Czech Technical University in Prague

Faculty of Nuclear Sciences and Physical

Engineering
Department of Physical Electronics

Design and Construction of Fluxgate

Magnetometer for Space Research
Bachelor's Degree Project

Author: Roman Pavelka

Submitted in: July 2011

Programme: Engineering Informatics

Branch of Study: Information Physics

Supervisor: Ing. Pavel M. Trávní£ek, Ph.D.

Consultant: Ing. Josef Blaºej, Ph.D.

Prohlá²ení

Prohla²uji, ºe jsem svou bakalá°skou práci vypracoval samostatn¥ a pouºil jsem

pouze podklady (literaturu, projekty, SW atd.) uvedené v p°iloºeném seznamu.

V Praze dne .................... ........................................

Roman Pavelka
Title:
Design and Construction of Fluxgate Magnetometer for Space Research
Author: Roman Pavelka
Programme: Engineering Informatics
Branch of Study: Information Physics
Supervisor: Ing. Pavel M. Trávní£ek, Ph.D.
Space Plasma Group
Astronomical Institute of Czech Academy of Sciences
Consultant: Ing. Josef Blaºej, Ph.D.
Faculty of Nuclear Sciences and Physical Engineering
Czech Technical University in Prague
Abstract:

The aim of this bachelor thesis is to summarize the general principle of uxgate
magnetometer with relevant physical and engineering knowledge, give an overview of
current space-born instruments and design and build simple magnetometer capable
of terrestrial measurement of sun-triggered activity of Earth's magnetic eld. The
design accents low cost, circuit's simplicity, digital implementation and usage of
accessible components. Complete description of the manufacturing process is given.

Key words: uxgate, magnetometer

Název práce:
Návrh a konstrukce uxgate magnetometru pro vesmírný výzkum
Autor: Roman Pavelka
Obor: Inºenýrská informatika
Druh práce: Bakalá°ská práce
Vedoucí práce: Ing. Pavel M. Trávní£ek, Ph.D.
Space Plasma Group
Astronomický ústav Akademie v¥d ƒR
Konzultant: Ing. Josef Blaºej, Ph.D.
Fakulta jaderná a fyzikáln¥ inºenýrská
ƒeské vysoké u£ení technické
Abstrakt:

Cílem této bakalá°ské práce je shrnout princip uxgate magnetometru se souvise-

jícími fyzikálními a inºenýrskými znalostmi, provést re²er²i stávajících návrh· pouºí-
vaných na sou£asných výzkumných satelitech a navrhnout a zkonstruovat prototyp
p°ístroje schopného pozemního m¥°ení uktuací zemského magnetického pole zp·-
sobených slune£ní aktivitou. Návrh klade d·raz na nízkou cenu, jednoduchost,
digitální zpracování signálu a vyuºití snadno dostupných komponent. Práce je v
angli£tin¥ a obsahuje kompletní popis postupu výroby.

Klí£ová slova: uxgate, magnetometr

Acknowledgment

I thank my friend Petr Waschinger for helping me to get into the world of mi-
crocontrollers, providing me various electronic material, for moral support and for
our invaluable technical discussions. Also I would like to thank Jakub Kákona for
providing me their great kit MLAB and for sharing his knowledge with others.
I would like to acknowledge Prof. Ing. Pavel Ripka, CSc. for our discussion, which
gave me a lot, and for providing me his comprehensive book Magnetic Sensors and
Magnetometers. I also appreciate valuable discussion with Ing. Jaroslav Pavel
about general electronics design. I thank Pavel Hude£ek for insightful advices and
his idea to optically interface driving electronics. I'm grateful to Mat¥j Laitl for his
invaluable help with GNU/Linux, especially for introducing me Gentoo distribution
and for a plenty of various programming knowledge and tricks, which he taught me.
Also I thank Ing. Josef Voltr, CSc. and Ing. David Vyhlídal for their help with
microcontroller-peripheries interfacing.
I would like to thank Bc. Adam Winsdor for reading the text, drawing gures
4.2, 4.3 and 4.4, taking all used photographies and teaching me some useful image
manipulation tricks. Also I would thank Petr Vála, Bc. Mirek Zima, my father
Ing. Roman Pavelka, PhD. and others who read this text and helped me to correct
mistakes.
I thank my family for all their support and understanding.
Contents

1 Introduction 6
2 Goals 7
3 Physical background 8
3.1 Magnetic eld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 Magnetic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.3 Transformer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4 Principle and basics of uxgate magnetometers 12
4.1 Non-linear transformer . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.2 Dierential measurement . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.3 Feedback measurement . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.4 Materials for uxgate sensors . . . . . . . . . . . . . . . . . . . . . . 15
4.5 Sensor designs and multi-axial measurement . . . . . . . . . . . . . . 16
5 Successful instruments, proposed designs and trends 18
5.1 Ørsted magnetometers . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.2 THEMIS magnetometers . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.3 Other designs and trends . . . . . . . . . . . . . . . . . . . . . . . . . 21
6 Design and prototyping 22
6.1 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
6.2 Prototyping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.3 Control unit and communication interface . . . . . . . . . . . . . . . 23
6.4 Drive electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6.5 Sense electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
6.6 Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
6.7 Firmware, software and operational procedure . . . . . . . . . . . . . 29
7 Testing 32
8 Conclusion 37
Bibliography 38
A Content of enclosed CD-ROM 40
B Used software 41
Chapter 1
Introduction

