STABLE Stabilization Table for Accurately Balancing a Loose Element Preliminary Design Review October 18th, 2012 William Brown Phillip Chen Eric Huckenpahler Laura Hughes Brian Ibeling Chris Johnson 11/7/2015

Download Report

Transcript STABLE Stabilization Table for Accurately Balancing a Loose Element Preliminary Design Review October 18th, 2012 William Brown Phillip Chen Eric Huckenpahler Laura Hughes Brian Ibeling Chris Johnson 11/7/2015

STABLE
Stabilization Table for Accurately Balancing a Loose Element
Preliminary Design Review
October 18th, 2012
William Brown
Phillip Chen
Eric Huckenpahler
Laura Hughes
Brian Ibeling
Chris Johnson
11/7/2015
1
Presentation Overview
• System Overview
• Subsystems
o
o
o
o
o
•
•
•
•
Sensors
Control
Motor/Mechanical
Power
User Interface (UI)
Team Roles
Risks and Mitigations
Schedule
Budget
11/7/2015
2
Demonstration
11/7/2015
3
System Objectives
• Maintain Desired Ball Position
o Knowledge of desired ball position
o Knowledge of current ball position
o Ability to move ball to desired position
• Ball Control
o Counter-act forces exerted on ball
o Follow desired user input path
o User capable to set desired ball position
• Unique and Engaging User Experience
o Games
o Artistic Application
11/7/2015
4
Functional Block Diagram
11/7/2015
5
System Responsibilities
11/7/2015
6
Sensor Subsystem
Position
Control System
Ball
Velocity
11/7/2015
7
Requirements & Purpose
• Purpose:
o Provide the Control Subsystem with time critical data
• Needed Information
o Gather Ball Position Data
o Calculate Ball Velocity Data
• Necessary Traits
o
o
o
o
o
High Speed
High Accuracy
High Precision
Large Size
Cost, under $300
11/7/2015
8
Sensor Method: Resistive Touch Screen
• Pros:
o Simplicity
o High resolution in range of 1024x1024 to
4096x4096 (for a 24” by 24” screen that is 170 bits per inch resolution)
o Linearity < 1.5% (acceptable)
o Price <$100
o Wide range of ball options, just need to be heavy
o Nice output format: easier to interface
• Cons:
o Resistive sensors can be noisy
o Cracks are fatal
11/7/2015
9
Sensor Method: Capacitive Touch Screen
• Pros:
o High Accuracy, 99% of true precision
o High Resolution 1024x1024 (for a 24” by 24” screen that is 42bits per inch
resolution)
o Linearity < 1.5% (acceptable)
o Less noisy
o Highly Durable
• Cons:
o Restrictive ball type
o More complex to interface directly
o Price < $300
11/7/2015
10
First Implementation Choice
• Resistive Touch Screen
• Necessary Hardware to implement:
o Dependent on touch screen purchased, some will include a controller
that can output RS232 data requiring a MAX232 IC to convert voltage
levels to UART acceptable levels.
o Others will require a touch screen controller, AR1000 series IC that can
directly output UART, I2C or SPI.
11/7/2015
11
Generating Velocity Data
• You can’t just sense velocity, it depends on time
• Use a timed interrupt to gather position data and
save a running average for its velocity.
• Frequency at which velocity data will be available
TBD (dependent on Control Subsystem’s needs)
Timer
Acquired
Position Data
Timer Length
11/7/2015
12
Control Subsystem
General outline of surrounding elements
Tilt Angle
Table Control
Subsystem
Control Subsystem
User Interface
Subsystem
Sensor Subsystem
Position & Velocity
11/7/2015
13
Control Subsystem Properties
• Settling time (movement within 0.5mm of desired
position) of 5 seconds.
• Maximum overshoot of around 2 cm (4%).
