Accelerometer Design and Accelerometer Design and Applications Applications

James Doscher James Doscher Micromachining Evangelist Micromachining Evangelist

Objectives Objectives
What can II do with an accelerometer? What can do with an accelerometer? Why consider an accelerometer? Why consider an accelerometer? Product technology overview Product technology overview Performance requirements for various applications Performance requirements for various applications Accelerometers from the Point of View of Signal Accelerometers from the Point of View of Signal Conditioning and Data Acquisition Conditioning and Data Acquisition
Inertial Inertial Vibration Vibration Tilt Tilt

What can II do with an What can do with an accelerometer? accelerometer?

Inertial measurement of velocity and position Inertial measurement of velocity and position
Acceleration single integrated for velocity Acceleration single integrated for velocity Acceleration double integrated for position Acceleration double integrated for position

Vibration and shock measurement Vibration and shock measurement

Measuring vibration for machine health Measuring vibration for machine health Motion and shock detection Motion and shock detection

Measurement of gravity to determine orientation Measurement of gravity to determine orientation

Tilt and inclination Tilt and inclination Position in 2 and 3 dimensional space Position in 2 and 3 dimensional space Can only be done with accelerometers that have DC response Can only be done with accelerometers that have DC response

Inertial measurement Inertial measurement

Non-contact method to Non-contact method to measure acceleration, measure acceleration, velocity, and distance velocity, and distance traveled traveled Velocity is the integral Velocity is the integral of Acceleration of Acceleration Distance is the integral Distance is the integral of Velocity (double of Velocity (double integral of integral of acceleration) acceleration)
1 1 0.5 0.5 0 0 -0.5 -0.5 -1 -10 0 Acceleration M/s2 Acceleration M/s2 5 5 10 10 15 15 20 20 25 25 30 30 Time Time 35 35 40 40 45 45 50 50

10 10 5 5 0 0 0 0 5 5 10 10 15 15

Velocity M/s Velocity M/s

20 20

25 25 30 30 Time Time

35 35

40 40

45 45

50 50

200 200 150 150 100 100 50 50 0 0 0 0

Distance M Distance M

5 5

10 10

15 15

20 20

25 25 30 30 Time Time

35 35

40 40

45 45

50 50

Inertial sensing for Inertial sensing for Airbag Crash Detection Airbag Crash Detection
Single accelerometer replaces 3 to 5 g switches Single accelerometer replaces 3 to 5 g switches Signature analysis of acceleration, velocity and Signature analysis of acceleration, velocity and distance profile is used to distinguish a crash distance profile is used to distinguish a crash from a non-crash event. from a non-crash event.
50 50 25 25

-25 -25

-50 0 -50 0

5 5

10 10

15 15

20 20

25 25

30 30

35 35

40 40

45 45

50 50

Using accelerometers in Using accelerometers in machine condition monitoring machine condition monitoring
Monitor and predict the health and condition of Monitor and predict the health and condition of moving machinery moving machinery
Predict failure before incurring expensive unplanned downtime Predict failure before incurring expensive unplanned downtime Increase time between preventative maintenance cycles Increase time between preventative maintenance cycles Improve product reliability and safety Improve product reliability and safety Gain competitive advantage Gain competitive advantage

Accelerometers monitor the vibration level. Accelerometers monitor the vibration level.
Analysis of acceleration, velocity, and displacement Analysis of acceleration, velocity, and displacement Time domain and frequency domain analysis Time domain and frequency domain analysis Detect out of balance, bearing failure, bushing failure etc. Detect out of balance, bearing failure, bushing failure etc.

Using accelerometers in Using accelerometers in machine condition monitoring machine condition monitoring
XL150/250 Capable of machine health measurements XL150/250 Capable of machine health measurements for motors, pumps, compressors for motors, pumps, compressors
Preconditioned Analog output eliminates need for charge amplifier Preconditioned Analog output eliminates need for charge amplifier Scale factor insensitive to temperature, 10mg resolution Scale factor insensitive to temperature, 10mg resolution Multi-axis sensor reduces cost, simplifies packaging Multi-axis sensor reduces cost, simplifies packaging XL105 will increase BW to 10kHz and increase SNR XL105 will increase BW to 10kHz and increase SNR
PC Data Acquisition PC Data Acquisition (VME, PMCIA,) (VME, PMCIA,) Data Analysis -Acceleration -Velocity -Displacement Data analyzed in: Time Domain Frequency Domain

Triaxial Accelerometer Triaxial Accelerometer Xl150+XL250 Xl150+XL250 50g FS with 10mg resolution 50g FS with 10mg resolution

Tilt sensing example Tilt sensing example

Pitch and Roll Sensing Pitch and Roll Sensing
Requires one accelerometer for each axis Requires one accelerometer for each axis Compass Correction Compass Correction Machine position Machine position Computer input devices Computer input devices