The rapid development and systematization of electromagnetic theory in the 19th

century (Maxwell, Faraday, Henry and others) revolutionized most elds of human
society. During the 20th century, new military, navigation and research problems
arose and precise and accurate measurement of electromagnetic quantities became
necessary.
The uxgate magnetometer is a device developed for precise vector measurement of
static magnetic elds. It was invented by Aschenbrenner and Goubau in 1936 and
improved during the WWII by Victor Vacquier. It was used for airborne detection
of submarines and exploration of mineral and oil deposits. The earliest investigation
of outer space environment revealed the importance of research of magnetic elds in
the Solar System. Earth's magnetic eld was found to be important for terrestrial
life because it forms a protective bubble with distinct regions, called magnetosphere,
and deects charged particle ux from the Sun and other extraterrestrial sources.
[16] [10]
Further exploration revealed strong coupling of dierent regions of Earth's magneto-
sphere to solar activity. Interesting results can be expected from exploring possible
connection between terrestrial and space weather. Physics of Earth's magnetosphere
became important part of geophysics. Fluxgate magnetometers are very valuable in
this eld because of their high precision, stability and reliability. [16]
This work studies general principle and current technology of uxgate magnetome-
ters. Simple digital device was designed and built to monitor local magnetic eld
variations caused by the solar, human and geological activity. I use available com-
ponents and digital design to achieve construction and calibration simplicity.

6
Chapter 2
Goals

Goals can be specied as

1. to describe the principle of the uxgate magnetometer with relevant physical
and engineering background,
2. to give an overview of successful space-born instruments and modern trends
in the eld of uxgate magnetometers dedicated to space research,
3. to design, construct and calibrate a uxgate magnetometer capable of terres-
trial measurement of variations of Earth's magnetic eld.

7
Chapter 3
Physical background

3.1 Magnetic eld

Magnetic eld B(r) is induced by a charge q moving with velocity v:
µ0 q
B(r) = (v × r), (3.1)
4π r3
where r is a vector pointing from the charge to a point of interest. This eld acts
on moving test charge qT with velocity u by magnetic force
F = qT (u × B). (3.2)
For magnetic eld induced by current I along path l we reformulate the formula
(3.1) to well known Biot-Savart law
dl × R
Z
µ0
B(r) = I , (3.3)
4π l R3

where R = r − l. Expressing the divergence and the curl of the formula (3.1) we
can write laws of magnetism in form
∇ · B = 0, (3.4)
∇ × B = µ0 j. (3.5)
j stands for the overall current density including free currents, bound currents, which
are results of electromagnetic properties of matter, and the Maxwell displacement
current 0 ∂E/∂t. The variable magnetic eld generates an electric eld given by
Faraday law of induction
∂B
∇×E=− . (3.6)
∂t
We deneR the inductance L for a current loop, which induces an overall magnetic
ux Φ = S B · S by a current I
Φ = LI (3.7)

8
For a system of current loops 1, ..., k we dene the inductance matrix Lij to express
a coupling between loops. The ux through the i-th loop is obtained as a sum of
uxes generated by currents in our system
k
(3.8)
X
Φi = Lij Ij .
j=1

Lij are for i = j called self inductances and for i 6= j mutual inductances. Taking
the derivative of (3.8) and the induction law (3.6), we obtain dynamic denition of
inductance, useful for circuit analysis,
k
dIj
(3.9)
X
Ui = − Lij .
j=1
dt

More detailed analysis is given in the publication [17].

3.2 Magnetic materials

Magnetic properties of matter result from both existing and induced bound currents
inside material bodies. This fact can be described by introducing the magnetizing
eld H(r) (generated by external currents jext ) and the magnetization eld M(r)
(generated by bound currents jb ).
∇ × H = jext , ∇ × M = jb . (3.10)
If we neglect the Maxwell displacement current, we can write laws of magnetism as
∇ × H = jext , ∇ · B = 0, B = µ0 (H + M). (3.11)
This simplies magnetic engineering by separating an external eld, which can have
simple description, and a matter response, which is described by a material function
M. The magnetization M is usually a function of the magnetizing eld H and is
useful to introduce the relative permeability µr by relation
B = µ0 µr H (3.12)
and the dierential permeability µd by
dB = µ0 µd dH. (3.13)
Material with weak magnetic interaction has µr ≈ µd ≈ 1.
Ferromagnetism is caused by presence of magnetic domains. Ferromagnetic bod-
ies react on applied magnetizing eld by its amplication with hysteresis and they
exhibit saturation magnetization. Dependence M(H) is linear for zero initial mag-
netization and small applied eld. Initial permeability µd ranges from 102 up to 105 .
Then dierential permeability continuously drops approaching 1 in full saturation.
If we remove the magnetizing eld, some remanent magnetic eld BR will remain
and we need to apply a coercive magnetizing eld HC in opposite direction to remove
it entirely. Example of the ferromagnetic magnetization curve is in gure 3.1. More
detailed overview can be found in the publication [17].

