Design and Implementation of LNA at 900MHz for GSM applications
1. EE Dept CIIT Islamabad 1
AbstractâThis report presents the design methodology,
simulation results and implementation details of an LNA at
900MHz RF frequency. The LNA is designed to work on
bandwidth of GSM system. Destined to be unconditionally stable
and work over a considerable bandwidth of about 300MHz with
gain of over 10dB and Noise Figure of less than 4, the achieved
results conformed quite well to the specifications. Final results
include a gain of 12 dB at the center frequency with a nominal
variation of ±1.3dB over the desired bandwidth. The Noise Figure
obtained was 3.9dB, evidence of a compromise between Noise
Figure and unconditional stability. Very high degree of linearity
was achieved with output one-dB compression at +15dB and
output third order intercept at +46dB.
The report contains extensive graphical and tabular
representation of results at every stage of design, simulation and
implementation along with an explanation of increased Noise
Figure in exchange for unconditional stability
Index TermsâAmplifier, noise figure, Low noise, Stability,
Impedance matching
I. INTRODUCTION
Low noise amplifiers (LNAs) are used in wide range in
wireless communications. RF receivers have a LNA at their
front-end. It is a special type of electronic amplifier which is
used in communication systems.
A low noise amplifier is a device that amplifies incoming
electronic signal while introducing little noise of its own.
Because of this property (high amplification but low noise
injection). It is usually placed as the first component in a
receiver chain. Low noise amplifier (LNA) is also used in both
commercial and military applications such as cellular phones,
WLANs, Doppler radars and signal interceptors. Depending
upon the system in which they are used, LNAs can be
implemented using different design topologies. The systems
deployed in commercial applications aim toward high
integration, and low voltage and bias currents.
Before going through, the title of this project define word by
word, Low noise amplifiers (LNAs) is that amplifying the
signal plus bring minimal amount of noise to the signal. The
aim of this project is to achieve maximum gain while
minimizing the noise and distortion as possible. Gain of the
amplifier is expressed as:
Stability, Noise figure, Impedance matching are some
important elements in Low noise amplifier design. Although
these elements are not much dependant on each other but
tradeoff between these elements must be understood while
designing a Low noise amplifier. Noise figure of two port
amplifier is expressed as
The scope of this project is to design a low noise amplifier
for GSM application. The LNA is designed to operate at a
frequency of 900MHz. The report presents the simulation, test
results and analysis for the designed of LNA. The performance
of the LNA is analyzed on the advance simulators.
II. LOW NOISE AMPLIFIER CIRCUIT DESIGN
Design and simulation of the LNA is accomplished by using
advance simulators. Use the transistor BFG135 and simulate
the S parameter file only. The simulation process consists of 4
stages. In each stage a single design is being implemented.
Whereas in every next stage, the previously designed LNA is
being customized to achieve the best results. The results, such
as gain, noise figure at each stage are also compared. Coming
up with different software models for the transistor, which
would also include searching for manufacturer datasheets and
s-parameter files with noise data on the internet. For the
Design and Implementation of Low Noise
Amplifier at 900MHz
Abdus Sami(samihundal@gmail.com), Nauman Azeem, Shabir Ahmad
COMSATS Institute of Information and Technology Islamabad, Pakistan
3. EE Dept CIIT Islamabad 3
Figure.2 Low Noise Amplifier Design
D. Transmission Line Effects
After the layout is prepared, there is still one effect that is
not yet taken into consideration; effect of transmission line
segments in the RF path. What this means is that, even though
none of the conductors was designed to be a transmission line
segment, when RF signals pass through the copper conductor,
the losses of copper conductor affect the results (effect is
directly proportional to the length of the transmission line
segment) have to be taken into account. Although simulation
results donât show significant change in gain but losses of
copper conductor have a little affect on Noise figure. In the
schematic, replace all copper conductors joining two
components in the RF path with transmission line segments of
length equal to the distance between the components and width
equal to the width of the copper conductor. The width and
length information can be obtained from the layout by placing
a SCALE at any conductor that is to be measured. Once all
such copper conductor segments in the RF path were replaced
with their transmission line equivalents, the LNA was
simulated again and then tune the passive components again
for maximum gain. It is not possible to completely model all
transmission line effects. The schematic after inserting
transmission lines is shown in Figure.3.
