Вы находитесь на странице: 1из 26

MKS Confidential 1

Process mprovement Team Project


Wave Soldering Thermal
Profiling Management mprovement
MKS Confidential 2
ProbIem Statement
PWI Introduction
Identify Performance Measures (Statistics)
Measurement Techniques for Performance
Measuring
Setting Metric for Performance Measures
W ;aIuating the current performance
W Data coIIection pIan de;eIopment
W CaIcuIating sampIe size
W CoIIected data sets
W AnaIyzed data
Define Process Limits
PiIoted to measure PWI
Define PWI ControI Limits
ControI pIan
Benefits
MKS Confidential 3
ProbIem Statement
":estions:
W Making ass:mption we get the machine profiles as shown, how can we
q:antificational q:alify the performance of the machine? Does the machine
work stable and meet specified criteria?
W ow can get profile #1 and profile #2 be compared, which one is better?
MKS Confidential 4
ProbIem Statement
There is no a q:antifiable system of ranking eq:ipment performance.
Meas:ring and comparing the thermal profile to its process window
is a s:bjective j:dgment with no real :niformity or statistically
reprod:cible res:lts; from one engineer or technician to the next;
from one prod:ct to the next.
Also, it is a time cons:ming process.
s there a q:antifying approach to calibrate the performance of
machine?
This project will introd:ce a q:antifying techniq:e PW to meas:re
the performance of wave soldering machine and reflow oven.
This presentation reported the implementation res:lts of wave
soldering only.
MKS Confidential 5
A Method for Quantifying ThermaI ProfiIe
Performance - PWI
What is PW
The Process Window Index
The PW is a q:antifiable, reprod:cible, statistical meas:re of how well a profile
performs relative to critical process limits. Every thermal profile is ranked on the
basis of how it "fits within the process window. The center of the process
window is defined as zero, and the extreme edge of the process window as
99%. A PW of 100% or more indicates that the profile will not process prod:ct
within specification. A PW of 99% indicates that the profile will process prod:ct
within spec, b:t it is r:nning at the very edge of the process window. A PW of
70% indicates a profile is :sing 70% of the process spec.
The PW tells :s exactly how m:ch of o:r process window a given profile :ses,
and th:s how rob:st that profile is. The lower the PW, the better the profile.
The thermal process can now be reliably meas:red, analyzed, compared and
tracked with the same level of SPC and ":ality Control available to other
man:fact:ring processes.
MKS Confidential 6
A Method for Quantifying ThermaI ProfiIe Performance
- PWI
CaIcuIating the PWI
The PW for a complete set of profile statistics is
calc:lated as the worst case (highest n:mber) in the
set of statistics. For example: if yo: r:n a profile with
three thermoco:ple, and fo:r profile statistics are
logged for each thermoco:ple, then there will be a
set of twelve statistics for that profile. The PW will be
the worst case (highest n:mber expressed as
a percentage) in that set of profile statistics.
(Meas:red_val:e[i,j] - average_limits[,j])
(range[,j]/2
PWI = 100 x MAX
,
j=1
N,
M
MKS Confidential 7
Identifying Performance Measures
Thermal profile specifications are ranges consisting of minim:m or
maxim:m val:es. These ranges apply to n:mero:s statistics s:ch
as soak time, slope, peak temperat:re and variety of others. Which
statistics were determined to be a CT" meas:res?
Project team selected below profile statistics as the performance
meas:re that was being :sed to calibrate the wave machine.
1. Top Side Preheat Temperat:re Rising Slope
2. Bottom Side Preheat Temperat:re Rising Slope
3. Top Side Preheat Peak Temperat:re
4. Bottom Side Preheat Peak Temperat:re
5. Top Side Wave Peak Temperat:re
6. Bottom Side Wave Peak Temperat:re
MKS Confidential 8
Measurement Techniques for Performance Measuring
R:n profiles :sing devices:
Profile Checker SlimKC 2000
Standard Profile Test board
Thermoco:ples: K Type
Bonding Techniq:es
Attaching the tip of the thermoco:ples to the
desired location with Al:min:m Tape or high
temperat:re solder.
Al:min:m Tape Size; Kept solder joint as
small as possible.
Thermoco:ples placements and KC
connections as shown.
Top thermo. measures -
Preheat SIope; Peak
Temp. & Top Wa;e
Peak Temp.
Bot. thermo.
measures - Bot.
Wa;e Peak Temp.
Bot. thermo.
measures - Bot.
Preheat Peak
Temp. & SIope
MKS Confidential 9
Setting Target for Performance Measures
Following activities have been done prior to set the limits for the
above performance meas:res:
Eval:ating the c:rrent performance of the machine.
Data Collection Plan Development
Calc:lating Sample Size
Collected data set
Analyzed data
Applied statistical tools to define the metric of the performance meas:re
MKS Confidential 10
Based on previo:s data knew the standard deviation of meas:res.
Refer to the appendix for the baseline data set
Variable StDev
W Top Max. Rising 0.0525
W Bot Max Rising 0.0540
W Top Preheat Temp. 0.725
W Bot Preheat Temp 0.793
W Top Wave Temp 2.37
W Bot Wave Temp 1.90
;aIuating the current performance of equipment
MKS Confidential 11
- Minimum SampIe Size CaIcuIation,
The minim:m sample size was determined to 26 per below calc:lation,
that means, 26 profiles req:ired in order to meet the req:ired precision.
The desired precision level was determined shown in below table.

