a

Monolithic Accelerometer
With Signal Conditioning
ADXL50*
For convenience, the ADXL50 has an internal buffer amplifier
with a full 0.25 V to 4.75 V output range. This may be used to
set the zero-g level and change the output sensitivity by using
external resistors. External capacitors may be added to the resistor network to provide 1 or 2 poles of filtering. No external
active components are required to interface directly to most
analog-to-digital converters (ADCs) or microcontrollers.

FEATURES
Complete Acceleration Measurement System
on a Single Monolithic IC
Full-Scale Measurement Range: 650 g
Self-Test on Digital Command
+5 V Single Supply Operation
Sensitivity Precalibrated to 19 mV/g
Internal Buffer Amplifier for User Adjustable Sensitivity
and Zero-g Level
Frequency Response: DC to 10 kHz
Post Filtering with External Passive Components
High Shock Survival: >2000 g Unpowered
Other Versions Available: ADXL05 (65 g)

The ADXL50 uses a capacitive measurement method. The analog output voltage is directly proportional to acceleration, and is
fully scaled, referenced and temperature compensated, resulting
in high accuracy and linearity over a wide temperature range.
Internal circuitry implements a forced-balance control loop that
improves accuracy by compensating for any mechanical sensor
variations.

GENERAL DESCRIPTION

The ADXL50 is powered from a standard +5 V supply and is

robust for use in harsh industrial and automotive environments
and will survive shocks of more than 2000 g unpowered.

The ADXL50 is a complete acceleration measurement system on

a single monolithic IC. Three external capacitors and a +5 volt
power supply are all that is required to measure accelerations up
to 50 g. Device sensitivity is factory trimmed to 19 mV/g,
resulting in a full-scale output swing of 0.95 volts for a 50 g
applied acceleration. Its zero g output level is +1.8 volts.

The ADXL50 is available in a hermetic 10-pin TO-100 metal

can, specified over the 0C to +70C commercial, and 40C to
+85C industrial temperature ranges. Contact factory for availability of devices specified for operation over the 40C to
+105C automotive temperature range.

A TTL compatible self-test function can electrostatically deflect

the sensor beam at any time to verify device functionality.

FUNCTIONAL BLOCK DIAGRAM

ADXL50

+3.4V
6

REFERENCE

VREF
OUTPUT

+1.8V

OSCILLATOR
DECOUPLING
CAPACITOR
4

OSCILLATOR

SENSOR

DEMODULATOR

C2
BUFFER
AMP

PREAMP
SELF TEST 7
(ST)
1

COM

C1

C3

+5V

C1

DEMODULATOR
CAPACITOR

VPR

10

R1

VIN

R3
VOUT
R2

*Patents pending.

REV. B
Information furnished by Analog Devices is believed to be accurate and
reliable. However, no responsibility is assumed by Analog Devices for its
use, nor for any infringements of patents or other rights of third parties
which may result from its use. No license is granted by implication or
otherwise under any patent or patent rights of Analog Devices.

Analog Devices, Inc., 1996

One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A.
Tel: 617/329-4700
Fax: 617/326-8703

(TA = TMIN to TMAX, TA = +258C for J Grade Only, VS = +5 V, @ Acceleration = 0 g,

ADXL50SPECIFICATIONS unless otherwise noted)

Parameter
SENSOR INPUT
Measurement Range
Nonlinearity
Alignment Error1
Transverse Sensitivity2

Conditions

Min

Guaranteed Full Scale

Best Fit Straight Line, 50 g FS

50

SENSITIVITY
Initial Sensitivity at VPR
Temperature Drift3

+25C

16.1

1.55/1.60

ZERO g BIAS LEVEL

Initial Offset
vs. Temperature3
vs. Supply

at VPR

NOISE PERFORMANCE
Voltage Noise Density
Noise in 100 Hz Bandwidth
Noise in 10 Hz Bandwidth

at VPR
BW = 10 Hz to 1 kHz

FREQUENCY RESPONSE
3 dB Bandwidth4
3 dB Bandwidth4
Sensor Resonant Frequency
SELF TEST INPUT
Output Change at VPR5
Logic 1 Voltage
Logic 0 Voltage
Input Resistance
+3.4 V REFERENCE
Output Voltage
Output Temperature Drift3
Power Supply Rejection
Output Current
PREAMPLIFIER OUTPUT
Voltage Swing
Current Output
Capacitive Load Drive
BUFFER AMPLIFIER
Input Offset Voltage6
Input Bias Current
Open-Loop Gain
Unity Gain Bandwidth
Output Voltage Swing
Capacitive Load Drive
Power Supply Rejection

ADXL50J/A
Typ

Max

Units

+50

g
% of FS
Degrees
%

19.0
0.75/1.0

21.9

mV/g
% of Reading

1.80
15/35
10

2.05/2.00

V
mV
mV/V

6.6
66
20

12

0.2
1
2

VS = 4.75 V to 5.25 V

C1 = 0.022 F (See Figure 22)

C1 = 0.0068 F

800

1300
10
24

ST Pin from Logic 0 to 1

0.85
2.0

1.00

32

Hz
kHz
kHz
1.15
0.8

To Common

50
3.350

DC, VS = +4.75 V to +5.25 V

Sourcing

500

Source or Sink

0.25
30

DC

DC, VS = +4.75 V to +5.25 V

0
40
40

V
mV
mV/V
A

VS 1.4

V
A
pF

25
20

10

10

5.25
13

V
mA

+70
+85
+125

C
C
C

VS 0.25

4.75

TEMPERATURE RANGE
Operating Range J
Specified Performance A
Automotive Grade*

10

V
V
V
k

mV
nA
dB
kHz
V
pF
mV/V

0.25
1000

POWER SUPPLY
Operating Voltage Range
Quiescent Supply Current

3.450

80
100
10
5
80
200

Delta from Nominal 1.800 V

IOUT = 100 A

3.400
10
1

mg/Hz
mg rms
mg rms

NOTES
1

Alignment error is specified as the angle between the true and indicated axis of sensitivity, (see Figure 2).
Transverse sensitivity is measured with an applied acceleration that is 90 from the indicated axis of sensitivity. Transverse sensitivity is specified as the percent of
transverse acceleration that appears at the V PR output. This is the algebraic sum of the alignment and the inherent sensor sensitivity errors, (see Figure 2).
3
Specification refers to the maximum change in parameter from its initial at +25C to its worst case value at T MIN to TMAX.
4
Frequency at which response is 3 dB down from dc response assuming an exact C1 value is used. Maximum recommended BW is 10 kHz using a 0.007 F capacitor, refer to
Figure 22.
5
Applying logic high to the self-test input has the effect of applying an acceleration of 52.6 g to the ADXL50.
6
Input offset voltage is defined as the output voltage differential from 1.800 V when the amplifier is connected as a follower (i.e., Pins 9 and 10 tied together). The voltage at
Pin 9 has a temperature drift proportional to that of the 3.4 V reference.
*Contact factory for availability of automotive grade devices.
2

All min and max specifications are guaranteed. Typical specifications are not tested or guaranteed.
Specifications subject to change without notice.

