Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
An interesting program:
Simulation and construction of a
Helix antenna for 2.45GHz using
4NEC2
This article represents the revised and below 1MHz) the wavelength rises to
extended version of a lecture at the 300m and a segment becomes at least 6m
UKW conference 2010 in Bensheim. long. The program cannot react to any
The article shows how a Helix antenna more refinements of the environment (it
can be sketched with the 4NEC2 soft- applies the rule of linear current on the
ware. Followed by the construction of smallest segment) and the simulation
a sample antenna, measuring its char- quickly becomes senseless.
acteristics and comparison with the
simulation. With high and very high frequencies the
concept of an infinitely thin antenna wire
does not work in the simulation any
more. A supplementary product for
1.0 thicker wires, the Extended thin wire
kernel can be used but as soon as the
Simulation of wire antennas relationship between of the segment
with NEC length to wire radius falls below 2.5, the
program ends with an error message with
wild warnings to the user about this
NEC (Numerical Electric Code) was de- problem
veloped in 1981 by the Lawrence Liver- NEC2 is the standard free application
more laboratory for the American Navy software. Development has continued
as a simulation method for wire antennas. and the weaknesses of NEC2 (e.g. incor-
The antenna is divided into very short rect computation of structures that cross
pieces (segments) where the current and very closely or wires buried in soil) were
voltage change is almost linear. Amaz- only corrected with NEC4. NEC4 was
ingly accurate simulations can be accom- restricted for a long time and was consid-
plished, but naturally there are limits. ered as secret. Today it is also available
Interestingly enough, a lower and an outside of the USA, but quite expensive
upper limit where the simulation be- (normally around the $2000).
comes ever more inaccurate.
NEC2 can be downloaded from the Inter-
The lower frequency limit is defined by net, but nothing can be done immedi-
the recommendation that there are ap- ately. It is a pure calculating machine that
proximately 10 to 20 segments per wave- was originally written in FORTRAN and
length but never more than 50 segments. available as a compiled coded program.
If very low frequencies are used (e.g. Therefore many people added extensive
89
VHF COMMUNICATIONS 2/2011
Fig 1: An online
calculator like this
makes the
determination of
the mechanical
antenna data
child's play.
91
VHF COMMUNICATIONS 2/2011
92
VHF COMMUNICATIONS 2/2011
Fig 3: Unbelievably
this short line
contains the
complete
description for the
Helix structure.
the antenna. Its upper end is con- “the Ground” and the fact that the
nected to the start of the Helix antenna wire is not infinitely thin.
(coordinates of Helix and the upper • From to Fig 7 the supply (Excitation)
end of the piece of wire the same). of the structure is programmed in the
The lower end of the wire is put on centre section of the connecting wire.
the Ground level that is formed by • ... the conclusion (Fig 8), the indica-
the reflector. NEC automatically puts tion of frequency as well as marking
the feed point into the centre seg-
ment. Thus the upper end of the feed the end of the NEC file.
wire will meet the start of the Helix The NEC file is now saved in a suitable
at a height of 5mm. place with the file extension “.nec” e.g.
• The next three entries; GE/GN/EK as “helix_2450MHz.nec”. 4NEC2X is
are all shown in Fig 6. They concern started and the file loaded using
the end of the geometry data, also “Main/file/open 4nec2 in/output file” and
93
VHF COMMUNICATIONS 2/2011
Fig 7: The exitation takes place in the Fig 8: There is no frequency sweep,
middle of the wire. the frequency is set to 2450MHz.
the path of this antenna file. Press F7 to One of the most beautiful options of
start the simulation after entering the 4NEC2 is the coloured 3D picture that
attribute (Far Field pattern/fill/resolution can be opened with the key F9. The first
= 5 degrees) as shown in Fig 9 and one (Fig 11) shows the antenna structure,
pressing “Generate”. it can easily be zoomed to show the
individual segments that the antenna di-
The simulated far field radiation pattern vided into for the simulation. Unfortu-
(far field pattern) at a frequency of nately the pictures in VHF Communica-
2450MHz is shown in Fig 10. The tions Magazine are in black and white
azimuth angle “Phi” = zero degree but the program colours the feed point in
(horizontal angle of rotation) is shown at violet at the centre segment of the feed
the bottom on the left. The elevation wire that makes it very easy to see. The
angle “Theta” is from 0 to 180 degrees or structure can be moved and rotated using
from 0 to -180 degrees is shown on the the mouse buttons.
bottom right.
