Term Project
Final Presentation
? Visual aids available on-campus
– Computer projection
– Document camera
– Camera
? Visual aids available off-campus
– Camera
– OR -- Send me your slides electronically and
I’ll project them from my laptop
16.881 MIT
Term Project
Grading
? Term project is 30% of course grade
? Written report is 75% of term project
– Due on last Lecture day.
– 10% penalty per day late
? Final presentation is 25% of term project
16.881 MIT
Term Project
Final Presentation Schedule
? Tom Hoag, “Designing a Robust Business”
? Chip Clampitt, “The Use of Orthogonal Arrays to Optimize Nonlinear
Functions Iteratively”
? Karl Hauenstein, “Robust Design of a Voltage Controlled Oscillator”
? Boran, Goran, Pepin, Shashlo, Wickenheiser, “Robust System Design
Application / Integration - Ford Motor Company”
? Joe Distefano, “Application of Robust Design Techniques to a Paper
Winding Simulation”
? Garth Grover, “HPT Dovetail 2-D Form Robust Design”
? Shelley Hayes, “Taguchi Method Meets Publish and Subscribe”
16.881 MIT
Term Project
Final Presentation Schedule, Cont.
? Wei Zhao, “Taguchi and Beyond - Methodologies for Experimental
Designs”
? J. Philip Perschbacher, “Robust Design of Blade Attachment Device”
? Michelle Martuccio, “Allied Signal's Six Sigma Initiative: A Robust
Design Case Study”
? Steve Sides, Bob Slack, “Coating Technology for Jet Aircraft Engines”
? Ebad Jahangir, “Robust Design and its Relationship with Axiomatic
Design”.
? Tom Courtney “Robust Thermal Inkjet Printhead Design”
? David Markham, “Robustness Testing of a Film-Scanner Magnetic
Module”
16.881 MIT
Robust Conceptual Design
Considering Variation
Early in the Design Process
16.881 MIT
Outline
? Motivation
? Tools and tricks -- TRIZ, etc.
? A framework -- RCDM & wafer handling case
? Case study -- VMA prehensor
? Case study -- Adhesive application in LBPs
16.881 MIT
Quality in Product Development
Quality efforts used to be
focussed here
determined here!
Concept
Development
System
Design
Detail
Design
Testing and
Refinement
Production
Ramp-up
But 80%
of quality is
Customer
use
Taguchi Methods
of parameter design
Induce noise
16.881
Source: Ulrich and Eppinger, “Product Design and Development”
MIT
Concept Design:
The Window of Opportunity
Quality determined &
costs committed
Design flexibility
Window of opportunity
100
Percentage of total
Problem
75
50
25
Manufacture
Use
definiti
on
Concept
Detail
design
design
Source:
Russell B.
Ford and
Philip Barkan16.881
Lifecycle phase
Concept versus Parameter Design
Concept Design
? Begins with broad specs
? Free wheeling, intuitive
? One off experiments
? Rough analysis
? Requires insight
Parameter Design
? Begins with system design
? Bounded, systematic
? Orthogonal arrays
? Precise analysis
? Can be implemented as a
“black box”
Source: Russell B. Ford and Philip Barkan
16.881 MIT
Biggest Roadblocks in Concept Design
? Poor problem formulation
? Stopping with too few alternatives
? Failure to search existing solutions
? Missing entire categories of solutions
? Inability to merge solutions
16.881
Source: Ulrich and Eppinger, “Product Design and Development”
MIT
Properties of a
Good Problem Statement
? Solution neutral
? Quantitative
?Clear
? Concise
? Complete
16.881 MIT
Techniques for Concept Generation
? Brainstorming
? Analogy
? Seek related and unrelated stimuli
? Use appropriate media to convey & explore
– Sketching / Foam / Lego
? Circulate concepts & create galleries
? Systematically classify & search
16.881
Source: Ulrich and Eppinger, “Product Design and Development”
MIT
Theory of Inventive Problem
Solving (TRIZ)
? Genrich Altshuller
– Sought to identify patterns in the patent literature
(1946)
– "Creativity as an Exact Science" translated in 1988.