9
 Figure 3.1: Common steel magnetization curves, credit: Wikipedia[18]

3.3 Transformer
We will use introduced theory to explain principle of a transformer. Our system
consists of two windings on a cylindrical core. One winding is called the primary,
has N1 turns and is connected to a resistor R1 and a voltage source U1 (t) in series.
Secondary winding with N2 turns is connected to a resistor R2 with a voltmeter in
parallel. The problem is sketched in gure 3.2.

Figure 3.2: Transformer analysis

We can describe both circuits

dI1 dI2
U1 (t) − R1 I1 − L1 −M = 0, (3.14)
dt dt
dI2 dI1
− L2 −M − R2 I2 = 0, (3.15)
dt dt
where L1 and L2 are self inductances and M is a mutual inductance.

10
For long solenoids with volumes V and no ux leakage, we can approximate [17]
Li = µr µ0 Ni2 V, M = µr µ0 N1 N2 V. (3.16)
Problem is transformed into a set of rst order ordinary dierential equations and
can be solved numerically.
If we assume R2 → +∞ (open circuit), we can set the second current zero and keep
constant the product Uind (t) = −R2 I2
dI1
U1 (t) − R1 I1 − L1 = 0, (3.17)
dt
dI1
M = Uind (t). (3.18)
dt
With solution
M
Uind (t) = (U1 (t) − R1 I1 ). (3.19)
L1
Assuming R1 = 0 and idealized inductances (2.15), we obtain well-known formula
for transformer with open secondary winding
N2
Uind (t) = U1 (t). (3.20)
N1

We assumed the full ux through both windings, but real transformers have some
ux leakage. This is the reason for high permeable cores keeping the most ux in
their volume and thus lowering a transfer loss. Saturation limits maximal power,
which the transformer is capable to eciently transfer. Those and other aspects are
analyzed in the publication [17].

11
Chapter 4
Principle and basics of uxgate
magnetometers

4.1 Non-linear transformer

The uxgate magnetometer is based on transformer eect with nonlinear permeabil-
ity µ(H) of the core. [15] We will analyze the preceding conguration with the core
made by idealized anisotropic material with linear and saturated regions as drawn
in gure 4.1.

Figure 4.1: Transformer analogy

Induced voltage in the sense winding is proportional to derivative of the ux Φ

through the core in longitudinal direction
dΦ ∂B d
U (t) = −N2 ≈ −N2 S = −µ0 N2 S (µr (H)H). (4.1)
dt ∂t dt
The magnetizing eld component H is a sum of the drive (excitation) winding
magnetization HD and the external magnetic eld component Hext
H = HD [I(t)] + Hext . (4.2)

12
Assuming Hext constant, we can rewrite the mechanism to
  
dHD dI dµr 
U (t) ≈ −µ0 N2 S H + µr |H . (4.3)
dI dt dH H
The dependence on the external eld is slightly hidden in permeability and its deriva-
tive parameters.
If we drive the primary winding with a sinusoidal current signal into saturation, we
obtain derivative of induced magnetic ux Φ. Undisturbed output is an odd function
relative to every zero-crossing point. It implies zero second harmonic signal in the
output.
An external magnetizing eld in longitudinal direction disturbs this symmetry be-
cause it shifts the core into the saturation earlier, when we drive it to the same
direction as the external eld, and later for the opposite direction. This brings
a second harmonic signal into the output. It is possible to lter out the second
harmonic, obtain its phase and rectify it to obtain a steady value proportional to
a magnitude of the external eld, but the presence of large original signal complicates
the extraction of measured signal.

4.2 Dierential measurement

Vacquier's design consists of two symmetrical anti-parallel drive windings and com-
mon sense winding. [15] See gure 4.2.
Without an external eld, the ux induced in the rst drive winding has an oppo-
site direction to that induced by the second drive winding with equal magnitudes
for both. Consequently, there is zero signal in the output. Let's drive our cores
periodically deep into saturation. An external magnetizing eld Hext shifts one core
into saturation earlier and other one later than in undisturbed case. This results in
a signal in the sense winding sketched in gure 4.2
Induced voltage in the sense winding with NS turns is proportional to derivative of
both uxes Φ = Φ1 + Φ2
dΦ dB d
U (t) = −NS ≈ −NS S = −µ0 NS S (µr |H1 H1 + µr |H2 H2 ), (4.4)
dt dt dt
where
H1 = HD + Hext , H2 = −HD + Hext , (4.5)
S is sum of cross-sections of both cores and HD is the magnetizing eld induced by
a drive signal.
Expanding the formula (4.4) and assuming Hext constant, we obtain
 
dHD dI dµr |H1 dµr |H2
U (t) ≈ −µ0 NS S H1 + H2 + µr |H1 − µr |H2 , (4.6)
dI dt dH dH
Let's assume µr (H) = µr (−H) (isotropy of the core) and |Hext |  |HD | (this
assumption is valid, when overall ux is not zero). Then we can approximate
 
dµr  dµr 
µr |H1 ≈ µr |HD + Hext , µr |H2 ≈ µr |HD − Hext , (4.7)
dH HD dH HD

13
Figure 4.2: Clockwise: Vacquier's conguration scheme, Φ(I) diagram undisturbed
(dashed) and disturbed (full), disturbed dΦ(I)
dI diagram, Example disturbed waveform

