ECE3305 - Lab 6
ECE3305 - Lab 6
Materials Needed:
MATLAB
Spectrum analyzer
RF signal generator
Overview: Antenna radiation and detection is crucial to the field of wireless communications. In this lab,
you will explore some basic antenna radiation properties and then construct a link budget for a hypothetical
communications link.
1
ECE 3305: Lab 6
The table below shows some conversions to demonstrate the concept. Notice that 0.0 dBm is defined as
exactly 1.0 mW of power. Every multiplication by a factor of 10 is then treated as an increase of exactly
+10 dB. Likewise, every reduction by a factor of 10 is treated as a decrease by -10 dB. This allows us to
easily express a huge range of power with an intuitive range of linear values.
dBm Watts
+30 100
+20 10−1
+10 10−2
0 10−3
-10 10−4
-20 10−5
-30 10−6
2
ECE 3305: Lab 6
Figure 1: Top: Agilent N1996A spectrum analyzer. Bottom: Agilent N5181A analog signal generator.
3
ECE 3305: Lab 6
To illustrate, consider the time-domain signal x(t) with Fourier transform F (f ). From your signal-
processing classes, you should already have a sense that F (f ) represents an infinite set of sinusoidal
functions adding together to produce x(t). Each value in F is also represented as complex number that
indicates the magnitude and phase of the sinusoidal contribution.
In practice, we are often not really interested in the phase information at each frequency, but only
magnitude. This motivates us to define a new function, S(f ), such that
S(f ) = |F (f )|2 .
Suppose further that x(t) happens to represent the voltage on some transmission line with Z0 = 50 Ω
impedance. In this case, a more appropriate calculation for S(f ) would scale the result by impedance such
that
1
S(f ) = |F (f )|2 .
2Z0
Under this interpretation, the units for S(f ) become W/Hz, and the term power spectral density begins
to make sense. Each value in the function S represents a kind of power contribution at frequency f to the
total time-domain signal.
SIDEBAR: Strictly speaking S(f ) is not the PSD per se, but only an estimate of the PSD. For now,
however, we can ignore that distinction and just treat S(f ) are the true PSD.
For most applications, the power content of a signal will not concentrate at a discrete set of frequencies
(i.e., delta functions). Instead, the power tends to “spread out” over a continuous range. In this context, it
is common to ask how much total power is being carried by the signal between two frequencies of interest.
In this case, all we need to do is integrate the PSD over some specified bandwidth, [f1 , f2 ]. For example,
Zf2
P = S(f )df .
f1
When measured by a spectrum analyzer, that little bit of bandwidth is often referred to as a bin. We can
then imagine the instrument scanning over a bunch of tiny little bins and then measuring the total power
within each one. The result is finally plotted on a screen over some broader span. That final plot is what
we ultimately refer to as a power spectrum.
4
ECE 3305: Lab 6
Begin by digging out another 5-ft length of RG-174 coaxial cable. Use it to connect the output of
the RF signal generator directly into the input of the spectrum analyzer.
NOTE: Both the RF signal generator and spectrum analyzer use a type-N connector at their ter-
minals. Since your piece of cable used SMA connectors, you will need to either find an adapter or
use one of the short conversion cables available in the lab.
Turn on the RF signal generator and set the frequency to 1.0 GHz with an amplitude of 0.0 dBm.
You may have noticed that the signal generator has a small activation button marked “RF On/Off.”
This is to help avoid accidentally blasting sensitive instruments with power until you are absolutely
ready. Leave it OFF for now.
On the spectrum analyzer, press the “FREQ” button and set the frequency span to [900, 1100] MHz.
There is also another button marked “BW” which can be used to specify the bandwidth of the
individual frequency bins. Set this to a value of 5 MHz.
At this point, you should just see a bunch of random noise getting sampled over and over again. For a
much smoother measurement, press “Trace/Detector” and then select “Average.” This will perform
a running average of the measurement sweeps. Feel free to play with the number of averages to get
a feel for this feature.
