Department of Computer Science, Electrical and Space Engineering
Electromagnetic compatibility (EMC) is an essential part in today’s society and there are more products around us that emit electromagnetic waves than ever before. To make sure that all these products function properly under all circumstances EMC testing is needed.
One test that is conducted is radiated immunity testing. A susceptibility antenna is needed to perform immunity testing. This thesis aims to show that a tuneable antenna could be used for immunity testing in the frequency band 26-150 MHz and in the future replace the current antenna, which is not tuneable, used at SAAB Support and Services EMC in ¨ Ostersund.
A simulation program called EZNEC+ was used to simulate different antennas that were
tested in the semi-anechoic chamber at SAAB. Two antenna types showed better efficiency
and reached lower in frequency than SAAB’s current antenna. These antennas were a bowtie
antenna and an x-shaped antenna, both extending in only two spacial directions instead of
the normal three. Their drawback was a less uniform E-field at the lowest frequencies, where
the E-field was much stronger to the sides compared to straight in front of the antenna. This
results in a narrow lobe width, but if this drawback could be accepted or mitigated both
antennas are possible replacements for the current antenna at SAAB and should be further
Abbreviations and Nomenclature 4
1 Introduction 5
1.1 Scope . . . . 5
1.2 Requirements . . . . 6
2 Theory 7 2.1 Antennas . . . . 7
2.1.1 Design Parameters . . . . 7
2.1.2 Near and Far Field . . . . 8
2.1.3 Balun . . . . 9
2.2 EMC . . . 10
2.2.1 Common EMC Tests . . . 10
2.2.2 Radiated Immunity Testing . . . 10
2.2.3 Anechoic Chamber . . . 11
3 Method 12 3.1 Studied Antennas . . . 12
3.2 FSA . . . 13
3.3 Simulation . . . 14
3.3.1 EZNEC+v6.0 . . . 14
3.3.2 Simulation Environment . . . 14
3.3.3 Electric Field . . . 16
3.4 Measurement . . . 16
3.4.1 Equipment . . . 17
3.4.2 Measurement Set-up . . . 17
3.4.3 Antenna Set-up . . . 19
3.4.4 Mimic Tuneability . . . 22
3.4.5 Scaling the E-field . . . 24
3.5 Constant Variables . . . 25
4 Result 26 4.1 Simulations . . . 26
4.1.1 VSWR and Resonance Frequency . . . 26
4.1.2 Homogeneity . . . 29
4.1.3 E-field Strength . . . 29
4.2 Measurements . . . 30
4.2.1 E-field strength . . . 30
4.2.2 Homogeneity . . . 32
4.2.3 VSWR . . . 34
5 Conclusions and Future Work 36
5.1 VSWR . . . 36
5.2 Homogeneity . . . 36
5.3 E-field Strength . . . 37
5.4 Comparison . . . 38
5.5 Conclusion . . . 39
5.6 Future Work . . . 39
5.6.1 Simulations . . . 39
5.6.2 Measurements . . . 39
5.6.3 Constructing the Antenna . . . 40
References 41 A Division of Labour 42 A.1 Lucas Svensson . . . 42
A.1.1 Responsibilities . . . 42
A.1.2 Problem Solving . . . 42
A.1.3 Main Author . . . 43
A.2 Gustaf Ljungn´ e . . . 43
A.2.1 Responsibilities . . . 43
A.2.2 Problem Solving . . . 44
A.2.3 Main Author . . . 44
B Equipment Information 45 B.1 Immunity Testing . . . 45
C Extra Results 46 C.1 Near field distance . . . 46
This master thesis was performed as a collaboration between Saab Support and Services EMC Laboratory in ¨ Ostersund and Lule˚ a University of Technology. Electromagnetic Waves (EMW) are all around us in our daily lives. These waves are created by electrical current and magnetism, and usually interact harmlessly with their environment. They can however, affect electronics with the results of decreased performance, a non-working product, or even a life-threatening hazard (page 3). This is why the study of Electromagnetic compatibility (EMC) is of major importance during the design of products containing electronic systems.
To be able to test some of a products’ EMC characteristics, antennas are needed. Both for receiving EMW from and emitting EMW to the test object. The rules regarding the EMC characteristics of products are set by different authorities and are in the European Union set by directive 2014/30/EU .
The goal of this master thesis was to study and propose a suitable antenna that could be used to perform radiating immunity tests (see Chapter 2.2.1) in the frequency band 26 - 150 MHz by physically changing the size of the antenna and thereby tune its resonance frequency. The antenna should also function better than the current antenna used at SAAB. The antenna is named S12017-02b but will be denoted as FSA in this report, see Figure 1.1.1. More information concerning the FSA will be presented in Chapter 3.2.
Figure 1.1.1: The FSA in horizontal orientation (human for scale).
Antennas for EMC measure- ments are traditionally con- structed to have a very broad bandwidth and thus not be optimized for each frequency.
Antennas that changes size and tunes itself to match the frequency does not ex- ist on the market as of today and very little re- search has been done in this area.
To summarize, the overall goal
with this thesis is to prove that
the concept of a tuneable an-
tenna can work better than, or
equally as good as the already
realized antenna, the FSA.
A list of requirements were put on the antenna, as seen in Table 1.2.1.
Table 1.2.1: Antenna requirements.
Design Parameter Requirement Level Voltage Standing
Wave Ratio (VSWR) between 26-150 MHz
Voltage at feed point <5 kV at 10 kW input power. Note: Requirement for fin- ished antenna, not yet applicable for the antennas in this thesis.
Current No limit
Electric Field Strength 200 V/m at 1 m distance with 5 kW input power. Higher in- put power (up to 10 kW) could be used if an antenna showed good results just below 200 V/m.
The difference between the center point field strength and the field strength at any given point away from the center should be at most ± 3 dB.
Effect endurance 10 kW, note: this will not be taken into consideration in this thesis.
Size Smaller than the theoretical box in Figure 3.3.1. That is, smaller than 3.64x3.64x2 m. (Height x Width x Depth) Frequency range Primary 26 - 150 MHz.
Note: Mainly the horizontal cases were simulated and tested. This was done to save time and it was deemed un- necessary to test both polarizations before making sure one polarization worked well.
Tuneable Be able to change size and thereby change its resonance fre-
This following chapter explains how an antenna works and also cover some basics concerning EMC.