REV. B

ADXL50
ABSOLUTE MAXIMUM RATINGS*

Package Characteristics

Acceleration (Any Axis, Unpowered for 0.5 ms) . . . . . . 2000 g

Acceleration (Any Axis, Powered for 0.5 ms) . . . . . . . . . . 500 g
+VS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0.3 V to +7.0 V
Output Short Circuit Duration
(VPR, VOUT, VREF Terminals to Common) . . . . . . . Indefinite
Operating Temperature . . . . . . . . . . . . . . . . . 55C to +125C
Storage Temperature . . . . . . . . . . . . . . . . . . . 65C to +150C

Package

uJA

uJC

Device Weight

10-Pin TO-100

130C/W

30C/W

5 Grams

ORDERING GUIDE

*Stresses above those listed under Absolute Maximum Ratings may cause
permanent damage to the device. This is a stress rating only; the functional
operation of the device at these or any other conditions above those indicated in the
operational sections of this specification is not implied. Exposure to absolute
maximum rating conditions for extended periods may affect device reliability.

Model

Temperature
Range

ADXL50JH
ADXL50AH

0C to +70C
40C to +85C

CAUTION
ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily
accumulate on the human body and test equipment and can discharge without detection.
Although the ADXL50 features proprietary ESD protection circuitry, permanent damage may
occur on devices subjected to high energy electrostatic discharges. Therefore, proper ESD
precautions are recommended to avoid performance degradation or loss of functionality.
PIN DESCRIPTION

C2

Connection for an external bypass capacitor (nominally

0.022 F) used to prevent oscillator switching noise
from interfering with other ADXL50 circuitry. Please
see the section on component selection.

C1

TOP VIEW
COM
VREF

ST
7
88

COM The power supply common (or ground) connection.

Output of the internal 3.4 V voltage reference.

ST

The digital self-test input. It is both CMOS and TTL

compatible.

VPR

The ADXL50 preamplifier output providing an output

voltage of 19 mV per g of acceleration.

VPR

C1

THE CASE OF THE METAL CAN

PACKAGE IS CONNECTED TO
PIN 5 (COMMON).

+5V
VIN

NOTES:
AXIS OF SENSITIVITY IS ALONG
A LINE BETWEEN PIN 5 AND
THE TAB.

10
VOUT

The inverting input of the uncommitted buffer amplifier.

REV. B

C1
3

VOUT Output of the uncommitted buffer amplifier.

VIN

44

66

Connections for the demodulator capacitor, nominally

0.022 F. See the section on component selection for
application information.

VREF

C2
5
SENSITIVITY

The power supply input pin.

ESD SENSITIVE DEVICE

CONNECTION DIAGRAM
10-Header (TO-100)

AXIS OF

+5 V

WARNING!

ARROW INDICATES DIRECTION

OF POSITIVE ACCELERATION
ALONG AXIS OF SENSITIVITY.

ADXL50
Polarity of the Acceleration Output
The polarity of the ADXL50 output is shown in the Figure 1.
When oriented to the earths gravity (and held in place), the
ADXL50 will experience an acceleration of +1 g. This corresponds to a change of approximately +19 mV at the VPR output
pin. Note that the polarity will be reversed to a negative going
signal at the buffer amplifier output VOUT, due to its inverting
configuration.

TAB

+1g
PIN 5

Mounting Considerations

There are three main causes of measurement error when using

accelerometers. The first two are alignment and transverse sensitivity errors. The third source of error is due to resonances or
vibrations of the sensor in its mounting fixture.
Errors Due to Misalignment

Figure 1. Output Polarity at VPR

The ADXL50 is a sensor designed to measure accelerations that

result from an applied force. Because these forces act on the
sensor in a vector manner, the alignment of the sensor to the
force to be measured may be critical.

Z
TRANSVERSE Z AXIS

SIDE VIEW

The ADXL50 responds to the component of acceleration on its

sensitive X axis. Figures 2a and 2b show the relationship between the sensitive X axis and the transverse Z and Y
axes as they relate to the TO-100 package.

PIN 5

TAB

SENSITIVE (X) AXIS

Figure 2c describes a three dimensional acceleration vector

(AXYZ) which might act on the sensor, where AX is the component of interest. To determine AX, first, the component of acceleration in the XY plane (AXY) is found using the cosine law:

Figure 2a. Sensitive X and Transverse Z Axis

AXY = AXYZ (cosXY) then

AX = AXY (cosX)

TRANSVERSE Y AXIS

Therefore: Typical VPR = 19 mV/g (AXYZ) (cosXY) cosX

TOP VIEW

Note that an ideal sensor will react to forces along or at angles

to its sensitive axis but will reject signals from its various transverse axes, i.e., those exactly 90 from the sensitive X axis.
But even an ideal sensor will produce output signals if the transverse signals are not exactly 90 to the sensitive axis. An acceleration that is acting on the sensor from a direction different
from the sensitive axis will show up at the ADXL50 output at a
reduced amplitude.

PIN 5

TAB

SENSITIVE (X) AXIS

Figure 2b. Sensitive X and Transverse Y Axis

Table I. Ideal Output Signals for Off Axis Applied

Accelerations Disregarding Device Alignment and
Transverse Sensitivity Errors

Z Axis
Axyz

xy
Ax
X Axis
x
Axy
Y Axis

Figure 2c. A Vector Analysis of an Acceleration Acting

Upon the ADXL50 in Three Dimensions

% of Signal Appearing
at Output

Output in gs for a 50 g
Applied Acceleration

0
1
2
3
5
10
30
45
60
80
85
87
88
89
90

100%
99 98%
99.94%
99.86%
99.62%
98.48%
86.60%
70.71%
50.00%
17.36%
8.72%
5.25%
3.49%
1.7%
0%

50 (On Axis)
49.99
49.97
49.93
49.81
49.24
43.30
35.36
25.00
8.68
4.36
2.63
1.75
0.85
0.00 (Transverse Axis)
REV. B

ADXL50
Table I shows the percentage signals resulting from various X
angles. Note that small errors in alignment have a negligible
effect on the output signal. A 1 error will only cause a 0.02%
error in the signal. Note, however, that a signal coming 1 off of
the transverse axis (i.e., 89 off the sensitive axis) will still contribute 1.7% of its signal to the output. Thus large transverse
signals could cause output signals as large as the signals of
interest.

Sensitivity: The output voltage change per g unit of acceleration applied, specified at the VPR pin in mV/g.
Sensitive Axis (X): The most sensitive axis of the accelerometer sensor. Defined by a line drawn between the package tab
and Pin 5 in the plane of the pin circle. See Figures 2a and 2b.
Sensor Alignment Error: Misalignment between the
ADXL50s on-chip sensor and the package axis, defined by
Pin 5 and the package tab.