? The basic concept
– Define problems as contradictions
– Compare them to solutions of a similar form
– Provide a large database of physical phenomena
– Anticipate trends in technical evolution
16.881 MIT
TRIZ Software
? Ideation International
(http://www.ideationtriz.com/)
? Invention Machine (http://www.invention-
machine.com/)
– Effects
–Principles
–Prediction
16.881 MIT
Tricks for
Robust Concept Design
? Create lots of concepts with noise in mind
? Build breadboards & experiment (quickly)
? Don’t be afraid to revisit concept design stage
? Eliminate dependence on non-robust physical
effects & technologies
? Design in non-linearities to exploit in
parameter design
16.881 MIT
Robust Concept Design
Methodology
? Russell B. Ford and Philip Barkan at
Stanford
? Four Stages
– Definition of the robustness problem
– Derivation of guiding principles
– New concept synthesis
– Concept evaluation and selection
16.881 MIT
Wafer Handling Robot
Rotating platform
Gear pair
Silicon Wafer
Double Parallelogram Linkage
Top View
Process A
Process B
Process C
Load
Store
16.881
Side View
MIT
Stage 1
Definition of the Robustness Problem
? Identify robustness as a primary goal
? Incorporate critical performance metrics
into the problem definition
? Target needed improvements in robustness
? Quantify key robustness goals
T
R =
y
6σ
y
16.881 MIT
Stage 1
Rotating platform
Silicon Wafer
Gear pair
Double Parallelogram Linkage
How will you
Top View
specify robustness?
Side View
MIT16.881
Stage 2
Derivation of Guiding Principles
? Identify dominant error propagation
mechanisms
? Derive insight into the root causes of
performance variation
? Predict the effect of design parameters and
error sources on performance variation
? Single out limiting constraints
? Substantiate the predicted behavior
16.881 MIT
Stage 2
Rotating platform Gear pair
Silicon Wafer
Double Parallelogram Linkage
Top View
Side View
16.881
What are the root causes?
What are the mechanisms of
propagation?
How would you predict
effects?
What are the constraints on
the design?
MIT
Stage 3
New Concept Synthesis
? Modify error propagation mechanisms to
reduce or eliminate transmission
? Eliminate or reduce error sources
? Circumvent limiting constraints
? Draw upon new technology
? Add extra degrees of freedom as necessary
16.881 MIT
Stage 3
Rotating platform Gear pair
Silicon Wafer
Double Parallelogram Linkage
Top View
16.881
How can you modify
propagation?
Can you circumvent
constraints?
Are there new technologies
to employ?
Develop 3 other concepts.
Side View
MIT
Stage 4
Concept Evaluation and Selection
? Reconcile robustness requirements with al
other critical performance specifications
? Select the best concept from all alternatives
? Predict the effect of design parameters and
error sources on performance variation
? Decide whether further improvement is
required
16.881 MIT
References -- Conceptual
Robustness
? Ford, Russell B., and Philip Barkan “Beyond Parameter
Design -- A Methodology Addressing Product Robustness
at the Concept Formation Stage”, DE-Vol. 81, Design for
Manufacturability, ASME, 1995.
? Andersson, Peder, “A Semi-Analytic Approach to Robust
Design in the Conceptual Design Phase”, Research in
Engineering Design, Research in Engineering Design, vol.
8, pp. 229-239.
? Stoll, Henry W., “Strategies for Robust Product Design,”
Journal of Applied Manufacturing Systems, Winter, 1994,
pp. 3-8.
16.881 MIT
Case Study
VMA Prehensor
? Dan Frey and Larry Carlson
? The authors wish to thank the NCMRR
(grant no. 1-RO1-HD30101-01) for its
financial support
? The contributions of Bob Radocy as both
design consultant and field evaluator are
gratefully acknowledged
16.881 MIT
Body Powered Prosthetic
Prehension
? Amputee wears a harness to which a cable is
attached
? Cable routed through a housing, down the arm, to
a prehensor
? Body motions create cable excursion & apply
force
16.881 MIT
The TRS Grip
? A “voluntary closing” prehensor
– Lightly spring loaded to open position
– User applies cable force
? Often users want to change body position
Cable
tension
while grasping objects
? How will variations in cable excursion
affect grip force?
