
the whole I am happy with what it can do. I 210615-01
don’t necessarily need higher sample rates,
as these would probably have a heavy impact
on battery life. In my opinion, Pokit Innova- Questions or Comments?
tions has done a lot of things right with the Do you have any technical questions or
hardware, but I get the impression we are comments prompted by this article? You
working with a beta version of the firmware. can email the editor at editor@elektor.
  com.
Am I looking forward to the Pokit Pro version?
Surprisingly, not so much even though it
promises real-time scope capability and
simultaneous operation of four units on one
smart device. For me the concept of a discreet
key fob that turns your smart device into a
sophisticated piece of test equipment is very
attractive.
 
I can see the Pokit Meter will find a place in
my growing collection of test gear for use in
the lab and especially in the field. For students
RELATED PRODUCTS
> Pokit Meter – Portable Multimeter, Oscilloscope and Logger (SKU 19854)
www.elektor.com/19854

WEB LINKS
[1] Pokit Innovations website: http://www.pokitinnovations.com
[2] The Pokit Meter in the Elektor store: http://www.elektor.com/pokit-meter-portable-multimeter-oscilloscope-and-logger

lektor July & August 2022 111

 store
www.elektor.com

The Elektor Store

Never expensive, always surprising
The Elektor Store has developed from the we simply want to try out. If you have a nice
community store for Elektor’s own products suggestion, we are here (sale@elektor.com).
like books, magazines, kits and modules, into Our main conditions:
a mature webshop that offers great value for never expensive, always surprising!
surprising electronics. We offer the products
that we ourselves are enthusiastic about or that

Miniware DS213 4-ch Mini Oscilloscope

(15 MHz)

Price: €249.00
Member Price: €224.10

www.elektor.com/20206

PiKVM v3 KVM over IP

for Raspberry Pi 4
(pre-assembled)

Price: €319.00
Member Price: €287.10

www.elektor.com/20199

112 July & August 2022 www.elektormagazine.com

Waveshare PicoGo Mobile Robot Ersa i-CON NANO
for Raspberry Pi Pico Soldering Station (80 W)

Price: €67.95
Member Price: €61.16 Price: €329.00
Member Price: €296.10

www.elektor.com/20196 www.elektor.com/20187

RangePi – LoRa USB Dongle

based on RP2040 (EU868)

Andonstar AD203
Handheld
Digital Microscope
with 4“ LCD Screen

Price: €54.95 Price: €69.95

Member Price: €49.46 Member Price: €62.96
www.elektor.com/20156 www.elektor.com/20146

lektor July & August 2022 113

 Hexadoku
Puzzles with an Electronic Touch

Traditionally, the last page of Elektor The Hexadoku puzzle employs numbers in the hexadecimal range 0
through F. In the diagram composed of 16 × 16 boxes, enter numbers
magazine is reserved for our puzzle with an such that all hexadecimal numbers 0 through F (that’s 0-9 and A-F)
electronics slant: welcome to Hexadoku! occur once only in each row, once in each column and in each of the
Find the solution in the gray boxes, submit it 4×4 boxes (marked by the thicker black lines). A number of clues
are given in the puzzle and these determine the start situation.
to us by email, and you automatically enter
the prize draw for one of five Elektor store Correct entries received enter a prize draw. All you need to do is
vouchers. send us the numbers in the gray boxes.

SOLVE HEXADOKU AND WIN! PARTICIPATE!

Correct solutions received from the
entire Elektor readership automati-
cally enter a prize draw for five
Elektor store vouchers worth €50.00
each, which should encourage all
Elektor readers to participate.
Ultimately August 15th, 2022,
supply your name, street address
and the solution (the numbers in
the gray boxes) by email to:
hexadoku@elektor.com

PRIZE WINNERS
The solution of Hexadoku in edition 04-05/2022 (May & June) is: 0FC3D.
Solutions submitted to us before June 15th were entered in a prize draw for 5 Elektor Store Vouchers.
The winners are posted at www.elektormagazine.com/hexadoku.
Congratulations everyone!

The competition is not open to employees of Elektor International Media, its subsidiaries, licensees and/or associated publishing houses.

114 July & August 2022 www.elektormagazine.com

 PROTEUS
DESIGN SUITE
Design Quality Assurance
Constraint Driven
Design
Flexible and scalable
Zone Inspector
rule system
Full support for design Analyze plane coverage and
rule rooms stitching
Manufacturing Grid view of plane
solder mask rules configurations
Live display of
Edit plane settings and
violation areas
draw order

Dedicated
Reporting
Module
Tables automatically
populate with design
Pre-Production data
Checklist
Compliance status for
diff pairs and length
Set of board tests
matched routes
before Gerber Output
Make custom
Includes placement, reports with data
connectivity and object tables
clearance testing
Generate reports
Completely independant from templates
code for clearance checks

info@labcenter.com
+44 (0)1756 753440
electronica fast
forward is back!

New Logo, New Concept

Do
you
the wa
be nt
st to
wi sta see
thi rt- the
nt up bes
he sa to
ele nd f
ctr sca
on le-
ics up
sce s
ne
th ?
b o o
s a t o
s i t u t t
Vi o g e
.
0 t e m
. 4 4 t h
B4 n ow
k

Save the date: 15th - 18th November 2022

Messe Munich - B4.440