Table I may also be used to approximate the effect of the

ADXL50s internal errors due to misalignment of the die to the
package. For example: a 1 degree sensor alignment error will
allow 1.7% of a transverse signal to appear at the output. In a
nonideal sensor, transverse sensitivity may also occur due to inherent sensor properties. That is, if the sensor physically moves
due to a force applied exactly 90 to its sensitive axis, then this
might be detected as an output signal, whereas an ideal sensor
would reject such signals. In every day use, alignment errors
may cause a small output peak with accelerations applied close
to the sensitive axis but the largest errors are normally due to
large accelerations applied close to the transverse axis.

Total Alignment Error: Net misalignment of the ADXL50s

on-chip sensor and the measurement axis of the application.
This error includes errors due to sensor die alignment to the
package, and any misalignment due to installation of the sensor
package in a circuit board or module.
Transverse Acceleration: Any acceleration applied 90 to the
axis of sensitivity.
Transverse Sensitivity Error: The percent of a transverse acceleration that appears at the VPR output. For example, if the
transverse sensitivity is 1%, then a +10 g transverse acceleration
will cause a 0.1 g signal to appear at VPR (1% of 10 g). Transverse sensitivity can result from a sensitivity of the sensor to
transverse forces or from misalignment of the internal sensor to
its package.

Errors Due to Mounting Fixture Resonances

A common source of error in acceleration sensing is resonance
of the mounting fixture. For example, the circuit board that the
ADXL50 mounts to may have resonant frequencies in the same
range as the signals of interest. This could cause the signals
measured to be larger than they really are. A common solution
to this problem is to dampen these resonances by mounting the
ADXL50 near a mounting post or by adding extra screws to
hold the board more securely in place.

Transverse Y Axis: The axis perpendicular (90) to the package axis of sensitivity in the plane of the package pin circle. See
Figure 2.
Transverse Z Axis: The axis perpendicular (90) to both the
package axis of sensitivity and the plane of the package pin
circle. See Figure 2.

When testing the accelerometer in your end application, it is

recommended that you test the application at a variety of frequencies in order to ensure that no major resonance problems
exist.

100

GLOSSARY OF TERMS

90

Acceleration: Change in velocity per unit time.

Acceleration Vector: Vector describing the net acceleration
acting upon the ADXL50 (AXYZ).
g: A unit of acceleration equal to the average force of gravity
occurring at the earths surface. A g is approximately equal to
32.17 feet/s2, or 9.807 meters/s2.

10
0%

Nonlinearity: The maximum deviation of the ADXL50 output

voltage from a best fit straight line fitted to a plot of acceleration
vs. output voltage, calculated as a % of the full-scale output
voltage (@ 50 g).

1V

0.5ms

Figure 3. 500 g Shock Overload Recovery. Top Trace:

ADXL50 Output. Bottom Trace: Reference Accelerometer
Output

Resonant Frequency: The natural frequency of vibration of

the ADXL50 sensors central plate (or beam). At its resonant
frequency of 24 kHz, the ADXL50s moving center plate has a
peak in its frequency response with a Q of 3 or 4.

REV. B

0.5V

ADXL50Typical Characteristics
7

9
5 g p-p SIGNAL T A = +25C

0
NOISE mV, RMS

NORMALIZED SENSITIVITY dB

C1 = C2 = 0.022F
3

3
6
9

12
3

15
18

2
0.010

21
1

10

100
FREQUENCY Hz

1k

10k

Figure 4. Normalized Sensitivity vs. Frequency

DEMODULATOR CAPACITANCE F

Figure 7. RMS Noise vs. Value of Demodulator

Capacitor, C1

0.25%

100

0.20%

TA = +25C, ACL = 2
80

0.15%

OUTPUT IMPEDANCE

LINEARITY IN % OF FULL SCALE

0.1

0.10%
0.05%
0.00%
0.05%
0.10%

60

40

20

0.15%
0.20%
0.25%

20

10

30

40

50

60

10

100
1000
FREQUENCY Hz

g LEVEL APPLIED

Figure 5. Linearity in Percent of Full Scale

10000

Figure 8. Buffer Amplifier Output Impedance vs. Frequency

1350

30
TA = +25C

25
G = 10
20

1300

15

1250

GAIN dB

3dB BW Hz

C1 = C2 = 0.022F

1200

10

G=2

5
0
5
10

1150

15

1100
60

20

40

20

20
40
60
80
TEMPERATURE C

100

120

10

140

Figure 6. 3 dB Bandwidth vs. Temperature at VPR

100

1k
10k
FREQUENCY Hz

100k

1M

Figure 9. Buffer Amplifier Closed-Loop Gain vs. Frequency

REV. B

ADXL50
0.5
TA = +25 C
C1 = C2 = 0.022F
CHANGE IN SENSITIVITY %

CHANGE IN SENSITIVITY AT VPR %

+0.50

+0.25

0.25

0.0

0.5

1.0

1.5

0.50
4.8

4.9

5.0
5.1
SUPPLY VOLTAGE V

5.2

5.3

60

Figure 10. Change in Sensitivity vs. Supply Voltage

40

+3.4V REF PSRR dB

120

140

45

35

60

40

20

10

100
1k
FREQUENCY Hz

10k

100k

10

100
1k
FREQUENCY Hz

10k

100k

Figure 14. +3.4 V REF PSRR vs. Frequency

3.404

40

VREF

VREF OUTPUT Volts

20

10

60

1004

3.400

30

40

20

20
40
60
80
TEMPERATURE C

100

120

SELFTEST
3.392

0.996

3.388

0.992

3.384
60

140

Figure 12. 0 g Bias Level vs. Temperature

1000

3.396

CHANGE IN SELFTEST OUTPUT SWING mV

VPR 0g PSRR dB

100

TA = +25 C
VS = +5V + (0.5Vp-p)

Figure 11. VPR 0 g PSRR vs. Frequency

CHANGE IN 0g OUTPUT LEVEL mV

20
40
60
80
TEMPERATURE C

80

TA = +25 C
VS = +5V + (0.5Vp-p)
C1 = C2 = 0.022F

REV. B

Figure 13. Percent Change in Sensitivity at VPR vs.

Temperature

55

25

20

0.988
40

20

20
40
60
80
TEMPERATURE C

100

120

140

Figure 15. VREF Output and Change in Self-Test Output

Swing vs. Temperature

ADXL50
demodulator will rectify any voltage which is in sync with its
clock signal. If the applied voltage is in sync and in phase with
the clock, a positive output will result. If the applied voltage is in
sync but 180 out of phase with the clock, then the demodulators output will be negative. All other signals will be rejected.
An external capacitor, C1, sets the bandwidth of the demodulator.

THEORY OF OPERATION

The ADXL50 is a complete acceleration measurement system

on a single monolithic IC. It contains a polysilicon surface-micro machined sensor and signal conditioning circuitry. The
ADXL50 is capable of measuring both positive and negative acceleration to a maximum level of 50 g.

The output of the synchronous demodulator drives the preamp

an instrumentation amplifier buffer which is referenced to
+1.8 volts. The output of the preamp is fed back to the sensor
through a 3 M isolation resistor. The correction voltage required to hold the sensors center plate in the 0 g position is a
direct measure of the applied acceleration and appears at the
VPR pin.