16.881 MIT
Testing Apparatus
? Lead screw applies force / displacement
? Load cell measures applied tension
? LVTD measures applied displacement
? Resulting grip force measured
MIT
Testing the Grip
Grip Force (lbs)
Grip Force (lbs)
Cable Tension (lbs)
Amputees can generate
2” of excursion
and 40 lbs tension
How would you design the Grip?
What form will the plots take?
What determines robustness
Cable Excursion (inches)
to body motions?
16.881 MIT
Pre-existing Approaches
APRL Hook
Northwestern U.
“Synergetic Prehensor”
16.881
Allows the user to lock the prehensor
- First stroke applies force and locks
- Second, harder stoke unlocks
- Safety compromised!
- Poor reliability
Myo-electrically operated hand
-Sizing and gripping are distinct
phases of grasp
- Both require minimal
mechanical energy
- Longer battery life
MIT
Variable Mechanical Advantage
? Idea -- break up the
task into sizing and
gripping
? How can one use this
to improve
Cable Tension (lbs)
robustness to body
position error?
Grip Force (lbs)
Grip Force (lbs)
16.881
Cable Excursion (inches)
MIT
VMA Design Concepts
Gear Based Design
Linkage Based Design
(Carlson)
(Frey / Carlson)
Simplified Linkage
Based Design
(Frey / Carlson)
16.881 MIT
Operation of the VMA Prehensor
MIT16.881
? Over-running clutch
used to hold force
? Performance very
sensitive to shape of
rollers
? Flat spots due to wear
rendered design
unreliable
MIT
Holding Assist Concept
0
5
10
15
20
25
GRIP FORCE (lbs.)
0 10 20 30 40
CABLE TENSION (lbs.)
Carlson’s Design
16.881
VMA Prehensor
First Prototype
? 2D profile allowed quick CNC prototyping
? $200 in machining costs
? Aluminum components
? Stock bearings
? ~$100 materials
VMA prototype with
face plate removed
MIT16.881
Robustness to Error in Excursion
? Excursion saved in sizing
? Employed later to lower sensitivity to excursion
by more than a factor of three
25
20
15
10
5
0
Grip Force at Figertip (lbs)
0 0.5 1 1.5 2
Input Cable Excursion (in)
16.881
VMA Prototype TRS GRIP II
MIT
MIT16.881
Robustness to Environment
? Users subject prehensors to varying conditions
? Such conditions adversely affected performance
VMA I PERFORMANCE
UNDER VARYING ENVIRONMENTAL CONDITIONS
0
5
10
15
20
25
0 5 10 15 20 25 30 35 40 45
INPUT CABLE TENSION (lbs)
G
R
IP
F
O
R
C
E
(l
b
s
)
CLEAN SOAPY WET DRY DEGREASED Series6
Ratchet Teeth
? Broached fine
teeth into mating
surfaces
? Friction no longer
determines
performance
16.881 MIT
VMA Prehensor
Second Generation Prototype
? More agressive increase in mechanical
advantage
? Holding assist enhanced through
mechanism design
16.881 MIT
VMA II PREHENSOR
COMPARED TO VMA I & GRIP II
25
20
15
10
5
0
0 10 20 30 40
INPUT CABLE TENSION (lbs)
GRIP FORCE AT FINGER TIP (lbs)
VMA II VMA I GRIP II
16.881 MIT
Results of Amputee Evaluation
VMA Prehensor
? Provides greater range of motion while
maintaining grasp
? Works reliably under wide range of
environmental conditions
? Shifts prematurely with compliant objects
? “Free-wheel” switch convenient to use
– Provides alternate mode of operation
16.881 MIT
References -- VMA Prehensor
? Frey, D. D. and L. E. Carlson, 1994, "A body powered prehensor with variable
mechanical advantage," Prosthetics and Orthotics International, vol. 18, pp.
118-123.
? Carlson, L. E. and R. Heim (1989). "Holding assist for a voluntary-closing
prosthetic prehensor," Issues in the Modeling and Control of Biomechanical
Systems, American Society of Mechanical Engineers, DSC-Vol. 17:79-87.
? Childress, D. S., and E. C. Grahn (1985). "Development of a powered
prehensor". In 38th Annual Conference on Engineering in Medicine and
Biology, p. 50.