3D Orientation in Space 3D Orientation in Space

Requires three accelerometers, X,Y, Z Requires three accelerometers, X,Y, Z

6 Degree of Freedom Sensing 6 Degree of Freedom Sensing

Requires 3 accelerometers and 3 gyros Requires 3 accelerometers and 3 gyros

Why should II consider Why should consider using an accelerometer? using an accelerometer?
A variety of new devices and technologies are on A variety of new devices and technologies are on the market the market Prices have gone down dramatically Prices have gone down dramatically Cost/Performance ratio is improving Cost/Performance ratio is improving

What is a Micromachine? What is a Micromachine? ADIs implementation ADIs implementation

Surface Micromachined Mechanical Element Surface Micromachined Mechanical Element
Combination of springs, masses, motion sensing and actuation cells Combination of springs, masses, motion sensing and actuation cells

BiMOS Circuitry for On-Board Signal Conditioning BiMOS Circuitry for On-Board Signal Conditioning
Oscillators and Demodulators Oscillators and Demodulators Amplifiers and Filters Amplifiers and Filters Self Test Circuitry Self Test Circuitry

All On a Single I.C. Chip All On a Single I.C. Chip

Fabricated in a Standard I.C. Wafer Fab Fabricated in a Standard I.C. Wafer Fab Uses Standard IC Photo Lithographic, Etch and Implant Technologies Uses Standard IC Photo Lithographic, Etch and Implant Technologies Packaged Using Specially Developed Techniques to Protect Moving Parts Packaged Using Specially Developed Techniques to Protect Moving Parts Highly Reliable, Low Cost Solution to Motion Sensors Highly Reliable, Low Cost Solution to Motion Sensors

Die size and cost per axis Die size and cost per axis is decreasing is decreasing

XL50 Original

XL150 New

XL250 2 Channel New

New technology reduces the New technology reduces the cost of sensing cost of sensing
Micromachining uses standard integrated Micromachining uses standard integrated processing techniques processing techniques Leverages existing technology base Leverages existing technology base Benefits of batch processing Benefits of batch processing Uses standard IC packages Uses standard IC packages Easy to handle and integrate with other IC devices Easy to handle and integrate with other IC devices Some devices now have PWM outputs that can Some devices now have PWM outputs that can interface to a microcontroller // DSP without an interface to a microcontroller DSP without an A to D converter A to D converter

Sensor operation Sensor operation

Sensor Forms Differential Capacitor Sensor Forms Differential Capacitor
TOP VIEW TOP VIEW
TETHER TETHER PROOF MASS PROOF MASS (BEAM) (BEAM) APPLIED APPLIED ACCELERATION ACCELERATION

FIXED FIXED OUTER OUTER PLATES PLATES

ANCHOR ANCHOR

CS1 < CS2 CS1 < CS2

SENSOR AT REST SENSOR AT REST

RESPONDING TO AN APPLIED ACCELERATION RESPONDING TO AN APPLIED ACCELERATION (MOVEMENT SHOWN IS GREATLY EXAGERATED) (MOVEMENT SHOWN IS GREATLY EXAGERATED)

Sensor operation; Sensor operation; ADIs implementation ADIs implementation

Folded tethers have more consistent spring constants, Folded tethers have more consistent spring constants, leading to better part to part consistency leading to better part to part consistency
TOP VIEW TOP VIEW
PROOF MASS PROOF MASS (BEAM) (BEAM) TETHER TETHER APPLIED APPLIED ACCELERATION ACCELERATION

CS1 < CS2 CS1 < CS2 FIXED FIXED OUTER OUTER PLATES PLATES ANCHOR ANCHOR

Self test operation Self test operation

Extra fixed outer plates may be added which when exited, Extra fixed outer plates may be added which when exited, force the proof mass to move. So you can electronically test force the proof mass to move. So you can electronically test the accelerometer the accelerometer

Additional fixed outer plates are electrically Additional fixed outer plates are electrically excited to induce movement of the proof mass. excited to induce movement of the proof mass. Acceleration is measured by the standard fixed Acceleration is measured by the standard fixed plates as usual. plates as usual.

Surface Micromachining Surface Micromachining

Selective Etching Selective Etching
Standard IC Photolithographic Standard IC Photolithographic Processes Processes 1.6m Thick Sacrificial Oxide Layer 1.6m Thick Sacrificial Oxide Layer Deposited 2.0m Polysilicon Layer Deposited 2.0m Polysilicon Layer
SENSOR SAC. OXIDE SUBSTRATE

Resulting in 3 Dimensional Resulting in 3 Dimensional Structure Structure

Suspended Above Substrate Suspended Above Substrate Free to Move in 3 Dimensions Free to Move in 3 Dimensions
SENSOR SUBSTRATE

Surrounding Silicon for Surrounding Silicon for Integrated Circuits Integrated Circuits
BiMOS-2 with 3m Design Rules BiMOS-2 with 3m Design Rules