!
d2 µr 
  
dµr |H1 dµr  dµr 
= H1 ≈ + Hext (HD + Hext ) (4.8)
dH dH H1 dH HD dH 2 HD
and
!
d2 µr 
  
dµr |H2 dµr  dµr 
=− H2 ≈ − Hext (−HD + Hext ). (4.9)
dH dH H2 dH HD dH 2 HD

Notice HD ≈ ND I , where ND is number of turns in each drive winding. Neglecting

higher orders gives us
!
d2 µr 
 
dI dµr 
U (t) ≈ −µ0 NS ND S Hext 4 +2 HD . (4.10)
dt dH HD dH 2 HD

First term in parentheses is our signal. For our idealized material and symmetrical
square wave, it is phase synchronized with driving signal and occurs with twice-
frequency of the drive. Second term is more problematical, but simple considerations
for our idealized material shows it is a quick phenomenon, which will disappear after
integration. Further examination is needed for real materials.
To improve sensitivity, we can add more turns to both windings, increase the sensor
cross-section, use faster transition driving signal and high-permeable core with sharp
transition into saturation. It is important to drive the core deep into saturation.

14
The output signal is present when permeability is changing (linear/saturated tran-
sition) and is proportional to the asymmetry introduced by the external eld. The
sign is determined by a current direction, an external eld orientation and by the
core transition in/out of the saturation.
One way of output processing is the second harmonic demodulation. We integrate
the signal only in moments of positive polarity of the output signal for chosen ex-
ternal eld orientation. Measuring electronics needs to be phase synchronized with
driving electronics. This principle is drawn in gure 4.3.

Figure 4.3: Principle of second harmonic demodulation uxgate magnetometer

4.3 Feedback measurement

To increase the range and the linearity we can add the Helmholtz coils pair as
a compensation winding. The signal proportional to the eld measured by the
uxgate is amplied and fed to Helmholtz coils against the external eld. This
forms a negative feedback compensating an external eld. It is simple to measure
the compensating current, which is directly proportional to the external eld. Two
extra windings can be spared by using DC isolated measurement path and feeding
the compensation directly to the sense winding. [15] Such system is sketched in
gure 4.4.

4.4 Materials for uxgate sensors

From preceding facts we can state some requirements for core materials. High per-
meable core magnetizes quickly and for low magnetizing eld is saturated. This
means that transition occurs for lower ratio Hext /HD . Also it lowers required mag-
netization current and thus power consumption of drive electronics.
Low coercive force and low remanence leads to separation of transitions of dierent
polarities.
Other required properties are low Barkhausen noise (caused by discrete nature of
the process), low magnetostriction and low internal stresses to decrease sensor noise.

15
Figure 4.4: Second harmonic demodulation uxgate magnetometer with feedback mea-
surement

Also low structural imperfections, uniform properties, low deformations and low
mechanical stress decrease sensor oset and improve stability.[15]

4.5 Sensor designs and multi-axial measurement

There are other uni-axial designs exploiting similar principles: racetrack, ring core
or Förster's anti-parallel conguration. See gure 4.5.

Figure 4.5: Dierent uni-axial sensors, credit: Fritz Primdahl [14]

16
Bi-axial magnetometer can be implemented by the ring core design with two or-
thogonal sense windings. Tri-axial measurement is more complicated. Multiple ring
cores in an orthogonal layout are used. Figure 4.6 shows example of a successful
design.

Figure 4.6: THEMIS sensor, credit: Auster et al. [1]

Advanced uxgate magnetometer design called compact spherical coil (CSC) is an

extension of this approach. It consists of the tri-axial magnetometer inside a spher-
ical system of compensation (feedback) coils. [12] Photography is in gure 4.7.

Figure 4.7: Øersted Compact spherical coil, credit: Nielsen et al.[12]

17
Chapter 5
Successful instruments, proposed
designs and trends

5.1 Ørsted magnetometers

The Danish satellite Ørsted is dedicated to precise mapping of Earth's magnetic eld
and was launched in 1997. It orbits our planet in height ranging from 600 to 850
km. The scientic payload consists of star camera, particle detector and vector and
scalar magnetometers. These instruments are placed on a deployable boom eight
meters long to avoid interfering with on-board electronics. The Ørsted measurement
revealed accelerating shift of North Earth's magnetic pole. Earth's magnetic eld
reversal is now subject of intensive research and can have a signicant eect on
terrestrial life. [6]
The Overhauser scalar magnetometer is used for scalar calibration of the vector
magnetometer. Advanced vector magnetometer is of the CSC uxgate type. Block
diagram of one channel of the tri-axial uxgate magnetometer is in gure 5.1. Ex-
citation electronics is implemented as full MOSFET H-bridge and utilizes single +5
V supply. The excitation circuit is on gure 5.2.
The sense electronics is implemented as an analog switched integrator. The mea-
surement is feedback based. Figure 5.3 shows the sense electronics.
The output is sampled by 16-bit ADC which limits the resolution to 0.5 nT. Range
of the uxgate magnetometer is ±65 µT. Power consumption is about 1 W. This
device shows high reliability and is still operating from 1999, 6x times exceeding
required operating life. [12]