8
n=
PIVs StdDe;. n
Top Max. Rising 0.0525 0.05 5
Bot Max Rising 0.054 0.05 5
Top Preheat Temp. 0.725 0.5 8
Bot Preheat Temp 0.793 0.5 10
Top Wave Temp. 2.07 0.8 26
Bot Wave Temp 1.9 0.8
MKS Confidential 12
Measurement
/Metric
X or
Y
Operationa
I Definition
Type of
Data
(Discrete/
Continuous
)
Data Source
and
Location
SampIe
Size
Who WiII
CoIIect the
Data?
When WiII
Data
be
CoIIected?
How WiII
Data
be
CoIIected?
Is the
Measureme
nt
System
CapabIe?
GraphicaI
and/or
StatisticaI
TooIs to be
Used
Top Max.
Rising X
Wave
Soldering
process Contin:o:s
Wave
Soldering
Station
David
Zhang J:ne, 2011
Randomly
selected es Excel form
Bot Max
Rising X
Wave
Soldering
process Contin:o:s
Wave
Soldering
Station
David
Zhang J:ne, 2011
Randomly
selected es Excel form
Top
Preheat
Temp. X
Wave
Soldering
process Contin:o:s
Wave
Soldering
Station
David
Zhang J:ne, 2011
Randomly
selected es Excel form
Bot
Preheat
Temp X
Wave
Soldering
process Contin:o:s
Wave
Soldering
Station
David
Zhang J:ne, 2011
Randomly
selected es Excel form
Top Wave
Temp X
Wave
Soldering
process Contin:o:s
Wave
Soldering
Station
David
Zhang J:ne, 2011
Randomly
selected es Excel form
Bot Wave
Temp X
Wave
Soldering
process Contin:o:s
Wave
Soldering
Station
David
Zhang J:ne, 2011
Randomly
selected es Excel form
Data Measurement PIan
MKS Confidential 13
Setting Pre-heat Zone
1 Temp. ()
Pre-heat Zone 2
Temp.( )
Pre-heat Zone 3
Temp.( )
Wave Temp. ( ) Speed ( cm / min)
Top Side / 140 145
242 103
Bottom Side 130 140 145
R:n profile per below machine setting for data collection process.
Machine Settings
MKS Confidential 14
Date Top Max. Rising Bot Max Rising Top Preheat Temp. Bot Preheat Temp Top Wave Temp. Bot Wave Temp
1.67 1.73 119.1 120.2 157.6 238.7
1.64 1.7 119.73 120.3 163.1 238.7
1.61 1.69 119.2 120.2 160.3 238.9
1.65 1.68 119.1 120.17 159.7 236.9
1.59 1.56 119.1 119.7 166 235
1.59 1.73 119.5 120.3 158.7 238.2
1.61 1.59 119.2 119.5 160.7 237.5
1.56 1.6 118.7 158.3 233.73
1.66 1.58 119.8 119.9 165 234.1
1.64 1.63 118.8 119.1 164 237.6
1.6 1.62 118.2 119.2 159.9 235.2
1.64 1.65 119 120 159.5 234.2
1.61 1.64 118.9 119.4 156.2 238.2
1.63 1.61 119.5 120.5 161.9 236.2
1.63 1.62 119.4 119.67 163.9 238.6
1.61 1.66 118.9 119.7 158.6 238.8
1.61 1.58 119.9 119.6 161.1 235.7
1.64 1.76 118.7 119.7 157.6 237.7
1.61 1.61 118.8 119.6 157.5 237.1
1.62 1.62 119 119.2 161.5 236.5
1.68 1.67 118.2 155.8 235
1.62 1.61 119.3 119.4 160.6 235.9
1.64 1.66 119 118.8 156.1 237.7
1.65 1.67 119 119 158.4 240
2011/6/13 1.64 1.65 119.4 119.3 163.5 234.8
1.64 1.65 119.1 119.2 161 235
1.64 1.67 119.1 119.1 157.6 237.1
2011/5/24
2011/5/26
2011/5/27
2011/5/30
2011/5/31
2011/6/1
2011/6/2
2011/6/3
2011/6/7
2011/6/8
2011/6/9
2011/6/10
2011/6/14
CoIIected Data Sets
MKS Confidential 15
1.99 1. 1.9 1.? 1.99 1. 1.9
99
9
99
89
9
9
9
49
`9
?9
19

