University of California, Berkeley Fall 2010

EE142, Problem Set #1 Solutions Amin Arbabian

Prof. Jan Rabaey
Problem 1
An antenna is an energy transformation system which converts driving current into EM radiation.
It can be modeled by a simple schematic as shown below. Resistor RX refers to the energy loss in
the form of heat caused within the antenna conductor. RR is called radiation resistance which is
dened as the total EM power radiated in all directions divided by the square of the driving current.
The radiation resistance increases with antenna dimension and thus the eciency of the antenna is
increased. Antennas with very small dimension have small radiation resistance, often comparable to
RX , which results in poor eciency.

Half wavelength at 900MHz :

c 3 × 108 m/s
0.5λ = 0.5 = 0.5 × = 16.67cm
f 900MHz
Half wavelength at 2.4GHz :

c 3 × 108 m/s
0.5λ = 0.5 = 0.5 × = 6.25cm
f 2.4GHz
Half wavelength at 10MHz :

c 3 × 108 m/s
0.5λ = 0.5 = 0.5 × = 15m
f 10MHz
A practical size of stand-alone antenna in portable wireless devices ranges from 1cm to several 10s
cm. This translates to a carrier frequency from several 100s MHz to 15GHz.
Pure water acts as an insulator whereas saline water has a very high conductivity of the order of
0.25 ohms/meter. The loss of radio waves is proportional to the operating frequency and also the
conductivity of the medium. Hence, by transmitting at a low frequency, the loss is reduced. Typical
antenna lengths for submarines are around 2000 ft ≈ 666 m.

Problem 2
Signal at the receiver output needs to meet a certain SNR (signal-to-noise ratio) so that the information
it carries can be retrieved at a satisfactory BER (bit error rate). The extra noise added by electronic

1
devices (transistors, resistors, diodes...) in the circuits deteriorates SNR as signal moves along the
receiver. As such, there's a minimum requirement on the input signal strength that makes sure a
worst-case SNR still be met.
On the other hand, as signal strength increases, the ratio of output signal to input signal will deviate
from the linear relationship and start to experience  compression or  expansion due to nonlinearity
of the circuits. The output signal is no more a faithful amplied copy of the original input but much
distorted. The distortion grows so much that it overwhelms the signal at receiver output. Thus
 distortion determines the largest signal a receiver can operate on.

Problem 3
Poynting's Theorem

Assuming plane waves and using Poynting's Theorem

~=E
S ~ ×H
~
→
− ~
In EM plane waves, E and H are perpendicular to each other and the direction of propagation.
Therefore,
~ ~r) = E0 cos(ωt − ~k.~r)
E(t,

~ ~r) = H0 cos(ωt − ~k.~r)

H(t,

where ~k is the wave vector.

Hence,
S(t) = E0 H0 cos2 (ωt − ~k.~r)
Average energy propagation per unit area < S >= 12 E0 H0 .
q
B0 E0 1
H0 = µ0 and B0 = c , where c is speed of light. Also, c= 0 µ0 .
Therefore,
1 E02
r
1 0 2
< S >= = E
2 µ0 c 2 µ0 0
If the transmitted power Pt = 1W and we consider a sphere with area 4πR2 ,we get

r
Pt 1 0 2
< S >= = E
4πR2 2 µ0 0

s r
Pt µ0
⇒ E0 =
2πR2 0
q
µ0
With Pt = 1W, 0 = 377Ω, we get

E100m = 0.0774 V/m

and
E1000m = 7.74 × 10−3 V/m

2
RFID Tag

If the transmitted power is Pt , the received power Pr and the eective area is Aef f , assuming that the
energy ows in a sphere as before, we have the relation

Pr Aef f
=
Pt 4πr2
2
With Aef f = 100 cm , Pt = 1 W, r = 10 m,

Pr = 7.96µW

Problem 4
In United States, Wireless LAN (WLAN) works between 2.412 GHz and 2.462 GHz with 20 MHz chan-
nel bandwidth (802.11b,g,n). 802.11a,h,j,n work between 5.2GHz and 5.8 GHz and 802.11y operates
at 3.6 GHz with 5 MHz bandwidth.
Cellphones work with a channel spacing of 200 kHz at carrier frequencies 850 MHz (824 − 849 MHz
uplink (mobile station to base station), 869 − 894 MHz downlink) and 1900 MHz (1850 − 1910 MHz
uplink, 1930 − 1990 MHz downlink).
Cellular network may have a Receiver Signal Strength (RSS) ranging from −40 dBm to −110 dBm.
Assuming a received signal strength of −100 dBm on a 50Ω antenna,