Changing over the menu to “Multi Col-
our” shows the radiation pattern as well
Fig 9: Pressing F7 displays these Fig 10: The simulated far field is
simulation parameters. displayed as a polar diagram.
94
VHF COMMUNICATIONS 2/2011
95
VHF COMMUNICATIONS 2/2011
accessed using the “Show” menu. The ance behaves if the frequency is changed.
wide bandwidth of the antenna is shown
because the gain rises from 10.5dBi at 3.3. Results of measurement of the
2300MHz to approximately 12dBi at experimental model
2600MHz. At 2450MHz it is exactly
11dBi. The lower diagram is empty be- 3.3.1. Feed impedance
cause the earlier message stated that no An HP 8410 network analyser with a
front to back data would be calculated. magnitude and phase display module was
A Smith chart can also be displayed used with a transmission reflection
using the relevant button (Fig 17). This bridge for 2 to 12.4GHz. An HP 8690
shows even better how the feed imped- sweeper supplied the excitation signal
with a module for 2 to 4GHz. The
Fig 18: This shows
the measured
impedance on a
Smith Chart from
2300 to 2600MHz.
97
VHF COMMUNICATIONS 2/2011
desired frequency range was carefully ishes to a bright spot). This exactly
controlled from 2300 to 2600MHz by a measures at the beginning of the Helix
frequency counter. This counter received spiral when the antenna is attached for
its signal from the sweeper using a 10dB measurement. The result of the measure-
directional coupler. ment can be seen in Fig 18 and the
comment: “Not at all bad” is surely
The measuring set was calibrated care- deserved.
fully:
Perfect adjustment using a Watkins John- 3.3.2. Determining the antenna gain
son SMA termination with more than An antenna measuring range would nor-
30dB reflection attenuation up to more mally be required to determine the gain
than 12GHz (the expert immediately rec- of this antenna. Unfortunately this is
ognises these rare parts at the electronics missing in the domestic cellar workshop.
flea market by the blue lacquer finish). Thus the following cheat was used:
Thus the bright spot was centred care-
fully in the centre of the magnitude and The wavelength is 122.4mm for a fre-
phase display. quency of 2450MHz. If a transmission
circuit consisting of transmitter, receiver
Now the termination was replaced by a and two identical antennas is developed
short piece of SMA semi rigid cable with with a path distance between the trans-
the same SMA connector as used on the mitting and receiving antenna of 120cm
Helix antenna fitted to the end. First that corresponds to a distance of approxi-
100% reflection is exactly calibrated mately 10 wavelengths and thus “genu-
(bright curved path must be exactly on ine” far field propagation. But the fol-
the outside diameter of the magnitude lowing applies:
and phase display) AND the datum plane
set by rotating the correct control on the Since the radiated energy distributes over
measuring bridge to the point of the inner an ever-increasing solid angles when
pin of the SMA socket (the curve dimin- moving away from the transmitting an-
tenna the energy per unit area decreases
98
VHF COMMUNICATIONS 2/2011
according to the Friis transmission equa- With a level of 0dBm and the antenna
tion at the receiving antenna. This is the distance of 120cm the result is:
famous formula for free space path loss.
⎛ 12.24cm ⎞
If the gain of transmitting and receiving Transmitga in − 20 ⋅ log⎜ ⎟
⎝ 4π ÷ 120cm ⎠ = Transmitga in + 41.8dB
Antennagai n =
antenna is included it reads: 2 2
2
⎛ λ ⎞
PRe ceiver = PTransmitte r ⋅ GTransmit − antenna ⋅ GRe ceive − antenna ⋅ ⎜ ⎟
With a measurement of -29dBm at the
⎝ 4π ⋅ d ⎠ receiver (Spectrum Analyser calibrated at
2.45GHz) gives an antenna gain of
“d” is the distance between sending and 6.4dBi for the patch antenna that is about
receiving antennas and “λ” is the wave- the value in the textbooks (usually: 6.5 to
length that will be used for the calcula- 7dBi, depending upon form, material and
tion being the speed of light divided by losses).
the transmitter frequency.