1
Tr H=x. 8+>+u

u
f
Mean 1.627
StDev 0.02628
N 27
AD 0.550
P-Val:e 0.142
r=+l+f lf Tr H=x. 8+>+u
\rm=l
1?1.9 1?9. 1?9.9 119. 119.9 118. 118.9
99
9
99
89
9
9
9
49
`9
?9
19

1
f r h= f T mr

u
f
Mean 119.5
StDev 0.5533
N 27
AD 0.234
P-Val:e 0.772
r=+l+f lf f rh=f Tmr
\rm=l
1. 1.9 1?. 19.9 1. 1.9
99
9
99
89
9
9
9
49
`9
?9
19

1
Tr "=v Tmr.

u
f
Mean 160.2
StDev 2.792
N 27
AD 0.303
P-Val:e 0.549
r=+l+f lf Tr "=v Tmr.
\rm=l
?41 ?49 ?`9 ?`8 ?` ?` ?` ?`4 ?`` ?`?
99
9
99
89
9
9
9
49
`9
?9
19

1
f "=v T mr

u
f
Mean 236.8
StDev 1.734
N 27
AD 0.444
P-Val:e 0.264
r=+l+f lf f "=v Tmr
\rm=l
Data AnaIysis ormaIity Test
1?9.9 119. 119.9 118. 118.9
99
9
99
89
9
9
9
49
`9
?9
19

1
T r r h = f = l T mr .

u
f
Mean 119.1
StDev 0.4081
N 26
AD 0.504
P-Val:e 0.186
r=+l+f lf Tr rh=f =l Tmr.
\rm=l
W P-;aIue > 0.05
W AII data are normaIIy distributed
1. 1.9 1. 1.9 1. 1.9
99
9
99
89
9
9
9
49
`9
?9
19

1
f r h =f H= x. 8+ >+ u l r

u
f
Mean 1.646
StDev 0.04917
N 27
AD 0.258
P-Val:e 0.692
r=+l+f lf f rh=f H=x. 8+>+u lr
\rm=l
MKS Confidential 16
Data AnaIysis - Hypothesis Test
Verified the two data sets Top Preheat Max. Rising Slope
and Bot. Preheat Max. Rising, if both witho:t a statistical
significant difference, than two meas:res co:ld be
combined and defined with an identical process limits.
Paired T-Test and C: Top Max. Rising, Bot Max Rising
N Mean StDev SE Mean
Top Preheat Max. Rising 27 1.62704 0.02628 0.00506
Bot Preheat Max. Rising 27 1.64593 0.04917 0.00946
Difference 27 -0.01889 0.04627 0.00890
95% C for mean difference: (-0.03719, -0.00058)
T-Test of mean difference = 0 (vs not = 0): T-Val:e = -2.12 P-Val:e = 0.044
Since P-;aIue is < 0.05, rejected the uII Hypothesis. That is the two measures
- Top Max. Rising and the Bottom Max. Rising with statisticaI significant
difference.
MKS Confidential 17
Verified the two data sets Top Preheat Max. Peak Temp. and Bot.
Preheat Peak Temp. if both meas:res witho:t a statistical significant
difference, then these two meas:res co:ld be combined and defined
with an identical process limit.
Paired T-Test and C: Top Preheat Temp., Bot Preheat Temp
N Mean StDev SE Mean
Top Preheat Temp. 27 119.052 0.421 0.092
Bot Preheat Temp 27 119.511 0.535 0.117
Difference 27 -0.459 0.493 0.108
95% C for mean difference: (-0.683, -0.235)
T-Test of mean difference = 0 (vs not = 0): T-Val:e = -4.27 P-Val:e = 0.000
Since P-;aIue is < 0.05, Reject the uII Hypothesis. That mean the two measures -
Top Max. Preheat Temp. and the Bottom Max. Preheat Temp. with statisticaI
significant difference.
Data AnaIysis - Hypothesis Test
MKS Confidential 18
?8 ? ?? 19 1 1` 19 4 1
1.8
1.
1.
1.
> r v =f + u
T
u
d
+
v
+
d
v
=
l

