
ADXL50
–14–
REV. B
Capacitors C3 and C4 are chosen to provide the desired 3 dB
bandwidth. Component values are specified for bandwidths of
10 Hz, 30 Hz, 100 Hz, and 300 Hz. For other 3 dB bandwidths
simply scale the capacitor values; i.e., for a 3 dB bandwidth of
20 Hz, divide the 10 Hz bandwidth numbers by 2.0. T he nomi-
nal buffer amplifier output will be +1.8 V
±
19 mV/g. Note that
the ADX L50’s self-test will be fully functional since the buffer
amplifier is operated at unity gain and resistor R1 is large. T he
external op amp offsets and scales the output to provide a +2.5 V
±
2 V output over a wide range of full-scale g levels. T he exter-
nal op amp may be omitted in high g, low gain applications.
NOISE CONSIDE RAT IONS
T he output noise of the ADX L50 scales with the square root of
its bandwidth. T he noise floor may be reduced by lowering the
bandwidth of the ADX L50 either by increasing the value of the
demodulator capacitor or by adding an external filter.
T he typical rms noise of the ADX L50J with a bandwidth of
100 Hz and a noise density of 125
μ
V/
√
Hz
is estimated as
follows:
Noise (rms) = (
125
μ
V/
√
Hz
)
√
100
= 1.25
mV rms
Peak-to-peak noise may be estimated with the following
equation:
Noise p-p = (
6.6)
Noise rms
Peak-to-peak noise is thus estimated at 8.25 mV or approxi-
mately 0.4 g p-p. T he ADX L50 noise is characteristic of white
noise. T ypical rms and p-p noise for various 3 dB bandwidths is
estimated in Figure 25.
1.7
0
10
100
1k
0.73
0.33
1.0
1.4
3dB BANDWIDTH – Hz
N
N
0.26
0
0.11
0.05
0.16
0.21
Figure 25. ADXL50 Noise Level and Resolution vs. –3 dB
Bandwidth
Because the ADX L50’s noise is for all practical purposes Gaus-
sian in amplitude distribution, the highest noise amplitudes have
the smallest (yet nonzero) probability. Peak-to-peak noise is,
therefore, difficult to measure and can only be estimated due to
its statistical nature. T able III is useful for estimating the prob-
abilities of exceeding various peak values, given the rms value.
T able III.
Nominal Peak-to-
Peak Value
% of T ime that Noise will E xceed
Nominal Peak-to-Peak Value
2.0
×
rms
3.0
×
rms
4.0
×
rms
5.0
×
rms
6.0
×
rms
6.6
×
rms
7.0
×
rms
8.0
×
rms
32%
13%
4.6%
1.2%
0.27%
0.1%
0.046%
0.006%
AC COUPLING V
PR
T O BUFFE R INPUT
If a dc response is not required, as in applications such as mo-
tion detection or vibration measurement, then ac coupling
should be considered. In low g applications, the output voltage
change due to acceleration is small compared to the 0 g offset
voltage drift. Because ac coupling removes the dc component of
the output, the preamp output signal may be amplified consid-
erably without increasing the 0 g level drift. T he most effective
way to ac couple the ADX L50 is between the preamp output at
V
PR
and the buffer input, V
IN–
, as shown in Figure 26.
BUFFER
AMP
ADXL50
PRE-AMP
V
OUT
V
IN–
V
PR
R1
R2
R3
1.8V
C4
V
PR
8
9
10
Figure 26. AC Coupling the V
PR
Output to the Buffer Input
Using this configuration, the system’s ac response is now rolled
off—at the low frequency end at F
L
, and at the high frequency
end at F
H
. T he normalized frequency response of the system
can be seen in Figure 27.
T he low frequency roll-off, F
L
, due to the ac coupling network
is:
F
L
= 1/(2
π
R
1
C
4)
T he high frequency roll-off F
H
is determined by the dominant
pole of the system which is controlled by either the demodulator
capacitor and its associated time-constant or by a dominant post
filter.
As a consequence of ac coupling, any constant acceleration
component gravity will not be detected (because this too is a dc
voltage present at the V
PR
output). T he self-test feature, if used,
must be monitored at V
PR
, rather than at the buffer output.