p
V = 2 ∗ 50 ∗ 1mW ∗ 10(−100dBm/10) = 3.16µV

For WLAN, the RSS may vary from −30 dBm to −60 dBm. Assuming a received signal strength
of −45 dBm on a 50Ω antenna,

p
V = 2 ∗ 50 ∗ 1mW ∗ 10(−45dBm/10) = 1.77mV

Problem 5
Optical ber loss is around 0.25 − 0.4dB/km. Typical wavelengths of operation are around 1300 nm
and 1550 nm, as the loss is minimum at these wavelengths.
For a cross-Atlantic ber, the required length of cable ≈ 6000 km (distance across Atlantic ocean)
Hence, the loss is around 2400 dBwhich is the major disadvantage of such a link. Hence, we require
the use of repeaters which makes the cost of the link very high. However, as it is a guided link it is
more reliable and also the channel capacity/data rates achievable with optical bers is very high.

Problem 6
The frequency used for uplink communication was 7.2 GHz and 8.4 GHz downlink.

Mars closer to earth

The frequency used isν = 8.4 GHz.

With a transmitted power of Pt = 1 kW and earth-mars distance r = 56 × 109 m,
the delay
r 56 × 109
τ= = =3 mins
c 3 × 108

3
 Using Friis equation, the received power

λ 2
Pr = Pt Gt Gr ( )
4πr
where Gt = Gr = 105 (50 dB gain).
c 3×108
The wavelength of signal λ= ν = 8.4×109 = 0.03571 m.
Hencem the received power
Pr = 2.575 × 10−14 W
.

Mars away from earth

Doing the same calculation as above,

ν = 8.4 GHz
Pt = 1 kW
r = 400 × 109 m
The delay
r 400 × 109
τ= = = 22 mins
c 3 × 108
Received power,

Pr = 5.047 × 10−16 W

Problem 7
Typical loss of a coaxial cable at 1GHz is from 0.15dB/meter to 0.4dB/meter. The maximum commu-
nication distance over a cable is given by :

Maximum transmitted signal-Minimum detectable signal by the receiver

Loss of the cable per distance

The minimum detectable is determined by noise.

The maximum signal that can be transmitted is of the order of several 1000 W and is determined
by :

1. Maximum current density that can be carried by the inner conductor

2. Maximum electric eld that the dielectric between the inner and outer conductor can tolerate

3. Maximum operating temperature

The maximum communication distance is given by :

2
5000V
10log( 2∗75Ω∗1mW ) − (−90dBm)
= 1.148 km
0.15dB/m

2
5000V
10log( 2∗75Ω∗1mW ) − (−90dBm)
= 430 m
0.4dB/m

4
Problem 8
The communication distance can be increased by

1. increase transmitter power

2. lowering minimum detectable signal strength (called sensitivity)

3. reducing cable loss

Thermal noise sets an ultimate limit in (2) and (3) because of electron vibration at any temperature
other than zero K degree. On the other hand, electro-migration and dielectric breakdown sets the
limit on the max current and max voltage that can be put out into the cable. Therefore the cable
transmission distance can not be increased indenitely.

Problem 9
Termination at the end of the cables is needed in order to prevent the incoming signal from bouncing
back to the source. This makes sure most signal energy is conveyed into the receiver. The amount
of reection is a relationship between the input (termination) resistance and the cable characteristic
resistance. Impedance match is a circumstance where 2 resistances are equal and no reection is
caused.

Problem 10
The bad connection causes an impedance mismatch between the antenna and the television set. Pre-
vious image signals get reected and reappear together with the current images.

Problem 11
Wireless communication over the ocean is near line-of-sight propagation because there're no terrain
obstacles. So it can transmit at a longer distance. In a densely populated city like SF, very often the
radio propagation is reected or deected by many buildings. The transmitted signal nds dierent
paths to arrive at the destination at dierent time instant. This is called multi-path eect. Multi-path
increases the diculty of recovering the information successfully. Therefore for the same transmission
rate, the transmission distance drops at rural area.