? Taylor, C.L. (1954). "The biomechanics of the normal and of the amputated
upper extremity," Human Limbs and Their Substitutes, McGraw Hill, New
York, pp. 169-221.
16.881 MIT
Case Study
Adhesive Application for Surface
Mount of Large Body Packages
? Dan Frey and Stan Taketani
? The authors wish to thank the Hughes
Doctoral Fellowship program for its
financial support.
16.881 MIT
Adhesive Application
Design Issues
? Adhesives are required to:
– Support mechanical loads
– Transfer heat to sink
? Robustness problems
Typical adhesive pattern
– Epoxy thickens during application
– Air sometimes “burps”
– Air gap height not repeatible
–LBPs
16.881 MIT
P-Diagram
Noise Factors
Product / Process
Response
Signal Factor
Control Factors
16.881 MIT
Compression of a Single, Long Bead
F/L (lbs/in)
Navier-Stokes
highly viscous
x
y
h(t)
W(t)
dh/dt
MCM Body
Adhesive Bead
u(x,t,y)
h
(,,
? h
3
dp
∫
uxt h) =
12μ
?
dx
0
conserve mass
PWA Upper Surface
F
=
?μ
?
dh
?W
3
L h
3
dt
Pressure (psi)
x
integrate
?15
?
5FL 1
?
/
ht
/
() =
?
?
μ
o
3
o
3
t +
h
o
5
?
?
Wh
16.881 MIT
Multiple Beads with Air Pockets
F/L
x
1
x
2
x
n
p
1
p
2
p
n
h
W
gauge pressure onder MCM (atm)
Pressure Profile During Rampup
0.15
0.1
0.05
0
1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
x (inches)
16.881 MIT
Eliminating “Squeeze-Out”
Despite Viscosity Variation
? When beads touch one another, downward motion
is arrested
d
dt
touching
1
h
=
2
d h
n
P
P
d t
nottouching
x
not touching
touching
? Design rules exploit this phenomena
4
margin>bead pitch
∫
Ft =
?3πμ
?
?
W
?
? W ?
d
2h
f
2
?
?
2
?
?
?
?
1 ?
W + margin?
?
16.881 MIT
Estimating Percent Coverage
?h
sparse
coverage =
h
h
h +?h
sparse dense
sparse
coverage coverage
coverage
? Thinnest air gaps set component height
? Wider air gaps are areas of sparse coverage
16.881 MIT
Adhesive Flow Model
Preliminary Verification
? Used dispense test data to estimate μ
?Used μ, P, and V to calculate bead shape
? Used force schedule to estimate final height
and percent coverage
16.881 MIT
Postage Stamp and Tape
? Postage stamp protects the circuitry
? Tape allows easier rework
? BUT -- MCM to PWA gap cut from 13 mils to 7.5
mils
?Fα1/h
3
-- over 400%more force req’d
Force
MCM Body
Leads
Leads
PWA
Adhesive Bead
Pads
Tape
Pads
Postage Stamp
16.881 MIT
Accomodating Equipment Limitations
? Robot can only apply 7 lbs seating force
? Air pockets support substantial load (>50%)
– Open air gaps (when practicable)
?
F ∝
1
- Switch to thinner beads
W
3
16.881 MIT
Gaps in Adhesive Coverage
? Model predicted existence of gaps in
coverage under certain conditions
? Experimentally observed later
? Given gap location, they might not have
been detected early enough
16.881 MIT
Dispense Problems
Due to PWA Waviness
Nozzle too close to PWA --
Nozzle too far above PWA --
flow partially blocked.
Just right.
bead breaking and dragging.
16.881 MIT
Dispense Parameter Selection
Force
h
Force
Plugging
Dragging
Gapping
A < A
max
( F
max
, μ, airgap)
Dragging
2 A
h < ??h
D
nozzle
Plugging
h > ?h +
D
nozzle
2
Gapping
A
(1 ? cosα)
sin
2
1 ? α
A
2
> airgap +?h
16.881 MIT
Next Steps
? Next off-campus session
? Course evaluations
? Term project presentations
? Good luck!
16.881 MIT