Interesting facts Interesting facts

0.1grams Proof Mass 0.1grams Proof Mass 0.1pF per Side for the Differential Capacitor 0.1pF per Side for the Differential Capacitor 20aF (10-18f) Smallest Detectable Capacitance Change 20aF (10-18f) Smallest Detectable Capacitance Change Total Capacitance Change for Full-scale is 10fF Total Capacitance Change for Full-scale is 10fF 1.3m Gaps Between Capacitor Plates 1.3m Gaps Between Capacitor Plates 0.2 Minimum Detectable Beam Deflection (one tenth 0.2 Minimum Detectable Beam Deflection (one tenth of an Atomic diameter) of an Atomic diameter) 1.6 m Between the Suspended Beam and Substrate 1.6 m Between the Suspended Beam and Substrate 10 to 22kHz Resonant Frequency of Beam 10 to 22kHz Resonant Frequency of Beam

Design focus Design focus

Tilt and Inertial Tilt and Inertial
DC Accuracy DC Accuracy Null Drift with time and temperature Null Drift with time and temperature mg Resolution mg Resolution

Vibration and shock Vibration and shock

Dynamic Range 1ug to 100g Dynamic Range 1ug to 100g Bandwidth Bandwidth Amplitude Stability Amplitude Stability

Common Issues in Data Acquisition Common Issues in Data Acquisition

Bandwidth Bandwidth Resolution Resolution Accuracy Accuracy

Choosing an appropriate Choosing an appropriate technology technology

Will you be measuring an AC phenomenon like Will you be measuring an AC phenomenon like vibration, or a DC phenomenon like gravity or vibration, or a DC phenomenon like gravity or constant acceleration? constant acceleration? What is the maximum g range you expect? What is the maximum g range you expect? What is the smallest signal you need to detect? What is the smallest signal you need to detect? What is the maximum frequency required? What is the maximum frequency required? What level of stability is required for the null and What level of stability is required for the null and sensitivity of the accelerometer? sensitivity of the accelerometer? What about size and power consumption? What about size and power consumption? Will you mount to a circuit board ? Will you mount to a circuit board ?

Comparison of approaches Comparison of approaches to acceleration sensing to acceleration sensing

Piezo-film (Vibration, shock) Piezo-film (Vibration, shock)
Inexpensive Inexpensive AC Response only AC Response only Sense many things besides motion (sound, temperature, pressure) Sense many things besides motion (sound, temperature, pressure)

Electromechanical Servo (Tilt, Inertial) Electromechanical Servo (Tilt, Inertial)

DC accurate, low frequency only DC accurate, low frequency only Tend to be fragile Tend to be fragile Expensive Expensive

Comparison of approaches Comparison of approaches to acceleration sensing to acceleration sensing

Piezoelectric (Vibration, Shock) Piezoelectric (Vibration, Shock)
Wide-dynamic range Wide-dynamic range AC Response only AC Response only Can be expensive, depending on the package Can be expensive, depending on the package

Liquid tilt sensors (Tilt) Liquid tilt sensors (Tilt)

DC response DC response Inexpensive Inexpensive Fragile Fragile Slosh problems Slosh problems

Comparison of approaches Comparison of approaches to acceleration sensing to acceleration sensing

Bulk Micromachined Piezo Resistive (Tilt, Vibration, Inertial) Bulk Micromachined Piezo Resistive (Tilt, Vibration, Inertial)
DC Response DC Response Difficult to trim, poor temperature performance Difficult to trim, poor temperature performance Can be expensive Can be expensive

Bulk Micromachined Capacitive (Tilt, Vibration, Inertial) Bulk Micromachined Capacitive (Tilt, Vibration, Inertial)
DC Response DC Response Good DC accuracy, low noise Good DC accuracy, low noise Expensive Expensive

Surface Micromachined Capacitive (Tilt, Vibration, Inertial) Surface Micromachined Capacitive (Tilt, Vibration, Inertial)
DC Response DC Response Standard IC form factors Standard IC form factors Inexpensive Inexpensive

Comparison of various Comparison of various technologies technologies

High

There are orders of magnitude of cost and performance!

Piezo-electric, Electromechanical Servo Vibrational Bulk Capacitive

Performance (Accuracy)

Surface MM Capacitive

Peizo-film Low $1.00 $10.00 $ $100.00 $1000.00

Future roadmap Future roadmap

High

There are orders of magnitude of cost and performance!

Piezo-electric, Electromechanical Servo Vibrational Bulk Capacitive

Performance (Accuracy)

Surface MM Capacitive

Peizo-film Low $1.00 $10.00 $ $100.00 $1000.00

Overview: Specifications Overview: Specifications

Zero g offset (i.e. voltage output at 0g) Zero g offset (i.e. voltage output at 0g)
May vary from sensor to sensor May vary from sensor to sensor Can be trimed out at system level using the Earths gravity as Can be trimed out at system level using the Earths gravity as reference input reference input

Sensitivity (i.e. voltage output per g) Sensitivity (i.e. voltage output per g)
May vary from sensor to sensor May vary from sensor to sensor Can be adjusted at system level using the Earths gravity as Can be adjusted at system level using the Earths gravity as reference input reference input