Figure.3 Transmission line effects
III. HARDWARE IMPLEMENTATION
After the completion of design stage of LNA, second phase
is hardware implementation. In this phase the layout design is
the first step for hardware implementation of LNA. Following
are the steps for layout design and hardware implementation of
LNA: Make a list of all passive components that required on a
piece of paper. Now, include âSMT-PADâ from the library of
simulation software in the schematic and in the properties for
this âSMT-PADâ object specify the size of the various
dimensions by checking for the same component in their
respective MuRata datasheets. If more than one size
components are needed, include one âSMT-PADâ object for
every size and once again fill all dimension data. For every
passive component, in its properties, include one of these PAD
objects according to the size available in the lab. Now once all
components are assigned their respective size PADs, generate
the layout in simulation software. Straighten out the layout and
keep distance between components as small as possible so that
overall size of LNA is optimum. However, do not keep
components too close otherwise their pads will merge during
etching. Assign proper width to the RF path and DC signal
paths. RF paths from input RF connector to the first
component and from the last component to the output RF
connector should be a 50 ohm transmission line (for FR4 the
width come to about 2.96mm). Once the layout is done, create
the artwork which will create a plan showing just the copper
area and non-copper area. Print this artwork on a transparency.
Layout is shown in figure.4:
4. EE Dept CIIT Islamabad 4
Figure.4 Layout Design
Take a small piece of PCB board. Cut the PCB board of size
that can fit the layout and remove the blue plastic covering
from only one side. Now place the artwork transparency on the
exposed copper and photo-process the copper in the photo
etching machine. Then wear all the protective gear to protect
from harmful acids and other chemicals. The PCB has to be
âwashedâ in two liquid solutions: one, the developer and
second, a mix of water, hydrochloric acid (HCL) and hydrogen
peroxide. First, place the photo-processed PCB in the
developer and wait until the purple layer is completely
removed. The outline of the layout will now appear on the
PCB. Place this in the second solution and shake the container
to speed the chemical reaction between HCL and copper on
the PCB. Only that portion which was exposed to light is
removed by the acid, thus keeping the copper tracts of the
circuit schematic unharmed. Now that the PCB is ready, it is
roughened on both sides with sandpaper and holes are drilled
in the ground pads. Now all that is left is to solder the
components in their respective places.
After soldering the components final hardware is shown in
figure.5:
Figure.5 Low Noise Amplifier hardware implementation
IV. SIMULATION RESULTS
Given below are the simulation results for the LNA
designed.
A. Gain
Simulation results show that small signal gain of the LNA is
12.043dB at 900MHz(Figure.6)
Figure.6 Gain of LNA
Noise Figure
Simulation results show that noise figure for the LNA
designed is 3.93 (Figure.7)
Figure.7 Noise figure of LNA
Stability factor(”)
Stability factor measured for biasing circuit is show in
figure.8. Results show that â”â is greater than 1 for almost
complete frequency range but below 348MHz frequency the
value of stability factor(”) is less than one which shows that
network is not unconditionally stable so we need to apply a
stability sub circuit.
After applying stability sub circuit the results are shown in
Figure.9. Results show that the stability factor(”) is greater
than 1 for complete frequency range which is the condition for
unconditional stability.
After applying the matching circuit the stability measure was
enhanced because of less value of return loss, results are
shown in Figure.10
5. EE Dept CIIT Islamabad 5
Figure.8 Stability factor â”â of LNA for biasing nework
Figure.9 Stability factor â”â of LNA After applying
stability sub circuit
Figure.10 Stability factor â”â of LNA after applying
matching network
B. Reflection coefficients
Value of input and output reflection coefficients before
applying matching circuit are shown on smith chart in
Figure.11 and Figure.12
Figure.11 Input reflection coefficient before applying
matching network
Figure.12 output reflection coefficient before applying
matching network
Results shows that input and output is not 50⊠matched
so we need to apply appropriate LC circuit for impedance
matching.