Take a note of the measured power spectrum across your entire frequency sweep. For Gaussian
random noise, the power spectrum should be completely flat. This is called the noise floor. Use the
cursor to pick out some random frequency bin and then record the measured power of the background
noise.
Reduce the bin bandwidth to 1.0 MHz and repeat the power measurement of the background noise.
Did it increase or decrease? Why do you think this happened?
Finally, activate the 1.0 GHz signal by pressing “RF On/Off.” You should immediately see a sharp
peak appear at the center of your frequency sweep. Use the “Peak Search” button to automatically
move the cursor to this peak and then measure the total detected power from the RF generator.
Use this measurement to determine the total loss of the RG-174 and extrapolate out to dB/100 ft.
Compare the results against your previous measurements from Lab 1 and see how well they agree.
Document any relevant calculations and discussion your report.
5
ECE 3305: Lab 6
Start by gathering two of the matched monopole antennas from Lab 5. One can be from your own
group, and the other will be from another student group. Using two long cables of RG-174, connect
one antenna to the signal generator and the other to the spectrum analyzer. The resonant frequency
of your WiFi antenna should have fallen somewhere around 900 MHz, so let us focus on the immediate
bandwidth around it. Set the frequency sweep of the spectrum analyzer to [850,950] MHz with 1 MHz
bandwidth for the bins.
Place both antennas on the floor and separate them by 3 ft (HINT: Each floor tile is exactly one
foot). The monopoles should be oriented perpendicular to the floor and parallel with each other.
This is called a co-polarized orientation. Use the signal generator to excite the transmit antenna at
its resonant frequency with a power of 0.0 dBm. This will radiate an electromagnetic wave and get
picked up by the receive antenna.
Measure the total detected power on the spectrum analyzer. How much power is lost in the commu-
nication link? Give your answer in decibels.
Pick up the receive antenna and orient it so that the monopole is perpendicular with the transmitter.
This is called a cross-polarized orientation. What is the power now? What does this tell you about
antenna polarization?
Rotate the receive antenna so that its top is pointed directly at the transmit antenna. What is the
power now? What does this tell you about the directivity of antenna radiation?
Place the Rx antenna as far away from the Tx antenna as you can using the cables. Pick a few
random orientations between the two antennas and record your measurements. What does this tell
you about the effect of the environment on your measurement?
HINT: Think about the radiation as it bounces off the floor and walls. How strong of a signal do
you think this will yield, and how do you think it might scramble the polarization of your signal?
6
ECE 3305: Lab 6
7
ECE 3305: Lab 6
Follow the procedure below to fill in the appropriate table entries on the previous page. Be sure to
express each entry with units of dB or dBm.
Begin by imagining some input power P1 generated by the voltage source and injected down the
input line. Assume a value of 0.0 dBm to reflect your earlier test. As the signal propagates down the
line, some energy is lost due to cable attenuation. Use the value from your earlier measurement to
fill this in and express the loss in decibels.
Ideally, the impedance match between the antenna and the transmission line would be perfect, and all
power would radiate from the antenna. In practice, there is likely some mismatch, which you measured
in Lab 5. Use your characterization data to estimate the mismatch between the transmission line and
the transmit antenna at 900 MHz. Estimate the return loss with the ratio P3 /P2 such that
P3
RL = = (1 − |Γ1 |2 ) .
P2
Note: The reflected power technically also has to propagate back up the transmission line, reflect
off the generator, and then propagate down to the end of the cable again. In practice, the generator
is likely well-matched to the cable, and so the any secondary reflections are going to be negligible.
If the Tx antenna were perfect, all of the energy that excites it would radiate away. However, for any
real antenna, some of that energy may get absorbed in the conductors. In our case, the little strip
of microstrip transmission line plus the matching network is likely absorbing some power here. This
motivates us to define the antenna efficiency ε such that
P4 = ε 1 P3 ,
where P4 is the total radiated power out of the Tx antenna. Assume a value of 1 = 0.95 and use
that to calculate the loss due to antenna inefficiency.