18
Figure 5.1: One channel of Ørsted uxgate magnetometer electronics,
credit: Nielsen et al.[12]

Figure 5.2: Ørsted drive (excitation) electronics, credit: Nielsen et al.[12]

19
 Figure 5.3: Ørsted sense electronics, credit: Nielsen et al.[12]

5.2 THEMIS magnetometers

THEMIS stands for Time History of Events and Macroscale Interactions during
Substorms. This NASA mission consists of ve satellites on dierent orbits inside
the magnetosphere and around the Moon. It is dedicated to study overall evolution
of space weather. The whole system is designed to measure conditions in the solar
wind, the magnetosheath, the magnetotail and in other parts of the magnetosphere
in between 1RE (Earth radii) and 30RE in various congurations at the same time.
Probes frequently cross radiation belts and all instrumentation works in broad range
of conditions. Very advanced electronic design digitalizes most processes on a eld
programmable array (FPGA). Figure 5.4 shows a block diagram of the system.
Design of the uxgate sensor is based on three concentric orthogonal ring cores
with dierent diameters. Sense windings and feedback windings are separated. See
gure 4.6. Range of uxgate magnetometer is ±25 µT and resolution: 3 pT. Proven
stability is 0.5 nT per half a year. [1]

20
 Figure 5.4: THEMIS magnetometer electronics, credit: Auster et al. [1]

5.3 Other designs and trends

Review of some other ongoing and future missions shows general shift from analog
devices to digital instrumentation. Large FPGA and fast ADC and DAC converters
implement digital signal processing, drive and compensation control, sensors and
electronics temperature control, output data compression, power management, data
interfacing to the spacecraft and other tasks. Mechanisms for on-ight calibration
were developed. [8][5]
There is a general requirement for reduction of power consumption, weight and
dimensions.

21
Chapter 6
Design and prototyping

6.1 Approach
Chosen electronic design exploits digital control of the excitation electronics together
with digital output signal processing. This approach is a compromise between an
analog 2nd harmonic demodulation system and modern digital design implemented
on FPGA, which I am unfortunately not experienced with.
This approach will require digitally interfaced drive electronics, sense amplier, con-
ditioning circuit and fast analog-digital converter (further only ADC ). Simple com-
munication interfaces, as is serial peripheral interface (further only SPI ), are pre-
ferred. Block diagram of this general idea is sketched in gure 6.1.

Digital supply Drive symmetrical supply

ATmega8, 16 MHz EN Drive Sense

POS/NEG Drive control

SPI
Display&UI

ADC ADC
conditioning

Ref. Analog symmetrical supply

Figure 6.1: General idea of digitally controlled 2nd harmonic demodulation uxgate
magnetometer

22
6.2 Prototyping
I used to develop electronic devices by the following procedure. First, all required
abilities and properties of device are written out and a block diagram is sketched.
Then electronics is divided into subsystems with dened tasks. Then again, all
required functions are written out for every subsystem with input and output spec-
ications.
Schematics is designed by hand and the proposed circuit is simulated. I use simple
open source circuit simulator by Paul Falstad [7]. Component values are adjusted
to improve performance and to minimize currents in the system. Then the circuit
is tested on a breadboard and soldered on a prototyping board. SMD components
are preferred because of compactness of the resulting device.
Useful tool is open source electronic construction kit MLAB. It provides reliable and
universal prototyping platform. MLAB kit consists of a base board with 3 mm holes
aligned to 400 mil (0.4 inch, 10.16 mm) grid and set of various modules as power sup-
plies, basic microcontroller circuits of dierent architectures (AVR, PIC and ARM),
their programmers, communication devices, ADC and DAC modules, sensors, hu-
man interface devices and many others. See MLAB homepage for exhausting list
and description of modules [11].