Noise Noise
Noise determines the minimum resolution of the sensor Noise determines the minimum resolution of the sensor Noise floor can be lowered by restricting bandwidth if the noise Noise floor can be lowered by restricting bandwidth if the noise is Gaussian is Gaussian

Overview: Specifications Overview: Specifications

Temperature Range Temperature Range Bias drift with temperature Bias drift with temperature
How does the zero g output change with temperature How does the zero g output change with temperature Can be trimmed out at system level by several methods Can be trimmed out at system level by several methods

Sensitivity drift with temperature Sensitivity drift with temperature

How does the output per g change with temperature How does the output per g change with temperature Difficult to trim out at system level Difficult to trim out at system level

Bandwidth Bandwidth Power consumption Power consumption

Can power cycling be used to minimize power consumption Can power cycling be used to minimize power consumption

Acceleration is a Vector Acceleration is a Vector

Accelerometer Acceleration Vector A Ax Ay

Accelerometer Axis of Sensitivity

Ax = A Cos ()

The accelerometer has an axis of sensitivity The accelerometer has an axis of sensitivity An acceleration has a direction and a An acceleration has a direction and a magnitude relative to the accelerometer axis magnitude relative to the accelerometer axis of sensitivity (or other reference frame). of sensitivity (or other reference frame).

Acceleration is a vector Acceleration is a vector

theta 0 1 2 3 4 5 10 20 30 45 60 75 85 86 87 88 89 90 % of signal appearing at output 100.00% 99.98% 99.94% 99.86% 99.76% 99.62% 98.48% 93.97% 86.60% 70.71% 50.00% 25.88% 8.72% 6.98% 5.23% 3.49% 1.75% 0.00% Output in g's for a 1g applied acceleration 1.0000 0.9998 0.9994 0.9986 0.9976 0.9962 0.9848 0.9397 0.8660 0.7071 0.5000 0.2588 0.0872 0.0698 0.0523 0.0349 0.0175 0.0000

How the accelerometer How the accelerometer output changes with tilt output changes with tilt
Table 4: How the X and Y axis respond to changes in tilt

X Output X axis Delta per Orientation to X Output (g) Degree of Tilt (mg) Horizon (deg) -90 -1.000 -0.2 -75 -0.966 4.4 -60 -0.866 8.6 -45 -0.707 12.2 -30 -0.500 15.0 -15 -0.259 16.8 0 0.000 17.5 15 0.259 16.9 30 0.500 15.2 45 0.707 12.4 60 0.866 8.9 75 0.966 4.7 90 1.000 0.2

+90 deg 0 deg -90 deg

1g

Y Output (g) Delta per Y Output (g) Degree of Tilt (mg) 0.000 17.5 0.259 16.9 0.500 15.2 0.707 12.4 0.866 8.9 0.966 4.7 1.000 0.2 0.966 -4.4 0.866 -8.6 0.707 -12.2 0.500 -15.0 0.259 -16.8 0.000 -17.5

How a DC accelerometer How a DC accelerometer responds to tilt responds to tilt

V(out) at 500mV/g

Output of Accelerometer Used as a Tilt Sensor

500 400 300 200

= arcsin[(V(out)-V(zero g))/ (1g x Scale factor(V/g))] = arcsin[(V(out)-V(zero g))/ (1g x Scale factor(V/g))]

1g 1g Acceleration Acceleration

100 0 -100 -200 -300 -400 -500 Degrees 90 80 70 60 50 40 30 20 10 0 -10 -20 -30 -40 -50 -60 -70 -80 -90

Data acquisition: Tilt Data acquisition: Tilt

How does it work How does it work
Accelerometer measures the static gravity field Accelerometer measures the static gravity field Acceleration of 9.8m/s 2 = 1g Acceleration of 9.8m/s 2 = 1g Changing the tilt (along the sensitive axis) changes Changing the tilt (along the sensitive axis) changes acceleration vector acceleration vector = arcsin[(V(out)-V(zero g))/ (1g x Scale factor(V/g))] = arcsin[(V(out)-V(zero g))/ (1g x Scale factor(V/g))]

Important Specifications Important Specifications

Since gravity is static (DC), absolute accuracy is important Since gravity is static (DC), absolute accuracy is important
Drift of zero g bias affects how accurately you can measure tilt Drift of zero g bias affects how accurately you can measure tilt Drift of zero g bias affects how accurately you can measure tilt Low hysteresis & repeatability are important Low hysteresis & repeatability are important Low hysteresis & repeatability are important

High resolution is important as 1 degree of tilt is 17mg High resolution is important as 1 degree of tilt is 17mg
Noise level is the chief determinant of resolution Noise level is the chief determinant of resolution Noise level is the chief determinant of resolution

Data acquisition: Inertial Data acquisition: Inertial

How does it work How does it work
Integrate acceleration once for velocity, twice for distance Integrate acceleration once for velocity, twice for distance Relative measurement from an initial position Relative measurement from an initial position Distance = 1/2 AT^2 Distance = 1/2 AT^2