For an isotropic radiator, all radiated power would distribute uniformly in all directions. Thus, if we
imagine a sphere with radius R, the radiated power density S at distance R away becomes
P4
S= ,
4πR2
where the units of S are W/m2 . In practice, however, all antennas possess some intrinsic gain,
or directivity, which means power tends to radiate better in some directions rather than others.
Assuming a directivity of D1 in the direction of the receive antenna (Rx), the power density then
becomes
P4
S = D1 .
4πR2
It is helpful to think of the Rx antenna as a small, rectangular aperture that captures energy as it is
illuminated by the radiator. This is called the effective area, or the effective aperture Ae . The
total energy captured by the Rx antenna thus satisfies
P5 = Ae S.
8
ECE 3305: Lab 6
The value Ae is directly proportional to the directivity D2 of the Rx antenna, and it is given as
D2 λ2
Ae = ,
4π
where λ is the wavelength of propagation in free space. Since the Rx antenna is identical to the Tx
antenna, you may assume that D2 = D1 = 5.14 dB. This implies a total received power of
2
λ
P5 = P4 D1 D2 .
4πR
The above expression is known as the Friis transmission equation. Assume a separation distance
of R = 1.0 m to approximate the 3-ft separation of your monopoles from before. The term in
parentheses at the end is called the free-space path loss (FSPL).
In principle, we could also include a corrective factor in the event that the two antennas are misaligned
in polarity. Since the antennas were co-polarized, however, this is not a significant issue. We may
therefore assume no loss of power due to polarization mismatch.
Strictly speaking, the Friis equation is only supposed to describe the special case of two antennas
floating in a free-space environment. In our situation, however, the presence of the ground plane
violates this assumption, and so we need to add a correction factor in order to obtain correct results
(see the supplemental articles on the CANVAS page for more information). For the special case of
two monopole antennas, the Friis equation will over-estimate the true received power by a factor
of four (≈ 6 dB). Thus, we need to subtract this from the link budget in order to obtain accurate
results.
As before, some energy is lost in the Rx antenna before reaching the coaxial cable. Let ε2 indicate
the Rx antenna efficiency so that
P6 = ε 2 P5 .
The mismatch between the coaxial cable and the Rx antenna implies another loss of power. Assuming
a reflection coefficient Γ2 , the transmitted power into the receiver cable becomes
P7 = P6 (1 − |Γ2 |2 ) .
This power again propagates down some length L2 of the second coaxial cable. Assume the cables
have the same attenuation properties to fill in the cable loss.
Finally, there is yet another reflection between the receive cable and some load impedance ZL . This is
often the equivalent input impedance to an amplifier/receiver circuit, which may or may not match
the cable impedance. Since you measured the result on a spectrum analyzer, assume there is no
reflection here.
9
ECE 3305: Lab 6
In principle, the receiver circuit could just amplify the detected signal by some arbitrary gain factor
G, thereby undoing all of the power loss along the communication link. In practice, however, there
will always exist some random noise in the system, which we can represent by the noise power Pn .
This noise would thus get amplified with the signal, effectively placing a lower-limit on the amount
of power loss that the system can tolerate. The signal-to-noise ratio is therefore defined as
Pr
SNR = ,
Pn
where Pr is the total power that reaches the detector/receiver circuit. Use your earlier measurement
with the spectrum analyzer to calculate the noise power Pn under 1.0 MHz of bandwidth. Give the
resulting SNR in decibels.
Discussion: Show your calculations, plus the completed link budget, in your report. Discuss how well
the prediction matched your measured outcome from Part 4 (anything within ±3 dB should be acceptable).
10
ECE 3305: Lab 6
What modulation scheme did you pick? How did you represent each digital bit?
How successful was your transfer of information? Were there any bit errors?
11