11/7/2015
14
Free Body Diagram
Drawing and force vectors not to scale
Fn
Fr
Fg

Free body diagram of the ball rolling in one dimension
11/7/2015
15
Mass-spring-damper 1D Model
External force
Wall
Model rolling
friction as a
damper
Ball of mass
M
x
(Force of gravity
along the plate)
U = M*g*sin(θ)
B
M
U
𝑀 𝑥 = −𝐵 𝑥 − 𝑈
11/7/2015
16
Solving for Proportional Control
Root Locus
• 𝑀𝑠 2 𝑋 = −𝐵𝑠𝑋 + 𝑈
𝑆𝑜𝑙𝑣𝑖𝑛𝑔 𝑌 = 𝑋 and substituting
𝑀𝑔 0.9 ∗ 𝜃 for U (to linearize).
• 𝑌=
1
𝑀𝑠 2 +𝐵𝑠
∗ 0.9𝑀𝑔
• Transfer function: P(s) =
0.9𝑀𝑔
𝑀𝑠 2 +𝐵𝑠
• Mass: assume the ball’s mass will be around 0.3 kg.
• Damper: the damper represents the force of rolling
friction, which is small (𝐹 = 𝐶𝑟𝑟 ∗ 𝑁, where N is the
normal force and Crr  0.003). B  0.008.
11/7/2015
17
Root Locus Graph (Proportional
Control)
11/7/2015
18
Root Locus Analysis
(Proportional control)
• Proportional control doesn’t yield a desirable
transient behavior in the theoretical plant.
o This can be seen by the poles being close to the imaginary axis in the Root
Locus graph. Increasing the proportional control parameter drives the
system to have more oscillatory behavior.
• Must start a lead-lag design for controller’s transfer
function.
o The lag portion is unnecessary, as the function already has step tracking
due to the naked integrator.
11/7/2015
19
Root Locus Setup (Lead
Component)
• Desires:
o to place a closed loop zero very far to the left to allow for poles to move
further left (thereby becoming more stable).
o to then place a closed loop pole even further to the right to minimize its
effect on the system.
• 𝐶 𝑠 =
𝐾(𝑠+100)
,
𝑠+50000
• 𝑃 𝑠 𝐶 𝑠 =𝐾∗
where K is the root locus parameter.
0.9𝑀𝑔(𝑠+10)
𝑀𝑠 3 + 𝐵+5000𝑀 𝑠 2 +(5000𝐵)𝑠
11/7/2015
20
Control Subsystem
Root Locus Graph (Lead Component)
11/7/2015
21
Step Response Graph (Lead
Component, K = 1.3E7)
11/7/2015
22
Root Locus Analysis of
Lead-Lag Controller
𝐾(𝑠+100)
• Choosing 𝐶 𝑠 =
and 𝐾 = 1.3 ∗ 107 , we arrive
𝑠+50000
at a step response with an overshoot of 4% and a
settling time of essentially 0.02 seconds
• Problems
o This model assumes we can actuate the plate instantaneously, when in
reality the motor setup will respond much more slowly.
o We have approximated the sine function, which may lead to more
overshoot.
11/7/2015
23
Motor Subsystem
11/7/2015
24
Requirements & Purpose
• Purpose:
o Actuate the plate according to instructions from the Control system
• Needed Information
o Desired position of each axis independently, relative to current position
11/7/2015
25
Restrictions
•
•
•
•
High Speed
High Accuracy
Preferably smaller and non-conducting
Cost, including replacements
Footer Text
11/7/2015
26
DC Brushed
Basics:
When the motor windings become energized, a temporary
magnetic field is created that repels(and/or attracts) against
permanent magnets. This force is converted into shaft rotation, which allows
the motor to do work. As the shaft rotates, electric current is routed to
different sets of windings, maintaining electromotive
repulsion/attraction, forcing the rotor to continually turn.
Pros:
• Cost. Cheaper than stepper or servo motors
• Super easy to work. Connect one wire high, one low, watch it go!
Cons:
• Torque vs Size. Many of these will be for smaller hobby applications and
it’s hard to find a low RPM, high torque DC motor in a smallish package.
• Accuracy. It’s going to be pretty hard to calibrate this, and that could
change with differences in load like a different ball.
• Can be electrically noisy and could interfere with the uC or any wireless
network we set up.
• A physical commentator is going to eventually wear out. New motor =
new calibration process.
