Professional Documents
Culture Documents
Lecture 11 Attribute Charts PDF
Lecture 11 Attribute Charts PDF
Yield Control
100 100
80 80
60 60
Yield
40 40
20 20
0 0
0 10 20 30 0 10 20 30
Months of Production
The P-Chart
The P chart is based on the binomial distribution:
n x n-x
P{D = X} = p (1-p) x = 0,1,...,n
x
mean np
variance np(1-p)
the sample fraction p= D
n
mean p
variance p(1-p)
n
m
mean p
Σ pi
p=
i=1
variance p(1-p)
m nm
(in
(in this
this and
and the
the following
following discussion,
discussion, "n" "n" isis the
the number
number ofof
samples
samples in in each
each group
group and
and "m"
"m" isis the
the number
number of of groups
groups
that
thatweweuseusein
inorder
orderto todetermine
determinethe
thecontrol
controllimits)
limits)
p can be estimated:
Di
pi = i = 1,...,m (m = 20 ~25)
p(1-p) n
UCL = p + 3 n m
LCL = p - 3
p(1-p)
n
Σ pi
i=1
p=
m
Lecture 11: Attribute Charts 6
EE290H F05 Spanos
0.3
0.232
0.2
0.1
LCL 0.053
0.0
0 10 20 30
Count
"Out
"Out of
of control
control points"
points" must
must be
be explained
explained and and eliminated
eliminated before
before we
we
recalculate
recalculatethe
thecontrol
controllimits.
limits.
This
Thismeans
meansthat
thatsetting
settingthe
thecontrol
controllimits
limitsisisan
aniterative
iterativeprocess!
process!
Special
Specialpatterns
patternsmust
mustalso
alsobe
beexplained.
explained.
Lecture 11: Attribute Charts 7
EE290H F05 Spanos
Example (cont.)
After the original problems have been corrected, the
limits must be evaluated again.
60 45
3.0
55 40
35
2.0 50
30
45
1.0 25
40
20
35
0.0 15
As
Assample
samplesize
sizeincreases,
increases,the
theNormal
Normalapproximation
approximationbecomes
becomesreasonable...
reasonable...
p = 0.20,
LCL=0.0303,
UCL=0.3697
The C-Chart
Sometimes we want to actually count the number of defects.
This gives us more information about the process.
The basic assumption is that defects "arrive" according to a
Poisson model:
e-c cx
p(x) = x = 0,1,2,..
x!
μ = c, σ 2 = c
This assumes that defects are independent and that they
arrive uniformly over time and space. Under these
assumptions: UCL = c + 3 c
center at c
LCL = c - 3 c
10 130
30 120
8
110
6
20 100
4 90
2 80
10
70
0
As
Asthe
themean
meanincreases,
increases,the
theNormal
Normalapproximation
approximationbecomes
becomesreasonable...
reasonable...
UCL 0.454
0.4
Fraction Nonconforming
0.3 ¯ 0.306
0.2
LCL 0.157
0.1
0 2 4 6 8 10 12 14
Ý 74.08
LCL 48.26
0
0 2 4 6 8 10 12 14
Wafer No
Counting particles
y
x
• Particle Count
• Particle Count by Size (histogram)
• Particle Density
• Particle Density variation by sub area (clustering)
• Cluster Count
• Cluster Classification
• Background Count
Whatever
Whatever we
we use
use (and
(and we
we might
might have
have to
to use
use more
more
than
thanone),
one),must
mustfollow
followaaknown,
known,usable
usabledistribution.
distribution.
to pump
Detector
The U-Chart
We could condense the information and avoid outliers by
using the “average” defect density u = Σc/n. It can be
shown that u obeys a Poisson "type" distribution with:
μu = u, σ2u = u
n
so
UCL =u + 3 u n
LCL =u - 3 u n
where u is the estimated value of the unknown u.
The sample size n may vary. This can easily be
accommodated.
poisson 2 average 5
5.0
10
4.0
8
3.0
6
4 2.0
2 1.0
0 0.0
Quantiles Quantiles
Moments Moments
By
Byexploiting
exploitingthe
thecentral
centrallimit
limittheorem,
theorem,ififsmall-sample
small-samplepoisson
poissonvariables
variables
can
canbe
bemade
madeto toapproach
approachnormal
normalbybygrouping
groupingandandaveraging
averaging
0.2
LCL 0.157
0.1
0 2 4 6 8 10 12 14
200
Defect Count (C-chart)
Number of Defects
Ý 74.08
LCL 48.26
0
0 2 4 6 8 10 12 14
6
Defect Density (U-chart)
5
Defects per Unit
4
UCL 3.76
3
× 2.79
2
LCL 1.82
1
0 2 4 6 8 10 12 14
Wafer No
Lecture 11: Attribute Charts 27
EE290H F05 Spanos
Input Output
Process
Measurement System
Detect
Verify and assignable
follow up cause
Diagrams by Phenomena
Diagrams by Causes
80
60
Percentage
occurence
cummulative
40
20
others
ontamination
sed contacts
ss problems
se particles
ern bridging
evious layer
s scratches
An Actual Example
80
60
percentage
occurence
cummulative
40
20
0
loading
equipmnet
others
inspection
utilities
smiff boxes
Defect Control
In general, statistical tools like control charts must be
combined with the rest of the "magnificent seven":
• Histograms
• Check Sheet
• Pareto Chart
• Cause and effect diagrams
• Defect Concentration Diagram
• Scatter Diagram
• Control Chart
Y = [ (1-e-AD)/AD ]2