Important Specifications Important Specifications

Bias stability: any offset is integrated as an acceleration Bias stability: any offset is integrated as an acceleration Noise and resolution; Small signals must be measured Noise and resolution; Small signals must be measured Similar to tilt, but need 10X or better accuracy Similar to tilt, but need 10X or better accuracy

Gravity and orientation effects on acceleration Gravity and orientation effects on acceleration
A tilt will look like an acceleration and thus a change in position A tilt will look like an acceleration and thus a change in position A rotation will change the accelerometer axis of sensitivity A rotation will change the accelerometer axis of sensitivity Gyroscopes are needed to determine tilt Vs acceleration Gyroscopes are needed to determine tilt Vs acceleration

Practical inertial Practical inertial applications applications

Most problems will require 6 degrees of Freedom Most problems will require 6 degrees of Freedom
3 Axes of lateral acceleration 3 Axes of lateral acceleration 3 Axes of rotation 3 Axes of rotation

Attempt to constrain the degrees of freedom Attempt to constrain the degrees of freedom
Constrain acceleration or rotation axes Constrain acceleration or rotation axes Allows fewer sensors to be used Allows fewer sensors to be used Simplifies mathematics Simplifies mathematics

Try to take advantage of the constraints of the Try to take advantage of the constraints of the mechanical system mechanical system
I.e. cars can only pitch or roll a few degrees I.e. cars can only pitch or roll a few degrees

Practical inertial Practical inertial applications applications

Attempt to limit time between absolute updates Attempt to limit time between absolute updates
Attempt to have periodic absolute position updates Attempt to have periodic absolute position updates Reduce dead reckoning time to a minimum Reduce dead reckoning time to a minimum Combine absolute and relative position sensors Combine absolute and relative position sensors
Example: Magnetics, encoders, LEDs etc. Example: Magnetics, encoders, LEDs etc.

Use signature analysis Use signature analysis

Look at the spectral response of one or more axis to make a Look at the spectral response of one or more axis to make a very educated guess as to what the mechanical system is doing very educated guess as to what the mechanical system is doing

Designing for tilt sensing Designing for tilt sensing

How can II digitize my signal and achieve the BW, How can digitize my signal and achieve the BW, resolution and accuracy required? resolution and accuracy required? What are my requirements? What are my requirements?
Resolution defined by bits, mg or degrees Resolution defined by bits, mg or degrees BW defined in Hz BW defined in Hz Accuracy in bits, mg or degrees Accuracy in bits, mg or degrees

What are the limitations What are the limitations

Accelerometer resolution and noise Accelerometer resolution and noise Accelerometer short term and long term stability Accelerometer short term and long term stability Accelerometer Sensitivity Accelerometer Sensitivity A/D converter bits A/D converter bits

Resolution limited by Resolution limited by accelerometer noise accelerometer noise

Short Term <1s Short Term <1s
Dominated by 1/f and broadband noise Dominated by 1/f and broadband noise Noise trades off with BW Noise trades off with BW

Long Term > minutes to hours Long Term > minutes to hours
Dominated by temperature changes due to bias drift with Dominated by temperature changes due to bias drift with temperature (can be corrected at system level) temperature (can be corrected at system level) Hysteresis (non-repeatability) generated by temperature, shock Hysteresis (non-repeatability) generated by temperature, shock or vibration or vibration Often affected by packaging technology, or device technology Often affected by packaging technology, or device technology

Noise in micro-machined Noise in micro-machined accelerometers accelerometers

White noise characteristic defined by mg/ Hz White noise characteristic defined by mg/ Hz Noise rolls off with accelerometer BW Noise rolls off with accelerometer BW Trade-off between Noise and BW Trade-off between Noise and BW
10mg Noise RMS = Noise * SQR(BW*1.5) (for a single pole filter) Noise RMS = Noise * SQR(BW*1.5) (for a single pole filter)

1mg 48mg RMS Noise 48mg RMS Noise with 1kHz device with 1kHz device

1mg/ Root Hz 1mg/ Root Hz

4.8mg RMS Noise 4.8mg RMS Noise with 10Hz single pole filter with 10Hz single pole filter

0.1mg 1Hz 10Hz 100Hz 1kHz 10kHz

Resolution is generally Resolution is generally limited by noise limited by noise

Given a 500ug/Rt Hz Accelerometer, what is the resolution Vs. Given a 500ug/Rt Hz Accelerometer, what is the resolution Vs. bandwidth trade-off? bandwidth trade-off?
BW(Hz) BW(Hz) RMS Noise RMS Noise P-P Estimate (4x) P-P Estimate (4x) Degree of Uncertainty Degree of Uncertainty

1Hz 1Hz 10Hz 10Hz 100Hz 100Hz

0.5mg 0.5mg 1.6mg 1.6mg 5.0mg 5.0mg

2.0mg 2.0mg 6.3mg 6.3mg 20mg 20mg

0.1 degree 0.1 degree 0.4 degree 0.4 degree 1.2 degree 1.2 degree

Use a peak to peak estimation; measurement is absolute Use a peak to peak estimation; measurement is absolute Be sure to account for an additional filtering due to averaging. Be sure to account for an additional filtering due to averaging. Example 100hz averaged to 1hz will give 1 Hz noise Example 100hz averaged to 1hz will give 1 Hz noise