11/7/2015
27
DC Brushed
Basics:
When the motor windings become energized, a temporary
magnetic field is created that repels(and/or attracts) against
permanent magnets. This force is converted into shaft rotation, which allows
the motor to do work. As the shaft rotates, electric current is routed to
different sets of windings, maintaining electromotive
repulsion/attraction, forcing the rotor to continually turn.
Pros:
• Cost. Cheaper than stepper or servo motors
• Super easy to work. Connect one wire high, one low, watch it go!
Cons:
• Torque vs Size. Many of these will be for smaller hobby applications and
it’s hard to find a low RPM, high torque DC motor in a smallish package.
• Accuracy. It’s going to be pretty hard to calibrate this, and that could
change with differences in load like a different ball.
• Can be electrically noisy and could interfere with the uC or any wireless
network we set up.
• A physical commentator is going to eventually wear out. New motor =
new calibration process.
11/7/2015
28
DC Brushed
Basics:
When the motor windings become energized, a temporary
magnetic field is created that repels(and/or attracts) against
permanent magnets. This force is converted into shaft rotation, which allows
the motor to do work. As the shaft rotates, electric current is routed to
different sets of windings, maintaining electromotive
repulsion/attraction, forcing the rotor to continually turn.
Pros:
•
•
Cost. Cheaper than stepper or servo motors
Super easy to work. Connect one wire high, one low, watch it go!
Cons:
• Torque vs Size. Many of these will be for smaller hobby applications and
it’s hard to find a low RPM, high torque DC motor in a smallish package.
• Accuracy. It’s going to be pretty hard to calibrate this, and that could
change with differences in load like a different ball.
• Can be electrically noisy and could interfere with the uC or any wireless
network we set up.
• A physical commentator is going to eventually wear out. New motor =
new calibration process.
11/7/2015
29
DC Brushed
Basics:
When the motor windings become energized, a temporary
magnetic field is created that repels(and/or attracts) against
permanent magnets. This force is converted into shaft rotation, which allows
the motor to do work. As the shaft rotates, electric current is routed to
different sets of windings, maintaining electromotive
repulsion/attraction, forcing the rotor to continually turn.
Pros:
•
•
Cost. Cheaper than stepper or servo motors
Super easy to work. Connect one wire high, one low, watch it go!
Cons:
•
•
•
•
Torque vs Size. Many of these will be for smaller hobby applications and
it’s hard to find a low RPM, high torque DC motor in a smallish package.
Accuracy. It’s going to be pretty hard to calibrate this, and that could
change with differences in load like a different ball.
Can be electrically noisy and could interfere with the uC or any wireless
network we set up.
A physical commentator is going to eventually wear out. New motor =
new calibration process.
11/7/2015
30
DC Brushless
Basics:
Same as brushed, but the commutator is realized with a
switch. The field inside a brushless motor is switched via an
amplifier triggered by a commutating device, such as an
optical encoder.
Pros:
• Cost. Cheaper than stepper or servo motors.
• Won’t wear out nearly as quickly as a brushed motor.
• Fewer physical parts means fewer factors that carry their own mess of
randomization.
• Cyprus has a great dev kit and software suite for $99 that we could play
with (proof of concept)
• I know how to keep these from breaking things by limiting the torque.
Cons:
• Torque vs Size. Many of these will be for smaller hobby applications and
it’s hard to find a low RPM, high torque DC motor in a smallish package.
• Accuracy. It’s going to be pretty hard to calibrate this, and that could
change with differences in load like a different ball. However, because it
is essentially PWM controlled, it’s going to be easier than brushed.
• Can be electrically noisy and could interfere with the uC or any wireless
network we set up.
11/7/2015
31
DC Brushless
Basics:
Same as brushed, but the commutator is realized with a
switch. The field inside a brushless motor is switched via an
amplifier triggered by a commutating device, such as an
optical encoder.
Pros:
• Cost. Cheaper than stepper or servo motors.
• Won’t wear out nearly as quickly as a brushed motor.
• Fewer physical parts means fewer factors that carry their own mess of
randomization.
• Cyprus has a great dev kit and software suite for $99 that we could play
with (proof of concept)
• I know how to keep these from breaking things by limiting the torque.