Figure 16 is a simplified view of the ADXL50s acceleration

sensor at rest. The actual structure of the sensor consists of 42
unit cells and a common beam. The differential capacitor sensor
consists of independent fixed plates and a movable floating
central plate which deflects in response to changes in relative
motion. The two capacitors are series connected, forming a capacitive divider with a common movable central plate. A force
balance technique counters any impeding deflection due to acceleration and servos the sensor back to its 0 g position.

TOP VIEW

TOP VIEW

CS1

APPLIED
ACCELERATION
TETHER

CS2

CENTER
PLATE

BEAM
BEAM
CENTER
PLATE

FIXED
OUTER
PLATES

UNIT CELL
CS1 < CS2

CS1

CS2
CS1
CS2

DENOTES ANCHOR

UNIT CELL
CS1 = CS2

Figure 17. The ADXL50 Sensor Momentarily Responding

to an Externally Applied Acceleration

DENOTES ANCHOR

When the ADXL50 is subjected to an acceleration, its capacitive

sensor begins to move creating a momentary output signal. This
is signal conditioned and amplified by the demodulator and
preamp circuits. The dc voltage appearing at the preamp output
is then fed back to the sensor and electrostatically forces the
center plate back to its original center position.

Figure 16. A Simplified Diagram of the ADXL50 Sensor at

Rest

Figure 17 shows the sensor responding to an applied acceleration. When this occurs, the common central plate or beam
moves closer to one of the fixed plates while moving further
from the other. The sensors fixed capacitor plates are driven
deferentially by a 1 MHz square wave: the two square wave amplitudes are equal but are 180 out of phase from one another.
When at rest, the values of the two capacitors are the same and
therefore, the voltage output at their electrical center (i.e., at the
center plate) is zero.

At 0 g the ADXL50 is calibrated to provide +1.8 volts at the

VPR pin. With an applied acceleration, the VPR voltage changes
to the voltage required to hold the sensor stationary for the duration of the acceleration and provides an output which varies
directly with applied acceleration.
The loop bandwidth corresponds to the time required to apply
feedback to the sensor and is set by external capacitor C1. The
loop response is fast enough to follow changes in g level up to
and exceeding 1 kHz. The ADXL50s ability to maintain a flat
response over this bandwidth keeps the sensor virtually motionless. This essentially eliminates any nonlinearity or aging effects
due to the sensor beams mechanical spring constant, as compared to an open-loop sensor.

When the sensor begins to move, a mismatch in the value of

their capacitance is created producing an output signal at the
central plate. The output amplitude will increase with the
amount of acceleration experienced by the sensor. Information
concerning the direction of beam motion is contained in the
phase of the signal with synchronous demodulation being used
to extract this information. Note that the sensor needs to be positioned so that the measured acceleration is along its sensitive
axis.

An uncommitted buffer amplifier provides the capability to adjust the scale factor and 0 g offset level over a wide range. An internal reference supplies the necessary regulated voltages for
powering the chip and +3.4 volts for external use.

Figure 18 shows a block diagram of the ADXL50. The voltage

output from the central plate of the sensor is buffered and then
applied to a synchronous demodulator. The demodulator is also
supplied with a (nominal) 1 MHz clock signal from the same
oscillator which drives the fixed plates of the sensor. The

REV. B

ADXL50
+3.4V

DENOTES EXTERNAL
PIN CONNECTION

+3.4V

+5V
33k

75

33k
PREAMP

C2
C1

EXTERNAL
OSCILLATOR
DECOUPLING
CAPACITOR

CS1

180

CS2

1MHz
OSCILLATOR

EXTERNAL
DEMODULATION
CAPACITOR

+1.8V

+1.8V

BEAM
SYNCHRONOUS
DEMODULATOR

INTERNAL
FEEDBACK
LOOP

LOOP GAIN = 10
+5V

+5V

SYNC
3M
+5V

COMMON

RST

VREF
+3.4V

VPR

C1

+0.2V

INTERNAL
REFERENCE

+3.4V
BUFFER
AMPLIFIER

+1.8V
+3.4V

50k

+1.8V +0.2V
VOUT

VIN

COM

SELFTEST
(ST)

Figure 18. Functional Block Diagram

the inverting input and the output of this amplifier via pins
VOUT and VIN, while the noninverting input is connected internally to a +1.8 V reference. The +1.8 V is derived from a
resistor divider connected to the 3.4 V reference.

The sensors tight mechanical spacing allows it to be electrostatically deflected to full scale while operating on a 5 volt supply. A self-test is initiated by applying a TTL high level
voltage (>+2.0 V) to the ADXL50s self-test pin which causes
the chip to apply a deflection voltage to the beam which moves
it an amount equal to 50 g (the negative full-scale output of the
device). Note that the 10% tolerance of the self-test circuit is
not proportional to the sensitivity error, see Self-Test section.

BASIC CONNECTIONS FOR THE ADXL50

Figure 19 shows the basic connections needed for the ADXL50

to measure accelerations in the 50 g range with an output scale
factor 40 mV/g corresponding to a 2.5 V 0 g level, a 2.0 V fullscale swing around 0 g and a 3 dB bandwidth of approximately
1 kHz.

The output of the ADXL50s preamplifier is 1.8 V at 0 g acceleration with an output range of 0.95 V for a 50 g input, i.e.,
19 mV/g. An uncommitted buffer amplifier has been included
on-chip to enhance the users ability to offset the 0 g signal level
and to amplify and filter the signal. Access is provided to both

In general, the designer will need to take into account the initial
zero g bias when designing circuits. For the ADXL50J this offset is 1.8 V 250 mV. When microprocessors and software

ADXL50

+3.4V
REFERENCE

OSCILLATOR

SENSOR

DEMODULATOR

C2
0.022F

BUFFER
AMP

PREAMP

5
COM

2
C1

C3

+5V

8
C1

0.022F
DEMODULATOR
CAPACITOR

10

VPR

9
VIN

R3
105k
VOUT

R1
49.9k
R2
274k

Figure 19. ADXL50 Application Providing an Output Sensitivity of

40 mV/g, a +2.5 V 0 g Level and a Bandwidth of 1 kHz

REV. B

VREF
OUTPUT

+1.8V

OSCILLATOR
DECOUPLING
CAPACITOR

SELF TEST
(ST)

ADXL50
calibration are used and there is a desire to eliminate trim potentiometers, the design should leave room at either supply rail
to account for signal swing and or variations in initial zero g bias.
For example, in the circuit in Figure 19, the initial zero g bias of
250 mV will be reflected to the output by the gain of the R3/R1
network, resulting in an output offset of 526 mV worst case.
The offset, combined with a full-scale signal of 50 g, (+2.0 V)
will cause the output buffer amplifier to saturate at the supply
rail.

ranges listed by keeping R1 > 49.9 k, with the subsequent

tradeoff that the required values for R3 will become very large.
The user always has the option of adding external gain and filtering stages after the ADXL50 to make lower full-scale ranges.
Measuring Full-Scale Accelerations Less than 65 g

Applications, such as motion detection, and tilt sensing, have

signal amplitudes in the 1 g to 2 g range. Although designed for
higher full-scale ranges, the ADXL50 may be adapted for use in
C2

The full 2.25 V output swing of the buffer amplifier can be

utilized if the user is able to trim the zero-g bias to exactly
2.5 V. In applications where the full-scale range will be 25 g or
less, a bias trim such as that shown in Figure 20 will almost always be required.