Antennas are devices that interact with the EMW in a medium (usually air or vacuum), ei- ther transmitting or receiving signals (page 1). The antenna translates the EMW to electric currents or the other way round. An antenna consists of one or several electrically connected elements that is optimized to receive or transmit at different frequencies. The most common parameters that affect how well an antenna will work for its intended purpose are frequency, gain, bandwidth, impedance, and polarization .
In an antenna an electric current excite the elements and the change in potential in turn creates EMW, as seen in Figure 2.1.1. Antennas can be constructed in many ways; with reflectors, in a log-periodic fashion, like micro strips, and many others.
Figure 2.1.1: A simple dipole transmitting .
2.1.1 Design Parameters
A few concepts must be taken into account to get an antenna to work well at a desired
frequency, these will be discussed in the following section. The design parameters that have
been under consideration during the thesis work are; resonance frequency, Voltage Standing
Wave Ratio (VSWR), electric field strength, homogeneity, the shape of the created electric
field, the size of the antenna, and how easily it could be manufactured.
The resonance frequency is the frequency where an antenna radiates (or receives) an electric field most efficiently. It is the impedance, and thereby also the electrical length of the antenna that determines where the resonance frequency is. The wavelength can be calculated using Equation 1. Where λ is the wavelength, c is the speed of light in vacuum, and f is the frequency. (page 260).
λ = c
To get an antenna to radiate efficiently at a frequency f, the electrical length has to be, in most cases, half a wavelength. Resonance frequency ties into VSWR; an antenna that tries to radiate outside its resonance will have a high VSWR. The VSWR itself is calculated by Equation 2 (page 518). Gamma is the reflection coefficient known as S11 or Γ
. VSWR is used to measure how much of the input power that is not transmitted from the source to the antenna but is reflected back along the transmission line. The aim is for the VSWR to be as low as possible with 1 being the lowest and the ideal case.
V SW R = 1 + |Γ
1 − |Γ
A high VSWR is due to poor impedance matching, i.e. the difference between the antenna’s impedance and the transmission line’s impedance is large. Γ
is calculated with Equation 3 (page 61). It shows that with more equal impedances, less power will be reflected and therefore lost. The two impedances in Equation 3, Z
, represents the impedance in the transmission line and the antenna respectively. Z
is usually 50 Ω , so Z
should be a value close to 50 Ω to make an antenna radiate efficiently.
The shape of the electric field, its strength, and its homogeneity are dependent on the shape of the antenna radiating it.
2.1.2 Near and Far Field
An antenna’s radiated field is generally said to have two parts, the near field and the far
field. In the far field the electric- and magnetic fields are orthogonal and closely related to
each other. The rate of which the far field decreases is the inverse of the distance (1/R) from
the source. In the near field the electric- and magnetic fields can exist independently of each
other and their relation can be highly complex making it often unknown (page 150). The
fall off of the field is also faster, decreasing with 1/R
, or even faster. The different fall
offs of the far field and the near field derives from what component of the field equations that
are the most significant (pages 145-155).
There is no sharp edge between the two field regions and sometimes the near field is split into two regions (page 31). Even so, there are two equations that describes where the far field starts, Equations 4 and 5 (page 269).
d > λ
d > 2D
Where λ is the wavelength, d is the distance from the antenna, and D is the maximum dimension of the antenna. Using these two equations with the antennas in this thesis shows that all measurements were performed in the near field, see Appendix C.1 for calculations.
A coaxial cable is often used to feed an antenna. The cable consist of two parts, an inner and an outer conductor. The inner conductor is the copper core seen in Figure 2.1.2 and the outer conductor is the shield. These two different conductors are connected to one side of the antenna each. In the simplest case, one antenna element each. A part of the current that runs on the inside of the shield can, instead of going out to the element, run back along the outside of the shield. This means that the antenna is unbalanced and this current is unwanted. A balun is used to create balance between the two conductors (balun = bal ance to unbalance 
(page 521)). The balun is made to remove the current that moves in the shield, this is done by increasing the impedance of the shield to such a high value that the current is choked 
(page 521). A ferrite core is most commonly used to choke the current.
Figure 2.1.2: A coaxial cable. 
EMC is the study of how to protect a product from an electromagnetic environment and as well as protecting said environment from electromagnetic pollution from the product  (page 3).
2.2.1 Common EMC Tests
Two of the concepts that are studied concerning EMC are emission and susceptibility (immu- nity) (page 10). To better understand these two concepts let us introduce two definitions:
source and victim. The source emits EMW at certain frequencies. These EMW are measured and compared with requirements in a standard to see if the electromagnetic field levels are too high, this is called emission testing. Susceptibility testing is done the other way round. The source is now the victim and is irradiated by an immunity testing antenna. This is done to
Figure 2.2.1: Radiated immunity testing set up .
see if the victim behaves normally within cer- tain limits stated in standards. Measure- ments are taken according to standards so it is possible to compare different measure- ments taken at different places with each other. Some standards are but not limited to EN/IEC 61000-4-3 2007 or MIL STD 461 (pages 89,115).
2.2.2 Radiated Immunity Testing
The MIL-STD-461G method RS103 states
that all products shall be able to function
normally when exposed to EMW in the fre-
quency range of 2 MHz - 40 GHz. A gen-
eral setup of a radiated immunity test con-
sists of an anechoic chamber, an Equipment
Under Test (EUT), an antenna, and some
auxiliary equipment, see Figure 2.2.1. The
antenna irradiates the EUT at different fre-
quencies and the EUT is observed for any
unusual behaviour that might occur.
2.2.3 Anechoic Chamber
Many EMC tests are, and all tests in this thesis were, performed inside an anechoic chamber.
It is a room shielded from all outside interference caused by electromagnetic waves and that filters all signals that enter the chamber (page 201). The room is usually made of steel with dampening material (absorbers) on the walls, ceiling and possibly the floor to stop or reduce the reflections of the electromagnetic waves inside the chamber. Otherwise these reflections would cause positive and negative interference and thus produce peaks and nulls in the field which would make it difficult to achieve a uniform E-field (page 169).