6.3 Control unit and communication interface

I chose 8-bit RISC microcontroller (further only µC) architecture Atmel AVR which
I am experienced with. It executes most instructions in single cycle including mul-
tiplication and there is a rich on-chip peripheral selection. It is simple to develop
a rmware in C programming language. There is a complete open source devel-
opment toolchain consisting of avr-gcc compiler, avr-libc library and avr-binutils
tools. [3] Also there are uisp and avrdude uploaders compatible with wide range of
programming hardware.[13]
ATmega µCs are simple to program directly in application (ISP feature, in-system
programming), they have sucient internal program memory and SRAM and µC
includes on-chip timers, counters and hardware SPI, I2 C and USART communication
devices.
The rst stage of development utilized my own ATmega8 µC prototyping circuit
and a simple homemade programmer. It is described in my tutorial [13] including
installation of the complete AVR development toolchain in GNU/Linux. In later
development stage I moved to MLAB modules ATMEGA801B, basic ATmega8 µC
circuit, and ATprogISPUSB02A USB programmer compatible with AVR STK500v2
interface. Complete documentation is available on MLAB website [11]. µC is clocked
by 16 MHz crystal.
All user interfacing is implemented as a serial communication between the ATmega8
USART (universal synchronous/asynchronous receiver/transmitter) communication
device and a personal computer (further only PC). In the earliest stage of develop-
ment I had been using a simple MAX232 based TTL-RS232 translator and depre-

23
cated PC serial port with baud rate 115.2 kBd. Current development stage utilizes
USB232R01B MLAB module. This module is a basic circuit for the FT232RL
USART-USB translator IC. This device has a native Linux driver translating the
communication to a common serial terminal, which appears as a /dev/ttyUSBX de-
vice le (where X stands for 0, 1, 2...). All communication is simplied to reading
and writing this device le. FT232RL module allows baud rate up to 1 MBd.
It is important to disable character echoing and terminal control signals for full
binary data transfer. This together with baud rate setting can be done on PC side
using GNU stty utility:
stty -F /dev/ttyUSB0 1000000 raw -hupcl
See stty manual for details [9]. µC setting is described in rmware section.
µC, communication and programmer modules are supplied from USB +5 V power
line.
Fluxgate magnetometer performs requested action after receiving a specied byte.
Data exchange is implemented as a binary transfer. The whole operation is described
in the last section of this chapter. Block diagram of implemented system is in gure
6.2. Photography of assembled MLAB board is in gure 6.3.
Interfacing Development
bash scope logger kate nano
avr-binutils, avr-gcc
avrdude
stty /dev/ttyUSB0
/dev/ttyUSB1
1 MBd
GNU/LINUX user space

LINUX kernel FT232RL driver

Personal computer USB0 USB1

USB232R01B ATprogISPUSB02A
ATX supply
+5V RX TX
+/-12V 1 MBd
RX TX AVR ISP 6pin
Analog supply
USART
+5V +/-8V ATMEGA801B
16 MHz
SPI
pos neg PORTC 1 MHz
Drive Sense
Fluxgate sensor 12-bit ADC
250 kS/s

Figure 6.2: System architecture and communication

24
Figure 6.3: Control and communication system with the ADC build from MLAB kit
parts

6.4 Drive electronics

I proposed simple optically-triggered bipolar half bridge with a symmetrical supply
as an alternative to commonly used H-bridge drive. Advantages are simplication of
digital interface, physically smaller electronics and optical isolation from the digital
circuit. The drive electronics scheme is in gure 6.4. It is supplied from the drive
symmetrical supply based on LM7808 /LM7908 pair.

25
 Figure 6.4: Drive electronics

6.5 Sense electronics

The sense electronics consists of a DC blocking passive lter followed by a non-
inverting amplier based on the low noise, low oset OP07C bipolar operational
amplier (further only opAmp). Next stage is a summing amplier. It shifts the
signal zero level to 2.5 V because of single-supply ADC operation. Last section before
ADC is simple a Zener diode clamp, protecting the ADC input against voltages above
5.1 V and under -1 V.
The gain of amplier is set to cover ADC range. This analog path is supplied by
another LM7808/LM7908 pair.
The ADC LTC1864 was chosen because of its 16-bit resolution, 250 kS/s sampling
rate, serial SPI-compatible output and its availability as ADCmonoSPI01B MLAB
module. Disadvantage for this application is single supply operation. The ADC is
powered by LM317T based +5 V supply. The sense scheme is in gure 6.5.

26
 Figure 6.5: Sense electronics

6.6 Sensor
The rst sensor was constructed for testing of electronics. It consists of the iron
wire 1.5 mm in diameter, 45 cm long. Excitation winding was winded along its full
length by approximately 400 turns. Then the wire was bended in its half to form
parallel symmetrical pair of long solenoids connected in sharp knee. Then the sense
(pick-up) coil was winded over the excitation pair. See photography in gure 6.6.
Similar sensor is described in the article [4].

27
Figure 6.6: My sensor

28
6.7 Firmware, software and operational procedure
µC is programmed in the C programming language. Its operation can be described
by following pseudocode:
main() {
set_pins_directions();
USART_init(1 MBd, 8 bits, no parity, 1 stop bit,
asynchronous, USART receive interrupt enable);
SPI_init(Master, 1 Mhz);
set_interrupt();

while (1) {};