C1

ADXL50

0.022F

+5V
0.1F

PRE-AMP

1.8V

BUFFER
AMP

0.022F
3
5

Table II. Recommended Resistor Values for Setting the

Circuit of Figure 20 to Several Common Full-Scale Ranges

Buffer
Gain

SF in
mV/g

R1

R3

R2

50.0
40.0
30.8
26.7
20.0
10.0

2.11
2.63
3.42
3.95
5.26
10.53

40
50
65
75
100
200

49.9 k
39.2 k
40.2 k
28.7 k
26.1 k
23.7 k

105 k
103 k
137 k
113 k
137 k
249 k

100 k
100 k
100 k
100 k
100 k
100 k

6
VREF

+3.4V
0g
LEVEL 50k
TRIM

10
VIN

VPR
R1

+1.8V

R3

R2
VX

Figure 20. ADXL50 Circuit Using the Buffer Amplifier to

Set the Output Scaling and 0 g Offset Level

low g applications; the two main design considerations are noise

and 0 g offset drift (BH, KH grades recommended).
At its full 1 kHz bandwidth, the ADXL50 will typically exhibit
1 g p-p of noise. With 50 g accelerations this is generally not a
problem, but at a 2 g full-scale level the signal-to-noise ratio
will be very poor. However, reducing the bandwidth to 100 Hz
or less considerably improves the S/N ratio. Figure 25 shows the
relationship between ADXL50 bandwidth and noise.
The ADXL50 exhibits offset drifts that are typically 0.02 g per
C but which may be as large as 0.1 g per C. With the buffer
amplifier configured for a 2 g full scale, the ADXL50 will typically drift 1/2 of its full-scale range with a 50C increase in
temperature.

Note that the value of resistor R1 should be selected to limit the

output current flowing into VPR to less than 25 A (to provide a
safety margin). For a J grade device, this current is equal to:
I PR =

VOUT

COM

The uncommitted buffer amplifier may be used to change the

output sensitivity to provide useful full-scale ranges of 50 g
and below. Table II provides recommended resistor values for
several standard ranges down to 10 g. As the full-scale range is
decreased, buffer amplifier gain is increased, and the noise contribution as a percentage of full scale will also increase. For all
ranges, the signal-to-noise ratio can be improved by reducing
the circuit bandwidth, either by increasing the demodulator capacitor, C1, or by adding a post filter using the buffer amplifier.

FS (g)

C1

VARYING THE OUTPUT SENSITIVITY AND 0 g LEVEL

USING THE INTERNAL BUFFER AMPLIFIER

There are several cures for offset drift. If a dc response is not

required, for example in motion sensing or vibration measurement
applications, consider ac coupling the acceleration signal to remove the effects of offset drift. See the section on ac coupling.

(2.05 V The peak full -scale output voltage at VPR ) 1.8 V

R1 in ohms

For a 50 g full-scale range, R1 needs to be 49.9 k or larger in

value; but at the lower full-scale g ranges, if the VPR swing is
much less, then it is possible to use much lower resistance values. For this table, the circuit of Figure 20 is used, as a 0 g offset trim will be required for most applications. In all cases, it is
assumed that the zero-g bias level is 2.5 V with an output span
of 2 V.
Note that for full scales below 20 g the self-test is unlikely to
operate correctly because the VPR pull-down current is not guaranteed to be large enough to drive R1 to the required 1.0 V
swing. In these cases, the self-test command will cause VOUT to
saturate at the rail, and it will be necessary to monitor the selftest at VPR. Self-test can remain operational at VPR for all g

Periodically recalibrating the accelerometers 0 g level is another

option. Autozero or long term averaging can be used to remove
long term drift using a microprocessor or the autozero circuit of
Figure 29. Be sure to keep the buffer amplifiers full-scale output range much larger than the measurement range to allow for
the 0 g level drift.
CALCULATING COMPONENT VALUES FOR SCALE
FACTOR AND 0 g SIGNAL LEVEL

The ADXL50 buffers scale factor is set by R3/R1 (since the

amplifier is in the inverter mode).

10

REV. B

ADXL50
As an example, if the desired span is 2.0 V for a = 50 g input,
then R3/R1 should be chosen such that
R3/R1 = VOUT Span/VPR Span = 2.00/0.95 = 2.105

(1)

where VPR span is the output from the preamplifier and VOUT
span is the buffer amplifiers output, giving
R3 = 2.105 R1

(2)

In noncritical applications, a resistor, R2, may simply be connected between VIN and common to provide an approximate
0 g offset level (see Figure 19). In this simplified configuration
R2 is found using:

SELF-TEST FUNCTION

R2 = (1.8 V R3)/(VOUT @ 0 g 1.8 V)

When used with a trim potentiometer, as in Figure 20, resistor
R2 sets the 0 g offset range and also sets the resolution of the
offset trim. A value of 100 k is typical. Increasing R2 above
this value makes trimming the offset easier, but may not provide
enough trim range to set VOUT equal to +2.5 V for all devices.
To provide an output span of 2.00 V, with a 0 g output of
+2.5 V, R1 could be set to the standard value of 49.9 k and
from Equation 2, R3 = 105 k.
R3
R3

V OUT =
(1.8 V VPR ) +
(1.8 V VX ) + 1.8 V
R1
R2

Operating the ADXL50s buffer amplifier at Gains > 2, to provide full-scale outputs of less than 50 g, may cause the self-test
output to overdrive the buffer into saturation. The self-test may
still be used in the case, but the change in the output must then
be monitored at the VPR pin instead of the buffer output.

The summing amplifier configuration allows noninteractive

trimming of offset and span. Since VPR is not always exactly
1.8 V at 0 g, it will contribute to output offset. Therefore, span
must be trimmed first, followed by 0 g offset adjustment.

Note that the value of the self-test delta is not an exact indication of the sensitivity (mV/g) of the ADXL50 and, therefore,
may not be used to calibrate the device for sensitivity error.

LOAD DRIVE CAPABILITIES OF THE V PR AND BUFFER

OUTPUTS

In critical applications, it may be desirable to monitor shifts in

the zero-g bias voltage from its initial value. A shift in the 0 g
bias level may indicate that the 0 g level has shifted which may
warrant an alarm.

The VPR and the buffer amplifier outputs are both capable of
driving a load to voltage levels approaching that of the supply
rail. However, both outputs are limited in how much current
they can supply, affecting component selection.