\
=
l
v

_
X=1.6464
UCL=1.7815
LCL=1.5114
?8 ? ?? 19 1 1` 19 4 1
9.?9
9.1
9.19
9.9
9.99
> r v =f + u
H

v
+
u


8
=
u

__
MR=0.0508
UCL=0.1659
LCL=0
6
1
2
1
I-MR Chart of Bot Preheat Max. Rising SIope
Define Process Limits -
In-controI/Out of ControI and StabiIity Verification
W Test the six statistical :sing r:les as shown.
W With exception of the statistical listed below others
are in control and stable.
~ Bot. Preheat Max. Rising Slope
~ Bot. Preheat Peak Temp.
? ?? 19 1 1` 19 4 1
1.9
1.
1.9
1.
>rv=f+u
T
u
d
+
v
+
d
v
=
l

\
=
l
v

_
X-1.?9
' l-1.9
l l-1.`
? ?? 19 1 1` 19 4 1
9.98
9.9
9.94
9.9?
9.99
>rv=f+u
H

v
+
u


8
=
u

__
H 8 -9.9?4
' l-9.981
l l-9
TH 8 h=rf T r rh=f H =x. 8 +>+u lr
MKS Confidential 19
? ?? 19 1 1` 19 4 1
1?9.
1?9.9
119.
119.9
118.
Obser;ation
I
n
d
i
;
i
d
u
a
I

V
a
I
u
e
UCL=120.827
LCL=118.257
_
X=119.542
? ?? 19 1 1` 19 4 1
1.
1.?
9.8
9.4
9.9
Obser;ation
M
o
;
i
n
g

R
a
n
g
e
__
MR=0.483
UCL=1.578
LCL=0
1
6
6
I-MR Chart of Bot Preheat Peak Temp
? ?? 19 1 1` 19 4 1
19
1
19
1
19
Obser;ation
I
n
d
i
;
i
d
u
a
I

V
a
I
u
e
UCL=169.75
LCL=150.56
_
X=160.15
? ?? 19 1 1` 19 4 1
1?
9

`
9
Obser;ation
M
o
;
i
n
g

R
a
n
g
e
__
MR=3.61
UCL=11.79
LCL=0
I-MR Chart of Top Wa;e Temp.
? ?? 19 1 1` 19 4 1
?4`
?49
?`
?`4
?`1
Obser;ation
I
n
d
i
;
i
d
u
a
I

V
a
I
u
e
_
X=236.78
UCL=241.68
LCL=231.88
? ?? 19 1 1` 19 4 1
.9
4.
`.9
1.
9.9
Obser;ation
M
o
;
i
n
g

R
a
n
g
e
__
MR=1.844
UCL=6.024
LCL=0
I-MR Chart of Bot Wa;e Temp
Define Process Limits -
? ?? 19 1 1` 19 4 1
1?9
119
118
>rv=f+u
T
u
d
+
v
+
d
v
=
l

\
=
l
v

_
X-119.1`
' l-1?9.4?8
l l-11.88
? ?? 19 1 1` 19 4 1
1.
1.?
9.8
9.4
9.9
>rv=f+u
H