The test chamber at SAAB in ¨ Ostersund is a semi-anechoic chamber with walls and ceil- ing damped by pyramidal foam, as seen in Figure 2.2.2. By SAAB’s collected experience and measurements of the Normalized Site Attenuation (NSA), done according to CISPR 16-1-4 , the chamber works well above 100 MHz. So for a large portion of the tested frequency span the chamber’s performance is not ideal. At lower frequencies the absorbers are too small to fully absorb the EMW:s and reflections will occur.
Figure 2.2.2: SAAB’s anechoic chamber.
Simulations were done in EZNEC+ v6.0  and the majority of all tests were performed in the semi-anechoic chamber at SAAB in ¨ Ostersund.
3.1 Studied Antennas
Several antennas were studied in this thesis, all with a simple geometric design to make tuneability achievable. That is, a simple mechanical system could be set up to change the size of the tested antenna so that the resonance frequency changed as well. The tested shapes can be seen in Figure 3.1.1, antennas A-E are seen from above, and antennas F-G are seen from behind. The antennas were constructed in two dimensions so from other angles they would be seen as lines. The name and description of the antennas in Figure 3.1.1 can be seen in Table 3.1.1.
Figure 3.1.1: The examined antennas.
Table 3.1.1: Examined Antennas.
Antenna Figure 3.1.1
A The simplest kind of dipole antenna and can be used as a reference value.
Folded Vee B A vee antenna that folds and with 5-20 cm distance runs back parallel to itself, a way to create long elements while still staying inside the designated area (explained in Chapter 3.3.2.
Bend C The longest the elements can be inside the designated area (explained in Chapter 3.3.2) without being folded or going back and forth.
Half Circle D Long elements without sharp corners as in B and C.
Inverted Half Circle
E Long elements without sharp corners but also the same geometric shape as the FSA antenna.
Bowtie F Two mirrored triangles with an angle of 45
between the angled elements. That is, the angle is between the two elements that connects with the vertical element.
Cross G Same properties as the bowtie but without the two ver- tical elements.
Figure 3.2.1: The FSA in vertical orientation.
The antenna in this thesis that is called FSA
is an antenna constructed by the German
company FS Antennentechnik . The offi-
cial name of the antenna is S12017-02b and
it can be seen in Figure 3.2.1. This antenna
is used at SAAB but have some disadvan-
tages; it is a large antenna that takes up too
much space, it does not work well at the low-
est part of the frequency band (26-150MHz),
and its feed point can not handle 10 kW in-
put power. Due to its size, at around 80 MHz
a whole wavelength can fit inside the antenna
and this causes a voltage maximum to appear
at the feed point. Using high power this max-
imum can amount to several thousand volts
which can result in destroyed electronics.
The pilot study conducted as a preparation for this thesis concluded that EZNEC+ v6.0 and MATLAB would be sufficient to simulate, process, and plot the data.
EZNEC is the most suggested simulation tool to use concerning antenna design in ham radio (amateur radio) communities. What mainly makes EZNEC more viable than its competitors is its easy to use Graphical User Interface (GUI) and its pedagogic user manual. EZNEC utilizes the calculating software Numerical Electromagnetics Code (NEC) and EZNEC itself is only a GUI that runs NEC.(page 3)
3.3.2 Simulation Environment
To start with, a size limiting box was defined in the semi-anechoic chamber, see Figure 3.3.1.
Figure 3.3.1: Size restrictions in the anechoic chamber.
The requirements put on the size of the antenna came from the size of the semi-anechoic chamber with a large EUT inside it. The volume dedicated to the antenna was decided to be 3.64 m in height, 3.64 m in width and 2.00 m in depth at a maximum, see Figure 3.3.1.
Since the antenna needs to be turned 90
for polarization the height and width limits are the same. The 3.64 m is derived from the distance from the floor to the peaks of the absorbers in the ceiling being at a height of 4.24 m, and a 0.3 m margin both ways (4.24-0.3x2=3.64).
The depth restriction is not as strict but came from the experience of using the FSA with a depth just above 2 m and the wish to not have it deeper than that. All simulated antenna types fits in this box of the size 3.64x3.64x2.00 m. Note that this only included the actual antenna and not the antenna mount or any auxiliary equipment.
In all simulations a perfect ground were used, that is, a perfectly conducting infinite ground
plane. This is not as far from reality as it sounds when considering a semi-anechoic chamber,
which will absorb the EMW to imitate infinite space. The antenna elements in the simula-
tions were ideal and lossless. But simulations with elements made out of copper and aluminum
showed that this amounted in little difference in the results compared to an ideal case. All
losses between the amplifier and the antenna were in the simulation concentrated in one 100
m long transmission line with 1 dB attenuation per 100 m. This 1 dB loss was estimated from
SAAB’s internal attenuation tests for some of the cables that were used in the measurements.
3.3.3 Electric Field
To save calculation time only half the Electric field were simulated, the symmetry of the antenna would make the left and right sides of the center identical. So, looking at the antenna from behind, only the left part of the antenna’s E-field was plotted in Figure 3.3.2. The dot in the figure shows where the E-field has dropped with 3 dB compared to the value at the center.
Figure 3.3.2: Horizontal plane of a straight dipole at 134 MHz.
To understand Figure 3.3.2 better, refer to the coordinate system. The x-coordinate is the length of the chamber, i.e. the distance in front of the antenna. The x-direction is also where the depth of the antenna varies. The y-coordinate is the lateral distance measured from the center of the antennas. And finally the z-coordinate represents representing the strength of the electric field.
To measure the electric field as well as the forward and reflected power, a field probe  and
SAAB’s in house equipment were used. This chapter will also cover how the measurements
of the tested antennas was set-up.
SAAB uses a software called RSUS  that controls equipment used for EMC testing via General Purpose Interface Bus (GPIB) and RS232. Figure 3.4.1 depicts how the computer running RSUS gives input values to the signal generator. The power meter checks the for- ward/reflected power and sends it to the computer. Inside the semi-anechoic chamber is the field probe measuring the strength of the E-field.
Figure 3.4.1: Schematic of SAAB’s anechoic chamber with auxiliary equipment.
See Appendix B and Table B.1.1 for more detailed information concerning Figure 3.4.1 and the equipment used.
3.4.2 Measurement Set-up
The measured positions are found in Figure 3.4.2. The field probe started at position 1, one meter in front of the antenna, and measured the E-field. It was then moved one step at a time up until position 9. The antenna was now moved backwards to a two meter distance and the field probe cycled through the positions again, which now are named 10-18 (because of the increased distance to the antenna). This was done again at three meters distance, creating positions 19-27. The total number of measured points thus became 27. The feed point (the middle of the antenna) was always located right in front of position 7. So, facing the field probe only the right part of the antenna was measured (this is due to symmetry).
