
Functional Description
(Continued)
THE LNA
The LNA is a common emitter stage with active feedback.
This feedback network allows for wide bandwidth operation
while providing the necessary optimal input impedance for
low noise performance. The power down feature is imple-
mented using a CMOS buffer and a power-down switch. The
power down switch is implemented with CMOS devices.
During power down, the switch is open and only leakage
currents are drawn from the supply.
THE MIXER
The mixer is a Gilbert cell architecture, with the RF input
signal modulating the LO signal and single ended output
taken from the collector of one of the upper four transistors.
The power down circuitry of the mixer is similar to that of the
LNA. The power down switch is used to provide or cut off
bias to the Gilbert cell.
Typical Low Noise Amplifier
TL/W/11814–14
FIGURE 4. Typical LNA Structure
A typical low noise amplifier consists of an active amplifying
element and input and output matching networks. The input
matching network is usually optimized for noise perform-
ance, and the output matching network for gain. The active
element is chosen such that it has the lowest optimal noise
figure, F
MIN
, an intrinsic property of the device. The noise
figure of a linear two-port is a function of the source admit-
tance and can be expressed by
F
e
F
MIN
a
R
n
G
G
[
(G
ON
b
G
G
)
2
a
(B
ON
b
B
G
)
2
]
where G
G
a
jB
G
e
generator admittance presented to
the input of the two port,
G
ON
a
jB
ON
e
generator admittance at which op-
timum noise figure occurs,
e
empirical
sensitivity of the noise figure to
generator admittance.
R
n
constant
relating
the
Typical Gilbert Cell
TL/W/11814–15
FIGURE 5. Typical Gilbert Cell Circuit Diagram
The Gilbert cell shown above is a circuit which multiplies
two input signals, RF and LO. The input RF voltage differen-
tially modulates the currents on the collectors of the transis-
tors Q1 and Q2, which in turn modulate the LO voltage by
varying the bias currents of the transistors Q3, Q4, Q5, and
Q6. Assuming that the two signals are small, the result is a
product of the two signals, producing at the output a sum
and difference of the frequencies of the two input signals. If
either of these two signals are much larger than the thresh-
old voltage V
T
, the output will contain other mixing products
and higher order terms which are undesirable and may need
to be attenuated or filtered out.
Analysis of the Gilbert cell shows that the output, which is
the difference of the collector currents of Q3 and Q6, is
related to the two inputs by the equation:
D
I
e
I
C3
b
I
C6
e
I
EE
tanh
V
RF
2V
T
tanh
V
LO
2V
T
and the hyperbolic tangent function can be expressed as a
Taylor series
tanh(x)
e
x
b
x
3
3
a
x
5
5
b
. . .
Assuming that the RF and LO signals are sinusoids.
V
RF
e
Acos (
0
RF
t
a
w
RF
)
V
LO
e
Bcos (
0
LO
t
a
w
LO
)
then
D
I
e
I
EE
Acos (
0
RF
t
a
w
RF
)
b
A
3
2
3
cos
3
(
0
RF
t
a
w
RF
)
a
. . .
(
#
D
Bcos (
0
LO
t
a
w
LO
)
b
B
3
3
cos
3
(
0
LO
t
a
w
LO
)
a
. . .
(
The lowest order term is a product of two sinusoids, yielding
a sum of two sinusoids,
cos ((
0
RF
a
0
LO
) t
a
w
RF
a
w
LO
)
I
EE
AB
a
cos ((
0
RF
b
0
LO
) t
a
w
RF
b
w
LO
)
(
one of which is the desired intermediate frequency signal.
9