Problem 12
Radio spectrum is a government regulated resource. Wi-Fi communication is constrained in an ISM
band (Industrial, Scientic and Medical) around 2.4GHz due to FCC regulation. All laptop users in
the café share the same xed bandwidth. As more people turn on their WLAN adapter, the bandwidth
available to each user drops. The following picture shows the RSS at max download speed. Notice
a wider bandwidth (wider bell shape and more peak tones) is allocated to in order to achieve higher
speed.

5
 The second cause for reduced throughput is the interference from adjacent users. The more laptops
turns on nearby, the more intermodulation distortion (IM) created by them. Intermodulation distortion
degrades the bit error rate at the receiver output and slows down the speed.

Problem 13
Mobile jammer sends out radio waves at the same frequencies that cellular phones use. This causes
enough interference with the communication between cell phones and towers. Strong jamming signal
provides big distortion and can saturate the cellular phone receiver to disable its functionality.

Problem 14
Assume the AM transmitted signal has the following form:
Transmit AM signal =
x(t)cos(ω0 t)
At the receiver we will have:
Received signal=
x(t) ∗ cos(ω0 t) ∗ cos(ω0 + δω)t
After lowpass ltering, we have
1
x(t) ∗ cos(δωt)
2
This shows that in time domain analysis, the received signal is modulated by the error term cos(δωt)
. If δω = 0 we get the desired signal x(t). Equivalently in frequency domain the down-converted signal
spectrum won't exactly add up at baseband. We will have spectrum extension at baseband equal to
δω as well. We want δω as small as possible. To solve this problem, carrier recover network and phase
locked loop is used in the receiver. For a 900MHz PLL utilizing a 10MHz XTAL ( ±20ppm) as the

6
reference, the frequency error of the XTAL is transferred to the LO. The LO error will therefore also
be ±20ppm. In the worst case the bounds on the LO frequency are:
900M Hz ± 20ppm = 900M Hz ∗ (1 ± 20/1e6) = 900M Hz ± 18kHz

Problem 15
Let's say what we receive is x(t) ∗ cos(ω0 t + Φ).
After multiplying the received signal by cos(ωo t) at the receiver we will have :
Received signal =

1
x(t) ∗ cos(ω0 t) ∗ cos(ω0 t + Φ) = x(t)[cos(2ω0 t + Φ) + cos(Φ)]
2
After lowpass ltering we will have:
1
x(t)cos(Φ)
2
If Φ = 0 we have the maximum received signal and if Φ = 90◦ then we will have zero output.
To solve the problem we can generate a 90 degree phase shifted signal cos(ω0 t + 90) = sin(ω0 t) from
cos(ω0 t) at the receiver. Then we divide the input received signal in to two paths. On one path we
multiply the input by cos(ω0 t) and after lowpass ltering we will have: x(t)cos(Φ). On the other path,
we multiply the received signal by sin(ω0 t) and after lowpass ltering we will have: x(t)sin(Φ). Having
these two nal signals, we can recover signal x(t) and phase shift Φ . Another approach for solving
this problem is using carrier recovery and phased locked loop.

Problem 16
Ideally the frequency and phase of 2 clocks at transmitter and receiver are made identical such that the
received data is sampled at maximum signal-to-noise ratio. In practice, Receiver's locally generated
clock will have both an unknown phase oset and an unknown frequency oset from the transmitter's
carrier signal. These osets are the result of real-world issues such as crystal oscillator frequency drift
due to temperature or the propagation delay from the transmitter to the receiver. In the worst case,
one XTAL will be o by +20ppm (100MHz+2kHz) and the other by -20ppm (100MHz-2kHz). The
period of the fast XTAL will be 9.9998ns, while for the slow XTAL it would be 10.0002ns, a dierence
of 400fs. At 100Mb/s, one bit period is equal to 10ns.

7
 Let's assume the slow XTAL is used in the TX and the fast in the RX. The TX bit length is then
increased slightly to 10.0002ns as shown above. If we start sampling in the middle of the bit using the
faster RX XTAL the rst sample will be right in the middle of that bit but each subsequent sample
will be 400fs earlier than the middle of the next bit. The sampling point would move further to the
left, closer to the bit transition region. Eventually, we would end up with two samples within the same
bit. This is the point where synchronization would be lost between the TX and RX.
n = (10.002ns/2)/400fs = 12,501
T = (12,501 samples)*(9.9998ns/sample) = 125μs
The link would be synchronized for only 125μs. A potential solution is to extract the timing
information from the received signal itself instead of from an uncorrelated stand-alone crystal source.
This process is called clock recovery.