After applying matching network the value of input and
output reflection coefficients on smith chart and linear scale
(Figure.13, 14, 15, 16). Smith chart results show that input
is exactly matched to 50⊠and output is also matched but not
exactly matched to 50âŠ. Linear scale results show that value
of reflection coefficient at 900MHz is -38.56dB and the
frequency range for which value of reflection coefficient is
less than -10dB is bandwidth of LNA designed which is
305MHz from 782MHz to 1.087GHz.
Figure.13 Input reflection coefficient after applying
matching network
6. EE Dept CIIT Islamabad 6
Figure.14 output reflection coefficient after applying
matching network
Figure.15 Input reflection coefficient after applying
matching network
Figure.16 Output reflection coefficient after applying
matching network
C. Non-linear simulation results
Active RF devices are ultimately non-linear in operation
When driven with a large enough RF signal the device will
generate undesirable signals. If an amplifier is driven hard
enough the output power will begin to roll off resulting in a
drop of gain known as gain compression. The point at which
gain of LNA drops 1dB from its linear Gain is called 1dB
compression point and Output power at which 3rd
order
intermodulation distortion products become equal in amplitude
to the main signal power is called third order intercept point
OIP3.
Figure.17 Shows the gain compression point of LNA and
input power which corresponds to 1dB compression point is
15.55dB. Figure.18 Shows the results for two tone test for
third order intercept point and the value of 3rd order intercept
point is 46.44dBm.
Figure.17 1dB Compression point
Figure.18 OIP3
V. CONCLUSION
In conclusion, an RF LNA at 900MHz was designed,
simulated, and built. The LNA specifications are shown in
Table.. However, there is a fitting explanation for high
Noise figure; the amplifier has a relatively high gain â to
achieve this gain and yet keep the amplifier inherently stable
over all frequencies is a difficult requirement. The only
suitable method for satisfying both these divergent
specifications simultaneously was to use a series resistor in the
input of the transistor. This series resistance has its own
undesirable side-effects, notably increase in the noise figure of
the entire LNA.
However, this research work has taught a great deal about RF
circuit design and implementation. It taught about the
7. EE Dept CIIT Islamabad 7
dynamics and relations between the different constituent
blocks of an RF system and how each component affects (or
does not affect!) the overall output.
Another learning is that, simulation software are mainly for
understanding these inter-block dynamics and not for
calculating the final values of the components unless ALL
parasitic and other similar RF effects are accounted for. Trial
and error in the final implementation stage is just as important
for obtaining optimum results; where the simulation software
predicts in which direction (greater or lower) should the next
trial component be
chosen.
ACKNOWLEDGMENT
First of all, we thank Allah All-Mighty for giving us the
strength and paving new ways for us. We would like to express
our deepest of gratitude to Mr. Haider Ali for his constant
encouragement and belief in us. He has been everything that a
group could wish for in a supervisor.
We are thankful to Mr. Safwan Khalid who has been extremely
helpful to us both inside and outside the institute premises. Our
classmate Mr. Waseem Ali deserves our thanks for going
through the theoretical work, research work and assisting us in
simulation phase of our project. We also appreciate the help of
our senior course mates who were helping us during the
hardware implementation of the project. Thanks to Dr. Syed
Irfan Ahmed for reviewing our thesis and providing us with
valuable suggestions. Finally, we would like to thank our
parents for their unconditional love and the faith which they
posses in us.
REFERENCES
[1] Microwave engineering 3rd
Ed by David M.pozar
[2] http://en.wikipedia.org/wiki/Low_noise_amplifier
[3] http://en.wikipedia.org/wiki/OIP3
[4] http://en.wikipedia.org/wiki/Gain_compression
[5] http://www.downeastmicrowave.com/PDF/IP3.PDF
[6] http://comsec.com/usrp/microtune/NF_tutorial.pdf
[7] Aleksandar TasiÄ. "PERFORMANCE PARAMETERS
OF RF CIRCUITS", Analog Circuits and Signal Processing
Series, 2006
[8] Domine Leenaerts, Jos Bergervoet, Jan-Willem Lobeek,
Marek Schmidt-Szalowskiâ900MHz/1800MHz GSM Base
Station LNA with Sub-1dB Noise Figure and +36dBm OIP3â
NXP Semiconductors, Eindhoven, 5656AE, the Netherlands