RMS noise Vs. peak to RMS noise Vs. peak to peak noise peak noise
Noise is Gaussian in distribution and described statistically Peak to Peak Noise can only be estimated from statistics
Nominal Peak-to Peak Value 2.0 x rms 4.0 x rms 6.0 x rms 8.0 x rms % of Time that Noise Will Exceed Nominal Peak-to Peak Value 32% 4.6% 0.27% 0.006%

A/D converter resolution A/D converter resolution

Given a 5V reference and a 500mV/g accelerometer Given a 5V reference and a 500mV/g accelerometer A/D Bits Resolution gs per bit Degrees/bit A/D Bits Resolution gs per bit Degrees/bit 8 Bits 19.5mV/bit 39mg 2.3 8 Bits 19.5mV/bit 39mg 2.3 10 Bits 4.88mV/bit 10mg 0.6 10 Bits 4.88mV/bit 10mg 0.6 12 Bits 1.22mV/bit 2.4mg 0.14 12 Bits 1.22mV/bit 2.4mg 0.14 Accelerometer offsets can limit resolution Accelerometer offsets can limit resolution
Accelerometer signal can be amplified, but initial zero g offset is also Accelerometer signal can be amplified, but initial zero g offset is also amplified. amplified. Low cost accelerometers have initial offset equal to full scale. Low cost accelerometers have initial offset equal to full scale. HW offset trimming is required even for moderate amplification HW offset trimming is required even for moderate amplification

Digital trickery Digital trickery

A to D converter resolution may be increased by using A to D converter resolution may be increased by using digital signal processing techniques digital signal processing techniques
Oversampling is a simple way to improve resolution at the expense of Oversampling is a simple way to improve resolution at the expense of bandwidth bandwidth

Digital filters are very effective in very low speed, very Digital filters are very effective in very low speed, very high resolution applications high resolution applications Zero g bias offsets and scale errors are easy to deal with Zero g bias offsets and scale errors are easy to deal with digitally digitally
Calibrations can be automated Calibrations can be automated

Zero g bias shift over temperature can be eliminated at Zero g bias shift over temperature can be eliminated at system level without temperature sensing in some cases system level without temperature sensing in some cases

Some possible designs Some possible designs

Trade-off noise and BW to achieve resolution Trade A/D converter bits Vs cost, and trimming
Accel Accel Amplifier Amplifier with Trim with Trim Circuit Circuit Accel Accel 12 bit A/D 12 bit A/D

Lower cost Lower cost Micro Micro

Trim Trim

Micro with Micro with 8 bit A/d 8 bit A/d

Amplifier allows use of lower resolution A/D, but requires trim resistor circuit to trim out offset of the accelerometer

More expensive A/D but trims are done in software. Possibility to use cheaper microcontroller

Some possible designs Some possible designs

Using oversampling, resolution of an 8 bit Using oversampling, resolution of an 8 bit converter can usually be increased to 10 bits or converter can usually be increased to 10 bits or more more
16 averaged samples required for 2 bits of resolution 16 averaged samples required for 2 bits of resolution improvement improvement Bandwidth is reduced proportionately (by a factor of 16 in this Bandwidth is reduced proportionately (by a factor of 16 in this example) example)

Micro with Micro with 8 bit A/d 8 bit A/d

More digital trickery More digital trickery

A TrimDAC may be used (controlled by the microcontroller) A TrimDAC may be used (controlled by the microcontroller) to trim for zero g offset. to trim for zero g offset.

TrimDAC TrimDAC

Micro with Micro with 8 bit A/d 8 bit A/d

Pitfalls: Acceleration Pitfalls: Acceleration interferes with tilt signal interferes with tilt signal
Both gravity and motion produce accelerations Both gravity and motion produce accelerations Accelerometer (or any tilt sensor) cant tell the Accelerometer (or any tilt sensor) cant tell the difference between motion and tilt difference between motion and tilt
However 2 accelerometers cleverly arranged (or a gyroscope) However 2 accelerometers cleverly arranged (or a gyroscope) can. (Note: lots of digital trickery may be required) can. (Note: lots of digital trickery may be required)
0.3g Acceleration 0.3g Acceleration Due to Motion Due to Motion

Apparent Tilt Apparent Tilt 1g 1g Acceleration Acceleration Due to Gravity Due to Gravity

Solutions to the problem Solutions to the problem of bias stability of bias stability
Correcting for temperature drift and long term Correcting for temperature drift and long term stability stability
Buy an expensive sensor ($100 +) Buy an expensive sensor ($100 +) Use a microcontroller with a temperature sensor Use a microcontroller with a temperature sensor Use an Analog correction circuit Use an Analog correction circuit Use a temperature controlled crystal oven Use a temperature controlled crystal oven Allow a user recalibration Allow a user recalibration Be very clever in your application Be very clever in your application