POWER SUPPLY DECOUPLING

VPR Output

The VPR pin has the ability to source current up to 500 A but
only has a sinking capability of 30 A which limits its ability to
drive loads. It is recommended that the buffer amplifier be used
in most applications, to avoid loading down VPR. In standard
50 g applications, the resistor R1 from VPR to VIN is recommended to have a value greater than 50 k to reduce loading
effects.
Capacitive loading of the VPR pin should be minimized. A load
capacitance between the VPR pin and common will introduce an
offset of approximately 1 mV for every 10 pF of load. The VPR
pin may be used to directly drive an A/D input or other source
as long as these sensitivities are taken into account. It is always
preferable to drive A/D converters or other sources using the
buffer amplifier (or an external op amp) instead of the VPR pin.
The buffer output can drive a load to within 0.25 V of either
power supply rail and is capable of driving 1000 pF capacitive

REV. B

The digital self-test input is compatible with both CMOS and

TTL signals. A Logic l applied to the self-test (ST) input will
cause an electrostatic force to be applied to the sensor which
will cause it to deflect to the approximate negative full-scale output of the device. Accordingly, a correctly functioning accelerometer will respond by initiating an approximate 1 volt output
change at VPR. If the ADXL50 is experiencing an acceleration
when the self-test is initiated, the VPR output will equal the algebraic sum of the two inputs. The output will stay at the self-test
level as long as the ST input remains high and will return to the
0 g level when the ST voltage is removed.
A self-test output that varies more than 10% from the nominal
1.0 V change indicates a defective beam or a circuit problem
such as an open or shorted pin or component.

For Figure 20, the circuit transfer function is:

Buffer Amplifier Output

loads. Note that a capacitance connected across the buffer feedback resistor for low-pass filtering does not appear as a capacitive load to the buffer. The buffer amplifier is limited to
sourcing or sinking a maximum of 100 A. Component values
for the resistor network should be selected to ensure that the
buffer amplifier can drive the filter under worst case transient
conditions.

The ADXL50 power supply should be decoupled with a 0.1 F

ceramic capacitor from +5 V pin of the ADXL50 to common
using very short component leads. For other decoupling considerations, see EMI/RFI section.
OSCILLATOR DECOUPLING CAPACITOR, C2

An oscillator decoupling capacitor, C2, is used to remove

1 MHz switching transients in the sensor excitation signal, and
is required for proper operation of the ADXL50. A ceramic capacitor with a minimum value of 0.022 F is recommended
from the oscillator decoupling capacitor pin to common. Small
amounts of capacitor leakage due to a dc resistance greater than
l M will not affect operation (i.e., a high quality capacitor is
not needed here). As with the power supply bypass capacitor,
very short component leads are recommended. Although
0.022 F is a good typical value, it may be increased for reasons
of convenience, but doing this will not improve the noise performance of the ADXL50.

11

ADXL50
DEMODULATOR CAPACITOR, C1

For example, to reduce the average power to 5 mW from its

typical 50 mW, the power should be on 10% of the time. With
the power on for 1 ms and off for 9 ms, a maximum sample rate
of 100 Hz is achievable. Further reduction in average power can
be realized with lower sample rates.

The demodulator capacitor is connected across Pins 2 and 3 to

filter the demodulated signal from the sensor beam and to set
the bandwidth of the force balance control loop. This capacitor
may be used to approximately set the bandwidth of the accelerometer. A capacitor is always required for proper operation.
The frequency response of the ADXL50 exhibits a single pole
roll-off response whose nominal 3 dB frequency is set by the
following equation:

+5V

POWER
SUPPLY
(V)
0V

f3 dB = (28.60/C1 in F) 40%
VFINAL

A nominal value of 0.022 F is recommended for C1. In general, the design bandwidth should be set 40% higher than the
minimum desired system bandwidth due to the 40% tolerance.
A minimum value of 0.015 F is required, (over temperature
and system life), to prevent device instability or oscillation. The
demodulation capacitor should be a low leakage, low drift ceramic type with an NPO (best) or X7R (good) dielectric.

VOUT
(V)

0.4
0.6
TIME ms

0.8

1.0

Figure 21. Power-On Settling Time when Power Cycling

SYSTEM BANDWIDTH CONTROL AND POST
FILTERING

Unlike piezoresistive sensors, the resonant frequency of the

ADXL50s capacitive sensor element is typically greater than
20 kHz and does not limit the useful bandwidth of the device.
Usually, the resonant frequency of the beam appears as a peak
in the bandwidth response at approximately 24 kHz with a Q of
3 to 4, as shown in Figure 22.
When using the recommended 0.022 F demodulator capacitor,
be advised that the nominal 1300 Hz pole it establishes within
the device can vary 40%. Therefore, if additional low-pass
filtering is usedat frequencies much above 600 Hzthe two
poles may interact and result in a net circuit bandwidth that is
lower than expected.
190

REDUCING THE AVERAGE POWER CONSUMPTION OF

THE ADXL50

AMPLITUDE
19
0.022F
1.9

0.015F
0.010F
0.005F

0.19

The ADXL50 is a versatile accelerometer that can be used in a

wide variety of applications. In some battery powered applications, such as shipping recorders, power consumption is a critical parameter. The ADXL50 typically draws 10 mA current
from a 5 V power supply which may exceed the power budgeted
for the accelerometer.
For such applications, the ADXL50 can be successfully power
cycled, where the power is turned on only during the period
when data is sampled. Figure 21 illustrates the power-on settling
of the ADXL50 during cycling where the output amplifier has
a gain of one with no filtering. The settling time-constant is
approximately 0.12 ms, waiting l ms before sampling ensures
maximally accurate readings.

0.2

90
PHASE
180
100

1k
FREQUENCY Hz

PHASE SHIFT Degrees

Care should be taken to reduce or eliminate any leakage paths

from the demodulator capacitor pins to common or to the +5 V
pin. Even a small imbalance in the leakage paths from these pins
will result in offset shifts in the zero-g bias level. As an example,
an unbalanced parasitic resistance of 30 M from either
demodulator pin to ground will result in an offset shift at VPR of
approximately 50 mV. Conformal coating of PC boards with a
high impedance material is recommended to avoid leakage problems due to aging or moisture.

T0

SENSITIVITY mV/g

In general, its best to use the recommended 0.022 F capacitor

across the demodulator pins and perform any additional lowpass filtering using the buffer amplifier. Using a large denominator capacitor for low-pass filtering has the disadvantage that the
capacitive sensor will be slow to respond to rapid changes in
acceleration and, therefore, the full shock survivability of the
device could be compromised. The use of the buffer for lowpass filtering generally results in smaller capacitance values and
better overall performance. It is also a convenient and more precise way to set the system bandwidth. Post filtering allows bandwidth to be controlled accurately by component selection and
avoids the 40% demodulation tolerance. Note that signal noise
is proportional to the square root of the bandwidth of the
ADXL50 and may be a consideration in component selection
see section on noise.