This also means that only the fourth quadrant of a XY-coordinate system was measured. See
Figure 3.4.2a to easier follow the numbering system of the measurement points. Power to the
antenna was regulated to always be 10 W.
Note that in Figure 3.4.2b, the small movable wall with absorbers standing perpendicular with the actual wall was not present during the tests (compare with Figure 3.2.1). All 27 positions were only measured for the FSA and for most other antennas only position 1, 3, 7, and 9 were measured.
(a) Grid dimensions.
(b) Actual setup.
Figure 3.4.2: The positions used during the measurements.
After placing the electric field probe at a measuring position in the foam wall in Figure 3.4.2b, a predetermined frequency sweep was run. The first frequency was 20 MHz and the following values were incremented with 4 % until the last frequency which was 150 MHz. This makes the total number of frequencies measured 53.
3.4.3 Antenna Set-up
The various antennas were constructed of one of two types of materials. Referring to Figure 3.1.1, A-F used a silver plated flattened copper braid cable shield ( = 6mm) and the cross used a 4043 aluminum welding rod ( = 3mm), see Figure 3.4.3.
Figure 3.4.3: Cable shield and welding rod.
The dipoles in Figure 3.4.4a have their braided cable shield run through plastic tubes con- figured to whatever shape the dipole needed to be (A-E in Figure 3.1.1). The bowtie used the same cable shield but were set up with duct tape instead of plastic tubes. The cross, as seen in Figure 3.4.4b, were set up with welding rods and could therefore not use the plastic tube system but rather had to be cut to the correct length between measurements. The white arrows indicate which direction the antenna would retract towards the feed point, which is indicated with an red arrow. The change of size was made between tests to get measurements for many different lengths of the same antenna. All these measurements would together show the results from the antenna if it was tuneable. The antenna was fed through a coaxial cable that was brought up to the height of the antenna (2.1 m) and then kept at this height moving back approximately 2 m to avoid interference.
(a) General setup of the dipoles. (b) Setup for the cross and bowtie.
Figure 3.4.4: Antenna mountings.
Figure 3.4.5 explains the feed setup. The blue arrow points to the incoming coaxial cable, white arrow points at the balun used, yellow arrow points at the spool that had the antenna wire, and finally the red arrow points to the actual feed point. Figures 3.4.5a and 3.4.5b are close ups of the feed point which consists of a copper plate, coaxial connector, and two clamps to connect to the antenna wires. Connected to the balun is the feed point, and under it is the spool. The spool’s purpose was twofold, firstly to simulate a tuneable antenna. Such an antenna has to have its antenna elements winded up around something and will be a chunk of metal just hanging under the antenna. The second purpose was to speed up the change between different antennas and sizes.
(a) Front of the copper plate. (b) Back of the copper plate.
(c) Whole feed setup.
Figure 3.4.5: All parts of the antenna feed.
The used balun consisted of a two part coiled coaxial cable winded around two ferrite cores, see Figure 3.4.6. The insertion loss of the coaxial when measured with a network analyzer became 0.4 dB. Adding the ferrite cores dampened the shield currents with 15-20 dB on a frequency band between 20-200 MHz.
Figure 3.4.6: The balun used.
3.4.4 Mimic Tuneability
Figure 3.4.7 displays how the data is plotted. The red line shows the FSA’s electric field
strength and the other line is the tested antenna. In many results the FSA is used as a
reference to the measurements taken inside the anechoic chamber. Similar plots were also
made to compare homogeneity and VSWR. The positions mentioned in figures containing
E-fields are the positions displayed in Figure 3.4.2.
Figure 3.4.7: Example of plotted field strength data.
To mimic an antenna with the ability to con- tinuously change size, several sizes of each antenna types were tested. These different sizes perform better at different frequencies.
Looking at the results for all sizes of one type of antenna and plotting the highest E-field at each frequency produces a combined plot.
With infinitely many sizes this would pro- duce the same plot as a continuously chang- ing antenna. Since only a discrete number of tests could be performed certain dips only exist because the size that performs best at that frequency was not tested. An example of this kind of plot is Figure 3.4.8 where the thin lines are the individual results for each size and the thick line is the combined re- sult, i.e. a tuneable antenna. The vertical lines are the different resonant frequencies of the various antennas used. Plots used in the Result chapter will only show the thicker line and the vertical resonance frequencies.
Figure 3.4.8: Maximal achieved E-field across different sized dipoles.
3.4.5 Scaling the E-field
The examined antennas could not be tested at full power (5 kW, according to Table 1.2.1) but instead were tested with 10 W. To get an idea of what the E-field strength would be at 5kW the plots were scaled. Using Equation 6 (p.519) for two different powers and calculating their ratio results in Equation 7. Where E is the E-field, P
is the equivalent power and R is the distance from the source. Equation 6 are designed for far field calculations, but in the near field the concept would be the same. How the equation depends on the distance might change to R
but they would still cancel each other out in the same way as below. The same would be true for a different scalar and for the equivalent power so the reasoning below stands in the near field as well.
√ 30 ∗ P
√ 30 ∗ P
√ 30 ∗ P
Since the distance is the same in both cases, R
they cancel out. The antennas are exactly the same with their input power being the only change which results in the P
ratio becoming only the input power ratio. Lastly 30/30 cancel out and what is left is Equation 8.
Calculating the scale between 10 W and 5kW with Equation 9 would result in the E-field ratio being:
r 5 ∗ 10
10 ≈ 22.36 (9)
This scalar, 22.36, will approximate the E-fields’ strengths produced by the antennas using 5
kW input power, by multiplying it with the E-field’ strength from 10 W input power. This
would of course not represent the actual results exactly at that power. The extra heating
of the antenna and cables that would occur due to this power increase was not accounted
for. The resistance of materials depends on its temperature so in the cables and the antenna,
which are conductors, the increase in temperature would result in a higher resistance (pages
43, 226). A higher resistance would mean more losses as the current runs through it but it will
also change the impedance matching of the antenna and the feed cable. This change could
both increase or decrease the VSWR depending on how the impedances were matched from
the beginning. Anyhow both of these effects would be minor and were therefore ignored.