Correcting for temperature Correcting for temperature drift and long term stability drift and long term stability
Buy an expensive sensor Buy an expensive sensor
End product too expensive leading to bankruptcy and End product too expensive leading to bankruptcy and personal humiliation personal humiliation

Use a microcontroller and Use a microcontroller and temperature sensor temperature sensor
Zero g bias drift can be measured over temperature Zero g bias drift can be measured over temperature and compensated for by the microcontroller and compensated for by the microcontroller Temperature induced effects are relatively linear so Temperature induced effects are relatively linear so they are easy to handle using either a look-up table or they are easy to handle using either a look-up table or a mathematical solution a mathematical solution

Correcting for temperature Correcting for temperature drift and long term stability drift and long term stability
Use an analog correction circuit Use an analog correction circuit
I.e. add a temperature controlled gain stage with an op-amp and I.e. add a temperature controlled gain stage with an op-amp and a thermistor a thermistor Hard to match response for sensor to sensor variance Hard to match response for sensor to sensor variance

Use a temperature controlled crystal oven Use a temperature controlled crystal oven
Holds the accelerometer at a constant temp, so drift due to Holds the accelerometer at a constant temp, so drift due to temp doesnt exist temp doesnt exist Crystal oven may consume a lot of power (500 mA) Crystal oven may consume a lot of power (500 mA)

Correcting for temperature Correcting for temperature drift and long term stability drift and long term stability
Allow a user recalibration Allow a user recalibration
Let the user hit a reset button from time to time when the Let the user hit a reset button from time to time when the accelerometer is experiencing zero g accelerometer is experiencing zero g

Be very clever in your application Be very clever in your application

Does your application really need to be DC coupled (i.e. do Does your application really need to be DC coupled (i.e. do you need to measure tilt)? If not you should AC couple the you need to measure tilt)? If not you should AC couple the accelerometer, via a capacitor, to your signal conditioning accelerometer, via a capacitor, to your signal conditioning stage to eliminate bias drift stage to eliminate bias drift Use software to detect very low speed, long term, zero Use software to detect very low speed, long term, zero changes changes

Using accelerometers as Using accelerometers as tilt sensors tilt sensors

Advantages Advantages
Accelerometers have faster response and settling than traditional Accelerometers have faster response and settling than traditional liquid tilt sensors, up to 100hz response liquid tilt sensors, up to 100hz response No slosh or leaking problems associated with liquid tilt sensors No slosh or leaking problems associated with liquid tilt sensors Sensor + Signal conditioning on a single chip in an IC form factor Sensor + Signal conditioning on a single chip in an IC form factor

Disadvantages Disadvantages
Acceleration signals may interfere with tilt signal, (problem for all tilt Acceleration signals may interfere with tilt signal, (problem for all tilt sensors) sensors) Currently available low cost devices difficult to use for sub 1 degree Currently available low cost devices difficult to use for sub 1 degree resolution (except for very low speed applications) resolution (except for very low speed applications)

Data acquisition: Data acquisition: Vibration and shock Vibration and shock
How does it work How does it work
Acceleration used to measure vibration in x, y or z axis Acceleration used to measure vibration in x, y or z axis Often analyzed in frequency domain as acceleration, velocity or Often analyzed in frequency domain as acceleration, velocity or distance (FFT or spectrum analysis) distance (FFT or spectrum analysis)

Important specifications Important specifications

Bandwidth (minimum and maximum) Bandwidth (minimum and maximum) Dynamic Range (noise limited) Dynamic Range (noise limited) Amplitude Stability (Sensitivity changes with temperature, time) Amplitude Stability (Sensitivity changes with temperature, time)

Characteristics of vibration Characteristics of vibration

For rotating equipment For rotating equipment
Acceleration and velocity of a vibration are the result of Acceleration and velocity of a vibration are the result of displacements with time. displacements with time. Peak velocity is proportional to displacement and frequency Peak velocity is proportional to displacement and frequency Vpeak = 2f*(peak displacement) Vpeak = 2f*(peak displacement) Peak acceleration is proportional to displacement and Peak acceleration is proportional to displacement and frequency squared frequency squared Apeak = 4*(^2)*(f^2)*(peak displacement) Apeak = 4*(^2)*(f^2)*(peak displacement) Energy is proportional to frequency and amplitude Energy is proportional to frequency and amplitude

Characteristics of vibration Characteristics of vibration

Examples Examples Fan Motor Fan Motor
30 Hz (1800 RPM) with a displacement of 1mm peak 30 Hz (1800 RPM) with a displacement of 1mm peak Velocity = 0.2m/s Velocity = 0.2m/s Acceleration = 35.5 m/s^2 = 3.62g peak Acceleration = 35.5 m/s^2 = 3.62g peak

Turbine Turbine
10,000 Hz with a displacement of 0.0001mm peak 10,000 Hz with a displacement of 0.0001mm peak Velocity = 0.006m/s Velocity = 0.006m/s Acceleration = 394m/s^2 = 40g Acceleration = 394m/s^2 = 40g

