Application Notes
(Continued)
The output voltage variation
V
is thus identical for RF
input signals that fall within the linear range (in dB) of the
detector. In other words, the output variation is independent
of the absolute RF input signal:
V
O
= V
Y
·
P
IN
In which V
is the slope of the curve. The log-conformance
error is usually much smaller than the ripple due to AM
modulation. In case of the LMV225/LMV228, V
Y
= 40 mV/
dB. With
P
IN
= 5 dB for CDMA,
V
OUT
= 200 mV
PP
. This is
valid for all V
OUT
.
(7)
Output Ripple with Additional Filtering
The calculated result above is for an unfiltered configuration.
When a low pass filter is used by shunting a capacitor of e.g.
C
OUT
= 1.5 nF at the output of the LMV225/LMV228 to
ground, this ripple is further attenuated. The cut-off fre-
quency follows from:
(8)
With the output resistance of the LMV225/LMV228 R
=
19.8 k
typical and C
= 1.5 nF, the cut-off frequency
equals f
= 5.36 kHz. A 100 kHz AM signal then gets attenu-
ated by 5.36/100 or 25.4 dB. The remaining ripple will be
less than 20 mV. With a slope of 40 mV/dB this translates
into an error of less than
±
0.5 dB. Since the LMV226 has a
low output impedance buffer, a capacitor to reduce the ripple
will not be effective.
Output Ripple Measurement
Figure 6
shows the ripple reduction that can be achieved by
adding additional capacitance at the output of the LMV225/
LMV228. The RF signal of 900 MHz is AM modulated with a
100 kHz sinewave and a modulation index of 0.3. The RF
input power is swept while the modulation index remains
unchanged. Without the output capacitor the ripple is about
200 mV
PP
. Connecting a capacitor of 1.5 nF at the output to
ground, results in a ripple of 12 mV
PP
. The attenuation with
a 1.5 nF capacitor is then 20
log (200/12) = 24.4 dB. This
is very close to the calculated number of the previous para-
graph.
PRINCIPLE OF OPERATION
The logarithmic response of the LMV225/LMV226/LMV228
is implemented by a logarithmic amplifier as shown in
Figure
7
. The logarithmic amplifier consists of a number of cas-
caded linear gain cells. With these gain cells, a piecewise
approximation of the logarithmic function is constructed.
Every gain cell has a response according to
Figure 8
. At a
certain threshold (E
), the gain cell starts to saturate, which
means that the gain drops to zero. The output of gain cell 1
is connected to the input of gain cell 2 and so on.
20076018
FIGURE 5. V
OUT
vs. RF Input Power P
IN
20076025
FIGURE 6. Output Ripple vs. RF Input Power
20076019
FIGURE 7. Logarithmic Amplifier
L
www.national.com
18