10k

Figure 22. Frequency Response of the ADXL50 for Various

Demodulator Capacitors

12

REV. B

ADXL50
C2
4
RECOMMENDED COMPONENT VALUES FOR VARIOUS
FULL SCALE RANGES AND A 300Hz BANDWIDTH

ADXL50

0.022F
C1

FULL
SCALE

mV
per g

3dB
BW (Hz)

R1a
k

R1b
k

R3
k

R2
k

C4
F

10 g

200

300

21.5

249

100

0.0022

20 g

100

300

23.7

137

100

0.0039

40 g

50

300

10

34

105

100

0.0056

50 g

40

300

10

45.3

105

100

0.0056

0.1F

BUFFER
AMP

PRE-AMP

+5V

1.8V

0.022F

9
VOUT

3
C1
5
COM
6

1
3dB BW =
2 R3 C4

10

VPR

+3.4V
REF

VIN

C4

R1a R1b
OPTIONAL SCALE
FACTOR TRIM*

0g
LEVEL
TRIM

R3

50k
R2

*TO OMIT THE OPTIONAL SCALE FACTOR TRIM,

REPLACE R1a AND R1b WITH A FIXED VALUE 1%
METAL FILM RESISTOR. SEE VALUES SPECIFIED
IN TABLE II.

Figure 23. Using the Buffer Amplifier to Provide One Pole Post Filtering Plus Scale Factor and 0 g Level Trimming

need to be twice as large as its 100 Hz value or 0.012 F 2 =

0.024 F. The closest standard value of 0.022 F should then
be used.

ONE POLE POST FILTERING

Figure 23 shows the ADXL50 buffer amplifier connected to

provide one pole post filtering, 0 g offset trimming, and output
scaling. The table included with the figure lists practical component values for various full-scale g levels and approximate circuit
bandwidths. For bandwidths other than those listed, use the
formula:

TWO POLE POST FILTERING

Figure 24 shows a circuit which uses the ADXL50s buffer amplifier to provide two pole post filtering. An AD820 external op
amp allows noninteractive adjustment of 0 g offset and scale
factor. Component values for the two pole filter were selected to
operate the buffer at unity gain with a Q of one.

Capacitor C4 in Farads = 1/(2 R3 in Ohms

3 dB BW in Hertz)
or simply scale the value of capacitor C4 accordingly; i.e., for a
20 g application with a 50 Hz bandwidth, the value of C4 will

0.022F

ADXL50
4

C2
C1
0.022F

2 POLE FILTER
COMPONENT VALUES
3dB
BW(Hz) C3F

0.1F

BUFFER
AMP

PRE-AMP

+5V

1.8V

OPTIONAL CAPACITOR
FOR 3 POLE FILTERING

9
VOUT

R5

C1
C4F

5
COM

+5V
8

10

300

0.027

0.0033

100

0.082

0.01

30

0.27

0.033

R1

10

0.82

0.1

82.5k

VREF

C4

VIN

VPR

42.2k

R5
C3

0.01F
R4a

R3
82.5k

SCALE
FACTOR
TRIM

2 POLE FILTER
+3.4V

OFFSET & SCALING AMPLIFIER

COMPONENT VALUES
FULL
SCALE

mV
per g

GAIN

R4a
k

R4b
k

R5
k

40.2k

10 g

200

10.53

21.5

249

20k

20 g

100

5.26

23.7

137

40 g

50

2.63

10

34

105

50 g

40

2.11

10

45.3

105

71.5k

R4b

3 AD820
4

OFFSET AND
SCALING
AMPLIFIER

R6

R7

0g
LEVEL
TRIM

Figure 24. Circuit Providing Two Pole Post Filtering and 0 g Offset and Scale Factor Trimming

REV. B

13

OUTPUT

ADXL50
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. The nominal buffer amplifier output will be +1.8 V 19 mV/g. Note that
the ADXL50s self-test will be fully functional since the buffer
amplifier is operated at unity gain and resistor R1 is large. The
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. The external op amp may be omitted in high g, low gain applications.
NOISE CONSIDERATIONS

The output noise of the ADXL50 scales with the square root of
its bandwidth. The noise floor may be reduced by lowering the
bandwidth of the ADXL50 either by increasing the value of the
demodulator capacitor or by adding an external filter.

Table III.

Nominal Peak-toPeak Value

% of Time that Noise will Exceed

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 VPR TO BUFFER INPUT

If a dc response is not required, as in applications such as motion 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 considerably without increasing the 0 g level drift. The most effective
way to ac couple the ADXL50 is between the preamp output at
VPR and the buffer input, VIN, as shown in Figure 26.

The typical rms noise of the ADXL50J with a bandwidth of

100 Hz and a noise density of 125 V/Hz is estimated as
follows:
Noise (rms) = (125 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 approximately 0.4 g p-p. The ADXL50 noise is characteristic of white
noise. Typical rms and p-p noise for various 3 dB bandwidths is
estimated in Figure 25.

ADXL50
1.8V

PRE-AMP

BUFFER
AMP

VPR

0.26

1.4

0.21

1.0

0.16

0.73

0.11

0.33

8
VPR

NOISE LEVEL g rms

NOISE LEVEL g p-p

VOUT

1.7

100
3dB BANDWIDTH Hz

R3

Figure 26. AC Coupling the VPR Output to the Buffer Input

Using this configuration, the systems ac response is now rolled

offat the low frequency end at FL, and at the high frequency
end at FH. The normalized frequency response of the system
can be seen in Figure 27.

0
10

VIN

R1

R2

0.05

10
C4

1k

The low frequency roll-off, FL, due to the ac coupling network

is:
FL = 1/(2 R1 C4)

Figure 25. ADXL50 Noise Level and Resolution vs. 3 dB

Bandwidth

Because the ADXL50s noise is for all practical purposes Gaussian 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. Table III is useful for estimating the probabilities of exceeding various peak values, given the rms value.

The high frequency roll-off FH 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 VPR output). The self-test feature, if used,
must be monitored at VPR, rather than at the buffer output.

14

REV. B

ADXL50
and package orientation affect the ADXL50s output (TO-100
package shown). Note that the output polarity is that which appears at VPR; the output at VOUT will have the opposite sign.
With its axis of sensitivity in the vertical plane, the ADXL50
should register a 1 g acceleration, either positive or negative, depending on orientation. With the axis of sensitivity in the horizontal plane, no acceleration (0 g) should be indicated.

NORMALIZED OUTPUT LEVEL dB

+20

+10
LOW FREQUENCY ROLL-OFF
(F L ) DUE TO AC COUPLING
0

10
HIGH FREQUENCY ROLL-OFF (FH )
DUE TO DEMODULATOR BANDWIDTH
20

0g
(a)

0g
(b)

1g
(c)

+1g
(d)

30
1

10

100
FREQUENCY Hz

1k

10k

INDICATED POLARITY IS THAT

OCCURING AT V PR .