Cons:
• Torque vs Size. Many of these will be for smaller hobby applications and
it’s hard to find a low RPM, high torque DC motor in a smallish package.
• Accuracy. It’s going to be pretty hard to calibrate this, and that could
change with differences in load like a different ball. However, because it
is essentially PWM controlled, it’s going to be easier than brushed.
• Can be electrically noisy and could interfere with the uC or any wireless
network we set up.
11/7/2015
32
Stepper
Basics:
Takes input from a uC to move in
precise, accurate, pre-determined steps.
Pros:
• Easier to code for b/c based on simple integer values
• Higher chance of startup jerk than DC motors, but we actually
might want that to negate static friction
Cons:
• Accuracy. Even with 200 steps/rotation (typical) moving a
single step would tilt the edge of the plate 9*arcsin(1.8deg) or
about 1/3”. This could be good enough, but we may run into
a maximum update rate depending on our controller.
• Cost. Somewhat more expensive than DC motors.
11/7/2015
33
Stepper
Basics:
Takes input from a uC to move in
precise, accurate, pre-determined steps.
Pros:
• Easier to code for b/c based on simple integer values
• Higher chance of startup jerk than DC motors, but we actually
might want that to negate static friction
Cons:
• Accuracy. Even with 200 steps/rotation (typical) moving a
single step would tilt the edge of the plate 9*arcsin(1.8deg) or
about 1/3”. This could be good enough, but we may run into
a maximum update rate depending on our controller.
• Cost. Somewhat more expensive than DC motors.
11/7/2015
34
Stepper
Basics:
Takes input from a uC to move in
precise, accurate, pre-determined steps.
Pros:
• Easier to code for b/c based on simple integer values
• Higher chance of startup jerk than DC motors, but we actually
might want that to negate static friction of the ball
Cons:
• Accuracy. Even with 200 steps/rotation (typical) moving a
single step would tilt the edge of the plate 9*arcsin(1.8deg) or
about 1/3”. This could be good enough, but we may run into
a maximum update rate depending on our controller.
• Cost. Somewhat more expensive than DC motors.
11/7/2015
35
Voice Coils
Basics:
Voice coils use a magnetic field to push
a metal piston up and down.
Pros:
• The cool factor
• High precision and extremely fast
Cons:
• Range. They are designed more for small vibration control than ballcatching angles. The best we can get is about 1” of lift.
• We would be fighting gravity instead of working with it.
• Calibration would need to be ongoing and the plate would need to
consciously return to a level state.
• More difficult to discretely control.
11/7/2015
36
Voice Coils
Basics:
Voice coils use a magnetic field to push
a metal piston up and down.
Pros:
•
•
The cool factor
High precision and extremely fast
Cons:
• Range. They are designed more for small vibration control than ballcatching angles. The best we can get is about 1” of lift.
• We would be fighting gravity instead of working with it.
• Calibration would need to be ongoing and the plate would need to
consciously return to a level state.
• More difficult to discretely control.
11/7/2015
37
Voice Coils
Basics:
Voice coils use a magnetic field to push
a metal piston up and down.
Pros:
•
•
The cool factor
High precision and extremely fast
Cons:
• Range. They are designed more for small vibration control than ball•
•
•
catching angles. The best we can get is about 1” of lift.
We would be fighting gravity instead of working with it.
Calibration would need to be ongoing and the plate would need to
consciously return to a level state.
More difficult to discretely control.
11/7/2015
38
Voice Coils
Basics:
Voice coils use a magnetic field to push
a metal piston up and down.
Pros:
•
•
The cool factor
High precision and extremely fast
Cons:
• Range. They are designed more for small vibration control than ballcatching angles. The best we can get is about 1” of lift.
•
•
•
We would be fighting gravity instead of working with it.
Calibration would need to be ongoing and the plate would need to
consciously return to a level state.
More difficult to discretely control.
11/7/2015
39
Servo
Basics:
These are really a subset of other continuous
motors, unique only because they have a strict,
built in feedback system.
Pros:
• THE hobbyist motor, so there’s a ton of application specific information
out there.
• High precision and fast
• LEGO makes one, so we should be able to figure it out.
