![](http://datasheet.mmic.net.cn/230000/LMX2240M_datasheet_15593808/LMX2240M_10.png)
Applications Information
THE INTERMEDIATE FREQUENCY LIMITER
The IF limiter has a large amount of gain at high enough
frequency to cause concern about oscillation. To ensure
that the limiter does not oscillate, a few precautions should
be taken. The compensation capacitors that are used
should be chosen to roll off any unwanted frequencies be-
low the band of interest. The capacitor should be a high Q,
RF type ceramic chip capacitor. For DECT, the capacitor
value should be 100 pF, and the capacitors should be sol-
dered as close to the LMX2240 as possible. This will create
a pass band from 40 MHz to 150 MHz. The AC coupling
capacitor at the input to the limiter (from the SAW filter)
should be the same value as the compensation capacitors.
THE DISCRIMINATOR
There are two types of discriminator that can be used to
demodulate FM signals. The first is a delay line discrimina-
tor, which uses a delay in one path of the received signal to
introduce a phase difference between it and the received
signal. The operation of the delay line discriminator is de-
rived in the inset box. The other type of discriminator relies
on a quadrature tank to directly introduce a phase shift in
the received signal. This is the type of implementation that
is commonly used in mobile communications because of its
relative ease of construction and low cost.
The discriminator operates best when the inputs to it are
hard-limited (i.e., square edges). If the input signal is small
enough such that the IF amplifier cannot limit it, the output
voltage swing of the limiter will suffer. Typically, the mini-
mum voltage swing the discriminator can see and still fully
switch is about 100 mV
PP
. The two inputs to the discrimina-
tor can be of different peak-to-peak voltage swings as long
as both are over the lower limit. This allows the quadrature
tank circuit to have some insertion loss. In fact, up to 8 dB
insertion loss can be tolerated while still ensuring that the
discriminator output won’t suffer.
The quadrature circuit can also affect the discriminator out-
put voltage swing. The discriminator output voltage swing
specified assumes perfect quadrature at the frequency of
interest (mixer operation). With available analog compo-
nents, perfect quadrature is not possible. This is due in part
to the high frequency of the IF and the proportionally very
narrow bandwidth of the desired signal. For example, a
DECT signal is about 1 MHz wide, which is
k
1% of the IF
at which the demodulation occurs. This makes the quadra-
ture circuit difficult to achieve. With moderately high Q com-
ponents, however, a reasonable phase shift can be
achieved with a single pole tank. This is illustrated by the
following equation: the output of the discriminator is given
by
DISC
out
e
cos(
0
c
t )
#
cos(
0
c
t
a
w
),
which results in
DISC
out
e
cos(
0
c
t
a
0
c
t
a
w
)
a
cos(
0
c
t
b
0
c
t
b
w
).
When the double frequency component is filtered out with a
low pass filter, the cosine of the phase remains
DISC
out
e
cos(
b
w
)
e
cos(
w
).
It can be seen that at 90
§
phase shift, the output will be zero.
At 0
§
, the output will be 0.5, and at 180
§
, it will be
b
0.5. The
output swing is then set by the multiplication of the cosine
term with the discriminator output amplifier’s gain.
(1)
(2)
(3)
With a circuit that gives an output peak-to-peak voltage of
1.0 V
PP
(min) with ideal quadrature, the slope is seen to be
5.5 mV/degree. With a practical quadrature tank circuit at
110.6 MHz, the phase shift over a 1 MHz bandwidth is about
45
§
–50
§
, which translates to an output peak-to-peak voltage
of about 250 mV
PP
.
m(t)
e
m
#
and b(t) is the modulating baseband signal. The con-
stant m is defined as m
e
2
D
fTb. The signal s(t) must
be delayed by some
u
so that
l(t)
e
s(t
a
u
)
e
cos (
0
c(t
a
u
)
a
m(t
a
u
)) .
If the delay
u
is such that
0
ct
e
2n
q
a
q
2
s(t
a
u
)
e
sin(
0
ct
a
m(t
a
u
)),
and multiplying (4) and (7) yields
s(t) l(t)
e
cos (
0
ct
a
m(t)) sin (
0
ct
a
m(t
a
u
))
e
1
2
Assume the FM modulated signal is denoted as
s(t)
e
cos (
0
ct
a
m(t)) ,
t
b
%
b(t) dt ,
(4)
where
(5)
,
n
e
0, 1, 2, 3, . . . ,
(6)
then
(7)
sin (2
0
ct
a
m(t)
a
m(t
a
u
))
(8)
a
1
2
sin (m(t
a
u
)
b
m(t)) .
The double frequency component can be filtered off
with a lowpass filter. If
u
is kept small,
1
sin (m(t
a
u
)
b
m(t))
&
1
2
2
2
#
[
m(t
a
u
)
b
m(t)
]
e
m
2
2
#
t
a
u
b
%
b(t) dt
b
t
b
%
b(t) dt
t
a
u
b(t) dt
m
(9)
e
m
t
&
u
m
2
b(t) .
The object for a delay line, then, is to maximize the de-
lay while retaining the approximations necessary to sat-
isfy (9),
u
k
0.1 Tb.
10