Figure 27. Normalized Output Level vs. Frequency for a

Typical Application Using AC Coupling Between VPR and
Buffer Amplifier
MINIMIZING EMI/RFI

The architecture of the ADXL50 and its use of synchronous demodulation make the device immune to most electromagnetic
(EMI) and radio frequency (RFI) interference. The use of synchronous demodulation allows the circuit to reject all signals except those at the frequency of the oscillator driving the sensor
element. However, the ADXL50 does have a sensitivity to RFI
that is within 5 kHz of the internal oscillators nominal frequency of 1 MHz. The internal oscillator frequency will exhibit
part to part variation in the range of 0.6 MHz to 1.4 MHz.
In general the effect is difficult to notice as the interference
must match the internal oscillator within 5 kHz and must be
large in amplitude. For example: a 1 MHz interference signal of
20 mV p-p applied to the +5 V power supply pin will produce a
200 mV p-p signal at the VPR pin if the internal oscillator and
interference signals are matched exactly. If the same 20 mV interference is applied but 5 kHz above or below the internal
oscillators frequency, the signal level at VPR will only be 20 mV
p-p in amplitude.
Power supply decoupling, short component leads (especially for
capacitors C1 and C2), physically small (surface mount, etc.)
components and attention to good grounding practices all help
to prevent RFI and EMI problems. Please consult the factory
for applications assistance in instances where this may be of
concern.
SELF-CALIBRATING THE ADXL50

If a calibrated shaker is not available, both the 0 g level and

scale factor of the ADXL50 may be easily set to fair accuracy by
using a self-calibration technique based on the 1 g (average) acceleration of the earths gravity. Figure 28 shows how gravity

REV. B

Figure 28. Using the Earths Gravity to Self-Calibrate the

ADXL50

To self-calibrate the ADXL50, place the accelerometer on its

side with its axis of sensitivity oriented as shown in a. The 0 g
offset potentiometer, Rt, is then roughly adjusted for midscale:
+2.5 V at the buffer output. If the optional scale factor trimming is to be used, it should be adjusted next.
Next, the package axis should be oriented as in c (pointing
down) and the output reading noted. The package axis should
then be rotated 180 to position d and the scale factor potentiometer, R1a, adjusted so that the output voltage indicates a
change of 2 gs in acceleration. For example, if the circuit scale
factor at the buffer output is 100 mV per g, then the scale factor
trim should be adjusted so that an output change of 200 mV is
indicated.
Adjusting the circuits scale factor will have some effect on its 0
g level so this should be readjusted, as before, but this time
checked in both positions a and b. If there is a difference in
the 0 g reading, a compromise setting should be selected so that
the reading in each direction is equidistant from +2.5 V. Scale
factor and 0 g offset adjustments should be repeated until both
are correct. Temporarily placing a capacitor across the buffer
amplifiers feedback resistor will reduce output noise and so aid
in trimming the device. Note that, for high full-scale g ranges,
2 g may be a very small fraction of the full-scale range and
device nonlinearity will, therefore, affect the circuits high g level
accuracy.

15

ADXL50
Compensating for the 0 g Drift of the ADXL50 Accelerometer

+15

The circuit of Figure 29 provides a linear temperature compensation for the ADXL50. Figure 30 shows the 0 g drift over temperature for a typical ADXL50 with and without this circuit. As
shown by Figure 30, the linear portion of the drift curve has
been subtracted out. In effect, the curve has been rotated counterclockwise until it is horizontal, leaving just the bow of the
curve: that portion which is not linear. As shown by Figure 30,
over a +25C to +70C range, a 10 reduction in drift is achieved.

+10
LOW TEMP TRIM

0
5
10

COMPENSATED XL50
15
20
25

UNCOMPENSATED XL50

30
35
40
45
55

35

+85

+105

Dimensions shown in inches and (mm).

REFERENCE PLANE

0.185 (4.70)
0.165 (4.19)

VOUT

0.750 (19.05)
0.500 (12.70)

0.160 (4.06)
0.110 (2.79)

10

8
VPR

49.9k

VIN

499k

R1

R3

+5V

TEMPERATURE
COMPENSATED
ACCELERATION
OUTPUT

1A/K
3101k
R5
TC
COMP
SET

10k

RB

TEST
0.01F POINT
"A"
7

2 AD820
4

BRIDGE
BALANCE
RA

30k
R6

30k
R8

0.370 (9.40)
0.335 (8.51)

20k

0.045 (1.14)
0.027 (0.69)

9
3

0.040 (1.02) MAX

0.045 (1.14)
0.010 (0.25)

0.230 (5.84)
BSC

10
1
0.034 (0.86)
0.027 (0.69)

SEATING PLANE

0g OUTPUT
LEVEL

R9
6
TEMPCO
AMPLIFIER

8
0.115
(2.92)
BSC

49.9k
R2

310k
R7
+5V

5
0.335 (8.51)
0.305 (7.75)

R10
RC
20k

25k

PRINTED IN U.S.A.

6
+3.4V
REF

1k

+65

Contact the Analog Devices Literature Center for a copy of

publication number G2112, the Accelerometer Application
Guide. This includes all current application notes and data
sheets for Analog Devices accelerometers.

PRE-AMP

C1

R4

+45

ADXL50 Application Literature Available

0.1F
C3

BUFFER
AMP

0.022F

500

+25

Figure 30. ADXL50 0 g Drift With and Without the Compensation Circuit of Figure 29

+5V

1.8V

C2

AD590

+5

OUTLINE DIMENSIONS
ADXL50

COM

15

TEMPERATURE C

0.022F

C1

C1808b53/96

MEASURED 0g DRIFT mV

The circuit of Figure 29 is essentially a temperature sensor

coupled to a Whetstone bridge. The AD590 provides a 1 A/K
current output whose voltage scale factor is set by resistor RA.
The bridge circuit subtracts out the nominal 298 mV output of
the AD590 at +25C and leaves only the change in temperature,
which is what is needed. Without the bridge, the 298 mV room
temperature offset would swamp the much smaller change
in output with temperature.
Resistors R5 and R6 form a resistor divider (one half of the
bridge) which divides down the +3.4 V reference output of the
ADXL50 to 0.3 V which appears at the noninverting input of
the AD820 op amp. Resistors R7 and R8 form the other half of
the bridge, and because they have the same ratio as R5 and R6,
the op amp will have a +3.4 V output at room temperature.

HIGH TEMP TRIM

+5

CALIBRATION PROCEEDURE:
AT T MIN OR LOWER TEMP CAL POINT...
1.
2.
3.
4.
5.
6.
7.
8.
9.

SET RB ALL THE WAY TO ONE SIDE.

ADJUST RA FOR +3.4V AT TEST POINT "A."
SET RC FOR +2.5V V OUT (AT PIN 9 OF ADXL50).
TEMPORARILY CONNECT A 1.5k RESISTOR BETWEEN
THE CENTER OF RB AND GROUND.
ADJUST RB FOR +2.5V AT VOUT .
REMOVE THE 1.5k RESISTOR. VOUT SHOULD NOT CHANGE.
GO TO T MAX OR HIGH TEMP CAL POINT.
READJUST RB FOR +2.5V.
CALIBRATION COMPLETE.

Figure 29. ADXL50 0 g Drift Compensation Circuit

16

REV. B