Cons:
• Accuracy. They could have the same issues as stepper motors or DC,
depending on what the servo base is.
• Maximum Update rate – as Michael points out, we are going to battle
the real-time aspect of this every step of the way. The motors are a bad
place to start.
• Cost. Built in feedback is apparently pretty expensive.
• Size and weight. These are typically much larger than their counterparts.
11/7/2015
40
Servo
Basics:
These are really a subset of other continuous
motors, unique only because they have a strict,
built in feedback system.
Pros:
• THE hobbyist motor, so there’s a ton of application specific information
out there.
• High precision and fast
•
LEGO makes one, so we should be able to figure it out.
Cons:
• Accuracy. They could have the same issues as stepper motors or DC,
depending on what the servo base is.
• Maximum Update rate – as Michael points out, we are going to battle
the real-time aspect of this every step of the way. The motors are a bad
place to start.
• Cost. Built in feedback is apparently pretty expensive.
• Size and weight. These are typically much larger than their counterparts.
11/7/2015
41
Servo
Basics:
These are really a subset of other continuous
motors, unique only because they have a strict,
built in feedback system.
Pros:
• THE hobbyist motor, so there’s a ton of application specific information
out there.
• High precision and fast
•
LEGO makes one, so we should be able to figure it out.
Cons:
• Accuracy. They could have the same issues as stepper motors or DC,
depending on what the servo base is.
• Maximum Update rate – as Michael points out, we are going to battle
the real-time aspect of this every step of the way. The motors are a bad
place to start.
• Cost. Built in feedback is apparently pretty expensive.
• Size and weight. These are typically much larger than their counterparts.
11/7/2015
42
Servo
Basics:
These are really a subset of other continuous
motors, unique only because they have a strict,
built in feedback system.
Pros:
• THE hobbyist motor, so there’s a ton of application specific information
out there.
• High precision and fast
•
LEGO makes one, so we should be able to figure it out.
Cons:
• Accuracy. They could have the same issues as stepper motors or DC,
depending on what the servo base is.
• Maximum Update rate – as Michael points out, we are going to
battle the real-time aspect of this every step of the way. The motors are
a bad place to start.
• Cost. Built in feedback is apparently pretty expensive.
• Size and weight. These are typically much larger than their counterparts.
11/7/2015
43
Servo
11/7/2015
44
Servo
11/7/2015
45
Servo
• 4.8 to 6.0V
• Torque around
6kg*cm
• Small Footprint
• $12
• 180 degrees of
movement
11/7/2015
46
Pivot
• Work WITH
Gravity
• Introduce
minimal
friction
• Can limit
motion of
plate with slant
of pivot tip
11/7/2015
47
Contact with Plate
• Cannot
interfere with
resistance or
capacitance
of plate
• Introduce
minimal
friction
• Allow full
range of
motion
• Be easy to
implement
and repair
11/7/2015
48
Servo
+
+
= proof of concept
11/7/2015
49
Power Subsystem
Footer Text
11/7/2015
50
Power Overview
• Voltage supplied from standard 120V 60Hz AC
Outlet
o AC-DC converter necessary for delicate instruments
o High efficiency
• Needs to prevent excessive heat buildup on user accessible surfaces
• Possible shielding to prevent noise from entering control subsystem
Footer Text
11/7/2015
51
Supply Rails
• AC-DC Converter will supply
o 5V for touch sensor and microprocessors
o 12V for motor power supply
• Motor power and motor control could possibly be combined into 1
unit. Noise suppression will be very important for servo drive voltage
o Aesthetic Supply depends on aesthetics. Unimportant for
prototype/preliminary design
• Possible aesthetic application:
o LED
o Speakers
o Displays
o LASERS
o Fog machine
Footer Text
11/7/2015