With the transmitting antenna replaced
Using two completely identically manu- by the Helix antenna the level measured
factured patch antennas with a resonant at the receiver rose to around +2dB. The
frequency of 2.45GHz and an input re- input impedance of the Helix must be
flection of less than 3% at this frequency, considered, this is approximately 150Ω
in the measuring position. An advantage therefore a large part of the available
of the patch antenna is their affiliation to power (0dBm = 1mW) is reflected back
the family of the planar array antennas. into the generator. Calculating the addi-
That means that the patch surface is at tional attenuation is therefore child’s
the starting point for the radiation leaving play, because on the left of the Smith
the antenna (technically the phase cen- diagram of the 4NEC2-Simulation there
tre). are some nomograms (Fig 19). The two
lower ones are important and they will be
Using logarithmic representation in dB considered more closely:
(with gain in dBm) for this arrangement:
⎛ λ ⎞
For the point on the curve for 2450MHz
Re ceivepower = Transmitpo wer + 2 xAntennaga in + 20 ⋅ log⎜ ⎟ there is a point on the right hand scale
⎝ 4π ⋅ d ⎠
“return loss” of approximately 5.5dB that
in dB corresponds to the negative value of S11.
Re-arranging for the gain: The left scale gives (by the reflections
⎛ λ ⎞
causing auxiliary absorption = reflection
Re ceivepower − Transmitpo wer − 20 ⋅ log⎜ ⎟
⎝ 4π ⋅ d ⎠ loss) the value of approximately 1.5dB
Antennagai n =
2 and thus a measured gain for the Helix
in dB antenna of 6.4dBi + 2dB +1.5dB =
9.9dBi.
99
VHF COMMUNICATIONS 2/2011
That is approximately 1dB less than the as well as a beautiful auxiliary attenua-
simulation stated (11 dBi) but not too far tion diagram for the consideration of this
away. 1dB deviation correspond to 10% unpleasant effect.
change of the voltage or approximately
20% of gain… and the losses in the
polystyrene and the mounting plate have 4.0
not been considered… and the skin effect
at this frequency… Conclusion
The layout of the measuring setup is
shown in Fig 20. The principle is quite
clear: NEC is also useful for wire antennas; no
blind tests are necessary, because the
A counter is connected via a directional NEC simulation decreases the largest
coupler to indicate the frequency of the part of the work and supplies, as shown,
transmitter signal. After an attenuator (to useful results.
improve adjustment) there is a coax
relay. In the lower switch position the However the learning effort is substantial
transmit level can be measured and ad- and pitfalls lurk everywhere that lead to
justed to 0dBm. In the upper switch simulation errors. Often NEC or 4NEC2
position this output of 1mW goes to the does not warn of these and special
transmitting antenna. The inputs to the knowledge of the user is required.
wattmeter and the patch antenna are two Also only a tiny fraction of this tremen-
identical, equal lengths but as short as dously efficient program with its possi-
possible, low-loss, SMA semi rigid ca- bilities could be presented. A constantly
bles. The accurately measured level of growing 4NEC2 tutorial can be down-
0dBm = 1mW is fed to the antenna and is loaded from [3] with all possible simula-
radiated. On the receive side the patch tion examples and appropriate practical
antenna is screwed directly onto the references. However if all possibilities
analyser input. The two antennas are and refinement of the program were to be
accurately aligned at a distance 1.2m discussed it would take more than 1000
apart in poor reflection environment. sides.
(Note: The connection between the gen- Designing wire antennas with 4NEC2
erator and the antenna was made as short makes for much joy and increases your
as possible otherwise the mismatch (it personal knowledge. Much fun is due to
has a radiation resistance of 150Ω) may my friend and 4NEC2 specialist, Hardy
cause further unexpected attenuations. Lau of the Dualen University Baden
The reflections lead too “standing Württemberg in Friedrichshafen. Without
waves” in the cable and each maximum his knowledge and patient help with
causes losses in the cable (because the problems or questions and his notes on
voltage and current amplitudes are in- critical cases when the author was not
creased there) with this increase the SWR sure or made incorrect assumptions this
will raise substantial. They can increase antenna would not work probably yet.
the overall attenuation by over 4dB. Even
if the cable attenuation is only 1dB with
correct adjustment. A gain (in this case
unfortunately energy dissipation) unfor- 5.0
tunately always rises as a square of the
current or voltage amplitude. This effect Literature
is described in the chapter “Additional
power Loss due to SWR” in [2], there are
also formulas and arithmetical examples [1] John D. Kraus and Ronald J. Marhe-
fka: “Antennas for all Applications”.
100
VHF COMMUNICATIONS 2/2011
101