v
+
u


8
=
u

__
H 8 -9.49
' l-1.
l l-9
TH 8 h=rf T r rh=f =l Tm r.
MKS Confidential 20
Define Process Limits -
The team performed a ca:se & effect analysis and determined the
most possible root ca:se as following.
Actions have been taken to eliminate the special ca:se variations
so that the process co:ld be drawn back :nder control and stable.
W The freq:ency of changing the al:min:m tape has been specified at 1
test / 1 times
W Tape dimension req:ired: 6mm x 6mm
W R:n test :ntil setting temp. being reached
Defined the Process Limits for the six statistical per the table as
shown below.
Spec. Top Max. Rising Bot Max Rising Top Preheat Temp. Bot Preheat Temp Top Wa;e Temp. Bot Wa;e Temp
UCL= x' + 3s 1.71 1.79 120.4 120.8 169.9 241.8
x' 1.63 1.65 119.1 119.6 160.2 236.8
LCL= x' - 3s 1.55 1.51 117.8 118.4 150.5 231.8
MKS Confidential 21
PiIoted to Measure PWI
The profile checker, machine
setting and sample size (26) :sed
to collect the PW was identical to
the baseline data collection. The
pilot was r:n for 13 days starting
on J:n 13, r:n two profiles per
each day.
The collected 26 PWs were less
than 100%, which reveals that the
pre-defined Process Limits are
rob:st being :sed to meas:re the
performance of the wave soldering
machine.
?` ?1 19 1 1 1` 11 9 ` 1
199.99
.99
9.99
?.99
9.99
>rv=f+u
T
u
d
+
v
+
d
v
=
l

\
=
l
v

_
X-1.9
' l-9.
l l-.4
?` ?1 19 1 1 1` 11 9 ` 1
9.99
4.99
`9.99
1.99
9.99
>rv=f+u
H

v
+
u


8
=
u

__
H 8 -1.`?
' l-.8
l l-9.99
TH8 h=rf "Tl+fd
MKS Confidential 22
Define ControI Limits for PWI
The control limits of PW is specified as following based on the
piloted meas:rements.
Targeted PW: >75% No need to take action
Process ndicator: >75% Keep close monitor the next data
O:t of Spec: >97% Root ca:se analysis & take corrective action
Wa;e SoIdering PWI DaiIy Performance
PWI (%)
0
15
30
45
60
75
90
105
120
26
May
26
May
27
May
27
May
30
May
30
May
31
May
31
May
01
J:n
01
J:n
02
J:n
02
J:n
03
J:n
03
J:n
07
J:n
07
J:n
08
J:n
08
J:n
09
J:n
09
J:n
10
J:n
10
J:n
13
J:n
13
J:n
14
J:n
14
J:n
GoodPWI
WarningPWI
OutOfSpec
MKS Confidential 23
Process ControI PIan
PW is indentified as the indicator to meas:re the
performance of the wave soldering machine.
PE technician m:st to meas:re the PW by the mean of
r:nning thermal profile every day.
Profile daily record sho:ld be reviewed by engineer.
Any o:t of control limits data sho:ld be analyzed by
PCBA PE engineer.
MKS Confidential 24
Process ControI PIan
Project
ame:
ntrod:ction of PW methodology for monitoring the
performance of Wave Soldering Machine
Process Step Key ndicator
X
(control)
or
Y
(monitor)
Prod:ct/Pr
ocess
Specificatio
ns/Target
Eval:ati
on/
Meas:re
ment
Techniq
:e
%P/Total
(R&R)
%P/Toleranc
e
Samp
le
Size
Sample
Freq:enc
y
Respons
ibility
Control
Method
Contingency
Action Plan
Set wave
soldering
machine
Conveyor Speed x 1.03m/min
Speedo
meter 1 Day David
Profile
Program
Calibrate
machine
Bot. Pre-heat Zone
#1 Temp. x 130
T
h
e
r
m
o
m
e
t
e
r
1 Day David
Profile
Program
Top Pre-heat Zone
#2 Temp. x 130 1 Day David
Profile
Program
Bot. Pre-heat Zone
#2 Temp. x 140 1 Day David
Profile
Program
Top Pre-heat Zone
#3 Temp. x 140 1 Day David
Profile
Program
Bot Pre-heat Zone #3
Temp. x 145 1 Day David
Profile
Program
,;0%025 x 145 1 Day David
Profile
Program
Profiling
PW Y 0% - 99% 1 Day David
Profile
Record & -
MR Control
Chart
Calibrate
machine
&
Check Profile
Test Board
MKS Confidential 25
ControI PIan Machine ProfiIe DaiIy Records
Print the Machine Profile o:t and send it to the engineer for
approval.
PWI ResuIt
Accept Criteria
Signoff
MKS Confidential 26
Benefits of Ranking with PWI
reatly simplifies the profiling process.
Confidence in thermal process capability
Profiles can be easily compared
mproved ":ality Control
Significant prod:ction savings

Вам также может понравиться