52
Possible Concerns
• Noise interference in motors
o Highly precise positioning is open to noise
• Heat generation
o Cautious for user interaction
Footer Text
11/7/2015
53
User Interface Subsystem
User
Desired
Position
Interaction
User Interface
Control System
11/7/2015
54
Requirements
• Interactive
• Intuitive
o Easy to use
o Total response time under a second
• Generates a Desired Position (and possibly velocity
and acceleration)
11/7/2015
55
Hardware Options
11/7/2015
56
Standard Mode
• User is free to disrupt the ball’s position in whatever
way they see fit
• User can also change balance point
11/7/2015
57
Game Mode
11/7/2015
58
Art Mode
11/7/2015
59
Risks
• Sensors
o Risk
• Filtering noise and isolating analog sensors from high noise devices
o Solution
• Short and isolated channel from sensor output to ADC
• Filter relevant data in software
o Risk
• Not being able to detect the ball with high enough precision when
the ball is moving at a high velocity
o Solution
• High sampling rate
• Average the samples
Footer Text
11/7/2015
60
Risks
• Control System
o Risk
• Plant perturbations
o Solution
• Detailed analysis on prototype and proper adjustment of model to
accurately define control model
• Motor Control
o Risk
• Not being able to actuate motor quickly and accurately enough to
counteract high forces exerted on ball
o Solution
• High torque motors
• Fast response time
• High update rate
11/7/2015
61
Risks
• Power
o Risk
• Providing high instantaneous current spikes due to accelerations
acted upon the ball
o Solution
• High gauge wires
• High power rectifier
• UI
o Risk
• Control system responding to appropriate interface device if using
multiple devices
o Solution
• Use of physical switches
11/7/2015
62
Team Roles
Task
Brian
Ibeling
Sensor Board
S
Motor Control
S
Mechanical
P
Laura
Hughes
William
Brown
Phillip
Chen
Eric
Huckenpahler
Chris
Johnson
S
S
P
P
S
P
S
Power
S
P
S
User Interface
S
S
P
S
P
Joystick Control
TouchScreen
Control
P
S
S
P = Primary
S = Secondary
S
11/7/2015
63
Schedule
11/7/2015
64
Budget
Budget
Unit Price
Quantity
Total Price
1.31
5.51
60
22
4
8
1
3
5.24
44.08
60
66
33
33
3
3
100
100
150
2
300
18.79
9.11
5.24
5.24
2
2
2
2
37.58
18.22
10.48
10.48
33
3
100
4
15
100
5
25
4
100
60
100
10
29
100
1
1
0.15
300
Sensors and Data Processing
ADC (Analog to Digital Converters)
Gyros
PIC24F Starter Kit
PiC24FJ25GB106 Microcontrollers
Sensor PCB Fabrication (2-layer)
Command and Data Handling PCB Fabrication
(2-layer)
Touch Screen for Ball Detection
Power Board
12V Converter
9V Converter
5V Converter
3.3V Converter
Power PCB Fabrication
Mechanical and Building Materials
Servo Motors
Motor Drivers
Aluminum
Steel Ball
2
User Interface
Joy Stick
Small Touch Screen
Custom User Interface Materials
LEDs
Printing
29
100
60
45
100
Poster Board Printing
100
Net Budget
1456.08
11/7/2015
65
Questions?
Footer Text
11/7/2015
66
Extra Slides
11/7/2015
67
Where I got pictures from:
(for Sensor Subsystem)
•
How it works:
o
•
Controller board:
o
•
http://officialandreascy.blogspot.com/2012/01/touch-screen-technology-how-it-works.html#.UH0AYcXA-rI
http://www.aliexpress.com/item/4-wire-Resistive-RS232-touch-screen-panel-controller-Serial-touch-panelcontroller/547670399.html
AR1000:
o
http://www2.electronicproducts.com/Resistive_touch_screen_controller_uses_USB-article-ICDJH08_Oct2011html.aspx
Footer Text
11/7/2015
68
Functional Decomposition: Level 0
Footer Text
11/7/2015
69
Functional Decomposition: Level 1
Footer Text
11/7/2015
70
Communicating with Control
Subsystem
• Ideal System Model:
o Save values as global variables for the control subsystem to interpret
• Distributed System Model:
o Send variables to Control Subsystem processor via I2C, SPI, or UART
Global
Variables
Control Logic
Sensor
Subsystem
Sensor
Subsystem
Footer Text
Control
Subsystem
Digital I/O
Control
Subsystem
Position & Velocity
11/7/2015
71