Characteristics of vibration Characteristics of vibration

Notice how high frequencies combined with small Notice how high frequencies combined with small displacements cause large acceleration displacements cause large acceleration Even low energy inputs applied to an absolutely Even low energy inputs applied to an absolutely fixed accelerometer can cause very large fixed accelerometer can cause very large acceleration (this is common to all acceleration (this is common to all accelerometers) accelerometers) Many applications can tolerate a small amount of Many applications can tolerate a small amount of cushioning of the accelerometer to protect it cushioning of the accelerometer to protect it from damage due to exposure to very large from damage due to exposure to very large acceleration (cushioning acts as a low pass filter) acceleration (cushioning acts as a low pass filter)

Special techniques to Special techniques to recover signal from noise recover signal from noise
Example: Example: Detect a 10mg signal at 1khz using an accelerometer Detect a 10mg signal at 1khz using an accelerometer with a 1mg/Root Hz noise floor. with a 1mg/Root Hz noise floor. Noise (rms) = 1mg/root Hz * sqr(1000) = 32mg rms Noise (rms) = 1mg/root Hz * sqr(1000) = 32mg rms = 128mg p-p = 128mg p-p recall that 4.6% of the time the peak to peak noise recall that 4.6% of the time the peak to peak noise will exceed 128mg will exceed 128mg

What if we add a What if we add a bandpass filter? bandpass filter?

Use a 10 Hz bandpass filter Use a 10 Hz bandpass filter Noise (rms) = 1mg/root hz * sqr(10) = 3.2mg rms Noise (rms) = 1mg/root hz * sqr(10) = 3.2mg rms = 12.8mg p-p = 12.8mg p-p But how do we look at multiple frequencies? But how do we look at multiple frequencies?

Use an FFT to recover Use an FFT to recover signals from noise signals from noise
An FFT is essentially a series of narrow band pass An FFT is essentially a series of narrow band pass filters. filters. Low speed FFTs can be performed by many 8-bit micros Low speed FFTs can be performed by many 8-bit micros
10mg Noise RMS = Noise * SQR(BW*1.5) (for a single pole filter) Noise RMS = Noise * SQR(BW*1.5) (for a single pole filter)

1mg

1mg/ Root Hz 1mg/ Root Hz

0.1mg 1Hz 10Hz 100Hz 1kHz 10kHz

Pros and Cons of Pros and Cons of Micromachined Sensor Micromachined Sensor
Advantages Advantages
Low cost (can even be made disposable) Low cost (can even be made disposable) FFTs can be used to increase the performance FFTs can be used to increase the performance Will work for many machine health applications Will work for many machine health applications On board signal conditioning. No charge amplifiers required On board signal conditioning. No charge amplifiers required

Disadvantages Disadvantages
Performance still below that of more expensive sensors Performance still below that of more expensive sensors May not be available in industrial hardened packages May not be available in industrial hardened packages

Calibration methods Calibration methods

Shaker with reference device Shaker with reference device
An AC test that can be done at various frequencies An AC test that can be done at various frequencies Required a calibrated reference device Required a calibrated reference device Can test to higher g levels than 1g Can test to higher g levels than 1g Fixturing can be an issue for higher g levels Fixturing can be an issue for higher g levels

Gravity Test Gravity Test

Test sensor at +1g and -1g to calibrate offset and sensitivity Test sensor at +1g and -1g to calibrate offset and sensitivity using Earths gravity using Earths gravity No need to calibrate the reference! No need to calibrate the reference! Fairly insensitive to alignment errors Fairly insensitive to alignment errors Only works for DC sensors Only works for DC sensors Will not test the full scale of a high g device. Will not test the full scale of a high g device.

Calibration methods Calibration methods

Gravity test method Gravity test method
Rotate the accelerometer at least 180 normal to the axis of Rotate the accelerometer at least 180 normal to the axis of sensitivity sensitivity Look for the minimum and maximum values Look for the minimum and maximum values Zero g output = (maxout+minout)/2 Zero g output = (maxout+minout)/2 Scale = (maxout-minout)/2 Volts per g Scale = (maxout-minout)/2 Volts per g

Calibration can be performed by the system Calibration can be performed by the system microcontroller microcontroller Easy fixture construction Easy fixture construction Dual axis accelerometers can calibrate both axis Dual axis accelerometers can calibrate both axis by extending the rotation to at least 360 by extending the rotation to at least 360

Summary Summary
New accelerometers open the door for new New accelerometers open the door for new applications in tilt, inertial and vibration applications in tilt, inertial and vibration
Low cost Low cost High level of integration: Multiple sensors, + signal High level of integration: Multiple sensors, + signal conditioning conditioning

Clever design can allow use of a less accurate, but Clever design can allow use of a less accurate, but less expensive sensor less expensive sensor
Using microcontrollers for calibration and algorithms Using microcontrollers for calibration and algorithms Using signal analysis to improve noise levels Using signal analysis to improve noise levels Taking new approaches to traditional problems Taking new approaches to traditional problems