}

USART_receive_interrupt_routine() {
clear_interrupt;

if (recieved=='a' )
send_waveform();
else if (recieved=='v' )
send_value();

set_interrupt();
}

send_waveform() {
for(j=0;j<32;j++) { // to charge blocking capacitors
for(i=0;i<128;i++) {
drive(i);
_delay_us(5);
}
}

for(i=0;i<128;i++) {
drive(i);
data[i]=SPI_get_ADC_value();
delay_us(4);
}

drive_off();
USART_transmit(data);
}

send_value() {

29
 for(j=0;j<32;j++) { // to charge blocking capacitors
for(i=0;i<128;i++) {
drive(i);
_delay_us(5);
}
}

for(i=0;i<128;i++) {
drive(i);
if (i<32) // even harmonic demodulation:
result = result + SPI_get_ADC_value();
else if ( (i>=64) && (i<96) )
result = result - SPI_get_ADC_value();
delay_us(4);
}

drive_off();
USART_transmit(result);
}

drive(i) {
if (i<64) {
negative_drive_pin=0;
positive_drive_pin=1;
} else {
positive_drive_pin=0;
negative_drive_pin=1;
}
}

drive_off() {
positive_drive_pin=0;
negative_drive_pin=0;
}
The actual code is included on the enclosed CD-ROM.
Debugging is possible by simple bash commands (don't forget stty setting described
in the communication section)
cat /dev/ttyUSB0 &
echo -n a > /dev/ttyUSB0
echo -n v > /dev/ttyUSB0
This is simple and powerful tool in combination with output redirecting and utilities
as GNU gnuplot, but it shows it is more comfortable to write two graphing PC pro-
grams. Scope and logger are written in C++ using the qT graphical toolkit to direct
plotting of measured data. (The qT is for its simplicity, programming paradigm,

30
documentation and completeness the best graphical API I have ever worked with.)
The scope periodically asks by sending 'a' for a measured waveform and plots it.
Nothing can be simpler:
void Scope::getData() { // Called every 100 ms
char c='a' ;
data=new char[LEN*sizeof(char)];

FILE *serial=fopen("/dev/ttyUSB0" ,"r+" );

fwrite(&c,sizeof(char),1,serial);
fread(data,sizeof(char),LEN,serial);
fclose(serial);

dataAquired=1;
update(); // update the screen
}
The logger acts very similarly. By periodically sending 'v' it plots the evolution
of measured value in time. The complete code of both programs is on enclosed
CD-ROM.

31
Chapter 7
Testing

Procedure begins by setting the serial communication on the PC and switching on

the analog power supply. We can start the scope to see measured waveform before
the demodulation. Gain of sense amplier has to be set to maximize usage of full
ADC range. We see reserves on the following screen-shot. Also there is a strong
feed-through of the excitation signal. Those eects are investigated in the following
chapter.

Figure 7.1: Typical waveform plotted by the scope

We exit the scope and start the logger. The sensor sends now an integral value.
It can vary in dependence on actual setting of the device everywhere in between
±105 . This is the reason for adjustable parameters shif t and divisor. Graphed
y-coordinate is then computed as
input + shif t
y= (7.1)
divisor
We see two sliders, bigger for setting the shif t (oset) and smaller for the divisor.
First we set the shif t to move the signal level to central horizontal line, than lower

32
the divisor, than again adjust the oset and so on, until the divisor is set to minimal
value 1. Result looks as gure 5.2.

Figure 7.2: Adjusted shift and divisor of logger

We can verify the basic function by a common fridge magnet being rotated ca 15 cm
from the sensor head. This is a problematical verication because length of our
sensor is large in comparison to magnet's size and its characteristic distance of eld
variation, but we obtained expected result as seen in gure 5.3.

33
 Figure 7.3: Turning a small magnet about 15 cm from the head of the sensor

We will try another test. We align the sensor in the east-west direction, null the oset
and we let the sensor turn a full circle. We see we need to set the divisor at least 4.
Result can be seen in gure 5.4.

34
 Figure 7.4: The sensor turns one full circle in horizontal plane

First practical experiments revealed an unpleasing fact: the sensor is very sensible
to deformation and stress. Although it was mounted on plastic rule, any movement
of whole setup brings noise to measurement and device parameters are unstable in
time. See gure 7.5 where one horizontal full circle is followed by ve gentle touches
on sensor body. This fact forces us to develop a new sensor.

35
Figure 7.5: The sensor turns one full circle in horizontal plane, then it faces few soft
touches on sensor body

36
Chapter 8
Conclusion

The rst goal of my thesis, to give an explanation of the uxgate magnetometer's

operation, was done. Magnetic engineering basics were introduced and the uxgate
operation was described. The formula (4.10) for symmetrical dierential sensor
was derived. The second harmonic demodulation method and feedback usage were
introduced. Several sensor designs were described.
The second goal, to give an overview of modern uxgate magnetometers dedicated
to space research, was completed as well. I studied uxgate magnetometers used
or proposed for Øersted, THEMIS, Double Star, Bepi Colombo and other missions.
General shift from analog to digital devices was recognized. Parameters and features
of modern space-born instruments were investigated.
The third goal, to design and construct a prototype of the uxgate magnetometer
capable of measuring solar triggered activity of Earth's magnetic eld, is not com-
pletely fullled yet. System architecture was designed. Communication, control and
interface systems were successfully implemented. Simple sense and drive electronics
were designed and implemented.
The sensor was manufactured from available materials. It proves correct function of
all other subsystems, but the sensor instability and sensitivity to mechanical stress
made measurement impossible. This failure was partially corrected by mounting the
sensor on a plastic ruler. It allows some experimentation, but the sensor oset and
gain is instable in time. I observed strong feed-through from the excitation winding
to the sense winding. This points to the sensor asymmetry. This eect has to be
examined.
New sensor has to be more stable. I will try to achieve it by a dierent core material
and a sensor mount design. Also mechanism to adjust a possible asymmetry has to
be considered. The sensor mount has to be designed with the calibration procedure
in mind. Sensor input and output tuning [15] has to be surveyed. Printed circuit
boards for all devices are planned to be designed and manufactured.
It is worth having detailed magnetization curves of used materials. The device for
studying magnetization curves of various materials is planned to be constructed.
Then the formula (4.10) can be further investigated. Work continues to have the
operating device ready in September 2011.