3.5 Constant Variables
Tables 3.5.1 and 3.5.2 show a summary of variables held constant. These variables were held constant so there were as little difference as possible between different simulations and different measurements.
Table 3.5.1: Summary of simulated variables.
Variable [unit] Explanation
Diameter of antenna elements [mm]
3 mm, to match the measurements Antenna height over ground
Was set to 2.1 m due to the size of the anechoic chamber
Ground plane ”Perfect” ground, infinitely long ground plane with perfect con- ductibility.
Measure distance [m] Set to take values concerning the electric field at 1, 2, and 3 meters away from the antenna.
Measurement grid [m] See Figure 3.4.2.
Input power [W] To match the requirements in Table 1.2.1 concerning power, the initial simulations were run at 5 kW (to see which antennas that would reach the requirement of 200 V/m stated in Table 1.2.1).
Later, and ALL figures in this report, were run with 10 W output power to match the power in the tests.
Table 3.5.2: Summary of measurement variables.
Diameter of antenna [mm] 3 mm or 6mm, depending on whether the 4043 aluminum welding rod or the copper cable shield were used.
Height over ground plane [m]
See Table 3.5.1 Measure distance [m] See Table 3.5.1 Measurement grid [m] See Table 3.5.1
Input Power [W] 10 W
E-field scalar 22.36
By first going through the simulations and then the measurements it will be clear which antennas could not reach the requirements and therefore could be discarded.
Simulations for the VSWR and an antennas resonance frequency were found to be fairly accurate while the simulated E-field strength often showed different results compared to the measurements. From the VSWR, one antenna could be discarded immediately and the effect of capacitance hats on the cross will be shown through simulations.
4.1.1 VSWR and Resonance Frequency
Changing an antenna’s size and for each size plot the lowest VSWR against the frequency were it was taken, this resulted in Figures 4.1.1 and 4.1.2. The VSWR requirement set was
<5 so the inverted half-circle could immediately be discarded.
Figure 4.1.1: Simulated VSWR depending on the resonance frequency.
The left most data point for each data line in Figure 4.1.2 corresponds to the maximum size of that antenna. Thus giving a good indication of how low in frequency an antenna type can radiate effectively. With the cross antenna and the straight dipole being the only ones not reaching close to or below the required 26 MHz.
Figure 4.1.2: Simulated VSWR depending on the resonance frequency, zoomed in.
To reach a lower frequency with the cross, capacitance hats were added. Two kinds of simu- lated hats can be seen in Figure 4.1.3 and their affect on the cross are documented in Tables 4.1.1 and 4.1.2.
(a) The normal cross antenna. (b) Cross antenna with capacitance hats made of
(c) Cross antenna with capacitance hats made of wires with connected ends to closer simulate a solid surface.
Figure 4.1.3: Models used in the simulations.
Figure 4.1.3a shows the plain cross antenna and in Figure 4.1.3b ten elements with a length of 30 cm were added to each end to create four capacitance hats with radii of 30 cm. In Figure 4.1.3c these elements are connected to closer reassemble a solid surface. The two tables below show that both kind of hats lower the resonance frequency significantly. With the solid hats reaching lower than the wire hats and with larger hats reaching lower than smaller ones.
Table 4.1.1: The affect of capacitance hats with radius=30 cm at lower frequencies.
Approximate width Resonance Frequency [MHz] VSWR at Resonance Frequency Cross Wire hat Solid hat Cross Wire hat Solid hat
3.6 m 33.1 22.3 20.4 1.07 1.74 2.53
3.2 m 38.1 24.3 22.1 1.17 1.64 1.96
2.8 m 43.4 26.5 24.2 1.22 1.22 1.88
Table 4.1.2: The affect of capacitance hats with radius=15 cm at lower frequencies.
Approximate width Resonance Frequency [MHz] VSWR at Resonance Frequency Cross Wire hat Solid hat Cross Wire hat Solid hat
3.6 m 33.1 26.2 24.8 1.07 1.26 1.38
3.2 m 38.1 28.8 27.2 1.17 1.17 1.29
2.8 m 43.4 32.0 30.0 1.22 1.09 1.20
The simulated E-field of a dipole at 134 MHz can be seen in Figure 4.1.4 and is one example of an antenna with poor lobe width. The -3 dB dot (red dot) shows that less than a meter to the side the field are already too weak. Several antennas showed similar problems with the lobe width when they were simulated. More plots of the simulated E-field can be found in Appendix C.2. But these problems were not as prevalent during the tests so no antennas were discarded due to these simulations
Figure 4.1.4: Horizontal plane from a dipole at 134 MHz.
4.1.3 E-field Strength
The simulated E-field from an antenna sometimes differed significantly from the measured E-field and sometimes did not. In Figure 4.1.5 the measurements and simulations of the E- field in position 1, 3, 7, and 9 are shown. The vertical dotted line is the resonance frequency which is the most important frequency for the simulation to be close to reality. It is around the resonance frequency the antenna works best and where an actual antenna would be used.
The figure shows the large difference in accuracy for the simulation even within the same
E-field. Similar effects were found in several different antennas. Even in, but less significant,
the simplest of straight dipoles this problem was found (can be seen in Appendix C.7).
Figure 4.1.5: Tested and simulated E-Field from a bowtie with a tested resonance frequency of 24 MHz.
Based on the measured results several of the studied antennas could be discarded due to not fulfilling the requirements.
4.2.1 E-field strength
Figure 4.2.1 displays the measured antennas and how they behave in comparison with the
FSA. It can clearly be seen that the antennas in Figure 4.2.1b have severe problems of reaching
higher strengths in their E-fields. On the other hand Figure 4.2.1a displays the cross, bowtie,
and straight dipole reaching the same level as the FSA or higher. Figure 4.2.2 shows how the
E-field, for all antennas, might look like when subjected to 5kW input power. Note that this is only a scaled (according to Chapter 3.4.5) result of Figure 4.2.1, 5 kW of input power were never actually tested. All antennas in Figure 4.2.2b except the straight dipole are below the 200 V/m limit (set in the requirements) at most frequencies. Because of this result, only the cross, bowtie, and straight dipole would be further investigated. For all positions see Appendix C.5.
(a) Bowtie and cross. (b) 90
bend, folded vee, half circle, straight dipole, and FSA.