Problem 17
Circuit (a)

1
vo (jω) jωC 1
H(jω) = = 1 =
vi (jω) R + jωC 1 + jωRC

Circuit (b)

1
vo (jω) R2 || jωC
H(jω) = = 1
vi (jω) R1 + R2 || jωC

R2 2 R
vo (jω) 1+jωR2 C R1 +R2
H(jω) = = R2
=
vi (jω) R1 + 1+jωR 2C
1 + jω RR11+R
R2
2
C

Circuit (c)

1
vo (jω) R|| jωC 2
H(jω) = = 1 1
vi (jω) jωC1 + R|| jωC 2

R
vo (jω) 1+jωRC2 jωRC1
H(jω) = = 1 R
=
vi (jω) jωC1 + 1+jωRC2
1 + jωR(C1 + C2 )

Circuit (d)

io (jω) R 1
H(jω) = =− =− L
is (jω) R + jωL 1 + jω R

The Bode plots are shown below :

8
 |H(jω)| |H(jω)|
(dB) (dB)

R1R2
Reff=
R1+R2

1/(RC) 1/(ReffC)
0 ω (rad/s) 0 ω (rad/s)
R2
20 log ( )
R1+R2

(a) (b)
|H(jω)| |H(jω)|
(dB) (dB)

1/(R(C1+C2)) R/L
0 ω (rad/s) 0 ω (rad/s)
C1
20 log ( )
C1+C2

(c) (d)

Problem 18
PART A

The impedance at point A is Zin . Since the network is innite, the impedance at point B is also Zin .

Hence, the input impedance is

1
Zin = Z +
Y + Yin
1
where Yin =
Zin .
This gives quadratic equation
2 Z
Zin − ZZin − =0
Y
Therefore,
r
Z 4
Zin = (1 + 1+ )
2 YZ
√
√
r
Z Z 1
⇒ Zin = Z( + + )
4 4 Y

9
 r √ r
Z ZY ZY
⇒ Zin = ( + + 1)
Y 4 4
r r r
Z ZY ZY
⇒ Zin = 1+ (1 + ) (1)
Y 4 ZY + 4
Here, the characteristic impedance of the transmission line is

r r
Z ZY
Z0 = 1+
Y 4
q
Z
As the number of sections tend to innity, Z and Y tend to zero and Z0 → Y , only the ratio
remains.
With Z = jωL0 δz and Y = jωC 0 δz ,
r r
L0 ω 2 L0 C 0 (δz)2
Z0 = 1−
C0 4
q
L0
This indicates that the transmission line has a characteristic impedance of approximately
C 0 with
2
cuto frequency ωc = √ . As the number of sections tends to innity δz → 0 and ωc → ∞,
L0 C 0 (δz)2
q
L0
Z0 → C 0 . The second term multiplying the characteristic term in (1) indicates the phase dierence
or delay through the transmission line.
The network must be terminated with Zin when it is nite so that the calculate Zin holds true.

PART B

We have Z = jωL0 δz and Y = jωC 0 δz .

Using KCL at v(z), we have

i(z) − i(z + δz) = (jωC 0 δz)v(z)

i(z) − i(z + δz)

⇒ = (jωC 0 )v(z)
δz
Taking the limit as the number of section tends to innity or δz → 0, we get

10
 ∂i(z)
− = (jωC 0 )v(z) (2)
∂z
Using KVL for the inductor section, we have

v(z) − v(z + δz) = (jωL0 δz)i(z + δz)

v(z) − v(z + δz)

⇒ = (jωL0 )i(z + δz)
δz
Taking the limit as the number of section tends to innity or δz → 0, we get

∂v(z)
− = (jωL0 )i(z) (3)
∂z
Subsitute (2) in (3), we get

∂ 2 i(z)
= −ω 2 L0 C 0 i(z)
∂z 2
or

∂ 2 i(z)
+ ω 2 L0 C 0 i(z) = 0
∂z 2
Similarly,

∂ 2 v(z)
+ ω 2 L0 C 0 v(z) = 0
∂z 2

11