37
Bibliography

[1] AUSTER et al. The THEMIS Fluxgate Magnetometer. Space Sci. Rev. 2008.
[2] AVRDUDE - AVR Downloader/UploaDEr [online]. last revision 8th of January
2010.
URL: <[Link] [cit. 2011-7-16]
[3] AVR Libc Home Page [online]. last revision 16th of February 2010.
URL: <[Link] [cit. 2011-7-16]
[4] BROWN UNIVERSITY. A Simple Fluxgate Magnetometer [online].
last revision 2006. URL: <[Link]
download/attachments/29069/Fluxgate+[Link]?version=
1&modificationDate=1183038899000> [cit. 2011-7-16]
[5] CARR et al. The Double Star magnetic eld investigation: instrument design,
performance and highlights of the rst years observations. Annales Geophysicae.
2005.
[6] DANMARKS METEOROLOGISKE INSTITUT. Øersted mission web page
[online]. last revision 4th of July 2004.
URL: <[Link] [cit. 2011-6-3]
[7] FALSTAD, Paul. Math, Physics, and Engineering Applets [online]. last revision
date not available.
URL: <[Link] [cit. 2011-7-16]
[8] GLASSMEIER et al. The Fluxgate magnetometer of the BepiColombo Mercury
Planetary Orbiter. Planet. Space Sci. 2008.
[9] MACKENZIE, David stty man page [online]. last revision January 2011.
URL: <[Link] [cit. 2011-7-16]
[10] MAUGH, Thomas. Victor Vacquier Sr. dies at 101; geophysicist was a
master of magnetics. [online]. Los Angeles Times, January 24, 2009. URL:
<[Link]
0,[Link]> [cit. 2011-6-3]
[11] MLAB homepage [online]. last revision 16th of July 2011.
URL: <[Link] [cit. 2011-7-16]
[12] NIELSEN et al. Development, construction and analysis of the 'Øersted' ux-
gate magnetometer. Meas. Sci. Technol. 1995

38
[13] PAVELKA, Roman. Atmel AVR development on GNU/Linux [online]. last
revision 5th December 2010.
URL: <[Link]
[cit. 2011-7-16]
[14] PRIMDAHL, Fritz. The Fluxgate Mechanism, Part I: The Gating Curves of
Parallel and Orthogonal Fluxgates. IEEE Trans. Magn. 1970
[15] RIPKA, Pavel, Magnetic Sensors and Magnetometers. Artech House Publish-
ers. ISBN-10: 1580530575. 494 p.
[16] SNARE, Robert. A History of Vector Magnetometry in Space [online]. last
revision date not available.
URL:<[Link]
[Link]> [cit. 2011-6-3]
[17] ’TOLL, Ivan. Elekt°ina a magnetismus. 2nd edition. Nakladatelství ƒVUT.
2003. 215 p. ISBN 80-01-02693-0.
[18] WIKIMEDIA COMMONS. B-H Loop [online]. last revision 20th May of 2006.
URL:
<[Link] [cit. 2011-6-3]

39
Appendix A
Content of enclosed CD-ROM

• rmware Directory with the C code and Makele of microcontroller rmware

• logger Directory with the C++ code and Makele of the logger program
• schematics Directory with all used schematics
• scope Directory with the C++ code and Makele of the scope program
• [Link] This thesis in PDF format
• [Link] This thesis in PS format
• stty_setting.txt Text le with a bash command to serial port setting

40
Appendix B
Used software

Development
• bash GNU Bourne-Again SHell
• nano Nano's ANOther editor, an enhanced free Pico clone
• kate Advanced text editor for KDE
• make GNU make utility to maintain groups of programs
• avr-gcc GNU C Compiler for AVR microcontroller
• avrdude AVR Downloader/UploaDEr
• gnuplot An interactive plotting program
• kicad Open source software suite for electronic design automation
• CicuitSimulator Paul Falstadt's circuit simulation applet
• Qt cross-platform application framework

Publication
• nano Nano's ANOther editor, an enhanced free Pico clone
• kate Advanced text editor for KDE
• LATEX Structured text formatting and typesetting system
• pdatex PDF output front-end for LATEX
• gimp GNU Image Manipulation Program
• pygments Python-based syntax highlighter

41