Figure 4.2.1: Measured E-field for antennas at position 1 with 10 W input power.
(a) Bowtie and cross. (b) 90
bend, folded vee, half circle, straight dipole, and FSA.
Figure 4.2.2: Scaled E-field values at 5kW input power.
Figure 4.2.3 shows the E-field at three different positions, 1 (lateral center), 2 (one meter right of 1), and 3 (two meters right of 1). This gives insight into the horizontal homogeneity of the FSA with horizontal polarization. In this figure the dotted lines are 3 dB and -3 dB limits compared to the E-field in position 1, the dashed line. The solid lines are the E-field at the other positions (2 and 3) and they are desired to stay between the dotted lines to fulfill the requirement mentioned earlier in Table 1.2.1. Figure 4.2.3 are the FSA’s worst homogeneity result and it can be seen that for the upper part of the frequency band the -3 dB limit is between pos 2 and pos 3. In other words, the -3 dB limit is reached between these positions and thus the lobe width also ends there. This case would result in a lobe width of approximately 1.5 m to one side (3 m lobe width in total) at 1 m in front of the antenna. All homogenieties and polarizations for the FSA can be seen in Appendix C.6.
Figure 4.2.3: Horizontal homogeneity for the FSA with horizontal polarization.
For the cross and bowtie antennas the E-field in four positions are shown in Figure 4.2.4a
respective 4.2.4b. The positions are 1, 3, 7, 9, which are explained in Chapter 3.4.2. Note
that these positions are not the same positions as earlier in Figure 4.2.3. For both antennas
position 3 and 9 are below the limit between 60-110 MHz while for the bowtie the same
positions at low frequencies goes far over the limit as well.
(a) Cross (b) Bowtie
(c) Capacitance hat
Figure 4.2.4: Homogeniety of the Cross and Bowtie with horizontal polarization.
Adding capacitance hats to the cross makes its resonance frequency lower but also affects the
homogeneity significantly. Figure 4.2.4c shows the E-field at the different positions for the
cross antenna with 15 cm radius hats, seen in Figure 4.2.5. At the lowest frequencies there
is a major difference in the E-field between positions 1/7 and 3/9. Note that the X-axis for
Figure 4.2.4c differ from the others since the hats purpose is to be used at and affect the lower
Figure 4.2.5: Cross antenna with large capacitance hats.
The three antennas (bowtie, cross, and straight dipole) displayed a VSWR that fulfilled the <5 re- quirement, as seen in Figure 4.2.6.
Anywhere where that was not the case, was due to a missing resonance frequency measurement.
The cross has a slightly lower VSWR compared to the straight dipole which could explain the higher E-field back in Figure 4.2.1.
Two reference lines are drawn at VSWR = 2 & 5.
Figure 4.2.6: VSWR for bowtie, cross, and straight dipole.
To be able to tune the antennas a pulley system would be used. Strings would attach to the end of the antenna elements, reach around the pulley, and back towards the feed point in the center. An electric motor would roll up the string which would drag the antenna elements (the aluminum braid) outwards to make the antenna longer, see 4.3.1. The aluminum braid would be rolled up on the spool close to the center and it would be spring loaded to roll it back in when the motor releases the wire to make the antenna shorter. Note that the antenna elements and the pulleys would be located differently for each antenna and that Figure 4.3.1 shows what the cross antenna concept looks like. The motor would tune the antenna by either a preset size setting that are time dependant or match the size to its corresponding frequency when that frequency is detected.
Figure 4.3.1: Pulley concept.
5 Conclusions and Future Work
This final chapter will cover what could be said based on the results from the thesis, what is missing, and what can be done in the future to improve the results. Both simulations and measurements will be covered and they will specifically be compared in one chapter.
The VSWR looked good for almost all the antennas, both in simulations and the tests. Staying far below the limit of 5. A low VSWR is needed for good efficiency and a high E-field but is in no way a guarantee for it. But the VSWR is easy to simulate in EZNEC and easy to test as well. This makes it a good starting point for testing new ideas and antennas.
The homogeneity simulations showed that most antennas would have a very low acceptable test area due to quickly reaching the -3 dB limit compared to the center. Some antennas seemed to fare very bad due to a sharp valley in the E-field, such as in Figure C.2.1f and C.3.1f.
However, all these problems in the simulations were a lot smaller when looking at the ac-
tual test results. Looking at the test results in Figure 5.2.1, position 3 and 9 were below the
-3 dB limit in the center of the frequency band for both the cross and bowtie antennas. This is
not too bad though since the limit was based on position 1’s value, which at those frequencies
was the highest value. What this means is that if the field strength at position 1 was seen as
the 3 dB limit, the other positions would be above the -3 dB limit or close to it. This logic
however, does not help the bowtie or the cross with added hats at the lower frequencies. Here
the difference between the positions in the center (1/7) and to the side (3/9) too large.
(a) Cross (b) Bowtie
Figure 5.2.1: Homogeniety of the Cross and Bowtie with horizontal polarization, taken from Chapter 4.2.2.
What this means is that the two antennas that reach down to 26 MHz have a bad lobe width.
To determine the actual lobe width more measuring points would be needed. Now it can only be said that it is between 0 and 4 m, but an estimate would be 1.5-2 m. Valleys in between the measuring positions could also exist and would not be found without adding more measuring positions closer together.
The reason for this large difference could be because the outer parts of these antennas; the vertical wires and the actual hats, results in a large radiating area of the antenna that are much closer to the positions to the side. A possible solution would be to angle the elements slightly backwards to create more distance between the outer parts and the EUT. This would have to be done with care though since the same positions have the weakest E-field in the middle of the frequency band and can not be lowered too much. Smaller hats would help by lowering the the E-field but also not reach as low in frequency, so a balance would have to be found.
5.3 E-field Strength
The cross, bowtie, and straight dipole were the antennas that was the most efficient and pro- duced a significantly stronger E-field than the other tested antennas. Compared to the FSA they were stronger around the middle frequencies but weaker above 120 MHz and around 30 MHz. The straight dipole produces a strong E-field but can not reach close to 26 MHz while staying inside the size limit. It drops completely below 60 MHz, where its resonance frequency is at maximum size.
The reason for the E-field of the cross being stronger than a straight dipole’s is not clear.
Nothing could be found that indicated that this would be the case. The most likely reason
is the small difference in VSWR that can be seen in Figure 4.2.6. Other possible reasons are that the cross has a larger area that can radiate compared to the straight dipole, or the fact that the cross’s lowest point is closer to the ground plane affect the results, see Figure 5.3.1.
Figure 5.3.1: Height difference
That the E-field of the bowtie is a bit weaker compared to the two earlier mentioned antennas is probably due to the VSWR. The bowtie’s VSWR are considerably worse at various frequencies which also can be seen in Figure 4.2.6.
As seen in Figure 4.1.5 the simulations sometimes matched well with the tests but were also sometimes very different. The accuracy when simulating the strength of the E-field were different between antenna types, antenna lengths, frequency, and positions. No relationship between the variables and the accuracy could be found which made the simulations of the E-field strength unreliable.
A first thought to explain the differences were that they were simulated with a perfect, infinite ground plane and the semi-anechoic chamber does not work very well below 100 MHz. This however could not be the case since all tests were done sweeping from 20 MHz to 150 MHz and the results does not seem to match up better at higher frequencies. Another reasoning could be that the simulations were closer to reality around the antennas resonance frequency and then became worse further from it. But as seen in Figure 4.1.5 they could be close and far apart even for the same antenna but at different measured positions. It was e.g. almost perfect in position 9 and differ with twice the amount in position 1.
In contrast to the E-field strength the tested and simulated VSWR showed consistency. The
dips in VSWR were always close to or just a few MHz apart when comparing them. The
VSWR differed more at frequencies far from the resonance point but this was of lesser impor-
tance since the antennas were not planned to be used there anyway.
not the NEC code) it was argued that in his experience the simulations can be very good and even represent reality better than many tests, which can have many error sources. In this thesis however, very good simulations were not able to be achieved. To improve the simulations the transmission line loss could be better modeled by having several of them with attenuation corresponding to each cable used in the tests.
To conclude everything in this thesis, it can be said that the cross and bowtie antennas were the two best antennas found. They both reached a strong E-field and a low VSWR. The cross being the best in these areas. With capacitance hats added to the cross both it and the bowtie had resonance frequencies below 26 MHz within the size restrictions. The tuneability of the cross antenna would work exactly as the concept presented in Chapter 4.3. The bowtie would be slightly more complex to make tuneable with the pulleys needing to move in and out.
The problems the cross and bowtie had were with the homogeneity. The E-field was much stronger to the sides than in the middle when radiating at low frequencies. But to fully see how significant this problem is more tests with more measured positions would be needed.
If this issue with homogeneity could be solved and the antennas showed similar results in a vertical polarization i.e. a strong E-field and low VSWR. Both antennas would be able to replace the FSA.
5.6 Future Work
There exist several possibilities to continue the work in this thesis. From better simulations of the chosen antennas to actually realizing a control system for a tuneable antenna.
There exist lots of possibilities of continued work within this area. If improvement in the simulation models could be done to the point of them almost always being close to the test results (if that is possible) continued work would go much faster. Then the tests would only be needed to verify in the end and not to make comparisons between designs. Although to get to the point were you would trust the simulations you would have to have compared them to a lot of test results.
With or without perfect simulations next step would be to test more kinds of antennas similar to the cross or bowtie which gave best results in this thesis. For these more positions should be measured, primarily in a straight vertical and horizontal direction from position 01. This would give a much better overview of the homogeneity and lobe width which were impossible to get with only the four measurement positions used for the cross and bowtie.
Testing more sizes and thus more resonance frequencies could be made to confirm which
dips in the E-field strength that are due to no matching length and which exist for some other
reason. The E-field strength also needs to be tested when the antennas are in vertical polar- ization as this was not done in this thesis. Their performance can vary significantly between polarizations and both need to meet the requirements.
5.6.3 Constructing the Antenna
A whole other challenge would be to realize the actual tuneable antenna, both mechanically
and electrically. Design decisions concerning if the electronics should measure the frequency
sent to the antenna and from that change the size of it or if it should run on a predetermined,
time based change would have to be made. With the former being more compatible and
needing less work to adapt but being harder to implement from the beginning. One could
also imagine that the antenna would tune itself to the most effective size depending on the
frequency by measuring the output. Everything also has to survive the high power and
radiation that will be used which in itself would be a challenge.
 Tim Williams. EMC For Product Designers. Elsevier, Reading, Massachusetts, fifth edition, 2017.
 Directive 2014/30/eu of the european parliament and of the council of 26 february 2014 on the harmonisation of the laws of the member states relating to electromagnetic com- patibility (recast). Journal L 096, Volume 57:Pages 79-106, 2014-03-29.
 Constantine A. Balanis. Antenna Theory. John Wiley & Sons, Inc., fourth edition, 2016.
 Peter Joseph Bevelacqua. http://www.Antenna-theory.com/basics/main.php, December 2018.
 Chetvorno. https://en.wikipedia.org/wiki/Antenna (radio), December 2018.
 Johnny ¨ Ostermalm Carl Nordling. Physics Handbook for Science and Engineering. Stu- dentlitteratur AB, 8:8 edition, 2006.
 Unknown. https://en.wikipedia.org/wiki/Coaxial cable, January 2019.
 USA Department of Defense. Requirements for the control of electromagnetic interference characteristiscs of subsystems and equipment. In MIL-STD-461G, page 131. Department of Defense, USA, December 2015.
 International Electrotechnical Commission. Specification for radio disturbance and im- munity measuring apparatus and methods - part 1-4: Radio disturbance and immunity measuring apparatus - antennas and test sites for radiated disturbance measurements.
In CISPR 16-1-4:2019. International Electrotechnical Commission, 2019.
 Roy W. Lewallen W7EL. https://www.eznec.com/, January 2019.
 Marcus Spielmann. http://www.fs-antennentechnik.de/, January 2019.
 Roy W. Lewallen W7EL. EZNEC User Manual, 6.0 edition, 2018.
 Advanced Test Equipment Rentals. https://www.atecorp.com/products/amplifier- research/fp7018, January 2019.
 Ingenieub¨ uro Bauer. www.bauer-emv.de/International/SF RSUS.html, January 2019.
A Division of Labour
This appendix will clarify how the workloads were divided between the two authors, Gustaf Ljungn´ e and Lucas Svensson. That is the following text discuss who was responsible for what during the master thesis and also who wrote what in the final report.
Both authors have had weekly meetings with SAAB and been in ¨ Ostersund to perform tests in the semi-anechoic chamber. They both formulated the problem and shaped the scope of the thesis together with their supervisors. Most chapters in the report were in part contributed to by both authors and the conclusion together with future work were contributed to in equal amount.
A.1 Lucas Svensson A.1.1 Responsibilities
• Programming in MATLAB to read .txt files
• Controlling LTU’s instruments over GPIB and RS-232 to automate measurements
• Data and name keeping for the large number of tests and simulations performed
• Organize the test procedure at SAAB
• Dipole antenna simulations in EZNEC
• Deeper study of simple dipoles antennas A.1.2 Problem Solving
The simulation program used, EZNEC, gave all E-field results as tables in .txt files. To be able to plot any of these results they had to be imported into MATLAB. This was made especially hard since doing frequency sweeps made a small mini-table for each frequency with information in running text in between them. The problem was solved with a MATLAB script importing and reading the file in a very specific way to then edit it into a usable MATLAB table.
The anechoic chamber in Lule˚ a was set up to only show the current E-field and not save any data. This would make the sweeps of 53 frequencies that were needed take a long time and manpower to perform. Especially since the sweeps should be performed with the field probe at different positions. To save time we had to get around having to change the frequency by hand all the time, change the output power to correspond to 10 W for every frequency, and write down the results by hand.
The solution was to, through GPIB via MATLAB, program the signal generator to step
used by the antenna could be calculated in MATLAB. The script then changed the output power from the signal generator to get closer to 10 W of actual used power. This process were repeated a few times to create a sort of control system with a feedback loop. Lastly the E-field probe sent its data over RS-232 which were read and converted to a usable file in MATLAB. By doing all this the tests were reduced to just one button press.
The simulations in EZNEC were initially very poor and had to be improved. I improved how the segments in the model lined up which is important where there are wires close together, as in folded antennas. The simulations were still pretty bad but I realized that this was because in the simulation model there were no loss between the source and the antenna. Of course this loss exist in reality but so does lots of other losses that can be ignored. The reason this loss is so important is because with imperfect VSWR power will be reflected back from the antenna through the loss again. This power will at the source reflect again and go through the loss towards the antenna. This continues forever and without any loss all the power would always be used by the antenna. In short, without this loss the VSWR performance of an antenna never affected the E-field simulations at all. The solution was to add a transmission line between the source and the antenna and give it a loss that were close to the measured loss in SAAB’s cables.
A.1.3 Main Author
Chapters written are as follow: 2.2.3, 3.2, 3.3.3, 3.4.4, 3.4.5, 4.1, 4.1.1, 4.1.2, 4.1.3, 4.2.2, 5.1, 5.2, 5.3, 5.4, 5.6.1, 5.6.2
A.2 Gustaf Ljungn´ e A.2.1 Responsibilities
• Format the files from the tests to MATLAB usable code
• Handle the large amount of test data in MATLAB and plotting it
• Automate MATLAB to plot 3D plots and calculate wanted values for further studies
• Using MATLAB to sort data into plots so analysis could be performed
• Simulate bowtie/cross antennas in EZNEC
• Antenna measurement set-up
• Manufacture antenna prototypes
• Deeper studies concerning bowties and crosses
• Responsible for e-mail communication
• Final report structure (Set up LaTeX code)
A.2.2 Problem Solving Programming MATLAB
The only output from EZNEC when simulating the E-field was .txt files. To be able to analyze any of the data a script to import the data and later plot needed information was programmed.
The input in this script was: EZNEC data, wanted positions in the simulated room. With this the script approximated the position that the user wanted to know information about.
The approx. position calculated lines in xyz directions that contained the E-field values along those lines. These lines were then used to calculate ±3 dB limits, max/min values, and plot the shapes of the E-fields at wanted heights/lateral direction from the center.
To be able to analyze measured data an import script was made. This script contained a very basic GUI to simplify the process of selecting many files and safety features to make sure that correct files were selected and that the right number of files were read. The files were then plotted with calculated VSWR, ±3 dB, and E-field strengths at different probe positions.
Several different scrips were also used to handle all antenna data and later put them into plots presented in this report. Most of my coding was used to handle all data (both measured and simulated) and plot it in various ways needed.
A.2.3 Main Author
Chapters written are as follow: 1, 1.1, 1.2, 2.1, 2.1.1, 2.1.3, 2.2.1, 2.2.2, 3.1, 3.3.1, 3.3.2, 3.4.1,
3.4.2, 3.4.3, 3.5, 4.2.1, 4.2.3, 4.3, 5.5, 5.6.3
B Equipment Information
B.1 Immunity Testing
Table B.1.1 details exactly what equipment that is used at SAAB when conducting immunity testing.
DC2000 14122 2018-03-08 2019-03-08 M249698 Power Me-
- Rohde &
URV5 + 2 URV5-Z2
835294/001 2018-03-27 2019-03-27 M379787 Coaxial
- SAAB SCF SCF-101-
2018-06-28 2019-06-28 M380272 Coaxial
- SAAB RG-214
214-67-5 2018-06-21 2019-06-21 M379960 Field
2 MHz - 18 GHz
AR FP7018 311426 2018-07-24 2019-07-24
C Extra Results
C.1 Near field distance
Using Equations 10 and 11, the distance before the far field starts can be calculated. For the antennas in this thesis it is Equation 11 that sets that distance (it will always give a larger d than Equation 10). The length and distances for the antennas are documented in Table C.1.1.
d > λ
d > 2D
Table C.1.1: Distance to far field for the cross, bowtie and straight dipole antennas.
Antenna type Wave Length, λ Maximum
Resulting distance to far field, d
Straight Dipole, large 14.99 m 3.64 m 1.77 m
Straight Dipole, small 1.99 m 1.06 m 1.13 m
Cross, large 14.99 m 3.93 m 2.06 m
Cross, small 1.99 m 1.06 m 1.13 m
Bowtie, large 14.99 m 3.24 m 1.40 m
Bowtie, small 1.99 m 1.25 m 1.57 m
All the resulting distances to the far field, d, are larger than the 1 meter distance between the
tested antennas and the measuring field probe. This means that at resonance frequency the
measurements are always done in the near field but some times close to the start of the far